From 8fd00b4d7014b00448eb33cf0590815304769798 Mon Sep 17 00:00:00 2001 From: Jiri Slaby Date: Wed, 26 Aug 2009 18:41:16 +0200 Subject: rlimits: security, add task_struct to setrlimit Add task_struct to task_setrlimit of security_operations to be able to set rlimit of task other than current. Signed-off-by: Jiri Slaby Acked-by: Eric Paris Acked-by: James Morris --- kernel/sys.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/sys.c b/kernel/sys.c index e83ddbbaf89d..1ba4522689d4 100644 --- a/kernel/sys.c +++ b/kernel/sys.c @@ -1290,7 +1290,7 @@ SYSCALL_DEFINE2(setrlimit, unsigned int, resource, struct rlimit __user *, rlim) if (resource == RLIMIT_NOFILE && new_rlim.rlim_max > sysctl_nr_open) return -EPERM; - retval = security_task_setrlimit(resource, &new_rlim); + retval = security_task_setrlimit(current, resource, &new_rlim); if (retval) return retval; -- cgit v1.2.3 From 5ab46b345e418747b3a52f0892680c0745c4223c Mon Sep 17 00:00:00 2001 From: Jiri Slaby Date: Fri, 28 Aug 2009 14:05:12 +0200 Subject: rlimits: add task_struct to update_rlimit_cpu Add task_struct as a parameter to update_rlimit_cpu to be able to set rlimit_cpu of different task than current. Signed-off-by: Jiri Slaby Acked-by: James Morris --- kernel/posix-cpu-timers.c | 8 ++++---- kernel/sys.c | 2 +- 2 files changed, 5 insertions(+), 5 deletions(-) (limited to 'kernel') diff --git a/kernel/posix-cpu-timers.c b/kernel/posix-cpu-timers.c index 9829646d399c..0513900995ce 100644 --- a/kernel/posix-cpu-timers.c +++ b/kernel/posix-cpu-timers.c @@ -16,13 +16,13 @@ * siglock protection since other code may update expiration cache as * well. */ -void update_rlimit_cpu(unsigned long rlim_new) +void update_rlimit_cpu(struct task_struct *task, unsigned long rlim_new) { cputime_t cputime = secs_to_cputime(rlim_new); - spin_lock_irq(¤t->sighand->siglock); - set_process_cpu_timer(current, CPUCLOCK_PROF, &cputime, NULL); - spin_unlock_irq(¤t->sighand->siglock); + spin_lock_irq(&task->sighand->siglock); + set_process_cpu_timer(task, CPUCLOCK_PROF, &cputime, NULL); + spin_unlock_irq(&task->sighand->siglock); } static int check_clock(const clockid_t which_clock) diff --git a/kernel/sys.c b/kernel/sys.c index 1ba4522689d4..f5183b08adfc 100644 --- a/kernel/sys.c +++ b/kernel/sys.c @@ -1320,7 +1320,7 @@ SYSCALL_DEFINE2(setrlimit, unsigned int, resource, struct rlimit __user *, rlim) if (new_rlim.rlim_cur == RLIM_INFINITY) goto out; - update_rlimit_cpu(new_rlim.rlim_cur); + update_rlimit_cpu(current, new_rlim.rlim_cur); out: return 0; } -- cgit v1.2.3 From 2fb9d2689a0041b88b25bc3187eada2968e25995 Mon Sep 17 00:00:00 2001 From: Oleg Nesterov Date: Thu, 3 Sep 2009 19:21:45 +0200 Subject: rlimits: make sure ->rlim_max never grows in sys_setrlimit Mostly preparation for Jiri's changes, but probably makes sense anyway. sys_setrlimit() checks new_rlim.rlim_max <= old_rlim->rlim_max, but when it takes task_lock() old_rlim->rlim_max can be already lowered. Move this check under task_lock(). Currently this is not important, we can only race with our sub-thread, this means the application is stupid. But when we change the code to allow the update of !current task's limits, it becomes important to make sure ->rlim_max can be lowered "reliably" even if we race with the application doing sys_setrlimit(). Signed-off-by: Oleg Nesterov Signed-off-by: Jiri Slaby --- kernel/sys.c | 15 ++++++++------- 1 file changed, 8 insertions(+), 7 deletions(-) (limited to 'kernel') diff --git a/kernel/sys.c b/kernel/sys.c index f5183b08adfc..f2b2d7aa3818 100644 --- a/kernel/sys.c +++ b/kernel/sys.c @@ -1283,10 +1283,6 @@ SYSCALL_DEFINE2(setrlimit, unsigned int, resource, struct rlimit __user *, rlim) return -EFAULT; if (new_rlim.rlim_cur > new_rlim.rlim_max) return -EINVAL; - old_rlim = current->signal->rlim + resource; - if ((new_rlim.rlim_max > old_rlim->rlim_max) && - !capable(CAP_SYS_RESOURCE)) - return -EPERM; if (resource == RLIMIT_NOFILE && new_rlim.rlim_max > sysctl_nr_open) return -EPERM; @@ -1304,11 +1300,16 @@ SYSCALL_DEFINE2(setrlimit, unsigned int, resource, struct rlimit __user *, rlim) new_rlim.rlim_cur = 1; } + old_rlim = current->signal->rlim + resource; task_lock(current->group_leader); - *old_rlim = new_rlim; + if (new_rlim.rlim_max > old_rlim->rlim_max && + !capable(CAP_SYS_RESOURCE)) + retval = -EPERM; + else + *old_rlim = new_rlim; task_unlock(current->group_leader); - if (resource != RLIMIT_CPU) + if (retval || resource != RLIMIT_CPU) goto out; /* @@ -1322,7 +1323,7 @@ SYSCALL_DEFINE2(setrlimit, unsigned int, resource, struct rlimit __user *, rlim) update_rlimit_cpu(current, new_rlim.rlim_cur); out: - return 0; + return retval; } /* -- cgit v1.2.3 From 7855c35da7ba16b389d17710401c4a55a3ea2102 Mon Sep 17 00:00:00 2001 From: Jiri Slaby Date: Wed, 26 Aug 2009 23:45:34 +0200 Subject: rlimits: split sys_setrlimit Create do_setrlimit from sys_setrlimit and declare do_setrlimit in the resource header. This is the first phase to have generic do_prlimit which allows to be called from read, write and compat rlimits code. The new do_setrlimit also accepts a task pointer to change the limits of. Currently, it cannot be other than current, but this will change with locking later. Also pass tsk->group_leader to security_task_setrlimit to check whether current is allowed to change rlimits of the process and not its arbitrary thread because it makes more sense given that rlimit are per process and not per-thread. Signed-off-by: Jiri Slaby --- kernel/sys.c | 40 ++++++++++++++++++++++++---------------- 1 file changed, 24 insertions(+), 16 deletions(-) (limited to 'kernel') diff --git a/kernel/sys.c b/kernel/sys.c index f2b2d7aa3818..b5b96e30e0d6 100644 --- a/kernel/sys.c +++ b/kernel/sys.c @@ -1272,42 +1272,41 @@ SYSCALL_DEFINE2(old_getrlimit, unsigned int, resource, #endif -SYSCALL_DEFINE2(setrlimit, unsigned int, resource, struct rlimit __user *, rlim) +int do_setrlimit(struct task_struct *tsk, unsigned int resource, + struct rlimit *new_rlim) { - struct rlimit new_rlim, *old_rlim; + struct rlimit *old_rlim; int retval; if (resource >= RLIM_NLIMITS) return -EINVAL; - if (copy_from_user(&new_rlim, rlim, sizeof(*rlim))) - return -EFAULT; - if (new_rlim.rlim_cur > new_rlim.rlim_max) + if (new_rlim->rlim_cur > new_rlim->rlim_max) return -EINVAL; - if (resource == RLIMIT_NOFILE && new_rlim.rlim_max > sysctl_nr_open) + if (resource == RLIMIT_NOFILE && new_rlim->rlim_max > sysctl_nr_open) return -EPERM; - retval = security_task_setrlimit(current, resource, &new_rlim); + retval = security_task_setrlimit(tsk->group_leader, resource, new_rlim); if (retval) return retval; - if (resource == RLIMIT_CPU && new_rlim.rlim_cur == 0) { + if (resource == RLIMIT_CPU && new_rlim->rlim_cur == 0) { /* * The caller is asking for an immediate RLIMIT_CPU * expiry. But we use the zero value to mean "it was * never set". So let's cheat and make it one second * instead */ - new_rlim.rlim_cur = 1; + new_rlim->rlim_cur = 1; } - old_rlim = current->signal->rlim + resource; - task_lock(current->group_leader); - if (new_rlim.rlim_max > old_rlim->rlim_max && + old_rlim = tsk->signal->rlim + resource; + task_lock(tsk->group_leader); + if (new_rlim->rlim_max > old_rlim->rlim_max && !capable(CAP_SYS_RESOURCE)) retval = -EPERM; else - *old_rlim = new_rlim; - task_unlock(current->group_leader); + *old_rlim = *new_rlim; + task_unlock(tsk->group_leader); if (retval || resource != RLIMIT_CPU) goto out; @@ -1318,14 +1317,23 @@ SYSCALL_DEFINE2(setrlimit, unsigned int, resource, struct rlimit __user *, rlim) * very long-standing error, and fixing it now risks breakage of * applications, so we live with it */ - if (new_rlim.rlim_cur == RLIM_INFINITY) + if (new_rlim->rlim_cur == RLIM_INFINITY) goto out; - update_rlimit_cpu(current, new_rlim.rlim_cur); + update_rlimit_cpu(tsk, new_rlim->rlim_cur); out: return retval; } +SYSCALL_DEFINE2(setrlimit, unsigned int, resource, struct rlimit __user *, rlim) +{ + struct rlimit new_rlim; + + if (copy_from_user(&new_rlim, rlim, sizeof(*rlim))) + return -EFAULT; + return do_setrlimit(current, resource, &new_rlim); +} + /* * It would make sense to put struct rusage in the task_struct, * except that would make the task_struct be *really big*. After -- cgit v1.2.3 From 1c1e618ddd15f69fd87ccea596769f78c8065504 Mon Sep 17 00:00:00 2001 From: Jiri Slaby Date: Fri, 28 Aug 2009 14:08:17 +0200 Subject: rlimits: allow setrlimit to non-current tasks Add locking to allow setrlimit accept task parameter other than current. Namely, lock tasklist_lock for read and check whether the task structure has sighand non-null. Do all the signal processing under that lock still held. There are some points: 1) security_task_setrlimit is now called with that lock held. This is not new, many security_* functions are called with this lock held already so it doesn't harm (all this security_* stuff does almost the same). 2) task->sighand->siglock (in update_rlimit_cpu) is nested in tasklist_lock. This dependence is already existing. 3) tsk->alloc_lock is nested in tasklist_lock. This is OK too, already existing dependence. Signed-off-by: Jiri Slaby Cc: Oleg Nesterov --- kernel/sys.c | 11 ++++++++++- 1 file changed, 10 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/sys.c b/kernel/sys.c index b5b96e30e0d6..9dbcbbcce153 100644 --- a/kernel/sys.c +++ b/kernel/sys.c @@ -1272,6 +1272,7 @@ SYSCALL_DEFINE2(old_getrlimit, unsigned int, resource, #endif +/* make sure you are allowed to change @tsk limits before calling this */ int do_setrlimit(struct task_struct *tsk, unsigned int resource, struct rlimit *new_rlim) { @@ -1285,9 +1286,16 @@ int do_setrlimit(struct task_struct *tsk, unsigned int resource, if (resource == RLIMIT_NOFILE && new_rlim->rlim_max > sysctl_nr_open) return -EPERM; + /* protect tsk->signal and tsk->sighand from disappearing */ + read_lock(&tasklist_lock); + if (!tsk->sighand) { + retval = -ESRCH; + goto out; + } + retval = security_task_setrlimit(tsk->group_leader, resource, new_rlim); if (retval) - return retval; + goto out; if (resource == RLIMIT_CPU && new_rlim->rlim_cur == 0) { /* @@ -1322,6 +1330,7 @@ int do_setrlimit(struct task_struct *tsk, unsigned int resource, update_rlimit_cpu(tsk, new_rlim->rlim_cur); out: + read_unlock(&tasklist_lock); return retval; } -- cgit v1.2.3 From 86f162f4c75ceb6daf43165469eeeca1bc3d4639 Mon Sep 17 00:00:00 2001 From: Jiri Slaby Date: Sat, 14 Nov 2009 17:37:04 +0100 Subject: rlimits: do security check under task_lock Do security_task_setrlimit under task_lock. Other tasks may change limits under our hands while we are checking limits inside the function. From now on, they can't. Note that all the security work is done under a spinlock here now. Security hooks count with that, they are called from interrupt context (like security_task_kill) and with spinlocks already held (e.g. capable->security_capable). Signed-off-by: Jiri Slaby Acked-by: James Morris Cc: Heiko Carstens --- kernel/sys.c | 20 ++++++++++---------- 1 file changed, 10 insertions(+), 10 deletions(-) (limited to 'kernel') diff --git a/kernel/sys.c b/kernel/sys.c index 9dbcbbcce153..c762eebdebf7 100644 --- a/kernel/sys.c +++ b/kernel/sys.c @@ -1277,7 +1277,7 @@ int do_setrlimit(struct task_struct *tsk, unsigned int resource, struct rlimit *new_rlim) { struct rlimit *old_rlim; - int retval; + int retval = 0; if (resource >= RLIM_NLIMITS) return -EINVAL; @@ -1293,9 +1293,14 @@ int do_setrlimit(struct task_struct *tsk, unsigned int resource, goto out; } - retval = security_task_setrlimit(tsk->group_leader, resource, new_rlim); - if (retval) - goto out; + old_rlim = tsk->signal->rlim + resource; + task_lock(tsk->group_leader); + if (new_rlim->rlim_max > old_rlim->rlim_max && + !capable(CAP_SYS_RESOURCE)) + retval = -EPERM; + if (!retval) + retval = security_task_setrlimit(tsk->group_leader, resource, + new_rlim); if (resource == RLIMIT_CPU && new_rlim->rlim_cur == 0) { /* @@ -1307,12 +1312,7 @@ int do_setrlimit(struct task_struct *tsk, unsigned int resource, new_rlim->rlim_cur = 1; } - old_rlim = tsk->signal->rlim + resource; - task_lock(tsk->group_leader); - if (new_rlim->rlim_max > old_rlim->rlim_max && - !capable(CAP_SYS_RESOURCE)) - retval = -EPERM; - else + if (!retval) *old_rlim = *new_rlim; task_unlock(tsk->group_leader); -- cgit v1.2.3 From 5b41535aac0c07135ff6a4c5c2ae115d1c20c0bc Mon Sep 17 00:00:00 2001 From: Jiri Slaby Date: Wed, 24 Mar 2010 16:11:29 +0100 Subject: rlimits: redo do_setrlimit to more generic do_prlimit It now allows also reading of limits. I.e. all read and writes will later use this function. It takes two parameters, new and old limits which can be both NULL. If new is non-NULL, the value in it is set to rlimits. If old is non-NULL, current rlimits are stored there. If both are non-NULL, old are stored prior to setting the new ones, atomically. (Similar to sigaction.) Signed-off-by: Jiri Slaby --- kernel/sys.c | 71 +++++++++++++++++++++++++++++++----------------------------- 1 file changed, 37 insertions(+), 34 deletions(-) (limited to 'kernel') diff --git a/kernel/sys.c b/kernel/sys.c index c762eebdebf7..bc7d1be0960e 100644 --- a/kernel/sys.c +++ b/kernel/sys.c @@ -1273,18 +1273,21 @@ SYSCALL_DEFINE2(old_getrlimit, unsigned int, resource, #endif /* make sure you are allowed to change @tsk limits before calling this */ -int do_setrlimit(struct task_struct *tsk, unsigned int resource, - struct rlimit *new_rlim) +int do_prlimit(struct task_struct *tsk, unsigned int resource, + struct rlimit *new_rlim, struct rlimit *old_rlim) { - struct rlimit *old_rlim; + struct rlimit *rlim; int retval = 0; if (resource >= RLIM_NLIMITS) return -EINVAL; - if (new_rlim->rlim_cur > new_rlim->rlim_max) - return -EINVAL; - if (resource == RLIMIT_NOFILE && new_rlim->rlim_max > sysctl_nr_open) - return -EPERM; + if (new_rlim) { + if (new_rlim->rlim_cur > new_rlim->rlim_max) + return -EINVAL; + if (resource == RLIMIT_NOFILE && + new_rlim->rlim_max > sysctl_nr_open) + return -EPERM; + } /* protect tsk->signal and tsk->sighand from disappearing */ read_lock(&tasklist_lock); @@ -1293,42 +1296,42 @@ int do_setrlimit(struct task_struct *tsk, unsigned int resource, goto out; } - old_rlim = tsk->signal->rlim + resource; + rlim = tsk->signal->rlim + resource; task_lock(tsk->group_leader); - if (new_rlim->rlim_max > old_rlim->rlim_max && - !capable(CAP_SYS_RESOURCE)) - retval = -EPERM; - if (!retval) - retval = security_task_setrlimit(tsk->group_leader, resource, - new_rlim); - - if (resource == RLIMIT_CPU && new_rlim->rlim_cur == 0) { - /* - * The caller is asking for an immediate RLIMIT_CPU - * expiry. But we use the zero value to mean "it was - * never set". So let's cheat and make it one second - * instead - */ - new_rlim->rlim_cur = 1; + if (new_rlim) { + if (new_rlim->rlim_max > rlim->rlim_max && + !capable(CAP_SYS_RESOURCE)) + retval = -EPERM; + if (!retval) + retval = security_task_setrlimit(tsk->group_leader, + resource, new_rlim); + if (resource == RLIMIT_CPU && new_rlim->rlim_cur == 0) { + /* + * The caller is asking for an immediate RLIMIT_CPU + * expiry. But we use the zero value to mean "it was + * never set". So let's cheat and make it one second + * instead + */ + new_rlim->rlim_cur = 1; + } + } + if (!retval) { + if (old_rlim) + *old_rlim = *rlim; + if (new_rlim) + *rlim = *new_rlim; } - - if (!retval) - *old_rlim = *new_rlim; task_unlock(tsk->group_leader); - if (retval || resource != RLIMIT_CPU) - goto out; - /* * RLIMIT_CPU handling. Note that the kernel fails to return an error * code if it rejected the user's attempt to set RLIMIT_CPU. This is a * very long-standing error, and fixing it now risks breakage of * applications, so we live with it */ - if (new_rlim->rlim_cur == RLIM_INFINITY) - goto out; - - update_rlimit_cpu(tsk, new_rlim->rlim_cur); + if (!retval && new_rlim && resource == RLIMIT_CPU && + new_rlim->rlim_cur != RLIM_INFINITY) + update_rlimit_cpu(tsk, new_rlim->rlim_cur); out: read_unlock(&tasklist_lock); return retval; @@ -1340,7 +1343,7 @@ SYSCALL_DEFINE2(setrlimit, unsigned int, resource, struct rlimit __user *, rlim) if (copy_from_user(&new_rlim, rlim, sizeof(*rlim))) return -EFAULT; - return do_setrlimit(current, resource, &new_rlim); + return do_prlimit(current, resource, &new_rlim, NULL); } /* -- cgit v1.2.3 From b95183453af2ed14a5c7027e58049c9fd17e92ce Mon Sep 17 00:00:00 2001 From: Jiri Slaby Date: Tue, 4 May 2010 11:28:25 +0200 Subject: rlimits: switch more rlimit syscalls to do_prlimit After we added more generic do_prlimit, switch sys_getrlimit to that. Also switch compat handling, so we can get rid of ugly __user casts and avoid setting process' address limit to kernel data and back. Signed-off-by: Jiri Slaby --- kernel/compat.c | 17 +++-------------- kernel/sys.c | 17 ++++++++--------- 2 files changed, 11 insertions(+), 23 deletions(-) (limited to 'kernel') diff --git a/kernel/compat.c b/kernel/compat.c index 5adab05a3172..e167efce8423 100644 --- a/kernel/compat.c +++ b/kernel/compat.c @@ -279,11 +279,6 @@ asmlinkage long compat_sys_setrlimit(unsigned int resource, struct compat_rlimit __user *rlim) { struct rlimit r; - int ret; - mm_segment_t old_fs = get_fs (); - - if (resource >= RLIM_NLIMITS) - return -EINVAL; if (!access_ok(VERIFY_READ, rlim, sizeof(*rlim)) || __get_user(r.rlim_cur, &rlim->rlim_cur) || @@ -294,10 +289,7 @@ asmlinkage long compat_sys_setrlimit(unsigned int resource, r.rlim_cur = RLIM_INFINITY; if (r.rlim_max == COMPAT_RLIM_INFINITY) r.rlim_max = RLIM_INFINITY; - set_fs(KERNEL_DS); - ret = sys_setrlimit(resource, (struct rlimit __user *) &r); - set_fs(old_fs); - return ret; + return do_prlimit(current, resource, &r, NULL); } #ifdef COMPAT_RLIM_OLD_INFINITY @@ -329,16 +321,13 @@ asmlinkage long compat_sys_old_getrlimit(unsigned int resource, #endif -asmlinkage long compat_sys_getrlimit (unsigned int resource, +asmlinkage long compat_sys_getrlimit(unsigned int resource, struct compat_rlimit __user *rlim) { struct rlimit r; int ret; - mm_segment_t old_fs = get_fs(); - set_fs(KERNEL_DS); - ret = sys_getrlimit(resource, (struct rlimit __user *) &r); - set_fs(old_fs); + ret = do_prlimit(current, resource, NULL, &r); if (!ret) { if (r.rlim_cur > COMPAT_RLIM_INFINITY) r.rlim_cur = COMPAT_RLIM_INFINITY; diff --git a/kernel/sys.c b/kernel/sys.c index bc7d1be0960e..9da98dd47276 100644 --- a/kernel/sys.c +++ b/kernel/sys.c @@ -1236,15 +1236,14 @@ SYSCALL_DEFINE2(setdomainname, char __user *, name, int, len) SYSCALL_DEFINE2(getrlimit, unsigned int, resource, struct rlimit __user *, rlim) { - if (resource >= RLIM_NLIMITS) - return -EINVAL; - else { - struct rlimit value; - task_lock(current->group_leader); - value = current->signal->rlim[resource]; - task_unlock(current->group_leader); - return copy_to_user(rlim, &value, sizeof(*rlim)) ? -EFAULT : 0; - } + struct rlimit value; + int ret; + + ret = do_prlimit(current, resource, NULL, &value); + if (!ret) + ret = copy_to_user(rlim, &value, sizeof(*rlim)) ? -EFAULT : 0; + + return ret; } #ifdef __ARCH_WANT_SYS_OLD_GETRLIMIT -- cgit v1.2.3 From c022a0acad534fd5f5d5f17280f6d4d135e74e81 Mon Sep 17 00:00:00 2001 From: Jiri Slaby Date: Tue, 4 May 2010 18:03:50 +0200 Subject: rlimits: implement prlimit64 syscall This patch adds the code to support the sys_prlimit64 syscall which modifies-and-returns the rlim values of a selected process atomically. The first parameter, pid, being 0 means current process. Unlike the current implementation, it is a generic interface, architecture indepentent so that we needn't handle compat stuff anymore. In the future, after glibc start to use this we can deprecate sys_setrlimit and sys_getrlimit in favor to clean up the code finally. It also adds a possibility of changing limits of other processes. We check the user's permissions to do that and if it succeeds, the new limits are propagated online. This is good for large scale applications such as SAP or databases where administrators need to change limits time by time (e.g. on crashes increase core size). And it is unacceptable to restart the service. For safety, all rlim users now either use accessors or doesn't need them due to - locking - the fact a process was just forked and nobody else knows about it yet (and nobody can't thus read/write limits) hence it is safe to modify limits now. The limitation is that we currently stay at ulong internal representation. So the rlim64_is_infinity check is used where value is compared against ULONG_MAX on 32-bit which is the maximum value there. And since internally the limits are held in struct rlimit, converters which are used before and after do_prlimit call in sys_prlimit64 are introduced. Signed-off-by: Jiri Slaby --- kernel/sys.c | 94 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 94 insertions(+) (limited to 'kernel') diff --git a/kernel/sys.c b/kernel/sys.c index 9da98dd47276..e9ad44489828 100644 --- a/kernel/sys.c +++ b/kernel/sys.c @@ -1271,6 +1271,39 @@ SYSCALL_DEFINE2(old_getrlimit, unsigned int, resource, #endif +static inline bool rlim64_is_infinity(__u64 rlim64) +{ +#if BITS_PER_LONG < 64 + return rlim64 >= ULONG_MAX; +#else + return rlim64 == RLIM64_INFINITY; +#endif +} + +static void rlim_to_rlim64(const struct rlimit *rlim, struct rlimit64 *rlim64) +{ + if (rlim->rlim_cur == RLIM_INFINITY) + rlim64->rlim_cur = RLIM64_INFINITY; + else + rlim64->rlim_cur = rlim->rlim_cur; + if (rlim->rlim_max == RLIM_INFINITY) + rlim64->rlim_max = RLIM64_INFINITY; + else + rlim64->rlim_max = rlim->rlim_max; +} + +static void rlim64_to_rlim(const struct rlimit64 *rlim64, struct rlimit *rlim) +{ + if (rlim64_is_infinity(rlim64->rlim_cur)) + rlim->rlim_cur = RLIM_INFINITY; + else + rlim->rlim_cur = (unsigned long)rlim64->rlim_cur; + if (rlim64_is_infinity(rlim64->rlim_max)) + rlim->rlim_max = RLIM_INFINITY; + else + rlim->rlim_max = (unsigned long)rlim64->rlim_max; +} + /* make sure you are allowed to change @tsk limits before calling this */ int do_prlimit(struct task_struct *tsk, unsigned int resource, struct rlimit *new_rlim, struct rlimit *old_rlim) @@ -1336,6 +1369,67 @@ out: return retval; } +/* rcu lock must be held */ +static int check_prlimit_permission(struct task_struct *task) +{ + const struct cred *cred = current_cred(), *tcred; + + tcred = __task_cred(task); + if ((cred->uid != tcred->euid || + cred->uid != tcred->suid || + cred->uid != tcred->uid || + cred->gid != tcred->egid || + cred->gid != tcred->sgid || + cred->gid != tcred->gid) && + !capable(CAP_SYS_RESOURCE)) { + return -EPERM; + } + + return 0; +} + +SYSCALL_DEFINE4(prlimit64, pid_t, pid, unsigned int, resource, + const struct rlimit64 __user *, new_rlim, + struct rlimit64 __user *, old_rlim) +{ + struct rlimit64 old64, new64; + struct rlimit old, new; + struct task_struct *tsk; + int ret; + + if (new_rlim) { + if (copy_from_user(&new64, new_rlim, sizeof(new64))) + return -EFAULT; + rlim64_to_rlim(&new64, &new); + } + + rcu_read_lock(); + tsk = pid ? find_task_by_vpid(pid) : current; + if (!tsk) { + rcu_read_unlock(); + return -ESRCH; + } + ret = check_prlimit_permission(tsk); + if (ret) { + rcu_read_unlock(); + return ret; + } + get_task_struct(tsk); + rcu_read_unlock(); + + ret = do_prlimit(tsk, resource, new_rlim ? &new : NULL, + old_rlim ? &old : NULL); + + if (!ret && old_rlim) { + rlim_to_rlim64(&old, &old64); + if (copy_to_user(old_rlim, &old64, sizeof(old64))) + ret = -EFAULT; + } + + put_task_struct(tsk); + return ret; +} + SYSCALL_DEFINE2(setrlimit, unsigned int, resource, struct rlimit __user *, rlim) { struct rlimit new_rlim; -- cgit v1.2.3 From ae7b8f4108bcffb42173f867ce845268c7202d48 Mon Sep 17 00:00:00 2001 From: Eric Paris Date: Thu, 17 Dec 2009 20:12:04 -0500 Subject: Audit: clean up the audit_watch split No real changes, just cleanup to the audit_watch split patch which we done with minimal code changes for easy review. Now fix interfaces to make things work better. Signed-off-by: Eric Paris --- kernel/audit.c | 1 - kernel/audit.h | 13 ++++------ kernel/audit_watch.c | 67 ++++++++++++++++++++++++++++++---------------------- kernel/auditfilter.c | 41 ++++++++++++-------------------- kernel/auditsc.c | 5 ++-- 5 files changed, 60 insertions(+), 67 deletions(-) (limited to 'kernel') diff --git a/kernel/audit.c b/kernel/audit.c index c71bd26631a2..05a32f0d87dc 100644 --- a/kernel/audit.c +++ b/kernel/audit.c @@ -56,7 +56,6 @@ #include #include #include -#include #include #include diff --git a/kernel/audit.h b/kernel/audit.h index 208687be4f30..82c8a09099f4 100644 --- a/kernel/audit.h +++ b/kernel/audit.h @@ -104,20 +104,15 @@ extern void audit_free_rule_rcu(struct rcu_head *); extern struct list_head audit_filter_list[]; /* audit watch functions */ -extern unsigned long audit_watch_inode(struct audit_watch *watch); -extern dev_t audit_watch_dev(struct audit_watch *watch); extern void audit_put_watch(struct audit_watch *watch); extern void audit_get_watch(struct audit_watch *watch); extern int audit_to_watch(struct audit_krule *krule, char *path, int len, u32 op); -extern int audit_add_watch(struct audit_krule *krule); -extern void audit_remove_watch(struct audit_watch *watch); +extern int audit_add_watch(struct audit_krule *krule, struct list_head **list); extern void audit_remove_watch_rule(struct audit_krule *krule, struct list_head *list); -extern void audit_inotify_unregister(struct list_head *in_list); +extern void audit_watch_inotify_unregister(struct list_head *in_list); extern char *audit_watch_path(struct audit_watch *watch); -extern struct list_head *audit_watch_rules(struct audit_watch *watch); - -extern struct audit_entry *audit_dupe_rule(struct audit_krule *old, - struct audit_watch *watch); +extern int audit_watch_compare(struct audit_watch *watch, unsigned long ino, dev_t dev); +extern struct audit_entry *audit_dupe_rule(struct audit_krule *old); #ifdef CONFIG_AUDIT_TREE extern struct audit_chunk *audit_tree_lookup(const struct inode *); diff --git a/kernel/audit_watch.c b/kernel/audit_watch.c index 8df43696f4ba..c2ca7168bfd1 100644 --- a/kernel/audit_watch.c +++ b/kernel/audit_watch.c @@ -51,12 +51,12 @@ struct audit_watch { unsigned long ino; /* associated inode number */ struct audit_parent *parent; /* associated parent */ struct list_head wlist; /* entry in parent->watches list */ - struct list_head rules; /* associated rules */ + struct list_head rules; /* anchor for krule->rlist */ }; struct audit_parent { - struct list_head ilist; /* entry in inotify registration list */ - struct list_head watches; /* associated watches */ + struct list_head ilist; /* tmp list used to free parents */ + struct list_head watches; /* anchor for audit_watch->wlist */ struct inotify_watch wdata; /* inotify watch data */ unsigned flags; /* status flags */ }; @@ -78,13 +78,18 @@ struct inotify_handle *audit_ih; /* Inotify events we care about. */ #define AUDIT_IN_WATCH IN_MOVE|IN_CREATE|IN_DELETE|IN_DELETE_SELF|IN_MOVE_SELF -static void audit_free_parent(struct inotify_watch *i_watch) +static void audit_free_parent(struct audit_parent *parent) +{ + WARN_ON(!list_empty(&parent->watches)); + kfree(parent); +} + +static void audit_destroy_watch(struct inotify_watch *i_watch) { struct audit_parent *parent; parent = container_of(i_watch, struct audit_parent, wdata); - WARN_ON(!list_empty(&parent->watches)); - kfree(parent); + audit_free_parent(parent); } void audit_get_watch(struct audit_watch *watch) @@ -115,19 +120,11 @@ char *audit_watch_path(struct audit_watch *watch) return watch->path; } -struct list_head *audit_watch_rules(struct audit_watch *watch) +int audit_watch_compare(struct audit_watch *watch, unsigned long ino, dev_t dev) { - return &watch->rules; -} - -unsigned long audit_watch_inode(struct audit_watch *watch) -{ - return watch->ino; -} - -dev_t audit_watch_dev(struct audit_watch *watch) -{ - return watch->dev; + return (watch->ino != (unsigned long)-1) && + (watch->ino == ino) && + (watch->dev == dev); } /* Initialize a parent watch entry. */ @@ -149,7 +146,7 @@ static struct audit_parent *audit_init_parent(struct nameidata *ndp) wd = inotify_add_watch(audit_ih, &parent->wdata, ndp->path.dentry->d_inode, AUDIT_IN_WATCH); if (wd < 0) { - audit_free_parent(&parent->wdata); + audit_free_parent(parent); return ERR_PTR(wd); } @@ -251,15 +248,19 @@ static void audit_update_watch(struct audit_parent *parent, struct audit_entry *oentry, *nentry; mutex_lock(&audit_filter_mutex); + /* Run all of the watches on this parent looking for the one that + * matches the given dname */ list_for_each_entry_safe(owatch, nextw, &parent->watches, wlist) { if (audit_compare_dname_path(dname, owatch->path, NULL)) continue; /* If the update involves invalidating rules, do the inode-based * filtering now, so we don't omit records. */ - if (invalidating && current->audit_context) + if (invalidating && !audit_dummy_context()) audit_filter_inodes(current, current->audit_context); + /* updating ino will likely change which audit_hash_list we + * are on so we need a new watch for the new list */ nwatch = audit_dupe_watch(owatch); if (IS_ERR(nwatch)) { mutex_unlock(&audit_filter_mutex); @@ -275,12 +276,21 @@ static void audit_update_watch(struct audit_parent *parent, list_del(&oentry->rule.rlist); list_del_rcu(&oentry->list); - nentry = audit_dupe_rule(&oentry->rule, nwatch); + nentry = audit_dupe_rule(&oentry->rule); if (IS_ERR(nentry)) { list_del(&oentry->rule.list); audit_panic("error updating watch, removing"); } else { int h = audit_hash_ino((u32)ino); + + /* + * nentry->rule.watch == oentry->rule.watch so + * we must drop that reference and set it to our + * new watch. + */ + audit_put_watch(nentry->rule.watch); + audit_get_watch(nwatch); + nentry->rule.watch = nwatch; list_add(&nentry->rule.rlist, &nwatch->rules); list_add_rcu(&nentry->list, &audit_inode_hash[h]); list_replace(&oentry->rule.list, @@ -329,14 +339,14 @@ static void audit_remove_parent_watches(struct audit_parent *parent) /* Unregister inotify watches for parents on in_list. * Generates an IN_IGNORED event. */ -void audit_inotify_unregister(struct list_head *in_list) +void audit_watch_inotify_unregister(struct list_head *in_list) { struct audit_parent *p, *n; list_for_each_entry_safe(p, n, in_list, ilist) { list_del(&p->ilist); inotify_rm_watch(audit_ih, &p->wdata); - /* the unpin matching the pin in audit_do_del_rule() */ + /* the unpin matching the pin in audit_remove_watch_rule() */ unpin_inotify_watch(&p->wdata); } } @@ -423,13 +433,13 @@ static void audit_add_to_parent(struct audit_krule *krule, /* Find a matching watch entry, or add this one. * Caller must hold audit_filter_mutex. */ -int audit_add_watch(struct audit_krule *krule) +int audit_add_watch(struct audit_krule *krule, struct list_head **list) { struct audit_watch *watch = krule->watch; struct inotify_watch *i_watch; struct audit_parent *parent; struct nameidata *ndp = NULL, *ndw = NULL; - int ret = 0; + int h, ret = 0; mutex_unlock(&audit_filter_mutex); @@ -475,6 +485,8 @@ int audit_add_watch(struct audit_krule *krule) /* match get in audit_init_parent or inotify_find_watch */ put_inotify_watch(&parent->wdata); + h = audit_hash_ino((u32)watch->ino); + *list = &audit_inode_hash[h]; error: audit_put_nd(ndp, ndw); /* NULL args OK */ return ret; @@ -514,8 +526,7 @@ static void audit_handle_ievent(struct inotify_watch *i_watch, u32 wd, u32 mask, parent = container_of(i_watch, struct audit_parent, wdata); if (mask & (IN_CREATE|IN_MOVED_TO) && inode) - audit_update_watch(parent, dname, inode->i_sb->s_dev, - inode->i_ino, 0); + audit_update_watch(parent, dname, inode->i_sb->s_dev, inode->i_ino, 0); else if (mask & (IN_DELETE|IN_MOVED_FROM)) audit_update_watch(parent, dname, (dev_t)-1, (unsigned long)-1, 1); /* inotify automatically removes the watch and sends IN_IGNORED */ @@ -531,7 +542,7 @@ static void audit_handle_ievent(struct inotify_watch *i_watch, u32 wd, u32 mask, static const struct inotify_operations audit_inotify_ops = { .handle_event = audit_handle_ievent, - .destroy_watch = audit_free_parent, + .destroy_watch = audit_destroy_watch, }; static int __init audit_watch_init(void) diff --git a/kernel/auditfilter.c b/kernel/auditfilter.c index ce08041f578d..ac87577f36b5 100644 --- a/kernel/auditfilter.c +++ b/kernel/auditfilter.c @@ -71,6 +71,7 @@ static inline void audit_free_rule(struct audit_entry *e) { int i; struct audit_krule *erule = &e->rule; + /* some rules don't have associated watches */ if (erule->watch) audit_put_watch(erule->watch); @@ -746,8 +747,7 @@ static inline int audit_dupe_lsm_field(struct audit_field *df, * rule with the new rule in the filterlist, then free the old rule. * The rlist element is undefined; list manipulations are handled apart from * the initial copy. */ -struct audit_entry *audit_dupe_rule(struct audit_krule *old, - struct audit_watch *watch) +struct audit_entry *audit_dupe_rule(struct audit_krule *old) { u32 fcount = old->field_count; struct audit_entry *entry; @@ -769,8 +769,8 @@ struct audit_entry *audit_dupe_rule(struct audit_krule *old, new->prio = old->prio; new->buflen = old->buflen; new->inode_f = old->inode_f; - new->watch = NULL; new->field_count = old->field_count; + /* * note that we are OK with not refcounting here; audit_match_tree() * never dereferences tree and we can't get false positives there @@ -811,9 +811,9 @@ struct audit_entry *audit_dupe_rule(struct audit_krule *old, } } - if (watch) { - audit_get_watch(watch); - new->watch = watch; + if (old->watch) { + audit_get_watch(old->watch); + new->watch = old->watch; } return entry; @@ -866,7 +866,7 @@ static inline int audit_add_rule(struct audit_entry *entry) struct audit_watch *watch = entry->rule.watch; struct audit_tree *tree = entry->rule.tree; struct list_head *list; - int h, err; + int err; #ifdef CONFIG_AUDITSYSCALL int dont_count = 0; @@ -889,15 +889,11 @@ static inline int audit_add_rule(struct audit_entry *entry) if (watch) { /* audit_filter_mutex is dropped and re-taken during this call */ - err = audit_add_watch(&entry->rule); + err = audit_add_watch(&entry->rule, &list); if (err) { mutex_unlock(&audit_filter_mutex); goto error; } - /* entry->rule.watch may have changed during audit_add_watch() */ - watch = entry->rule.watch; - h = audit_hash_ino((u32)audit_watch_inode(watch)); - list = &audit_inode_hash[h]; } if (tree) { err = audit_add_tree_rule(&entry->rule); @@ -949,7 +945,7 @@ static inline int audit_del_rule(struct audit_entry *entry) struct audit_watch *watch = entry->rule.watch; struct audit_tree *tree = entry->rule.tree; struct list_head *list; - LIST_HEAD(inotify_list); + LIST_HEAD(inotify_unregister_list); int ret = 0; #ifdef CONFIG_AUDITSYSCALL int dont_count = 0; @@ -969,7 +965,7 @@ static inline int audit_del_rule(struct audit_entry *entry) } if (e->rule.watch) - audit_remove_watch_rule(&e->rule, &inotify_list); + audit_remove_watch_rule(&e->rule, &inotify_unregister_list); if (e->rule.tree) audit_remove_tree_rule(&e->rule); @@ -987,8 +983,8 @@ static inline int audit_del_rule(struct audit_entry *entry) #endif mutex_unlock(&audit_filter_mutex); - if (!list_empty(&inotify_list)) - audit_inotify_unregister(&inotify_list); + if (!list_empty(&inotify_unregister_list)) + audit_watch_inotify_unregister(&inotify_unregister_list); out: if (watch) @@ -1323,30 +1319,23 @@ static int update_lsm_rule(struct audit_krule *r) { struct audit_entry *entry = container_of(r, struct audit_entry, rule); struct audit_entry *nentry; - struct audit_watch *watch; - struct audit_tree *tree; int err = 0; if (!security_audit_rule_known(r)) return 0; - watch = r->watch; - tree = r->tree; - nentry = audit_dupe_rule(r, watch); + nentry = audit_dupe_rule(r); if (IS_ERR(nentry)) { /* save the first error encountered for the * return value */ err = PTR_ERR(nentry); audit_panic("error updating LSM filters"); - if (watch) + if (r->watch) list_del(&r->rlist); list_del_rcu(&entry->list); list_del(&r->list); } else { - if (watch) { - list_add(&nentry->rule.rlist, audit_watch_rules(watch)); - list_del(&r->rlist); - } else if (tree) + if (r->watch || r->tree) list_replace_init(&r->rlist, &nentry->rule.rlist); list_replace_rcu(&entry->list, &nentry->list); list_replace(&r->list, &nentry->rule.list); diff --git a/kernel/auditsc.c b/kernel/auditsc.c index 3828ad5fb8f1..240063c370e6 100644 --- a/kernel/auditsc.c +++ b/kernel/auditsc.c @@ -549,9 +549,8 @@ static int audit_filter_rules(struct task_struct *tsk, } break; case AUDIT_WATCH: - if (name && audit_watch_inode(rule->watch) != (unsigned long)-1) - result = (name->dev == audit_watch_dev(rule->watch) && - name->ino == audit_watch_inode(rule->watch)); + if (name) + result = audit_watch_compare(rule->watch, name->ino, name->dev); break; case AUDIT_DIR: if (ctx) -- cgit v1.2.3 From e9fd702a58c49dbb14481dca88dad44758da393a Mon Sep 17 00:00:00 2001 From: Eric Paris Date: Thu, 17 Dec 2009 20:12:04 -0500 Subject: audit: convert audit watches to use fsnotify instead of inotify Audit currently uses inotify to pin inodes in core and to detect when watched inodes are deleted or unmounted. This patch uses fsnotify instead of inotify. Signed-off-by: Eric Paris --- kernel/audit_watch.c | 208 ++++++++++++++++++++++++++++++++++++--------------- 1 file changed, 148 insertions(+), 60 deletions(-) (limited to 'kernel') diff --git a/kernel/audit_watch.c b/kernel/audit_watch.c index c2ca7168bfd1..ff5be849473d 100644 --- a/kernel/audit_watch.c +++ b/kernel/audit_watch.c @@ -24,18 +24,18 @@ #include #include #include +#include #include #include #include #include -#include #include #include "audit.h" /* * Reference counting: * - * audit_parent: lifetime is from audit_init_parent() to receipt of an IN_IGNORED + * audit_parent: lifetime is from audit_init_parent() to receipt of an FS_IGNORED * event. Each audit_watch holds a reference to its associated parent. * * audit_watch: if added to lists, lifetime is from audit_init_watch() to @@ -57,26 +57,27 @@ struct audit_watch { struct audit_parent { struct list_head ilist; /* tmp list used to free parents */ struct list_head watches; /* anchor for audit_watch->wlist */ - struct inotify_watch wdata; /* inotify watch data */ + struct fsnotify_mark_entry mark; /* fsnotify mark on the inode */ unsigned flags; /* status flags */ }; -/* Inotify handle. */ -struct inotify_handle *audit_ih; +/* fsnotify handle. */ +struct fsnotify_group *audit_watch_group; /* * audit_parent status flags: * * AUDIT_PARENT_INVALID - set anytime rules/watches are auto-removed due to * a filesystem event to ensure we're adding audit watches to a valid parent. - * Technically not needed for IN_DELETE_SELF or IN_UNMOUNT events, as we cannot - * receive them while we have nameidata, but must be used for IN_MOVE_SELF which + * Technically not needed for FS_DELETE_SELF or FS_UNMOUNT events, as we cannot + * receive them while we have nameidata, but must be used for FS_MOVE_SELF which * we can receive while holding nameidata. */ #define AUDIT_PARENT_INVALID 0x001 -/* Inotify events we care about. */ -#define AUDIT_IN_WATCH IN_MOVE|IN_CREATE|IN_DELETE|IN_DELETE_SELF|IN_MOVE_SELF +/* fsnotify events we care about. */ +#define AUDIT_FS_WATCH (FS_MOVE | FS_CREATE | FS_DELETE | FS_DELETE_SELF |\ + FS_MOVE_SELF | FS_EVENT_ON_CHILD) static void audit_free_parent(struct audit_parent *parent) { @@ -84,14 +85,45 @@ static void audit_free_parent(struct audit_parent *parent) kfree(parent); } -static void audit_destroy_watch(struct inotify_watch *i_watch) +static void audit_watch_free_mark(struct fsnotify_mark_entry *entry) { struct audit_parent *parent; - parent = container_of(i_watch, struct audit_parent, wdata); + parent = container_of(entry, struct audit_parent, mark); audit_free_parent(parent); } +static void audit_get_parent(struct audit_parent *parent) +{ + if (likely(parent)) + fsnotify_get_mark(&parent->mark); +} + +static void audit_put_parent(struct audit_parent *parent) +{ + if (likely(parent)) + fsnotify_put_mark(&parent->mark); +} + +/* + * Find and return the audit_parent on the given inode. If found a reference + * is taken on this parent. + */ +static inline struct audit_parent *audit_find_parent(struct inode *inode) +{ + struct audit_parent *parent = NULL; + struct fsnotify_mark_entry *entry; + + spin_lock(&inode->i_lock); + entry = fsnotify_find_mark_entry(audit_watch_group, inode); + spin_unlock(&inode->i_lock); + + if (entry) + parent = container_of(entry, struct audit_parent, mark); + + return parent; +} + void audit_get_watch(struct audit_watch *watch) { atomic_inc(&watch->count); @@ -110,7 +142,7 @@ void audit_put_watch(struct audit_watch *watch) void audit_remove_watch(struct audit_watch *watch) { list_del(&watch->wlist); - put_inotify_watch(&watch->parent->wdata); + audit_put_parent(watch->parent); watch->parent = NULL; audit_put_watch(watch); /* match initial get */ } @@ -130,8 +162,9 @@ int audit_watch_compare(struct audit_watch *watch, unsigned long ino, dev_t dev) /* Initialize a parent watch entry. */ static struct audit_parent *audit_init_parent(struct nameidata *ndp) { + struct inode *inode = ndp->path.dentry->d_inode; struct audit_parent *parent; - s32 wd; + int ret; parent = kzalloc(sizeof(*parent), GFP_KERNEL); if (unlikely(!parent)) @@ -140,14 +173,14 @@ static struct audit_parent *audit_init_parent(struct nameidata *ndp) INIT_LIST_HEAD(&parent->watches); parent->flags = 0; - inotify_init_watch(&parent->wdata); - /* grab a ref so inotify watch hangs around until we take audit_filter_mutex */ - get_inotify_watch(&parent->wdata); - wd = inotify_add_watch(audit_ih, &parent->wdata, - ndp->path.dentry->d_inode, AUDIT_IN_WATCH); - if (wd < 0) { + fsnotify_init_mark(&parent->mark, audit_watch_free_mark); + parent->mark.mask = AUDIT_FS_WATCH; + /* grab a ref so fsnotify mark hangs around until we take audit_filter_mutex */ + audit_get_parent(parent); + ret = fsnotify_add_mark(&parent->mark, audit_watch_group, inode); + if (ret < 0) { audit_free_parent(parent); - return ERR_PTR(wd); + return ERR_PTR(ret); } return parent; @@ -176,7 +209,7 @@ int audit_to_watch(struct audit_krule *krule, char *path, int len, u32 op) { struct audit_watch *watch; - if (!audit_ih) + if (!audit_watch_group) return -EOPNOTSUPP; if (path[0] != '/' || path[len-1] == '/' || @@ -214,7 +247,7 @@ static struct audit_watch *audit_dupe_watch(struct audit_watch *old) new->dev = old->dev; new->ino = old->ino; - get_inotify_watch(&old->parent->wdata); + audit_get_parent(old->parent); new->parent = old->parent; out: @@ -335,19 +368,21 @@ static void audit_remove_parent_watches(struct audit_parent *parent) audit_remove_watch(w); } mutex_unlock(&audit_filter_mutex); + + fsnotify_destroy_mark_by_entry(&parent->mark); } /* Unregister inotify watches for parents on in_list. - * Generates an IN_IGNORED event. */ + * Generates an FS_IGNORED event. */ void audit_watch_inotify_unregister(struct list_head *in_list) { struct audit_parent *p, *n; list_for_each_entry_safe(p, n, in_list, ilist) { list_del(&p->ilist); - inotify_rm_watch(audit_ih, &p->wdata); - /* the unpin matching the pin in audit_remove_watch_rule() */ - unpin_inotify_watch(&p->wdata); + fsnotify_destroy_mark_by_entry(&p->mark); + /* matches the get in audit_remove_watch_rule() */ + audit_put_parent(p); } } @@ -399,7 +434,7 @@ static void audit_put_nd(struct nameidata *ndp, struct nameidata *ndw) } } -/* Associate the given rule with an existing parent inotify_watch. +/* Associate the given rule with an existing parent. * Caller must hold audit_filter_mutex. */ static void audit_add_to_parent(struct audit_krule *krule, struct audit_parent *parent) @@ -407,6 +442,8 @@ static void audit_add_to_parent(struct audit_krule *krule, struct audit_watch *w, *watch = krule->watch; int watch_found = 0; + BUG_ON(!mutex_is_locked(&audit_filter_mutex)); + list_for_each_entry(w, &parent->watches, wlist) { if (strcmp(watch->path, w->path)) continue; @@ -423,7 +460,7 @@ static void audit_add_to_parent(struct audit_krule *krule, } if (!watch_found) { - get_inotify_watch(&parent->wdata); + audit_get_parent(parent); watch->parent = parent; list_add(&watch->wlist, &parent->watches); @@ -436,7 +473,6 @@ static void audit_add_to_parent(struct audit_krule *krule, int audit_add_watch(struct audit_krule *krule, struct list_head **list) { struct audit_watch *watch = krule->watch; - struct inotify_watch *i_watch; struct audit_parent *parent; struct nameidata *ndp = NULL, *ndw = NULL; int h, ret = 0; @@ -462,8 +498,8 @@ int audit_add_watch(struct audit_krule *krule, struct list_head **list) * inotify watch is found, inotify_find_watch() grabs a reference before * returning. */ - if (inotify_find_watch(audit_ih, ndp->path.dentry->d_inode, - &i_watch) < 0) { + parent = audit_find_parent(ndp->path.dentry->d_inode); + if (!parent) { parent = audit_init_parent(ndp); if (IS_ERR(parent)) { /* caller expects mutex locked */ @@ -471,8 +507,7 @@ int audit_add_watch(struct audit_krule *krule, struct list_head **list) ret = PTR_ERR(parent); goto error; } - } else - parent = container_of(i_watch, struct audit_parent, wdata); + } mutex_lock(&audit_filter_mutex); @@ -482,8 +517,8 @@ int audit_add_watch(struct audit_krule *krule, struct list_head **list) else audit_add_to_parent(krule, parent); - /* match get in audit_init_parent or inotify_find_watch */ - put_inotify_watch(&parent->wdata); + /* match get in audit_find_parent or audit_init_parent */ + audit_put_parent(parent); h = audit_hash_ino((u32)watch->ino); *list = &audit_inode_hash[h]; @@ -504,52 +539,105 @@ void audit_remove_watch_rule(struct audit_krule *krule, struct list_head *list) audit_remove_watch(watch); if (list_empty(&parent->watches)) { - /* Put parent on the inotify un-registration - * list. Grab a reference before releasing + /* Put parent on the un-registration list. + * Grab a reference before releasing * audit_filter_mutex, to be released in - * audit_inotify_unregister(). + * audit_watch_inotify_unregister(). * If filesystem is going away, just leave * the sucker alone, eviction will take * care of it. */ - if (pin_inotify_watch(&parent->wdata)) - list_add(&parent->ilist, list); + audit_get_parent(parent); + list_add(&parent->ilist, list); } } } -/* Update watch data in audit rules based on inotify events. */ -static void audit_handle_ievent(struct inotify_watch *i_watch, u32 wd, u32 mask, - u32 cookie, const char *dname, struct inode *inode) +static bool audit_watch_should_send_event(struct fsnotify_group *group, struct inode *inode, __u32 mask) { + struct fsnotify_mark_entry *entry; + bool send; + + spin_lock(&inode->i_lock); + entry = fsnotify_find_mark_entry(group, inode); + spin_unlock(&inode->i_lock); + if (!entry) + return false; + + mask = (mask & ~FS_EVENT_ON_CHILD); + send = (entry->mask & mask); + + /* find took a reference */ + fsnotify_put_mark(entry); + + return send; +} + +/* Update watch data in audit rules based on fsnotify events. */ +static int audit_watch_handle_event(struct fsnotify_group *group, struct fsnotify_event *event) +{ + struct inode *inode; + __u32 mask = event->mask; + const char *dname = event->file_name; struct audit_parent *parent; - parent = container_of(i_watch, struct audit_parent, wdata); + BUG_ON(group != audit_watch_group); + + parent = audit_find_parent(event->to_tell); + if (unlikely(!parent)) + return 0; + + switch (event->data_type) { + case (FSNOTIFY_EVENT_PATH): + inode = event->path.dentry->d_inode; + break; + case (FSNOTIFY_EVENT_INODE): + inode = event->inode; + break; + default: + BUG(); + inode = NULL; + break; + }; - if (mask & (IN_CREATE|IN_MOVED_TO) && inode) + if (mask & (FS_CREATE|FS_MOVED_TO) && inode) audit_update_watch(parent, dname, inode->i_sb->s_dev, inode->i_ino, 0); - else if (mask & (IN_DELETE|IN_MOVED_FROM)) + else if (mask & (FS_DELETE|FS_MOVED_FROM)) audit_update_watch(parent, dname, (dev_t)-1, (unsigned long)-1, 1); - /* inotify automatically removes the watch and sends IN_IGNORED */ - else if (mask & (IN_DELETE_SELF|IN_UNMOUNT)) + else if (mask & (FS_DELETE_SELF|FS_UNMOUNT|FS_MOVE_SELF)) audit_remove_parent_watches(parent); - /* inotify does not remove the watch, so remove it manually */ - else if(mask & IN_MOVE_SELF) { - audit_remove_parent_watches(parent); - inotify_remove_watch_locked(audit_ih, i_watch); - } else if (mask & IN_IGNORED) - put_inotify_watch(i_watch); + /* moved put_inotify_watch to freeing mark */ + + /* matched the ref taken by audit_find_parent */ + audit_put_parent(parent); + + return 0; +} + +static void audit_watch_freeing_mark(struct fsnotify_mark_entry *entry, struct fsnotify_group *group) +{ + struct audit_parent *parent; + + parent = container_of(entry, struct audit_parent, mark); + /* taken from audit_handle_ievent & FS_IGNORED please figure out what I match... */ + audit_put_parent(parent); } -static const struct inotify_operations audit_inotify_ops = { - .handle_event = audit_handle_ievent, - .destroy_watch = audit_destroy_watch, +static const struct fsnotify_ops audit_watch_fsnotify_ops = { + .should_send_event = audit_watch_should_send_event, + .handle_event = audit_watch_handle_event, + .free_group_priv = NULL, + .freeing_mark = audit_watch_freeing_mark, + .free_event_priv = NULL, }; static int __init audit_watch_init(void) { - audit_ih = inotify_init(&audit_inotify_ops); - if (IS_ERR(audit_ih)) - audit_panic("cannot initialize inotify handle"); + audit_watch_group = fsnotify_obtain_group(AUDIT_WATCH_GROUP_NUM, AUDIT_FS_WATCH, + &audit_watch_fsnotify_ops); + if (IS_ERR(audit_watch_group)) { + audit_watch_group = NULL; + audit_panic("cannot create audit fsnotify group"); + } return 0; } subsys_initcall(audit_watch_init); -- cgit v1.2.3 From e118e9c5638bbe877aa26b5cd2fd223cc24cdc8a Mon Sep 17 00:00:00 2001 From: Eric Paris Date: Thu, 17 Dec 2009 20:12:04 -0500 Subject: audit: redo audit watch locking and refcnt in light of fsnotify fsnotify can handle mutexes to be held across all fsnotify operations since it deals strickly in spinlocks. This can simplify and reduce some of the audit_filter_mutex taking and dropping. Signed-off-by: Eric Paris --- kernel/audit_watch.c | 45 +++++---------------------------------------- 1 file changed, 5 insertions(+), 40 deletions(-) (limited to 'kernel') diff --git a/kernel/audit_watch.c b/kernel/audit_watch.c index ff5be849473d..da66197e3abc 100644 --- a/kernel/audit_watch.c +++ b/kernel/audit_watch.c @@ -58,23 +58,11 @@ struct audit_parent { struct list_head ilist; /* tmp list used to free parents */ struct list_head watches; /* anchor for audit_watch->wlist */ struct fsnotify_mark_entry mark; /* fsnotify mark on the inode */ - unsigned flags; /* status flags */ }; /* fsnotify handle. */ struct fsnotify_group *audit_watch_group; -/* - * audit_parent status flags: - * - * AUDIT_PARENT_INVALID - set anytime rules/watches are auto-removed due to - * a filesystem event to ensure we're adding audit watches to a valid parent. - * Technically not needed for FS_DELETE_SELF or FS_UNMOUNT events, as we cannot - * receive them while we have nameidata, but must be used for FS_MOVE_SELF which - * we can receive while holding nameidata. - */ -#define AUDIT_PARENT_INVALID 0x001 - /* fsnotify events we care about. */ #define AUDIT_FS_WATCH (FS_MOVE | FS_CREATE | FS_DELETE | FS_DELETE_SELF |\ FS_MOVE_SELF | FS_EVENT_ON_CHILD) @@ -171,12 +159,9 @@ static struct audit_parent *audit_init_parent(struct nameidata *ndp) return ERR_PTR(-ENOMEM); INIT_LIST_HEAD(&parent->watches); - parent->flags = 0; fsnotify_init_mark(&parent->mark, audit_watch_free_mark); parent->mark.mask = AUDIT_FS_WATCH; - /* grab a ref so fsnotify mark hangs around until we take audit_filter_mutex */ - audit_get_parent(parent); ret = fsnotify_add_mark(&parent->mark, audit_watch_group, inode); if (ret < 0) { audit_free_parent(parent); @@ -355,7 +340,6 @@ static void audit_remove_parent_watches(struct audit_parent *parent) struct audit_entry *e; mutex_lock(&audit_filter_mutex); - parent->flags |= AUDIT_PARENT_INVALID; list_for_each_entry_safe(w, nextw, &parent->watches, wlist) { list_for_each_entry_safe(r, nextr, &w->rules, rlist) { e = container_of(r, struct audit_entry, rule); @@ -487,35 +471,25 @@ int audit_add_watch(struct audit_krule *krule, struct list_head **list) goto error; } + mutex_lock(&audit_filter_mutex); + /* update watch filter fields */ if (ndw) { watch->dev = ndw->path.dentry->d_inode->i_sb->s_dev; watch->ino = ndw->path.dentry->d_inode->i_ino; } - /* The audit_filter_mutex must not be held during inotify calls because - * we hold it during inotify event callback processing. If an existing - * inotify watch is found, inotify_find_watch() grabs a reference before - * returning. - */ + /* either find an old parent or attach a new one */ parent = audit_find_parent(ndp->path.dentry->d_inode); if (!parent) { parent = audit_init_parent(ndp); if (IS_ERR(parent)) { - /* caller expects mutex locked */ - mutex_lock(&audit_filter_mutex); ret = PTR_ERR(parent); goto error; } } - mutex_lock(&audit_filter_mutex); - - /* parent was moved before we took audit_filter_mutex */ - if (parent->flags & AUDIT_PARENT_INVALID) - ret = -ENOENT; - else - audit_add_to_parent(krule, parent); + audit_add_to_parent(krule, parent); /* match get in audit_find_parent or audit_init_parent */ audit_put_parent(parent); @@ -613,20 +587,11 @@ static int audit_watch_handle_event(struct fsnotify_group *group, struct fsnotif return 0; } -static void audit_watch_freeing_mark(struct fsnotify_mark_entry *entry, struct fsnotify_group *group) -{ - struct audit_parent *parent; - - parent = container_of(entry, struct audit_parent, mark); - /* taken from audit_handle_ievent & FS_IGNORED please figure out what I match... */ - audit_put_parent(parent); -} - static const struct fsnotify_ops audit_watch_fsnotify_ops = { .should_send_event = audit_watch_should_send_event, .handle_event = audit_watch_handle_event, .free_group_priv = NULL, - .freeing_mark = audit_watch_freeing_mark, + .freeing_mark = NULL, .free_event_priv = NULL, }; -- cgit v1.2.3 From a05fb6cc573130915380e00d182a4c6571cec6b2 Mon Sep 17 00:00:00 2001 From: Eric Paris Date: Thu, 17 Dec 2009 20:12:05 -0500 Subject: audit: do not get and put just to free a watch deleting audit watch rules is not currently done under audit_filter_mutex. It was done this way because we could not hold the mutex during inotify manipulation. Since we are using fsnotify we don't need to do the extra get/put pair nor do we need the private list on which to store the parents while they are about to be freed. Signed-off-by: Eric Paris --- kernel/audit.h | 3 +-- kernel/audit_watch.c | 27 +++------------------------ kernel/auditfilter.c | 6 +----- 3 files changed, 5 insertions(+), 31 deletions(-) (limited to 'kernel') diff --git a/kernel/audit.h b/kernel/audit.h index 82c8a09099f4..100b454a7354 100644 --- a/kernel/audit.h +++ b/kernel/audit.h @@ -108,8 +108,7 @@ extern void audit_put_watch(struct audit_watch *watch); extern void audit_get_watch(struct audit_watch *watch); extern int audit_to_watch(struct audit_krule *krule, char *path, int len, u32 op); extern int audit_add_watch(struct audit_krule *krule, struct list_head **list); -extern void audit_remove_watch_rule(struct audit_krule *krule, struct list_head *list); -extern void audit_watch_inotify_unregister(struct list_head *in_list); +extern void audit_remove_watch_rule(struct audit_krule *krule); extern char *audit_watch_path(struct audit_watch *watch); extern int audit_watch_compare(struct audit_watch *watch, unsigned long ino, dev_t dev); extern struct audit_entry *audit_dupe_rule(struct audit_krule *old); diff --git a/kernel/audit_watch.c b/kernel/audit_watch.c index da66197e3abc..75ab53987ece 100644 --- a/kernel/audit_watch.c +++ b/kernel/audit_watch.c @@ -55,7 +55,6 @@ struct audit_watch { }; struct audit_parent { - struct list_head ilist; /* tmp list used to free parents */ struct list_head watches; /* anchor for audit_watch->wlist */ struct fsnotify_mark_entry mark; /* fsnotify mark on the inode */ }; @@ -356,20 +355,6 @@ static void audit_remove_parent_watches(struct audit_parent *parent) fsnotify_destroy_mark_by_entry(&parent->mark); } -/* Unregister inotify watches for parents on in_list. - * Generates an FS_IGNORED event. */ -void audit_watch_inotify_unregister(struct list_head *in_list) -{ - struct audit_parent *p, *n; - - list_for_each_entry_safe(p, n, in_list, ilist) { - list_del(&p->ilist); - fsnotify_destroy_mark_by_entry(&p->mark); - /* matches the get in audit_remove_watch_rule() */ - audit_put_parent(p); - } -} - /* Get path information necessary for adding watches. */ static int audit_get_nd(char *path, struct nameidata **ndp, struct nameidata **ndw) { @@ -502,7 +487,7 @@ error: } -void audit_remove_watch_rule(struct audit_krule *krule, struct list_head *list) +void audit_remove_watch_rule(struct audit_krule *krule) { struct audit_watch *watch = krule->watch; struct audit_parent *parent = watch->parent; @@ -513,15 +498,9 @@ void audit_remove_watch_rule(struct audit_krule *krule, struct list_head *list) audit_remove_watch(watch); if (list_empty(&parent->watches)) { - /* Put parent on the un-registration list. - * Grab a reference before releasing - * audit_filter_mutex, to be released in - * audit_watch_inotify_unregister(). - * If filesystem is going away, just leave - * the sucker alone, eviction will take - * care of it. */ audit_get_parent(parent); - list_add(&parent->ilist, list); + fsnotify_destroy_mark_by_entry(&parent->mark); + audit_put_parent(parent); } } } diff --git a/kernel/auditfilter.c b/kernel/auditfilter.c index ac87577f36b5..eb7675499fb5 100644 --- a/kernel/auditfilter.c +++ b/kernel/auditfilter.c @@ -945,7 +945,6 @@ static inline int audit_del_rule(struct audit_entry *entry) struct audit_watch *watch = entry->rule.watch; struct audit_tree *tree = entry->rule.tree; struct list_head *list; - LIST_HEAD(inotify_unregister_list); int ret = 0; #ifdef CONFIG_AUDITSYSCALL int dont_count = 0; @@ -965,7 +964,7 @@ static inline int audit_del_rule(struct audit_entry *entry) } if (e->rule.watch) - audit_remove_watch_rule(&e->rule, &inotify_unregister_list); + audit_remove_watch_rule(&e->rule); if (e->rule.tree) audit_remove_tree_rule(&e->rule); @@ -983,9 +982,6 @@ static inline int audit_del_rule(struct audit_entry *entry) #endif mutex_unlock(&audit_filter_mutex); - if (!list_empty(&inotify_unregister_list)) - audit_watch_inotify_unregister(&inotify_unregister_list); - out: if (watch) audit_put_watch(watch); /* match initial get */ -- cgit v1.2.3 From 40554c3dae83bd892b7fbfaa2ea9de739cbcf065 Mon Sep 17 00:00:00 2001 From: Eric Paris Date: Thu, 17 Dec 2009 20:12:05 -0500 Subject: fsnotify: allow addition of duplicate fsnotify marks This patch allows a task to add a second fsnotify mark to an inode for the same group. This mark will be added to the end of the inode's list and this will never be found by the stand fsnotify_find_mark() function. This is useful if a user wants to add a new mark before removing the old one. Signed-off-by: Eric Paris --- kernel/audit_watch.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/audit_watch.c b/kernel/audit_watch.c index 75ab53987ece..c44de0c4fc47 100644 --- a/kernel/audit_watch.c +++ b/kernel/audit_watch.c @@ -161,7 +161,7 @@ static struct audit_parent *audit_init_parent(struct nameidata *ndp) fsnotify_init_mark(&parent->mark, audit_watch_free_mark); parent->mark.mask = AUDIT_FS_WATCH; - ret = fsnotify_add_mark(&parent->mark, audit_watch_group, inode); + ret = fsnotify_add_mark(&parent->mark, audit_watch_group, inode, 0); if (ret < 0) { audit_free_parent(parent); return ERR_PTR(ret); -- cgit v1.2.3 From 28a3a7eb3b1f3e7d834e19f06e794e429058a4dd Mon Sep 17 00:00:00 2001 From: Eric Paris Date: Thu, 17 Dec 2009 20:12:05 -0500 Subject: audit: reimplement audit_trees using fsnotify rather than inotify Simply switch audit_trees from using inotify to using fsnotify for it's inode pinning and disappearing act information. Signed-off-by: Eric Paris --- kernel/audit_tree.c | 234 +++++++++++++++++++++++++++++----------------------- kernel/auditsc.c | 4 +- 2 files changed, 132 insertions(+), 106 deletions(-) (limited to 'kernel') diff --git a/kernel/audit_tree.c b/kernel/audit_tree.c index 46a57b57a335..a164600dd82e 100644 --- a/kernel/audit_tree.c +++ b/kernel/audit_tree.c @@ -1,5 +1,5 @@ #include "audit.h" -#include +#include #include #include #include @@ -22,7 +22,7 @@ struct audit_tree { struct audit_chunk { struct list_head hash; - struct inotify_watch watch; + struct fsnotify_mark_entry mark; struct list_head trees; /* with root here */ int dead; int count; @@ -59,7 +59,7 @@ static LIST_HEAD(prune_list); * tree is refcounted; one reference for "some rules on rules_list refer to * it", one for each chunk with pointer to it. * - * chunk is refcounted by embedded inotify_watch + .refs (non-zero refcount + * chunk is refcounted by embedded fsnotify_mark + .refs (non-zero refcount * of watch contributes 1 to .refs). * * node.index allows to get from node.list to containing chunk. @@ -68,7 +68,7 @@ static LIST_HEAD(prune_list); * that makes a difference. Some. */ -static struct inotify_handle *rtree_ih; +static struct fsnotify_group *audit_tree_group; static struct audit_tree *alloc_tree(const char *s) { @@ -111,29 +111,6 @@ const char *audit_tree_path(struct audit_tree *tree) return tree->pathname; } -static struct audit_chunk *alloc_chunk(int count) -{ - struct audit_chunk *chunk; - size_t size; - int i; - - size = offsetof(struct audit_chunk, owners) + count * sizeof(struct node); - chunk = kzalloc(size, GFP_KERNEL); - if (!chunk) - return NULL; - - INIT_LIST_HEAD(&chunk->hash); - INIT_LIST_HEAD(&chunk->trees); - chunk->count = count; - atomic_long_set(&chunk->refs, 1); - for (i = 0; i < count; i++) { - INIT_LIST_HEAD(&chunk->owners[i].list); - chunk->owners[i].index = i; - } - inotify_init_watch(&chunk->watch); - return chunk; -} - static void free_chunk(struct audit_chunk *chunk) { int i; @@ -157,6 +134,35 @@ static void __put_chunk(struct rcu_head *rcu) audit_put_chunk(chunk); } +static void audit_tree_destroy_watch(struct fsnotify_mark_entry *entry) +{ + struct audit_chunk *chunk = container_of(entry, struct audit_chunk, mark); + call_rcu(&chunk->head, __put_chunk); +} + +static struct audit_chunk *alloc_chunk(int count) +{ + struct audit_chunk *chunk; + size_t size; + int i; + + size = offsetof(struct audit_chunk, owners) + count * sizeof(struct node); + chunk = kzalloc(size, GFP_KERNEL); + if (!chunk) + return NULL; + + INIT_LIST_HEAD(&chunk->hash); + INIT_LIST_HEAD(&chunk->trees); + chunk->count = count; + atomic_long_set(&chunk->refs, 1); + for (i = 0; i < count; i++) { + INIT_LIST_HEAD(&chunk->owners[i].list); + chunk->owners[i].index = i; + } + fsnotify_init_mark(&chunk->mark, audit_tree_destroy_watch); + return chunk; +} + enum {HASH_SIZE = 128}; static struct list_head chunk_hash_heads[HASH_SIZE]; static __cacheline_aligned_in_smp DEFINE_SPINLOCK(hash_lock); @@ -167,10 +173,15 @@ static inline struct list_head *chunk_hash(const struct inode *inode) return chunk_hash_heads + n % HASH_SIZE; } -/* hash_lock is held by caller */ +/* hash_lock & entry->lock is held by caller */ static void insert_hash(struct audit_chunk *chunk) { - struct list_head *list = chunk_hash(chunk->watch.inode); + struct fsnotify_mark_entry *entry = &chunk->mark; + struct list_head *list; + + if (!entry->inode) + return; + list = chunk_hash(entry->inode); list_add_rcu(&chunk->hash, list); } @@ -181,7 +192,8 @@ struct audit_chunk *audit_tree_lookup(const struct inode *inode) struct audit_chunk *p; list_for_each_entry_rcu(p, list, hash) { - if (p->watch.inode == inode) { + /* mark.inode may have gone NULL, but who cares? */ + if (p->mark.inode == inode) { atomic_long_inc(&p->refs); return p; } @@ -210,38 +222,19 @@ static struct audit_chunk *find_chunk(struct node *p) static void untag_chunk(struct node *p) { struct audit_chunk *chunk = find_chunk(p); + struct fsnotify_mark_entry *entry = &chunk->mark; struct audit_chunk *new; struct audit_tree *owner; int size = chunk->count - 1; int i, j; - if (!pin_inotify_watch(&chunk->watch)) { - /* - * Filesystem is shutting down; all watches are getting - * evicted, just take it off the node list for this - * tree and let the eviction logics take care of the - * rest. - */ - owner = p->owner; - if (owner->root == chunk) { - list_del_init(&owner->same_root); - owner->root = NULL; - } - list_del_init(&p->list); - p->owner = NULL; - put_tree(owner); - return; - } + fsnotify_get_mark(entry); spin_unlock(&hash_lock); - /* - * pin_inotify_watch() succeeded, so the watch won't go away - * from under us. - */ - mutex_lock(&chunk->watch.inode->inotify_mutex); - if (chunk->dead) { - mutex_unlock(&chunk->watch.inode->inotify_mutex); + spin_lock(&entry->lock); + if (chunk->dead || !entry->inode) { + spin_unlock(&entry->lock); goto out; } @@ -256,16 +249,17 @@ static void untag_chunk(struct node *p) list_del_init(&p->list); list_del_rcu(&chunk->hash); spin_unlock(&hash_lock); - inotify_evict_watch(&chunk->watch); - mutex_unlock(&chunk->watch.inode->inotify_mutex); - put_inotify_watch(&chunk->watch); + spin_unlock(&entry->lock); + fsnotify_destroy_mark_by_entry(entry); + fsnotify_put_mark(entry); goto out; } new = alloc_chunk(size); if (!new) goto Fallback; - if (inotify_clone_watch(&chunk->watch, &new->watch) < 0) { + fsnotify_duplicate_mark(&new->mark, entry); + if (fsnotify_add_mark(&new->mark, new->mark.group, new->mark.inode, 1)) { free_chunk(new); goto Fallback; } @@ -298,9 +292,9 @@ static void untag_chunk(struct node *p) list_for_each_entry(owner, &new->trees, same_root) owner->root = new; spin_unlock(&hash_lock); - inotify_evict_watch(&chunk->watch); - mutex_unlock(&chunk->watch.inode->inotify_mutex); - put_inotify_watch(&chunk->watch); + spin_unlock(&entry->lock); + fsnotify_destroy_mark_by_entry(entry); + fsnotify_put_mark(entry); goto out; Fallback: @@ -314,31 +308,33 @@ Fallback: p->owner = NULL; put_tree(owner); spin_unlock(&hash_lock); - mutex_unlock(&chunk->watch.inode->inotify_mutex); + spin_unlock(&entry->lock); out: - unpin_inotify_watch(&chunk->watch); + fsnotify_put_mark(entry); spin_lock(&hash_lock); } static int create_chunk(struct inode *inode, struct audit_tree *tree) { + struct fsnotify_mark_entry *entry; struct audit_chunk *chunk = alloc_chunk(1); if (!chunk) return -ENOMEM; - if (inotify_add_watch(rtree_ih, &chunk->watch, inode, IN_IGNORED | IN_DELETE_SELF) < 0) { + entry = &chunk->mark; + if (fsnotify_add_mark(entry, audit_tree_group, inode, 0)) { free_chunk(chunk); return -ENOSPC; } - mutex_lock(&inode->inotify_mutex); + spin_lock(&entry->lock); spin_lock(&hash_lock); if (tree->goner) { spin_unlock(&hash_lock); chunk->dead = 1; - inotify_evict_watch(&chunk->watch); - mutex_unlock(&inode->inotify_mutex); - put_inotify_watch(&chunk->watch); + spin_unlock(&entry->lock); + fsnotify_destroy_mark_by_entry(entry); + fsnotify_put_mark(entry); return 0; } chunk->owners[0].index = (1U << 31); @@ -351,30 +347,33 @@ static int create_chunk(struct inode *inode, struct audit_tree *tree) } insert_hash(chunk); spin_unlock(&hash_lock); - mutex_unlock(&inode->inotify_mutex); + spin_unlock(&entry->lock); return 0; } /* the first tagged inode becomes root of tree */ static int tag_chunk(struct inode *inode, struct audit_tree *tree) { - struct inotify_watch *watch; + struct fsnotify_mark_entry *old_entry, *chunk_entry; struct audit_tree *owner; struct audit_chunk *chunk, *old; struct node *p; int n; - if (inotify_find_watch(rtree_ih, inode, &watch) < 0) + spin_lock(&inode->i_lock); + old_entry = fsnotify_find_mark_entry(audit_tree_group, inode); + spin_unlock(&inode->i_lock); + if (!old_entry) return create_chunk(inode, tree); - old = container_of(watch, struct audit_chunk, watch); + old = container_of(old_entry, struct audit_chunk, mark); /* are we already there? */ spin_lock(&hash_lock); for (n = 0; n < old->count; n++) { if (old->owners[n].owner == tree) { spin_unlock(&hash_lock); - put_inotify_watch(&old->watch); + fsnotify_put_mark(old_entry); return 0; } } @@ -382,25 +381,44 @@ static int tag_chunk(struct inode *inode, struct audit_tree *tree) chunk = alloc_chunk(old->count + 1); if (!chunk) { - put_inotify_watch(&old->watch); + fsnotify_put_mark(old_entry); return -ENOMEM; } - mutex_lock(&inode->inotify_mutex); - if (inotify_clone_watch(&old->watch, &chunk->watch) < 0) { - mutex_unlock(&inode->inotify_mutex); - put_inotify_watch(&old->watch); + chunk_entry = &chunk->mark; + + spin_lock(&old_entry->lock); + if (!old_entry->inode) { + /* old_entry is being shot, lets just lie */ + spin_unlock(&old_entry->lock); + fsnotify_put_mark(old_entry); free_chunk(chunk); + return -ENOENT; + } + + fsnotify_duplicate_mark(chunk_entry, old_entry); + if (fsnotify_add_mark(chunk_entry, chunk_entry->group, chunk_entry->inode, 1)) { + spin_unlock(&old_entry->lock); + free_chunk(chunk); + fsnotify_put_mark(old_entry); return -ENOSPC; } + + /* even though we hold old_entry->lock, this is safe since chunk_entry->lock could NEVER have been grabbed before */ + spin_lock(&chunk_entry->lock); spin_lock(&hash_lock); + + /* we now hold old_entry->lock, chunk_entry->lock, and hash_lock */ if (tree->goner) { spin_unlock(&hash_lock); chunk->dead = 1; - inotify_evict_watch(&chunk->watch); - mutex_unlock(&inode->inotify_mutex); - put_inotify_watch(&old->watch); - put_inotify_watch(&chunk->watch); + spin_unlock(&chunk_entry->lock); + spin_unlock(&old_entry->lock); + + fsnotify_destroy_mark_by_entry(chunk_entry); + + fsnotify_put_mark(chunk_entry); + fsnotify_put_mark(old_entry); return 0; } list_replace_init(&old->trees, &chunk->trees); @@ -426,10 +444,11 @@ static int tag_chunk(struct inode *inode, struct audit_tree *tree) list_add(&tree->same_root, &chunk->trees); } spin_unlock(&hash_lock); - inotify_evict_watch(&old->watch); - mutex_unlock(&inode->inotify_mutex); - put_inotify_watch(&old->watch); /* pair to inotify_find_watch */ - put_inotify_watch(&old->watch); /* and kill it */ + spin_unlock(&chunk_entry->lock); + spin_unlock(&old_entry->lock); + fsnotify_destroy_mark_by_entry(old_entry); + fsnotify_put_mark(old_entry); /* pair to fsnotify_find mark_entry */ + fsnotify_put_mark(old_entry); /* and kill it */ return 0; } @@ -584,7 +603,9 @@ void audit_trim_trees(void) spin_lock(&hash_lock); list_for_each_entry(node, &tree->chunks, list) { - struct inode *inode = find_chunk(node)->watch.inode; + struct audit_chunk *chunk = find_chunk(node); + /* this could be NULL if the watch is dieing else where... */ + struct inode *inode = chunk->mark.inode; node->index |= 1U<<31; if (iterate_mounts(compare_root, inode, root_mnt)) node->index &= ~(1U<<31); @@ -846,7 +867,6 @@ void audit_kill_trees(struct list_head *list) * Here comes the stuff asynchronous to auditctl operations */ -/* inode->inotify_mutex is locked */ static void evict_chunk(struct audit_chunk *chunk) { struct audit_tree *owner; @@ -885,35 +905,41 @@ static void evict_chunk(struct audit_chunk *chunk) mutex_unlock(&audit_filter_mutex); } -static void handle_event(struct inotify_watch *watch, u32 wd, u32 mask, - u32 cookie, const char *dname, struct inode *inode) +static int audit_tree_handle_event(struct fsnotify_group *group, struct fsnotify_event *event) { - struct audit_chunk *chunk = container_of(watch, struct audit_chunk, watch); + BUG(); + return -EOPNOTSUPP; +} - if (mask & IN_IGNORED) { - evict_chunk(chunk); - put_inotify_watch(watch); - } +static void audit_tree_freeing_mark(struct fsnotify_mark_entry *entry, struct fsnotify_group *group) +{ + struct audit_chunk *chunk = container_of(entry, struct audit_chunk, mark); + + evict_chunk(chunk); + fsnotify_put_mark(entry); } -static void destroy_watch(struct inotify_watch *watch) +static bool audit_tree_send_event(struct fsnotify_group *group, struct inode *inode, __u32 mask) { - struct audit_chunk *chunk = container_of(watch, struct audit_chunk, watch); - call_rcu(&chunk->head, __put_chunk); + return 0; } -static const struct inotify_operations rtree_inotify_ops = { - .handle_event = handle_event, - .destroy_watch = destroy_watch, +static const struct fsnotify_ops audit_tree_ops = { + .handle_event = audit_tree_handle_event, + .should_send_event = audit_tree_send_event, + .free_group_priv = NULL, + .free_event_priv = NULL, + .freeing_mark = audit_tree_freeing_mark, }; static int __init audit_tree_init(void) { int i; - rtree_ih = inotify_init(&rtree_inotify_ops); - if (IS_ERR(rtree_ih)) - audit_panic("cannot initialize inotify handle for rectree watches"); + audit_tree_group = fsnotify_obtain_group(AUDIT_TREE_GROUP_NUM, + 0, &audit_tree_ops); + if (IS_ERR(audit_tree_group)) + audit_panic("cannot initialize fsnotify group for rectree watches"); for (i = 0; i < HASH_SIZE; i++) INIT_LIST_HEAD(&chunk_hash_heads[i]); diff --git a/kernel/auditsc.c b/kernel/auditsc.c index 240063c370e6..786901cd8217 100644 --- a/kernel/auditsc.c +++ b/kernel/auditsc.c @@ -1725,7 +1725,7 @@ static inline void handle_one(const struct inode *inode) struct audit_tree_refs *p; struct audit_chunk *chunk; int count; - if (likely(list_empty(&inode->inotify_watches))) + if (likely(hlist_empty(&inode->i_fsnotify_mark_entries))) return; context = current->audit_context; p = context->trees; @@ -1768,7 +1768,7 @@ retry: seq = read_seqbegin(&rename_lock); for(;;) { struct inode *inode = d->d_inode; - if (inode && unlikely(!list_empty(&inode->inotify_watches))) { + if (inode && unlikely(!hlist_empty(&inode->i_fsnotify_mark_entries))) { struct audit_chunk *chunk; chunk = audit_tree_lookup(inode); if (chunk) { -- cgit v1.2.3 From 939a67fc4cbab8ca11c90da8a769d7e965d66a9b Mon Sep 17 00:00:00 2001 From: Eric Paris Date: Thu, 17 Dec 2009 20:12:06 -0500 Subject: Audit: split audit watch Kconfig Audit watch should depend on CONFIG_AUDIT_SYSCALL and should select FSNOTIFY. This splits the spagetti like mixing of audit_watch and audit_filter code so they can be configured seperately. Signed-off-by: Eric Paris --- kernel/Makefile | 5 +++-- kernel/audit.h | 14 +++++++++++++- 2 files changed, 16 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/Makefile b/kernel/Makefile index 057472fbc272..202df4ece6a5 100644 --- a/kernel/Makefile +++ b/kernel/Makefile @@ -70,10 +70,11 @@ obj-$(CONFIG_IKCONFIG) += configs.o obj-$(CONFIG_RESOURCE_COUNTERS) += res_counter.o obj-$(CONFIG_SMP) += stop_machine.o obj-$(CONFIG_KPROBES_SANITY_TEST) += test_kprobes.o -obj-$(CONFIG_AUDIT) += audit.o auditfilter.o audit_watch.o +obj-$(CONFIG_AUDIT) += audit.o auditfilter.o obj-$(CONFIG_AUDITSYSCALL) += auditsc.o -obj-$(CONFIG_GCOV_KERNEL) += gcov/ +obj-$(CONFIG_AUDIT_WATCH) += audit_watch.o obj-$(CONFIG_AUDIT_TREE) += audit_tree.o +obj-$(CONFIG_GCOV_KERNEL) += gcov/ obj-$(CONFIG_KPROBES) += kprobes.o obj-$(CONFIG_KGDB) += debug/ obj-$(CONFIG_DETECT_SOFTLOCKUP) += softlockup.o diff --git a/kernel/audit.h b/kernel/audit.h index 100b454a7354..f7206db4e13d 100644 --- a/kernel/audit.h +++ b/kernel/audit.h @@ -103,7 +103,10 @@ extern struct mutex audit_filter_mutex; extern void audit_free_rule_rcu(struct rcu_head *); extern struct list_head audit_filter_list[]; +extern struct audit_entry *audit_dupe_rule(struct audit_krule *old); + /* audit watch functions */ +#ifdef CONFIG_AUDIT_WATCH extern void audit_put_watch(struct audit_watch *watch); extern void audit_get_watch(struct audit_watch *watch); extern int audit_to_watch(struct audit_krule *krule, char *path, int len, u32 op); @@ -111,7 +114,16 @@ extern int audit_add_watch(struct audit_krule *krule, struct list_head **list); extern void audit_remove_watch_rule(struct audit_krule *krule); extern char *audit_watch_path(struct audit_watch *watch); extern int audit_watch_compare(struct audit_watch *watch, unsigned long ino, dev_t dev); -extern struct audit_entry *audit_dupe_rule(struct audit_krule *old); +#else +#define audit_put_watch(w) {} +#define audit_get_watch(w) {} +#define audit_to_watch(k, p, l, o) (-EINVAL) +#define audit_add_watch(k, l) (-EINVAL) +#define audit_remove_watch_rule(k) BUG() +#define audit_watch_path(w) "" +#define audit_watch_compare(w, i, d) 0 + +#endif /* CONFIG_AUDIT_WATCH */ #ifdef CONFIG_AUDIT_TREE extern struct audit_chunk *audit_tree_lookup(const struct inode *); -- cgit v1.2.3 From 1a3aedbce416dfdbd5d5ac14a0edbcf21a62ee50 Mon Sep 17 00:00:00 2001 From: Eric Paris Date: Thu, 17 Dec 2009 20:12:06 -0500 Subject: Audit: audit watch init should not be before fsnotify init Audit watch init and fsnotify init both use subsys_initcall() but since the audit watch code is linked in before the fsnotify code the audit watch code would be using the fsnotify srcu struct before it was initialized. This patch fixes that problem by moving audit watch init to device_initcall() so it happens after fsnotify is ready. Reported-by: Stephen Rothwell Signed-off-by: Eric Paris Tested-by : Sachin Sant --- kernel/audit_watch.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/audit_watch.c b/kernel/audit_watch.c index c44de0c4fc47..f8543a41115b 100644 --- a/kernel/audit_watch.c +++ b/kernel/audit_watch.c @@ -584,4 +584,4 @@ static int __init audit_watch_init(void) } return 0; } -subsys_initcall(audit_watch_init); +device_initcall(audit_watch_init); -- cgit v1.2.3 From 2dfc1cae4c42b93b831b2417540df2b895ab7108 Mon Sep 17 00:00:00 2001 From: Eric Paris Date: Thu, 17 Dec 2009 20:30:52 -0500 Subject: inotify: remove inotify in kernel interface nothing uses inotify in the kernel, drop it! Signed-off-by: Eric Paris --- kernel/auditsc.c | 1 - 1 file changed, 1 deletion(-) (limited to 'kernel') diff --git a/kernel/auditsc.c b/kernel/auditsc.c index 786901cd8217..853185f7ba7e 100644 --- a/kernel/auditsc.c +++ b/kernel/auditsc.c @@ -65,7 +65,6 @@ #include #include #include -#include #include #include -- cgit v1.2.3 From 7b0a04fbfb35650941af87728d4891515b4fc179 Mon Sep 17 00:00:00 2001 From: Eric Paris Date: Thu, 17 Dec 2009 21:24:21 -0500 Subject: fsnotify: provide the data type to should_send_event fanotify is only interested in event types which contain enough information to open the original file in the context of the fanotify listener. Since fanotify may not want to send events if that data isn't present we pass the data type to the should_send_event function call so fanotify can express its lack of interest. Signed-off-by: Eric Paris --- kernel/audit_tree.c | 3 ++- kernel/audit_watch.c | 3 ++- 2 files changed, 4 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/audit_tree.c b/kernel/audit_tree.c index a164600dd82e..b5417cd65216 100644 --- a/kernel/audit_tree.c +++ b/kernel/audit_tree.c @@ -919,7 +919,8 @@ static void audit_tree_freeing_mark(struct fsnotify_mark_entry *entry, struct fs fsnotify_put_mark(entry); } -static bool audit_tree_send_event(struct fsnotify_group *group, struct inode *inode, __u32 mask) +static bool audit_tree_send_event(struct fsnotify_group *group, struct inode *inode, + __u32 mask, int data_type) { return 0; } diff --git a/kernel/audit_watch.c b/kernel/audit_watch.c index f8543a41115b..67d8f2f52874 100644 --- a/kernel/audit_watch.c +++ b/kernel/audit_watch.c @@ -505,7 +505,8 @@ void audit_remove_watch_rule(struct audit_krule *krule) } } -static bool audit_watch_should_send_event(struct fsnotify_group *group, struct inode *inode, __u32 mask) +static bool audit_watch_should_send_event(struct fsnotify_group *group, struct inode *inode, + __u32 mask, int data_type) { struct fsnotify_mark_entry *entry; bool send; -- cgit v1.2.3 From 8112e2d6a7356e8c3ff1f7f3c86f375ed0305705 Mon Sep 17 00:00:00 2001 From: Eric Paris Date: Thu, 17 Dec 2009 21:24:21 -0500 Subject: fsnotify: include data in should_send calls fanotify is going to need to look at file->private_data to know if an event should be sent or not. This passes the data (which might be a file, dentry, inode, or none) to the should_send function calls so fanotify can get that information when available Signed-off-by: Eric Paris --- kernel/audit_tree.c | 2 +- kernel/audit_watch.c | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/audit_tree.c b/kernel/audit_tree.c index b5417cd65216..e3d63b596ef0 100644 --- a/kernel/audit_tree.c +++ b/kernel/audit_tree.c @@ -920,7 +920,7 @@ static void audit_tree_freeing_mark(struct fsnotify_mark_entry *entry, struct fs } static bool audit_tree_send_event(struct fsnotify_group *group, struct inode *inode, - __u32 mask, int data_type) + __u32 mask, void *data, int data_type) { return 0; } diff --git a/kernel/audit_watch.c b/kernel/audit_watch.c index 67d8f2f52874..85c43aa292e0 100644 --- a/kernel/audit_watch.c +++ b/kernel/audit_watch.c @@ -506,7 +506,7 @@ void audit_remove_watch_rule(struct audit_krule *krule) } static bool audit_watch_should_send_event(struct fsnotify_group *group, struct inode *inode, - __u32 mask, int data_type) + __u32 mask, void *data, int data_type) { struct fsnotify_mark_entry *entry; bool send; -- cgit v1.2.3 From 74be0cc82835aecad332a29896b0f212ba893403 Mon Sep 17 00:00:00 2001 From: Eric Paris Date: Thu, 17 Dec 2009 21:24:22 -0500 Subject: fsnotify: remove group_num altogether The original fsnotify interface has a group-num which was intended to be able to find a group after it was added. I no longer think this is a necessary thing to do and so we remove the group_num. Signed-off-by: Eric Paris --- kernel/audit_tree.c | 3 +-- kernel/audit_watch.c | 2 +- 2 files changed, 2 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/audit_tree.c b/kernel/audit_tree.c index e3d63b596ef0..59065e72a2eb 100644 --- a/kernel/audit_tree.c +++ b/kernel/audit_tree.c @@ -937,8 +937,7 @@ static int __init audit_tree_init(void) { int i; - audit_tree_group = fsnotify_obtain_group(AUDIT_TREE_GROUP_NUM, - 0, &audit_tree_ops); + audit_tree_group = fsnotify_obtain_group(0, &audit_tree_ops); if (IS_ERR(audit_tree_group)) audit_panic("cannot initialize fsnotify group for rectree watches"); diff --git a/kernel/audit_watch.c b/kernel/audit_watch.c index 85c43aa292e0..c500104d38c2 100644 --- a/kernel/audit_watch.c +++ b/kernel/audit_watch.c @@ -577,7 +577,7 @@ static const struct fsnotify_ops audit_watch_fsnotify_ops = { static int __init audit_watch_init(void) { - audit_watch_group = fsnotify_obtain_group(AUDIT_WATCH_GROUP_NUM, AUDIT_FS_WATCH, + audit_watch_group = fsnotify_obtain_group(AUDIT_FS_WATCH, &audit_watch_fsnotify_ops); if (IS_ERR(audit_watch_group)) { audit_watch_group = NULL; -- cgit v1.2.3 From ffab83402f01555a5fa32efb48a4dd0ce8d12ef5 Mon Sep 17 00:00:00 2001 From: Eric Paris Date: Thu, 17 Dec 2009 21:24:22 -0500 Subject: fsnotify: fsnotify_obtain_group should be fsnotify_alloc_group fsnotify_obtain_group was intended to be able to find an already existing group. Nothing uses that functionality. This just renames it to fsnotify_alloc_group so it is clear what it is doing. Signed-off-by: Eric Paris --- kernel/audit_tree.c | 2 +- kernel/audit_watch.c | 4 ++-- 2 files changed, 3 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/audit_tree.c b/kernel/audit_tree.c index 59065e72a2eb..813274d4edad 100644 --- a/kernel/audit_tree.c +++ b/kernel/audit_tree.c @@ -937,7 +937,7 @@ static int __init audit_tree_init(void) { int i; - audit_tree_group = fsnotify_obtain_group(0, &audit_tree_ops); + audit_tree_group = fsnotify_alloc_group(0, &audit_tree_ops); if (IS_ERR(audit_tree_group)) audit_panic("cannot initialize fsnotify group for rectree watches"); diff --git a/kernel/audit_watch.c b/kernel/audit_watch.c index c500104d38c2..0f03a6ab96ed 100644 --- a/kernel/audit_watch.c +++ b/kernel/audit_watch.c @@ -577,8 +577,8 @@ static const struct fsnotify_ops audit_watch_fsnotify_ops = { static int __init audit_watch_init(void) { - audit_watch_group = fsnotify_obtain_group(AUDIT_FS_WATCH, - &audit_watch_fsnotify_ops); + audit_watch_group = fsnotify_alloc_group(AUDIT_FS_WATCH, + &audit_watch_fsnotify_ops); if (IS_ERR(audit_watch_group)) { audit_watch_group = NULL; audit_panic("cannot create audit fsnotify group"); -- cgit v1.2.3 From 220d14df0dc587c06b97762829a41157c9375b94 Mon Sep 17 00:00:00 2001 From: Eric Paris Date: Thu, 17 Dec 2009 21:24:22 -0500 Subject: Audit: only set group mask when something is being watched Currently the audit watch group always sets a mask equal to all events it might care about. We instead should only set the group mask if we are actually watching inodes. This should be a perf win when audit watches are compiled in. Signed-off-by: Eric Paris --- kernel/audit_watch.c | 11 +++++++++-- 1 file changed, 9 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/audit_watch.c b/kernel/audit_watch.c index 0f03a6ab96ed..87408b282118 100644 --- a/kernel/audit_watch.c +++ b/kernel/audit_watch.c @@ -167,6 +167,8 @@ static struct audit_parent *audit_init_parent(struct nameidata *ndp) return ERR_PTR(ret); } + fsnotify_recalc_group_mask(audit_watch_group); + return parent; } @@ -353,6 +355,9 @@ static void audit_remove_parent_watches(struct audit_parent *parent) mutex_unlock(&audit_filter_mutex); fsnotify_destroy_mark_by_entry(&parent->mark); + + fsnotify_recalc_group_mask(audit_watch_group); + } /* Get path information necessary for adding watches. */ @@ -503,6 +508,9 @@ void audit_remove_watch_rule(struct audit_krule *krule) audit_put_parent(parent); } } + + fsnotify_recalc_group_mask(audit_watch_group); + } static bool audit_watch_should_send_event(struct fsnotify_group *group, struct inode *inode, @@ -577,8 +585,7 @@ static const struct fsnotify_ops audit_watch_fsnotify_ops = { static int __init audit_watch_init(void) { - audit_watch_group = fsnotify_alloc_group(AUDIT_FS_WATCH, - &audit_watch_fsnotify_ops); + audit_watch_group = fsnotify_alloc_group(0, &audit_watch_fsnotify_ops); if (IS_ERR(audit_watch_group)) { audit_watch_group = NULL; audit_panic("cannot create audit fsnotify group"); -- cgit v1.2.3 From 0d2e2a1d00d7d23e5bd9bb0935cde7c3d5835c56 Mon Sep 17 00:00:00 2001 From: Eric Paris Date: Thu, 17 Dec 2009 21:24:22 -0500 Subject: fsnotify: drop mask argument from fsnotify_alloc_group Nothing uses the mask argument to fsnotify_alloc_group. This patch drops that argument. Signed-off-by: Eric Paris --- kernel/audit_tree.c | 2 +- kernel/audit_watch.c | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/audit_tree.c b/kernel/audit_tree.c index 813274d4edad..04f16887406b 100644 --- a/kernel/audit_tree.c +++ b/kernel/audit_tree.c @@ -937,7 +937,7 @@ static int __init audit_tree_init(void) { int i; - audit_tree_group = fsnotify_alloc_group(0, &audit_tree_ops); + audit_tree_group = fsnotify_alloc_group(&audit_tree_ops); if (IS_ERR(audit_tree_group)) audit_panic("cannot initialize fsnotify group for rectree watches"); diff --git a/kernel/audit_watch.c b/kernel/audit_watch.c index 87408b282118..83d5f9674cec 100644 --- a/kernel/audit_watch.c +++ b/kernel/audit_watch.c @@ -585,7 +585,7 @@ static const struct fsnotify_ops audit_watch_fsnotify_ops = { static int __init audit_watch_init(void) { - audit_watch_group = fsnotify_alloc_group(0, &audit_watch_fsnotify_ops); + audit_watch_group = fsnotify_alloc_group(&audit_watch_fsnotify_ops); if (IS_ERR(audit_watch_group)) { audit_watch_group = NULL; audit_panic("cannot create audit fsnotify group"); -- cgit v1.2.3 From 3a9fb89f4cd04c23e16397befba92efb5d989b74 Mon Sep 17 00:00:00 2001 From: Eric Paris Date: Thu, 17 Dec 2009 21:24:23 -0500 Subject: fsnotify: include vfsmount in should_send_event when appropriate To ensure that a group will not duplicate events when it receives it based on the vfsmount and the inode should_send_event test we should distinguish those two cases. We pass a vfsmount to this function so groups can make their own determinations. Signed-off-by: Eric Paris --- kernel/audit_tree.c | 3 ++- kernel/audit_watch.c | 3 ++- 2 files changed, 4 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/audit_tree.c b/kernel/audit_tree.c index 04f16887406b..ecf0bf260d09 100644 --- a/kernel/audit_tree.c +++ b/kernel/audit_tree.c @@ -920,7 +920,8 @@ static void audit_tree_freeing_mark(struct fsnotify_mark_entry *entry, struct fs } static bool audit_tree_send_event(struct fsnotify_group *group, struct inode *inode, - __u32 mask, void *data, int data_type) + struct vfsmount *mnt, __u32 mask, void *data, + int data_type) { return 0; } diff --git a/kernel/audit_watch.c b/kernel/audit_watch.c index 83d5f9674cec..6304ee5d7642 100644 --- a/kernel/audit_watch.c +++ b/kernel/audit_watch.c @@ -514,7 +514,8 @@ void audit_remove_watch_rule(struct audit_krule *krule) } static bool audit_watch_should_send_event(struct fsnotify_group *group, struct inode *inode, - __u32 mask, void *data, int data_type) + struct vfsmount *mnt, __u32 mask, void *data, + int data_type) { struct fsnotify_mark_entry *entry; bool send; -- cgit v1.2.3 From 2823e04de4f1a49087b58ff2bb8f61361ffd9321 Mon Sep 17 00:00:00 2001 From: Eric Paris Date: Thu, 17 Dec 2009 21:24:23 -0500 Subject: fsnotify: put inode specific fields in an fsnotify_mark in a union The addition of marks on vfs mounts will be simplified if the inode specific parts of a mark and the vfsmnt specific parts of a mark are actually in a union so naming can be easy. This patch just implements the inode struct and the union. Signed-off-by: Eric Paris --- kernel/audit_tree.c | 16 ++++++++-------- 1 file changed, 8 insertions(+), 8 deletions(-) (limited to 'kernel') diff --git a/kernel/audit_tree.c b/kernel/audit_tree.c index ecf0bf260d09..c21b05d25224 100644 --- a/kernel/audit_tree.c +++ b/kernel/audit_tree.c @@ -179,9 +179,9 @@ static void insert_hash(struct audit_chunk *chunk) struct fsnotify_mark_entry *entry = &chunk->mark; struct list_head *list; - if (!entry->inode) + if (!entry->i.inode) return; - list = chunk_hash(entry->inode); + list = chunk_hash(entry->i.inode); list_add_rcu(&chunk->hash, list); } @@ -193,7 +193,7 @@ struct audit_chunk *audit_tree_lookup(const struct inode *inode) list_for_each_entry_rcu(p, list, hash) { /* mark.inode may have gone NULL, but who cares? */ - if (p->mark.inode == inode) { + if (p->mark.i.inode == inode) { atomic_long_inc(&p->refs); return p; } @@ -233,7 +233,7 @@ static void untag_chunk(struct node *p) spin_unlock(&hash_lock); spin_lock(&entry->lock); - if (chunk->dead || !entry->inode) { + if (chunk->dead || !entry->i.inode) { spin_unlock(&entry->lock); goto out; } @@ -259,7 +259,7 @@ static void untag_chunk(struct node *p) if (!new) goto Fallback; fsnotify_duplicate_mark(&new->mark, entry); - if (fsnotify_add_mark(&new->mark, new->mark.group, new->mark.inode, 1)) { + if (fsnotify_add_mark(&new->mark, new->mark.group, new->mark.i.inode, 1)) { free_chunk(new); goto Fallback; } @@ -388,7 +388,7 @@ static int tag_chunk(struct inode *inode, struct audit_tree *tree) chunk_entry = &chunk->mark; spin_lock(&old_entry->lock); - if (!old_entry->inode) { + if (!old_entry->i.inode) { /* old_entry is being shot, lets just lie */ spin_unlock(&old_entry->lock); fsnotify_put_mark(old_entry); @@ -397,7 +397,7 @@ static int tag_chunk(struct inode *inode, struct audit_tree *tree) } fsnotify_duplicate_mark(chunk_entry, old_entry); - if (fsnotify_add_mark(chunk_entry, chunk_entry->group, chunk_entry->inode, 1)) { + if (fsnotify_add_mark(chunk_entry, chunk_entry->group, chunk_entry->i.inode, 1)) { spin_unlock(&old_entry->lock); free_chunk(chunk); fsnotify_put_mark(old_entry); @@ -605,7 +605,7 @@ void audit_trim_trees(void) list_for_each_entry(node, &tree->chunks, list) { struct audit_chunk *chunk = find_chunk(node); /* this could be NULL if the watch is dieing else where... */ - struct inode *inode = chunk->mark.inode; + struct inode *inode = chunk->mark.i.inode; node->index |= 1U<<31; if (iterate_mounts(compare_root, inode, root_mnt)) node->index &= ~(1U<<31); -- cgit v1.2.3 From e61ce86737b4d60521e4e71f9892fe4bdcfb688b Mon Sep 17 00:00:00 2001 From: Eric Paris Date: Thu, 17 Dec 2009 21:24:24 -0500 Subject: fsnotify: rename fsnotify_mark_entry to just fsnotify_mark The name is long and it serves no real purpose. So rename fsnotify_mark_entry to just fsnotify_mark. Signed-off-by: Eric Paris --- kernel/audit_tree.c | 14 +++++++------- kernel/audit_watch.c | 8 ++++---- kernel/auditsc.c | 4 ++-- 3 files changed, 13 insertions(+), 13 deletions(-) (limited to 'kernel') diff --git a/kernel/audit_tree.c b/kernel/audit_tree.c index c21b05d25224..f16f909fbbc1 100644 --- a/kernel/audit_tree.c +++ b/kernel/audit_tree.c @@ -22,7 +22,7 @@ struct audit_tree { struct audit_chunk { struct list_head hash; - struct fsnotify_mark_entry mark; + struct fsnotify_mark mark; struct list_head trees; /* with root here */ int dead; int count; @@ -134,7 +134,7 @@ static void __put_chunk(struct rcu_head *rcu) audit_put_chunk(chunk); } -static void audit_tree_destroy_watch(struct fsnotify_mark_entry *entry) +static void audit_tree_destroy_watch(struct fsnotify_mark *entry) { struct audit_chunk *chunk = container_of(entry, struct audit_chunk, mark); call_rcu(&chunk->head, __put_chunk); @@ -176,7 +176,7 @@ static inline struct list_head *chunk_hash(const struct inode *inode) /* hash_lock & entry->lock is held by caller */ static void insert_hash(struct audit_chunk *chunk) { - struct fsnotify_mark_entry *entry = &chunk->mark; + struct fsnotify_mark *entry = &chunk->mark; struct list_head *list; if (!entry->i.inode) @@ -222,7 +222,7 @@ static struct audit_chunk *find_chunk(struct node *p) static void untag_chunk(struct node *p) { struct audit_chunk *chunk = find_chunk(p); - struct fsnotify_mark_entry *entry = &chunk->mark; + struct fsnotify_mark *entry = &chunk->mark; struct audit_chunk *new; struct audit_tree *owner; int size = chunk->count - 1; @@ -316,7 +316,7 @@ out: static int create_chunk(struct inode *inode, struct audit_tree *tree) { - struct fsnotify_mark_entry *entry; + struct fsnotify_mark *entry; struct audit_chunk *chunk = alloc_chunk(1); if (!chunk) return -ENOMEM; @@ -354,7 +354,7 @@ static int create_chunk(struct inode *inode, struct audit_tree *tree) /* the first tagged inode becomes root of tree */ static int tag_chunk(struct inode *inode, struct audit_tree *tree) { - struct fsnotify_mark_entry *old_entry, *chunk_entry; + struct fsnotify_mark *old_entry, *chunk_entry; struct audit_tree *owner; struct audit_chunk *chunk, *old; struct node *p; @@ -911,7 +911,7 @@ static int audit_tree_handle_event(struct fsnotify_group *group, struct fsnotify return -EOPNOTSUPP; } -static void audit_tree_freeing_mark(struct fsnotify_mark_entry *entry, struct fsnotify_group *group) +static void audit_tree_freeing_mark(struct fsnotify_mark *entry, struct fsnotify_group *group) { struct audit_chunk *chunk = container_of(entry, struct audit_chunk, mark); diff --git a/kernel/audit_watch.c b/kernel/audit_watch.c index 6304ee5d7642..d8cb55a5c059 100644 --- a/kernel/audit_watch.c +++ b/kernel/audit_watch.c @@ -56,7 +56,7 @@ struct audit_watch { struct audit_parent { struct list_head watches; /* anchor for audit_watch->wlist */ - struct fsnotify_mark_entry mark; /* fsnotify mark on the inode */ + struct fsnotify_mark mark; /* fsnotify mark on the inode */ }; /* fsnotify handle. */ @@ -72,7 +72,7 @@ static void audit_free_parent(struct audit_parent *parent) kfree(parent); } -static void audit_watch_free_mark(struct fsnotify_mark_entry *entry) +static void audit_watch_free_mark(struct fsnotify_mark *entry) { struct audit_parent *parent; @@ -99,7 +99,7 @@ static void audit_put_parent(struct audit_parent *parent) static inline struct audit_parent *audit_find_parent(struct inode *inode) { struct audit_parent *parent = NULL; - struct fsnotify_mark_entry *entry; + struct fsnotify_mark *entry; spin_lock(&inode->i_lock); entry = fsnotify_find_mark_entry(audit_watch_group, inode); @@ -517,7 +517,7 @@ static bool audit_watch_should_send_event(struct fsnotify_group *group, struct i struct vfsmount *mnt, __u32 mask, void *data, int data_type) { - struct fsnotify_mark_entry *entry; + struct fsnotify_mark *entry; bool send; spin_lock(&inode->i_lock); diff --git a/kernel/auditsc.c b/kernel/auditsc.c index 853185f7ba7e..b87a63beb66c 100644 --- a/kernel/auditsc.c +++ b/kernel/auditsc.c @@ -1724,7 +1724,7 @@ static inline void handle_one(const struct inode *inode) struct audit_tree_refs *p; struct audit_chunk *chunk; int count; - if (likely(hlist_empty(&inode->i_fsnotify_mark_entries))) + if (likely(hlist_empty(&inode->i_fsnotify_marks))) return; context = current->audit_context; p = context->trees; @@ -1767,7 +1767,7 @@ retry: seq = read_seqbegin(&rename_lock); for(;;) { struct inode *inode = d->d_inode; - if (inode && unlikely(!hlist_empty(&inode->i_fsnotify_mark_entries))) { + if (inode && unlikely(!hlist_empty(&inode->i_fsnotify_marks))) { struct audit_chunk *chunk; chunk = audit_tree_lookup(inode); if (chunk) { -- cgit v1.2.3 From d07754412f9cdc2f4a99318d5ee81ace6715ea99 Mon Sep 17 00:00:00 2001 From: Eric Paris Date: Thu, 17 Dec 2009 21:24:24 -0500 Subject: fsnotify: rename fsnotify_find_mark_entry to fsnotify_find_mark the _entry portion of fsnotify functions is useless. Drop it. Signed-off-by: Eric Paris --- kernel/audit_tree.c | 12 ++++++------ kernel/audit_watch.c | 8 ++++---- 2 files changed, 10 insertions(+), 10 deletions(-) (limited to 'kernel') diff --git a/kernel/audit_tree.c b/kernel/audit_tree.c index f16f909fbbc1..b20fb055d712 100644 --- a/kernel/audit_tree.c +++ b/kernel/audit_tree.c @@ -250,7 +250,7 @@ static void untag_chunk(struct node *p) list_del_rcu(&chunk->hash); spin_unlock(&hash_lock); spin_unlock(&entry->lock); - fsnotify_destroy_mark_by_entry(entry); + fsnotify_destroy_mark(entry); fsnotify_put_mark(entry); goto out; } @@ -293,7 +293,7 @@ static void untag_chunk(struct node *p) owner->root = new; spin_unlock(&hash_lock); spin_unlock(&entry->lock); - fsnotify_destroy_mark_by_entry(entry); + fsnotify_destroy_mark(entry); fsnotify_put_mark(entry); goto out; @@ -333,7 +333,7 @@ static int create_chunk(struct inode *inode, struct audit_tree *tree) spin_unlock(&hash_lock); chunk->dead = 1; spin_unlock(&entry->lock); - fsnotify_destroy_mark_by_entry(entry); + fsnotify_destroy_mark(entry); fsnotify_put_mark(entry); return 0; } @@ -361,7 +361,7 @@ static int tag_chunk(struct inode *inode, struct audit_tree *tree) int n; spin_lock(&inode->i_lock); - old_entry = fsnotify_find_mark_entry(audit_tree_group, inode); + old_entry = fsnotify_find_mark(audit_tree_group, inode); spin_unlock(&inode->i_lock); if (!old_entry) return create_chunk(inode, tree); @@ -415,7 +415,7 @@ static int tag_chunk(struct inode *inode, struct audit_tree *tree) spin_unlock(&chunk_entry->lock); spin_unlock(&old_entry->lock); - fsnotify_destroy_mark_by_entry(chunk_entry); + fsnotify_destroy_mark(chunk_entry); fsnotify_put_mark(chunk_entry); fsnotify_put_mark(old_entry); @@ -446,7 +446,7 @@ static int tag_chunk(struct inode *inode, struct audit_tree *tree) spin_unlock(&hash_lock); spin_unlock(&chunk_entry->lock); spin_unlock(&old_entry->lock); - fsnotify_destroy_mark_by_entry(old_entry); + fsnotify_destroy_mark(old_entry); fsnotify_put_mark(old_entry); /* pair to fsnotify_find mark_entry */ fsnotify_put_mark(old_entry); /* and kill it */ return 0; diff --git a/kernel/audit_watch.c b/kernel/audit_watch.c index d8cb55a5c059..24ecbebf4354 100644 --- a/kernel/audit_watch.c +++ b/kernel/audit_watch.c @@ -102,7 +102,7 @@ static inline struct audit_parent *audit_find_parent(struct inode *inode) struct fsnotify_mark *entry; spin_lock(&inode->i_lock); - entry = fsnotify_find_mark_entry(audit_watch_group, inode); + entry = fsnotify_find_mark(audit_watch_group, inode); spin_unlock(&inode->i_lock); if (entry) @@ -354,7 +354,7 @@ static void audit_remove_parent_watches(struct audit_parent *parent) } mutex_unlock(&audit_filter_mutex); - fsnotify_destroy_mark_by_entry(&parent->mark); + fsnotify_destroy_mark(&parent->mark); fsnotify_recalc_group_mask(audit_watch_group); @@ -504,7 +504,7 @@ void audit_remove_watch_rule(struct audit_krule *krule) if (list_empty(&parent->watches)) { audit_get_parent(parent); - fsnotify_destroy_mark_by_entry(&parent->mark); + fsnotify_destroy_mark(&parent->mark); audit_put_parent(parent); } } @@ -521,7 +521,7 @@ static bool audit_watch_should_send_event(struct fsnotify_group *group, struct i bool send; spin_lock(&inode->i_lock); - entry = fsnotify_find_mark_entry(group, inode); + entry = fsnotify_find_mark(group, inode); spin_unlock(&inode->i_lock); if (!entry) return false; -- cgit v1.2.3 From 35566087099c3ff8901d65ee98af56347ee66e5a Mon Sep 17 00:00:00 2001 From: Andreas Gruenbacher Date: Thu, 17 Dec 2009 21:24:25 -0500 Subject: fsnotify: take inode->i_lock inside fsnotify_find_mark_entry() All callers to fsnotify_find_mark_entry() except one take and release inode->i_lock around the call. Take the lock inside fsnotify_find_mark_entry() instead. Signed-off-by: Andreas Gruenbacher Signed-off-by: Eric Paris --- kernel/audit_tree.c | 2 -- kernel/audit_watch.c | 5 ----- 2 files changed, 7 deletions(-) (limited to 'kernel') diff --git a/kernel/audit_tree.c b/kernel/audit_tree.c index b20fb055d712..80f8ac328aad 100644 --- a/kernel/audit_tree.c +++ b/kernel/audit_tree.c @@ -360,9 +360,7 @@ static int tag_chunk(struct inode *inode, struct audit_tree *tree) struct node *p; int n; - spin_lock(&inode->i_lock); old_entry = fsnotify_find_mark(audit_tree_group, inode); - spin_unlock(&inode->i_lock); if (!old_entry) return create_chunk(inode, tree); diff --git a/kernel/audit_watch.c b/kernel/audit_watch.c index 24ecbebf4354..d85fa538a722 100644 --- a/kernel/audit_watch.c +++ b/kernel/audit_watch.c @@ -101,10 +101,7 @@ static inline struct audit_parent *audit_find_parent(struct inode *inode) struct audit_parent *parent = NULL; struct fsnotify_mark *entry; - spin_lock(&inode->i_lock); entry = fsnotify_find_mark(audit_watch_group, inode); - spin_unlock(&inode->i_lock); - if (entry) parent = container_of(entry, struct audit_parent, mark); @@ -520,9 +517,7 @@ static bool audit_watch_should_send_event(struct fsnotify_group *group, struct i struct fsnotify_mark *entry; bool send; - spin_lock(&inode->i_lock); entry = fsnotify_find_mark(group, inode); - spin_unlock(&inode->i_lock); if (!entry) return false; -- cgit v1.2.3 From 11637e4b7dc098e9a863f0a619d55ebc60f5949e Mon Sep 17 00:00:00 2001 From: Eric Paris Date: Thu, 17 Dec 2009 21:24:25 -0500 Subject: fanotify: fanotify_init syscall declaration This patch defines a new syscall fanotify_init() of the form: int sys_fanotify_init(unsigned int flags, unsigned int event_f_flags, unsigned int priority) This syscall is used to create and fanotify group. This is very similar to the inotify_init() syscall. Signed-off-by: Eric Paris --- kernel/sys_ni.c | 3 +++ 1 file changed, 3 insertions(+) (limited to 'kernel') diff --git a/kernel/sys_ni.c b/kernel/sys_ni.c index 70f2ea758ffe..2c4adc2decc3 100644 --- a/kernel/sys_ni.c +++ b/kernel/sys_ni.c @@ -181,3 +181,6 @@ cond_syscall(sys_eventfd2); /* performance counters: */ cond_syscall(sys_perf_event_open); + +/* fanotify! */ +cond_syscall(sys_fanotify_init); -- cgit v1.2.3 From bbaa4168b2d2d8cc674e6d35806e8426aef464b8 Mon Sep 17 00:00:00 2001 From: Eric Paris Date: Thu, 17 Dec 2009 21:24:26 -0500 Subject: fanotify: sys_fanotify_mark declartion This patch simply declares the new sys_fanotify_mark syscall int fanotify_mark(int fanotify_fd, unsigned int flags, u64_mask, int dfd const char *pathname) Signed-off-by: Eric Paris --- kernel/sys_ni.c | 1 + 1 file changed, 1 insertion(+) (limited to 'kernel') diff --git a/kernel/sys_ni.c b/kernel/sys_ni.c index 2c4adc2decc3..bad369ec5403 100644 --- a/kernel/sys_ni.c +++ b/kernel/sys_ni.c @@ -184,3 +184,4 @@ cond_syscall(sys_perf_event_open); /* fanotify! */ cond_syscall(sys_fanotify_init); +cond_syscall(sys_fanotify_mark); -- cgit v1.2.3 From 5444e2981c31d0ed7465475e451b8437084337e5 Mon Sep 17 00:00:00 2001 From: Eric Paris Date: Thu, 17 Dec 2009 21:24:27 -0500 Subject: fsnotify: split generic and inode specific mark code currently all marking is done by functions in inode-mark.c. Some of this is pretty generic and should be instead done in a generic function and we should only put the inode specific code in inode-mark.c Signed-off-by: Eric Paris --- kernel/audit_tree.c | 8 ++++---- kernel/audit_watch.c | 6 +++--- 2 files changed, 7 insertions(+), 7 deletions(-) (limited to 'kernel') diff --git a/kernel/audit_tree.c b/kernel/audit_tree.c index 80f8ac328aad..cfb97d752a61 100644 --- a/kernel/audit_tree.c +++ b/kernel/audit_tree.c @@ -259,7 +259,7 @@ static void untag_chunk(struct node *p) if (!new) goto Fallback; fsnotify_duplicate_mark(&new->mark, entry); - if (fsnotify_add_mark(&new->mark, new->mark.group, new->mark.i.inode, 1)) { + if (fsnotify_add_mark(&new->mark, new->mark.group, new->mark.i.inode, NULL, 1)) { free_chunk(new); goto Fallback; } @@ -322,7 +322,7 @@ static int create_chunk(struct inode *inode, struct audit_tree *tree) return -ENOMEM; entry = &chunk->mark; - if (fsnotify_add_mark(entry, audit_tree_group, inode, 0)) { + if (fsnotify_add_mark(entry, audit_tree_group, inode, NULL, 0)) { free_chunk(chunk); return -ENOSPC; } @@ -360,7 +360,7 @@ static int tag_chunk(struct inode *inode, struct audit_tree *tree) struct node *p; int n; - old_entry = fsnotify_find_mark(audit_tree_group, inode); + old_entry = fsnotify_find_inode_mark(audit_tree_group, inode); if (!old_entry) return create_chunk(inode, tree); @@ -395,7 +395,7 @@ static int tag_chunk(struct inode *inode, struct audit_tree *tree) } fsnotify_duplicate_mark(chunk_entry, old_entry); - if (fsnotify_add_mark(chunk_entry, chunk_entry->group, chunk_entry->i.inode, 1)) { + if (fsnotify_add_mark(chunk_entry, chunk_entry->group, chunk_entry->i.inode, NULL, 1)) { spin_unlock(&old_entry->lock); free_chunk(chunk); fsnotify_put_mark(old_entry); diff --git a/kernel/audit_watch.c b/kernel/audit_watch.c index d85fa538a722..7499397a6100 100644 --- a/kernel/audit_watch.c +++ b/kernel/audit_watch.c @@ -101,7 +101,7 @@ static inline struct audit_parent *audit_find_parent(struct inode *inode) struct audit_parent *parent = NULL; struct fsnotify_mark *entry; - entry = fsnotify_find_mark(audit_watch_group, inode); + entry = fsnotify_find_inode_mark(audit_watch_group, inode); if (entry) parent = container_of(entry, struct audit_parent, mark); @@ -158,7 +158,7 @@ static struct audit_parent *audit_init_parent(struct nameidata *ndp) fsnotify_init_mark(&parent->mark, audit_watch_free_mark); parent->mark.mask = AUDIT_FS_WATCH; - ret = fsnotify_add_mark(&parent->mark, audit_watch_group, inode, 0); + ret = fsnotify_add_mark(&parent->mark, audit_watch_group, inode, NULL, 0); if (ret < 0) { audit_free_parent(parent); return ERR_PTR(ret); @@ -517,7 +517,7 @@ static bool audit_watch_should_send_event(struct fsnotify_group *group, struct i struct fsnotify_mark *entry; bool send; - entry = fsnotify_find_mark(group, inode); + entry = fsnotify_find_inode_mark(group, inode); if (!entry) return false; -- cgit v1.2.3 From 6e006701ccc1590500186ef21e074bd900c5dd67 Mon Sep 17 00:00:00 2001 From: Alexey Dobriyan Date: Wed, 20 Jan 2010 22:27:56 +0200 Subject: dnotify: move dir_notify_enable declaration Move dir_notify_enable declaration to where it belongs -- dnotify.h . Signed-off-by: Alexey Dobriyan Signed-off-by: Eric Paris --- kernel/sysctl.c | 1 + 1 file changed, 1 insertion(+) (limited to 'kernel') diff --git a/kernel/sysctl.c b/kernel/sysctl.c index d24f761f4876..7b983dbfe0ec 100644 --- a/kernel/sysctl.c +++ b/kernel/sysctl.c @@ -44,6 +44,7 @@ #include #include #include +#include #include #include #include -- cgit v1.2.3 From d14f1729483fad3a8817fbbcbd017678b7d1ad26 Mon Sep 17 00:00:00 2001 From: Dave Young Date: Thu, 25 Feb 2010 20:28:57 -0500 Subject: sysctl extern cleanup: inotify Extern declarations in sysctl.c should be move to their own head file, and then include them in relavant .c files. Move inotify_table extern declaration to linux/inotify.h Signed-off-by: Dave Young Signed-off-by: Andrew Morton Signed-off-by: Eric Paris --- kernel/sysctl.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/sysctl.c b/kernel/sysctl.c index 7b983dbfe0ec..fe30db7bdb0a 100644 --- a/kernel/sysctl.c +++ b/kernel/sysctl.c @@ -131,6 +131,9 @@ static int min_percpu_pagelist_fract = 8; static int ngroups_max = NGROUPS_MAX; +#ifdef CONFIG_INOTIFY_USER +#include +#endif #ifdef CONFIG_SPARC #include #endif @@ -207,9 +210,6 @@ static struct ctl_table fs_table[]; static struct ctl_table debug_table[]; static struct ctl_table dev_table[]; extern struct ctl_table random_table[]; -#ifdef CONFIG_INOTIFY_USER -extern struct ctl_table inotify_table[]; -#endif #ifdef CONFIG_EPOLL extern struct ctl_table epoll_table[]; #endif -- cgit v1.2.3 From 3bcf3860a4ff9bbc522820b4b765e65e4deceb3e Mon Sep 17 00:00:00 2001 From: Eric Paris Date: Wed, 28 Jul 2010 10:18:37 -0400 Subject: fsnotify: store struct file not struct path Al explains that calling dentry_open() with a mnt/dentry pair is only garunteed to be safe if they are already used in an open struct file. To make sure this is the case don't store and use a struct path in fsnotify, always use a struct file. Signed-off-by: Eric Paris --- kernel/audit_watch.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/audit_watch.c b/kernel/audit_watch.c index 7499397a6100..b955a22d8ff1 100644 --- a/kernel/audit_watch.c +++ b/kernel/audit_watch.c @@ -545,8 +545,8 @@ static int audit_watch_handle_event(struct fsnotify_group *group, struct fsnotif return 0; switch (event->data_type) { - case (FSNOTIFY_EVENT_PATH): - inode = event->path.dentry->d_inode; + case (FSNOTIFY_EVENT_FILE): + inode = event->file->f_path.dentry->d_inode; break; case (FSNOTIFY_EVENT_INODE): inode = event->inode; -- cgit v1.2.3 From 3a9b16b407f10b2a771bcae13fb5791e527d6bcf Mon Sep 17 00:00:00 2001 From: Eric Paris Date: Wed, 28 Jul 2010 10:18:38 -0400 Subject: fsnotify: send fsnotify_mark to groups in event handling functions With the change of fsnotify to use srcu walking the marks list instead of walking the global groups list we now know the mark in question. The code can send the mark to the group's handling functions and the groups won't have to find those marks themselves. Signed-off-by: Eric Paris --- kernel/audit_tree.c | 8 +++++--- kernel/audit_watch.c | 8 +++++--- 2 files changed, 10 insertions(+), 6 deletions(-) (limited to 'kernel') diff --git a/kernel/audit_tree.c b/kernel/audit_tree.c index cfb97d752a61..584b94360217 100644 --- a/kernel/audit_tree.c +++ b/kernel/audit_tree.c @@ -903,7 +903,9 @@ static void evict_chunk(struct audit_chunk *chunk) mutex_unlock(&audit_filter_mutex); } -static int audit_tree_handle_event(struct fsnotify_group *group, struct fsnotify_event *event) +static int audit_tree_handle_event(struct fsnotify_group *group, + struct fsnotify_mark *mark, + struct fsnotify_event *event) { BUG(); return -EOPNOTSUPP; @@ -918,8 +920,8 @@ static void audit_tree_freeing_mark(struct fsnotify_mark *entry, struct fsnotify } static bool audit_tree_send_event(struct fsnotify_group *group, struct inode *inode, - struct vfsmount *mnt, __u32 mask, void *data, - int data_type) + struct vfsmount *mnt, struct fsnotify_mark *mark, + __u32 mask, void *data, int data_type) { return 0; } diff --git a/kernel/audit_watch.c b/kernel/audit_watch.c index b955a22d8ff1..4d5ea0319a6c 100644 --- a/kernel/audit_watch.c +++ b/kernel/audit_watch.c @@ -511,8 +511,8 @@ void audit_remove_watch_rule(struct audit_krule *krule) } static bool audit_watch_should_send_event(struct fsnotify_group *group, struct inode *inode, - struct vfsmount *mnt, __u32 mask, void *data, - int data_type) + struct vfsmount *mnt, struct fsnotify_mark *mark, + __u32 mask, void *data, int data_type) { struct fsnotify_mark *entry; bool send; @@ -531,7 +531,9 @@ static bool audit_watch_should_send_event(struct fsnotify_group *group, struct i } /* Update watch data in audit rules based on fsnotify events. */ -static int audit_watch_handle_event(struct fsnotify_group *group, struct fsnotify_event *event) +static int audit_watch_handle_event(struct fsnotify_group *group, + struct fsnotify_mark *mark, + struct fsnotify_event *event) { struct inode *inode; __u32 mask = event->mask; -- cgit v1.2.3 From 4cd76a47924cd966799402d0f2bba356cde5c1b3 Mon Sep 17 00:00:00 2001 From: Eric Paris Date: Wed, 28 Jul 2010 10:18:38 -0400 Subject: audit: use the mark in handler functions audit now gets a mark in the should_send_event and handle_event functions. Rather than look up the mark themselves audit should just use the mark it was handed. Signed-off-by: Eric Paris --- kernel/audit_watch.c | 20 +++----------------- 1 file changed, 3 insertions(+), 17 deletions(-) (limited to 'kernel') diff --git a/kernel/audit_watch.c b/kernel/audit_watch.c index 4d5ea0319a6c..9173bcf33763 100644 --- a/kernel/audit_watch.c +++ b/kernel/audit_watch.c @@ -514,18 +514,10 @@ static bool audit_watch_should_send_event(struct fsnotify_group *group, struct i struct vfsmount *mnt, struct fsnotify_mark *mark, __u32 mask, void *data, int data_type) { - struct fsnotify_mark *entry; bool send; - entry = fsnotify_find_inode_mark(group, inode); - if (!entry) - return false; - mask = (mask & ~FS_EVENT_ON_CHILD); - send = (entry->mask & mask); - - /* find took a reference */ - fsnotify_put_mark(entry); + send = (mark->mask & mask); return send; } @@ -540,11 +532,9 @@ static int audit_watch_handle_event(struct fsnotify_group *group, const char *dname = event->file_name; struct audit_parent *parent; - BUG_ON(group != audit_watch_group); + parent = container_of(mark, struct audit_parent, mark); - parent = audit_find_parent(event->to_tell); - if (unlikely(!parent)) - return 0; + BUG_ON(group != audit_watch_group); switch (event->data_type) { case (FSNOTIFY_EVENT_FILE): @@ -565,10 +555,6 @@ static int audit_watch_handle_event(struct fsnotify_group *group, audit_update_watch(parent, dname, (dev_t)-1, (unsigned long)-1, 1); else if (mask & (FS_DELETE_SELF|FS_UNMOUNT|FS_MOVE_SELF)) audit_remove_parent_watches(parent); - /* moved put_inotify_watch to freeing mark */ - - /* matched the ref taken by audit_find_parent */ - audit_put_parent(parent); return 0; } -- cgit v1.2.3 From 2612abb51b11ffd2d75c472b11178115f5808909 Mon Sep 17 00:00:00 2001 From: Eric Paris Date: Wed, 28 Jul 2010 10:18:39 -0400 Subject: fsnotify: cleanup should_send_event The change to use srcu and walk the object list rather than the global fsnotify_group list means that should_send_event is no longer needed for a number of groups and can be simplified for others. Do that. Signed-off-by: Eric Paris --- kernel/audit_tree.c | 2 +- kernel/audit_watch.c | 7 +------ 2 files changed, 2 insertions(+), 7 deletions(-) (limited to 'kernel') diff --git a/kernel/audit_tree.c b/kernel/audit_tree.c index 584b94360217..2abb99f3459d 100644 --- a/kernel/audit_tree.c +++ b/kernel/audit_tree.c @@ -923,7 +923,7 @@ static bool audit_tree_send_event(struct fsnotify_group *group, struct inode *in struct vfsmount *mnt, struct fsnotify_mark *mark, __u32 mask, void *data, int data_type) { - return 0; + return false; } static const struct fsnotify_ops audit_tree_ops = { diff --git a/kernel/audit_watch.c b/kernel/audit_watch.c index 9173bcf33763..097a61c65fe0 100644 --- a/kernel/audit_watch.c +++ b/kernel/audit_watch.c @@ -514,12 +514,7 @@ static bool audit_watch_should_send_event(struct fsnotify_group *group, struct i struct vfsmount *mnt, struct fsnotify_mark *mark, __u32 mask, void *data, int data_type) { - bool send; - - mask = (mask & ~FS_EVENT_ON_CHILD); - send = (mark->mask & mask); - - return send; + return true; } /* Update watch data in audit rules based on fsnotify events. */ -- cgit v1.2.3 From 43709a288ed03aa0e2979ab63dd089b3889645c4 Mon Sep 17 00:00:00 2001 From: Eric Paris Date: Wed, 28 Jul 2010 10:18:39 -0400 Subject: fsnotify: remove group->mask group->mask is now useless. It was originally a shortcut for fsnotify to save on performance. These checks are now redundant, so we remove them. Signed-off-by: Eric Paris --- kernel/audit_watch.c | 8 -------- 1 file changed, 8 deletions(-) (limited to 'kernel') diff --git a/kernel/audit_watch.c b/kernel/audit_watch.c index 097a61c65fe0..1b87e757845d 100644 --- a/kernel/audit_watch.c +++ b/kernel/audit_watch.c @@ -164,8 +164,6 @@ static struct audit_parent *audit_init_parent(struct nameidata *ndp) return ERR_PTR(ret); } - fsnotify_recalc_group_mask(audit_watch_group); - return parent; } @@ -352,9 +350,6 @@ static void audit_remove_parent_watches(struct audit_parent *parent) mutex_unlock(&audit_filter_mutex); fsnotify_destroy_mark(&parent->mark); - - fsnotify_recalc_group_mask(audit_watch_group); - } /* Get path information necessary for adding watches. */ @@ -505,9 +500,6 @@ void audit_remove_watch_rule(struct audit_krule *krule) audit_put_parent(parent); } } - - fsnotify_recalc_group_mask(audit_watch_group); - } static bool audit_watch_should_send_event(struct fsnotify_group *group, struct inode *inode, -- cgit v1.2.3 From ce8f76fb7320297ccbe7c950fd9a2d727dd6a5a0 Mon Sep 17 00:00:00 2001 From: Eric Paris Date: Wed, 28 Jul 2010 10:18:39 -0400 Subject: fsnotify: pass both the vfsmount mark and inode mark should_send_event() and handle_event() will both need to look up the inode event if they get a vfsmount event. Lets just pass both at the same time since we have them both after walking the lists in lockstep. Signed-off-by: Eric Paris --- kernel/audit_tree.c | 6 ++++-- kernel/audit_watch.c | 8 +++++--- 2 files changed, 9 insertions(+), 5 deletions(-) (limited to 'kernel') diff --git a/kernel/audit_tree.c b/kernel/audit_tree.c index 2abb99f3459d..781ab7f4e35c 100644 --- a/kernel/audit_tree.c +++ b/kernel/audit_tree.c @@ -904,7 +904,8 @@ static void evict_chunk(struct audit_chunk *chunk) } static int audit_tree_handle_event(struct fsnotify_group *group, - struct fsnotify_mark *mark, + struct fsnotify_mark *inode_mark, + struct fsnotify_mark *vfsmonut_mark, struct fsnotify_event *event) { BUG(); @@ -920,7 +921,8 @@ static void audit_tree_freeing_mark(struct fsnotify_mark *entry, struct fsnotify } static bool audit_tree_send_event(struct fsnotify_group *group, struct inode *inode, - struct vfsmount *mnt, struct fsnotify_mark *mark, + struct vfsmount *mnt, struct fsnotify_mark *inode_mark, + struct fsnotify_mark *vfsmount_mark, __u32 mask, void *data, int data_type) { return false; diff --git a/kernel/audit_watch.c b/kernel/audit_watch.c index 1b87e757845d..a273cf340527 100644 --- a/kernel/audit_watch.c +++ b/kernel/audit_watch.c @@ -503,7 +503,8 @@ void audit_remove_watch_rule(struct audit_krule *krule) } static bool audit_watch_should_send_event(struct fsnotify_group *group, struct inode *inode, - struct vfsmount *mnt, struct fsnotify_mark *mark, + struct vfsmount *mnt, struct fsnotify_mark *inode_mark, + struct fsnotify_mark *vfsmount_mark, __u32 mask, void *data, int data_type) { return true; @@ -511,7 +512,8 @@ static bool audit_watch_should_send_event(struct fsnotify_group *group, struct i /* Update watch data in audit rules based on fsnotify events. */ static int audit_watch_handle_event(struct fsnotify_group *group, - struct fsnotify_mark *mark, + struct fsnotify_mark *inode_mark, + struct fsnotify_mark *vfsmount_mark, struct fsnotify_event *event) { struct inode *inode; @@ -519,7 +521,7 @@ static int audit_watch_handle_event(struct fsnotify_group *group, const char *dname = event->file_name; struct audit_parent *parent; - parent = container_of(mark, struct audit_parent, mark); + parent = container_of(inode_mark, struct audit_parent, mark); BUG_ON(group != audit_watch_group); -- cgit v1.2.3 From 1968f5eed54ce47bde488fd9a450912e4a2d7138 Mon Sep 17 00:00:00 2001 From: Eric Paris Date: Wed, 28 Jul 2010 10:18:39 -0400 Subject: fanotify: use both marks when possible fanotify currently, when given a vfsmount_mark will look up (if it exists) the corresponding inode mark. This patch drops that lookup and uses the mark provided. Signed-off-by: Eric Paris --- kernel/audit_tree.c | 2 +- kernel/audit_watch.c | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/audit_tree.c b/kernel/audit_tree.c index 781ab7f4e35c..7f18d3a4527e 100644 --- a/kernel/audit_tree.c +++ b/kernel/audit_tree.c @@ -921,7 +921,7 @@ static void audit_tree_freeing_mark(struct fsnotify_mark *entry, struct fsnotify } static bool audit_tree_send_event(struct fsnotify_group *group, struct inode *inode, - struct vfsmount *mnt, struct fsnotify_mark *inode_mark, + struct fsnotify_mark *inode_mark, struct fsnotify_mark *vfsmount_mark, __u32 mask, void *data, int data_type) { diff --git a/kernel/audit_watch.c b/kernel/audit_watch.c index a273cf340527..6bf2306be7d6 100644 --- a/kernel/audit_watch.c +++ b/kernel/audit_watch.c @@ -503,7 +503,7 @@ void audit_remove_watch_rule(struct audit_krule *krule) } static bool audit_watch_should_send_event(struct fsnotify_group *group, struct inode *inode, - struct vfsmount *mnt, struct fsnotify_mark *inode_mark, + struct fsnotify_mark *inode_mark, struct fsnotify_mark *vfsmount_mark, __u32 mask, void *data, int data_type) { -- cgit v1.2.3 From 575570f02761bd680ba5731c1dfd4701062e7fb2 Mon Sep 17 00:00:00 2001 From: Shaohua Li Date: Tue, 27 Jul 2010 16:06:34 +0800 Subject: tracing: Fix an unallocated memory access in function_graph With CONFIG_DEBUG_PAGEALLOC, I observed an unallocated memory access in function_graph trace. It appears we find a small size entry in ring buffer, but we access it as a big size entry. The access overflows the page size and touches an unallocated page. Cc: Signed-off-by: Shaohua Li LKML-Reference: <1280217994.32400.76.camel@sli10-desk.sh.intel.com> [ Added a comment to explain the problem - SDR ] Signed-off-by: Steven Rostedt --- kernel/trace/trace_functions_graph.c | 10 +++++++++- 1 file changed, 9 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/trace/trace_functions_graph.c b/kernel/trace/trace_functions_graph.c index 79f4bac99a94..b4c179ae4e45 100644 --- a/kernel/trace/trace_functions_graph.c +++ b/kernel/trace/trace_functions_graph.c @@ -507,7 +507,15 @@ get_return_for_leaf(struct trace_iterator *iter, * if the output fails. */ data->ent = *curr; - data->ret = *next; + /* + * If the next event is not a return type, then + * we only care about what type it is. Otherwise we can + * safely copy the entire event. + */ + if (next->ent.type == TRACE_GRAPH_RET) + data->ret = *next; + else + data->ret.ent.type = next->ent.type; } } -- cgit v1.2.3 From 18fab912d4fa70133df164d2dcf3310be0c38c34 Mon Sep 17 00:00:00 2001 From: Huang Ying Date: Wed, 28 Jul 2010 14:14:01 +0800 Subject: tracing: Fix ring_buffer_read_page reading out of page boundary With the configuration: CONFIG_DEBUG_PAGEALLOC=y and Shaohua's patch: [PATCH]x86: make spurious_fault check correct pte bit Function call graph trace with the following will trigger a page fault. # cd /sys/kernel/debug/tracing/ # echo function_graph > current_tracer # cat per_cpu/cpu1/trace_pipe_raw > /dev/null BUG: unable to handle kernel paging request at ffff880006e99000 IP: [] rb_event_length+0x1/0x3f PGD 1b19063 PUD 1b1d063 PMD 3f067 PTE 6e99160 Oops: 0000 [#1] SMP DEBUG_PAGEALLOC last sysfs file: /sys/devices/virtual/net/lo/operstate CPU 1 Modules linked in: Pid: 1982, comm: cat Not tainted 2.6.35-rc6-aes+ #300 /Bochs RIP: 0010:[] [] rb_event_length+0x1/0x3f RSP: 0018:ffff880006475e38 EFLAGS: 00010006 RAX: 0000000000000ff0 RBX: ffff88000786c630 RCX: 000000000000001d RDX: ffff880006e98000 RSI: 0000000000000ff0 RDI: ffff880006e99000 RBP: ffff880006475eb8 R08: 000000145d7008bd R09: 0000000000000000 R10: 0000000000008000 R11: ffffffff815d9336 R12: ffff880006d08000 R13: ffff880006e605d8 R14: 0000000000000000 R15: 0000000000000018 FS: 00007f2b83e456f0(0000) GS:ffff880002100000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b CR2: ffff880006e99000 CR3: 00000000064a8000 CR4: 00000000000006e0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400 Process cat (pid: 1982, threadinfo ffff880006474000, task ffff880006e40770) Stack: ffff880006475eb8 ffffffff8108730f 0000000000000ff0 000000145d7008bd <0> ffff880006e98010 ffff880006d08010 0000000000000296 ffff88000786c640 <0> ffffffff81002956 0000000000000000 ffff8800071f4680 ffff8800071f4680 Call Trace: [] ? ring_buffer_read_page+0x15a/0x24a [] ? return_to_handler+0x15/0x2f [] tracing_buffers_read+0xb9/0x164 [] vfs_read+0xaf/0x150 [] return_to_handler+0x0/0x2f [] __bad_area_nosemaphore+0x17e/0x1a1 [] return_to_handler+0x0/0x2f [] bad_area_nosemaphore+0x13/0x15 Code: 80 25 b2 16 b3 00 fe c9 c3 55 48 89 e5 f0 80 0d a4 16 b3 00 02 c9 c3 55 31 c0 48 89 e5 48 83 3d 94 16 b3 00 01 c9 0f 94 c0 c3 55 <8a> 0f 48 89 e5 83 e1 1f b8 08 00 00 00 0f b6 d1 83 fa 1e 74 27 RIP [] rb_event_length+0x1/0x3f RSP CR2: ffff880006e99000 ---[ end trace a6877bb92ccb36bb ]--- The root cause is that ring_buffer_read_page() may read out of page boundary, because the boundary checking is done after reading. This is fixed via doing boundary checking before reading. Reported-by: Shaohua Li Cc: Signed-off-by: Huang Ying LKML-Reference: <1280297641.2771.307.camel@yhuang-dev> Signed-off-by: Steven Rostedt --- kernel/trace/ring_buffer.c | 3 +++ 1 file changed, 3 insertions(+) (limited to 'kernel') diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c index 1da7b6ea8b85..5ec8f1d1480e 100644 --- a/kernel/trace/ring_buffer.c +++ b/kernel/trace/ring_buffer.c @@ -3868,6 +3868,9 @@ int ring_buffer_read_page(struct ring_buffer *buffer, rpos = reader->read; pos += size; + if (rpos >= commit) + break; + event = rb_reader_event(cpu_buffer); size = rb_event_length(event); } while (len > size); -- cgit v1.2.3 From 33659ebbae262228eef4e0fe990f393d1f0ed941 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Sat, 7 Aug 2010 18:17:56 +0200 Subject: block: remove wrappers for request type/flags Remove all the trivial wrappers for the cmd_type and cmd_flags fields in struct requests. This allows much easier grepping for different request types instead of unwinding through macros. Signed-off-by: Christoph Hellwig Signed-off-by: Jens Axboe --- kernel/trace/blktrace.c | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/blktrace.c b/kernel/trace/blktrace.c index 638711c17504..4f149944cb89 100644 --- a/kernel/trace/blktrace.c +++ b/kernel/trace/blktrace.c @@ -661,10 +661,10 @@ static void blk_add_trace_rq(struct request_queue *q, struct request *rq, if (likely(!bt)) return; - if (blk_discard_rq(rq)) + if (rq->cmd_flags & REQ_DISCARD) rw |= (1 << BIO_RW_DISCARD); - if (blk_pc_request(rq)) { + if (rq->cmd_type == REQ_TYPE_BLOCK_PC) { what |= BLK_TC_ACT(BLK_TC_PC); __blk_add_trace(bt, 0, blk_rq_bytes(rq), rw, what, rq->errors, rq->cmd_len, rq->cmd); @@ -925,7 +925,7 @@ void blk_add_driver_data(struct request_queue *q, if (likely(!bt)) return; - if (blk_pc_request(rq)) + if (rq->cmd_type == REQ_TYPE_BLOCK_PC) __blk_add_trace(bt, 0, blk_rq_bytes(rq), 0, BLK_TA_DRV_DATA, rq->errors, len, data); else @@ -1730,7 +1730,7 @@ void blk_dump_cmd(char *buf, struct request *rq) int len = rq->cmd_len; unsigned char *cmd = rq->cmd; - if (!blk_pc_request(rq)) { + if (rq->cmd_type != REQ_TYPE_BLOCK_PC) { buf[0] = '\0'; return; } @@ -1779,7 +1779,7 @@ void blk_fill_rwbs_rq(char *rwbs, struct request *rq) int rw = rq->cmd_flags & 0x03; int bytes; - if (blk_discard_rq(rq)) + if (rq->cmd_flags & REQ_DISCARD) rw |= (1 << BIO_RW_DISCARD); bytes = blk_rq_bytes(rq); -- cgit v1.2.3 From 7b6d91daee5cac6402186ff224c3af39d79f4a0e Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Sat, 7 Aug 2010 18:20:39 +0200 Subject: block: unify flags for struct bio and struct request Remove the current bio flags and reuse the request flags for the bio, too. This allows to more easily trace the type of I/O from the filesystem down to the block driver. There were two flags in the bio that were missing in the requests: BIO_RW_UNPLUG and BIO_RW_AHEAD. Also I've renamed two request flags that had a superflous RW in them. Note that the flags are in bio.h despite having the REQ_ name - as blkdev.h includes bio.h that is the only way to go for now. Signed-off-by: Christoph Hellwig Signed-off-by: Jens Axboe --- kernel/power/block_io.c | 2 +- kernel/trace/blktrace.c | 27 +++++++++++++++------------ 2 files changed, 16 insertions(+), 13 deletions(-) (limited to 'kernel') diff --git a/kernel/power/block_io.c b/kernel/power/block_io.c index 97024fd40cd5..83bbc7c02df9 100644 --- a/kernel/power/block_io.c +++ b/kernel/power/block_io.c @@ -28,7 +28,7 @@ static int submit(int rw, struct block_device *bdev, sector_t sector, struct page *page, struct bio **bio_chain) { - const int bio_rw = rw | (1 << BIO_RW_SYNCIO) | (1 << BIO_RW_UNPLUG); + const int bio_rw = rw | REQ_SYNC | REQ_UNPLUG; struct bio *bio; bio = bio_alloc(__GFP_WAIT | __GFP_HIGH, 1); diff --git a/kernel/trace/blktrace.c b/kernel/trace/blktrace.c index 4f149944cb89..3b4a695051b6 100644 --- a/kernel/trace/blktrace.c +++ b/kernel/trace/blktrace.c @@ -169,9 +169,12 @@ static int act_log_check(struct blk_trace *bt, u32 what, sector_t sector, static const u32 ddir_act[2] = { BLK_TC_ACT(BLK_TC_READ), BLK_TC_ACT(BLK_TC_WRITE) }; +#define BLK_TC_HARDBARRIER BLK_TC_BARRIER +#define BLK_TC_RAHEAD BLK_TC_AHEAD + /* The ilog2() calls fall out because they're constant */ -#define MASK_TC_BIT(rw, __name) ((rw & (1 << BIO_RW_ ## __name)) << \ - (ilog2(BLK_TC_ ## __name) + BLK_TC_SHIFT - BIO_RW_ ## __name)) +#define MASK_TC_BIT(rw, __name) ((rw & REQ_ ## __name) << \ + (ilog2(BLK_TC_ ## __name) + BLK_TC_SHIFT - __REQ_ ## __name)) /* * The worker for the various blk_add_trace*() types. Fills out a @@ -194,9 +197,9 @@ static void __blk_add_trace(struct blk_trace *bt, sector_t sector, int bytes, return; what |= ddir_act[rw & WRITE]; - what |= MASK_TC_BIT(rw, BARRIER); - what |= MASK_TC_BIT(rw, SYNCIO); - what |= MASK_TC_BIT(rw, AHEAD); + what |= MASK_TC_BIT(rw, HARDBARRIER); + what |= MASK_TC_BIT(rw, SYNC); + what |= MASK_TC_BIT(rw, RAHEAD); what |= MASK_TC_BIT(rw, META); what |= MASK_TC_BIT(rw, DISCARD); @@ -662,7 +665,7 @@ static void blk_add_trace_rq(struct request_queue *q, struct request *rq, return; if (rq->cmd_flags & REQ_DISCARD) - rw |= (1 << BIO_RW_DISCARD); + rw |= REQ_DISCARD; if (rq->cmd_type == REQ_TYPE_BLOCK_PC) { what |= BLK_TC_ACT(BLK_TC_PC); @@ -1755,20 +1758,20 @@ void blk_fill_rwbs(char *rwbs, u32 rw, int bytes) if (rw & WRITE) rwbs[i++] = 'W'; - else if (rw & 1 << BIO_RW_DISCARD) + else if (rw & REQ_DISCARD) rwbs[i++] = 'D'; else if (bytes) rwbs[i++] = 'R'; else rwbs[i++] = 'N'; - if (rw & 1 << BIO_RW_AHEAD) + if (rw & REQ_RAHEAD) rwbs[i++] = 'A'; - if (rw & 1 << BIO_RW_BARRIER) + if (rw & REQ_HARDBARRIER) rwbs[i++] = 'B'; - if (rw & 1 << BIO_RW_SYNCIO) + if (rw & REQ_SYNC) rwbs[i++] = 'S'; - if (rw & 1 << BIO_RW_META) + if (rw & REQ_META) rwbs[i++] = 'M'; rwbs[i] = '\0'; @@ -1780,7 +1783,7 @@ void blk_fill_rwbs_rq(char *rwbs, struct request *rq) int bytes; if (rq->cmd_flags & REQ_DISCARD) - rw |= (1 << BIO_RW_DISCARD); + rw |= REQ_DISCARD; bytes = blk_rq_bytes(rq); -- cgit v1.2.3 From 62c2a7d969f30163f733c81158254b3095b23e72 Mon Sep 17 00:00:00 2001 From: Arnd Bergmann Date: Wed, 7 Jul 2010 16:51:26 +0200 Subject: block: push BKL into blktrace ioctls The blktrace driver currently needs the BKL, but we should not need to take that in the block layer, so just push it down into the driver itself. It is quite likely that the BKL is not actually required in blktrace code and could be removed in a follow-on patch. Signed-off-by: Arnd Bergmann Acked-by: Christoph Hellwig Signed-off-by: Jens Axboe --- kernel/trace/blktrace.c | 43 +++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 43 insertions(+) (limited to 'kernel') diff --git a/kernel/trace/blktrace.c b/kernel/trace/blktrace.c index 3b4a695051b6..82499a5bdcb7 100644 --- a/kernel/trace/blktrace.c +++ b/kernel/trace/blktrace.c @@ -552,6 +552,41 @@ int blk_trace_setup(struct request_queue *q, char *name, dev_t dev, } EXPORT_SYMBOL_GPL(blk_trace_setup); +#if defined(CONFIG_COMPAT) && defined(CONFIG_X86_64) +static int compat_blk_trace_setup(struct request_queue *q, char *name, + dev_t dev, struct block_device *bdev, + char __user *arg) +{ + struct blk_user_trace_setup buts; + struct compat_blk_user_trace_setup cbuts; + int ret; + + if (copy_from_user(&cbuts, arg, sizeof(cbuts))) + return -EFAULT; + + buts = (struct blk_user_trace_setup) { + .act_mask = cbuts.act_mask, + .buf_size = cbuts.buf_size, + .buf_nr = cbuts.buf_nr, + .start_lba = cbuts.start_lba, + .end_lba = cbuts.end_lba, + .pid = cbuts.pid, + }; + memcpy(&buts.name, &cbuts.name, 32); + + ret = do_blk_trace_setup(q, name, dev, bdev, &buts); + if (ret) + return ret; + + if (copy_to_user(arg, &buts.name, 32)) { + blk_trace_remove(q); + return -EFAULT; + } + + return 0; +} +#endif + int blk_trace_startstop(struct request_queue *q, int start) { int ret; @@ -604,6 +639,7 @@ int blk_trace_ioctl(struct block_device *bdev, unsigned cmd, char __user *arg) if (!q) return -ENXIO; + lock_kernel(); mutex_lock(&bdev->bd_mutex); switch (cmd) { @@ -611,6 +647,12 @@ int blk_trace_ioctl(struct block_device *bdev, unsigned cmd, char __user *arg) bdevname(bdev, b); ret = blk_trace_setup(q, b, bdev->bd_dev, bdev, arg); break; +#if defined(CONFIG_COMPAT) && defined(CONFIG_X86_64) + case BLKTRACESETUP32: + bdevname(bdev, b); + ret = compat_blk_trace_setup(q, b, bdev->bd_dev, bdev, arg); + break; +#endif case BLKTRACESTART: start = 1; case BLKTRACESTOP: @@ -625,6 +667,7 @@ int blk_trace_ioctl(struct block_device *bdev, unsigned cmd, char __user *arg) } mutex_unlock(&bdev->bd_mutex); + unlock_kernel(); return ret; } -- cgit v1.2.3 From 426d31071ac476ea62c62656b242930c17b58c00 Mon Sep 17 00:00:00 2001 From: Paul Bolle Date: Sat, 7 Aug 2010 12:30:03 +0200 Subject: fix printk typo 'faild' Signed-off-by: Paul Bolle Signed-off-by: Jiri Kosina --- kernel/trace/trace_kprobe.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/trace/trace_kprobe.c b/kernel/trace/trace_kprobe.c index f52b5f50299d..58716e73e2a2 100644 --- a/kernel/trace/trace_kprobe.c +++ b/kernel/trace/trace_kprobe.c @@ -490,7 +490,7 @@ static int register_trace_probe(struct trace_probe *tp) } ret = register_probe_event(tp); if (ret) { - pr_warning("Faild to register probe event(%d)\n", ret); + pr_warning("Failed to register probe event(%d)\n", ret); goto end; } -- cgit v1.2.3 From ebabe9a9001af0af56c0c2780ca1576246e7a74b Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Wed, 7 Jul 2010 18:53:11 +0200 Subject: pass a struct path to vfs_statfs We'll need the path to implement the flags field for statvfs support. We do have it available in all callers except: - ecryptfs_statfs. This one doesn't actually need vfs_statfs but just needs to do a caller to the lower filesystem statfs method. - sys_ustat. Add a non-exported statfs_by_dentry helper for it which doesn't won't be able to fill out the flags field later on. In addition rename the helpers for statfs vs fstatfs to do_*statfs instead of the misleading vfs prefix. Signed-off-by: Christoph Hellwig Signed-off-by: Al Viro --- kernel/acct.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/acct.c b/kernel/acct.c index 385b88461c29..fa7eb3de2ddc 100644 --- a/kernel/acct.c +++ b/kernel/acct.c @@ -122,7 +122,7 @@ static int check_free_space(struct bsd_acct_struct *acct, struct file *file) spin_unlock(&acct_lock); /* May block */ - if (vfs_statfs(file->f_path.dentry, &sbuf)) + if (vfs_statfs(&file->f_path, &sbuf)) return res; suspend = sbuf.f_blocks * SUSPEND; resume = sbuf.f_blocks * RESUME; -- cgit v1.2.3 From 0caa621065b2cc05d4e53655a34fd989f500b040 Mon Sep 17 00:00:00 2001 From: Randy Dunlap Date: Mon, 9 Aug 2010 16:32:50 -0700 Subject: kernel/timer.c: fix kernel-doc function parameter warning Fix kernel-doc warning, add @timer description: Warning(kernel/timer.c:335): No description found for parameter 'timer' Signed-off-by: Randy Dunlap Cc: Thomas Gleixner Signed-off-by: Linus Torvalds --- kernel/timer.c | 1 + 1 file changed, 1 insertion(+) (limited to 'kernel') diff --git a/kernel/timer.c b/kernel/timer.c index f1b8afe1ad86..97bf05baade7 100644 --- a/kernel/timer.c +++ b/kernel/timer.c @@ -326,6 +326,7 @@ EXPORT_SYMBOL_GPL(round_jiffies_up_relative); /** * set_timer_slack - set the allowed slack for a timer + * @timer: the timer to be modified * @slack_hz: the amount of time (in jiffies) allowed for rounding * * Set the amount of time, in jiffies, that a certain timer has -- cgit v1.2.3 From f7ad3c6be90809b53b7f0ae9d4eaa45ce2564a79 Mon Sep 17 00:00:00 2001 From: Miklos Szeredi Date: Tue, 10 Aug 2010 11:41:36 +0200 Subject: vfs: add helpers to get root and pwd Add three helpers that retrieve a refcounted copy of the root and cwd from the supplied fs_struct. get_fs_root() get_fs_pwd() get_fs_root_and_pwd() Signed-off-by: Miklos Szeredi Signed-off-by: Al Viro --- kernel/auditsc.c | 9 ++------- 1 file changed, 2 insertions(+), 7 deletions(-) (limited to 'kernel') diff --git a/kernel/auditsc.c b/kernel/auditsc.c index b87a63beb66c..1b31c130d034 100644 --- a/kernel/auditsc.c +++ b/kernel/auditsc.c @@ -1835,13 +1835,8 @@ void __audit_getname(const char *name) context->names[context->name_count].ino = (unsigned long)-1; context->names[context->name_count].osid = 0; ++context->name_count; - if (!context->pwd.dentry) { - read_lock(¤t->fs->lock); - context->pwd = current->fs->pwd; - path_get(¤t->fs->pwd); - read_unlock(¤t->fs->lock); - } - + if (!context->pwd.dentry) + get_fs_pwd(current->fs, &context->pwd); } /* audit_putname - intercept a putname request -- cgit v1.2.3 From 2e9fb9953df91ef6310da22182ca8f4496907502 Mon Sep 17 00:00:00 2001 From: Rusty Russell Date: Wed, 11 Aug 2010 23:04:10 -0600 Subject: params: don't hand NULL values to param.set callbacks. MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit An audit by Dongdong Deng revealed that most driver-author-written param calls don't handle val == NULL (which happens when parameters are specified with no =, eg "foo" instead of "foo=1"). The only real case to use this is boolean, so handle it specially for that case and remove a source of bugs for everyone else. Signed-off-by: Rusty Russell Cc: Dongdong Deng Cc: Américo Wang --- kernel/params.c | 20 +++----------------- 1 file changed, 3 insertions(+), 17 deletions(-) (limited to 'kernel') diff --git a/kernel/params.c b/kernel/params.c index 0b30ecd53a52..3c4a9f1b095e 100644 --- a/kernel/params.c +++ b/kernel/params.c @@ -58,6 +58,9 @@ static int parse_one(char *param, /* Find parameter */ for (i = 0; i < num_params; i++) { if (parameq(param, params[i].name)) { + /* Noone handled NULL, so do it here. */ + if (!val && params[i].set != param_set_bool) + return -EINVAL; DEBUGP("They are equal! Calling %p\n", params[i].set); return params[i].set(val, ¶ms[i]); @@ -181,7 +184,6 @@ int parse_args(const char *name, tmptype l; \ int ret; \ \ - if (!val) return -EINVAL; \ ret = strtolfn(val, 0, &l); \ if (ret == -EINVAL || ((type)l != l)) \ return -EINVAL; \ @@ -203,12 +205,6 @@ STANDARD_PARAM_DEF(ulong, unsigned long, "%lu", unsigned long, strict_strtoul); int param_set_charp(const char *val, struct kernel_param *kp) { - if (!val) { - printk(KERN_ERR "%s: string parameter expected\n", - kp->name); - return -EINVAL; - } - if (strlen(val) > 1024) { printk(KERN_ERR "%s: string parameter too long\n", kp->name); @@ -309,12 +305,6 @@ static int param_array(const char *name, kp.arg = elem; kp.flags = flags; - /* No equals sign? */ - if (!val) { - printk(KERN_ERR "%s: expects arguments\n", name); - return -EINVAL; - } - *num = 0; /* We expect a comma-separated list of values. */ do { @@ -381,10 +371,6 @@ int param_set_copystring(const char *val, struct kernel_param *kp) { const struct kparam_string *kps = kp->str; - if (!val) { - printk(KERN_ERR "%s: missing param set value\n", kp->name); - return -EINVAL; - } if (strlen(val)+1 > kps->maxlen) { printk(KERN_ERR "%s: string doesn't fit in %u chars.\n", kp->name, kps->maxlen-1); -- cgit v1.2.3 From a14fe249a8f74269c9e636bcbaa78f5bdb354ce3 Mon Sep 17 00:00:00 2001 From: Rusty Russell Date: Wed, 11 Aug 2010 23:04:11 -0600 Subject: param: move the EXPORT_SYMBOL to after the definitions. This is modern style, and good to do before we start changing things. Signed-off-by: Rusty Russell Reviewed-by: Takashi Iwai Tested-by: Phil Carmody --- kernel/params.c | 39 +++++++++++++-------------------------- 1 file changed, 13 insertions(+), 26 deletions(-) (limited to 'kernel') diff --git a/kernel/params.c b/kernel/params.c index 3c4a9f1b095e..3e78fdb445e7 100644 --- a/kernel/params.c +++ b/kernel/params.c @@ -193,7 +193,9 @@ int parse_args(const char *name, int param_get_##name(char *buffer, struct kernel_param *kp) \ { \ return sprintf(buffer, format, *((type *)kp->arg)); \ - } + } \ + EXPORT_SYMBOL(param_set_##name); \ + EXPORT_SYMBOL(param_get_##name) STANDARD_PARAM_DEF(byte, unsigned char, "%c", unsigned long, strict_strtoul); STANDARD_PARAM_DEF(short, short, "%hi", long, strict_strtol); @@ -222,11 +224,13 @@ int param_set_charp(const char *val, struct kernel_param *kp) return 0; } +EXPORT_SYMBOL(param_set_charp); int param_get_charp(char *buffer, struct kernel_param *kp) { return sprintf(buffer, "%s", *((char **)kp->arg)); } +EXPORT_SYMBOL(param_get_charp); /* Actually could be a bool or an int, for historical reasons. */ int param_set_bool(const char *val, struct kernel_param *kp) @@ -254,6 +258,7 @@ int param_set_bool(const char *val, struct kernel_param *kp) *(int *)kp->arg = v; return 0; } +EXPORT_SYMBOL(param_set_bool); int param_get_bool(char *buffer, struct kernel_param *kp) { @@ -266,6 +271,7 @@ int param_get_bool(char *buffer, struct kernel_param *kp) /* Y and N chosen as being relatively non-coder friendly */ return sprintf(buffer, "%c", val ? 'Y' : 'N'); } +EXPORT_SYMBOL(param_get_bool); /* This one must be bool. */ int param_set_invbool(const char *val, struct kernel_param *kp) @@ -281,11 +287,13 @@ int param_set_invbool(const char *val, struct kernel_param *kp) *(bool *)kp->arg = !boolval; return ret; } +EXPORT_SYMBOL(param_set_invbool); int param_get_invbool(char *buffer, struct kernel_param *kp) { return sprintf(buffer, "%c", (*(bool *)kp->arg) ? 'N' : 'Y'); } +EXPORT_SYMBOL(param_get_invbool); /* We break the rule and mangle the string. */ static int param_array(const char *name, @@ -346,6 +354,7 @@ int param_array_set(const char *val, struct kernel_param *kp) arr->elemsize, arr->set, kp->flags, arr->num ?: &temp_num); } +EXPORT_SYMBOL(param_array_set); int param_array_get(char *buffer, struct kernel_param *kp) { @@ -366,6 +375,7 @@ int param_array_get(char *buffer, struct kernel_param *kp) buffer[off] = '\0'; return off; } +EXPORT_SYMBOL(param_array_get); int param_set_copystring(const char *val, struct kernel_param *kp) { @@ -379,12 +389,14 @@ int param_set_copystring(const char *val, struct kernel_param *kp) strcpy(kps->string, val); return 0; } +EXPORT_SYMBOL(param_set_copystring); int param_get_string(char *buffer, struct kernel_param *kp) { const struct kparam_string *kps = kp->str; return strlcpy(buffer, kps->string, kps->maxlen); } +EXPORT_SYMBOL(param_get_string); /* sysfs output in /sys/modules/XYZ/parameters/ */ #define to_module_attr(n) container_of(n, struct module_attribute, attr) @@ -754,28 +766,3 @@ static int __init param_sysfs_init(void) subsys_initcall(param_sysfs_init); #endif /* CONFIG_SYSFS */ - -EXPORT_SYMBOL(param_set_byte); -EXPORT_SYMBOL(param_get_byte); -EXPORT_SYMBOL(param_set_short); -EXPORT_SYMBOL(param_get_short); -EXPORT_SYMBOL(param_set_ushort); -EXPORT_SYMBOL(param_get_ushort); -EXPORT_SYMBOL(param_set_int); -EXPORT_SYMBOL(param_get_int); -EXPORT_SYMBOL(param_set_uint); -EXPORT_SYMBOL(param_get_uint); -EXPORT_SYMBOL(param_set_long); -EXPORT_SYMBOL(param_get_long); -EXPORT_SYMBOL(param_set_ulong); -EXPORT_SYMBOL(param_get_ulong); -EXPORT_SYMBOL(param_set_charp); -EXPORT_SYMBOL(param_get_charp); -EXPORT_SYMBOL(param_set_bool); -EXPORT_SYMBOL(param_get_bool); -EXPORT_SYMBOL(param_set_invbool); -EXPORT_SYMBOL(param_get_invbool); -EXPORT_SYMBOL(param_array_set); -EXPORT_SYMBOL(param_array_get); -EXPORT_SYMBOL(param_set_copystring); -EXPORT_SYMBOL(param_get_string); -- cgit v1.2.3 From 9bbb9e5a33109b2832e2e63dcc7a132924ab374b Mon Sep 17 00:00:00 2001 From: Rusty Russell Date: Wed, 11 Aug 2010 23:04:12 -0600 Subject: param: use ops in struct kernel_param, rather than get and set fns directly This is more kernel-ish, saves some space, and also allows us to expand the ops without breaking all the callers who are happy for the new members to be NULL. The few places which defined their own param types are changed to the new scheme (more which crept in recently fixed in following patches). Since we're touching them anyway, we change get() and set() to take a const struct kernel_param (which they really are). This causes some harmless warnings until we fix them (in following patches). To reduce churn, module_param_call creates the ops struct so the callers don't have to change (and casts the functions to reduce warnings). The modern version which takes an ops struct is called module_param_cb. Signed-off-by: Rusty Russell Reviewed-by: Takashi Iwai Tested-by: Phil Carmody Cc: "David S. Miller" Cc: Ville Syrjala Cc: Dmitry Torokhov Cc: Alessandro Rubini Cc: Michal Januszewski Cc: Trond Myklebust Cc: "J. Bruce Fields" Cc: Neil Brown Cc: linux-kernel@vger.kernel.org Cc: linux-input@vger.kernel.org Cc: linux-fbdev-devel@lists.sourceforge.net Cc: linux-nfs@vger.kernel.org Cc: netdev@vger.kernel.org --- kernel/params.c | 90 +++++++++++++++++++++++++++++++++++++++------------------ 1 file changed, 62 insertions(+), 28 deletions(-) (limited to 'kernel') diff --git a/kernel/params.c b/kernel/params.c index 3e78fdb445e7..a550698ae02d 100644 --- a/kernel/params.c +++ b/kernel/params.c @@ -59,11 +59,11 @@ static int parse_one(char *param, for (i = 0; i < num_params; i++) { if (parameq(param, params[i].name)) { /* Noone handled NULL, so do it here. */ - if (!val && params[i].set != param_set_bool) + if (!val && params[i].ops->set != param_set_bool) return -EINVAL; DEBUGP("They are equal! Calling %p\n", - params[i].set); - return params[i].set(val, ¶ms[i]); + params[i].ops->set); + return params[i].ops->set(val, ¶ms[i]); } } @@ -179,7 +179,7 @@ int parse_args(const char *name, /* Lazy bastard, eh? */ #define STANDARD_PARAM_DEF(name, type, format, tmptype, strtolfn) \ - int param_set_##name(const char *val, struct kernel_param *kp) \ + int param_set_##name(const char *val, const struct kernel_param *kp) \ { \ tmptype l; \ int ret; \ @@ -190,12 +190,18 @@ int parse_args(const char *name, *((type *)kp->arg) = l; \ return 0; \ } \ - int param_get_##name(char *buffer, struct kernel_param *kp) \ + int param_get_##name(char *buffer, const struct kernel_param *kp) \ { \ return sprintf(buffer, format, *((type *)kp->arg)); \ } \ + struct kernel_param_ops param_ops_##name = { \ + .set = param_set_##name, \ + .get = param_get_##name, \ + }; \ EXPORT_SYMBOL(param_set_##name); \ - EXPORT_SYMBOL(param_get_##name) + EXPORT_SYMBOL(param_get_##name); \ + EXPORT_SYMBOL(param_ops_##name) + STANDARD_PARAM_DEF(byte, unsigned char, "%c", unsigned long, strict_strtoul); STANDARD_PARAM_DEF(short, short, "%hi", long, strict_strtol); @@ -205,7 +211,7 @@ STANDARD_PARAM_DEF(uint, unsigned int, "%u", unsigned long, strict_strtoul); STANDARD_PARAM_DEF(long, long, "%li", long, strict_strtol); STANDARD_PARAM_DEF(ulong, unsigned long, "%lu", unsigned long, strict_strtoul); -int param_set_charp(const char *val, struct kernel_param *kp) +int param_set_charp(const char *val, const struct kernel_param *kp) { if (strlen(val) > 1024) { printk(KERN_ERR "%s: string parameter too long\n", @@ -226,14 +232,20 @@ int param_set_charp(const char *val, struct kernel_param *kp) } EXPORT_SYMBOL(param_set_charp); -int param_get_charp(char *buffer, struct kernel_param *kp) +int param_get_charp(char *buffer, const struct kernel_param *kp) { return sprintf(buffer, "%s", *((char **)kp->arg)); } EXPORT_SYMBOL(param_get_charp); +struct kernel_param_ops param_ops_charp = { + .set = param_set_charp, + .get = param_get_charp, +}; +EXPORT_SYMBOL(param_ops_charp); + /* Actually could be a bool or an int, for historical reasons. */ -int param_set_bool(const char *val, struct kernel_param *kp) +int param_set_bool(const char *val, const struct kernel_param *kp) { bool v; @@ -260,7 +272,7 @@ int param_set_bool(const char *val, struct kernel_param *kp) } EXPORT_SYMBOL(param_set_bool); -int param_get_bool(char *buffer, struct kernel_param *kp) +int param_get_bool(char *buffer, const struct kernel_param *kp) { bool val; if (kp->flags & KPARAM_ISBOOL) @@ -273,8 +285,14 @@ int param_get_bool(char *buffer, struct kernel_param *kp) } EXPORT_SYMBOL(param_get_bool); +struct kernel_param_ops param_ops_bool = { + .set = param_set_bool, + .get = param_get_bool, +}; +EXPORT_SYMBOL(param_ops_bool); + /* This one must be bool. */ -int param_set_invbool(const char *val, struct kernel_param *kp) +int param_set_invbool(const char *val, const struct kernel_param *kp) { int ret; bool boolval; @@ -289,18 +307,24 @@ int param_set_invbool(const char *val, struct kernel_param *kp) } EXPORT_SYMBOL(param_set_invbool); -int param_get_invbool(char *buffer, struct kernel_param *kp) +int param_get_invbool(char *buffer, const struct kernel_param *kp) { return sprintf(buffer, "%c", (*(bool *)kp->arg) ? 'N' : 'Y'); } EXPORT_SYMBOL(param_get_invbool); +struct kernel_param_ops param_ops_invbool = { + .set = param_set_invbool, + .get = param_get_invbool, +}; +EXPORT_SYMBOL(param_ops_invbool); + /* We break the rule and mangle the string. */ static int param_array(const char *name, const char *val, unsigned int min, unsigned int max, void *elem, int elemsize, - int (*set)(const char *, struct kernel_param *kp), + int (*set)(const char *, const struct kernel_param *kp), u16 flags, unsigned int *num) { @@ -345,18 +369,17 @@ static int param_array(const char *name, return 0; } -int param_array_set(const char *val, struct kernel_param *kp) +static int param_array_set(const char *val, const struct kernel_param *kp) { const struct kparam_array *arr = kp->arr; unsigned int temp_num; return param_array(kp->name, val, 1, arr->max, arr->elem, - arr->elemsize, arr->set, kp->flags, + arr->elemsize, arr->ops->set, kp->flags, arr->num ?: &temp_num); } -EXPORT_SYMBOL(param_array_set); -int param_array_get(char *buffer, struct kernel_param *kp) +static int param_array_get(char *buffer, const struct kernel_param *kp) { int i, off, ret; const struct kparam_array *arr = kp->arr; @@ -367,7 +390,7 @@ int param_array_get(char *buffer, struct kernel_param *kp) if (i) buffer[off++] = ','; p.arg = arr->elem + arr->elemsize * i; - ret = arr->get(buffer + off, &p); + ret = arr->ops->get(buffer + off, &p); if (ret < 0) return ret; off += ret; @@ -375,9 +398,14 @@ int param_array_get(char *buffer, struct kernel_param *kp) buffer[off] = '\0'; return off; } -EXPORT_SYMBOL(param_array_get); -int param_set_copystring(const char *val, struct kernel_param *kp) +struct kernel_param_ops param_array_ops = { + .set = param_array_set, + .get = param_array_get, +}; +EXPORT_SYMBOL(param_array_ops); + +int param_set_copystring(const char *val, const struct kernel_param *kp) { const struct kparam_string *kps = kp->str; @@ -391,13 +419,19 @@ int param_set_copystring(const char *val, struct kernel_param *kp) } EXPORT_SYMBOL(param_set_copystring); -int param_get_string(char *buffer, struct kernel_param *kp) +int param_get_string(char *buffer, const struct kernel_param *kp) { const struct kparam_string *kps = kp->str; return strlcpy(buffer, kps->string, kps->maxlen); } EXPORT_SYMBOL(param_get_string); +struct kernel_param_ops param_ops_string = { + .set = param_set_copystring, + .get = param_get_string, +}; +EXPORT_SYMBOL(param_ops_string); + /* sysfs output in /sys/modules/XYZ/parameters/ */ #define to_module_attr(n) container_of(n, struct module_attribute, attr) #define to_module_kobject(n) container_of(n, struct module_kobject, kobj) @@ -407,7 +441,7 @@ extern struct kernel_param __start___param[], __stop___param[]; struct param_attribute { struct module_attribute mattr; - struct kernel_param *param; + const struct kernel_param *param; }; struct module_param_attrs @@ -426,10 +460,10 @@ static ssize_t param_attr_show(struct module_attribute *mattr, int count; struct param_attribute *attribute = to_param_attr(mattr); - if (!attribute->param->get) + if (!attribute->param->ops->get) return -EPERM; - count = attribute->param->get(buf, attribute->param); + count = attribute->param->ops->get(buf, attribute->param); if (count > 0) { strcat(buf, "\n"); ++count; @@ -445,10 +479,10 @@ static ssize_t param_attr_store(struct module_attribute *mattr, int err; struct param_attribute *attribute = to_param_attr(mattr); - if (!attribute->param->set) + if (!attribute->param->ops->set) return -EPERM; - err = attribute->param->set(buf, attribute->param); + err = attribute->param->ops->set(buf, attribute->param); if (!err) return len; return err; @@ -473,7 +507,7 @@ static ssize_t param_attr_store(struct module_attribute *mattr, * if there's an error. */ static __modinit int add_sysfs_param(struct module_kobject *mk, - struct kernel_param *kp, + const struct kernel_param *kp, const char *name) { struct module_param_attrs *new; @@ -555,7 +589,7 @@ static void free_module_param_attrs(struct module_kobject *mk) * /sys/module/[mod->name]/parameters/ */ int module_param_sysfs_setup(struct module *mod, - struct kernel_param *kparam, + const struct kernel_param *kparam, unsigned int num_params) { int i, err; -- cgit v1.2.3 From e6df34a4429b77fdffb6e05adf263468a3dcda33 Mon Sep 17 00:00:00 2001 From: Rusty Russell Date: Wed, 11 Aug 2010 23:04:17 -0600 Subject: param: add a free hook to kernel_param_ops. This allows us to generalize the KPARAM_KMALLOCED flag, by calling a function on every parameter when a module is unloaded. Signed-off-by: Rusty Russell Reviewed-by: Takashi Iwai Tested-by: Phil Carmody --- kernel/params.c | 17 ++++++++++++++++- 1 file changed, 16 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/params.c b/kernel/params.c index a550698ae02d..458a09b886c4 100644 --- a/kernel/params.c +++ b/kernel/params.c @@ -399,9 +399,20 @@ static int param_array_get(char *buffer, const struct kernel_param *kp) return off; } +static void param_array_free(void *arg) +{ + unsigned int i; + const struct kparam_array *arr = arg; + + if (arr->ops->free) + for (i = 0; i < (arr->num ? *arr->num : arr->max); i++) + arr->ops->free(arr->elem + arr->elemsize * i); +} + struct kernel_param_ops param_array_ops = { .set = param_array_set, .get = param_array_get, + .free = param_array_free, }; EXPORT_SYMBOL(param_array_ops); @@ -634,7 +645,11 @@ void module_param_sysfs_remove(struct module *mod) void destroy_params(const struct kernel_param *params, unsigned num) { - /* FIXME: This should free kmalloced charp parameters. It doesn't. */ + unsigned int i; + + for (i = 0; i < num; i++) + if (params[i].ops->free) + params[i].ops->free(params[i].arg); } static void __init kernel_add_sysfs_param(const char *name, -- cgit v1.2.3 From a1054322afc8120ea5a50bc84e5beeda54571862 Mon Sep 17 00:00:00 2001 From: Rusty Russell Date: Wed, 11 Aug 2010 23:04:18 -0600 Subject: param: use free hook for charp (fix leak of charp parameters) Instead of using a "I kmalloced this" flag, we keep track of the kmalloced strings and use that list to check if we need to kfree (in practice, the list is very short). This means that kparams can be const again, and plugs a leak. This is important for drivers/usb/gadget/nokia.c which gets modprobe/rmmod'ed frequently on the N9000. Signed-off-by: Rusty Russell Reviewed-by: Takashi Iwai Cc: Artem Bityutskiy Tested-by: Phil Carmody --- kernel/params.c | 52 ++++++++++++++++++++++++++++++++++++++++++++++++++-- 1 file changed, 50 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/params.c b/kernel/params.c index 458a09b886c4..ef60db14fae0 100644 --- a/kernel/params.c +++ b/kernel/params.c @@ -31,6 +31,45 @@ #define DEBUGP(fmt, a...) #endif +/* This just allows us to keep track of which parameters are kmalloced. */ +struct kmalloced_param { + struct list_head list; + char val[]; +}; +static DEFINE_MUTEX(param_lock); +static LIST_HEAD(kmalloced_params); + +static void *kmalloc_parameter(unsigned int size) +{ + struct kmalloced_param *p; + + p = kmalloc(sizeof(*p) + size, GFP_KERNEL); + if (!p) + return NULL; + + mutex_lock(¶m_lock); + list_add(&p->list, &kmalloced_params); + mutex_unlock(¶m_lock); + + return p->val; +} + +/* Does nothing if parameter wasn't kmalloced above. */ +static void maybe_kfree_parameter(void *param) +{ + struct kmalloced_param *p; + + mutex_lock(¶m_lock); + list_for_each_entry(p, &kmalloced_params, list) { + if (p->val == param) { + list_del(&p->list); + kfree(p); + break; + } + } + mutex_unlock(¶m_lock); +} + static inline char dash2underscore(char c) { if (c == '-') @@ -219,12 +258,15 @@ int param_set_charp(const char *val, const struct kernel_param *kp) return -ENOSPC; } - /* This is a hack. We can't need to strdup in early boot, and we + maybe_kfree_parameter(*(char **)kp->arg); + + /* This is a hack. We can't kmalloc in early boot, and we * don't need to; this mangled commandline is preserved. */ if (slab_is_available()) { - *(char **)kp->arg = kstrdup(val, GFP_KERNEL); + *(char **)kp->arg = kmalloc_parameter(strlen(val)+1); if (!*(char **)kp->arg) return -ENOMEM; + strcpy(*(char **)kp->arg, val); } else *(const char **)kp->arg = val; @@ -238,9 +280,15 @@ int param_get_charp(char *buffer, const struct kernel_param *kp) } EXPORT_SYMBOL(param_get_charp); +static void param_free_charp(void *arg) +{ + maybe_kfree_parameter(*((char **)arg)); +} + struct kernel_param_ops param_ops_charp = { .set = param_set_charp, .get = param_get_charp, + .free = param_free_charp, }; EXPORT_SYMBOL(param_ops_charp); -- cgit v1.2.3 From 914dcaa84c53f2c3efa6016efcae13fd92a8a17c Mon Sep 17 00:00:00 2001 From: Rusty Russell Date: Wed, 11 Aug 2010 23:04:18 -0600 Subject: param: make param sections const. Since this section can be read-only (they're in .rodata), they should always have been const. Minor flow-through various functions. Signed-off-by: Rusty Russell Tested-by: Phil Carmody --- kernel/params.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/params.c b/kernel/params.c index ef60db14fae0..a3eeeefc9472 100644 --- a/kernel/params.c +++ b/kernel/params.c @@ -88,7 +88,7 @@ static inline int parameq(const char *input, const char *paramname) static int parse_one(char *param, char *val, - struct kernel_param *params, + const struct kernel_param *params, unsigned num_params, int (*handle_unknown)(char *param, char *val)) { @@ -170,7 +170,7 @@ static char *next_arg(char *args, char **param, char **val) /* Args looks like "foo=bar,bar2 baz=fuz wiz". */ int parse_args(const char *name, char *args, - struct kernel_param *params, + const struct kernel_param *params, unsigned num, int (*unknown)(char *param, char *val)) { -- cgit v1.2.3 From 907b29eb41aa604477a655bff7345731da94514d Mon Sep 17 00:00:00 2001 From: Rusty Russell Date: Wed, 11 Aug 2010 23:04:19 -0600 Subject: param: locking for kernel parameters There may be cases (most obviously, sysfs-writable charp parameters) where a module needs to prevent sysfs access to parameters. Rather than express this in terms of a big lock, the functions are expressed in terms of what they protect against. This is clearer, esp. if the implementation changes to a module-level or even param-level lock. Signed-off-by: Rusty Russell Reviewed-by: Takashi Iwai Tested-by: Phil Carmody --- kernel/params.c | 33 ++++++++++++++++++++++++++------- 1 file changed, 26 insertions(+), 7 deletions(-) (limited to 'kernel') diff --git a/kernel/params.c b/kernel/params.c index a3eeeefc9472..08107d181758 100644 --- a/kernel/params.c +++ b/kernel/params.c @@ -31,12 +31,14 @@ #define DEBUGP(fmt, a...) #endif +/* Protects all parameters, and incidentally kmalloced_param list. */ +static DEFINE_MUTEX(param_lock); + /* This just allows us to keep track of which parameters are kmalloced. */ struct kmalloced_param { struct list_head list; char val[]; }; -static DEFINE_MUTEX(param_lock); static LIST_HEAD(kmalloced_params); static void *kmalloc_parameter(unsigned int size) @@ -47,10 +49,7 @@ static void *kmalloc_parameter(unsigned int size) if (!p) return NULL; - mutex_lock(¶m_lock); list_add(&p->list, &kmalloced_params); - mutex_unlock(¶m_lock); - return p->val; } @@ -59,7 +58,6 @@ static void maybe_kfree_parameter(void *param) { struct kmalloced_param *p; - mutex_lock(¶m_lock); list_for_each_entry(p, &kmalloced_params, list) { if (p->val == param) { list_del(&p->list); @@ -67,7 +65,6 @@ static void maybe_kfree_parameter(void *param) break; } } - mutex_unlock(¶m_lock); } static inline char dash2underscore(char c) @@ -93,6 +90,7 @@ static int parse_one(char *param, int (*handle_unknown)(char *param, char *val)) { unsigned int i; + int err; /* Find parameter */ for (i = 0; i < num_params; i++) { @@ -102,7 +100,10 @@ static int parse_one(char *param, return -EINVAL; DEBUGP("They are equal! Calling %p\n", params[i].ops->set); - return params[i].ops->set(val, ¶ms[i]); + mutex_lock(¶m_lock); + err = params[i].ops->set(val, ¶ms[i]); + mutex_unlock(¶m_lock); + return err; } } @@ -400,6 +401,7 @@ static int param_array(const char *name, /* nul-terminate and parse */ save = val[len]; ((char *)val)[len] = '\0'; + BUG_ON(!mutex_is_locked(¶m_lock)); ret = set(val, &kp); if (ret != 0) @@ -438,6 +440,7 @@ static int param_array_get(char *buffer, const struct kernel_param *kp) if (i) buffer[off++] = ','; p.arg = arr->elem + arr->elemsize * i; + BUG_ON(!mutex_is_locked(¶m_lock)); ret = arr->ops->get(buffer + off, &p); if (ret < 0) return ret; @@ -522,7 +525,9 @@ static ssize_t param_attr_show(struct module_attribute *mattr, if (!attribute->param->ops->get) return -EPERM; + mutex_lock(¶m_lock); count = attribute->param->ops->get(buf, attribute->param); + mutex_unlock(¶m_lock); if (count > 0) { strcat(buf, "\n"); ++count; @@ -541,7 +546,9 @@ static ssize_t param_attr_store(struct module_attribute *mattr, if (!attribute->param->ops->set) return -EPERM; + mutex_lock(¶m_lock); err = attribute->param->ops->set(buf, attribute->param); + mutex_unlock(¶m_lock); if (!err) return len; return err; @@ -555,6 +562,18 @@ static ssize_t param_attr_store(struct module_attribute *mattr, #endif #ifdef CONFIG_SYSFS +void __kernel_param_lock(void) +{ + mutex_lock(¶m_lock); +} +EXPORT_SYMBOL(__kernel_param_lock); + +void __kernel_param_unlock(void) +{ + mutex_unlock(¶m_lock); +} +EXPORT_SYMBOL(__kernel_param_unlock); + /* * add_sysfs_param - add a parameter to sysfs * @mk: struct module_kobject -- cgit v1.2.3 From e400c28524af2d344b1663b27bf28984fa959a0e Mon Sep 17 00:00:00 2001 From: Dan Carpenter Date: Tue, 10 Aug 2010 18:02:54 -0700 Subject: cgroups: save space for the terminator The original code didn't leave enough space for a NULL terminator. These strings are copied with strcpy() into fixed length buffers in cgroup_root_from_opts(). Signed-off-by: Dan Carpenter Acked-by: Serge E. Hallyn Reviewd-by: KAMEZAWA Hiroyuki Cc: Paul Menage Cc: Li Zefan Cc: Ben Blum Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/cgroup.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/cgroup.c b/kernel/cgroup.c index d83cab06da87..192f88c5b0f9 100644 --- a/kernel/cgroup.c +++ b/kernel/cgroup.c @@ -1102,7 +1102,7 @@ static int parse_cgroupfs_options(char *data, struct cgroup_sb_opts *opts) if (opts->release_agent) return -EINVAL; opts->release_agent = - kstrndup(token + 14, PATH_MAX, GFP_KERNEL); + kstrndup(token + 14, PATH_MAX - 1, GFP_KERNEL); if (!opts->release_agent) return -ENOMEM; } else if (!strncmp(token, "name=", 5)) { @@ -1123,7 +1123,7 @@ static int parse_cgroupfs_options(char *data, struct cgroup_sb_opts *opts) if (opts->name) return -EINVAL; opts->name = kstrndup(name, - MAX_CGROUP_ROOT_NAMELEN, + MAX_CGROUP_ROOT_NAMELEN - 1, GFP_KERNEL); if (!opts->name) return -ENOMEM; -- cgit v1.2.3 From c7e49c1488ab20342eaaf38f1ca35a207f4c051d Mon Sep 17 00:00:00 2001 From: Oleg Nesterov Date: Tue, 10 Aug 2010 18:03:07 -0700 Subject: ptrace: optimize exit_ptrace() for the likely case exit_ptrace() takes tasklist_lock unconditionally. We need this lock to avoid the race with ptrace_traceme(), it acts as a barrier. Change its caller, forget_original_parent(), to call exit_ptrace() under tasklist_lock. Change exit_ptrace() to drop and reacquire this lock if needed. This allows us to add the fastpath list_empty(ptraced) check. In the likely no-tracees case exit_ptrace() just returns and we avoid the lock() + unlock() sequence. "Zhang, Yanmin" suggested to add this check, and he reports that this change adds about 11% improvement in some tests. Suggested-and-tested-by: "Zhang, Yanmin" Signed-off-by: Oleg Nesterov Acked-by: Roland McGrath Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/exit.c | 7 +++++-- kernel/ptrace.c | 12 +++++++++--- 2 files changed, 14 insertions(+), 5 deletions(-) (limited to 'kernel') diff --git a/kernel/exit.c b/kernel/exit.c index ceffc67b564a..671ed56e0a49 100644 --- a/kernel/exit.c +++ b/kernel/exit.c @@ -771,9 +771,12 @@ static void forget_original_parent(struct task_struct *father) struct task_struct *p, *n, *reaper; LIST_HEAD(dead_children); - exit_ptrace(father); - write_lock_irq(&tasklist_lock); + /* + * Note that exit_ptrace() and find_new_reaper() might + * drop tasklist_lock and reacquire it. + */ + exit_ptrace(father); reaper = find_new_reaper(father); list_for_each_entry_safe(p, n, &father->children, sibling) { diff --git a/kernel/ptrace.c b/kernel/ptrace.c index 74a3d693c196..f34d798ef4a2 100644 --- a/kernel/ptrace.c +++ b/kernel/ptrace.c @@ -324,26 +324,32 @@ int ptrace_detach(struct task_struct *child, unsigned int data) } /* - * Detach all tasks we were using ptrace on. + * Detach all tasks we were using ptrace on. Called with tasklist held + * for writing, and returns with it held too. But note it can release + * and reacquire the lock. */ void exit_ptrace(struct task_struct *tracer) { struct task_struct *p, *n; LIST_HEAD(ptrace_dead); - write_lock_irq(&tasklist_lock); + if (likely(list_empty(&tracer->ptraced))) + return; + list_for_each_entry_safe(p, n, &tracer->ptraced, ptrace_entry) { if (__ptrace_detach(tracer, p)) list_add(&p->ptrace_entry, &ptrace_dead); } - write_unlock_irq(&tasklist_lock); + write_unlock_irq(&tasklist_lock); BUG_ON(!list_empty(&tracer->ptraced)); list_for_each_entry_safe(p, n, &ptrace_dead, ptrace_entry) { list_del_init(&p->ptrace_entry); release_task(p); } + + write_lock_irq(&tasklist_lock); } int ptrace_readdata(struct task_struct *tsk, unsigned long src, char __user *dst, int len) -- cgit v1.2.3 From 5fdee8c4a5e1800489ce61963208f8cc55e42ea1 Mon Sep 17 00:00:00 2001 From: Salman Date: Tue, 10 Aug 2010 18:03:16 -0700 Subject: pids: fix a race in pid generation that causes pids to be reused immediately A program that repeatedly forks and waits is susceptible to having the same pid repeated, especially when it competes with another instance of the same program. This is really bad for bash implementation. Furthermore, many shell scripts assume that pid numbers will not be used for some length of time. Race Description: A B // pid == offset == n // pid == offset == n + 1 test_and_set_bit(offset, map->page) test_and_set_bit(offset, map->page); pid_ns->last_pid = pid; pid_ns->last_pid = pid; // pid == n + 1 is freed (wait()) // Next fork()... last = pid_ns->last_pid; // == n pid = last + 1; Code to reproduce it (Running multiple instances is more effective): #include #include #include #include #include #include // The distance mod 32768 between two pids, where the first pid is expected // to be smaller than the second. int PidDistance(pid_t first, pid_t second) { return (second + 32768 - first) % 32768; } int main(int argc, char* argv[]) { int failed = 0; pid_t last_pid = 0; int i; printf("%d\n", sizeof(pid_t)); for (i = 0; i < 10000000; ++i) { if (i % 32786 == 0) printf("Iter: %d\n", i/32768); int child_exit_code = i % 256; pid_t pid = fork(); if (pid == -1) { fprintf(stderr, "fork failed, iteration %d, errno=%d", i, errno); exit(1); } if (pid == 0) { // Child exit(child_exit_code); } else { // Parent if (i > 0) { int distance = PidDistance(last_pid, pid); if (distance == 0 || distance > 30000) { fprintf(stderr, "Unexpected pid sequence: previous fork: pid=%d, " "current fork: pid=%d for iteration=%d.\n", last_pid, pid, i); failed = 1; } } last_pid = pid; int status; int reaped = wait(&status); if (reaped != pid) { fprintf(stderr, "Wait return value: expected pid=%d, " "got %d, iteration %d\n", pid, reaped, i); failed = 1; } else if (WEXITSTATUS(status) != child_exit_code) { fprintf(stderr, "Unexpected exit status %x, iteration %d\n", WEXITSTATUS(status), i); failed = 1; } } } exit(failed); } Thanks to Ted Tso for the key ideas of this implementation. Signed-off-by: Salman Qazi Cc: Ingo Molnar Cc: Theodore Ts'o Cc: Peter Zijlstra Cc: Sukadev Bhattiprolu Cc: "Eric W. Biederman" Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/pid.c | 39 ++++++++++++++++++++++++++++++++++++++- 1 file changed, 38 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/pid.c b/kernel/pid.c index e9fd8c132d26..fbbd5f6b6f2f 100644 --- a/kernel/pid.c +++ b/kernel/pid.c @@ -122,6 +122,43 @@ static void free_pidmap(struct upid *upid) atomic_inc(&map->nr_free); } +/* + * If we started walking pids at 'base', is 'a' seen before 'b'? + */ +static int pid_before(int base, int a, int b) +{ + /* + * This is the same as saying + * + * (a - base + MAXUINT) % MAXUINT < (b - base + MAXUINT) % MAXUINT + * and that mapping orders 'a' and 'b' with respect to 'base'. + */ + return (unsigned)(a - base) < (unsigned)(b - base); +} + +/* + * We might be racing with someone else trying to set pid_ns->last_pid. + * We want the winner to have the "later" value, because if the + * "earlier" value prevails, then a pid may get reused immediately. + * + * Since pids rollover, it is not sufficient to just pick the bigger + * value. We have to consider where we started counting from. + * + * 'base' is the value of pid_ns->last_pid that we observed when + * we started looking for a pid. + * + * 'pid' is the pid that we eventually found. + */ +static void set_last_pid(struct pid_namespace *pid_ns, int base, int pid) +{ + int prev; + int last_write = base; + do { + prev = last_write; + last_write = cmpxchg(&pid_ns->last_pid, prev, pid); + } while ((prev != last_write) && (pid_before(base, last_write, pid))); +} + static int alloc_pidmap(struct pid_namespace *pid_ns) { int i, offset, max_scan, pid, last = pid_ns->last_pid; @@ -154,7 +191,7 @@ static int alloc_pidmap(struct pid_namespace *pid_ns) do { if (!test_and_set_bit(offset, map->page)) { atomic_dec(&map->nr_free); - pid_ns->last_pid = pid; + set_last_pid(pid_ns, last, pid); return pid; } offset = find_next_offset(map, offset); -- cgit v1.2.3 From c52b0b91ba1f4b7ea90e20385c0a6df0ba54aed4 Mon Sep 17 00:00:00 2001 From: Oleg Nesterov Date: Tue, 10 Aug 2010 18:03:17 -0700 Subject: pids: alloc_pidmap: remove the unnecessary boundary checks alloc_pidmap() calculates max_scan so that if the initial offset != 0 we inspect the first map->page twice. This is correct, we want to find the unused bits < offset in this bitmap block. Add the comment. But it doesn't make any sense to stop the find_next_offset() loop when we are looking into this map->page for the second time. We have already already checked the bits >= offset during the first attempt, it is fine to do this again, no matter if we succeed this time or not. Remove this hard-to-understand code. It optimizes the very unlikely case when we are going to fail, but slows down the more likely case. Signed-off-by: Oleg Nesterov Cc: Salman Qazi Cc: Ingo Molnar Cc: Sukadev Bhattiprolu Cc: "Eric W. Biederman" Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/pid.c | 17 +++++++---------- 1 file changed, 7 insertions(+), 10 deletions(-) (limited to 'kernel') diff --git a/kernel/pid.c b/kernel/pid.c index fbbd5f6b6f2f..d55c6fb8d087 100644 --- a/kernel/pid.c +++ b/kernel/pid.c @@ -169,7 +169,12 @@ static int alloc_pidmap(struct pid_namespace *pid_ns) pid = RESERVED_PIDS; offset = pid & BITS_PER_PAGE_MASK; map = &pid_ns->pidmap[pid/BITS_PER_PAGE]; - max_scan = (pid_max + BITS_PER_PAGE - 1)/BITS_PER_PAGE - !offset; + /* + * If last_pid points into the middle of the map->page we + * want to scan this bitmap block twice, the second time + * we start with offset == 0 (or RESERVED_PIDS). + */ + max_scan = DIV_ROUND_UP(pid_max, BITS_PER_PAGE) - !offset; for (i = 0; i <= max_scan; ++i) { if (unlikely(!map->page)) { void *page = kzalloc(PAGE_SIZE, GFP_KERNEL); @@ -196,15 +201,7 @@ static int alloc_pidmap(struct pid_namespace *pid_ns) } offset = find_next_offset(map, offset); pid = mk_pid(pid_ns, map, offset); - /* - * find_next_offset() found a bit, the pid from it - * is in-bounds, and if we fell back to the last - * bitmap block and the final block was the same - * as the starting point, pid is before last_pid. - */ - } while (offset < BITS_PER_PAGE && pid < pid_max && - (i != max_scan || pid < last || - !((last+1) & BITS_PER_PAGE_MASK))); + } while (offset < BITS_PER_PAGE && pid < pid_max); } if (map < &pid_ns->pidmap[(pid_max-1)/BITS_PER_PAGE]) { ++map; -- cgit v1.2.3 From c7ff0d9c92435e836e13aaa8d0e56d4000424bcc Mon Sep 17 00:00:00 2001 From: TAMUKI Shoichi Date: Tue, 10 Aug 2010 18:03:28 -0700 Subject: panic: keep blinking in spite of long spin timer mode To keep panic_timeout accuracy when running under a hypervisor, the current implementation only spins on long time (1 second) calls to mdelay. That brings a good effect, but the problem is the keyboard LEDs don't blink at all on that situation. This patch changes to call to panic_blink_enter() between every mdelay and keeps blinking in spite of long spin timer mode. The time to call to mdelay is now 100ms. Even this change will keep panic_timeout accuracy enough when running under a hypervisor. Signed-off-by: TAMUKI Shoichi Cc: Ben Dooks Cc: Russell King Acked-by: Dmitry Torokhov Cc: Anton Blanchard Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/panic.c | 58 ++++++++++++++++++++++++++-------------------------------- 1 file changed, 26 insertions(+), 32 deletions(-) (limited to 'kernel') diff --git a/kernel/panic.c b/kernel/panic.c index 3b16cd93fa7d..3e9037ae10e1 100644 --- a/kernel/panic.c +++ b/kernel/panic.c @@ -24,6 +24,9 @@ #include #include +#define PANIC_TIMER_STEP 100 +#define PANIC_BLINK_SPD 18 + int panic_on_oops; static unsigned long tainted_mask; static int pause_on_oops; @@ -36,36 +39,15 @@ ATOMIC_NOTIFIER_HEAD(panic_notifier_list); EXPORT_SYMBOL(panic_notifier_list); -/* Returns how long it waited in ms */ -long (*panic_blink)(long time); -EXPORT_SYMBOL(panic_blink); - -static void panic_blink_one_second(void) +static long no_blink(int state) { - static long i = 0, end; - - if (panic_blink) { - end = i + MSEC_PER_SEC; - - while (i < end) { - i += panic_blink(i); - mdelay(1); - i++; - } - } else { - /* - * When running under a hypervisor a small mdelay may get - * rounded up to the hypervisor timeslice. For example, with - * a 1ms in 10ms hypervisor timeslice we might inflate a - * mdelay(1) loop by 10x. - * - * If we have nothing to blink, spin on 1 second calls to - * mdelay to avoid this. - */ - mdelay(MSEC_PER_SEC); - } + return 0; } +/* Returns how long it waited in ms */ +long (*panic_blink)(int state); +EXPORT_SYMBOL(panic_blink); + /** * panic - halt the system * @fmt: The text string to print @@ -78,7 +60,8 @@ NORET_TYPE void panic(const char * fmt, ...) { static char buf[1024]; va_list args; - long i; + long i, i_next = 0; + int state = 0; /* * It's possible to come here directly from a panic-assertion and @@ -117,6 +100,9 @@ NORET_TYPE void panic(const char * fmt, ...) bust_spinlocks(0); + if (!panic_blink) + panic_blink = no_blink; + if (panic_timeout > 0) { /* * Delay timeout seconds before rebooting the machine. @@ -124,9 +110,13 @@ NORET_TYPE void panic(const char * fmt, ...) */ printk(KERN_EMERG "Rebooting in %d seconds..", panic_timeout); - for (i = 0; i < panic_timeout; i++) { + for (i = 0; i < panic_timeout * 1000; i += PANIC_TIMER_STEP) { touch_nmi_watchdog(); - panic_blink_one_second(); + if (i >= i_next) { + i += panic_blink(state ^= 1); + i_next = i + 3600 / PANIC_BLINK_SPD; + } + mdelay(PANIC_TIMER_STEP); } /* * This will not be a clean reboot, with everything @@ -152,9 +142,13 @@ NORET_TYPE void panic(const char * fmt, ...) } #endif local_irq_enable(); - while (1) { + for (i = 0; ; i += PANIC_TIMER_STEP) { touch_softlockup_watchdog(); - panic_blink_one_second(); + if (i >= i_next) { + i += panic_blink(state ^= 1); + i_next = i + 3600 / PANIC_BLINK_SPD; + } + mdelay(PANIC_TIMER_STEP); } } -- cgit v1.2.3 From 863a6049202412a6d655d052eb1c45ca7dd74a83 Mon Sep 17 00:00:00 2001 From: Anton Blanchard Date: Tue, 10 Aug 2010 18:03:30 -0700 Subject: lib/bug.c: add oops end marker to WARN implementation We are missing the oops end marker for the exception based WARN implementation in lib/bug.c. This is useful for logfile analysis tools. Signed-off-by: Anton Blanchard Cc: Ingo Molnar Cc: Arjan van de Ven Cc: "Kirill A. Shutemov" Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/panic.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/panic.c b/kernel/panic.c index 3e9037ae10e1..4c13b1a88ebb 100644 --- a/kernel/panic.c +++ b/kernel/panic.c @@ -338,7 +338,7 @@ static int init_oops_id(void) } late_initcall(init_oops_id); -static void print_oops_end_marker(void) +void print_oops_end_marker(void) { init_oops_id(); printk(KERN_WARNING "---[ end trace %016llx ]---\n", -- cgit v1.2.3 From f65a03f6ab6f53a6f2847dbac232dcb38b3b3642 Mon Sep 17 00:00:00 2001 From: Dan Carpenter Date: Tue, 10 Aug 2010 18:03:31 -0700 Subject: kexec: return -EFAULT on copy_to_user() failures copy_to/from_user() returns the number of bytes remaining to be copied. It never returns a negative value. The correct return code is -EFAULT and not -EIO. All the callers check for non-zero returns so that's Ok, but the return code is passed to the user so we should fix this. Signed-off-by: Dan Carpenter Cc: Hidetoshi Seto Cc: "Paul E. McKenney" Cc: "Eric W. Biederman" Cc: Simon Kagstrom Acked-by: WANG Cong Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/kexec.c | 8 +++++--- 1 file changed, 5 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/kexec.c b/kernel/kexec.c index 131b1703936f..c0613f7d6730 100644 --- a/kernel/kexec.c +++ b/kernel/kexec.c @@ -151,8 +151,10 @@ static int do_kimage_alloc(struct kimage **rimage, unsigned long entry, image->nr_segments = nr_segments; segment_bytes = nr_segments * sizeof(*segments); result = copy_from_user(image->segment, segments, segment_bytes); - if (result) + if (result) { + result = -EFAULT; goto out; + } /* * Verify we have good destination addresses. The caller is @@ -827,7 +829,7 @@ static int kimage_load_normal_segment(struct kimage *image, result = copy_from_user(ptr, buf, uchunk); kunmap(page); if (result) { - result = (result < 0) ? result : -EIO; + result = -EFAULT; goto out; } ubytes -= uchunk; @@ -882,7 +884,7 @@ static int kimage_load_crash_segment(struct kimage *image, kexec_flush_icache_page(page); kunmap(page); if (result) { - result = (result < 0) ? result : -EIO; + result = -EFAULT; goto out; } ubytes -= uchunk; -- cgit v1.2.3 From 4201d9a8e86b51dd40aa8a0dabd093376c859985 Mon Sep 17 00:00:00 2001 From: Stefani Seibold Date: Tue, 10 Aug 2010 18:03:38 -0700 Subject: kfifo: add the new generic kfifo API Add the new version of the kfifo API files kfifo.c and kfifo.h. Signed-off-by: Stefani Seibold Cc: Greg KH Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/kfifo-new.c | 602 +++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 602 insertions(+) create mode 100644 kernel/kfifo-new.c (limited to 'kernel') diff --git a/kernel/kfifo-new.c b/kernel/kfifo-new.c new file mode 100644 index 000000000000..02192dd905cc --- /dev/null +++ b/kernel/kfifo-new.c @@ -0,0 +1,602 @@ +/* + * A generic kernel FIFO implementation + * + * Copyright (C) 2009/2010 Stefani Seibold + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. + * + */ + +#include +#include +#include +#include +#include +#include +#include + +/* + * internal helper to calculate the unused elements in a fifo + */ +static inline unsigned int kfifo_unused(struct __kfifo *fifo) +{ + return (fifo->mask + 1) - (fifo->in - fifo->out); +} + +int __kfifo_alloc(struct __kfifo *fifo, unsigned int size, + size_t esize, gfp_t gfp_mask) +{ + /* + * round down to the next power of 2, since our 'let the indices + * wrap' technique works only in this case. + */ + if (!is_power_of_2(size)) + size = rounddown_pow_of_two(size); + + fifo->in = 0; + fifo->out = 0; + fifo->esize = esize; + + if (size < 2) { + fifo->data = NULL; + fifo->mask = 0; + return -EINVAL; + } + + fifo->data = kmalloc(size * esize, gfp_mask); + + if (!fifo->data) { + fifo->mask = 0; + return -ENOMEM; + } + fifo->mask = size - 1; + + return 0; +} +EXPORT_SYMBOL(__kfifo_alloc); + +void __kfifo_free(struct __kfifo *fifo) +{ + kfree(fifo->data); + fifo->in = 0; + fifo->out = 0; + fifo->esize = 0; + fifo->data = NULL; + fifo->mask = 0; +} +EXPORT_SYMBOL(__kfifo_free); + +int __kfifo_init(struct __kfifo *fifo, void *buffer, + unsigned int size, size_t esize) +{ + size /= esize; + + if (!is_power_of_2(size)) + size = rounddown_pow_of_two(size); + + fifo->in = 0; + fifo->out = 0; + fifo->esize = esize; + fifo->data = buffer; + + if (size < 2) { + fifo->mask = 0; + return -EINVAL; + } + fifo->mask = size - 1; + + return 0; +} +EXPORT_SYMBOL(__kfifo_init); + +static void kfifo_copy_in(struct __kfifo *fifo, const void *src, + unsigned int len, unsigned int off) +{ + unsigned int size = fifo->mask + 1; + unsigned int esize = fifo->esize; + unsigned int l; + + off &= fifo->mask; + if (esize != 1) { + off *= esize; + size *= esize; + len *= esize; + } + l = min(len, size - off); + + memcpy(fifo->data + off, src, l); + memcpy(fifo->data, src + l, len - l); + /* + * make sure that the data in the fifo is up to date before + * incrementing the fifo->in index counter + */ + smp_wmb(); +} + +unsigned int __kfifo_in(struct __kfifo *fifo, + const void *buf, unsigned int len) +{ + unsigned int l; + + l = kfifo_unused(fifo); + if (len > l) + len = l; + + kfifo_copy_in(fifo, buf, len, fifo->in); + fifo->in += len; + return len; +} +EXPORT_SYMBOL(__kfifo_in); + +static void kfifo_copy_out(struct __kfifo *fifo, void *dst, + unsigned int len, unsigned int off) +{ + unsigned int size = fifo->mask + 1; + unsigned int esize = fifo->esize; + unsigned int l; + + off &= fifo->mask; + if (esize != 1) { + off *= esize; + size *= esize; + len *= esize; + } + l = min(len, size - off); + + memcpy(dst, fifo->data + off, l); + memcpy(dst + l, fifo->data, len - l); + /* + * make sure that the data is copied before + * incrementing the fifo->out index counter + */ + smp_wmb(); +} + +unsigned int __kfifo_out_peek(struct __kfifo *fifo, + void *buf, unsigned int len) +{ + unsigned int l; + + l = fifo->in - fifo->out; + if (len > l) + len = l; + + kfifo_copy_out(fifo, buf, len, fifo->out); + return len; +} +EXPORT_SYMBOL(__kfifo_out_peek); + +unsigned int __kfifo_out(struct __kfifo *fifo, + void *buf, unsigned int len) +{ + len = __kfifo_out_peek(fifo, buf, len); + fifo->out += len; + return len; +} +EXPORT_SYMBOL(__kfifo_out); + +static unsigned long kfifo_copy_from_user(struct __kfifo *fifo, + const void __user *from, unsigned int len, unsigned int off, + unsigned int *copied) +{ + unsigned int size = fifo->mask + 1; + unsigned int esize = fifo->esize; + unsigned int l; + unsigned long ret; + + off &= fifo->mask; + if (esize != 1) { + off *= esize; + size *= esize; + len *= esize; + } + l = min(len, size - off); + + ret = copy_from_user(fifo->data + off, from, l); + if (unlikely(ret)) + ret = DIV_ROUND_UP(ret + len - l, esize); + else { + ret = copy_from_user(fifo->data, from + l, len - l); + if (unlikely(ret)) + ret = DIV_ROUND_UP(ret, esize); + } + /* + * make sure that the data in the fifo is up to date before + * incrementing the fifo->in index counter + */ + smp_wmb(); + *copied = len - ret; + /* return the number of elements which are not copied */ + return ret; +} + +int __kfifo_from_user(struct __kfifo *fifo, const void __user *from, + unsigned long len, unsigned int *copied) +{ + unsigned int l; + unsigned long ret; + unsigned int esize = fifo->esize; + int err; + + if (esize != 1) + len /= esize; + + l = kfifo_unused(fifo); + if (len > l) + len = l; + + ret = kfifo_copy_from_user(fifo, from, len, fifo->in, copied); + if (unlikely(ret)) { + len -= ret; + err = -EFAULT; + } else + err = 0; + fifo->in += len; + return err; +} +EXPORT_SYMBOL(__kfifo_from_user); + +static unsigned long kfifo_copy_to_user(struct __kfifo *fifo, void __user *to, + unsigned int len, unsigned int off, unsigned int *copied) +{ + unsigned int l; + unsigned long ret; + unsigned int size = fifo->mask + 1; + unsigned int esize = fifo->esize; + + off &= fifo->mask; + if (esize != 1) { + off *= esize; + size *= esize; + len *= esize; + } + l = min(len, size - off); + + ret = copy_to_user(to, fifo->data + off, l); + if (unlikely(ret)) + ret = DIV_ROUND_UP(ret + len - l, esize); + else { + ret = copy_to_user(to + l, fifo->data, len - l); + if (unlikely(ret)) + ret = DIV_ROUND_UP(ret, esize); + } + /* + * make sure that the data is copied before + * incrementing the fifo->out index counter + */ + smp_wmb(); + *copied = len - ret; + /* return the number of elements which are not copied */ + return ret; +} + +int __kfifo_to_user(struct __kfifo *fifo, void __user *to, + unsigned long len, unsigned int *copied) +{ + unsigned int l; + unsigned long ret; + unsigned int esize = fifo->esize; + int err; + + if (esize != 1) + len /= esize; + + l = fifo->in - fifo->out; + if (len > l) + len = l; + ret = kfifo_copy_to_user(fifo, to, len, fifo->out, copied); + if (unlikely(ret)) { + len -= ret; + err = -EFAULT; + } else + err = 0; + fifo->out += len; + return err; +} +EXPORT_SYMBOL(__kfifo_to_user); + +static int setup_sgl_buf(struct scatterlist *sgl, void *buf, + int nents, unsigned int len) +{ + int n; + unsigned int l; + unsigned int off; + struct page *page; + + if (!nents) + return 0; + + if (!len) + return 0; + + n = 0; + page = virt_to_page(buf); + off = offset_in_page(buf); + l = 0; + + while (len >= l + PAGE_SIZE - off) { + struct page *npage; + + l += PAGE_SIZE; + buf += PAGE_SIZE; + npage = virt_to_page(buf); + if (page_to_phys(page) != page_to_phys(npage) - l) { + sgl->page_link = 0; + sg_set_page(sgl++, page, l - off, off); + if (++n == nents) + return n; + page = npage; + len -= l - off; + l = off = 0; + } + } + sgl->page_link = 0; + sg_set_page(sgl++, page, len, off); + return n + 1; +} + +static unsigned int setup_sgl(struct __kfifo *fifo, struct scatterlist *sgl, + int nents, unsigned int len, unsigned int off) +{ + unsigned int size = fifo->mask + 1; + unsigned int esize = fifo->esize; + unsigned int l; + unsigned int n; + + off &= fifo->mask; + if (esize != 1) { + off *= esize; + size *= esize; + len *= esize; + } + l = min(len, size - off); + + n = setup_sgl_buf(sgl, fifo->data + off, nents, l); + n += setup_sgl_buf(sgl + n, fifo->data, nents - n, len - l); + + if (n) + sg_mark_end(sgl + n - 1); + return n; +} + +unsigned int __kfifo_dma_in_prepare(struct __kfifo *fifo, + struct scatterlist *sgl, int nents, unsigned int len) +{ + unsigned int l; + + l = kfifo_unused(fifo); + if (len > l) + len = l; + + return setup_sgl(fifo, sgl, nents, len, fifo->in); +} +EXPORT_SYMBOL(__kfifo_dma_in_prepare); + +unsigned int __kfifo_dma_out_prepare(struct __kfifo *fifo, + struct scatterlist *sgl, int nents, unsigned int len) +{ + unsigned int l; + + l = fifo->in - fifo->out; + if (len > l) + len = l; + + return setup_sgl(fifo, sgl, nents, len, fifo->out); +} +EXPORT_SYMBOL(__kfifo_dma_out_prepare); + +unsigned int __kfifo_max_r(unsigned int len, size_t recsize) +{ + unsigned int max = (1 << (recsize << 3)) - 1; + + if (len > max) + return max; + return len; +} + +#define __KFIFO_PEEK(data, out, mask) \ + ((data)[(out) & (mask)]) +/* + * __kfifo_peek_n internal helper function for determinate the length of + * the next record in the fifo + */ +static unsigned int __kfifo_peek_n(struct __kfifo *fifo, size_t recsize) +{ + unsigned int l; + unsigned int mask = fifo->mask; + unsigned char *data = fifo->data; + + l = __KFIFO_PEEK(data, fifo->out, mask); + + if (--recsize) + l |= __KFIFO_PEEK(data, fifo->out + 1, mask) << 8; + + return l; +} + +#define __KFIFO_POKE(data, in, mask, val) \ + ( \ + (data)[(in) & (mask)] = (unsigned char)(val) \ + ) + +/* + * __kfifo_poke_n internal helper function for storeing the length of + * the record into the fifo + */ +static void __kfifo_poke_n(struct __kfifo *fifo, unsigned int n, size_t recsize) +{ + unsigned int mask = fifo->mask; + unsigned char *data = fifo->data; + + __KFIFO_POKE(data, fifo->in, mask, n); + + if (recsize > 1) + __KFIFO_POKE(data, fifo->in + 1, mask, n >> 8); +} + +unsigned int __kfifo_len_r(struct __kfifo *fifo, size_t recsize) +{ + return __kfifo_peek_n(fifo, recsize); +} +EXPORT_SYMBOL(__kfifo_len_r); + +unsigned int __kfifo_in_r(struct __kfifo *fifo, const void *buf, + unsigned int len, size_t recsize) +{ + if (len + recsize > kfifo_unused(fifo)) + return 0; + + __kfifo_poke_n(fifo, len, recsize); + + kfifo_copy_in(fifo, buf, len, fifo->in + recsize); + fifo->in += len + recsize; + return len; +} +EXPORT_SYMBOL(__kfifo_in_r); + +static unsigned int kfifo_out_copy_r(struct __kfifo *fifo, + void *buf, unsigned int len, size_t recsize, unsigned int *n) +{ + *n = __kfifo_peek_n(fifo, recsize); + + if (len > *n) + len = *n; + + kfifo_copy_out(fifo, buf, len, fifo->out + recsize); + return len; +} + +unsigned int __kfifo_out_peek_r(struct __kfifo *fifo, void *buf, + unsigned int len, size_t recsize) +{ + unsigned int n; + + if (fifo->in == fifo->out) + return 0; + + return kfifo_out_copy_r(fifo, buf, len, recsize, &n); +} +EXPORT_SYMBOL(__kfifo_out_peek_r); + +unsigned int __kfifo_out_r(struct __kfifo *fifo, void *buf, + unsigned int len, size_t recsize) +{ + unsigned int n; + + if (fifo->in == fifo->out) + return 0; + + len = kfifo_out_copy_r(fifo, buf, len, recsize, &n); + fifo->out += n + recsize; + return len; +} +EXPORT_SYMBOL(__kfifo_out_r); + +int __kfifo_from_user_r(struct __kfifo *fifo, const void __user *from, + unsigned long len, unsigned int *copied, size_t recsize) +{ + unsigned long ret; + + len = __kfifo_max_r(len, recsize); + + if (len + recsize > kfifo_unused(fifo)) { + *copied = 0; + return 0; + } + + __kfifo_poke_n(fifo, len, recsize); + + ret = kfifo_copy_from_user(fifo, from, len, fifo->in + recsize, copied); + if (unlikely(ret)) { + *copied = 0; + return -EFAULT; + } + fifo->in += len + recsize; + return 0; +} +EXPORT_SYMBOL(__kfifo_from_user_r); + +int __kfifo_to_user_r(struct __kfifo *fifo, void __user *to, + unsigned long len, unsigned int *copied, size_t recsize) +{ + unsigned long ret; + unsigned int n; + + if (fifo->in == fifo->out) { + *copied = 0; + return 0; + } + + n = __kfifo_peek_n(fifo, recsize); + if (len > n) + len = n; + + ret = kfifo_copy_to_user(fifo, to, len, fifo->out + recsize, copied); + if (unlikely(ret)) { + *copied = 0; + return -EFAULT; + } + fifo->out += n + recsize; + return 0; +} +EXPORT_SYMBOL(__kfifo_to_user_r); + +unsigned int __kfifo_dma_in_prepare_r(struct __kfifo *fifo, + struct scatterlist *sgl, int nents, unsigned int len, size_t recsize) +{ + if (!nents) + BUG(); + + len = __kfifo_max_r(len, recsize); + + if (len + recsize > kfifo_unused(fifo)) + return 0; + + return setup_sgl(fifo, sgl, nents, len, fifo->in + recsize); +} +EXPORT_SYMBOL(__kfifo_dma_in_prepare_r); + +void __kfifo_dma_in_finish_r(struct __kfifo *fifo, + unsigned int len, size_t recsize) +{ + len = __kfifo_max_r(len, recsize); + __kfifo_poke_n(fifo, len, recsize); + fifo->in += len + recsize; +} +EXPORT_SYMBOL(__kfifo_dma_in_finish_r); + +unsigned int __kfifo_dma_out_prepare_r(struct __kfifo *fifo, + struct scatterlist *sgl, int nents, unsigned int len, size_t recsize) +{ + if (!nents) + BUG(); + + len = __kfifo_max_r(len, recsize); + + if (len + recsize > fifo->in - fifo->out) + return 0; + + return setup_sgl(fifo, sgl, nents, len, fifo->out + recsize); +} +EXPORT_SYMBOL(__kfifo_dma_out_prepare_r); + +void __kfifo_dma_out_finish_r(struct __kfifo *fifo, size_t recsize) +{ + unsigned int len; + + len = __kfifo_peek_n(fifo, recsize); + fifo->out += len + recsize; +} +EXPORT_SYMBOL(__kfifo_dma_out_finish_r); -- cgit v1.2.3 From 2e956fb320568cc70861761483e2f0e2db75fd66 Mon Sep 17 00:00:00 2001 From: Stefani Seibold Date: Tue, 10 Aug 2010 18:03:38 -0700 Subject: kfifo: replace the old non generic API Simply replace the whole kfifo.c and kfifo.h files with the new generic version and fix the kerneldoc API template file. Signed-off-by: Stefani Seibold Cc: Greg KH Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/kfifo-new.c | 602 ------------------------------------------ kernel/kfifo.c | 749 ++++++++++++++++++++++++++++++++--------------------- 2 files changed, 453 insertions(+), 898 deletions(-) delete mode 100644 kernel/kfifo-new.c (limited to 'kernel') diff --git a/kernel/kfifo-new.c b/kernel/kfifo-new.c deleted file mode 100644 index 02192dd905cc..000000000000 --- a/kernel/kfifo-new.c +++ /dev/null @@ -1,602 +0,0 @@ -/* - * A generic kernel FIFO implementation - * - * Copyright (C) 2009/2010 Stefani Seibold - * - * This program is free software; you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation; either version 2 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program; if not, write to the Free Software - * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. - * - */ - -#include -#include -#include -#include -#include -#include -#include - -/* - * internal helper to calculate the unused elements in a fifo - */ -static inline unsigned int kfifo_unused(struct __kfifo *fifo) -{ - return (fifo->mask + 1) - (fifo->in - fifo->out); -} - -int __kfifo_alloc(struct __kfifo *fifo, unsigned int size, - size_t esize, gfp_t gfp_mask) -{ - /* - * round down to the next power of 2, since our 'let the indices - * wrap' technique works only in this case. - */ - if (!is_power_of_2(size)) - size = rounddown_pow_of_two(size); - - fifo->in = 0; - fifo->out = 0; - fifo->esize = esize; - - if (size < 2) { - fifo->data = NULL; - fifo->mask = 0; - return -EINVAL; - } - - fifo->data = kmalloc(size * esize, gfp_mask); - - if (!fifo->data) { - fifo->mask = 0; - return -ENOMEM; - } - fifo->mask = size - 1; - - return 0; -} -EXPORT_SYMBOL(__kfifo_alloc); - -void __kfifo_free(struct __kfifo *fifo) -{ - kfree(fifo->data); - fifo->in = 0; - fifo->out = 0; - fifo->esize = 0; - fifo->data = NULL; - fifo->mask = 0; -} -EXPORT_SYMBOL(__kfifo_free); - -int __kfifo_init(struct __kfifo *fifo, void *buffer, - unsigned int size, size_t esize) -{ - size /= esize; - - if (!is_power_of_2(size)) - size = rounddown_pow_of_two(size); - - fifo->in = 0; - fifo->out = 0; - fifo->esize = esize; - fifo->data = buffer; - - if (size < 2) { - fifo->mask = 0; - return -EINVAL; - } - fifo->mask = size - 1; - - return 0; -} -EXPORT_SYMBOL(__kfifo_init); - -static void kfifo_copy_in(struct __kfifo *fifo, const void *src, - unsigned int len, unsigned int off) -{ - unsigned int size = fifo->mask + 1; - unsigned int esize = fifo->esize; - unsigned int l; - - off &= fifo->mask; - if (esize != 1) { - off *= esize; - size *= esize; - len *= esize; - } - l = min(len, size - off); - - memcpy(fifo->data + off, src, l); - memcpy(fifo->data, src + l, len - l); - /* - * make sure that the data in the fifo is up to date before - * incrementing the fifo->in index counter - */ - smp_wmb(); -} - -unsigned int __kfifo_in(struct __kfifo *fifo, - const void *buf, unsigned int len) -{ - unsigned int l; - - l = kfifo_unused(fifo); - if (len > l) - len = l; - - kfifo_copy_in(fifo, buf, len, fifo->in); - fifo->in += len; - return len; -} -EXPORT_SYMBOL(__kfifo_in); - -static void kfifo_copy_out(struct __kfifo *fifo, void *dst, - unsigned int len, unsigned int off) -{ - unsigned int size = fifo->mask + 1; - unsigned int esize = fifo->esize; - unsigned int l; - - off &= fifo->mask; - if (esize != 1) { - off *= esize; - size *= esize; - len *= esize; - } - l = min(len, size - off); - - memcpy(dst, fifo->data + off, l); - memcpy(dst + l, fifo->data, len - l); - /* - * make sure that the data is copied before - * incrementing the fifo->out index counter - */ - smp_wmb(); -} - -unsigned int __kfifo_out_peek(struct __kfifo *fifo, - void *buf, unsigned int len) -{ - unsigned int l; - - l = fifo->in - fifo->out; - if (len > l) - len = l; - - kfifo_copy_out(fifo, buf, len, fifo->out); - return len; -} -EXPORT_SYMBOL(__kfifo_out_peek); - -unsigned int __kfifo_out(struct __kfifo *fifo, - void *buf, unsigned int len) -{ - len = __kfifo_out_peek(fifo, buf, len); - fifo->out += len; - return len; -} -EXPORT_SYMBOL(__kfifo_out); - -static unsigned long kfifo_copy_from_user(struct __kfifo *fifo, - const void __user *from, unsigned int len, unsigned int off, - unsigned int *copied) -{ - unsigned int size = fifo->mask + 1; - unsigned int esize = fifo->esize; - unsigned int l; - unsigned long ret; - - off &= fifo->mask; - if (esize != 1) { - off *= esize; - size *= esize; - len *= esize; - } - l = min(len, size - off); - - ret = copy_from_user(fifo->data + off, from, l); - if (unlikely(ret)) - ret = DIV_ROUND_UP(ret + len - l, esize); - else { - ret = copy_from_user(fifo->data, from + l, len - l); - if (unlikely(ret)) - ret = DIV_ROUND_UP(ret, esize); - } - /* - * make sure that the data in the fifo is up to date before - * incrementing the fifo->in index counter - */ - smp_wmb(); - *copied = len - ret; - /* return the number of elements which are not copied */ - return ret; -} - -int __kfifo_from_user(struct __kfifo *fifo, const void __user *from, - unsigned long len, unsigned int *copied) -{ - unsigned int l; - unsigned long ret; - unsigned int esize = fifo->esize; - int err; - - if (esize != 1) - len /= esize; - - l = kfifo_unused(fifo); - if (len > l) - len = l; - - ret = kfifo_copy_from_user(fifo, from, len, fifo->in, copied); - if (unlikely(ret)) { - len -= ret; - err = -EFAULT; - } else - err = 0; - fifo->in += len; - return err; -} -EXPORT_SYMBOL(__kfifo_from_user); - -static unsigned long kfifo_copy_to_user(struct __kfifo *fifo, void __user *to, - unsigned int len, unsigned int off, unsigned int *copied) -{ - unsigned int l; - unsigned long ret; - unsigned int size = fifo->mask + 1; - unsigned int esize = fifo->esize; - - off &= fifo->mask; - if (esize != 1) { - off *= esize; - size *= esize; - len *= esize; - } - l = min(len, size - off); - - ret = copy_to_user(to, fifo->data + off, l); - if (unlikely(ret)) - ret = DIV_ROUND_UP(ret + len - l, esize); - else { - ret = copy_to_user(to + l, fifo->data, len - l); - if (unlikely(ret)) - ret = DIV_ROUND_UP(ret, esize); - } - /* - * make sure that the data is copied before - * incrementing the fifo->out index counter - */ - smp_wmb(); - *copied = len - ret; - /* return the number of elements which are not copied */ - return ret; -} - -int __kfifo_to_user(struct __kfifo *fifo, void __user *to, - unsigned long len, unsigned int *copied) -{ - unsigned int l; - unsigned long ret; - unsigned int esize = fifo->esize; - int err; - - if (esize != 1) - len /= esize; - - l = fifo->in - fifo->out; - if (len > l) - len = l; - ret = kfifo_copy_to_user(fifo, to, len, fifo->out, copied); - if (unlikely(ret)) { - len -= ret; - err = -EFAULT; - } else - err = 0; - fifo->out += len; - return err; -} -EXPORT_SYMBOL(__kfifo_to_user); - -static int setup_sgl_buf(struct scatterlist *sgl, void *buf, - int nents, unsigned int len) -{ - int n; - unsigned int l; - unsigned int off; - struct page *page; - - if (!nents) - return 0; - - if (!len) - return 0; - - n = 0; - page = virt_to_page(buf); - off = offset_in_page(buf); - l = 0; - - while (len >= l + PAGE_SIZE - off) { - struct page *npage; - - l += PAGE_SIZE; - buf += PAGE_SIZE; - npage = virt_to_page(buf); - if (page_to_phys(page) != page_to_phys(npage) - l) { - sgl->page_link = 0; - sg_set_page(sgl++, page, l - off, off); - if (++n == nents) - return n; - page = npage; - len -= l - off; - l = off = 0; - } - } - sgl->page_link = 0; - sg_set_page(sgl++, page, len, off); - return n + 1; -} - -static unsigned int setup_sgl(struct __kfifo *fifo, struct scatterlist *sgl, - int nents, unsigned int len, unsigned int off) -{ - unsigned int size = fifo->mask + 1; - unsigned int esize = fifo->esize; - unsigned int l; - unsigned int n; - - off &= fifo->mask; - if (esize != 1) { - off *= esize; - size *= esize; - len *= esize; - } - l = min(len, size - off); - - n = setup_sgl_buf(sgl, fifo->data + off, nents, l); - n += setup_sgl_buf(sgl + n, fifo->data, nents - n, len - l); - - if (n) - sg_mark_end(sgl + n - 1); - return n; -} - -unsigned int __kfifo_dma_in_prepare(struct __kfifo *fifo, - struct scatterlist *sgl, int nents, unsigned int len) -{ - unsigned int l; - - l = kfifo_unused(fifo); - if (len > l) - len = l; - - return setup_sgl(fifo, sgl, nents, len, fifo->in); -} -EXPORT_SYMBOL(__kfifo_dma_in_prepare); - -unsigned int __kfifo_dma_out_prepare(struct __kfifo *fifo, - struct scatterlist *sgl, int nents, unsigned int len) -{ - unsigned int l; - - l = fifo->in - fifo->out; - if (len > l) - len = l; - - return setup_sgl(fifo, sgl, nents, len, fifo->out); -} -EXPORT_SYMBOL(__kfifo_dma_out_prepare); - -unsigned int __kfifo_max_r(unsigned int len, size_t recsize) -{ - unsigned int max = (1 << (recsize << 3)) - 1; - - if (len > max) - return max; - return len; -} - -#define __KFIFO_PEEK(data, out, mask) \ - ((data)[(out) & (mask)]) -/* - * __kfifo_peek_n internal helper function for determinate the length of - * the next record in the fifo - */ -static unsigned int __kfifo_peek_n(struct __kfifo *fifo, size_t recsize) -{ - unsigned int l; - unsigned int mask = fifo->mask; - unsigned char *data = fifo->data; - - l = __KFIFO_PEEK(data, fifo->out, mask); - - if (--recsize) - l |= __KFIFO_PEEK(data, fifo->out + 1, mask) << 8; - - return l; -} - -#define __KFIFO_POKE(data, in, mask, val) \ - ( \ - (data)[(in) & (mask)] = (unsigned char)(val) \ - ) - -/* - * __kfifo_poke_n internal helper function for storeing the length of - * the record into the fifo - */ -static void __kfifo_poke_n(struct __kfifo *fifo, unsigned int n, size_t recsize) -{ - unsigned int mask = fifo->mask; - unsigned char *data = fifo->data; - - __KFIFO_POKE(data, fifo->in, mask, n); - - if (recsize > 1) - __KFIFO_POKE(data, fifo->in + 1, mask, n >> 8); -} - -unsigned int __kfifo_len_r(struct __kfifo *fifo, size_t recsize) -{ - return __kfifo_peek_n(fifo, recsize); -} -EXPORT_SYMBOL(__kfifo_len_r); - -unsigned int __kfifo_in_r(struct __kfifo *fifo, const void *buf, - unsigned int len, size_t recsize) -{ - if (len + recsize > kfifo_unused(fifo)) - return 0; - - __kfifo_poke_n(fifo, len, recsize); - - kfifo_copy_in(fifo, buf, len, fifo->in + recsize); - fifo->in += len + recsize; - return len; -} -EXPORT_SYMBOL(__kfifo_in_r); - -static unsigned int kfifo_out_copy_r(struct __kfifo *fifo, - void *buf, unsigned int len, size_t recsize, unsigned int *n) -{ - *n = __kfifo_peek_n(fifo, recsize); - - if (len > *n) - len = *n; - - kfifo_copy_out(fifo, buf, len, fifo->out + recsize); - return len; -} - -unsigned int __kfifo_out_peek_r(struct __kfifo *fifo, void *buf, - unsigned int len, size_t recsize) -{ - unsigned int n; - - if (fifo->in == fifo->out) - return 0; - - return kfifo_out_copy_r(fifo, buf, len, recsize, &n); -} -EXPORT_SYMBOL(__kfifo_out_peek_r); - -unsigned int __kfifo_out_r(struct __kfifo *fifo, void *buf, - unsigned int len, size_t recsize) -{ - unsigned int n; - - if (fifo->in == fifo->out) - return 0; - - len = kfifo_out_copy_r(fifo, buf, len, recsize, &n); - fifo->out += n + recsize; - return len; -} -EXPORT_SYMBOL(__kfifo_out_r); - -int __kfifo_from_user_r(struct __kfifo *fifo, const void __user *from, - unsigned long len, unsigned int *copied, size_t recsize) -{ - unsigned long ret; - - len = __kfifo_max_r(len, recsize); - - if (len + recsize > kfifo_unused(fifo)) { - *copied = 0; - return 0; - } - - __kfifo_poke_n(fifo, len, recsize); - - ret = kfifo_copy_from_user(fifo, from, len, fifo->in + recsize, copied); - if (unlikely(ret)) { - *copied = 0; - return -EFAULT; - } - fifo->in += len + recsize; - return 0; -} -EXPORT_SYMBOL(__kfifo_from_user_r); - -int __kfifo_to_user_r(struct __kfifo *fifo, void __user *to, - unsigned long len, unsigned int *copied, size_t recsize) -{ - unsigned long ret; - unsigned int n; - - if (fifo->in == fifo->out) { - *copied = 0; - return 0; - } - - n = __kfifo_peek_n(fifo, recsize); - if (len > n) - len = n; - - ret = kfifo_copy_to_user(fifo, to, len, fifo->out + recsize, copied); - if (unlikely(ret)) { - *copied = 0; - return -EFAULT; - } - fifo->out += n + recsize; - return 0; -} -EXPORT_SYMBOL(__kfifo_to_user_r); - -unsigned int __kfifo_dma_in_prepare_r(struct __kfifo *fifo, - struct scatterlist *sgl, int nents, unsigned int len, size_t recsize) -{ - if (!nents) - BUG(); - - len = __kfifo_max_r(len, recsize); - - if (len + recsize > kfifo_unused(fifo)) - return 0; - - return setup_sgl(fifo, sgl, nents, len, fifo->in + recsize); -} -EXPORT_SYMBOL(__kfifo_dma_in_prepare_r); - -void __kfifo_dma_in_finish_r(struct __kfifo *fifo, - unsigned int len, size_t recsize) -{ - len = __kfifo_max_r(len, recsize); - __kfifo_poke_n(fifo, len, recsize); - fifo->in += len + recsize; -} -EXPORT_SYMBOL(__kfifo_dma_in_finish_r); - -unsigned int __kfifo_dma_out_prepare_r(struct __kfifo *fifo, - struct scatterlist *sgl, int nents, unsigned int len, size_t recsize) -{ - if (!nents) - BUG(); - - len = __kfifo_max_r(len, recsize); - - if (len + recsize > fifo->in - fifo->out) - return 0; - - return setup_sgl(fifo, sgl, nents, len, fifo->out + recsize); -} -EXPORT_SYMBOL(__kfifo_dma_out_prepare_r); - -void __kfifo_dma_out_finish_r(struct __kfifo *fifo, size_t recsize) -{ - unsigned int len; - - len = __kfifo_peek_n(fifo, recsize); - fifo->out += len + recsize; -} -EXPORT_SYMBOL(__kfifo_dma_out_finish_r); diff --git a/kernel/kfifo.c b/kernel/kfifo.c index 35edbe22e9a9..02192dd905cc 100644 --- a/kernel/kfifo.c +++ b/kernel/kfifo.c @@ -1,8 +1,7 @@ /* - * A generic kernel FIFO implementation. + * A generic kernel FIFO implementation * - * Copyright (C) 2009 Stefani Seibold - * Copyright (C) 2004 Stelian Pop + * Copyright (C) 2009/2010 Stefani Seibold * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -24,422 +23,580 @@ #include #include #include -#include #include #include +#include -static void _kfifo_init(struct kfifo *fifo, void *buffer, - unsigned int size) -{ - fifo->buffer = buffer; - fifo->size = size; - - kfifo_reset(fifo); -} - -/** - * kfifo_init - initialize a FIFO using a preallocated buffer - * @fifo: the fifo to assign the buffer - * @buffer: the preallocated buffer to be used. - * @size: the size of the internal buffer, this has to be a power of 2. - * +/* + * internal helper to calculate the unused elements in a fifo */ -void kfifo_init(struct kfifo *fifo, void *buffer, unsigned int size) +static inline unsigned int kfifo_unused(struct __kfifo *fifo) { - /* size must be a power of 2 */ - BUG_ON(!is_power_of_2(size)); - - _kfifo_init(fifo, buffer, size); + return (fifo->mask + 1) - (fifo->in - fifo->out); } -EXPORT_SYMBOL(kfifo_init); -/** - * kfifo_alloc - allocates a new FIFO internal buffer - * @fifo: the fifo to assign then new buffer - * @size: the size of the buffer to be allocated, this have to be a power of 2. - * @gfp_mask: get_free_pages mask, passed to kmalloc() - * - * This function dynamically allocates a new fifo internal buffer - * - * The size will be rounded-up to a power of 2. - * The buffer will be release with kfifo_free(). - * Return 0 if no error, otherwise the an error code - */ -int kfifo_alloc(struct kfifo *fifo, unsigned int size, gfp_t gfp_mask) +int __kfifo_alloc(struct __kfifo *fifo, unsigned int size, + size_t esize, gfp_t gfp_mask) { - unsigned char *buffer; - /* - * round up to the next power of 2, since our 'let the indices + * round down to the next power of 2, since our 'let the indices * wrap' technique works only in this case. */ - if (!is_power_of_2(size)) { - BUG_ON(size > 0x80000000); - size = roundup_pow_of_two(size); + if (!is_power_of_2(size)) + size = rounddown_pow_of_two(size); + + fifo->in = 0; + fifo->out = 0; + fifo->esize = esize; + + if (size < 2) { + fifo->data = NULL; + fifo->mask = 0; + return -EINVAL; } - buffer = kmalloc(size, gfp_mask); - if (!buffer) { - _kfifo_init(fifo, NULL, 0); + fifo->data = kmalloc(size * esize, gfp_mask); + + if (!fifo->data) { + fifo->mask = 0; return -ENOMEM; } - - _kfifo_init(fifo, buffer, size); + fifo->mask = size - 1; return 0; } -EXPORT_SYMBOL(kfifo_alloc); +EXPORT_SYMBOL(__kfifo_alloc); -/** - * kfifo_free - frees the FIFO internal buffer - * @fifo: the fifo to be freed. - */ -void kfifo_free(struct kfifo *fifo) +void __kfifo_free(struct __kfifo *fifo) { - kfree(fifo->buffer); - _kfifo_init(fifo, NULL, 0); + kfree(fifo->data); + fifo->in = 0; + fifo->out = 0; + fifo->esize = 0; + fifo->data = NULL; + fifo->mask = 0; } -EXPORT_SYMBOL(kfifo_free); +EXPORT_SYMBOL(__kfifo_free); -/** - * kfifo_skip - skip output data - * @fifo: the fifo to be used. - * @len: number of bytes to skip - */ -void kfifo_skip(struct kfifo *fifo, unsigned int len) +int __kfifo_init(struct __kfifo *fifo, void *buffer, + unsigned int size, size_t esize) { - if (len < kfifo_len(fifo)) { - __kfifo_add_out(fifo, len); - return; + size /= esize; + + if (!is_power_of_2(size)) + size = rounddown_pow_of_two(size); + + fifo->in = 0; + fifo->out = 0; + fifo->esize = esize; + fifo->data = buffer; + + if (size < 2) { + fifo->mask = 0; + return -EINVAL; } - kfifo_reset_out(fifo); + fifo->mask = size - 1; + + return 0; } -EXPORT_SYMBOL(kfifo_skip); +EXPORT_SYMBOL(__kfifo_init); -static inline void __kfifo_in_data(struct kfifo *fifo, - const void *from, unsigned int len, unsigned int off) +static void kfifo_copy_in(struct __kfifo *fifo, const void *src, + unsigned int len, unsigned int off) { + unsigned int size = fifo->mask + 1; + unsigned int esize = fifo->esize; unsigned int l; + off &= fifo->mask; + if (esize != 1) { + off *= esize; + size *= esize; + len *= esize; + } + l = min(len, size - off); + + memcpy(fifo->data + off, src, l); + memcpy(fifo->data, src + l, len - l); /* - * Ensure that we sample the fifo->out index -before- we - * start putting bytes into the kfifo. + * make sure that the data in the fifo is up to date before + * incrementing the fifo->in index counter */ + smp_wmb(); +} - smp_mb(); - - off = __kfifo_off(fifo, fifo->in + off); +unsigned int __kfifo_in(struct __kfifo *fifo, + const void *buf, unsigned int len) +{ + unsigned int l; - /* first put the data starting from fifo->in to buffer end */ - l = min(len, fifo->size - off); - memcpy(fifo->buffer + off, from, l); + l = kfifo_unused(fifo); + if (len > l) + len = l; - /* then put the rest (if any) at the beginning of the buffer */ - memcpy(fifo->buffer, from + l, len - l); + kfifo_copy_in(fifo, buf, len, fifo->in); + fifo->in += len; + return len; } +EXPORT_SYMBOL(__kfifo_in); -static inline void __kfifo_out_data(struct kfifo *fifo, - void *to, unsigned int len, unsigned int off) +static void kfifo_copy_out(struct __kfifo *fifo, void *dst, + unsigned int len, unsigned int off) { + unsigned int size = fifo->mask + 1; + unsigned int esize = fifo->esize; unsigned int l; + off &= fifo->mask; + if (esize != 1) { + off *= esize; + size *= esize; + len *= esize; + } + l = min(len, size - off); + + memcpy(dst, fifo->data + off, l); + memcpy(dst + l, fifo->data, len - l); /* - * Ensure that we sample the fifo->in index -before- we - * start removing bytes from the kfifo. + * make sure that the data is copied before + * incrementing the fifo->out index counter */ + smp_wmb(); +} - smp_rmb(); +unsigned int __kfifo_out_peek(struct __kfifo *fifo, + void *buf, unsigned int len) +{ + unsigned int l; - off = __kfifo_off(fifo, fifo->out + off); + l = fifo->in - fifo->out; + if (len > l) + len = l; - /* first get the data from fifo->out until the end of the buffer */ - l = min(len, fifo->size - off); - memcpy(to, fifo->buffer + off, l); + kfifo_copy_out(fifo, buf, len, fifo->out); + return len; +} +EXPORT_SYMBOL(__kfifo_out_peek); - /* then get the rest (if any) from the beginning of the buffer */ - memcpy(to + l, fifo->buffer, len - l); +unsigned int __kfifo_out(struct __kfifo *fifo, + void *buf, unsigned int len) +{ + len = __kfifo_out_peek(fifo, buf, len); + fifo->out += len; + return len; } +EXPORT_SYMBOL(__kfifo_out); -static inline int __kfifo_from_user_data(struct kfifo *fifo, - const void __user *from, unsigned int len, unsigned int off, - unsigned *lenout) +static unsigned long kfifo_copy_from_user(struct __kfifo *fifo, + const void __user *from, unsigned int len, unsigned int off, + unsigned int *copied) { + unsigned int size = fifo->mask + 1; + unsigned int esize = fifo->esize; unsigned int l; - int ret; + unsigned long ret; + off &= fifo->mask; + if (esize != 1) { + off *= esize; + size *= esize; + len *= esize; + } + l = min(len, size - off); + + ret = copy_from_user(fifo->data + off, from, l); + if (unlikely(ret)) + ret = DIV_ROUND_UP(ret + len - l, esize); + else { + ret = copy_from_user(fifo->data, from + l, len - l); + if (unlikely(ret)) + ret = DIV_ROUND_UP(ret, esize); + } /* - * Ensure that we sample the fifo->out index -before- we - * start putting bytes into the kfifo. + * make sure that the data in the fifo is up to date before + * incrementing the fifo->in index counter */ + smp_wmb(); + *copied = len - ret; + /* return the number of elements which are not copied */ + return ret; +} - smp_mb(); +int __kfifo_from_user(struct __kfifo *fifo, const void __user *from, + unsigned long len, unsigned int *copied) +{ + unsigned int l; + unsigned long ret; + unsigned int esize = fifo->esize; + int err; - off = __kfifo_off(fifo, fifo->in + off); + if (esize != 1) + len /= esize; - /* first put the data starting from fifo->in to buffer end */ - l = min(len, fifo->size - off); - ret = copy_from_user(fifo->buffer + off, from, l); - if (unlikely(ret)) { - *lenout = ret; - return -EFAULT; - } - *lenout = l; + l = kfifo_unused(fifo); + if (len > l) + len = l; - /* then put the rest (if any) at the beginning of the buffer */ - ret = copy_from_user(fifo->buffer, from + l, len - l); - *lenout += ret ? ret : len - l; - return ret ? -EFAULT : 0; + ret = kfifo_copy_from_user(fifo, from, len, fifo->in, copied); + if (unlikely(ret)) { + len -= ret; + err = -EFAULT; + } else + err = 0; + fifo->in += len; + return err; } +EXPORT_SYMBOL(__kfifo_from_user); -static inline int __kfifo_to_user_data(struct kfifo *fifo, - void __user *to, unsigned int len, unsigned int off, unsigned *lenout) +static unsigned long kfifo_copy_to_user(struct __kfifo *fifo, void __user *to, + unsigned int len, unsigned int off, unsigned int *copied) { unsigned int l; - int ret; - + unsigned long ret; + unsigned int size = fifo->mask + 1; + unsigned int esize = fifo->esize; + + off &= fifo->mask; + if (esize != 1) { + off *= esize; + size *= esize; + len *= esize; + } + l = min(len, size - off); + + ret = copy_to_user(to, fifo->data + off, l); + if (unlikely(ret)) + ret = DIV_ROUND_UP(ret + len - l, esize); + else { + ret = copy_to_user(to + l, fifo->data, len - l); + if (unlikely(ret)) + ret = DIV_ROUND_UP(ret, esize); + } /* - * Ensure that we sample the fifo->in index -before- we - * start removing bytes from the kfifo. + * make sure that the data is copied before + * incrementing the fifo->out index counter */ + smp_wmb(); + *copied = len - ret; + /* return the number of elements which are not copied */ + return ret; +} - smp_rmb(); +int __kfifo_to_user(struct __kfifo *fifo, void __user *to, + unsigned long len, unsigned int *copied) +{ + unsigned int l; + unsigned long ret; + unsigned int esize = fifo->esize; + int err; - off = __kfifo_off(fifo, fifo->out + off); + if (esize != 1) + len /= esize; - /* first get the data from fifo->out until the end of the buffer */ - l = min(len, fifo->size - off); - ret = copy_to_user(to, fifo->buffer + off, l); - *lenout = l; + l = fifo->in - fifo->out; + if (len > l) + len = l; + ret = kfifo_copy_to_user(fifo, to, len, fifo->out, copied); if (unlikely(ret)) { - *lenout -= ret; - return -EFAULT; - } + len -= ret; + err = -EFAULT; + } else + err = 0; + fifo->out += len; + return err; +} +EXPORT_SYMBOL(__kfifo_to_user); - /* then get the rest (if any) from the beginning of the buffer */ - len -= l; - ret = copy_to_user(to + l, fifo->buffer, len); - if (unlikely(ret)) { - *lenout += len - ret; - return -EFAULT; +static int setup_sgl_buf(struct scatterlist *sgl, void *buf, + int nents, unsigned int len) +{ + int n; + unsigned int l; + unsigned int off; + struct page *page; + + if (!nents) + return 0; + + if (!len) + return 0; + + n = 0; + page = virt_to_page(buf); + off = offset_in_page(buf); + l = 0; + + while (len >= l + PAGE_SIZE - off) { + struct page *npage; + + l += PAGE_SIZE; + buf += PAGE_SIZE; + npage = virt_to_page(buf); + if (page_to_phys(page) != page_to_phys(npage) - l) { + sgl->page_link = 0; + sg_set_page(sgl++, page, l - off, off); + if (++n == nents) + return n; + page = npage; + len -= l - off; + l = off = 0; + } } - *lenout += len; - return 0; + sgl->page_link = 0; + sg_set_page(sgl++, page, len, off); + return n + 1; } -unsigned int __kfifo_in_n(struct kfifo *fifo, - const void *from, unsigned int len, unsigned int recsize) +static unsigned int setup_sgl(struct __kfifo *fifo, struct scatterlist *sgl, + int nents, unsigned int len, unsigned int off) { - if (kfifo_avail(fifo) < len + recsize) - return len + 1; + unsigned int size = fifo->mask + 1; + unsigned int esize = fifo->esize; + unsigned int l; + unsigned int n; - __kfifo_in_data(fifo, from, len, recsize); - return 0; + off &= fifo->mask; + if (esize != 1) { + off *= esize; + size *= esize; + len *= esize; + } + l = min(len, size - off); + + n = setup_sgl_buf(sgl, fifo->data + off, nents, l); + n += setup_sgl_buf(sgl + n, fifo->data, nents - n, len - l); + + if (n) + sg_mark_end(sgl + n - 1); + return n; } -EXPORT_SYMBOL(__kfifo_in_n); -/** - * kfifo_in - puts some data into the FIFO - * @fifo: the fifo to be used. - * @from: the data to be added. - * @len: the length of the data to be added. - * - * This function copies at most @len bytes from the @from buffer into - * the FIFO depending on the free space, and returns the number of - * bytes copied. - * - * Note that with only one concurrent reader and one concurrent - * writer, you don't need extra locking to use these functions. - */ -unsigned int kfifo_in(struct kfifo *fifo, const void *from, - unsigned int len) +unsigned int __kfifo_dma_in_prepare(struct __kfifo *fifo, + struct scatterlist *sgl, int nents, unsigned int len) { - len = min(kfifo_avail(fifo), len); + unsigned int l; - __kfifo_in_data(fifo, from, len, 0); - __kfifo_add_in(fifo, len); - return len; + l = kfifo_unused(fifo); + if (len > l) + len = l; + + return setup_sgl(fifo, sgl, nents, len, fifo->in); } -EXPORT_SYMBOL(kfifo_in); +EXPORT_SYMBOL(__kfifo_dma_in_prepare); -unsigned int __kfifo_in_generic(struct kfifo *fifo, - const void *from, unsigned int len, unsigned int recsize) +unsigned int __kfifo_dma_out_prepare(struct __kfifo *fifo, + struct scatterlist *sgl, int nents, unsigned int len) { - return __kfifo_in_rec(fifo, from, len, recsize); + unsigned int l; + + l = fifo->in - fifo->out; + if (len > l) + len = l; + + return setup_sgl(fifo, sgl, nents, len, fifo->out); } -EXPORT_SYMBOL(__kfifo_in_generic); +EXPORT_SYMBOL(__kfifo_dma_out_prepare); -unsigned int __kfifo_out_n(struct kfifo *fifo, - void *to, unsigned int len, unsigned int recsize) +unsigned int __kfifo_max_r(unsigned int len, size_t recsize) { - if (kfifo_len(fifo) < len + recsize) - return len; + unsigned int max = (1 << (recsize << 3)) - 1; - __kfifo_out_data(fifo, to, len, recsize); - __kfifo_add_out(fifo, len + recsize); - return 0; + if (len > max) + return max; + return len; } -EXPORT_SYMBOL(__kfifo_out_n); -/** - * kfifo_out - gets some data from the FIFO - * @fifo: the fifo to be used. - * @to: where the data must be copied. - * @len: the size of the destination buffer. - * - * This function copies at most @len bytes from the FIFO into the - * @to buffer and returns the number of copied bytes. - * - * Note that with only one concurrent reader and one concurrent - * writer, you don't need extra locking to use these functions. +#define __KFIFO_PEEK(data, out, mask) \ + ((data)[(out) & (mask)]) +/* + * __kfifo_peek_n internal helper function for determinate the length of + * the next record in the fifo */ -unsigned int kfifo_out(struct kfifo *fifo, void *to, unsigned int len) +static unsigned int __kfifo_peek_n(struct __kfifo *fifo, size_t recsize) { - len = min(kfifo_len(fifo), len); + unsigned int l; + unsigned int mask = fifo->mask; + unsigned char *data = fifo->data; - __kfifo_out_data(fifo, to, len, 0); - __kfifo_add_out(fifo, len); + l = __KFIFO_PEEK(data, fifo->out, mask); - return len; + if (--recsize) + l |= __KFIFO_PEEK(data, fifo->out + 1, mask) << 8; + + return l; } -EXPORT_SYMBOL(kfifo_out); - -/** - * kfifo_out_peek - copy some data from the FIFO, but do not remove it - * @fifo: the fifo to be used. - * @to: where the data must be copied. - * @len: the size of the destination buffer. - * @offset: offset into the fifo - * - * This function copies at most @len bytes at @offset from the FIFO - * into the @to buffer and returns the number of copied bytes. - * The data is not removed from the FIFO. + +#define __KFIFO_POKE(data, in, mask, val) \ + ( \ + (data)[(in) & (mask)] = (unsigned char)(val) \ + ) + +/* + * __kfifo_poke_n internal helper function for storeing the length of + * the record into the fifo */ -unsigned int kfifo_out_peek(struct kfifo *fifo, void *to, unsigned int len, - unsigned offset) +static void __kfifo_poke_n(struct __kfifo *fifo, unsigned int n, size_t recsize) { - len = min(kfifo_len(fifo), len + offset); + unsigned int mask = fifo->mask; + unsigned char *data = fifo->data; - __kfifo_out_data(fifo, to, len, offset); - return len; + __KFIFO_POKE(data, fifo->in, mask, n); + + if (recsize > 1) + __KFIFO_POKE(data, fifo->in + 1, mask, n >> 8); } -EXPORT_SYMBOL(kfifo_out_peek); -unsigned int __kfifo_out_generic(struct kfifo *fifo, - void *to, unsigned int len, unsigned int recsize, - unsigned int *total) +unsigned int __kfifo_len_r(struct __kfifo *fifo, size_t recsize) { - return __kfifo_out_rec(fifo, to, len, recsize, total); + return __kfifo_peek_n(fifo, recsize); } -EXPORT_SYMBOL(__kfifo_out_generic); +EXPORT_SYMBOL(__kfifo_len_r); -unsigned int __kfifo_from_user_n(struct kfifo *fifo, - const void __user *from, unsigned int len, unsigned int recsize) +unsigned int __kfifo_in_r(struct __kfifo *fifo, const void *buf, + unsigned int len, size_t recsize) { - unsigned total; + if (len + recsize > kfifo_unused(fifo)) + return 0; - if (kfifo_avail(fifo) < len + recsize) - return len + 1; + __kfifo_poke_n(fifo, len, recsize); - __kfifo_from_user_data(fifo, from, len, recsize, &total); - return total; + kfifo_copy_in(fifo, buf, len, fifo->in + recsize); + fifo->in += len + recsize; + return len; } -EXPORT_SYMBOL(__kfifo_from_user_n); - -/** - * kfifo_from_user - puts some data from user space into the FIFO - * @fifo: the fifo to be used. - * @from: pointer to the data to be added. - * @len: the length of the data to be added. - * @total: the actual returned data length. - * - * This function copies at most @len bytes from the @from into the - * FIFO depending and returns -EFAULT/0. - * - * Note that with only one concurrent reader and one concurrent - * writer, you don't need extra locking to use these functions. - */ -int kfifo_from_user(struct kfifo *fifo, - const void __user *from, unsigned int len, unsigned *total) +EXPORT_SYMBOL(__kfifo_in_r); + +static unsigned int kfifo_out_copy_r(struct __kfifo *fifo, + void *buf, unsigned int len, size_t recsize, unsigned int *n) { - int ret; - len = min(kfifo_avail(fifo), len); - ret = __kfifo_from_user_data(fifo, from, len, 0, total); - if (ret) - return ret; - __kfifo_add_in(fifo, len); - return 0; + *n = __kfifo_peek_n(fifo, recsize); + + if (len > *n) + len = *n; + + kfifo_copy_out(fifo, buf, len, fifo->out + recsize); + return len; +} + +unsigned int __kfifo_out_peek_r(struct __kfifo *fifo, void *buf, + unsigned int len, size_t recsize) +{ + unsigned int n; + + if (fifo->in == fifo->out) + return 0; + + return kfifo_out_copy_r(fifo, buf, len, recsize, &n); } -EXPORT_SYMBOL(kfifo_from_user); +EXPORT_SYMBOL(__kfifo_out_peek_r); -unsigned int __kfifo_from_user_generic(struct kfifo *fifo, - const void __user *from, unsigned int len, unsigned int recsize) +unsigned int __kfifo_out_r(struct __kfifo *fifo, void *buf, + unsigned int len, size_t recsize) { - return __kfifo_from_user_rec(fifo, from, len, recsize); + unsigned int n; + + if (fifo->in == fifo->out) + return 0; + + len = kfifo_out_copy_r(fifo, buf, len, recsize, &n); + fifo->out += n + recsize; + return len; } -EXPORT_SYMBOL(__kfifo_from_user_generic); +EXPORT_SYMBOL(__kfifo_out_r); -unsigned int __kfifo_to_user_n(struct kfifo *fifo, - void __user *to, unsigned int len, unsigned int reclen, - unsigned int recsize) +int __kfifo_from_user_r(struct __kfifo *fifo, const void __user *from, + unsigned long len, unsigned int *copied, size_t recsize) { - unsigned int ret, total; + unsigned long ret; - if (kfifo_len(fifo) < reclen + recsize) - return len; + len = __kfifo_max_r(len, recsize); - ret = __kfifo_to_user_data(fifo, to, reclen, recsize, &total); + if (len + recsize > kfifo_unused(fifo)) { + *copied = 0; + return 0; + } - if (likely(ret == 0)) - __kfifo_add_out(fifo, reclen + recsize); + __kfifo_poke_n(fifo, len, recsize); - return total; + ret = kfifo_copy_from_user(fifo, from, len, fifo->in + recsize, copied); + if (unlikely(ret)) { + *copied = 0; + return -EFAULT; + } + fifo->in += len + recsize; + return 0; } -EXPORT_SYMBOL(__kfifo_to_user_n); - -/** - * kfifo_to_user - gets data from the FIFO and write it to user space - * @fifo: the fifo to be used. - * @to: where the data must be copied. - * @len: the size of the destination buffer. - * @lenout: pointer to output variable with copied data - * - * This function copies at most @len bytes from the FIFO into the - * @to buffer and 0 or -EFAULT. - * - * Note that with only one concurrent reader and one concurrent - * writer, you don't need extra locking to use these functions. - */ -int kfifo_to_user(struct kfifo *fifo, - void __user *to, unsigned int len, unsigned *lenout) +EXPORT_SYMBOL(__kfifo_from_user_r); + +int __kfifo_to_user_r(struct __kfifo *fifo, void __user *to, + unsigned long len, unsigned int *copied, size_t recsize) { - int ret; - len = min(kfifo_len(fifo), len); - ret = __kfifo_to_user_data(fifo, to, len, 0, lenout); - __kfifo_add_out(fifo, *lenout); - return ret; + unsigned long ret; + unsigned int n; + + if (fifo->in == fifo->out) { + *copied = 0; + return 0; + } + + n = __kfifo_peek_n(fifo, recsize); + if (len > n) + len = n; + + ret = kfifo_copy_to_user(fifo, to, len, fifo->out + recsize, copied); + if (unlikely(ret)) { + *copied = 0; + return -EFAULT; + } + fifo->out += n + recsize; + return 0; } -EXPORT_SYMBOL(kfifo_to_user); +EXPORT_SYMBOL(__kfifo_to_user_r); -unsigned int __kfifo_to_user_generic(struct kfifo *fifo, - void __user *to, unsigned int len, unsigned int recsize, - unsigned int *total) +unsigned int __kfifo_dma_in_prepare_r(struct __kfifo *fifo, + struct scatterlist *sgl, int nents, unsigned int len, size_t recsize) { - return __kfifo_to_user_rec(fifo, to, len, recsize, total); + if (!nents) + BUG(); + + len = __kfifo_max_r(len, recsize); + + if (len + recsize > kfifo_unused(fifo)) + return 0; + + return setup_sgl(fifo, sgl, nents, len, fifo->in + recsize); } -EXPORT_SYMBOL(__kfifo_to_user_generic); +EXPORT_SYMBOL(__kfifo_dma_in_prepare_r); -unsigned int __kfifo_peek_generic(struct kfifo *fifo, unsigned int recsize) +void __kfifo_dma_in_finish_r(struct __kfifo *fifo, + unsigned int len, size_t recsize) { - if (recsize == 0) - return kfifo_avail(fifo); - - return __kfifo_peek_n(fifo, recsize); + len = __kfifo_max_r(len, recsize); + __kfifo_poke_n(fifo, len, recsize); + fifo->in += len + recsize; } -EXPORT_SYMBOL(__kfifo_peek_generic); +EXPORT_SYMBOL(__kfifo_dma_in_finish_r); -void __kfifo_skip_generic(struct kfifo *fifo, unsigned int recsize) +unsigned int __kfifo_dma_out_prepare_r(struct __kfifo *fifo, + struct scatterlist *sgl, int nents, unsigned int len, size_t recsize) { - __kfifo_skip_rec(fifo, recsize); + if (!nents) + BUG(); + + len = __kfifo_max_r(len, recsize); + + if (len + recsize > fifo->in - fifo->out) + return 0; + + return setup_sgl(fifo, sgl, nents, len, fifo->out + recsize); } -EXPORT_SYMBOL(__kfifo_skip_generic); +EXPORT_SYMBOL(__kfifo_dma_out_prepare_r); + +void __kfifo_dma_out_finish_r(struct __kfifo *fifo, size_t recsize) +{ + unsigned int len; + len = __kfifo_peek_n(fifo, recsize); + fifo->out += len + recsize; +} +EXPORT_SYMBOL(__kfifo_dma_out_finish_r); -- cgit v1.2.3 From d78a3eda6985e74bc21a23362f27526f73e71649 Mon Sep 17 00:00:00 2001 From: Stefani Seibold Date: Wed, 11 Aug 2010 14:17:27 -0700 Subject: kernel/kfifo.c: add handling of chained scatterlists The current kfifo scatterlist implementation will not work with chained scatterlists. It assumes that struct scatterlist arrays are allocated contiguously, which is not the case when chained scatterlists (struct sg_table) are in use. Signed-off-by: Stefani Seibold Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/kfifo.c | 13 ++++++------- 1 file changed, 6 insertions(+), 7 deletions(-) (limited to 'kernel') diff --git a/kernel/kfifo.c b/kernel/kfifo.c index 02192dd905cc..4502604ecadf 100644 --- a/kernel/kfifo.c +++ b/kernel/kfifo.c @@ -10,7 +10,7 @@ * * This program is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the * GNU General Public License for more details. * * You should have received a copy of the GNU General Public License @@ -333,17 +333,16 @@ static int setup_sgl_buf(struct scatterlist *sgl, void *buf, buf += PAGE_SIZE; npage = virt_to_page(buf); if (page_to_phys(page) != page_to_phys(npage) - l) { - sgl->page_link = 0; - sg_set_page(sgl++, page, l - off, off); - if (++n == nents) + sg_set_page(sgl, page, l - off, off); + sgl = sg_next(sgl); + if (++n == nents || sgl == NULL) return n; page = npage; len -= l - off; l = off = 0; } } - sgl->page_link = 0; - sg_set_page(sgl++, page, len, off); + sg_set_page(sgl, page, len, off); return n + 1; } @@ -363,7 +362,7 @@ static unsigned int setup_sgl(struct __kfifo *fifo, struct scatterlist *sgl, } l = min(len, size - off); - n = setup_sgl_buf(sgl, fifo->data + off, nents, l); + n = setup_sgl_buf(sgl, fifo->data + off, nents, l); n += setup_sgl_buf(sgl + n, fifo->data, nents - n, len - l); if (n) -- cgit v1.2.3 From 8d57a98ccd0b4489003473979da8f5a1363ba7a3 Mon Sep 17 00:00:00 2001 From: Adrian Hunter Date: Wed, 11 Aug 2010 14:17:49 -0700 Subject: block: add secure discard Secure discard is the same as discard except that all copies of the discarded sectors (perhaps created by garbage collection) must also be erased. Signed-off-by: Adrian Hunter Acked-by: Jens Axboe Cc: Kyungmin Park Cc: Madhusudhan Chikkature Cc: Christoph Hellwig Cc: Ben Gardiner Cc: Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/trace/blktrace.c | 8 ++++++++ 1 file changed, 8 insertions(+) (limited to 'kernel') diff --git a/kernel/trace/blktrace.c b/kernel/trace/blktrace.c index 82499a5bdcb7..959f8d6c8cc1 100644 --- a/kernel/trace/blktrace.c +++ b/kernel/trace/blktrace.c @@ -710,6 +710,9 @@ static void blk_add_trace_rq(struct request_queue *q, struct request *rq, if (rq->cmd_flags & REQ_DISCARD) rw |= REQ_DISCARD; + if (rq->cmd_flags & REQ_SECURE) + rw |= REQ_SECURE; + if (rq->cmd_type == REQ_TYPE_BLOCK_PC) { what |= BLK_TC_ACT(BLK_TC_PC); __blk_add_trace(bt, 0, blk_rq_bytes(rq), rw, @@ -1816,6 +1819,8 @@ void blk_fill_rwbs(char *rwbs, u32 rw, int bytes) rwbs[i++] = 'S'; if (rw & REQ_META) rwbs[i++] = 'M'; + if (rw & REQ_SECURE) + rwbs[i++] = 'E'; rwbs[i] = '\0'; } @@ -1828,6 +1833,9 @@ void blk_fill_rwbs_rq(char *rwbs, struct request *rq) if (rq->cmd_flags & REQ_DISCARD) rw |= REQ_DISCARD; + if (rq->cmd_flags & REQ_SECURE) + rw |= REQ_SECURE; + bytes = blk_rq_bytes(rq); blk_fill_rwbs(rwbs, rw, bytes); -- cgit v1.2.3 From 12fdff3fc2483f906ae6404a6e8dcf2550310b6f Mon Sep 17 00:00:00 2001 From: David Howells Date: Thu, 12 Aug 2010 16:54:57 +0100 Subject: Add a dummy printk function for the maintenance of unused printks Add a dummy printk function for the maintenance of unused printks through gcc format checking, and also so that side-effect checking is maintained too. Signed-off-by: David Howells Signed-off-by: Linus Torvalds --- kernel/cred.c | 4 ---- 1 file changed, 4 deletions(-) (limited to 'kernel') diff --git a/kernel/cred.c b/kernel/cred.c index 60bc8b1e32e6..9a3e22641fe7 100644 --- a/kernel/cred.c +++ b/kernel/cred.c @@ -22,10 +22,6 @@ #define kdebug(FMT, ...) \ printk("[%-5.5s%5u] "FMT"\n", current->comm, current->pid ,##__VA_ARGS__) #else -static inline __attribute__((format(printf, 1, 2))) -void no_printk(const char *fmt, ...) -{ -} #define kdebug(FMT, ...) \ no_printk("[%-5.5s%5u] "FMT"\n", current->comm, current->pid ,##__VA_ARGS__) #endif -- cgit v1.2.3 From deda2e81961e96be4f2c09328baca4710a2fd1a0 Mon Sep 17 00:00:00 2001 From: Jason Wessel Date: Mon, 9 Aug 2010 14:20:09 -0700 Subject: timekeeping: Fix overflow in rawtime tv_nsec on 32 bit archs The tv_nsec is a long and when added to the shifted interval it can wrap and become negative which later causes looping problems in the getrawmonotonic(). The edge case occurs when the system has slept for a short period of time of ~2 seconds. A trace printk of the values in this patch illustrate the problem: ftrace time stamp: log 43.716079: logarithmic_accumulation: raw: 3d0913 tv_nsec d687faa 43.718513: logarithmic_accumulation: raw: 3d0913 tv_nsec da588bd 43.722161: logarithmic_accumulation: raw: 3d0913 tv_nsec de291d0 46.349925: logarithmic_accumulation: raw: 7a122600 tv_nsec e1f9ae3 46.349930: logarithmic_accumulation: raw: 1e848980 tv_nsec 8831c0e3 The kernel starts looping at 46.349925 in the getrawmonotonic() due to the negative value from adding the raw value to tv_nsec. A simple solution is to accumulate into a u64, and then normalize it to a timespec_t. Signed-off-by: Jason Wessel [ Reworked variable names and simplified some of the code. - John ] Signed-off-by: John Stultz Cc: Thomas Gleixner Cc: H. Peter Anvin Signed-off-by: Linus Torvalds --- kernel/time/timekeeping.c | 11 +++++++---- 1 file changed, 7 insertions(+), 4 deletions(-) (limited to 'kernel') diff --git a/kernel/time/timekeeping.c b/kernel/time/timekeeping.c index e14c839e9faa..e960d824263f 100644 --- a/kernel/time/timekeeping.c +++ b/kernel/time/timekeeping.c @@ -690,6 +690,7 @@ static void timekeeping_adjust(s64 offset) static cycle_t logarithmic_accumulation(cycle_t offset, int shift) { u64 nsecps = (u64)NSEC_PER_SEC << timekeeper.shift; + u64 raw_nsecs; /* If the offset is smaller then a shifted interval, do nothing */ if (offset < timekeeper.cycle_interval<= NSEC_PER_SEC) { - raw_time.tv_nsec -= NSEC_PER_SEC; + /* Accumulate raw time */ + raw_nsecs = timekeeper.raw_interval << shift; + raw_nsecs += raw_time.tv_nsec; + while (raw_nsecs >= NSEC_PER_SEC) { + raw_nsecs -= NSEC_PER_SEC; raw_time.tv_sec++; } + raw_time.tv_nsec = raw_nsecs; /* Accumulate error between NTP and clock interval */ timekeeper.ntp_error += tick_length << shift; -- cgit v1.2.3 From 2a37a3df57c44e947271758a1aa4bea7bff9feab Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Thu, 3 Jun 2010 15:21:34 -0400 Subject: tracing/events: Convert format output to seq_file Two new events were added that broke the current format output. Both from the SCSI system: scsi_dispatch_cmd_done and scsi_dispatch_cmd_timeout The reason is that their print_fmt exceeded a page size. Since the output of the format used simple_read_from_buffer and trace_seq, it was limited to a page size in output. This patch converts the printing of the format of an event into seq_file, which allows greater than a page size to be shown. I diffed all event formats comparing the output with and without this patch. All matched except for the above two, which showed just: FORMAT TOO BIG without this patch, but now properly displays the output with this patch. v2: Remove updating *pos in seq start function. [ Thanks to Li Zefan for pointing that out ] Reviewed-by: Li Zefan Cc: Martin K. Petersen Cc: Kei Tokunaga Cc: James Bottomley Cc: Tomohiro Kusumi Cc: Xiao Guangrong Signed-off-by: Steven Rostedt --- kernel/trace/trace_events.c | 208 ++++++++++++++++++++++++++++++-------------- 1 file changed, 141 insertions(+), 67 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace_events.c b/kernel/trace/trace_events.c index 53cffc0b0801..45a8968707aa 100644 --- a/kernel/trace/trace_events.c +++ b/kernel/trace/trace_events.c @@ -29,6 +29,8 @@ DEFINE_MUTEX(event_mutex); LIST_HEAD(ftrace_events); +#define COMMON_FIELD_COUNT 5 + struct list_head * trace_get_fields(struct ftrace_event_call *event_call) { @@ -544,85 +546,155 @@ out: return ret; } -static ssize_t -event_format_read(struct file *filp, char __user *ubuf, size_t cnt, - loff_t *ppos) +enum { + FORMAT_HEADER = 1, + FORMAT_PRINTFMT = 2, +}; + +static void *f_next(struct seq_file *m, void *v, loff_t *pos) { - struct ftrace_event_call *call = filp->private_data; + struct ftrace_event_call *call = m->private; struct ftrace_event_field *field; struct list_head *head; - struct trace_seq *s; - int common_field_count = 5; - char *buf; - int r = 0; - - if (*ppos) - return 0; + loff_t index = *pos; - s = kmalloc(sizeof(*s), GFP_KERNEL); - if (!s) - return -ENOMEM; + (*pos)++; - trace_seq_init(s); + head = trace_get_fields(call); - trace_seq_printf(s, "name: %s\n", call->name); - trace_seq_printf(s, "ID: %d\n", call->event.type); - trace_seq_printf(s, "format:\n"); + switch ((unsigned long)v) { + case FORMAT_HEADER: - head = trace_get_fields(call); - list_for_each_entry_reverse(field, head, link) { - /* - * Smartly shows the array type(except dynamic array). - * Normal: - * field:TYPE VAR - * If TYPE := TYPE[LEN], it is shown: - * field:TYPE VAR[LEN] - */ - const char *array_descriptor = strchr(field->type, '['); - - if (!strncmp(field->type, "__data_loc", 10)) - array_descriptor = NULL; - - if (!array_descriptor) { - r = trace_seq_printf(s, "\tfield:%s %s;\toffset:%u;" - "\tsize:%u;\tsigned:%d;\n", - field->type, field->name, field->offset, - field->size, !!field->is_signed); - } else { - r = trace_seq_printf(s, "\tfield:%.*s %s%s;\toffset:%u;" - "\tsize:%u;\tsigned:%d;\n", - (int)(array_descriptor - field->type), - field->type, field->name, - array_descriptor, field->offset, - field->size, !!field->is_signed); - } + if (unlikely(list_empty(head))) + return NULL; - if (--common_field_count == 0) - r = trace_seq_printf(s, "\n"); + field = list_entry(head->prev, struct ftrace_event_field, link); + return field; - if (!r) - break; + case FORMAT_PRINTFMT: + /* all done */ + return NULL; } - if (r) - r = trace_seq_printf(s, "\nprint fmt: %s\n", - call->print_fmt); + /* + * To separate common fields from event fields, the + * LSB is set on the first event field. Clear it in case. + */ + v = (void *)((unsigned long)v & ~1L); - if (!r) { - /* - * ug! The format output is bigger than a PAGE!! - */ - buf = "FORMAT TOO BIG\n"; - r = simple_read_from_buffer(ubuf, cnt, ppos, - buf, strlen(buf)); - goto out; + field = v; + if (field->link.prev == head) + return (void *)FORMAT_PRINTFMT; + + field = list_entry(field->link.prev, struct ftrace_event_field, link); + + /* Set the LSB to notify f_show to print an extra newline */ + if (index == COMMON_FIELD_COUNT) + field = (struct ftrace_event_field *) + ((unsigned long)field | 1); + + return field; +} + +static void *f_start(struct seq_file *m, loff_t *pos) +{ + loff_t l = 0; + void *p; + + /* Start by showing the header */ + if (!*pos) + return (void *)FORMAT_HEADER; + + p = (void *)FORMAT_HEADER; + do { + p = f_next(m, p, &l); + } while (p && l < *pos); + + return p; +} + +static int f_show(struct seq_file *m, void *v) +{ + struct ftrace_event_call *call = m->private; + struct ftrace_event_field *field; + const char *array_descriptor; + + switch ((unsigned long)v) { + case FORMAT_HEADER: + seq_printf(m, "name: %s\n", call->name); + seq_printf(m, "ID: %d\n", call->event.type); + seq_printf(m, "format:\n"); + return 0; + + case FORMAT_PRINTFMT: + seq_printf(m, "\nprint fmt: %s\n", + call->print_fmt); + return 0; } - r = simple_read_from_buffer(ubuf, cnt, ppos, - s->buffer, s->len); - out: - kfree(s); - return r; + /* + * To separate common fields from event fields, the + * LSB is set on the first event field. Clear it and + * print a newline if it is set. + */ + if ((unsigned long)v & 1) { + seq_putc(m, '\n'); + v = (void *)((unsigned long)v & ~1L); + } + + field = v; + + /* + * Smartly shows the array type(except dynamic array). + * Normal: + * field:TYPE VAR + * If TYPE := TYPE[LEN], it is shown: + * field:TYPE VAR[LEN] + */ + array_descriptor = strchr(field->type, '['); + + if (!strncmp(field->type, "__data_loc", 10)) + array_descriptor = NULL; + + if (!array_descriptor) + seq_printf(m, "\tfield:%s %s;\toffset:%u;\tsize:%u;\tsigned:%d;\n", + field->type, field->name, field->offset, + field->size, !!field->is_signed); + else + seq_printf(m, "\tfield:%.*s %s%s;\toffset:%u;\tsize:%u;\tsigned:%d;\n", + (int)(array_descriptor - field->type), + field->type, field->name, + array_descriptor, field->offset, + field->size, !!field->is_signed); + + return 0; +} + +static void f_stop(struct seq_file *m, void *p) +{ +} + +static const struct seq_operations trace_format_seq_ops = { + .start = f_start, + .next = f_next, + .stop = f_stop, + .show = f_show, +}; + +static int trace_format_open(struct inode *inode, struct file *file) +{ + struct ftrace_event_call *call = inode->i_private; + struct seq_file *m; + int ret; + + ret = seq_open(file, &trace_format_seq_ops); + if (ret < 0) + return ret; + + m = file->private_data; + m->private = call; + + return 0; } static ssize_t @@ -820,8 +892,10 @@ static const struct file_operations ftrace_enable_fops = { }; static const struct file_operations ftrace_event_format_fops = { - .open = tracing_open_generic, - .read = event_format_read, + .open = trace_format_open, + .read = seq_read, + .llseek = seq_lseek, + .release = seq_release, }; static const struct file_operations ftrace_event_id_fops = { -- cgit v1.2.3 From 2069601b3f0ea38170d4b509b89f3ca0a373bdc1 Mon Sep 17 00:00:00 2001 From: Linus Torvalds Date: Thu, 12 Aug 2010 14:23:04 -0700 Subject: Revert "fsnotify: store struct file not struct path" This reverts commit 3bcf3860a4ff9bbc522820b4b765e65e4deceb3e (and the accompanying commit c1e5c954020e "vfs/fsnotify: fsnotify_close can delay the final work in fput" that was a horribly ugly hack to make it work at all). The 'struct file' approach not only causes that disgusting hack, it somehow breaks pulseaudio, probably due to some other subtlety with f_count handling. Fix up various conflicts due to later fsnotify work. Signed-off-by: Linus Torvalds --- kernel/audit_watch.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/audit_watch.c b/kernel/audit_watch.c index 6bf2306be7d6..f0c9b2e7542d 100644 --- a/kernel/audit_watch.c +++ b/kernel/audit_watch.c @@ -526,8 +526,8 @@ static int audit_watch_handle_event(struct fsnotify_group *group, BUG_ON(group != audit_watch_group); switch (event->data_type) { - case (FSNOTIFY_EVENT_FILE): - inode = event->file->f_path.dentry->d_inode; + case (FSNOTIFY_EVENT_PATH): + inode = event->path.dentry->d_inode; break; case (FSNOTIFY_EVENT_INODE): inode = event->inode; -- cgit v1.2.3 From c7dcf87a6881bf796faee83003163eb3de41a309 Mon Sep 17 00:00:00 2001 From: John Stultz Date: Fri, 13 Aug 2010 11:30:58 -0700 Subject: time: Workaround gcc loop optimization that causes 64bit div errors Early 4.3 versions of gcc apparently aggressively optimize the raw time accumulation loop, replacing it with a divide. On 32bit systems, this causes the following link errors: undefined reference to `__umoddi3' undefined reference to `__udivdi3' The gcc issue has been fixed in 4.4 and greater. This patch replaces the accumulation loop with a do_div, as suggested by Linus. Signed-off-by: John Stultz CC: Jason Wessel CC: Larry Finger CC: Ingo Molnar CC: Linus Torvalds Signed-off-by: Linus Torvalds --- kernel/time/timekeeping.c | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/time/timekeeping.c b/kernel/time/timekeeping.c index e960d824263f..49010d822f72 100644 --- a/kernel/time/timekeeping.c +++ b/kernel/time/timekeeping.c @@ -710,9 +710,10 @@ static cycle_t logarithmic_accumulation(cycle_t offset, int shift) /* Accumulate raw time */ raw_nsecs = timekeeper.raw_interval << shift; raw_nsecs += raw_time.tv_nsec; - while (raw_nsecs >= NSEC_PER_SEC) { - raw_nsecs -= NSEC_PER_SEC; - raw_time.tv_sec++; + if (raw_nsecs >= NSEC_PER_SEC) { + u64 raw_secs = raw_nsecs; + raw_nsecs = do_div(raw_secs, NSEC_PER_SEC); + raw_time.tv_sec += raw_secs; } raw_time.tv_nsec = raw_nsecs; -- cgit v1.2.3 From 1aa54bca6ee0d07ebcafb8ca8074b624d80724aa Mon Sep 17 00:00:00 2001 From: Marcin Slusarz Date: Wed, 28 Jul 2010 01:18:01 +0200 Subject: tracing: Sanitize value returned from write(trace_marker, "...", len) When userspace code writes non-new-line-terminated string to trace_marker file, write handler appends new-line and returns number of bytes written to trace buffer, so write(fd, "abc", 3) will return 4 That's unexpected and unfortunately it confuses glibc's fprintf function. Example: int main() { fprintf(stderr, "abc"); return 0; } $ gcc test.c -o test $ echo mmiotrace > /sys/kernel/debug/tracing/current_tracer $ ./test 2>/sys/kernel/debug/tracing/trace_marker results in infinite loop: write(fd, "abc", 3) = 4 write(fd, "", 1) = 0 write(fd, "", 1) = 0 write(fd, "", 1) = 0 write(fd, "", 1) = 0 write(fd, "", 1) = 0 write(fd, "", 1) = 0 write(fd, "", 1) = 0 (...) ...and kernel trace buffer full of empty markers. Fix it by sanitizing write return value. Signed-off-by: Marcin Slusarz LKML-Reference: <20100727231801.GB2826@joi.lan> Cc: Frederic Weisbecker Cc: Ingo Molnar Signed-off-by: Steven Rostedt --- kernel/trace/trace.c | 11 ++++++++--- 1 file changed, 8 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 086d36316805..88b42d14d32d 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -3498,6 +3498,7 @@ tracing_mark_write(struct file *filp, const char __user *ubuf, size_t cnt, loff_t *fpos) { char *buf; + size_t written; if (tracing_disabled) return -EINVAL; @@ -3519,11 +3520,15 @@ tracing_mark_write(struct file *filp, const char __user *ubuf, } else buf[cnt] = '\0'; - cnt = mark_printk("%s", buf); + written = mark_printk("%s", buf); kfree(buf); - *fpos += cnt; + *fpos += written; - return cnt; + /* don't tell userspace we wrote more - it might confuse them */ + if (written > cnt) + written = cnt; + + return written; } static int tracing_clock_show(struct seq_file *m, void *v) -- cgit v1.2.3 From 8d9df9f0844ed87541453a3ef91bfc9f487053b7 Mon Sep 17 00:00:00 2001 From: Xiaotian Feng Date: Mon, 16 Aug 2010 09:54:28 +0200 Subject: workqueue: free rescuer on destroy_workqueue wq->rescuer is not freed when wq is destroyed, leads a memory leak then. This patch also remove a redundant line. Signed-off-by: Xiaotian Feng Signed-off-by: Tejun Heo Cc: Oleg Nesterov --- kernel/workqueue.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/workqueue.c b/kernel/workqueue.c index 2994a0e3a61c..1001b6e3fcbd 100644 --- a/kernel/workqueue.c +++ b/kernel/workqueue.c @@ -2782,7 +2782,6 @@ struct workqueue_struct *__alloc_workqueue_key(const char *name, if (IS_ERR(rescuer->task)) goto err; - wq->rescuer = rescuer; rescuer->task->flags |= PF_THREAD_BOUND; wake_up_process(rescuer->task); } @@ -2848,6 +2847,7 @@ void destroy_workqueue(struct workqueue_struct *wq) if (wq->flags & WQ_RESCUER) { kthread_stop(wq->rescuer->task); free_mayday_mask(wq->mayday_mask); + kfree(wq->rescuer); } free_cwqs(wq); -- cgit v1.2.3 From b590cddfa6f40447158323b43a13cdae01d9a051 Mon Sep 17 00:00:00 2001 From: Jason Wessel Date: Mon, 16 Aug 2010 15:58:29 -0500 Subject: kdb: fix compile error without CONFIG_KALLSYMS If CONFIG_KGDB_KDB is set and CONFIG_KALLSYMS is not set the kernel will fail to build with the error: kernel/built-in.o: In function `kallsyms_symbol_next': kernel/debug/kdb/kdb_support.c:237: undefined reference to `kdb_walk_kallsyms' kernel/built-in.o: In function `kallsyms_symbol_complete': kernel/debug/kdb/kdb_support.c:193: undefined reference to `kdb_walk_kallsyms' The kdb_walk_kallsyms needs a #ifdef proper header to match the C implementation. This patch also fixes the compiler warnings in kdb_support.c when compiling without CONFIG_KALLSYMS set. The compiler warnings are a result of the kallsyms_lookup() macro not initializing the two of the pass by reference variables. Signed-off-by: Jason Wessel Reported-by: Michal Simek --- kernel/debug/kdb/kdb_private.h | 7 +++++++ kernel/debug/kdb/kdb_support.c | 4 ++-- 2 files changed, 9 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/debug/kdb/kdb_private.h b/kernel/debug/kdb/kdb_private.h index c438f545a321..be775f7e81e0 100644 --- a/kernel/debug/kdb/kdb_private.h +++ b/kernel/debug/kdb/kdb_private.h @@ -255,7 +255,14 @@ extern void kdb_ps1(const struct task_struct *p); extern void kdb_print_nameval(const char *name, unsigned long val); extern void kdb_send_sig_info(struct task_struct *p, struct siginfo *info); extern void kdb_meminfo_proc_show(void); +#ifdef CONFIG_KALLSYMS extern const char *kdb_walk_kallsyms(loff_t *pos); +#else /* ! CONFIG_KALLSYMS */ +static inline const char *kdb_walk_kallsyms(loff_t *pos) +{ + return NULL; +} +#endif /* ! CONFIG_KALLSYMS */ extern char *kdb_getstr(char *, size_t, char *); /* Defines for kdb_symbol_print */ diff --git a/kernel/debug/kdb/kdb_support.c b/kernel/debug/kdb/kdb_support.c index 45344d5c53dd..6b2485dcb050 100644 --- a/kernel/debug/kdb/kdb_support.c +++ b/kernel/debug/kdb/kdb_support.c @@ -82,8 +82,8 @@ static char *kdb_name_table[100]; /* arbitrary size */ int kdbnearsym(unsigned long addr, kdb_symtab_t *symtab) { int ret = 0; - unsigned long symbolsize; - unsigned long offset; + unsigned long symbolsize = 0; + unsigned long offset = 0; #define knt1_size 128 /* must be >= kallsyms table size */ char *knt1 = NULL; -- cgit v1.2.3 From f1b499f029c5dde85d46a8811353c62f29157541 Mon Sep 17 00:00:00 2001 From: John Kacur Date: Thu, 5 Aug 2010 17:10:53 +0200 Subject: lockdep: Remove __debug_show_held_locks There is no longer any functional difference between __debug_show_held_locks() and debug_show_held_locks(), so remove the former. Signed-off-by: John Kacur Cc: Peter Zijlstra LKML-Reference: <1281021054-4228-1-git-send-email-jkacur@redhat.com> Signed-off-by: Ingo Molnar --- kernel/hung_task.c | 2 +- kernel/lockdep.c | 8 +------- 2 files changed, 2 insertions(+), 8 deletions(-) (limited to 'kernel') diff --git a/kernel/hung_task.c b/kernel/hung_task.c index 0c642d51aac2..bca942379559 100644 --- a/kernel/hung_task.c +++ b/kernel/hung_task.c @@ -98,7 +98,7 @@ static void check_hung_task(struct task_struct *t, unsigned long timeout) printk(KERN_ERR "\"echo 0 > /proc/sys/kernel/hung_task_timeout_secs\"" " disables this message.\n"); sched_show_task(t); - __debug_show_held_locks(t); + debug_show_held_locks(t); touch_nmi_watchdog(); diff --git a/kernel/lockdep.c b/kernel/lockdep.c index f2852a510232..84baa71cfda5 100644 --- a/kernel/lockdep.c +++ b/kernel/lockdep.c @@ -3775,7 +3775,7 @@ EXPORT_SYMBOL_GPL(debug_show_all_locks); * Careful: only use this function if you are sure that * the task cannot run in parallel! */ -void __debug_show_held_locks(struct task_struct *task) +void debug_show_held_locks(struct task_struct *task) { if (unlikely(!debug_locks)) { printk("INFO: lockdep is turned off.\n"); @@ -3783,12 +3783,6 @@ void __debug_show_held_locks(struct task_struct *task) } lockdep_print_held_locks(task); } -EXPORT_SYMBOL_GPL(__debug_show_held_locks); - -void debug_show_held_locks(struct task_struct *task) -{ - __debug_show_held_locks(task); -} EXPORT_SYMBOL_GPL(debug_show_held_locks); void lockdep_sys_exit(void) -- cgit v1.2.3 From 6a103b0d44e9f97dc430002cf3ac7a7defa3819f Mon Sep 17 00:00:00 2001 From: John Kacur Date: Thu, 5 Aug 2010 17:10:54 +0200 Subject: lockup detector: Fix grammar by adding a missing "to" in the comments This fixes a minor grammar problem in the comments in hung_task.c Signed-off-by: John Kacur Cc: Peter Zijlstra LKML-Reference: <1281021054-4228-2-git-send-email-jkacur@redhat.com> Signed-off-by: Ingo Molnar --- kernel/hung_task.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/hung_task.c b/kernel/hung_task.c index bca942379559..53ead174da2f 100644 --- a/kernel/hung_task.c +++ b/kernel/hung_task.c @@ -111,7 +111,7 @@ static void check_hung_task(struct task_struct *t, unsigned long timeout) * periodically exit the critical section and enter a new one. * * For preemptible RCU it is sufficient to call rcu_read_unlock in order - * exit the grace period. For classic RCU, a reschedule is required. + * to exit the grace period. For classic RCU, a reschedule is required. */ static void rcu_lock_break(struct task_struct *g, struct task_struct *t) { -- cgit v1.2.3 From d7627467b7a8dd6944885290a03a07ceb28c10eb Mon Sep 17 00:00:00 2001 From: David Howells Date: Tue, 17 Aug 2010 23:52:56 +0100 Subject: Make do_execve() take a const filename pointer Make do_execve() take a const filename pointer so that kernel_execve() compiles correctly on ARM: arch/arm/kernel/sys_arm.c:88: warning: passing argument 1 of 'do_execve' discards qualifiers from pointer target type This also requires the argv and envp arguments to be consted twice, once for the pointer array and once for the strings the array points to. This is because do_execve() passes a pointer to the filename (now const) to copy_strings_kernel(). A simpler alternative would be to cast the filename pointer in do_execve() when it's passed to copy_strings_kernel(). do_execve() may not change any of the strings it is passed as part of the argv or envp lists as they are some of them in .rodata, so marking these strings as const should be fine. Further kernel_execve() and sys_execve() need to be changed to match. This has been test built on x86_64, frv, arm and mips. Signed-off-by: David Howells Tested-by: Ralf Baechle Acked-by: Russell King Signed-off-by: Linus Torvalds --- kernel/kmod.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/kmod.c b/kernel/kmod.c index 6e9b19667a8d..9cd0591c96a2 100644 --- a/kernel/kmod.c +++ b/kernel/kmod.c @@ -153,7 +153,9 @@ static int ____call_usermodehelper(void *data) goto fail; } - retval = kernel_execve(sub_info->path, sub_info->argv, sub_info->envp); + retval = kernel_execve(sub_info->path, + (const char *const *)sub_info->argv, + (const char *const *)sub_info->envp); /* Exec failed? */ fail: -- cgit v1.2.3 From f362b73244fb16ea4ae127ced1467dd8adaa7733 Mon Sep 17 00:00:00 2001 From: Daniel J Blueman Date: Tue, 17 Aug 2010 23:56:55 +0100 Subject: Fix unprotected access to task credentials in waitid() Using a program like the following: #include #include #include #include int main() { id_t id; siginfo_t infop; pid_t res; id = fork(); if (id == 0) { sleep(1); exit(0); } kill(id, SIGSTOP); alarm(1); waitid(P_PID, id, &infop, WCONTINUED); return 0; } to call waitid() on a stopped process results in access to the child task's credentials without the RCU read lock being held - which may be replaced in the meantime - eliciting the following warning: =================================================== [ INFO: suspicious rcu_dereference_check() usage. ] --------------------------------------------------- kernel/exit.c:1460 invoked rcu_dereference_check() without protection! other info that might help us debug this: rcu_scheduler_active = 1, debug_locks = 1 2 locks held by waitid02/22252: #0: (tasklist_lock){.?.?..}, at: [] do_wait+0xc5/0x310 #1: (&(&sighand->siglock)->rlock){-.-...}, at: [] wait_consider_task+0x19a/0xbe0 stack backtrace: Pid: 22252, comm: waitid02 Not tainted 2.6.35-323cd+ #3 Call Trace: [] lockdep_rcu_dereference+0xa4/0xc0 [] wait_consider_task+0xaf1/0xbe0 [] do_wait+0xf5/0x310 [] sys_waitid+0x86/0x1f0 [] ? child_wait_callback+0x0/0x70 [] system_call_fastpath+0x16/0x1b This is fixed by holding the RCU read lock in wait_task_continued() to ensure that the task's current credentials aren't destroyed between us reading the cred pointer and us reading the UID from those credentials. Furthermore, protect wait_task_stopped() in the same way. We don't need to keep holding the RCU read lock once we've read the UID from the credentials as holding the RCU read lock doesn't stop the target task from changing its creds under us - so the credentials may be outdated immediately after we've read the pointer, lock or no lock. Signed-off-by: Daniel J Blueman Signed-off-by: David Howells Acked-by: Paul E. McKenney Acked-by: Oleg Nesterov Signed-off-by: Linus Torvalds --- kernel/exit.c | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/exit.c b/kernel/exit.c index 671ed56e0a49..03120229db28 100644 --- a/kernel/exit.c +++ b/kernel/exit.c @@ -1386,8 +1386,7 @@ static int wait_task_stopped(struct wait_opts *wo, if (!unlikely(wo->wo_flags & WNOWAIT)) *p_code = 0; - /* don't need the RCU readlock here as we're holding a spinlock */ - uid = __task_cred(p)->uid; + uid = task_uid(p); unlock_sig: spin_unlock_irq(&p->sighand->siglock); if (!exit_code) @@ -1460,7 +1459,7 @@ static int wait_task_continued(struct wait_opts *wo, struct task_struct *p) } if (!unlikely(wo->wo_flags & WNOWAIT)) p->signal->flags &= ~SIGNAL_STOP_CONTINUED; - uid = __task_cred(p)->uid; + uid = task_uid(p); spin_unlock_irq(&p->sighand->siglock); pid = task_pid_vnr(p); -- cgit v1.2.3 From 2a4419b5b2a77f3f4537c14f7ad7df95770655dd Mon Sep 17 00:00:00 2001 From: Nick Piggin Date: Wed, 18 Aug 2010 04:37:33 +1000 Subject: fs: fs_struct rwlock to spinlock fs: fs_struct rwlock to spinlock struct fs_struct.lock is an rwlock with the read-side used to protect root and pwd members while taking references to them. Taking a reference to a path typically requires just 2 atomic ops, so the critical section is very small. Parallel read-side operations would have cacheline contention on the lock, the dentry, and the vfsmount cachelines, so the rwlock is unlikely to ever give a real parallelism increase. Replace it with a spinlock to avoid one or two atomic operations in typical path lookup fastpath. Signed-off-by: Nick Piggin Signed-off-by: Al Viro --- kernel/fork.c | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) (limited to 'kernel') diff --git a/kernel/fork.c b/kernel/fork.c index 98b450876f93..856eac3ec52e 100644 --- a/kernel/fork.c +++ b/kernel/fork.c @@ -752,13 +752,13 @@ static int copy_fs(unsigned long clone_flags, struct task_struct *tsk) struct fs_struct *fs = current->fs; if (clone_flags & CLONE_FS) { /* tsk->fs is already what we want */ - write_lock(&fs->lock); + spin_lock(&fs->lock); if (fs->in_exec) { - write_unlock(&fs->lock); + spin_unlock(&fs->lock); return -EAGAIN; } fs->users++; - write_unlock(&fs->lock); + spin_unlock(&fs->lock); return 0; } tsk->fs = copy_fs_struct(fs); @@ -1676,13 +1676,13 @@ SYSCALL_DEFINE1(unshare, unsigned long, unshare_flags) if (new_fs) { fs = current->fs; - write_lock(&fs->lock); + spin_lock(&fs->lock); current->fs = new_fs; if (--fs->users) new_fs = NULL; else new_fs = fs; - write_unlock(&fs->lock); + spin_unlock(&fs->lock); } if (new_mm) { -- cgit v1.2.3 From 86397dc3ccfc0e17b7550d05eaf15fe91f6498dd Mon Sep 17 00:00:00 2001 From: Li Zefan Date: Tue, 17 Aug 2010 13:53:06 +0800 Subject: tracing: Clean up seqfile code for format file Remove the nasty hack that marks a pointer's LSB to distinguish common fields from event fields. Replace it with a more sane approach. Signed-off-by: Li Zefan LKML-Reference: <4C6A23C2.9020606@cn.fujitsu.com> Signed-off-by: Steven Rostedt --- kernel/trace/trace_events.c | 55 +++++++++++++++------------------------------ 1 file changed, 18 insertions(+), 37 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace_events.c b/kernel/trace/trace_events.c index 4c758f146328..398c0e8b332c 100644 --- a/kernel/trace/trace_events.c +++ b/kernel/trace/trace_events.c @@ -600,21 +600,29 @@ out: enum { FORMAT_HEADER = 1, - FORMAT_PRINTFMT = 2, + FORMAT_FIELD_SEPERATOR = 2, + FORMAT_PRINTFMT = 3, }; static void *f_next(struct seq_file *m, void *v, loff_t *pos) { struct ftrace_event_call *call = m->private; struct ftrace_event_field *field; - struct list_head *head; + struct list_head *common_head = &ftrace_common_fields; + struct list_head *head = trace_get_fields(call); (*pos)++; switch ((unsigned long)v) { case FORMAT_HEADER: - head = &ftrace_common_fields; + if (unlikely(list_empty(common_head))) + return NULL; + + field = list_entry(common_head->prev, + struct ftrace_event_field, link); + return field; + case FORMAT_FIELD_SEPERATOR: if (unlikely(list_empty(head))) return NULL; @@ -626,31 +634,10 @@ static void *f_next(struct seq_file *m, void *v, loff_t *pos) return NULL; } - head = trace_get_fields(call); - - /* - * To separate common fields from event fields, the - * LSB is set on the first event field. Clear it in case. - */ - v = (void *)((unsigned long)v & ~1L); - field = v; - /* - * If this is a common field, and at the end of the list, then - * continue with main list. - */ - if (field->link.prev == &ftrace_common_fields) { - if (unlikely(list_empty(head))) - return NULL; - field = list_entry(head->prev, struct ftrace_event_field, link); - /* Set the LSB to notify f_show to print an extra newline */ - field = (struct ftrace_event_field *) - ((unsigned long)field | 1); - return field; - } - - /* If we are done tell f_show to print the format */ - if (field->link.prev == head) + if (field->link.prev == common_head) + return (void *)FORMAT_FIELD_SEPERATOR; + else if (field->link.prev == head) return (void *)FORMAT_PRINTFMT; field = list_entry(field->link.prev, struct ftrace_event_field, link); @@ -688,22 +675,16 @@ static int f_show(struct seq_file *m, void *v) seq_printf(m, "format:\n"); return 0; + case FORMAT_FIELD_SEPERATOR: + seq_putc(m, '\n'); + return 0; + case FORMAT_PRINTFMT: seq_printf(m, "\nprint fmt: %s\n", call->print_fmt); return 0; } - /* - * To separate common fields from event fields, the - * LSB is set on the first event field. Clear it and - * print a newline if it is set. - */ - if ((unsigned long)v & 1) { - seq_putc(m, '\n'); - v = (void *)((unsigned long)v & ~1L); - } - field = v; /* -- cgit v1.2.3 From 56962b4449af34070bb1994621ef4f0265eed4d8 Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Wed, 30 Jun 2010 23:03:51 +0200 Subject: perf: Generalize some arch callchain code - Most archs use one callchain buffer per cpu, except x86 that needs to deal with NMIs. Provide a default perf_callchain_buffer() implementation that x86 overrides. - Centralize all the kernel/user regs handling and invoke new arch handlers from there: perf_callchain_user() / perf_callchain_kernel() That avoid all the user_mode(), current->mm checks and so... - Invert some parameters in perf_callchain_*() helpers: entry to the left, regs to the right, following the traditional (dst, src). Signed-off-by: Frederic Weisbecker Acked-by: Paul Mackerras Tested-by: Will Deacon Cc: Ingo Molnar Cc: Peter Zijlstra Cc: Arnaldo Carvalho de Melo Cc: Stephane Eranian Cc: David Miller Cc: Paul Mundt Cc: Borislav Petkov --- kernel/perf_event.c | 40 ++++++++++++++++++++++++++++++++++++++-- 1 file changed, 38 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index c772a3d4000d..02efde6c8798 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -2937,13 +2937,49 @@ void perf_event_do_pending(void) __perf_pending_run(); } +DEFINE_PER_CPU(struct perf_callchain_entry, perf_callchain_entry); + /* * Callchain support -- arch specific */ -__weak struct perf_callchain_entry *perf_callchain(struct pt_regs *regs) +__weak struct perf_callchain_entry *perf_callchain_buffer(void) { - return NULL; + return &__get_cpu_var(perf_callchain_entry); +} + +__weak void perf_callchain_kernel(struct perf_callchain_entry *entry, + struct pt_regs *regs) +{ +} + +__weak void perf_callchain_user(struct perf_callchain_entry *entry, + struct pt_regs *regs) +{ +} + +static struct perf_callchain_entry *perf_callchain(struct pt_regs *regs) +{ + struct perf_callchain_entry *entry; + + entry = perf_callchain_buffer(); + if (!entry) + return NULL; + + entry->nr = 0; + + if (!user_mode(regs)) { + perf_callchain_kernel(entry, regs); + if (current->mm) + regs = task_pt_regs(current); + else + regs = NULL; + } + + if (regs) + perf_callchain_user(entry, regs); + + return entry; } -- cgit v1.2.3 From f72c1a931e311bb7780fee19e41a89ac42cab50e Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Thu, 1 Jul 2010 02:31:21 +0200 Subject: perf: Factorize callchain context handling Store the kernel and user contexts from the generic layer instead of archs, this gathers some repetitive code. Signed-off-by: Frederic Weisbecker Acked-by: Paul Mackerras Tested-by: Will Deacon Cc: Ingo Molnar Cc: Peter Zijlstra Cc: Arnaldo Carvalho de Melo Cc: Stephane Eranian Cc: David Miller Cc: Paul Mundt Cc: Borislav Petkov --- kernel/perf_event.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index 02efde6c8798..615d024894cf 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -2969,6 +2969,7 @@ static struct perf_callchain_entry *perf_callchain(struct pt_regs *regs) entry->nr = 0; if (!user_mode(regs)) { + perf_callchain_store(entry, PERF_CONTEXT_KERNEL); perf_callchain_kernel(entry, regs); if (current->mm) regs = task_pt_regs(current); @@ -2976,8 +2977,10 @@ static struct perf_callchain_entry *perf_callchain(struct pt_regs *regs) regs = NULL; } - if (regs) + if (regs) { + perf_callchain_store(entry, PERF_CONTEXT_USER); perf_callchain_user(entry, regs); + } return entry; } -- cgit v1.2.3 From 927c7a9e92c4f69097a6e9e086d11fc2f8a5b40b Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Thu, 1 Jul 2010 16:20:36 +0200 Subject: perf: Fix race in callchains Now that software events don't have interrupt disabled anymore in the event path, callchains can nest on any context. So seperating nmi and others contexts in two buffers has become racy. Fix this by providing one buffer per nesting level. Given the size of the callchain entries (2040 bytes * 4), we now need to allocate them dynamically. v2: Fixed put_callchain_entry call after recursion. Fix the type of the recursion, it must be an array. v3: Use a manual pr cpu allocation (temporary solution until NMIs can safely access vmalloc'ed memory). Do a better separation between callchain reference tracking and allocation. Make the "put" path lockless for non-release cases. v4: Protect the callchain buffers with rcu. v5: Do the cpu buffers allocations node affine. Signed-off-by: Frederic Weisbecker Tested-by: Will Deacon Cc: Ingo Molnar Cc: Peter Zijlstra Cc: Arnaldo Carvalho de Melo Cc: Paul Mackerras Cc: Stephane Eranian Cc: Paul Mundt Cc: David Miller Cc: Borislav Petkov --- kernel/perf_event.c | 298 ++++++++++++++++++++++++++++++++++++++++------------ 1 file changed, 229 insertions(+), 69 deletions(-) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index 615d024894cf..75ab8a2df6b2 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -1763,6 +1763,216 @@ static u64 perf_event_read(struct perf_event *event) return perf_event_count(event); } +/* + * Callchain support + */ + +struct callchain_cpus_entries { + struct rcu_head rcu_head; + struct perf_callchain_entry *cpu_entries[0]; +}; + +static DEFINE_PER_CPU(int, callchain_recursion[4]); +static atomic_t nr_callchain_events; +static DEFINE_MUTEX(callchain_mutex); +struct callchain_cpus_entries *callchain_cpus_entries; + + +__weak void perf_callchain_kernel(struct perf_callchain_entry *entry, + struct pt_regs *regs) +{ +} + +__weak void perf_callchain_user(struct perf_callchain_entry *entry, + struct pt_regs *regs) +{ +} + +static void release_callchain_buffers_rcu(struct rcu_head *head) +{ + struct callchain_cpus_entries *entries; + int cpu; + + entries = container_of(head, struct callchain_cpus_entries, rcu_head); + + for_each_possible_cpu(cpu) + kfree(entries->cpu_entries[cpu]); + + kfree(entries); +} + +static void release_callchain_buffers(void) +{ + struct callchain_cpus_entries *entries; + + entries = callchain_cpus_entries; + rcu_assign_pointer(callchain_cpus_entries, NULL); + call_rcu(&entries->rcu_head, release_callchain_buffers_rcu); +} + +static int alloc_callchain_buffers(void) +{ + int cpu; + int size; + struct callchain_cpus_entries *entries; + + /* + * We can't use the percpu allocation API for data that can be + * accessed from NMI. Use a temporary manual per cpu allocation + * until that gets sorted out. + */ + size = sizeof(*entries) + sizeof(struct perf_callchain_entry *) * + num_possible_cpus(); + + entries = kzalloc(size, GFP_KERNEL); + if (!entries) + return -ENOMEM; + + size = sizeof(struct perf_callchain_entry) * 4; + + for_each_possible_cpu(cpu) { + entries->cpu_entries[cpu] = kmalloc_node(size, GFP_KERNEL, + cpu_to_node(cpu)); + if (!entries->cpu_entries[cpu]) + goto fail; + } + + rcu_assign_pointer(callchain_cpus_entries, entries); + + return 0; + +fail: + for_each_possible_cpu(cpu) + kfree(entries->cpu_entries[cpu]); + kfree(entries); + + return -ENOMEM; +} + +static int get_callchain_buffers(void) +{ + int err = 0; + int count; + + mutex_lock(&callchain_mutex); + + count = atomic_inc_return(&nr_callchain_events); + if (WARN_ON_ONCE(count < 1)) { + err = -EINVAL; + goto exit; + } + + if (count > 1) { + /* If the allocation failed, give up */ + if (!callchain_cpus_entries) + err = -ENOMEM; + goto exit; + } + + err = alloc_callchain_buffers(); + if (err) + release_callchain_buffers(); +exit: + mutex_unlock(&callchain_mutex); + + return err; +} + +static void put_callchain_buffers(void) +{ + if (atomic_dec_and_mutex_lock(&nr_callchain_events, &callchain_mutex)) { + release_callchain_buffers(); + mutex_unlock(&callchain_mutex); + } +} + +static int get_recursion_context(int *recursion) +{ + int rctx; + + if (in_nmi()) + rctx = 3; + else if (in_irq()) + rctx = 2; + else if (in_softirq()) + rctx = 1; + else + rctx = 0; + + if (recursion[rctx]) + return -1; + + recursion[rctx]++; + barrier(); + + return rctx; +} + +static inline void put_recursion_context(int *recursion, int rctx) +{ + barrier(); + recursion[rctx]--; +} + +static struct perf_callchain_entry *get_callchain_entry(int *rctx) +{ + int cpu; + struct callchain_cpus_entries *entries; + + *rctx = get_recursion_context(__get_cpu_var(callchain_recursion)); + if (*rctx == -1) + return NULL; + + entries = rcu_dereference(callchain_cpus_entries); + if (!entries) + return NULL; + + cpu = smp_processor_id(); + + return &entries->cpu_entries[cpu][*rctx]; +} + +static void +put_callchain_entry(int rctx) +{ + put_recursion_context(__get_cpu_var(callchain_recursion), rctx); +} + +static struct perf_callchain_entry *perf_callchain(struct pt_regs *regs) +{ + int rctx; + struct perf_callchain_entry *entry; + + + entry = get_callchain_entry(&rctx); + if (rctx == -1) + return NULL; + + if (!entry) + goto exit_put; + + entry->nr = 0; + + if (!user_mode(regs)) { + perf_callchain_store(entry, PERF_CONTEXT_KERNEL); + perf_callchain_kernel(entry, regs); + if (current->mm) + regs = task_pt_regs(current); + else + regs = NULL; + } + + if (regs) { + perf_callchain_store(entry, PERF_CONTEXT_USER); + perf_callchain_user(entry, regs); + } + +exit_put: + put_callchain_entry(rctx); + + return entry; +} + /* * Initialize the perf_event context in a task_struct: */ @@ -1895,6 +2105,8 @@ static void free_event(struct perf_event *event) atomic_dec(&nr_comm_events); if (event->attr.task) atomic_dec(&nr_task_events); + if (event->attr.sample_type & PERF_SAMPLE_CALLCHAIN) + put_callchain_buffers(); } if (event->buffer) { @@ -2937,55 +3149,6 @@ void perf_event_do_pending(void) __perf_pending_run(); } -DEFINE_PER_CPU(struct perf_callchain_entry, perf_callchain_entry); - -/* - * Callchain support -- arch specific - */ - -__weak struct perf_callchain_entry *perf_callchain_buffer(void) -{ - return &__get_cpu_var(perf_callchain_entry); -} - -__weak void perf_callchain_kernel(struct perf_callchain_entry *entry, - struct pt_regs *regs) -{ -} - -__weak void perf_callchain_user(struct perf_callchain_entry *entry, - struct pt_regs *regs) -{ -} - -static struct perf_callchain_entry *perf_callchain(struct pt_regs *regs) -{ - struct perf_callchain_entry *entry; - - entry = perf_callchain_buffer(); - if (!entry) - return NULL; - - entry->nr = 0; - - if (!user_mode(regs)) { - perf_callchain_store(entry, PERF_CONTEXT_KERNEL); - perf_callchain_kernel(entry, regs); - if (current->mm) - regs = task_pt_regs(current); - else - regs = NULL; - } - - if (regs) { - perf_callchain_store(entry, PERF_CONTEXT_USER); - perf_callchain_user(entry, regs); - } - - return entry; -} - - /* * We assume there is only KVM supporting the callbacks. * Later on, we might change it to a list if there is @@ -3480,14 +3643,20 @@ static void perf_event_output(struct perf_event *event, int nmi, struct perf_output_handle handle; struct perf_event_header header; + /* protect the callchain buffers */ + rcu_read_lock(); + perf_prepare_sample(&header, data, event, regs); if (perf_output_begin(&handle, event, header.size, nmi, 1)) - return; + goto exit; perf_output_sample(&handle, &header, data, event); perf_output_end(&handle); + +exit: + rcu_read_unlock(); } /* @@ -4243,32 +4412,16 @@ end: int perf_swevent_get_recursion_context(void) { struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context); - int rctx; - if (in_nmi()) - rctx = 3; - else if (in_irq()) - rctx = 2; - else if (in_softirq()) - rctx = 1; - else - rctx = 0; - - if (cpuctx->recursion[rctx]) - return -1; - - cpuctx->recursion[rctx]++; - barrier(); - - return rctx; + return get_recursion_context(cpuctx->recursion); } EXPORT_SYMBOL_GPL(perf_swevent_get_recursion_context); void inline perf_swevent_put_recursion_context(int rctx) { struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context); - barrier(); - cpuctx->recursion[rctx]--; + + put_recursion_context(cpuctx->recursion, rctx); } void __perf_sw_event(u32 event_id, u64 nr, int nmi, @@ -4968,6 +5121,13 @@ done: atomic_inc(&nr_comm_events); if (event->attr.task) atomic_inc(&nr_task_events); + if (event->attr.sample_type & PERF_SAMPLE_CALLCHAIN) { + err = get_callchain_buffers(); + if (err) { + free_event(event); + return ERR_PTR(err); + } + } } return event; -- cgit v1.2.3 From 7ae07ea3a48d30689ee037cb136bc21f0b37d8ae Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Sat, 14 Aug 2010 20:45:13 +0200 Subject: perf: Humanize the number of contexts Instead of hardcoding the number of contexts for the recursions barriers, define a cpp constant to make the code more self-explanatory. Signed-off-by: Frederic Weisbecker Cc: Ingo Molnar Cc: Peter Zijlstra Cc: Arnaldo Carvalho de Melo Cc: Paul Mackerras Cc: Stephane Eranian --- kernel/perf_event.c | 4 ++-- kernel/trace/trace_event_perf.c | 8 ++++---- 2 files changed, 6 insertions(+), 6 deletions(-) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index 75ab8a2df6b2..f416aef242c3 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -1772,7 +1772,7 @@ struct callchain_cpus_entries { struct perf_callchain_entry *cpu_entries[0]; }; -static DEFINE_PER_CPU(int, callchain_recursion[4]); +static DEFINE_PER_CPU(int, callchain_recursion[PERF_NR_CONTEXTS]); static atomic_t nr_callchain_events; static DEFINE_MUTEX(callchain_mutex); struct callchain_cpus_entries *callchain_cpus_entries; @@ -1828,7 +1828,7 @@ static int alloc_callchain_buffers(void) if (!entries) return -ENOMEM; - size = sizeof(struct perf_callchain_entry) * 4; + size = sizeof(struct perf_callchain_entry) * PERF_NR_CONTEXTS; for_each_possible_cpu(cpu) { entries->cpu_entries[cpu] = kmalloc_node(size, GFP_KERNEL, diff --git a/kernel/trace/trace_event_perf.c b/kernel/trace/trace_event_perf.c index 000e6e85b445..db2eae2efcf2 100644 --- a/kernel/trace/trace_event_perf.c +++ b/kernel/trace/trace_event_perf.c @@ -9,7 +9,7 @@ #include #include "trace.h" -static char *perf_trace_buf[4]; +static char *perf_trace_buf[PERF_NR_CONTEXTS]; /* * Force it to be aligned to unsigned long to avoid misaligned accesses @@ -45,7 +45,7 @@ static int perf_trace_event_init(struct ftrace_event_call *tp_event, char *buf; int i; - for (i = 0; i < 4; i++) { + for (i = 0; i < PERF_NR_CONTEXTS; i++) { buf = (char *)alloc_percpu(perf_trace_t); if (!buf) goto fail; @@ -65,7 +65,7 @@ fail: if (!total_ref_count) { int i; - for (i = 0; i < 4; i++) { + for (i = 0; i < PERF_NR_CONTEXTS; i++) { free_percpu(perf_trace_buf[i]); perf_trace_buf[i] = NULL; } @@ -140,7 +140,7 @@ void perf_trace_destroy(struct perf_event *p_event) tp_event->perf_events = NULL; if (!--total_ref_count) { - for (i = 0; i < 4; i++) { + for (i = 0; i < PERF_NR_CONTEXTS; i++) { free_percpu(perf_trace_buf[i]); perf_trace_buf[i] = NULL; } -- cgit v1.2.3 From 6016ee13db518ab1cd0cbf43fc2ad5712021e338 Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Wed, 11 Aug 2010 12:47:59 +0900 Subject: perf, tracing: add missing __percpu markups ftrace_event_call->perf_events, perf_trace_buf, fgraph_data->cpu_data and some local variables are percpu pointers missing __percpu markups. Add them. Signed-off-by: Namhyung Kim Acked-by: Tejun Heo Cc: Steven Rostedt Cc: Ingo Molnar Cc: Peter Zijlstra Cc: Arnaldo Carvalho de Melo Cc: Paul Mackerras Cc: Stephane Eranian LKML-Reference: <1281498479-28551-1-git-send-email-namhyung@gmail.com> Signed-off-by: Frederic Weisbecker --- kernel/trace/trace_event_perf.c | 15 ++++++++------- kernel/trace/trace_functions_graph.c | 2 +- 2 files changed, 9 insertions(+), 8 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace_event_perf.c b/kernel/trace/trace_event_perf.c index db2eae2efcf2..92f5477a006a 100644 --- a/kernel/trace/trace_event_perf.c +++ b/kernel/trace/trace_event_perf.c @@ -9,7 +9,7 @@ #include #include "trace.h" -static char *perf_trace_buf[PERF_NR_CONTEXTS]; +static char __percpu *perf_trace_buf[PERF_NR_CONTEXTS]; /* * Force it to be aligned to unsigned long to avoid misaligned accesses @@ -24,7 +24,7 @@ static int total_ref_count; static int perf_trace_event_init(struct ftrace_event_call *tp_event, struct perf_event *p_event) { - struct hlist_head *list; + struct hlist_head __percpu *list; int ret = -ENOMEM; int cpu; @@ -42,11 +42,11 @@ static int perf_trace_event_init(struct ftrace_event_call *tp_event, tp_event->perf_events = list; if (!total_ref_count) { - char *buf; + char __percpu *buf; int i; for (i = 0; i < PERF_NR_CONTEXTS; i++) { - buf = (char *)alloc_percpu(perf_trace_t); + buf = (char __percpu *)alloc_percpu(perf_trace_t); if (!buf) goto fail; @@ -102,13 +102,14 @@ int perf_trace_init(struct perf_event *p_event) int perf_trace_enable(struct perf_event *p_event) { struct ftrace_event_call *tp_event = p_event->tp_event; + struct hlist_head __percpu *pcpu_list; struct hlist_head *list; - list = tp_event->perf_events; - if (WARN_ON_ONCE(!list)) + pcpu_list = tp_event->perf_events; + if (WARN_ON_ONCE(!pcpu_list)) return -EINVAL; - list = this_cpu_ptr(list); + list = this_cpu_ptr(pcpu_list); hlist_add_head_rcu(&p_event->hlist_entry, list); return 0; diff --git a/kernel/trace/trace_functions_graph.c b/kernel/trace/trace_functions_graph.c index 6bff23625781..fcb5a542cd21 100644 --- a/kernel/trace/trace_functions_graph.c +++ b/kernel/trace/trace_functions_graph.c @@ -23,7 +23,7 @@ struct fgraph_cpu_data { }; struct fgraph_data { - struct fgraph_cpu_data *cpu_data; + struct fgraph_cpu_data __percpu *cpu_data; /* Place to preserve last processed entry. */ struct ftrace_graph_ent_entry ent; -- cgit v1.2.3 From ca5ecddfa8fcbd948c95530e7e817cee9fb43a3d Mon Sep 17 00:00:00 2001 From: "Paul E. McKenney" Date: Wed, 28 Apr 2010 14:39:09 -0700 Subject: rcu: define __rcu address space modifier for sparse This commit provides definitions for the __rcu annotation defined earlier. This annotation permits sparse to check for correct use of RCU-protected pointers. If a pointer that is annotated with __rcu is accessed directly (as opposed to via rcu_dereference(), rcu_assign_pointer(), or one of their variants), sparse can be made to complain. To enable such complaints, use the new default-disabled CONFIG_SPARSE_RCU_POINTER kernel configuration option. Please note that these sparse complaints are intended to be a debugging aid, -not- a code-style-enforcement mechanism. There are special rcu_dereference_protected() and rcu_access_pointer() accessors for use when RCU read-side protection is not required, for example, when no other CPU has access to the data structure in question or while the current CPU hold the update-side lock. This patch also updates a number of docbook comments that were showing their age. Signed-off-by: Arnd Bergmann Signed-off-by: Paul E. McKenney Cc: Christopher Li Reviewed-by: Josh Triplett --- kernel/rcupdate.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/rcupdate.c b/kernel/rcupdate.c index 4d169835fb36..6c79e851521c 100644 --- a/kernel/rcupdate.c +++ b/kernel/rcupdate.c @@ -73,12 +73,14 @@ int debug_lockdep_rcu_enabled(void) EXPORT_SYMBOL_GPL(debug_lockdep_rcu_enabled); /** - * rcu_read_lock_bh_held - might we be in RCU-bh read-side critical section? + * rcu_read_lock_bh_held() - might we be in RCU-bh read-side critical section? * * Check for bottom half being disabled, which covers both the * CONFIG_PROVE_RCU and not cases. Note that if someone uses * rcu_read_lock_bh(), but then later enables BH, lockdep (if enabled) - * will show the situation. + * will show the situation. This is useful for debug checks in functions + * that require that they be called within an RCU read-side critical + * section. * * Check debug_lockdep_rcu_enabled() to prevent false positives during boot. */ -- cgit v1.2.3 From 67bdbffd696f29a0b68aa8daa285783a06651583 Mon Sep 17 00:00:00 2001 From: Arnd Bergmann Date: Thu, 25 Feb 2010 16:55:13 +0100 Subject: rculist: avoid __rcu annotations This avoids warnings from missing __rcu annotations in the rculist implementation, making it possible to use the same lists in both RCU and non-RCU cases. We can add rculist annotations later, together with lockdep support for rculist, which is missing as well, but that may involve changing all the users. Signed-off-by: Arnd Bergmann Signed-off-by: Paul E. McKenney Cc: Pavel Emelyanov Cc: Sukadev Bhattiprolu Reviewed-by: Josh Triplett --- kernel/pid.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/pid.c b/kernel/pid.c index d55c6fb8d087..0f90c2f713f1 100644 --- a/kernel/pid.c +++ b/kernel/pid.c @@ -401,7 +401,7 @@ struct task_struct *pid_task(struct pid *pid, enum pid_type type) struct task_struct *result = NULL; if (pid) { struct hlist_node *first; - first = rcu_dereference_check(pid->tasks[type].first, + first = rcu_dereference_check(hlist_first_rcu(&pid->tasks[type]), rcu_read_lock_held() || lockdep_tasklist_lock_is_held()); if (first) -- cgit v1.2.3 From 2c392b8c3450ceb69ba1b93cb0cddb3998fb8cdc Mon Sep 17 00:00:00 2001 From: Arnd Bergmann Date: Wed, 24 Feb 2010 19:41:39 +0100 Subject: cgroups: __rcu annotations Signed-off-by: Arnd Bergmann Signed-off-by: Paul E. McKenney Acked-by: Paul Menage Cc: Li Zefan Reviewed-by: Josh Triplett --- kernel/cgroup.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/cgroup.c b/kernel/cgroup.c index 192f88c5b0f9..e5c5497a7dca 100644 --- a/kernel/cgroup.c +++ b/kernel/cgroup.c @@ -138,7 +138,7 @@ struct css_id { * is called after synchronize_rcu(). But for safe use, css_is_removed() * css_tryget() should be used for avoiding race. */ - struct cgroup_subsys_state *css; + struct cgroup_subsys_state __rcu *css; /* * ID of this css. */ -- cgit v1.2.3 From e546f485e1d7520ca0200502cdcc11b503f4805c Mon Sep 17 00:00:00 2001 From: Lai Jiangshan Date: Mon, 21 Jun 2010 16:57:42 +0800 Subject: rcutorture: add random preemption Add random preemption to help we to torture the preemptable rcu. srcu_read_delay() also calls rcu_read_delay() for shorter delays. Added comment to preempt_schedule() call indicating that no quiescent states happen if preemption is disabled. Signed-off-by: Lai Jiangshan Signed-off-by: Paul E. McKenney Reviewed-by: Josh Triplett --- kernel/rcutorture.c | 6 ++++++ 1 file changed, 6 insertions(+) (limited to 'kernel') diff --git a/kernel/rcutorture.c b/kernel/rcutorture.c index 2e2726d790b9..729710273dcb 100644 --- a/kernel/rcutorture.c +++ b/kernel/rcutorture.c @@ -303,6 +303,10 @@ static void rcu_read_delay(struct rcu_random_state *rrsp) mdelay(longdelay_ms); if (!(rcu_random(rrsp) % (nrealreaders * 2 * shortdelay_us))) udelay(shortdelay_us); +#ifdef CONFIG_PREEMPT + if (!preempt_count() && !(rcu_random(rrsp) % (nrealreaders * 20000))) + preempt_schedule(); /* No QS if preempt_disable() in effect */ +#endif } static void rcu_torture_read_unlock(int idx) __releases(RCU) @@ -536,6 +540,8 @@ static void srcu_read_delay(struct rcu_random_state *rrsp) delay = rcu_random(rrsp) % (nrealreaders * 2 * longdelay * uspertick); if (!delay) schedule_timeout_interruptible(longdelay); + else + rcu_read_delay(rrsp); } static void srcu_torture_read_unlock(int idx) __releases(&srcu_ctl) -- cgit v1.2.3 From 394f99a9007d4274f7076bb8553ab0ff9707688b Mon Sep 17 00:00:00 2001 From: Lai Jiangshan Date: Mon, 28 Jun 2010 16:25:04 +0800 Subject: rcu: simplify the usage of percpu data &percpu_data is compatible with allocated percpu data. And we use it and remove the "->rda[NR_CPUS]" array, saving significant storage on systems with large numbers of CPUs. This does add an additional level of indirection and thus an additional cache line referenced, but because ->rda is not used on the read side, this is OK. Signed-off-by: Lai Jiangshan Reviewed-by: Tejun Heo Signed-off-by: Paul E. McKenney Reviewed-by: Josh Triplett --- kernel/rcutree.c | 42 +++++++++++++++--------------------------- kernel/rcutree.h | 2 +- kernel/rcutree_plugin.h | 4 ++-- kernel/rcutree_trace.c | 2 +- 4 files changed, 19 insertions(+), 31 deletions(-) (limited to 'kernel') diff --git a/kernel/rcutree.c b/kernel/rcutree.c index d5bc43976c5a..5b1c3c231bae 100644 --- a/kernel/rcutree.c +++ b/kernel/rcutree.c @@ -712,7 +712,7 @@ static void rcu_start_gp(struct rcu_state *rsp, unsigned long flags) __releases(rcu_get_root(rsp)->lock) { - struct rcu_data *rdp = rsp->rda[smp_processor_id()]; + struct rcu_data *rdp = this_cpu_ptr(rsp->rda); struct rcu_node *rnp = rcu_get_root(rsp); if (!cpu_needs_another_gp(rsp, rdp) || rsp->fqs_active) { @@ -960,7 +960,7 @@ rcu_check_quiescent_state(struct rcu_state *rsp, struct rcu_data *rdp) static void rcu_send_cbs_to_orphanage(struct rcu_state *rsp) { int i; - struct rcu_data *rdp = rsp->rda[smp_processor_id()]; + struct rcu_data *rdp = this_cpu_ptr(rsp->rda); if (rdp->nxtlist == NULL) return; /* irqs disabled, so comparison is stable. */ @@ -984,7 +984,7 @@ static void rcu_adopt_orphan_cbs(struct rcu_state *rsp) struct rcu_data *rdp; raw_spin_lock_irqsave(&rsp->onofflock, flags); - rdp = rsp->rda[smp_processor_id()]; + rdp = this_cpu_ptr(rsp->rda); if (rsp->orphan_cbs_list == NULL) { raw_spin_unlock_irqrestore(&rsp->onofflock, flags); return; @@ -1007,7 +1007,7 @@ static void __rcu_offline_cpu(int cpu, struct rcu_state *rsp) unsigned long flags; unsigned long mask; int need_report = 0; - struct rcu_data *rdp = rsp->rda[cpu]; + struct rcu_data *rdp = per_cpu_ptr(rsp->rda, cpu); struct rcu_node *rnp; /* Exclude any attempts to start a new grace period. */ @@ -1226,7 +1226,8 @@ static void force_qs_rnp(struct rcu_state *rsp, int (*f)(struct rcu_data *)) cpu = rnp->grplo; bit = 1; for (; cpu <= rnp->grphi; cpu++, bit <<= 1) { - if ((rnp->qsmask & bit) != 0 && f(rsp->rda[cpu])) + if ((rnp->qsmask & bit) != 0 && + f(per_cpu_ptr(rsp->rda, cpu))) mask |= bit; } if (mask != 0) { @@ -1402,7 +1403,7 @@ __call_rcu(struct rcu_head *head, void (*func)(struct rcu_head *rcu), * a quiescent state betweentimes. */ local_irq_save(flags); - rdp = rsp->rda[smp_processor_id()]; + rdp = this_cpu_ptr(rsp->rda); rcu_process_gp_end(rsp, rdp); check_for_new_grace_period(rsp, rdp); @@ -1701,7 +1702,7 @@ rcu_boot_init_percpu_data(int cpu, struct rcu_state *rsp) { unsigned long flags; int i; - struct rcu_data *rdp = rsp->rda[cpu]; + struct rcu_data *rdp = per_cpu_ptr(rsp->rda, cpu); struct rcu_node *rnp = rcu_get_root(rsp); /* Set up local state, ensuring consistent view of global state. */ @@ -1729,7 +1730,7 @@ rcu_init_percpu_data(int cpu, struct rcu_state *rsp, int preemptable) { unsigned long flags; unsigned long mask; - struct rcu_data *rdp = rsp->rda[cpu]; + struct rcu_data *rdp = per_cpu_ptr(rsp->rda, cpu); struct rcu_node *rnp = rcu_get_root(rsp); /* Set up local state, ensuring consistent view of global state. */ @@ -1865,7 +1866,8 @@ static void __init rcu_init_levelspread(struct rcu_state *rsp) /* * Helper function for rcu_init() that initializes one rcu_state structure. */ -static void __init rcu_init_one(struct rcu_state *rsp) +static void __init rcu_init_one(struct rcu_state *rsp, + struct rcu_data __percpu *rda) { static char *buf[] = { "rcu_node_level_0", "rcu_node_level_1", @@ -1918,37 +1920,23 @@ static void __init rcu_init_one(struct rcu_state *rsp) } } + rsp->rda = rda; rnp = rsp->level[NUM_RCU_LVLS - 1]; for_each_possible_cpu(i) { while (i > rnp->grphi) rnp++; - rsp->rda[i]->mynode = rnp; + per_cpu_ptr(rsp->rda, i)->mynode = rnp; rcu_boot_init_percpu_data(i, rsp); } } -/* - * Helper macro for __rcu_init() and __rcu_init_preempt(). To be used - * nowhere else! Assigns leaf node pointers into each CPU's rcu_data - * structure. - */ -#define RCU_INIT_FLAVOR(rsp, rcu_data) \ -do { \ - int i; \ - \ - for_each_possible_cpu(i) { \ - (rsp)->rda[i] = &per_cpu(rcu_data, i); \ - } \ - rcu_init_one(rsp); \ -} while (0) - void __init rcu_init(void) { int cpu; rcu_bootup_announce(); - RCU_INIT_FLAVOR(&rcu_sched_state, rcu_sched_data); - RCU_INIT_FLAVOR(&rcu_bh_state, rcu_bh_data); + rcu_init_one(&rcu_sched_state, &rcu_sched_data); + rcu_init_one(&rcu_bh_state, &rcu_bh_data); __rcu_init_preempt(); open_softirq(RCU_SOFTIRQ, rcu_process_callbacks); diff --git a/kernel/rcutree.h b/kernel/rcutree.h index 14c040b18ed0..5ce197e87792 100644 --- a/kernel/rcutree.h +++ b/kernel/rcutree.h @@ -283,7 +283,7 @@ struct rcu_state { struct rcu_node *level[NUM_RCU_LVLS]; /* Hierarchy levels. */ u32 levelcnt[MAX_RCU_LVLS + 1]; /* # nodes in each level. */ u8 levelspread[NUM_RCU_LVLS]; /* kids/node in each level. */ - struct rcu_data *rda[NR_CPUS]; /* array of rdp pointers. */ + struct rcu_data __percpu *rda; /* pointer of percu rcu_data. */ /* The following fields are guarded by the root rcu_node's lock. */ diff --git a/kernel/rcutree_plugin.h b/kernel/rcutree_plugin.h index 0e4f420245d9..9906f85c7780 100644 --- a/kernel/rcutree_plugin.h +++ b/kernel/rcutree_plugin.h @@ -154,7 +154,7 @@ static void rcu_preempt_note_context_switch(int cpu) (t->rcu_read_unlock_special & RCU_READ_UNLOCK_BLOCKED) == 0) { /* Possibly blocking in an RCU read-side critical section. */ - rdp = rcu_preempt_state.rda[cpu]; + rdp = per_cpu_ptr(rcu_preempt_state.rda, cpu); rnp = rdp->mynode; raw_spin_lock_irqsave(&rnp->lock, flags); t->rcu_read_unlock_special |= RCU_READ_UNLOCK_BLOCKED; @@ -771,7 +771,7 @@ static void rcu_preempt_send_cbs_to_orphanage(void) */ static void __init __rcu_init_preempt(void) { - RCU_INIT_FLAVOR(&rcu_preempt_state, rcu_preempt_data); + rcu_init_one(&rcu_preempt_state, &rcu_preempt_data); } /* diff --git a/kernel/rcutree_trace.c b/kernel/rcutree_trace.c index 36c95b45738e..458e032a3a30 100644 --- a/kernel/rcutree_trace.c +++ b/kernel/rcutree_trace.c @@ -262,7 +262,7 @@ static void print_rcu_pendings(struct seq_file *m, struct rcu_state *rsp) struct rcu_data *rdp; for_each_possible_cpu(cpu) { - rdp = rsp->rda[cpu]; + rdp = per_cpu_ptr(rsp->rda, cpu); if (rdp->beenonline) print_one_rcu_pending(m, rdp); } -- cgit v1.2.3 From 4221a9918e38b7494cee341dda7b7b4bb8c04bde Mon Sep 17 00:00:00 2001 From: Tetsuo Handa Date: Sat, 26 Jun 2010 01:08:19 +0900 Subject: Add RCU check for find_task_by_vpid(). find_task_by_vpid() says "Must be called under rcu_read_lock().". But due to commit 3120438 "rcu: Disable lockdep checking in RCU list-traversal primitives", we are currently unable to catch "find_task_by_vpid() with tasklist_lock held but RCU lock not held" errors due to the RCU-lockdep checks being suppressed in the RCU variants of the struct list_head traversals. This commit therefore places an explicit check for being in an RCU read-side critical section in find_task_by_pid_ns(). =================================================== [ INFO: suspicious rcu_dereference_check() usage. ] --------------------------------------------------- kernel/pid.c:386 invoked rcu_dereference_check() without protection! other info that might help us debug this: rcu_scheduler_active = 1, debug_locks = 1 1 lock held by rc.sysinit/1102: #0: (tasklist_lock){.+.+..}, at: [] sys_setpgid+0x40/0x160 stack backtrace: Pid: 1102, comm: rc.sysinit Not tainted 2.6.35-rc3-dirty #1 Call Trace: [] lockdep_rcu_dereference+0x94/0xb0 [] find_task_by_pid_ns+0x6d/0x70 [] find_task_by_vpid+0x18/0x20 [] sys_setpgid+0x47/0x160 [] sysenter_do_call+0x12/0x36 Commit updated to use a new rcu_lockdep_assert() exported API rather than the old internal __do_rcu_dereference(). Signed-off-by: Tetsuo Handa Signed-off-by: Paul E. McKenney Reviewed-by: Josh Triplett --- kernel/pid.c | 1 + 1 file changed, 1 insertion(+) (limited to 'kernel') diff --git a/kernel/pid.c b/kernel/pid.c index 0f90c2f713f1..39b65b69584f 100644 --- a/kernel/pid.c +++ b/kernel/pid.c @@ -416,6 +416,7 @@ EXPORT_SYMBOL(pid_task); */ struct task_struct *find_task_by_pid_ns(pid_t nr, struct pid_namespace *ns) { + rcu_lockdep_assert(rcu_read_lock_held()); return pid_task(find_pid_ns(nr, ns), PIDTYPE_PID); } -- cgit v1.2.3 From b163760e37047781b37c412cde54d146ac4b651f Mon Sep 17 00:00:00 2001 From: "Paul E. McKenney" Date: Wed, 2 Jun 2010 16:21:38 -0700 Subject: rcu: make CPU stall warning timeout configurable Also set the default to 60 seconds, up from the previous hard-coded timeout of 10 seconds. This allows people who care to set short timeouts, while avoiding people with unusual configurations (make randconfig!!!) from being bothered with spurious CPU stall warnings. Signed-off-by: Paul E. McKenney Reviewed-by: Josh Triplett --- kernel/rcutree.h | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/rcutree.h b/kernel/rcutree.h index 5ce197e87792..183ebf405315 100644 --- a/kernel/rcutree.h +++ b/kernel/rcutree.h @@ -254,9 +254,10 @@ struct rcu_data { #define RCU_STALL_DELAY_DELTA 0 #endif -#define RCU_SECONDS_TILL_STALL_CHECK (10 * HZ + RCU_STALL_DELAY_DELTA) +#define RCU_SECONDS_TILL_STALL_CHECK (CONFIG_RCU_CPU_STALL_TIMEOUT * HZ + \ + RCU_STALL_DELAY_DELTA) /* for rsp->jiffies_stall */ -#define RCU_SECONDS_TILL_STALL_RECHECK (30 * HZ + RCU_STALL_DELAY_DELTA) +#define RCU_SECONDS_TILL_STALL_RECHECK (3 * RCU_SECONDS_TILL_STALL_CHECK + 30) /* for rsp->jiffies_stall */ #define RCU_STALL_RAT_DELAY 2 /* Allow other CPUs time */ /* to take at least one */ -- cgit v1.2.3 From 742734eea0cf5314cde5945963ed964be167bd84 Mon Sep 17 00:00:00 2001 From: "Paul E. McKenney" Date: Wed, 30 Jun 2010 11:43:52 -0700 Subject: rcu: add boot parameter to suppress RCU CPU stall warning messages Although the RCU CPU stall warning messages are a very good way to alert people to a problem, once alerted, it is sometimes helpful to shut them off in order to avoid obscuring other messages that might be being used to track down the problem. Although you can rebuild the kernel with CONFIG_RCU_CPU_STALL_DETECTOR=n, this is sometimes inconvenient. This commit therefore adds a boot parameter named "rcu_cpu_stall_suppress" that shuts these messages off without requiring a rebuild (though a reboot might be needed for those not brave enough to patch their kernel while it is running). This message-suppression was already in place for the panic case, so this commit need only rename the variable and export it via module_param(). Signed-off-by: Paul E. McKenney --- kernel/rcutree.c | 11 ++++++++--- 1 file changed, 8 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/rcutree.c b/kernel/rcutree.c index 5b1c3c231bae..f3d5906cbc21 100644 --- a/kernel/rcutree.c +++ b/kernel/rcutree.c @@ -143,6 +143,11 @@ module_param(blimit, int, 0); module_param(qhimark, int, 0); module_param(qlowmark, int, 0); +#ifdef CONFIG_RCU_CPU_STALL_DETECTOR +int rcu_cpu_stall_suppress __read_mostly; +module_param(rcu_cpu_stall_suppress, int, 0); +#endif /* #ifdef CONFIG_RCU_CPU_STALL_DETECTOR */ + static void force_quiescent_state(struct rcu_state *rsp, int relaxed); static int rcu_pending(int cpu); @@ -450,7 +455,7 @@ static int rcu_implicit_dynticks_qs(struct rcu_data *rdp) #ifdef CONFIG_RCU_CPU_STALL_DETECTOR -int rcu_cpu_stall_panicking __read_mostly; +int rcu_cpu_stall_suppress __read_mostly; static void record_gp_stall_check_time(struct rcu_state *rsp) { @@ -530,7 +535,7 @@ static void check_cpu_stall(struct rcu_state *rsp, struct rcu_data *rdp) long delta; struct rcu_node *rnp; - if (rcu_cpu_stall_panicking) + if (rcu_cpu_stall_suppress) return; delta = jiffies - rsp->jiffies_stall; rnp = rdp->mynode; @@ -548,7 +553,7 @@ static void check_cpu_stall(struct rcu_state *rsp, struct rcu_data *rdp) static int rcu_panic(struct notifier_block *this, unsigned long ev, void *ptr) { - rcu_cpu_stall_panicking = 1; + rcu_cpu_stall_suppress = 1; return NOTIFY_DONE; } -- cgit v1.2.3 From 77d8485a8b5416c615b6acd95f01bfcacd7d81ff Mon Sep 17 00:00:00 2001 From: "Paul E. McKenney" Date: Thu, 8 Jul 2010 17:38:59 -0700 Subject: rcu: improve kerneldoc for rcu_read_lock(), call_rcu(), and synchronize_rcu() Make it explicit that new RCU read-side critical sections that start after call_rcu() and synchronize_rcu() start might still be running after the end of the relevant grace period. Signed-off-by: Paul E. McKenney Reviewed-by: Josh Triplett --- kernel/rcutree_plugin.h | 8 +++++--- 1 file changed, 5 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/rcutree_plugin.h b/kernel/rcutree_plugin.h index 9906f85c7780..63bb7714fdeb 100644 --- a/kernel/rcutree_plugin.h +++ b/kernel/rcutree_plugin.h @@ -546,9 +546,11 @@ EXPORT_SYMBOL_GPL(call_rcu); * * Control will return to the caller some time after a full grace * period has elapsed, in other words after all currently executing RCU - * read-side critical sections have completed. RCU read-side critical - * sections are delimited by rcu_read_lock() and rcu_read_unlock(), - * and may be nested. + * read-side critical sections have completed. Note, however, that + * upon return from synchronize_rcu(), the caller might well be executing + * concurrently with new RCU read-side critical sections that began while + * synchronize_rcu() was waiting. RCU read-side critical sections are + * delimited by rcu_read_lock() and rcu_read_unlock(), and may be nested. */ void synchronize_rcu(void) { -- cgit v1.2.3 From f2e0dd7090eddef427ab9d9f81de122244cded51 Mon Sep 17 00:00:00 2001 From: "Paul E. McKenney" Date: Wed, 14 Jul 2010 14:38:30 -0700 Subject: rcu: allow RCU CPU stall warning messages to be controlled in /sys Set the permissions of the rcu_cpu_stall_suppress to 644 to enable RCU CPU stall warnings to be enabled and disabled at runtime via sysfs. Suggested-by: Josh Triplett Signed-off-by: Paul E. McKenney --- kernel/rcutree.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/rcutree.c b/kernel/rcutree.c index f3d5906cbc21..5d910beefff2 100644 --- a/kernel/rcutree.c +++ b/kernel/rcutree.c @@ -145,7 +145,7 @@ module_param(qlowmark, int, 0); #ifdef CONFIG_RCU_CPU_STALL_DETECTOR int rcu_cpu_stall_suppress __read_mostly; -module_param(rcu_cpu_stall_suppress, int, 0); +module_param(rcu_cpu_stall_suppress, int, 0644); #endif /* #ifdef CONFIG_RCU_CPU_STALL_DETECTOR */ static void force_quiescent_state(struct rcu_state *rsp, int relaxed); -- cgit v1.2.3 From 910b1b7e19a292ff685001caf1bf1a9775b771a1 Mon Sep 17 00:00:00 2001 From: "Paul E. McKenney" Date: Wed, 21 Jul 2010 08:05:56 -0700 Subject: rcu: Allow RCU CPU stall warnings to be off at boot, but manually enablable Currently, if RCU CPU stall warnings are enabled, they are enabled immediately upon boot. They can be manually disabled via /sys (and also re-enabled via /sys), and are automatically disabled upon panic. However, some users need RCU CPU stalls to be disabled at boot time, but to be enabled without rebuilding/rebooting. For example, someone running a real-time application in production might not want the additional latency of RCU CPU stall detection in normal operation, but might need to enable it at any point for fault isolation purposes. This commit therefore provides a new CONFIG_RCU_CPU_STALL_DETECTOR_RUNNABLE kernel configuration parameter that maintains the current behavior (enable at boot) by default, but allows a kernel to be configured with RCU CPU stall detection built into the kernel, but disabled at boot time. Requested-by: Clark Williams Requested-by: John Kacur Signed-off-by: Paul E. McKenney --- kernel/rcutree.c | 2 +- kernel/rcutree.h | 6 ++++++ 2 files changed, 7 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/rcutree.c b/kernel/rcutree.c index 5d910beefff2..5aab7dabd0d5 100644 --- a/kernel/rcutree.c +++ b/kernel/rcutree.c @@ -144,7 +144,7 @@ module_param(qhimark, int, 0); module_param(qlowmark, int, 0); #ifdef CONFIG_RCU_CPU_STALL_DETECTOR -int rcu_cpu_stall_suppress __read_mostly; +int rcu_cpu_stall_suppress __read_mostly = RCU_CPU_STALL_SUPPRESS_INIT; module_param(rcu_cpu_stall_suppress, int, 0644); #endif /* #ifdef CONFIG_RCU_CPU_STALL_DETECTOR */ diff --git a/kernel/rcutree.h b/kernel/rcutree.h index 183ebf405315..bb4d08695c45 100644 --- a/kernel/rcutree.h +++ b/kernel/rcutree.h @@ -264,6 +264,12 @@ struct rcu_data { /* scheduling clock irq */ /* before ratting on them. */ +#ifdef CONFIG_RCU_CPU_STALL_DETECTOR_RUNNABLE +#define RCU_CPU_STALL_SUPPRESS_INIT 0 +#else +#define RCU_CPU_STALL_SUPPRESS_INIT 1 +#endif + #endif /* #ifdef CONFIG_RCU_CPU_STALL_DETECTOR */ #define ULONG_CMP_GE(a, b) (ULONG_MAX / 2 >= (a) - (b)) -- cgit v1.2.3 From 1495cc9df4e81f5a8fa9b0b8f1034b14d24b7d8c Mon Sep 17 00:00:00 2001 From: Dmitry Torokhov Date: Tue, 17 Aug 2010 21:15:46 -0700 Subject: Input: sysrq - drop tty argument from sysrq ops handlers Noone is using tty argument so let's get rid of it. Acked-by: Alan Cox Acked-by: Jason Wessel Acked-by: Greg Kroah-Hartman Signed-off-by: Dmitry Torokhov --- kernel/debug/debug_core.c | 2 +- kernel/power/poweroff.c | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/debug/debug_core.c b/kernel/debug/debug_core.c index 3c2d4972d235..de407c78178d 100644 --- a/kernel/debug/debug_core.c +++ b/kernel/debug/debug_core.c @@ -741,7 +741,7 @@ static struct console kgdbcons = { }; #ifdef CONFIG_MAGIC_SYSRQ -static void sysrq_handle_dbg(int key, struct tty_struct *tty) +static void sysrq_handle_dbg(int key) { if (!dbg_io_ops) { printk(KERN_CRIT "ERROR: No KGDB I/O module available\n"); diff --git a/kernel/power/poweroff.c b/kernel/power/poweroff.c index e8b337006276..d52359374e85 100644 --- a/kernel/power/poweroff.c +++ b/kernel/power/poweroff.c @@ -24,7 +24,7 @@ static void do_poweroff(struct work_struct *dummy) static DECLARE_WORK(poweroff_work, do_poweroff); -static void handle_poweroff(int key, struct tty_struct *tty) +static void handle_poweroff(int key) { /* run sysrq poweroff on boot cpu */ schedule_work_on(cpumask_first(cpu_online_mask), &poweroff_work); -- cgit v1.2.3 From 277b199800ac90811ac86d215063df1984f51619 Mon Sep 17 00:00:00 2001 From: Lin Ming Date: Fri, 20 Aug 2010 11:03:51 +0800 Subject: lockup_detector: Make callback function static watchdog_overflow_callback() is only used in kernel/watchdog.c. Signed-off-by: Lin Ming Cc: Peter Zijlstra Cc: Don Zickus LKML-Reference: <1282273431.16443.32.camel@minggr.sh.intel.com> Signed-off-by: Ingo Molnar --- kernel/watchdog.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/watchdog.c b/kernel/watchdog.c index 613bc1f04610..b60e2a869bba 100644 --- a/kernel/watchdog.c +++ b/kernel/watchdog.c @@ -202,7 +202,7 @@ static struct perf_event_attr wd_hw_attr = { }; /* Callback function for perf event subsystem */ -void watchdog_overflow_callback(struct perf_event *event, int nmi, +static void watchdog_overflow_callback(struct perf_event *event, int nmi, struct perf_sample_data *data, struct pt_regs *regs) { -- cgit v1.2.3 From 861d034ee814917a83bd5de4b26e3b8336ddeeb8 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Thu, 19 Aug 2010 13:31:43 +0200 Subject: sched: Fix rq->clock synchronization when migrating tasks sched_fork() -- we do task placement in ->task_fork_fair() ensure we update_rq_clock() so we work with current time. We leave the vruntime in relative state, so the time delay until wake_up_new_task() doesn't matter. wake_up_new_task() -- Since task_fork_fair() left p->vruntime in relative state we can safely migrate, the activate_task() on the remote rq will call update_rq_clock() and causes the clock to be synced (enough). Tested-by: Jack Daniel Tested-by: Philby John Signed-off-by: Peter Zijlstra LKML-Reference: <1281002322.1923.1708.camel@laptop> Signed-off-by: Ingo Molnar --- kernel/sched_fair.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'kernel') diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c index 806d1b227a21..ab661ebc4895 100644 --- a/kernel/sched_fair.c +++ b/kernel/sched_fair.c @@ -3752,6 +3752,8 @@ static void task_fork_fair(struct task_struct *p) raw_spin_lock_irqsave(&rq->lock, flags); + update_rq_clock(rq); + if (unlikely(task_cpu(p) != this_cpu)) __set_task_cpu(p, this_cpu); -- cgit v1.2.3 From a57eb940d130477a799dfb24a570ee04979c0f7f Mon Sep 17 00:00:00 2001 From: "Paul E. McKenney" Date: Tue, 29 Jun 2010 16:49:16 -0700 Subject: rcu: Add a TINY_PREEMPT_RCU MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Implement a small-memory-footprint uniprocessor-only implementation of preemptible RCU. This implementation uses but a single blocked-tasks list rather than the combinatorial number used per leaf rcu_node by TREE_PREEMPT_RCU, which reduces memory consumption and greatly simplifies processing. This version also takes advantage of uniprocessor execution to accelerate grace periods in the case where there are no readers. The general design is otherwise broadly similar to that of TREE_PREEMPT_RCU. This implementation is a step towards having RCU implementation driven off of the SMP and PREEMPT kernel configuration variables, which can happen once this implementation has accumulated sufficient experience. Removed ACCESS_ONCE() from __rcu_read_unlock() and added barrier() as suggested by Steve Rostedt in order to avoid the compiler-reordering issue noted by Mathieu Desnoyers (http://lkml.org/lkml/2010/8/16/183). As can be seen below, CONFIG_TINY_PREEMPT_RCU represents almost 5Kbyte savings compared to CONFIG_TREE_PREEMPT_RCU. Of course, for non-real-time workloads, CONFIG_TINY_RCU is even better. CONFIG_TREE_PREEMPT_RCU text data bss dec filename 13 0 0 13 kernel/rcupdate.o 6170 825 28 7023 kernel/rcutree.o ---- 7026 Total CONFIG_TINY_PREEMPT_RCU text data bss dec filename 13 0 0 13 kernel/rcupdate.o 2081 81 8 2170 kernel/rcutiny.o ---- 2183 Total CONFIG_TINY_RCU (non-preemptible) text data bss dec filename 13 0 0 13 kernel/rcupdate.o 719 25 0 744 kernel/rcutiny.o --- 757 Total Requested-by: Loïc Minier Signed-off-by: Paul E. McKenney --- kernel/Makefile | 1 + kernel/rcutiny.c | 33 ++- kernel/rcutiny_plugin.h | 582 +++++++++++++++++++++++++++++++++++++++++++++++- 3 files changed, 594 insertions(+), 22 deletions(-) (limited to 'kernel') diff --git a/kernel/Makefile b/kernel/Makefile index 0b72d1a74be0..17046b6e7c90 100644 --- a/kernel/Makefile +++ b/kernel/Makefile @@ -86,6 +86,7 @@ obj-$(CONFIG_TREE_RCU) += rcutree.o obj-$(CONFIG_TREE_PREEMPT_RCU) += rcutree.o obj-$(CONFIG_TREE_RCU_TRACE) += rcutree_trace.o obj-$(CONFIG_TINY_RCU) += rcutiny.o +obj-$(CONFIG_TINY_PREEMPT_RCU) += rcutiny.o obj-$(CONFIG_RELAY) += relay.o obj-$(CONFIG_SYSCTL) += utsname_sysctl.o obj-$(CONFIG_TASK_DELAY_ACCT) += delayacct.o diff --git a/kernel/rcutiny.c b/kernel/rcutiny.c index 196ec02f8be0..d806735342ac 100644 --- a/kernel/rcutiny.c +++ b/kernel/rcutiny.c @@ -59,6 +59,14 @@ int rcu_scheduler_active __read_mostly; EXPORT_SYMBOL_GPL(rcu_scheduler_active); #endif /* #ifdef CONFIG_DEBUG_LOCK_ALLOC */ +/* Forward declarations for rcutiny_plugin.h. */ +static void __rcu_process_callbacks(struct rcu_ctrlblk *rcp); +static void __call_rcu(struct rcu_head *head, + void (*func)(struct rcu_head *rcu), + struct rcu_ctrlblk *rcp); + +#include "rcutiny_plugin.h" + #ifdef CONFIG_NO_HZ static long rcu_dynticks_nesting = 1; @@ -140,6 +148,7 @@ void rcu_check_callbacks(int cpu, int user) rcu_sched_qs(cpu); else if (!in_softirq()) rcu_bh_qs(cpu); + rcu_preempt_check_callbacks(); } /* @@ -162,6 +171,7 @@ static void __rcu_process_callbacks(struct rcu_ctrlblk *rcp) *rcp->donetail = NULL; if (rcp->curtail == rcp->donetail) rcp->curtail = &rcp->rcucblist; + rcu_preempt_remove_callbacks(rcp); rcp->donetail = &rcp->rcucblist; local_irq_restore(flags); @@ -182,6 +192,7 @@ static void rcu_process_callbacks(struct softirq_action *unused) { __rcu_process_callbacks(&rcu_sched_ctrlblk); __rcu_process_callbacks(&rcu_bh_ctrlblk); + rcu_preempt_process_callbacks(); } /* @@ -223,15 +234,15 @@ static void __call_rcu(struct rcu_head *head, } /* - * Post an RCU callback to be invoked after the end of an RCU grace + * Post an RCU callback to be invoked after the end of an RCU-sched grace * period. But since we have but one CPU, that would be after any * quiescent state. */ -void call_rcu(struct rcu_head *head, void (*func)(struct rcu_head *rcu)) +void call_rcu_sched(struct rcu_head *head, void (*func)(struct rcu_head *rcu)) { __call_rcu(head, func, &rcu_sched_ctrlblk); } -EXPORT_SYMBOL_GPL(call_rcu); +EXPORT_SYMBOL_GPL(call_rcu_sched); /* * Post an RCU bottom-half callback to be invoked after any subsequent @@ -243,20 +254,6 @@ void call_rcu_bh(struct rcu_head *head, void (*func)(struct rcu_head *rcu)) } EXPORT_SYMBOL_GPL(call_rcu_bh); -void rcu_barrier(void) -{ - struct rcu_synchronize rcu; - - init_rcu_head_on_stack(&rcu.head); - init_completion(&rcu.completion); - /* Will wake me after RCU finished. */ - call_rcu(&rcu.head, wakeme_after_rcu); - /* Wait for it. */ - wait_for_completion(&rcu.completion); - destroy_rcu_head_on_stack(&rcu.head); -} -EXPORT_SYMBOL_GPL(rcu_barrier); - void rcu_barrier_bh(void) { struct rcu_synchronize rcu; @@ -289,5 +286,3 @@ void __init rcu_init(void) { open_softirq(RCU_SOFTIRQ, rcu_process_callbacks); } - -#include "rcutiny_plugin.h" diff --git a/kernel/rcutiny_plugin.h b/kernel/rcutiny_plugin.h index d223a92bc742..e6bc1b447c6c 100644 --- a/kernel/rcutiny_plugin.h +++ b/kernel/rcutiny_plugin.h @@ -1,7 +1,7 @@ /* - * Read-Copy Update mechanism for mutual exclusion (tree-based version) + * Read-Copy Update mechanism for mutual exclusion, the Bloatwatch edition * Internal non-public definitions that provide either classic - * or preemptable semantics. + * or preemptible semantics. * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -17,11 +17,587 @@ * along with this program; if not, write to the Free Software * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. * - * Copyright IBM Corporation, 2009 + * Copyright (c) 2010 Linaro * * Author: Paul E. McKenney */ +#ifdef CONFIG_TINY_PREEMPT_RCU + +#include + +/* FIXME: merge with definitions in kernel/rcutree.h. */ +#define ULONG_CMP_GE(a, b) (ULONG_MAX / 2 >= (a) - (b)) +#define ULONG_CMP_LT(a, b) (ULONG_MAX / 2 < (a) - (b)) + +/* Global control variables for preemptible RCU. */ +struct rcu_preempt_ctrlblk { + struct rcu_ctrlblk rcb; /* curtail: ->next ptr of last CB for GP. */ + struct rcu_head **nexttail; + /* Tasks blocked in a preemptible RCU */ + /* read-side critical section while an */ + /* preemptible-RCU grace period is in */ + /* progress must wait for a later grace */ + /* period. This pointer points to the */ + /* ->next pointer of the last task that */ + /* must wait for a later grace period, or */ + /* to &->rcb.rcucblist if there is no */ + /* such task. */ + struct list_head blkd_tasks; + /* Tasks blocked in RCU read-side critical */ + /* section. Tasks are placed at the head */ + /* of this list and age towards the tail. */ + struct list_head *gp_tasks; + /* Pointer to the first task blocking the */ + /* current grace period, or NULL if there */ + /* is not such task. */ + struct list_head *exp_tasks; + /* Pointer to first task blocking the */ + /* current expedited grace period, or NULL */ + /* if there is no such task. If there */ + /* is no current expedited grace period, */ + /* then there cannot be any such task. */ + u8 gpnum; /* Current grace period. */ + u8 gpcpu; /* Last grace period blocked by the CPU. */ + u8 completed; /* Last grace period completed. */ + /* If all three are equal, RCU is idle. */ +}; + +static struct rcu_preempt_ctrlblk rcu_preempt_ctrlblk = { + .rcb.donetail = &rcu_preempt_ctrlblk.rcb.rcucblist, + .rcb.curtail = &rcu_preempt_ctrlblk.rcb.rcucblist, + .nexttail = &rcu_preempt_ctrlblk.rcb.rcucblist, + .blkd_tasks = LIST_HEAD_INIT(rcu_preempt_ctrlblk.blkd_tasks), +}; + +static int rcu_preempted_readers_exp(void); +static void rcu_report_exp_done(void); + +/* + * Return true if the CPU has not yet responded to the current grace period. + */ +static int rcu_cpu_cur_gp(void) +{ + return rcu_preempt_ctrlblk.gpcpu != rcu_preempt_ctrlblk.gpnum; +} + +/* + * Check for a running RCU reader. Because there is only one CPU, + * there can be but one running RCU reader at a time. ;-) + */ +static int rcu_preempt_running_reader(void) +{ + return current->rcu_read_lock_nesting; +} + +/* + * Check for preempted RCU readers blocking any grace period. + * If the caller needs a reliable answer, it must disable hard irqs. + */ +static int rcu_preempt_blocked_readers_any(void) +{ + return !list_empty(&rcu_preempt_ctrlblk.blkd_tasks); +} + +/* + * Check for preempted RCU readers blocking the current grace period. + * If the caller needs a reliable answer, it must disable hard irqs. + */ +static int rcu_preempt_blocked_readers_cgp(void) +{ + return rcu_preempt_ctrlblk.gp_tasks != NULL; +} + +/* + * Return true if another preemptible-RCU grace period is needed. + */ +static int rcu_preempt_needs_another_gp(void) +{ + return *rcu_preempt_ctrlblk.rcb.curtail != NULL; +} + +/* + * Return true if a preemptible-RCU grace period is in progress. + * The caller must disable hardirqs. + */ +static int rcu_preempt_gp_in_progress(void) +{ + return rcu_preempt_ctrlblk.completed != rcu_preempt_ctrlblk.gpnum; +} + +/* + * Record a preemptible-RCU quiescent state for the specified CPU. Note + * that this just means that the task currently running on the CPU is + * in a quiescent state. There might be any number of tasks blocked + * while in an RCU read-side critical section. + * + * Unlike the other rcu_*_qs() functions, callers to this function + * must disable irqs in order to protect the assignment to + * ->rcu_read_unlock_special. + * + * Because this is a single-CPU implementation, the only way a grace + * period can end is if the CPU is in a quiescent state. The reason is + * that a blocked preemptible-RCU reader can exit its critical section + * only if the CPU is running it at the time. Therefore, when the + * last task blocking the current grace period exits its RCU read-side + * critical section, neither the CPU nor blocked tasks will be stopping + * the current grace period. (In contrast, SMP implementations + * might have CPUs running in RCU read-side critical sections that + * block later grace periods -- but this is not possible given only + * one CPU.) + */ +static void rcu_preempt_cpu_qs(void) +{ + /* Record both CPU and task as having responded to current GP. */ + rcu_preempt_ctrlblk.gpcpu = rcu_preempt_ctrlblk.gpnum; + current->rcu_read_unlock_special &= ~RCU_READ_UNLOCK_NEED_QS; + + /* + * If there is no GP, or if blocked readers are still blocking GP, + * then there is nothing more to do. + */ + if (!rcu_preempt_gp_in_progress() || rcu_preempt_blocked_readers_cgp()) + return; + + /* Advance callbacks. */ + rcu_preempt_ctrlblk.completed = rcu_preempt_ctrlblk.gpnum; + rcu_preempt_ctrlblk.rcb.donetail = rcu_preempt_ctrlblk.rcb.curtail; + rcu_preempt_ctrlblk.rcb.curtail = rcu_preempt_ctrlblk.nexttail; + + /* If there are no blocked readers, next GP is done instantly. */ + if (!rcu_preempt_blocked_readers_any()) + rcu_preempt_ctrlblk.rcb.donetail = rcu_preempt_ctrlblk.nexttail; + + /* If there are done callbacks, make RCU_SOFTIRQ process them. */ + if (*rcu_preempt_ctrlblk.rcb.donetail != NULL) + raise_softirq(RCU_SOFTIRQ); +} + +/* + * Start a new RCU grace period if warranted. Hard irqs must be disabled. + */ +static void rcu_preempt_start_gp(void) +{ + if (!rcu_preempt_gp_in_progress() && rcu_preempt_needs_another_gp()) { + + /* Official start of GP. */ + rcu_preempt_ctrlblk.gpnum++; + + /* Any blocked RCU readers block new GP. */ + if (rcu_preempt_blocked_readers_any()) + rcu_preempt_ctrlblk.gp_tasks = + rcu_preempt_ctrlblk.blkd_tasks.next; + + /* If there is no running reader, CPU is done with GP. */ + if (!rcu_preempt_running_reader()) + rcu_preempt_cpu_qs(); + } +} + +/* + * We have entered the scheduler, and the current task might soon be + * context-switched away from. If this task is in an RCU read-side + * critical section, we will no longer be able to rely on the CPU to + * record that fact, so we enqueue the task on the blkd_tasks list. + * If the task started after the current grace period began, as recorded + * by ->gpcpu, we enqueue at the beginning of the list. Otherwise + * before the element referenced by ->gp_tasks (or at the tail if + * ->gp_tasks is NULL) and point ->gp_tasks at the newly added element. + * The task will dequeue itself when it exits the outermost enclosing + * RCU read-side critical section. Therefore, the current grace period + * cannot be permitted to complete until the ->gp_tasks pointer becomes + * NULL. + * + * Caller must disable preemption. + */ +void rcu_preempt_note_context_switch(void) +{ + struct task_struct *t = current; + unsigned long flags; + + local_irq_save(flags); /* must exclude scheduler_tick(). */ + if (rcu_preempt_running_reader() && + (t->rcu_read_unlock_special & RCU_READ_UNLOCK_BLOCKED) == 0) { + + /* Possibly blocking in an RCU read-side critical section. */ + t->rcu_read_unlock_special |= RCU_READ_UNLOCK_BLOCKED; + + /* + * If this CPU has already checked in, then this task + * will hold up the next grace period rather than the + * current grace period. Queue the task accordingly. + * If the task is queued for the current grace period + * (i.e., this CPU has not yet passed through a quiescent + * state for the current grace period), then as long + * as that task remains queued, the current grace period + * cannot end. + */ + list_add(&t->rcu_node_entry, &rcu_preempt_ctrlblk.blkd_tasks); + if (rcu_cpu_cur_gp()) + rcu_preempt_ctrlblk.gp_tasks = &t->rcu_node_entry; + } + + /* + * Either we were not in an RCU read-side critical section to + * begin with, or we have now recorded that critical section + * globally. Either way, we can now note a quiescent state + * for this CPU. Again, if we were in an RCU read-side critical + * section, and if that critical section was blocking the current + * grace period, then the fact that the task has been enqueued + * means that current grace period continues to be blocked. + */ + rcu_preempt_cpu_qs(); + local_irq_restore(flags); +} + +/* + * Tiny-preemptible RCU implementation for rcu_read_lock(). + * Just increment ->rcu_read_lock_nesting, shared state will be updated + * if we block. + */ +void __rcu_read_lock(void) +{ + current->rcu_read_lock_nesting++; + barrier(); /* needed if we ever invoke rcu_read_lock in rcutiny.c */ +} +EXPORT_SYMBOL_GPL(__rcu_read_lock); + +/* + * Handle special cases during rcu_read_unlock(), such as needing to + * notify RCU core processing or task having blocked during the RCU + * read-side critical section. + */ +static void rcu_read_unlock_special(struct task_struct *t) +{ + int empty; + int empty_exp; + unsigned long flags; + struct list_head *np; + int special; + + /* + * NMI handlers cannot block and cannot safely manipulate state. + * They therefore cannot possibly be special, so just leave. + */ + if (in_nmi()) + return; + + local_irq_save(flags); + + /* + * If RCU core is waiting for this CPU to exit critical section, + * let it know that we have done so. + */ + special = t->rcu_read_unlock_special; + if (special & RCU_READ_UNLOCK_NEED_QS) + rcu_preempt_cpu_qs(); + + /* Hardware IRQ handlers cannot block. */ + if (in_irq()) { + local_irq_restore(flags); + return; + } + + /* Clean up if blocked during RCU read-side critical section. */ + if (special & RCU_READ_UNLOCK_BLOCKED) { + t->rcu_read_unlock_special &= ~RCU_READ_UNLOCK_BLOCKED; + + /* + * Remove this task from the ->blkd_tasks list and adjust + * any pointers that might have been referencing it. + */ + empty = !rcu_preempt_blocked_readers_cgp(); + empty_exp = rcu_preempt_ctrlblk.exp_tasks == NULL; + np = t->rcu_node_entry.next; + if (np == &rcu_preempt_ctrlblk.blkd_tasks) + np = NULL; + list_del(&t->rcu_node_entry); + if (&t->rcu_node_entry == rcu_preempt_ctrlblk.gp_tasks) + rcu_preempt_ctrlblk.gp_tasks = np; + if (&t->rcu_node_entry == rcu_preempt_ctrlblk.exp_tasks) + rcu_preempt_ctrlblk.exp_tasks = np; + INIT_LIST_HEAD(&t->rcu_node_entry); + + /* + * If this was the last task on the current list, and if + * we aren't waiting on the CPU, report the quiescent state + * and start a new grace period if needed. + */ + if (!empty && !rcu_preempt_blocked_readers_cgp()) { + rcu_preempt_cpu_qs(); + rcu_preempt_start_gp(); + } + + /* + * If this was the last task on the expedited lists, + * then we need wake up the waiting task. + */ + if (!empty_exp && rcu_preempt_ctrlblk.exp_tasks == NULL) + rcu_report_exp_done(); + } + local_irq_restore(flags); +} + +/* + * Tiny-preemptible RCU implementation for rcu_read_unlock(). + * Decrement ->rcu_read_lock_nesting. If the result is zero (outermost + * rcu_read_unlock()) and ->rcu_read_unlock_special is non-zero, then + * invoke rcu_read_unlock_special() to clean up after a context switch + * in an RCU read-side critical section and other special cases. + */ +void __rcu_read_unlock(void) +{ + struct task_struct *t = current; + + barrier(); /* needed if we ever invoke rcu_read_unlock in rcutiny.c */ + --t->rcu_read_lock_nesting; + barrier(); /* decrement before load of ->rcu_read_unlock_special */ + if (t->rcu_read_lock_nesting == 0 && + unlikely(ACCESS_ONCE(t->rcu_read_unlock_special))) + rcu_read_unlock_special(t); +#ifdef CONFIG_PROVE_LOCKING + WARN_ON_ONCE(t->rcu_read_lock_nesting < 0); +#endif /* #ifdef CONFIG_PROVE_LOCKING */ +} +EXPORT_SYMBOL_GPL(__rcu_read_unlock); + +/* + * Check for a quiescent state from the current CPU. When a task blocks, + * the task is recorded in the rcu_preempt_ctrlblk structure, which is + * checked elsewhere. This is called from the scheduling-clock interrupt. + * + * Caller must disable hard irqs. + */ +static void rcu_preempt_check_callbacks(void) +{ + struct task_struct *t = current; + + if (!rcu_preempt_running_reader() && rcu_preempt_gp_in_progress()) + rcu_preempt_cpu_qs(); + if (&rcu_preempt_ctrlblk.rcb.rcucblist != + rcu_preempt_ctrlblk.rcb.donetail) + raise_softirq(RCU_SOFTIRQ); + if (rcu_preempt_gp_in_progress() && rcu_preempt_running_reader()) + t->rcu_read_unlock_special |= RCU_READ_UNLOCK_NEED_QS; +} + +/* + * TINY_PREEMPT_RCU has an extra callback-list tail pointer to + * update, so this is invoked from __rcu_process_callbacks() to + * handle that case. Of course, it is invoked for all flavors of + * RCU, but RCU callbacks can appear only on one of the lists, and + * neither ->nexttail nor ->donetail can possibly be NULL, so there + * is no need for an explicit check. + */ +static void rcu_preempt_remove_callbacks(struct rcu_ctrlblk *rcp) +{ + if (rcu_preempt_ctrlblk.nexttail == rcp->donetail) + rcu_preempt_ctrlblk.nexttail = &rcp->rcucblist; +} + +/* + * Process callbacks for preemptible RCU. + */ +static void rcu_preempt_process_callbacks(void) +{ + __rcu_process_callbacks(&rcu_preempt_ctrlblk.rcb); +} + +/* + * Queue a preemptible -RCU callback for invocation after a grace period. + */ +void call_rcu(struct rcu_head *head, void (*func)(struct rcu_head *rcu)) +{ + unsigned long flags; + + debug_rcu_head_queue(head); + head->func = func; + head->next = NULL; + + local_irq_save(flags); + *rcu_preempt_ctrlblk.nexttail = head; + rcu_preempt_ctrlblk.nexttail = &head->next; + rcu_preempt_start_gp(); /* checks to see if GP needed. */ + local_irq_restore(flags); +} +EXPORT_SYMBOL_GPL(call_rcu); + +void rcu_barrier(void) +{ + struct rcu_synchronize rcu; + + init_rcu_head_on_stack(&rcu.head); + init_completion(&rcu.completion); + /* Will wake me after RCU finished. */ + call_rcu(&rcu.head, wakeme_after_rcu); + /* Wait for it. */ + wait_for_completion(&rcu.completion); + destroy_rcu_head_on_stack(&rcu.head); +} +EXPORT_SYMBOL_GPL(rcu_barrier); + +/* + * synchronize_rcu - wait until a grace period has elapsed. + * + * Control will return to the caller some time after a full grace + * period has elapsed, in other words after all currently executing RCU + * read-side critical sections have completed. RCU read-side critical + * sections are delimited by rcu_read_lock() and rcu_read_unlock(), + * and may be nested. + */ +void synchronize_rcu(void) +{ +#ifdef CONFIG_DEBUG_LOCK_ALLOC + if (!rcu_scheduler_active) + return; +#endif /* #ifdef CONFIG_DEBUG_LOCK_ALLOC */ + + WARN_ON_ONCE(rcu_preempt_running_reader()); + if (!rcu_preempt_blocked_readers_any()) + return; + + /* Once we get past the fastpath checks, same code as rcu_barrier(). */ + rcu_barrier(); +} +EXPORT_SYMBOL_GPL(synchronize_rcu); + +static DECLARE_WAIT_QUEUE_HEAD(sync_rcu_preempt_exp_wq); +static unsigned long sync_rcu_preempt_exp_count; +static DEFINE_MUTEX(sync_rcu_preempt_exp_mutex); + +/* + * Return non-zero if there are any tasks in RCU read-side critical + * sections blocking the current preemptible-RCU expedited grace period. + * If there is no preemptible-RCU expedited grace period currently in + * progress, returns zero unconditionally. + */ +static int rcu_preempted_readers_exp(void) +{ + return rcu_preempt_ctrlblk.exp_tasks != NULL; +} + +/* + * Report the exit from RCU read-side critical section for the last task + * that queued itself during or before the current expedited preemptible-RCU + * grace period. + */ +static void rcu_report_exp_done(void) +{ + wake_up(&sync_rcu_preempt_exp_wq); +} + +/* + * Wait for an rcu-preempt grace period, but expedite it. The basic idea + * is to rely in the fact that there is but one CPU, and that it is + * illegal for a task to invoke synchronize_rcu_expedited() while in a + * preemptible-RCU read-side critical section. Therefore, any such + * critical sections must correspond to blocked tasks, which must therefore + * be on the ->blkd_tasks list. So just record the current head of the + * list in the ->exp_tasks pointer, and wait for all tasks including and + * after the task pointed to by ->exp_tasks to drain. + */ +void synchronize_rcu_expedited(void) +{ + unsigned long flags; + struct rcu_preempt_ctrlblk *rpcp = &rcu_preempt_ctrlblk; + unsigned long snap; + + barrier(); /* ensure prior action seen before grace period. */ + + WARN_ON_ONCE(rcu_preempt_running_reader()); + + /* + * Acquire lock so that there is only one preemptible RCU grace + * period in flight. Of course, if someone does the expedited + * grace period for us while we are acquiring the lock, just leave. + */ + snap = sync_rcu_preempt_exp_count + 1; + mutex_lock(&sync_rcu_preempt_exp_mutex); + if (ULONG_CMP_LT(snap, sync_rcu_preempt_exp_count)) + goto unlock_mb_ret; /* Others did our work for us. */ + + local_irq_save(flags); + + /* + * All RCU readers have to already be on blkd_tasks because + * we cannot legally be executing in an RCU read-side critical + * section. + */ + + /* Snapshot current head of ->blkd_tasks list. */ + rpcp->exp_tasks = rpcp->blkd_tasks.next; + if (rpcp->exp_tasks == &rpcp->blkd_tasks) + rpcp->exp_tasks = NULL; + local_irq_restore(flags); + + /* Wait for tail of ->blkd_tasks list to drain. */ + if (rcu_preempted_readers_exp()) + wait_event(sync_rcu_preempt_exp_wq, + !rcu_preempted_readers_exp()); + + /* Clean up and exit. */ + barrier(); /* ensure expedited GP seen before counter increment. */ + sync_rcu_preempt_exp_count++; +unlock_mb_ret: + mutex_unlock(&sync_rcu_preempt_exp_mutex); + barrier(); /* ensure subsequent action seen after grace period. */ +} +EXPORT_SYMBOL_GPL(synchronize_rcu_expedited); + +/* + * Does preemptible RCU need the CPU to stay out of dynticks mode? + */ +int rcu_preempt_needs_cpu(void) +{ + if (!rcu_preempt_running_reader()) + rcu_preempt_cpu_qs(); + return rcu_preempt_ctrlblk.rcb.rcucblist != NULL; +} + +/* + * Check for a task exiting while in a preemptible -RCU read-side + * critical section, clean up if so. No need to issue warnings, + * as debug_check_no_locks_held() already does this if lockdep + * is enabled. + */ +void exit_rcu(void) +{ + struct task_struct *t = current; + + if (t->rcu_read_lock_nesting == 0) + return; + t->rcu_read_lock_nesting = 1; + rcu_read_unlock(); +} + +#else /* #ifdef CONFIG_TINY_PREEMPT_RCU */ + +/* + * Because preemptible RCU does not exist, it never has any callbacks + * to check. + */ +static void rcu_preempt_check_callbacks(void) +{ +} + +/* + * Because preemptible RCU does not exist, it never has any callbacks + * to remove. + */ +static void rcu_preempt_remove_callbacks(struct rcu_ctrlblk *rcp) +{ +} + +/* + * Because preemptible RCU does not exist, it never has any callbacks + * to process. + */ +static void rcu_preempt_process_callbacks(void) +{ +} + +#endif /* #else #ifdef CONFIG_TINY_PREEMPT_RCU */ + #ifdef CONFIG_DEBUG_LOCK_ALLOC #include -- cgit v1.2.3 From 8cdd32a918350430483751feaae1c19cef816f69 Mon Sep 17 00:00:00 2001 From: "Paul E. McKenney" Date: Mon, 9 Aug 2010 14:23:03 -0700 Subject: rcu: refer RCU CPU stall-warning victims to stallwarn.txt There is some documentation on RCU CPU stall warnings contained in Documentation/RCU/stallwarn.txt, but it will not be apparent to someone who runs into such a warning while under time pressure. This commit therefore adds comments preceding the printk()s pointing out the location of this documentation. Signed-off-by: Paul E. McKenney --- kernel/rcutree.c | 12 ++++++++++-- 1 file changed, 10 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/rcutree.c b/kernel/rcutree.c index 5aab7dabd0d5..ff214118e4b8 100644 --- a/kernel/rcutree.c +++ b/kernel/rcutree.c @@ -487,8 +487,11 @@ static void print_other_cpu_stall(struct rcu_state *rsp) rcu_print_task_stall(rnp); raw_spin_unlock_irqrestore(&rnp->lock, flags); - /* OK, time to rat on our buddy... */ - + /* + * OK, time to rat on our buddy... + * See Documentation/RCU/stallwarn.txt for info on how to debug + * RCU CPU stall warnings. + */ printk(KERN_ERR "INFO: %s detected stalls on CPUs/tasks: {", rsp->name); rcu_for_each_leaf_node(rsp, rnp) { @@ -517,6 +520,11 @@ static void print_cpu_stall(struct rcu_state *rsp) unsigned long flags; struct rcu_node *rnp = rcu_get_root(rsp); + /* + * OK, time to rat on ourselves... + * See Documentation/RCU/stallwarn.txt for info on how to debug + * RCU CPU stall warnings. + */ printk(KERN_ERR "INFO: %s detected stall on CPU %d (t=%lu jiffies)\n", rsp->name, smp_processor_id(), jiffies - rsp->gp_start); trigger_all_cpu_backtrace(); -- cgit v1.2.3 From 53d84e004d5e8c018be395c4330dc72fd60bd13e Mon Sep 17 00:00:00 2001 From: "Paul E. McKenney" Date: Tue, 10 Aug 2010 14:28:53 -0700 Subject: rcu: permit suppressing current grace period's CPU stall warnings When using a kernel debugger, a long sojourn in the debugger can get you lots of RCU CPU stall warnings once you resume. This might not be helpful, especially if you are using the system console. This patch therefore allows RCU CPU stall warnings to be suppressed, but only for the duration of the current set of grace periods. This differs from Jason's original patch in that it adds support for tiny RCU and preemptible RCU, and uses a slightly different method for suppressing the RCU CPU stall warning messages. Signed-off-by: Jason Wessel Signed-off-by: Paul E. McKenney Tested-by: Jason Wessel --- kernel/rcutree.c | 20 ++++++++++++++++++++ kernel/rcutree.h | 1 + kernel/rcutree_plugin.h | 18 ++++++++++++++++++ 3 files changed, 39 insertions(+) (limited to 'kernel') diff --git a/kernel/rcutree.c b/kernel/rcutree.c index ff214118e4b8..42140a860bb9 100644 --- a/kernel/rcutree.c +++ b/kernel/rcutree.c @@ -565,6 +565,22 @@ static int rcu_panic(struct notifier_block *this, unsigned long ev, void *ptr) return NOTIFY_DONE; } +/** + * rcu_cpu_stall_reset - prevent further stall warnings in current grace period + * + * Set the stall-warning timeout way off into the future, thus preventing + * any RCU CPU stall-warning messages from appearing in the current set of + * RCU grace periods. + * + * The caller must disable hard irqs. + */ +void rcu_cpu_stall_reset(void) +{ + rcu_sched_state.jiffies_stall = jiffies + ULONG_MAX / 2; + rcu_bh_state.jiffies_stall = jiffies + ULONG_MAX / 2; + rcu_preempt_stall_reset(); +} + static struct notifier_block rcu_panic_block = { .notifier_call = rcu_panic, }; @@ -584,6 +600,10 @@ static void check_cpu_stall(struct rcu_state *rsp, struct rcu_data *rdp) { } +void rcu_cpu_stall_reset(void) +{ +} + static void __init check_cpu_stall_init(void) { } diff --git a/kernel/rcutree.h b/kernel/rcutree.h index bb4d08695c45..7abd439a7573 100644 --- a/kernel/rcutree.h +++ b/kernel/rcutree.h @@ -372,6 +372,7 @@ static void rcu_report_unblock_qs_rnp(struct rcu_node *rnp, #ifdef CONFIG_RCU_CPU_STALL_DETECTOR static void rcu_print_detail_task_stall(struct rcu_state *rsp); static void rcu_print_task_stall(struct rcu_node *rnp); +static void rcu_preempt_stall_reset(void); #endif /* #ifdef CONFIG_RCU_CPU_STALL_DETECTOR */ static void rcu_preempt_check_blocked_tasks(struct rcu_node *rnp); #ifdef CONFIG_HOTPLUG_CPU diff --git a/kernel/rcutree_plugin.h b/kernel/rcutree_plugin.h index 63bb7714fdeb..561410f70d4a 100644 --- a/kernel/rcutree_plugin.h +++ b/kernel/rcutree_plugin.h @@ -417,6 +417,16 @@ static void rcu_print_task_stall(struct rcu_node *rnp) } } +/* + * Suppress preemptible RCU's CPU stall warnings by pushing the + * time of the next stall-warning message comfortably far into the + * future. + */ +static void rcu_preempt_stall_reset(void) +{ + rcu_preempt_state.jiffies_stall = jiffies + ULONG_MAX / 2; +} + #endif /* #ifdef CONFIG_RCU_CPU_STALL_DETECTOR */ /* @@ -867,6 +877,14 @@ static void rcu_print_task_stall(struct rcu_node *rnp) { } +/* + * Because preemptible RCU does not exist, there is no need to suppress + * its CPU stall warnings. + */ +static void rcu_preempt_stall_reset(void) +{ +} + #endif /* #ifdef CONFIG_RCU_CPU_STALL_DETECTOR */ /* -- cgit v1.2.3 From a3dc3fb161f9b4066c0fce22db72638af8baf83b Mon Sep 17 00:00:00 2001 From: "Paul E. McKenney" Date: Fri, 13 Aug 2010 16:16:25 -0700 Subject: rcu: repair code-duplication FIXMEs Combine the duplicate definitions of ULONG_CMP_GE(), ULONG_CMP_LT(), and rcu_preempt_depth() into include/linux/rcupdate.h. Signed-off-by: Paul E. McKenney --- kernel/rcutiny_plugin.h | 4 ---- kernel/rcutree.h | 3 --- 2 files changed, 7 deletions(-) (limited to 'kernel') diff --git a/kernel/rcutiny_plugin.h b/kernel/rcutiny_plugin.h index e6bc1b447c6c..c5bea1137dcb 100644 --- a/kernel/rcutiny_plugin.h +++ b/kernel/rcutiny_plugin.h @@ -26,10 +26,6 @@ #include -/* FIXME: merge with definitions in kernel/rcutree.h. */ -#define ULONG_CMP_GE(a, b) (ULONG_MAX / 2 >= (a) - (b)) -#define ULONG_CMP_LT(a, b) (ULONG_MAX / 2 < (a) - (b)) - /* Global control variables for preemptible RCU. */ struct rcu_preempt_ctrlblk { struct rcu_ctrlblk rcb; /* curtail: ->next ptr of last CB for GP. */ diff --git a/kernel/rcutree.h b/kernel/rcutree.h index 7abd439a7573..7918ba61873f 100644 --- a/kernel/rcutree.h +++ b/kernel/rcutree.h @@ -272,9 +272,6 @@ struct rcu_data { #endif /* #ifdef CONFIG_RCU_CPU_STALL_DETECTOR */ -#define ULONG_CMP_GE(a, b) (ULONG_MAX / 2 >= (a) - (b)) -#define ULONG_CMP_LT(a, b) (ULONG_MAX / 2 < (a) - (b)) - /* * RCU global state, including node hierarchy. This hierarchy is * represented in "heap" form in a dense array. The root (first level) -- cgit v1.2.3 From 7b0b759b65247cbc66384a912be9acf8d4800636 Mon Sep 17 00:00:00 2001 From: "Paul E. McKenney" Date: Tue, 17 Aug 2010 14:18:46 -0700 Subject: rcu: combine duplicate code, courtesy of CONFIG_PREEMPT_RCU The CONFIG_PREEMPT_RCU kernel configuration parameter was recently re-introduced, but as an indication of the type of RCU (preemptible vs. non-preemptible) instead of as selecting a given implementation. This commit uses CONFIG_PREEMPT_RCU to combine duplicate code from include/linux/rcutiny.h and include/linux/rcutree.h into include/linux/rcupdate.h. This commit also combines a few other pieces of duplicate code that have accumulated. Signed-off-by: Paul E. McKenney --- kernel/rcutree_plugin.h | 9 --------- 1 file changed, 9 deletions(-) (limited to 'kernel') diff --git a/kernel/rcutree_plugin.h b/kernel/rcutree_plugin.h index 561410f70d4a..87f60f06b18e 100644 --- a/kernel/rcutree_plugin.h +++ b/kernel/rcutree_plugin.h @@ -938,15 +938,6 @@ static void rcu_preempt_process_callbacks(void) { } -/* - * In classic RCU, call_rcu() is just call_rcu_sched(). - */ -void call_rcu(struct rcu_head *head, void (*func)(struct rcu_head *rcu)) -{ - call_rcu_sched(head, func); -} -EXPORT_SYMBOL_GPL(call_rcu); - /* * Wait for an rcu-preempt grace period, but make it happen quickly. * But because preemptable RCU does not exist, map to rcu-sched. -- cgit v1.2.3 From 80dcf60e6b97c7363971e7a0a788d8484d35f8a6 Mon Sep 17 00:00:00 2001 From: "Paul E. McKenney" Date: Thu, 19 Aug 2010 16:57:45 -0700 Subject: rcu: apply TINY_PREEMPT_RCU read-side speedup to TREE_PREEMPT_RCU Replace one of the ACCESS_ONCE() calls in each of __rcu_read_lock() and __rcu_read_unlock() with barrier() as suggested by Steve Rostedt in order to avoid the potential compiler-optimization-induced bug noted by Mathieu Desnoyers. Located-by: Mathieu Desnoyers Suggested-by: Steven Rostedt Signed-off-by: Paul E. McKenney --- kernel/rcutree_plugin.h | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/rcutree_plugin.h b/kernel/rcutree_plugin.h index 87f60f06b18e..e9e0bc74ff37 100644 --- a/kernel/rcutree_plugin.h +++ b/kernel/rcutree_plugin.h @@ -201,7 +201,7 @@ static void rcu_preempt_note_context_switch(int cpu) */ void __rcu_read_lock(void) { - ACCESS_ONCE(current->rcu_read_lock_nesting)++; + current->rcu_read_lock_nesting++; barrier(); /* needed if we ever invoke rcu_read_lock in rcutree.c */ } EXPORT_SYMBOL_GPL(__rcu_read_lock); @@ -344,7 +344,9 @@ void __rcu_read_unlock(void) struct task_struct *t = current; barrier(); /* needed if we ever invoke rcu_read_unlock in rcutree.c */ - if (--ACCESS_ONCE(t->rcu_read_lock_nesting) == 0 && + --t->rcu_read_lock_nesting; + barrier(); /* decrement before load of ->rcu_read_unlock_special */ + if (t->rcu_read_lock_nesting == 0 && unlikely(ACCESS_ONCE(t->rcu_read_unlock_special))) rcu_read_unlock_special(t); #ifdef CONFIG_PROVE_LOCKING -- cgit v1.2.3 From b35de43b31040828f83046f40fd34ba33146409d Mon Sep 17 00:00:00 2001 From: Andrea Righi Date: Thu, 19 Aug 2010 14:13:27 -0700 Subject: kfifo: implement missing __kfifo_skip_r() kfifo_skip() is currently broken, due to the missing of the internal helper function. Add it. Signed-off-by: Andrea Righi Cc: Greg KH Acked-by: Stefani Seibold Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/kfifo.c | 9 +++++++++ 1 file changed, 9 insertions(+) (limited to 'kernel') diff --git a/kernel/kfifo.c b/kernel/kfifo.c index 4502604ecadf..6b5580c57644 100644 --- a/kernel/kfifo.c +++ b/kernel/kfifo.c @@ -503,6 +503,15 @@ unsigned int __kfifo_out_r(struct __kfifo *fifo, void *buf, } EXPORT_SYMBOL(__kfifo_out_r); +void __kfifo_skip_r(struct __kfifo *fifo, size_t recsize) +{ + unsigned int n; + + n = __kfifo_peek_n(fifo, recsize); + fifo->out += n + recsize; +} +EXPORT_SYMBOL(__kfifo_skip_r); + int __kfifo_from_user_r(struct __kfifo *fifo, const void __user *from, unsigned long len, unsigned int *copied, size_t recsize) { -- cgit v1.2.3 From f335397d177c906256ee1bba28e8c49e8ec63817 Mon Sep 17 00:00:00 2001 From: Dmitry Torokhov Date: Tue, 17 Aug 2010 21:15:47 -0700 Subject: Input: sysrq - drop tty argument form handle_sysrq() Sysrq operations do not accept tty argument anymore so no need to pass it to us. [Stephen Rothwell : fix build breakage in drm code caused by sysrq using bool but not including linux/types.h] [Sachin Sant : fix build breakage in s390 keyboadr driver] Acked-by: Alan Cox Acked-by: Jason Wessel Acked-by: Greg Kroah-Hartman Signed-off-by: Dmitry Torokhov --- kernel/debug/kdb/kdb_main.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/debug/kdb/kdb_main.c b/kernel/debug/kdb/kdb_main.c index 28b844118bbd..caf057a3de0e 100644 --- a/kernel/debug/kdb/kdb_main.c +++ b/kernel/debug/kdb/kdb_main.c @@ -1929,7 +1929,7 @@ static int kdb_sr(int argc, const char **argv) if (argc != 1) return KDB_ARGCOUNT; kdb_trap_printk++; - __handle_sysrq(*argv[1], NULL, 0); + __handle_sysrq(*argv[1], false); kdb_trap_printk--; return 0; -- cgit v1.2.3 From 297c5eee372478fc32fec5fe8eed711eedb13f3d Mon Sep 17 00:00:00 2001 From: Linus Torvalds Date: Fri, 20 Aug 2010 16:24:55 -0700 Subject: mm: make the vma list be doubly linked It's a really simple list, and several of the users want to go backwards in it to find the previous vma. So rather than have to look up the previous entry with 'find_vma_prev()' or something similar, just make it doubly linked instead. Tested-by: Ian Campbell Signed-off-by: Linus Torvalds --- kernel/fork.c | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/fork.c b/kernel/fork.c index 856eac3ec52e..b7e9d60a675d 100644 --- a/kernel/fork.c +++ b/kernel/fork.c @@ -300,7 +300,7 @@ out: #ifdef CONFIG_MMU static int dup_mmap(struct mm_struct *mm, struct mm_struct *oldmm) { - struct vm_area_struct *mpnt, *tmp, **pprev; + struct vm_area_struct *mpnt, *tmp, *prev, **pprev; struct rb_node **rb_link, *rb_parent; int retval; unsigned long charge; @@ -328,6 +328,7 @@ static int dup_mmap(struct mm_struct *mm, struct mm_struct *oldmm) if (retval) goto out; + prev = NULL; for (mpnt = oldmm->mmap; mpnt; mpnt = mpnt->vm_next) { struct file *file; @@ -359,7 +360,7 @@ static int dup_mmap(struct mm_struct *mm, struct mm_struct *oldmm) goto fail_nomem_anon_vma_fork; tmp->vm_flags &= ~VM_LOCKED; tmp->vm_mm = mm; - tmp->vm_next = NULL; + tmp->vm_next = tmp->vm_prev = NULL; file = tmp->vm_file; if (file) { struct inode *inode = file->f_path.dentry->d_inode; @@ -392,6 +393,8 @@ static int dup_mmap(struct mm_struct *mm, struct mm_struct *oldmm) */ *pprev = tmp; pprev = &tmp->vm_next; + tmp->vm_prev = prev; + prev = tmp; __vma_link_rb(mm, tmp, rb_link, rb_parent); rb_link = &tmp->vm_rb.rb_right; -- cgit v1.2.3 From e36c886a0f9d624377977fa6cae309cfd7f362fa Mon Sep 17 00:00:00 2001 From: Arjan van de Ven Date: Sat, 21 Aug 2010 13:07:26 -0700 Subject: workqueue: Add basic tracepoints to track workqueue execution With the introduction of the new unified work queue thread pools, we lost one feature: It's no longer possible to know which worker is causing the CPU to wake out of idle. The result is that PowerTOP now reports a lot of "kworker/a:b" instead of more readable results. This patch adds a pair of tracepoints to the new workqueue code, similar in style to the timer/hrtimer tracepoints. With this pair of tracepoints, the next PowerTOP can correctly report which work item caused the wakeup (and how long it took): Interrupt (43) i915 time 3.51ms wakeups 141 Work ieee80211_iface_work time 0.81ms wakeups 29 Work do_dbs_timer time 0.55ms wakeups 24 Process Xorg time 21.36ms wakeups 4 Timer sched_rt_period_timer time 0.01ms wakeups 1 Signed-off-by: Arjan van de Ven Signed-off-by: Linus Torvalds --- kernel/workqueue.c | 9 +++++++++ 1 file changed, 9 insertions(+) (limited to 'kernel') diff --git a/kernel/workqueue.c b/kernel/workqueue.c index 2994a0e3a61c..8bd600c020e5 100644 --- a/kernel/workqueue.c +++ b/kernel/workqueue.c @@ -35,6 +35,9 @@ #include #include +#define CREATE_TRACE_POINTS +#include + #include "workqueue_sched.h" enum { @@ -1790,7 +1793,13 @@ static void process_one_work(struct worker *worker, struct work_struct *work) work_clear_pending(work); lock_map_acquire(&cwq->wq->lockdep_map); lock_map_acquire(&lockdep_map); + trace_workqueue_execute_start(work); f(work); + /* + * While we must be careful to not use "work" after this, the trace + * point will only record its address. + */ + trace_workqueue_execute_end(work); lock_map_release(&lockdep_map); lock_map_release(&cwq->wq->lockdep_map); -- cgit v1.2.3 From c6db67cda735d8ace5f19c3831240e1408679790 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Fri, 20 Aug 2010 11:49:15 +0200 Subject: watchdog: Don't throttle the watchdog Stephane reported that when the machine locks up, the regular ticks, which are responsible to resetting the throttle count, stop too. Hence the NMI watchdog can end up being throttled before it reports on the locked up state, and we end up being sad.. Cure this by having the watchdog overflow reset its own throttle count. Reported-by: Stephane Eranian Tested-by: Stephane Eranian Cc: Don Zickus Cc: Frederic Weisbecker Signed-off-by: Peter Zijlstra LKML-Reference: <1282215916.1926.4696.camel@laptop> Signed-off-by: Ingo Molnar --- kernel/watchdog.c | 3 +++ 1 file changed, 3 insertions(+) (limited to 'kernel') diff --git a/kernel/watchdog.c b/kernel/watchdog.c index 613bc1f04610..0d53c8e853b1 100644 --- a/kernel/watchdog.c +++ b/kernel/watchdog.c @@ -206,6 +206,9 @@ void watchdog_overflow_callback(struct perf_event *event, int nmi, struct perf_sample_data *data, struct pt_regs *regs) { + /* Ensure the watchdog never gets throttled */ + event->hw.interrupts = 0; + if (__get_cpu_var(watchdog_nmi_touch) == true) { __get_cpu_var(watchdog_nmi_touch) = false; return; -- cgit v1.2.3 From 9d0f4dcc5c4d1c5dd01172172684a45b5f49d740 Mon Sep 17 00:00:00 2001 From: Tim Chen Date: Wed, 18 Aug 2010 15:00:27 -0700 Subject: mutex: Improve the scalability of optimistic spinning MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit There is a scalability issue for current implementation of optimistic mutex spin in the kernel. It is found on a 8 node 64 core Nehalem-EX system (HT mode). The intention of the optimistic mutex spin is to busy wait and spin on a mutex if the owner of the mutex is running, in the hope that the mutex will be released soon and be acquired, without the thread trying to acquire mutex going to sleep. However, when we have a large number of threads, contending for the mutex, we could have the mutex grabbed by other thread, and then another ……, and we will keep spinning, wasting cpu cycles and adding to the contention. One possible fix is to quit spinning and put the current thread on wait-list if mutex lock switch to a new owner while we spin, indicating heavy contention (see the patch included). I did some testing on a 8 socket Nehalem-EX system with a total of 64 cores. Using Ingo's test-mutex program that creates/delete files with 256 threads (http://lkml.org/lkml/2006/1/8/50) , I see the following speed up after putting in the mutex spin fix: ./mutex-test V 256 10 Ops/sec 2.6.34 62864 With fix 197200 Repeating the test with Aim7 fserver workload, again there is a speed up with the fix: Jobs/min 2.6.34 91657 With fix 149325 To look at the impact on the distribution of mutex acquisition time, I collected the mutex acquisition time on Aim7 fserver workload with some instrumentation. The average acquisition time is reduced by 48% and number of contentions reduced by 32%. #contentions Time to acquire mutex (cycles) 2.6.34 72973 44765791 With fix 49210 23067129 The histogram of mutex acquisition time is listed below. The acquisition time is in 2^bin cycles. We see that without the fix, the acquisition time is mostly around 2^26 cycles. With the fix, we the distribution get spread out a lot more towards the lower cycles, starting from 2^13. However, there is an increase of the tail distribution with the fix at 2^28 and 2^29 cycles. It seems a small price to pay for the reduced average acquisition time and also getting the cpu to do useful work. Mutex acquisition time distribution (acq time = 2^bin cycles): 2.6.34 With Fix bin #occurrence % #occurrence % 11 2 0.00% 120 0.24% 12 10 0.01% 790 1.61% 13 14 0.02% 2058 4.18% 14 86 0.12% 3378 6.86% 15 393 0.54% 4831 9.82% 16 710 0.97% 4893 9.94% 17 815 1.12% 4667 9.48% 18 790 1.08% 5147 10.46% 19 580 0.80% 6250 12.70% 20 429 0.59% 6870 13.96% 21 311 0.43% 1809 3.68% 22 255 0.35% 2305 4.68% 23 317 0.44% 916 1.86% 24 610 0.84% 233 0.47% 25 3128 4.29% 95 0.19% 26 63902 87.69% 122 0.25% 27 619 0.85% 286 0.58% 28 0 0.00% 3536 7.19% 29 0 0.00% 903 1.83% 30 0 0.00% 0 0.00% I've done similar experiments with 2.6.35 kernel on smaller boxes as well. One is on a dual-socket Westmere box (12 cores total, with HT). Another experiment is on an old dual-socket Core 2 box (4 cores total, no HT) On the 12-core Westmere box, I see a 250% increase for Ingo's mutex-test program with my mutex patch but no significant difference in aim7's fserver workload. On the 4-core Core 2 box, I see the difference with the patch for both mutex-test and aim7 fserver are negligible. So far, it seems like the patch has not caused regression on smaller systems. Signed-off-by: Tim Chen Acked-by: Peter Zijlstra Cc: Linus Torvalds Cc: Andrew Morton Cc: Thomas Gleixner Cc: Frederic Weisbecker Cc: # .35.x LKML-Reference: <1282168827.9542.72.camel@schen9-DESK> Signed-off-by: Ingo Molnar --- kernel/sched.c | 10 +++++++++- 1 file changed, 9 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index 41541d79e3c8..09b574e7f4df 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -3865,8 +3865,16 @@ int mutex_spin_on_owner(struct mutex *lock, struct thread_info *owner) /* * Owner changed, break to re-assess state. */ - if (lock->owner != owner) + if (lock->owner != owner) { + /* + * If the lock has switched to a different owner, + * we likely have heavy contention. Return 0 to quit + * optimistic spinning and not contend further: + */ + if (lock->owner) + return 0; break; + } /* * Is that owner really running on that cpu? -- cgit v1.2.3 From 06bd6ebffae36d3b105677598c48e8bd0a10b205 Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Sun, 22 Aug 2010 23:19:42 +0900 Subject: workqueue: annotate lock context change Some of internal functions called within gcwq->lock context releases and regrabs the lock but were missing proper annotations. Add it. Signed-off-by: Namhyung Kim Signed-off-by: Tejun Heo --- kernel/workqueue.c | 6 ++++++ 1 file changed, 6 insertions(+) (limited to 'kernel') diff --git a/kernel/workqueue.c b/kernel/workqueue.c index 1001b6e3fcbd..7415f27a8aa7 100644 --- a/kernel/workqueue.c +++ b/kernel/workqueue.c @@ -1485,6 +1485,8 @@ static void gcwq_mayday_timeout(unsigned long __gcwq) * otherwise. */ static bool maybe_create_worker(struct global_cwq *gcwq) +__releases(&gcwq->lock) +__acquires(&gcwq->lock) { if (!need_to_create_worker(gcwq)) return false; @@ -1722,6 +1724,8 @@ static void cwq_dec_nr_in_flight(struct cpu_workqueue_struct *cwq, int color) * spin_lock_irq(gcwq->lock) which is released and regrabbed. */ static void process_one_work(struct worker *worker, struct work_struct *work) +__releases(&gcwq->lock) +__acquires(&gcwq->lock) { struct cpu_workqueue_struct *cwq = get_work_cwq(work); struct global_cwq *gcwq = cwq->gcwq; @@ -3230,6 +3234,8 @@ static int __cpuinit trustee_thread(void *__gcwq) * multiple times. To be used by cpu_callback. */ static void __cpuinit wait_trustee_state(struct global_cwq *gcwq, int state) +__releases(&gcwq->lock) +__acquires(&gcwq->lock) { if (!(gcwq->trustee_state == state || gcwq->trustee_state == TRUSTEE_DONE)) { -- cgit v1.2.3 From 972fa1c5316d18c8297123e08e9b6930ca34f888 Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Sun, 22 Aug 2010 23:19:43 +0900 Subject: workqueue: mark lock acquisition on worker_maybe_bind_and_lock() worker_maybe_bind_and_lock() actually grabs gcwq->lock but was missing proper annotation. Add it. So this patch will remove following sparse warnings: kernel/workqueue.c:1214:13: warning: context imbalance in 'worker_maybe_bind_and_lock' - wrong count at exit arch/x86/include/asm/irqflags.h:44:9: warning: context imbalance in 'worker_rebind_fn' - unexpected unlock kernel/workqueue.c:1991:17: warning: context imbalance in 'rescuer_thread' - unexpected unlock Signed-off-by: Namhyung Kim Signed-off-by: Tejun Heo --- kernel/workqueue.c | 1 + 1 file changed, 1 insertion(+) (limited to 'kernel') diff --git a/kernel/workqueue.c b/kernel/workqueue.c index 7415f27a8aa7..cc3456f96c56 100644 --- a/kernel/workqueue.c +++ b/kernel/workqueue.c @@ -1212,6 +1212,7 @@ static void worker_leave_idle(struct worker *worker) * bound), %false if offline. */ static bool worker_maybe_bind_and_lock(struct worker *worker) +__acquires(&gcwq->lock) { struct global_cwq *gcwq = worker->gcwq; struct task_struct *task = worker->task; -- cgit v1.2.3 From e41e704bc4f49057fc68b643108366e6e6781aa3 Mon Sep 17 00:00:00 2001 From: Tejun Heo Date: Tue, 24 Aug 2010 14:22:47 +0200 Subject: workqueue: improve destroy_workqueue() debuggability Now that the worklist is global, having works pending after wq destruction can easily lead to oops and destroy_workqueue() have several BUG_ON()s to catch these cases. Unfortunately, BUG_ON() doesn't tell much about how the work became pending after the final flush_workqueue(). This patch adds WQ_DYING which is set before the final flush begins. If a work is requested to be queued on a dying workqueue, WARN_ON_ONCE() is triggered and the request is ignored. This clearly indicates which caller is trying to queue a work on a dying workqueue and keeps the system working in most cases. Locking rule comment is updated such that the 'I' rule includes modifying the field from destruction path. Signed-off-by: Tejun Heo --- kernel/workqueue.c | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/workqueue.c b/kernel/workqueue.c index cc3456f96c56..362b50d092e2 100644 --- a/kernel/workqueue.c +++ b/kernel/workqueue.c @@ -87,7 +87,8 @@ enum { /* * Structure fields follow one of the following exclusion rules. * - * I: Set during initialization and read-only afterwards. + * I: Modifiable by initialization/destruction paths and read-only for + * everyone else. * * P: Preemption protected. Disabling preemption is enough and should * only be modified and accessed from the local cpu. @@ -944,6 +945,9 @@ static void __queue_work(unsigned int cpu, struct workqueue_struct *wq, debug_work_activate(work); + if (WARN_ON_ONCE(wq->flags & WQ_DYING)) + return; + /* determine gcwq to use */ if (!(wq->flags & WQ_UNBOUND)) { struct global_cwq *last_gcwq; @@ -2828,6 +2832,7 @@ void destroy_workqueue(struct workqueue_struct *wq) { unsigned int cpu; + wq->flags |= WQ_DYING; flush_workqueue(wq); /* -- cgit v1.2.3 From bac1e74dba9755128748b872a0f304dad4d198c6 Mon Sep 17 00:00:00 2001 From: David Alan Gilbert Date: Tue, 24 Aug 2010 20:22:18 +0200 Subject: PM QoS: Fix kzalloc() parameters swapped in pm_qos_power_open() sparse spotted that the kzalloc() in pm_qos_power_open() in the current Linus' git tree had its parameters swapped. Fix this. Signed-off-by: David Alan Gilbert Acked-by: mark gross Signed-off-by: Rafael J. Wysocki --- kernel/pm_qos_params.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/pm_qos_params.c b/kernel/pm_qos_params.c index 996a4dec5f96..9da439b419aa 100644 --- a/kernel/pm_qos_params.c +++ b/kernel/pm_qos_params.c @@ -348,7 +348,7 @@ static int pm_qos_power_open(struct inode *inode, struct file *filp) pm_qos_class = find_pm_qos_object_by_minor(iminor(inode)); if (pm_qos_class >= 0) { - struct pm_qos_request_list *req = kzalloc(GFP_KERNEL, sizeof(*req)); + struct pm_qos_request_list *req = kzalloc(sizeof(*req), GFP_KERNEL); if (!req) return -ENOMEM; -- cgit v1.2.3 From 8a2e8e5dec7e29c56a46ba176c664ab6a3d04118 Mon Sep 17 00:00:00 2001 From: Tejun Heo Date: Wed, 25 Aug 2010 10:33:56 +0200 Subject: workqueue: fix cwq->nr_active underflow cwq->nr_active is used to keep track of how many work items are active for the cpu workqueue, where 'active' is defined as either pending on global worklist or executing. This is used to implement the max_active limit and workqueue freezing. If a work item is queued after nr_active has already reached max_active, the work item doesn't increment nr_active and is put on the delayed queue and gets activated later as previous active work items retire. try_to_grab_pending() which is used in the cancellation path unconditionally decremented nr_active whether the work item being cancelled is currently active or delayed, so cancelling a delayed work item makes nr_active underflow. This breaks max_active enforcement and triggers BUG_ON() in destroy_workqueue() later on. This patch fixes this bug by adding a flag WORK_STRUCT_DELAYED, which is set while a work item in on the delayed list and making try_to_grab_pending() decrement nr_active iff the work item is currently active. The addition of the flag enlarges cwq alignment to 256 bytes which is getting a bit too large. It's scheduled to be reduced back to 128 bytes by merging WORK_STRUCT_PENDING and WORK_STRUCT_CWQ in the next devel cycle. Signed-off-by: Tejun Heo Reported-by: Johannes Berg --- kernel/workqueue.c | 30 ++++++++++++++++++++---------- 1 file changed, 20 insertions(+), 10 deletions(-) (limited to 'kernel') diff --git a/kernel/workqueue.c b/kernel/workqueue.c index 362b50d092e2..a2dccfca03ba 100644 --- a/kernel/workqueue.c +++ b/kernel/workqueue.c @@ -941,6 +941,7 @@ static void __queue_work(unsigned int cpu, struct workqueue_struct *wq, struct global_cwq *gcwq; struct cpu_workqueue_struct *cwq; struct list_head *worklist; + unsigned int work_flags; unsigned long flags; debug_work_activate(work); @@ -990,14 +991,17 @@ static void __queue_work(unsigned int cpu, struct workqueue_struct *wq, BUG_ON(!list_empty(&work->entry)); cwq->nr_in_flight[cwq->work_color]++; + work_flags = work_color_to_flags(cwq->work_color); if (likely(cwq->nr_active < cwq->max_active)) { cwq->nr_active++; worklist = gcwq_determine_ins_pos(gcwq, cwq); - } else + } else { + work_flags |= WORK_STRUCT_DELAYED; worklist = &cwq->delayed_works; + } - insert_work(cwq, work, worklist, work_color_to_flags(cwq->work_color)); + insert_work(cwq, work, worklist, work_flags); spin_unlock_irqrestore(&gcwq->lock, flags); } @@ -1666,6 +1670,7 @@ static void cwq_activate_first_delayed(struct cpu_workqueue_struct *cwq) struct list_head *pos = gcwq_determine_ins_pos(cwq->gcwq, cwq); move_linked_works(work, pos, NULL); + __clear_bit(WORK_STRUCT_DELAYED_BIT, work_data_bits(work)); cwq->nr_active++; } @@ -1673,6 +1678,7 @@ static void cwq_activate_first_delayed(struct cpu_workqueue_struct *cwq) * cwq_dec_nr_in_flight - decrement cwq's nr_in_flight * @cwq: cwq of interest * @color: color of work which left the queue + * @delayed: for a delayed work * * A work either has completed or is removed from pending queue, * decrement nr_in_flight of its cwq and handle workqueue flushing. @@ -1680,19 +1686,22 @@ static void cwq_activate_first_delayed(struct cpu_workqueue_struct *cwq) * CONTEXT: * spin_lock_irq(gcwq->lock). */ -static void cwq_dec_nr_in_flight(struct cpu_workqueue_struct *cwq, int color) +static void cwq_dec_nr_in_flight(struct cpu_workqueue_struct *cwq, int color, + bool delayed) { /* ignore uncolored works */ if (color == WORK_NO_COLOR) return; cwq->nr_in_flight[color]--; - cwq->nr_active--; - if (!list_empty(&cwq->delayed_works)) { - /* one down, submit a delayed one */ - if (cwq->nr_active < cwq->max_active) - cwq_activate_first_delayed(cwq); + if (!delayed) { + cwq->nr_active--; + if (!list_empty(&cwq->delayed_works)) { + /* one down, submit a delayed one */ + if (cwq->nr_active < cwq->max_active) + cwq_activate_first_delayed(cwq); + } } /* is flush in progress and are we at the flushing tip? */ @@ -1823,7 +1832,7 @@ __acquires(&gcwq->lock) hlist_del_init(&worker->hentry); worker->current_work = NULL; worker->current_cwq = NULL; - cwq_dec_nr_in_flight(cwq, work_color); + cwq_dec_nr_in_flight(cwq, work_color, false); } /** @@ -2388,7 +2397,8 @@ static int try_to_grab_pending(struct work_struct *work) debug_work_deactivate(work); list_del_init(&work->entry); cwq_dec_nr_in_flight(get_work_cwq(work), - get_work_color(work)); + get_work_color(work), + *work_data_bits(work) & WORK_STRUCT_DELAYED); ret = 1; } } -- cgit v1.2.3 From 151772dbfad4dbe81721e40f9b3d588ea77bb7aa Mon Sep 17 00:00:00 2001 From: Anton Blanchard Date: Wed, 25 Aug 2010 11:32:38 +1000 Subject: tracing/trace_stack: Fix stack trace on ppc64 save_stack_trace() stores the instruction pointer, not the function descriptor. On ppc64 the trace stack code currently dereferences the instruction pointer and shows 8 bytes of instructions in our backtraces: # cat /sys/kernel/debug/tracing/stack_trace Depth Size Location (26 entries) ----- ---- -------- 0) 5424 112 0x6000000048000004 1) 5312 160 0x60000000ebad01b0 2) 5152 160 0x2c23000041c20030 3) 4992 240 0x600000007c781b79 4) 4752 160 0xe84100284800000c 5) 4592 192 0x600000002fa30000 6) 4400 256 0x7f1800347b7407e0 7) 4144 208 0xe89f0108f87f0070 8) 3936 272 0xe84100282fa30000 Since we aren't dealing with function descriptors, use %pS instead of %pF to fix it: # cat /sys/kernel/debug/tracing/stack_trace Depth Size Location (26 entries) ----- ---- -------- 0) 5424 112 ftrace_call+0x4/0x8 1) 5312 160 .current_io_context+0x28/0x74 2) 5152 160 .get_io_context+0x48/0xa0 3) 4992 240 .cfq_set_request+0x94/0x4c4 4) 4752 160 .elv_set_request+0x60/0x84 5) 4592 192 .get_request+0x2d4/0x468 6) 4400 256 .get_request_wait+0x7c/0x258 7) 4144 208 .__make_request+0x49c/0x610 8) 3936 272 .generic_make_request+0x390/0x434 Signed-off-by: Anton Blanchard Cc: rostedt@goodmis.org Cc: fweisbec@gmail.com LKML-Reference: <20100825013238.GE28360@kryten> Signed-off-by: Ingo Molnar --- kernel/trace/trace_stack.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/trace/trace_stack.c b/kernel/trace/trace_stack.c index 056468eae7cf..a6b7e0e0f3eb 100644 --- a/kernel/trace/trace_stack.c +++ b/kernel/trace/trace_stack.c @@ -249,7 +249,7 @@ static int trace_lookup_stack(struct seq_file *m, long i) { unsigned long addr = stack_dump_trace[i]; - return seq_printf(m, "%pF\n", (void *)addr); + return seq_printf(m, "%pS\n", (void *)addr); } static void print_disabled(struct seq_file *m) -- cgit v1.2.3 From 25cc69ec34a563e943e85b3b68a79a8aac7f076d Mon Sep 17 00:00:00 2001 From: Saravana Kannan Date: Thu, 26 Aug 2010 20:18:43 +0200 Subject: PM QoS: Fix inline documentation. Fix the pm_qos_add_request() kerneldoc comment that doesn't reflect the behavior of the function after the last PM QoS update. Signed-off-by: Saravana Kannan Acked-by: mark gross Signed-off-by: Rafael J. Wysocki --- kernel/pm_qos_params.c | 10 ++++++---- 1 file changed, 6 insertions(+), 4 deletions(-) (limited to 'kernel') diff --git a/kernel/pm_qos_params.c b/kernel/pm_qos_params.c index 9da439b419aa..b7e4c362361b 100644 --- a/kernel/pm_qos_params.c +++ b/kernel/pm_qos_params.c @@ -212,15 +212,17 @@ EXPORT_SYMBOL_GPL(pm_qos_request_active); /** * pm_qos_add_request - inserts new qos request into the list - * @pm_qos_class: identifies which list of qos request to us + * @dep: pointer to a preallocated handle + * @pm_qos_class: identifies which list of qos request to use * @value: defines the qos request * * This function inserts a new entry in the pm_qos_class list of requested qos * performance characteristics. It recomputes the aggregate QoS expectations - * for the pm_qos_class of parameters, and returns the pm_qos_request list - * element as a handle for use in updating and removal. Call needs to save - * this handle for later use. + * for the pm_qos_class of parameters and initializes the pm_qos_request_list + * handle. Caller needs to save this handle for later use in updates and + * removal. */ + void pm_qos_add_request(struct pm_qos_request_list *dep, int pm_qos_class, s32 value) { -- cgit v1.2.3 From dd7c4d89730a1be2c1d361a8ae1f0fe9465ccf9c Mon Sep 17 00:00:00 2001 From: "Paul E. McKenney" Date: Fri, 27 Aug 2010 10:51:17 -0700 Subject: rcu: performance fixes to TINY_PREEMPT_RCU callback checking This commit tightens up checks in rcu_preempt_check_callbacks() to avoid unnecessary special handling at rcu_read_unlock() time. Signed-off-by: Paul E. McKenney --- kernel/rcutiny_plugin.h | 12 ++++++++---- 1 file changed, 8 insertions(+), 4 deletions(-) (limited to 'kernel') diff --git a/kernel/rcutiny_plugin.h b/kernel/rcutiny_plugin.h index c5bea1137dcb..6ceca4f745ff 100644 --- a/kernel/rcutiny_plugin.h +++ b/kernel/rcutiny_plugin.h @@ -72,7 +72,7 @@ static void rcu_report_exp_done(void); /* * Return true if the CPU has not yet responded to the current grace period. */ -static int rcu_cpu_cur_gp(void) +static int rcu_cpu_blocking_cur_gp(void) { return rcu_preempt_ctrlblk.gpcpu != rcu_preempt_ctrlblk.gpnum; } @@ -229,7 +229,7 @@ void rcu_preempt_note_context_switch(void) * cannot end. */ list_add(&t->rcu_node_entry, &rcu_preempt_ctrlblk.blkd_tasks); - if (rcu_cpu_cur_gp()) + if (rcu_cpu_blocking_cur_gp()) rcu_preempt_ctrlblk.gp_tasks = &t->rcu_node_entry; } @@ -368,12 +368,16 @@ static void rcu_preempt_check_callbacks(void) { struct task_struct *t = current; - if (!rcu_preempt_running_reader() && rcu_preempt_gp_in_progress()) + if (rcu_preempt_gp_in_progress() && + (!rcu_preempt_running_reader() || + !rcu_cpu_blocking_cur_gp())) rcu_preempt_cpu_qs(); if (&rcu_preempt_ctrlblk.rcb.rcucblist != rcu_preempt_ctrlblk.rcb.donetail) raise_softirq(RCU_SOFTIRQ); - if (rcu_preempt_gp_in_progress() && rcu_preempt_running_reader()) + if (rcu_preempt_gp_in_progress() && + rcu_cpu_blocking_cur_gp() && + rcu_preempt_running_reader()) t->rcu_read_unlock_special |= RCU_READ_UNLOCK_NEED_QS; } -- cgit v1.2.3 From a587d2daebcd2bc159d4348b6a7b028950a6d803 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Wed, 25 Aug 2010 13:39:18 -0700 Subject: x86: Remove not used early_res code and some functions in e820.c that are not used anymore Signed-off-by: Yinghai Lu Signed-off-by: H. Peter Anvin --- kernel/Makefile | 1 - kernel/early_res.c | 590 ----------------------------------------------------- 2 files changed, 591 deletions(-) delete mode 100644 kernel/early_res.c (limited to 'kernel') diff --git a/kernel/Makefile b/kernel/Makefile index 057472fbc272..80e61c3e44ae 100644 --- a/kernel/Makefile +++ b/kernel/Makefile @@ -11,7 +11,6 @@ obj-y = sched.o fork.o exec_domain.o panic.o printk.o \ hrtimer.o rwsem.o nsproxy.o srcu.o semaphore.o \ notifier.o ksysfs.o pm_qos_params.o sched_clock.o cred.o \ async.o range.o -obj-$(CONFIG_HAVE_EARLY_RES) += early_res.o obj-y += groups.o ifdef CONFIG_FUNCTION_TRACER diff --git a/kernel/early_res.c b/kernel/early_res.c deleted file mode 100644 index 7bfae887f211..000000000000 --- a/kernel/early_res.c +++ /dev/null @@ -1,590 +0,0 @@ -/* - * early_res, could be used to replace bootmem - */ -#include -#include -#include -#include -#include -#include -#include -#include - -/* - * Early reserved memory areas. - */ -/* - * need to make sure this one is bigger enough before - * find_fw_memmap_area could be used - */ -#define MAX_EARLY_RES_X 32 - -struct early_res { - u64 start, end; - char name[15]; - char overlap_ok; -}; -static struct early_res early_res_x[MAX_EARLY_RES_X] __initdata; - -static int max_early_res __initdata = MAX_EARLY_RES_X; -static struct early_res *early_res __initdata = &early_res_x[0]; -static int early_res_count __initdata; - -static int __init find_overlapped_early(u64 start, u64 end) -{ - int i; - struct early_res *r; - - for (i = 0; i < max_early_res && early_res[i].end; i++) { - r = &early_res[i]; - if (end > r->start && start < r->end) - break; - } - - return i; -} - -/* - * Drop the i-th range from the early reservation map, - * by copying any higher ranges down one over it, and - * clearing what had been the last slot. - */ -static void __init drop_range(int i) -{ - int j; - - for (j = i + 1; j < max_early_res && early_res[j].end; j++) - ; - - memmove(&early_res[i], &early_res[i + 1], - (j - 1 - i) * sizeof(struct early_res)); - - early_res[j - 1].end = 0; - early_res_count--; -} - -static void __init drop_range_partial(int i, u64 start, u64 end) -{ - u64 common_start, common_end; - u64 old_start, old_end; - - old_start = early_res[i].start; - old_end = early_res[i].end; - common_start = max(old_start, start); - common_end = min(old_end, end); - - /* no overlap ? */ - if (common_start >= common_end) - return; - - if (old_start < common_start) { - /* make head segment */ - early_res[i].end = common_start; - if (old_end > common_end) { - char name[15]; - - /* - * Save a local copy of the name, since the - * early_res array could get resized inside - * reserve_early_without_check() -> - * __check_and_double_early_res(), which would - * make the current name pointer invalid. - */ - strncpy(name, early_res[i].name, - sizeof(early_res[i].name) - 1); - /* add another for left over on tail */ - reserve_early_without_check(common_end, old_end, name); - } - return; - } else { - if (old_end > common_end) { - /* reuse the entry for tail left */ - early_res[i].start = common_end; - return; - } - /* all covered */ - drop_range(i); - } -} - -/* - * Split any existing ranges that: - * 1) are marked 'overlap_ok', and - * 2) overlap with the stated range [start, end) - * into whatever portion (if any) of the existing range is entirely - * below or entirely above the stated range. Drop the portion - * of the existing range that overlaps with the stated range, - * which will allow the caller of this routine to then add that - * stated range without conflicting with any existing range. - */ -static void __init drop_overlaps_that_are_ok(u64 start, u64 end) -{ - int i; - struct early_res *r; - u64 lower_start, lower_end; - u64 upper_start, upper_end; - char name[15]; - - for (i = 0; i < max_early_res && early_res[i].end; i++) { - r = &early_res[i]; - - /* Continue past non-overlapping ranges */ - if (end <= r->start || start >= r->end) - continue; - - /* - * Leave non-ok overlaps as is; let caller - * panic "Overlapping early reservations" - * when it hits this overlap. - */ - if (!r->overlap_ok) - return; - - /* - * We have an ok overlap. We will drop it from the early - * reservation map, and add back in any non-overlapping - * portions (lower or upper) as separate, overlap_ok, - * non-overlapping ranges. - */ - - /* 1. Note any non-overlapping (lower or upper) ranges. */ - strncpy(name, r->name, sizeof(name) - 1); - - lower_start = lower_end = 0; - upper_start = upper_end = 0; - if (r->start < start) { - lower_start = r->start; - lower_end = start; - } - if (r->end > end) { - upper_start = end; - upper_end = r->end; - } - - /* 2. Drop the original ok overlapping range */ - drop_range(i); - - i--; /* resume for-loop on copied down entry */ - - /* 3. Add back in any non-overlapping ranges. */ - if (lower_end) - reserve_early_overlap_ok(lower_start, lower_end, name); - if (upper_end) - reserve_early_overlap_ok(upper_start, upper_end, name); - } -} - -static void __init __reserve_early(u64 start, u64 end, char *name, - int overlap_ok) -{ - int i; - struct early_res *r; - - i = find_overlapped_early(start, end); - if (i >= max_early_res) - panic("Too many early reservations"); - r = &early_res[i]; - if (r->end) - panic("Overlapping early reservations " - "%llx-%llx %s to %llx-%llx %s\n", - start, end - 1, name ? name : "", r->start, - r->end - 1, r->name); - r->start = start; - r->end = end; - r->overlap_ok = overlap_ok; - if (name) - strncpy(r->name, name, sizeof(r->name) - 1); - early_res_count++; -} - -/* - * A few early reservtations come here. - * - * The 'overlap_ok' in the name of this routine does -not- mean it - * is ok for these reservations to overlap an earlier reservation. - * Rather it means that it is ok for subsequent reservations to - * overlap this one. - * - * Use this entry point to reserve early ranges when you are doing - * so out of "Paranoia", reserving perhaps more memory than you need, - * just in case, and don't mind a subsequent overlapping reservation - * that is known to be needed. - * - * The drop_overlaps_that_are_ok() call here isn't really needed. - * It would be needed if we had two colliding 'overlap_ok' - * reservations, so that the second such would not panic on the - * overlap with the first. We don't have any such as of this - * writing, but might as well tolerate such if it happens in - * the future. - */ -void __init reserve_early_overlap_ok(u64 start, u64 end, char *name) -{ - drop_overlaps_that_are_ok(start, end); - __reserve_early(start, end, name, 1); -} - -static void __init __check_and_double_early_res(u64 ex_start, u64 ex_end) -{ - u64 start, end, size, mem; - struct early_res *new; - - /* do we have enough slots left ? */ - if ((max_early_res - early_res_count) > max(max_early_res/8, 2)) - return; - - /* double it */ - mem = -1ULL; - size = sizeof(struct early_res) * max_early_res * 2; - if (early_res == early_res_x) - start = 0; - else - start = early_res[0].end; - end = ex_start; - if (start + size < end) - mem = find_fw_memmap_area(start, end, size, - sizeof(struct early_res)); - if (mem == -1ULL) { - start = ex_end; - end = get_max_mapped(); - if (start + size < end) - mem = find_fw_memmap_area(start, end, size, - sizeof(struct early_res)); - } - if (mem == -1ULL) - panic("can not find more space for early_res array"); - - new = __va(mem); - /* save the first one for own */ - new[0].start = mem; - new[0].end = mem + size; - new[0].overlap_ok = 0; - /* copy old to new */ - if (early_res == early_res_x) { - memcpy(&new[1], &early_res[0], - sizeof(struct early_res) * max_early_res); - memset(&new[max_early_res+1], 0, - sizeof(struct early_res) * (max_early_res - 1)); - early_res_count++; - } else { - memcpy(&new[1], &early_res[1], - sizeof(struct early_res) * (max_early_res - 1)); - memset(&new[max_early_res], 0, - sizeof(struct early_res) * max_early_res); - } - memset(&early_res[0], 0, sizeof(struct early_res) * max_early_res); - early_res = new; - max_early_res *= 2; - printk(KERN_DEBUG "early_res array is doubled to %d at [%llx - %llx]\n", - max_early_res, mem, mem + size - 1); -} - -/* - * Most early reservations come here. - * - * We first have drop_overlaps_that_are_ok() drop any pre-existing - * 'overlap_ok' ranges, so that we can then reserve this memory - * range without risk of panic'ing on an overlapping overlap_ok - * early reservation. - */ -void __init reserve_early(u64 start, u64 end, char *name) -{ - if (start >= end) - return; - - __check_and_double_early_res(start, end); - - drop_overlaps_that_are_ok(start, end); - __reserve_early(start, end, name, 0); -} - -void __init reserve_early_without_check(u64 start, u64 end, char *name) -{ - struct early_res *r; - - if (start >= end) - return; - - __check_and_double_early_res(start, end); - - r = &early_res[early_res_count]; - - r->start = start; - r->end = end; - r->overlap_ok = 0; - if (name) - strncpy(r->name, name, sizeof(r->name) - 1); - early_res_count++; -} - -void __init free_early(u64 start, u64 end) -{ - struct early_res *r; - int i; - - kmemleak_free_part(__va(start), end - start); - - i = find_overlapped_early(start, end); - r = &early_res[i]; - if (i >= max_early_res || r->end != end || r->start != start) - panic("free_early on not reserved area: %llx-%llx!", - start, end - 1); - - drop_range(i); -} - -void __init free_early_partial(u64 start, u64 end) -{ - struct early_res *r; - int i; - - kmemleak_free_part(__va(start), end - start); - - if (start == end) - return; - - if (WARN_ONCE(start > end, " wrong range [%#llx, %#llx]\n", start, end)) - return; - -try_next: - i = find_overlapped_early(start, end); - if (i >= max_early_res) - return; - - r = &early_res[i]; - /* hole ? */ - if (r->end >= end && r->start <= start) { - drop_range_partial(i, start, end); - return; - } - - drop_range_partial(i, start, end); - goto try_next; -} - -#ifdef CONFIG_NO_BOOTMEM -static void __init subtract_early_res(struct range *range, int az) -{ - int i, count; - u64 final_start, final_end; - int idx = 0; - - count = 0; - for (i = 0; i < max_early_res && early_res[i].end; i++) - count++; - - /* need to skip first one ?*/ - if (early_res != early_res_x) - idx = 1; - -#define DEBUG_PRINT_EARLY_RES 1 - -#if DEBUG_PRINT_EARLY_RES - printk(KERN_INFO "Subtract (%d early reservations)\n", count); -#endif - for (i = idx; i < count; i++) { - struct early_res *r = &early_res[i]; -#if DEBUG_PRINT_EARLY_RES - printk(KERN_INFO " #%d [%010llx - %010llx] %15s\n", i, - r->start, r->end, r->name); -#endif - final_start = PFN_DOWN(r->start); - final_end = PFN_UP(r->end); - if (final_start >= final_end) - continue; - subtract_range(range, az, final_start, final_end); - } - -} - -int __init get_free_all_memory_range(struct range **rangep, int nodeid) -{ - int i, count; - u64 start = 0, end; - u64 size; - u64 mem; - struct range *range; - int nr_range; - - count = 0; - for (i = 0; i < max_early_res && early_res[i].end; i++) - count++; - - count *= 2; - - size = sizeof(struct range) * count; - end = get_max_mapped(); -#ifdef MAX_DMA32_PFN - if (end > (MAX_DMA32_PFN << PAGE_SHIFT)) - start = MAX_DMA32_PFN << PAGE_SHIFT; -#endif - mem = find_fw_memmap_area(start, end, size, sizeof(struct range)); - if (mem == -1ULL) - panic("can not find more space for range free"); - - range = __va(mem); - /* use early_node_map[] and early_res to get range array at first */ - memset(range, 0, size); - nr_range = 0; - - /* need to go over early_node_map to find out good range for node */ - nr_range = add_from_early_node_map(range, count, nr_range, nodeid); -#ifdef CONFIG_X86_32 - subtract_range(range, count, max_low_pfn, -1ULL); -#endif - subtract_early_res(range, count); - nr_range = clean_sort_range(range, count); - - /* need to clear it ? */ - if (nodeid == MAX_NUMNODES) { - memset(&early_res[0], 0, - sizeof(struct early_res) * max_early_res); - early_res = NULL; - max_early_res = 0; - } - - *rangep = range; - return nr_range; -} -#else -void __init early_res_to_bootmem(u64 start, u64 end) -{ - int i, count; - u64 final_start, final_end; - int idx = 0; - - count = 0; - for (i = 0; i < max_early_res && early_res[i].end; i++) - count++; - - /* need to skip first one ?*/ - if (early_res != early_res_x) - idx = 1; - - printk(KERN_INFO "(%d/%d early reservations) ==> bootmem [%010llx - %010llx]\n", - count - idx, max_early_res, start, end); - for (i = idx; i < count; i++) { - struct early_res *r = &early_res[i]; - printk(KERN_INFO " #%d [%010llx - %010llx] %16s", i, - r->start, r->end, r->name); - final_start = max(start, r->start); - final_end = min(end, r->end); - if (final_start >= final_end) { - printk(KERN_CONT "\n"); - continue; - } - printk(KERN_CONT " ==> [%010llx - %010llx]\n", - final_start, final_end); - reserve_bootmem_generic(final_start, final_end - final_start, - BOOTMEM_DEFAULT); - } - /* clear them */ - memset(&early_res[0], 0, sizeof(struct early_res) * max_early_res); - early_res = NULL; - max_early_res = 0; - early_res_count = 0; -} -#endif - -/* Check for already reserved areas */ -static inline int __init bad_addr(u64 *addrp, u64 size, u64 align) -{ - int i; - u64 addr = *addrp; - int changed = 0; - struct early_res *r; -again: - i = find_overlapped_early(addr, addr + size); - r = &early_res[i]; - if (i < max_early_res && r->end) { - *addrp = addr = round_up(r->end, align); - changed = 1; - goto again; - } - return changed; -} - -/* Check for already reserved areas */ -static inline int __init bad_addr_size(u64 *addrp, u64 *sizep, u64 align) -{ - int i; - u64 addr = *addrp, last; - u64 size = *sizep; - int changed = 0; -again: - last = addr + size; - for (i = 0; i < max_early_res && early_res[i].end; i++) { - struct early_res *r = &early_res[i]; - if (last > r->start && addr < r->start) { - size = r->start - addr; - changed = 1; - goto again; - } - if (last > r->end && addr < r->end) { - addr = round_up(r->end, align); - size = last - addr; - changed = 1; - goto again; - } - if (last <= r->end && addr >= r->start) { - (*sizep)++; - return 0; - } - } - if (changed) { - *addrp = addr; - *sizep = size; - } - return changed; -} - -/* - * Find a free area with specified alignment in a specific range. - * only with the area.between start to end is active range from early_node_map - * so they are good as RAM - */ -u64 __init find_early_area(u64 ei_start, u64 ei_last, u64 start, u64 end, - u64 size, u64 align) -{ - u64 addr, last; - - addr = round_up(ei_start, align); - if (addr < start) - addr = round_up(start, align); - if (addr >= ei_last) - goto out; - while (bad_addr(&addr, size, align) && addr+size <= ei_last) - ; - last = addr + size; - if (last > ei_last) - goto out; - if (last > end) - goto out; - - return addr; - -out: - return -1ULL; -} - -u64 __init find_early_area_size(u64 ei_start, u64 ei_last, u64 start, - u64 *sizep, u64 align) -{ - u64 addr, last; - - addr = round_up(ei_start, align); - if (addr < start) - addr = round_up(start, align); - if (addr >= ei_last) - goto out; - *sizep = ei_last - addr; - while (bad_addr_size(&addr, sizep, align) && addr + *sizep <= ei_last) - ; - last = addr + *sizep; - if (last > ei_last) - goto out; - - return addr; - -out: - return -1ULL; -} -- cgit v1.2.3 From fa66f07aa1f0950e1dc78b7ab39728b3f8aa77a1 Mon Sep 17 00:00:00 2001 From: Stephane Eranian Date: Thu, 26 Aug 2010 16:40:01 +0200 Subject: perf_events: Fix time tracking for events with pid != -1 and cpu != -1 Per-thread events with a cpu filter, i.e., cpu != -1, were not reporting correct timings when the thread never ran on the monitored cpu. The time enabled was reported as a negative value. This patch fixes the problem by updating tstamp_stopped, tstamp_running in event_sched_out() for events with filters and which are marked as INACTIVE. The function group_sched_out() is modified to systematically call into event_sched_out() to avoid duplicating the timing adjustment code twice. With the patch, I now get: $ task_cpu -i -e unhalted_core_cycles,unhalted_core_cycles noploop 2 noploop for 2 seconds CPU0 0 unhalted_core_cycles (ena=1,991,136,594, run=0) CPU0 0 unhalted_core_cycles (ena=1,991,136,594, run=0) CPU1 0 unhalted_core_cycles (ena=1,991,136,594, run=0) CPU1 0 unhalted_core_cycles (ena=1,991,136,594, run=0) CPU2 0 unhalted_core_cycles (ena=1,991,136,594, run=0) CPU2 0 unhalted_core_cycles (ena=1,991,136,594, run=0) CPU3 4,747,990,931 unhalted_core_cycles (ena=1,991,136,594, run=1,991,136,594) CPU3 4,747,990,931 unhalted_core_cycles (ena=1,991,136,594, run=1,991,136,594) Signed-off-by: Stephane Eranian Acked-by: Peter Zijlstra Cc: paulus@samba.org Cc: davem@davemloft.net Cc: fweisbec@gmail.com Cc: perfmon2-devel@lists.sf.net Cc: eranian@google.com LKML-Reference: <4c76802d.aae9d80a.115d.70fe@mx.google.com> Signed-off-by: Ingo Molnar --- kernel/perf_event.c | 26 ++++++++++++++++++++++---- 1 file changed, 22 insertions(+), 4 deletions(-) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index 403d1804b198..657555a5f30f 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -402,11 +402,31 @@ static void perf_group_detach(struct perf_event *event) } } +static inline int +event_filter_match(struct perf_event *event) +{ + return event->cpu == -1 || event->cpu == smp_processor_id(); +} + static void event_sched_out(struct perf_event *event, struct perf_cpu_context *cpuctx, struct perf_event_context *ctx) { + u64 delta; + /* + * An event which could not be activated because of + * filter mismatch still needs to have its timings + * maintained, otherwise bogus information is return + * via read() for time_enabled, time_running: + */ + if (event->state == PERF_EVENT_STATE_INACTIVE + && !event_filter_match(event)) { + delta = ctx->time - event->tstamp_stopped; + event->tstamp_running += delta; + event->tstamp_stopped = ctx->time; + } + if (event->state != PERF_EVENT_STATE_ACTIVE) return; @@ -432,9 +452,7 @@ group_sched_out(struct perf_event *group_event, struct perf_event_context *ctx) { struct perf_event *event; - - if (group_event->state != PERF_EVENT_STATE_ACTIVE) - return; + int state = group_event->state; event_sched_out(group_event, cpuctx, ctx); @@ -444,7 +462,7 @@ group_sched_out(struct perf_event *group_event, list_for_each_entry(event, &group_event->sibling_list, group_entry) event_sched_out(event, cpuctx, ctx); - if (group_event->attr.exclusive) + if (state == PERF_EVENT_STATE_ACTIVE && group_event->attr.exclusive) cpuctx->exclusive = 0; } -- cgit v1.2.3 From 477a3c33d1efa0342a74bd02da2e049191993e2c Mon Sep 17 00:00:00 2001 From: Tejun Heo Date: Tue, 31 Aug 2010 10:54:35 +0200 Subject: workqueue: fix GCWQ_DISASSOCIATED initialization init_workqueues() incorrectly marks workqueues for all possible CPUs associated. Combined with mayday_mask initialization bug, this can make rescuers keep trying to bind to an offline gcwq indefinitely. Fix init_workqueues() such that only online CPUs have their gcwqs have GCWQ_DISASSOCIATED cleared. Signed-off-by: Tejun Heo Reported-by: CAI Qian --- kernel/workqueue.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/workqueue.c b/kernel/workqueue.c index a2dccfca03ba..c8183b235d16 100644 --- a/kernel/workqueue.c +++ b/kernel/workqueue.c @@ -3558,8 +3558,7 @@ static int __init init_workqueues(void) spin_lock_init(&gcwq->lock); INIT_LIST_HEAD(&gcwq->worklist); gcwq->cpu = cpu; - if (cpu == WORK_CPU_UNBOUND) - gcwq->flags |= GCWQ_DISASSOCIATED; + gcwq->flags |= GCWQ_DISASSOCIATED; INIT_LIST_HEAD(&gcwq->idle_list); for (i = 0; i < BUSY_WORKER_HASH_SIZE; i++) @@ -3583,6 +3582,8 @@ static int __init init_workqueues(void) struct global_cwq *gcwq = get_gcwq(cpu); struct worker *worker; + if (cpu != WORK_CPU_UNBOUND) + gcwq->flags &= ~GCWQ_DISASSOCIATED; worker = create_worker(gcwq, true); BUG_ON(!worker); spin_lock_irq(&gcwq->lock); -- cgit v1.2.3 From 9c37547ab62f88aac3e1e3c2065b611f811de9b5 Mon Sep 17 00:00:00 2001 From: Tejun Heo Date: Tue, 31 Aug 2010 11:18:34 +0200 Subject: workqueue: use zalloc_cpumask_var() for gcwq->mayday_mask alloc_mayday_mask() was using alloc_cpumask_var() making gcwq->mayday_mask contain garbage after initialization on CONFIG_CPUMASK_OFFSTACK=y configurations. This combined with the previously fixed GCWQ_DISASSOCIATED initialization bug could make rescuers fall into infinite loop trying to bind to an offline cpu. Signed-off-by: Tejun Heo Reported-by: CAI Qian --- kernel/workqueue.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/workqueue.c b/kernel/workqueue.c index c8183b235d16..785542976b00 100644 --- a/kernel/workqueue.c +++ b/kernel/workqueue.c @@ -196,7 +196,7 @@ typedef cpumask_var_t mayday_mask_t; cpumask_test_and_set_cpu((cpu), (mask)) #define mayday_clear_cpu(cpu, mask) cpumask_clear_cpu((cpu), (mask)) #define for_each_mayday_cpu(cpu, mask) for_each_cpu((cpu), (mask)) -#define alloc_mayday_mask(maskp, gfp) alloc_cpumask_var((maskp), (gfp)) +#define alloc_mayday_mask(maskp, gfp) zalloc_cpumask_var((maskp), (gfp)) #define free_mayday_mask(mask) free_cpumask_var((mask)) #else typedef unsigned long mayday_mask_t; -- cgit v1.2.3 From 3aaba20f26f58843e8f20611e5c0b1c06954310f Mon Sep 17 00:00:00 2001 From: Li Zefan Date: Mon, 23 Aug 2010 16:50:12 +0800 Subject: tracing: Fix a race in function profile While we are reading trace_stat/functionX and someone just disabled function_profile at that time, we can trigger this: divide error: 0000 [#1] PREEMPT SMP ... EIP is at function_stat_show+0x90/0x230 ... This fix just takes the ftrace_profile_lock and checks if rec->counter is 0. If it's 0, we know the profile buffer has been reset. Signed-off-by: Li Zefan Cc: stable@kernel.org LKML-Reference: <4C723644.4040708@cn.fujitsu.com> Signed-off-by: Steven Rostedt --- kernel/trace/ftrace.c | 15 +++++++++++---- 1 file changed, 11 insertions(+), 4 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index 0d88ce9b9fb8..7cb1f45a1de1 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -381,12 +381,19 @@ static int function_stat_show(struct seq_file *m, void *v) { struct ftrace_profile *rec = v; char str[KSYM_SYMBOL_LEN]; + int ret = 0; #ifdef CONFIG_FUNCTION_GRAPH_TRACER - static DEFINE_MUTEX(mutex); static struct trace_seq s; unsigned long long avg; unsigned long long stddev; #endif + mutex_lock(&ftrace_profile_lock); + + /* we raced with function_profile_reset() */ + if (unlikely(rec->counter == 0)) { + ret = -EBUSY; + goto out; + } kallsyms_lookup(rec->ip, NULL, NULL, NULL, str); seq_printf(m, " %-30.30s %10lu", str, rec->counter); @@ -408,7 +415,6 @@ static int function_stat_show(struct seq_file *m, void *v) do_div(stddev, (rec->counter - 1) * 1000); } - mutex_lock(&mutex); trace_seq_init(&s); trace_print_graph_duration(rec->time, &s); trace_seq_puts(&s, " "); @@ -416,11 +422,12 @@ static int function_stat_show(struct seq_file *m, void *v) trace_seq_puts(&s, " "); trace_print_graph_duration(stddev, &s); trace_print_seq(m, &s); - mutex_unlock(&mutex); #endif seq_putc(m, '\n'); +out: + mutex_unlock(&ftrace_profile_lock); - return 0; + return ret; } static void ftrace_profile_reset(struct ftrace_profile_stat *stat) -- cgit v1.2.3 From 950eaaca681c44aab87a46225c9e44f902c080aa Mon Sep 17 00:00:00 2001 From: "Paul E. McKenney" Date: Tue, 31 Aug 2010 17:00:18 -0700 Subject: pid: make setpgid() system call use RCU read-side critical section [ 23.584719] [ 23.584720] =================================================== [ 23.585059] [ INFO: suspicious rcu_dereference_check() usage. ] [ 23.585176] --------------------------------------------------- [ 23.585176] kernel/pid.c:419 invoked rcu_dereference_check() without protection! [ 23.585176] [ 23.585176] other info that might help us debug this: [ 23.585176] [ 23.585176] [ 23.585176] rcu_scheduler_active = 1, debug_locks = 1 [ 23.585176] 1 lock held by rc.sysinit/728: [ 23.585176] #0: (tasklist_lock){.+.+..}, at: [] sys_setpgid+0x5f/0x193 [ 23.585176] [ 23.585176] stack backtrace: [ 23.585176] Pid: 728, comm: rc.sysinit Not tainted 2.6.36-rc2 #2 [ 23.585176] Call Trace: [ 23.585176] [] lockdep_rcu_dereference+0x99/0xa2 [ 23.585176] [] find_task_by_pid_ns+0x50/0x6a [ 23.585176] [] find_task_by_vpid+0x1d/0x1f [ 23.585176] [] sys_setpgid+0x67/0x193 [ 23.585176] [] system_call_fastpath+0x16/0x1b [ 24.959669] type=1400 audit(1282938522.956:4): avc: denied { module_request } for pid=766 comm="hwclock" kmod="char-major-10-135" scontext=system_u:system_r:hwclock_t:s0 tcontext=system_u:system_r:kernel_t:s0 tclas It turns out that the setpgid() system call fails to enter an RCU read-side critical section before doing a PID-to-task_struct translation. This commit therefore does rcu_read_lock() before the translation, and also does rcu_read_unlock() after the last use of the returned pointer. Reported-by: Andrew Morton Signed-off-by: Paul E. McKenney Acked-by: David Howells --- kernel/sys.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'kernel') diff --git a/kernel/sys.c b/kernel/sys.c index e9ad44489828..7f5a0cd296a9 100644 --- a/kernel/sys.c +++ b/kernel/sys.c @@ -931,6 +931,7 @@ SYSCALL_DEFINE2(setpgid, pid_t, pid, pid_t, pgid) pgid = pid; if (pgid < 0) return -EINVAL; + rcu_read_lock(); /* From this point forward we keep holding onto the tasklist lock * so that our parent does not change from under us. -DaveM @@ -984,6 +985,7 @@ SYSCALL_DEFINE2(setpgid, pid_t, pid, pid_t, pgid) out: /* All paths lead to here, thus we are safe. -DaveM */ write_unlock_irq(&tasklist_lock); + rcu_read_unlock(); return err; } -- cgit v1.2.3 From eac243355a99d6b9d41bbeba4fc83e7f735485f9 Mon Sep 17 00:00:00 2001 From: Akinobu Mita Date: Tue, 31 Aug 2010 23:00:08 -0400 Subject: lockup_detector: Convert cpu notifier to return encapsulate errno value By the commit e6bde73b07edeb703d4c89c1daabc09c303de11f ("cpu-hotplug: return better errno on cpu hotplug failure"), the cpu notifier can return encapsulate errno value, resulting in more meaningful error codes for CPU hotplug failures. This converts the cpu notifier to return encapsulate errno value for the lockup_detector as well. Signed-off-by: Akinobu Mita Cc: peterz@infradead.org Cc: gorcunov@gmail.com Cc: fweisbec@gmail.com LKML-Reference: <1283310009-22168-3-git-send-email-dzickus@redhat.com> Signed-off-by: Don Zickus Signed-off-by: Ingo Molnar --- kernel/watchdog.c | 21 +++++++++++---------- 1 file changed, 11 insertions(+), 10 deletions(-) (limited to 'kernel') diff --git a/kernel/watchdog.c b/kernel/watchdog.c index 0d53c8e853b1..501cb6edca4b 100644 --- a/kernel/watchdog.c +++ b/kernel/watchdog.c @@ -371,7 +371,7 @@ static int watchdog_nmi_enable(int cpu) } printk(KERN_ERR "NMI watchdog failed to create perf event on cpu%i: %p\n", cpu, event); - return -1; + return PTR_ERR(event); /* success path */ out_save: @@ -415,17 +415,19 @@ static int watchdog_prepare_cpu(int cpu) static int watchdog_enable(int cpu) { struct task_struct *p = per_cpu(softlockup_watchdog, cpu); + int err; /* enable the perf event */ - if (watchdog_nmi_enable(cpu) != 0) - return -1; + err = watchdog_nmi_enable(cpu); + if (err) + return err; /* create the watchdog thread */ if (!p) { p = kthread_create(watchdog, (void *)(unsigned long)cpu, "watchdog/%d", cpu); if (IS_ERR(p)) { printk(KERN_ERR "softlockup watchdog for %i failed\n", cpu); - return -1; + return PTR_ERR(p); } kthread_bind(p, cpu); per_cpu(watchdog_touch_ts, cpu) = 0; @@ -519,17 +521,16 @@ static int __cpuinit cpu_callback(struct notifier_block *nfb, unsigned long action, void *hcpu) { int hotcpu = (unsigned long)hcpu; + int err = 0; switch (action) { case CPU_UP_PREPARE: case CPU_UP_PREPARE_FROZEN: - if (watchdog_prepare_cpu(hotcpu)) - return NOTIFY_BAD; + err = watchdog_prepare_cpu(hotcpu); break; case CPU_ONLINE: case CPU_ONLINE_FROZEN: - if (watchdog_enable(hotcpu)) - return NOTIFY_BAD; + err = watchdog_enable(hotcpu); break; #ifdef CONFIG_HOTPLUG_CPU case CPU_UP_CANCELED: @@ -542,7 +543,7 @@ cpu_callback(struct notifier_block *nfb, unsigned long action, void *hcpu) break; #endif /* CONFIG_HOTPLUG_CPU */ } - return NOTIFY_OK; + return notifier_from_errno(err); } static struct notifier_block __cpuinitdata cpu_nfb = { @@ -558,7 +559,7 @@ static int __init spawn_watchdog_task(void) return 0; err = cpu_callback(&cpu_nfb, CPU_UP_PREPARE, cpu); - WARN_ON(err == NOTIFY_BAD); + WARN_ON(notifier_to_errno(err)); cpu_callback(&cpu_nfb, CPU_ONLINE, cpu); register_cpu_notifier(&cpu_nfb); -- cgit v1.2.3 From 14416c35b6b9975c9593d7ecc8382d1ecaa0b598 Mon Sep 17 00:00:00 2001 From: Akinobu Mita Date: Tue, 31 Aug 2010 23:00:09 -0400 Subject: lockup_detector: Remove unused panic_notifier The panic notifer in lockup_detector just set did_panic to 1. But did_panic is not used anywhere so we can just remove it. Signed-off-by: Akinobu Mita Cc: peterz@infradead.org Cc: gorcunov@gmail.com Cc: fweisbec@gmail.com LKML-Reference: <1283310009-22168-4-git-send-email-dzickus@redhat.com> Signed-off-by: Don Zickus Signed-off-by: Ingo Molnar --- kernel/watchdog.c | 15 --------------- 1 file changed, 15 deletions(-) (limited to 'kernel') diff --git a/kernel/watchdog.c b/kernel/watchdog.c index 501cb6edca4b..5b1ee4f4ca0d 100644 --- a/kernel/watchdog.c +++ b/kernel/watchdog.c @@ -43,7 +43,6 @@ static DEFINE_PER_CPU(unsigned long, hrtimer_interrupts_saved); static DEFINE_PER_CPU(struct perf_event *, watchdog_ev); #endif -static int __read_mostly did_panic; static int __initdata no_watchdog; @@ -180,18 +179,6 @@ static int is_softlockup(unsigned long touch_ts) return 0; } -static int -watchdog_panic(struct notifier_block *this, unsigned long event, void *ptr) -{ - did_panic = 1; - - return NOTIFY_DONE; -} - -static struct notifier_block panic_block = { - .notifier_call = watchdog_panic, -}; - #ifdef CONFIG_HARDLOCKUP_DETECTOR static struct perf_event_attr wd_hw_attr = { .type = PERF_TYPE_HARDWARE, @@ -564,8 +551,6 @@ static int __init spawn_watchdog_task(void) cpu_callback(&cpu_nfb, CPU_ONLINE, cpu); register_cpu_notifier(&cpu_nfb); - atomic_notifier_chain_register(&panic_notifier_list, &panic_block); - return 0; } early_initcall(spawn_watchdog_task); -- cgit v1.2.3 From 68d3f1d810500e8b975bdf0b20dd83d060076b4b Mon Sep 17 00:00:00 2001 From: Don Zickus Date: Tue, 31 Aug 2010 23:00:07 -0400 Subject: lockup_detector: Sync touch_*_watchdog back to old semantics During my rewrite, the semantics of touch_nmi_watchdog and touch_softlockup_watchdog changed enough to break some drivers (mostly over preemptable regions). These are cases where long delays on one CPU (due to print_delay for example) can cause long delays on other CPUs - so we must 'touch' the nmi_watchdog flag of those other CPUs as well. This change brings those touch_*_watchdog() functions back in line with to how they used to work. Signed-off-by: Don Zickus Acked-by: Cyrill Gorcunov Cc: peterz@infradead.org Cc: fweisbec@gmail.com LKML-Reference: <1283310009-22168-2-git-send-email-dzickus@redhat.com> Signed-off-by: Ingo Molnar --- kernel/watchdog.c | 17 ++++++++++++----- 1 file changed, 12 insertions(+), 5 deletions(-) (limited to 'kernel') diff --git a/kernel/watchdog.c b/kernel/watchdog.c index 0d53c8e853b1..7f9c3c52ecc1 100644 --- a/kernel/watchdog.c +++ b/kernel/watchdog.c @@ -122,7 +122,7 @@ static void __touch_watchdog(void) void touch_softlockup_watchdog(void) { - __get_cpu_var(watchdog_touch_ts) = 0; + __raw_get_cpu_var(watchdog_touch_ts) = 0; } EXPORT_SYMBOL(touch_softlockup_watchdog); @@ -142,7 +142,14 @@ void touch_all_softlockup_watchdogs(void) #ifdef CONFIG_HARDLOCKUP_DETECTOR void touch_nmi_watchdog(void) { - __get_cpu_var(watchdog_nmi_touch) = true; + if (watchdog_enabled) { + unsigned cpu; + + for_each_present_cpu(cpu) { + if (per_cpu(watchdog_nmi_touch, cpu) != true) + per_cpu(watchdog_nmi_touch, cpu) = true; + } + } touch_softlockup_watchdog(); } EXPORT_SYMBOL(touch_nmi_watchdog); @@ -433,6 +440,9 @@ static int watchdog_enable(int cpu) wake_up_process(p); } + /* if any cpu succeeds, watchdog is considered enabled for the system */ + watchdog_enabled = 1; + return 0; } @@ -455,9 +465,6 @@ static void watchdog_disable(int cpu) per_cpu(softlockup_watchdog, cpu) = NULL; kthread_stop(p); } - - /* if any cpu succeeds, watchdog is considered enabled for the system */ - watchdog_enabled = 1; } static void watchdog_enable_all_cpus(void) -- cgit v1.2.3 From f6195aa09e618d712f52bf4fa33b5293820eb93d Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Wed, 1 Sep 2010 12:23:12 -0400 Subject: ring-buffer: Place duplicate expression into a single function While discussing the strictness of the 80 character limit on the Kernel Summit Discussion mailing list, I showed examples that I broke that limit slightly with some algorithms. In discussing with John Linville, what looked better, I realized that two of the 80 char breaking culprits were an identical expression. As a clean up, this patch moves the identical expression into its own helper function and that is used instead. As a side effect, the offending code is now under the 80 character limit. :-) This clean up code also changes the expression from (A - B) - C to A - (B + C) This makes the code look a little nicer too. Cc: John W. Linville Signed-off-by: Steven Rostedt --- kernel/trace/ring_buffer.c | 21 +++++++++++++++------ 1 file changed, 15 insertions(+), 6 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c index 19cccc3c3028..ef27017caa56 100644 --- a/kernel/trace/ring_buffer.c +++ b/kernel/trace/ring_buffer.c @@ -2606,6 +2606,19 @@ void ring_buffer_record_enable_cpu(struct ring_buffer *buffer, int cpu) } EXPORT_SYMBOL_GPL(ring_buffer_record_enable_cpu); +/* + * The total entries in the ring buffer is the running counter + * of entries entered into the ring buffer, minus the sum of + * the entries read from the ring buffer and the number of + * entries that were overwritten. + */ +static inline unsigned long +rb_num_of_entries(struct ring_buffer_per_cpu *cpu_buffer) +{ + return local_read(&cpu_buffer->entries) - + (local_read(&cpu_buffer->overrun) + cpu_buffer->read); +} + /** * ring_buffer_entries_cpu - get the number of entries in a cpu buffer * @buffer: The ring buffer @@ -2614,16 +2627,13 @@ EXPORT_SYMBOL_GPL(ring_buffer_record_enable_cpu); unsigned long ring_buffer_entries_cpu(struct ring_buffer *buffer, int cpu) { struct ring_buffer_per_cpu *cpu_buffer; - unsigned long ret; if (!cpumask_test_cpu(cpu, buffer->cpumask)) return 0; cpu_buffer = buffer->buffers[cpu]; - ret = (local_read(&cpu_buffer->entries) - local_read(&cpu_buffer->overrun)) - - cpu_buffer->read; - return ret; + return rb_num_of_entries(cpu_buffer); } EXPORT_SYMBOL_GPL(ring_buffer_entries_cpu); @@ -2684,8 +2694,7 @@ unsigned long ring_buffer_entries(struct ring_buffer *buffer) /* if you care about this being correct, lock the buffer */ for_each_buffer_cpu(buffer, cpu) { cpu_buffer = buffer->buffers[cpu]; - entries += (local_read(&cpu_buffer->entries) - - local_read(&cpu_buffer->overrun)) - cpu_buffer->read; + entries += rb_num_of_entries(cpu_buffer); } return entries; -- cgit v1.2.3 From 09bfafac3e237415cc4b6adde49f9f28b3a42659 Mon Sep 17 00:00:00 2001 From: Rabin Vincent Date: Tue, 10 Aug 2010 19:32:37 +0100 Subject: ARM: 6314/1: ftrace: allow build without frame pointers on ARM With a new enough GCC, ARM function tracing can be supported without the need for frame pointers. This is essential for Thumb-2 support, since frame pointers aren't available then. Acked-by: Catalin Marinas Acked-by: Steven Rostedt Signed-off-by: Rabin Vincent Signed-off-by: Russell King --- kernel/trace/Kconfig | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/trace/Kconfig b/kernel/trace/Kconfig index 538501c6ea50..6329d063b5e4 100644 --- a/kernel/trace/Kconfig +++ b/kernel/trace/Kconfig @@ -121,7 +121,7 @@ if FTRACE config FUNCTION_TRACER bool "Kernel Function Tracer" depends on HAVE_FUNCTION_TRACER - select FRAME_POINTER + select FRAME_POINTER if (!ARM_UNWIND) select KALLSYMS select GENERIC_TRACER select CONTEXT_SWITCH_TRACER -- cgit v1.2.3 From 81a294c44e973dc7182e4733421b7cb2daba3c29 Mon Sep 17 00:00:00 2001 From: "Paul E. McKenney" Date: Mon, 30 Aug 2010 09:52:50 -0700 Subject: rcu: fix _oddness handling of verbose stall warnings CONFIG_RCU_CPU_STALL_VERBOSE depends on CONFIG_TREE_PREEMPT_RCU, but rcu_bootup_announce_oddness() complains if CONFIG_RCU_CPU_STALL_VERBOSE is not set even in the case of CONFIG_TREE_RCU. This commit therefore fixes rcu_bootup_announce_oddness() to avoid insisting on impossibilities. Reported-by: Guy Martin Signed-off-by: Paul E. McKenney --- kernel/rcutree_plugin.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/rcutree_plugin.h b/kernel/rcutree_plugin.h index e9e0bc74ff37..71a4147473f9 100644 --- a/kernel/rcutree_plugin.h +++ b/kernel/rcutree_plugin.h @@ -57,7 +57,7 @@ static void __init rcu_bootup_announce_oddness(void) printk(KERN_INFO "\tRCU-based detection of stalled CPUs is disabled.\n"); #endif -#ifndef CONFIG_RCU_CPU_STALL_VERBOSE +#if defined(CONFIG_TREE_PREEMPT_RCU) && !defined(CONFIG_RCU_CPU_STALL_VERBOSE) printk(KERN_INFO "\tVerbose stalled-CPUs detection is disabled.\n"); #endif #if NUM_RCU_LVL_4 != 0 -- cgit v1.2.3 From ef5dc121d5a0bb1fa477c5395277259f07d318a3 Mon Sep 17 00:00:00 2001 From: Randy Dunlap Date: Thu, 2 Sep 2010 15:48:16 -0700 Subject: mutex: Fix annotations to include it in kernel-locking docbook Fix kernel-doc notation in linux/mutex.h and kernel/mutex.c, then add these 2 files to the kernel-locking docbook as the Mutex API reference chapter. Add one API function to mutex-design.txt and correct a typo in that file. Signed-off-by: Randy Dunlap Cc: Rusty Russell LKML-Reference: <20100902154816.6cc2f9ad.randy.dunlap@oracle.com> Signed-off-by: Ingo Molnar --- kernel/mutex.c | 23 +++++++---------------- 1 file changed, 7 insertions(+), 16 deletions(-) (limited to 'kernel') diff --git a/kernel/mutex.c b/kernel/mutex.c index 4c0b7b3e6d2e..200407c1502f 100644 --- a/kernel/mutex.c +++ b/kernel/mutex.c @@ -36,15 +36,6 @@ # include #endif -/*** - * mutex_init - initialize the mutex - * @lock: the mutex to be initialized - * @key: the lock_class_key for the class; used by mutex lock debugging - * - * Initialize the mutex to unlocked state. - * - * It is not allowed to initialize an already locked mutex. - */ void __mutex_init(struct mutex *lock, const char *name, struct lock_class_key *key) { @@ -68,7 +59,7 @@ EXPORT_SYMBOL(__mutex_init); static __used noinline void __sched __mutex_lock_slowpath(atomic_t *lock_count); -/*** +/** * mutex_lock - acquire the mutex * @lock: the mutex to be acquired * @@ -105,7 +96,7 @@ EXPORT_SYMBOL(mutex_lock); static __used noinline void __sched __mutex_unlock_slowpath(atomic_t *lock_count); -/*** +/** * mutex_unlock - release the mutex * @lock: the mutex to be released * @@ -364,8 +355,8 @@ __mutex_lock_killable_slowpath(atomic_t *lock_count); static noinline int __sched __mutex_lock_interruptible_slowpath(atomic_t *lock_count); -/*** - * mutex_lock_interruptible - acquire the mutex, interruptable +/** + * mutex_lock_interruptible - acquire the mutex, interruptible * @lock: the mutex to be acquired * * Lock the mutex like mutex_lock(), and return 0 if the mutex has @@ -456,15 +447,15 @@ static inline int __mutex_trylock_slowpath(atomic_t *lock_count) return prev == 1; } -/*** - * mutex_trylock - try acquire the mutex, without waiting +/** + * mutex_trylock - try to acquire the mutex, without waiting * @lock: the mutex to be acquired * * Try to acquire the mutex atomically. Returns 1 if the mutex * has been acquired successfully, and 0 on contention. * * NOTE: this function follows the spin_trylock() convention, so - * it is negated to the down_trylock() return values! Be careful + * it is negated from the down_trylock() return values! Be careful * about this when converting semaphore users to mutexes. * * This function must not be used in interrupt context. The -- cgit v1.2.3 From b3bd3de66f60df4c9a2076e2886a622458929056 Mon Sep 17 00:00:00 2001 From: Andi Kleen Date: Tue, 10 Aug 2010 14:17:51 -0700 Subject: gcc-4.6: kernel/*: Fix unused but set warnings No real bugs I believe, just some dead code. Signed-off-by: Andi Kleen Cc: Peter Zijlstra Cc: andi@firstfloor.org Signed-off-by: Andrew Morton Signed-off-by: Ingo Molnar --- kernel/debug/kdb/kdb_bp.c | 2 -- kernel/hrtimer.c | 3 +-- kernel/sched_fair.c | 3 +-- kernel/sysctl.c | 5 +---- kernel/trace/ring_buffer.c | 2 -- 5 files changed, 3 insertions(+), 12 deletions(-) (limited to 'kernel') diff --git a/kernel/debug/kdb/kdb_bp.c b/kernel/debug/kdb/kdb_bp.c index 75bd9b3ebbb7..20059ef4459a 100644 --- a/kernel/debug/kdb/kdb_bp.c +++ b/kernel/debug/kdb/kdb_bp.c @@ -274,7 +274,6 @@ static int kdb_bp(int argc, const char **argv) int i, bpno; kdb_bp_t *bp, *bp_check; int diag; - int free; char *symname = NULL; long offset = 0ul; int nextarg; @@ -305,7 +304,6 @@ static int kdb_bp(int argc, const char **argv) /* * Find an empty bp structure to allocate */ - free = KDB_MAXBPT; for (bpno = 0, bp = kdb_breakpoints; bpno < KDB_MAXBPT; bpno++, bp++) { if (bp->bp_free) break; diff --git a/kernel/hrtimer.c b/kernel/hrtimer.c index ce669174f355..1decafbb6b1a 100644 --- a/kernel/hrtimer.c +++ b/kernel/hrtimer.c @@ -1091,11 +1091,10 @@ EXPORT_SYMBOL_GPL(hrtimer_cancel); */ ktime_t hrtimer_get_remaining(const struct hrtimer *timer) { - struct hrtimer_clock_base *base; unsigned long flags; ktime_t rem; - base = lock_hrtimer_base(timer, &flags); + lock_hrtimer_base(timer, &flags); rem = hrtimer_expires_remaining(timer); unlock_hrtimer_base(timer, &flags); diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c index ab661ebc4895..134f7edb30c6 100644 --- a/kernel/sched_fair.c +++ b/kernel/sched_fair.c @@ -1313,7 +1313,7 @@ static struct sched_group * find_idlest_group(struct sched_domain *sd, struct task_struct *p, int this_cpu, int load_idx) { - struct sched_group *idlest = NULL, *this = NULL, *group = sd->groups; + struct sched_group *idlest = NULL, *group = sd->groups; unsigned long min_load = ULONG_MAX, this_load = 0; int imbalance = 100 + (sd->imbalance_pct-100)/2; @@ -1348,7 +1348,6 @@ find_idlest_group(struct sched_domain *sd, struct task_struct *p, if (local_group) { this_load = avg_load; - this = group; } else if (avg_load < min_load) { min_load = avg_load; idlest = group; diff --git a/kernel/sysctl.c b/kernel/sysctl.c index ca38e8e3e907..f88552c6d227 100644 --- a/kernel/sysctl.c +++ b/kernel/sysctl.c @@ -1713,10 +1713,7 @@ static __init int sysctl_init(void) { sysctl_set_parent(NULL, root_table); #ifdef CONFIG_SYSCTL_SYSCALL_CHECK - { - int err; - err = sysctl_check_table(current->nsproxy, root_table); - } + sysctl_check_table(current->nsproxy, root_table); #endif return 0; } diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c index 19cccc3c3028..492197e2f86c 100644 --- a/kernel/trace/ring_buffer.c +++ b/kernel/trace/ring_buffer.c @@ -2985,13 +2985,11 @@ static void rb_advance_reader(struct ring_buffer_per_cpu *cpu_buffer) static void rb_advance_iter(struct ring_buffer_iter *iter) { - struct ring_buffer *buffer; struct ring_buffer_per_cpu *cpu_buffer; struct ring_buffer_event *event; unsigned length; cpu_buffer = iter->cpu_buffer; - buffer = cpu_buffer->buffer; /* * Check if we are at the end of the buffer. -- cgit v1.2.3 From ed2d372c0738386b8a184a6a6bea9c16df6ffb68 Mon Sep 17 00:00:00 2001 From: Christian Dietrich Date: Mon, 6 Sep 2010 16:37:05 +0200 Subject: sched: Remove unnecessary #ifdef CONFIG_SMP The CONFIG_SMP ifdef isn't necessary at this point, because it is checked in an outer ifdef level already and has no effect here. Cleanup only, no functional effect. Signed-off-by: Christian Dietrich Cc: vamos-dev@i4.informatik.uni-erlangen.de Cc: Peter Zijlstra Cc: Mike Galbraith Cc: Tejun Heo LKML-Reference: <7a3a39ef3f765a4473cb026b1f204059568a7098.1283782701.git.qy03fugy@stud.informatik.uni-erlangen.de> Signed-off-by: Ingo Molnar --- kernel/sched.c | 4 +--- 1 file changed, 1 insertion(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index 09b574e7f4df..8eef8e5512d4 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -426,9 +426,7 @@ struct root_domain { */ cpumask_var_t rto_mask; atomic_t rto_count; -#ifdef CONFIG_SMP struct cpupri cpupri; -#endif }; /* @@ -437,7 +435,7 @@ struct root_domain { */ static struct root_domain def_root_domain; -#endif +#endif /* CONFIG_SMP */ /* * This is the main, per-CPU runqueue data structure. -- cgit v1.2.3 From 9c55cb12c1c172e2d51e85fbb5a4796ca86b77e7 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Wed, 8 Sep 2010 11:20:37 -0400 Subject: tracing: Do not allow llseek to set_ftrace_filter Reading the file set_ftrace_filter does three things. 1) shows whether or not filters are set for the function tracer 2) shows what functions are set for the function tracer 3) shows what triggers are set on any functions 3 is independent from 1 and 2. The way this file currently works is that it is a state machine, and as you read it, it may change state. But this assumption breaks when you use lseek() on the file. The state machine gets out of sync and the t_show() may use the wrong pointer and cause a kernel oops. Luckily, this will only kill the app that does the lseek, but the app dies while holding a mutex. This prevents anyone else from using the set_ftrace_filter file (or any other function tracing file for that matter). A real fix for this is to rewrite the code, but that is too much for a -rc release or stable. This patch simply disables llseek on the set_ftrace_filter() file for now, and we can do the proper fix for the next major release. Reported-by: Robert Swiecki Cc: Chris Wright Cc: Tavis Ormandy Cc: Eugene Teo Cc: vendor-sec@lst.de Cc: Signed-off-by: Steven Rostedt --- kernel/trace/ftrace.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index 7cb1f45a1de1..83a16e9ee518 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -2416,7 +2416,7 @@ static const struct file_operations ftrace_filter_fops = { .open = ftrace_filter_open, .read = seq_read, .write = ftrace_filter_write, - .llseek = ftrace_regex_lseek, + .llseek = no_llseek, .release = ftrace_filter_release, }; -- cgit v1.2.3 From 61a527362234ac3352a91ac67c50c6f7cd248eb1 Mon Sep 17 00:00:00 2001 From: Masami Hiramatsu Date: Fri, 27 Aug 2010 20:38:46 +0900 Subject: tracing/kprobe: Fix a memory leak in error case Fix a memory leak which happens when a field name conflicts with others. In error case, free_trace_probe() will free all arguments until nr_args, so this increments nr_args the begining of the loop instead of the end. Cc: Steven Rostedt Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mathieu Desnoyers LKML-Reference: <20100827113846.22882.12670.stgit@ltc236.sdl.hitachi.co.jp> Signed-off-by: Masami Hiramatsu Signed-off-by: Arnaldo Carvalho de Melo --- kernel/trace/trace_kprobe.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace_kprobe.c b/kernel/trace/trace_kprobe.c index 8b27c9849b42..0116c038b0bc 100644 --- a/kernel/trace/trace_kprobe.c +++ b/kernel/trace/trace_kprobe.c @@ -992,6 +992,9 @@ static int create_trace_probe(int argc, char **argv) /* parse arguments */ ret = 0; for (i = 0; i < argc && i < MAX_TRACE_ARGS; i++) { + /* Increment count for freeing args in error case */ + tp->nr_args++; + /* Parse argument name */ arg = strchr(argv[i], '='); if (arg) @@ -1021,11 +1024,8 @@ static int create_trace_probe(int argc, char **argv) ret = parse_probe_arg(arg, tp, &tp->args[i], is_return); if (ret) { pr_info("Parse error at argument%d. (%d)\n", i, ret); - kfree(tp->args[i].name); goto error; } - - tp->nr_args++; } ret = register_trace_probe(tp); -- cgit v1.2.3 From aba91595cfcebd193425e20aabc407531526a1c5 Mon Sep 17 00:00:00 2001 From: Masami Hiramatsu Date: Fri, 27 Aug 2010 20:39:06 +0900 Subject: tracing/kprobes: Fix handling of argument names Set "argN" name for each argument automatically if it has no specified name. Since dynamic trace event(kprobe_events) accepts special characters for its argument, its format can show those special characters (e.g. '$', '%', '+'). However, perf can't parse those format because of the character (especially '%') mess up the format. This sets "argX" name for those arguments if user omitted the argument names. E.g. # echo 'p do_fork %ax IP=%ip $stack' > tracing/kprobe_events # cat tracing/kprobe_events p:kprobes/p_do_fork_0 do_fork arg1=%ax IP=%ip arg3=$stack Reported-by: Srikar Dronamraju Cc: Steven Rostedt Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mathieu Desnoyers LKML-Reference: <20100827113906.22882.59312.stgit@ltc236.sdl.hitachi.co.jp> Signed-off-by: Masami Hiramatsu Signed-off-by: Arnaldo Carvalho de Melo --- kernel/trace/trace_kprobe.c | 17 ++++++++++------- 1 file changed, 10 insertions(+), 7 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace_kprobe.c b/kernel/trace/trace_kprobe.c index 0116c038b0bc..a39251ef1a7b 100644 --- a/kernel/trace/trace_kprobe.c +++ b/kernel/trace/trace_kprobe.c @@ -997,15 +997,18 @@ static int create_trace_probe(int argc, char **argv) /* Parse argument name */ arg = strchr(argv[i], '='); - if (arg) + if (arg) { *arg++ = '\0'; - else + tp->args[i].name = kstrdup(argv[i], GFP_KERNEL); + } else { arg = argv[i]; + /* If argument name is omitted, set "argN" */ + snprintf(buf, MAX_EVENT_NAME_LEN, "arg%d", i + 1); + tp->args[i].name = kstrdup(buf, GFP_KERNEL); + } - tp->args[i].name = kstrdup(argv[i], GFP_KERNEL); if (!tp->args[i].name) { - pr_info("Failed to allocate argument%d name '%s'.\n", - i, argv[i]); + pr_info("Failed to allocate argument[%d] name.\n", i); ret = -ENOMEM; goto error; } @@ -1014,7 +1017,7 @@ static int create_trace_probe(int argc, char **argv) *tmp = '_'; /* convert : to _ */ if (conflict_field_name(tp->args[i].name, tp->args, i)) { - pr_info("Argument%d name '%s' conflicts with " + pr_info("Argument[%d] name '%s' conflicts with " "another field.\n", i, argv[i]); ret = -EINVAL; goto error; @@ -1023,7 +1026,7 @@ static int create_trace_probe(int argc, char **argv) /* Parse fetch argument */ ret = parse_probe_arg(arg, tp, &tp->args[i], is_return); if (ret) { - pr_info("Parse error at argument%d. (%d)\n", i, ret); + pr_info("Parse error at argument[%d]. (%d)\n", i, ret); goto error; } } -- cgit v1.2.3 From da34634fd39958725310d2c30c9b4543945f968b Mon Sep 17 00:00:00 2001 From: Masami Hiramatsu Date: Fri, 27 Aug 2010 20:39:12 +0900 Subject: tracing/kprobe: Fix handling of C-unlike argument names Check the argument name whether it is invalid (not C-like symbol name). This makes event format simple. Reported-by: Srikar Dronamraju Cc: Steven Rostedt Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mathieu Desnoyers LKML-Reference: <20100827113912.22882.62313.stgit@ltc236.sdl.hitachi.co.jp> Signed-off-by: Masami Hiramatsu Signed-off-by: Arnaldo Carvalho de Melo --- kernel/trace/trace_kprobe.c | 20 ++++++++++++-------- 1 file changed, 12 insertions(+), 8 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace_kprobe.c b/kernel/trace/trace_kprobe.c index a39251ef1a7b..544301d29dee 100644 --- a/kernel/trace/trace_kprobe.c +++ b/kernel/trace/trace_kprobe.c @@ -514,8 +514,8 @@ static int kprobe_dispatcher(struct kprobe *kp, struct pt_regs *regs); static int kretprobe_dispatcher(struct kretprobe_instance *ri, struct pt_regs *regs); -/* Check the name is good for event/group */ -static int check_event_name(const char *name) +/* Check the name is good for event/group/fields */ +static int is_good_name(const char *name) { if (!isalpha(*name) && *name != '_') return 0; @@ -557,7 +557,7 @@ static struct trace_probe *alloc_trace_probe(const char *group, else tp->rp.kp.pre_handler = kprobe_dispatcher; - if (!event || !check_event_name(event)) { + if (!event || !is_good_name(event)) { ret = -EINVAL; goto error; } @@ -567,7 +567,7 @@ static struct trace_probe *alloc_trace_probe(const char *group, if (!tp->call.name) goto error; - if (!group || !check_event_name(group)) { + if (!group || !is_good_name(group)) { ret = -EINVAL; goto error; } @@ -883,7 +883,7 @@ static int create_trace_probe(int argc, char **argv) int i, ret = 0; int is_return = 0, is_delete = 0; char *symbol = NULL, *event = NULL, *group = NULL; - char *arg, *tmp; + char *arg; unsigned long offset = 0; void *addr = NULL; char buf[MAX_EVENT_NAME_LEN]; @@ -1012,9 +1012,13 @@ static int create_trace_probe(int argc, char **argv) ret = -ENOMEM; goto error; } - tmp = strchr(tp->args[i].name, ':'); - if (tmp) - *tmp = '_'; /* convert : to _ */ + + if (!is_good_name(tp->args[i].name)) { + pr_info("Invalid argument[%d] name: %s\n", + i, tp->args[i].name); + ret = -EINVAL; + goto error; + } if (conflict_field_name(tp->args[i].name, tp->args, i)) { pr_info("Argument[%d] name '%s' conflicts with " -- cgit v1.2.3 From 9cb627d5f38830ca19aa0dca52d1d3a633018bf7 Mon Sep 17 00:00:00 2001 From: Li Zefan Date: Wed, 1 Sep 2010 12:58:43 +0200 Subject: perf, trace: Fix module leak Commit 1c024eca (perf, trace: Optimize tracepoints by using per-tracepoint-per-cpu hlist to track events) caused a module refcount leak. Reported-And-Tested-by: Avi Kivity Signed-off-by: Peter Zijlstra LKML-Reference: <4C7E1F12.8030304@cn.fujitsu.com> Signed-off-by: Ingo Molnar --- kernel/trace/trace_event_perf.c | 3 +++ 1 file changed, 3 insertions(+) (limited to 'kernel') diff --git a/kernel/trace/trace_event_perf.c b/kernel/trace/trace_event_perf.c index 000e6e85b445..31cc4cb0dbf2 100644 --- a/kernel/trace/trace_event_perf.c +++ b/kernel/trace/trace_event_perf.c @@ -91,6 +91,8 @@ int perf_trace_init(struct perf_event *p_event) tp_event->class && tp_event->class->reg && try_module_get(tp_event->mod)) { ret = perf_trace_event_init(tp_event, p_event); + if (ret) + module_put(tp_event->mod); break; } } @@ -146,6 +148,7 @@ void perf_trace_destroy(struct perf_event *p_event) } } out: + module_put(tp_event->mod); mutex_unlock(&event_mutex); } -- cgit v1.2.3 From 5e11637e2c929e34dcc0fbbfb48bdb638937701a Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Fri, 11 Jun 2010 13:35:08 +0200 Subject: perf: Fix CPU hotplug Since we have UP_PREPARE, we should also have UP_CANCELED. Signed-off-by: Peter Zijlstra Cc: paulus LKML-Reference: Signed-off-by: Ingo Molnar --- kernel/perf_event.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index 657555a5f30f..db5b56064687 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -5761,15 +5761,15 @@ perf_cpu_notify(struct notifier_block *self, unsigned long action, void *hcpu) { unsigned int cpu = (long)hcpu; - switch (action) { + switch (action & ~CPU_TASKS_FROZEN) { case CPU_UP_PREPARE: - case CPU_UP_PREPARE_FROZEN: + case CPU_DOWN_FAILED: perf_event_init_cpu(cpu); break; + case CPU_UP_CANCELED: case CPU_DOWN_PREPARE: - case CPU_DOWN_PREPARE_FROZEN: perf_event_exit_cpu(cpu); break; -- cgit v1.2.3 From da2b71edd8a7db44fe1746261410a981f3e03632 Mon Sep 17 00:00:00 2001 From: Suresh Siddha Date: Mon, 23 Aug 2010 13:42:51 -0700 Subject: sched: Move sched_avg_update() to update_cpu_load() Currently sched_avg_update() (which updates rt_avg stats in the rq) is getting called from scale_rt_power() (in the load balance context) which doesn't take rq->lock. Fix it by moving the sched_avg_update() to more appropriate update_cpu_load() where the CFS load gets updated as well. Signed-off-by: Suresh Siddha Signed-off-by: Peter Zijlstra LKML-Reference: <1282596171.2694.3.camel@sbsiddha-MOBL3> Signed-off-by: Ingo Molnar --- kernel/sched.c | 6 ++++++ kernel/sched_fair.c | 2 -- 2 files changed, 6 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index 09b574e7f4df..ed09d4f2a69c 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -1294,6 +1294,10 @@ static void resched_task(struct task_struct *p) static void sched_rt_avg_update(struct rq *rq, u64 rt_delta) { } + +static void sched_avg_update(struct rq *rq) +{ +} #endif /* CONFIG_SMP */ #if BITS_PER_LONG == 32 @@ -3182,6 +3186,8 @@ static void update_cpu_load(struct rq *this_rq) this_rq->cpu_load[i] = (old_load * (scale - 1) + new_load) >> i; } + + sched_avg_update(this_rq); } static void update_cpu_load_active(struct rq *this_rq) diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c index ab661ebc4895..f53ec7550056 100644 --- a/kernel/sched_fair.c +++ b/kernel/sched_fair.c @@ -2268,8 +2268,6 @@ unsigned long scale_rt_power(int cpu) struct rq *rq = cpu_rq(cpu); u64 total, available; - sched_avg_update(rq); - total = sched_avg_period() + (rq->clock - rq->age_stamp); available = total - rq->rt_avg; -- cgit v1.2.3 From f269893c575167447cc9f6d1867e639fb5b6f0c5 Mon Sep 17 00:00:00 2001 From: Heiko Carstens Date: Tue, 31 Aug 2010 10:28:15 +0200 Subject: sched: Merge cpu_to_core_group functions Merge and simplify the two cpu_to_core_group variants so that the resulting function follows the same pattern like cpu_to_phys_group. Signed-off-by: Heiko Carstens Signed-off-by: Peter Zijlstra LKML-Reference: <20100831082843.953617555@de.ibm.com> Signed-off-by: Ingo Molnar --- kernel/sched.c | 18 +++++------------- 1 file changed, 5 insertions(+), 13 deletions(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index 8eef8e5512d4..1a0c084b1cf9 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -6552,31 +6552,23 @@ cpu_to_cpu_group(int cpu, const struct cpumask *cpu_map, #ifdef CONFIG_SCHED_MC static DEFINE_PER_CPU(struct static_sched_domain, core_domains); static DEFINE_PER_CPU(struct static_sched_group, sched_group_core); -#endif /* CONFIG_SCHED_MC */ -#if defined(CONFIG_SCHED_MC) && defined(CONFIG_SCHED_SMT) static int cpu_to_core_group(int cpu, const struct cpumask *cpu_map, struct sched_group **sg, struct cpumask *mask) { int group; - +#ifdef CONFIG_SCHED_SMT cpumask_and(mask, topology_thread_cpumask(cpu), cpu_map); group = cpumask_first(mask); +#else + group = cpu; +#endif if (sg) *sg = &per_cpu(sched_group_core, group).sg; return group; } -#elif defined(CONFIG_SCHED_MC) -static int -cpu_to_core_group(int cpu, const struct cpumask *cpu_map, - struct sched_group **sg, struct cpumask *unused) -{ - if (sg) - *sg = &per_cpu(sched_group_core, cpu).sg; - return cpu; -} -#endif +#endif /* CONFIG_SCHED_MC */ static DEFINE_PER_CPU(struct static_sched_domain, phys_domains); static DEFINE_PER_CPU(struct static_sched_group, sched_group_phys); -- cgit v1.2.3 From 01a08546af311c065f34727787dd0cc8dc0c216f Mon Sep 17 00:00:00 2001 From: Heiko Carstens Date: Tue, 31 Aug 2010 10:28:16 +0200 Subject: sched: Add book scheduling domain On top of the SMT and MC scheduling domains this adds the BOOK scheduling domain. This is useful for NUMA like machines which do not have an interface which tells which piece of memory is attached to which node or where the hardware performs striping. Signed-off-by: Heiko Carstens Signed-off-by: Peter Zijlstra LKML-Reference: <20100831082844.253053798@de.ibm.com> Signed-off-by: Ingo Molnar --- kernel/sched.c | 77 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++-- 1 file changed, 75 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index 1a0c084b1cf9..26f83e2f1534 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -6506,6 +6506,7 @@ struct s_data { cpumask_var_t nodemask; cpumask_var_t this_sibling_map; cpumask_var_t this_core_map; + cpumask_var_t this_book_map; cpumask_var_t send_covered; cpumask_var_t tmpmask; struct sched_group **sched_group_nodes; @@ -6517,6 +6518,7 @@ enum s_alloc { sa_rootdomain, sa_tmpmask, sa_send_covered, + sa_this_book_map, sa_this_core_map, sa_this_sibling_map, sa_nodemask, @@ -6570,6 +6572,31 @@ cpu_to_core_group(int cpu, const struct cpumask *cpu_map, } #endif /* CONFIG_SCHED_MC */ +/* + * book sched-domains: + */ +#ifdef CONFIG_SCHED_BOOK +static DEFINE_PER_CPU(struct static_sched_domain, book_domains); +static DEFINE_PER_CPU(struct static_sched_group, sched_group_book); + +static int +cpu_to_book_group(int cpu, const struct cpumask *cpu_map, + struct sched_group **sg, struct cpumask *mask) +{ + int group = cpu; +#ifdef CONFIG_SCHED_MC + cpumask_and(mask, cpu_coregroup_mask(cpu), cpu_map); + group = cpumask_first(mask); +#elif defined(CONFIG_SCHED_SMT) + cpumask_and(mask, topology_thread_cpumask(cpu), cpu_map); + group = cpumask_first(mask); +#endif + if (sg) + *sg = &per_cpu(sched_group_book, group).sg; + return group; +} +#endif /* CONFIG_SCHED_BOOK */ + static DEFINE_PER_CPU(struct static_sched_domain, phys_domains); static DEFINE_PER_CPU(struct static_sched_group, sched_group_phys); @@ -6578,7 +6605,10 @@ cpu_to_phys_group(int cpu, const struct cpumask *cpu_map, struct sched_group **sg, struct cpumask *mask) { int group; -#ifdef CONFIG_SCHED_MC +#ifdef CONFIG_SCHED_BOOK + cpumask_and(mask, cpu_book_mask(cpu), cpu_map); + group = cpumask_first(mask); +#elif defined(CONFIG_SCHED_MC) cpumask_and(mask, cpu_coregroup_mask(cpu), cpu_map); group = cpumask_first(mask); #elif defined(CONFIG_SCHED_SMT) @@ -6839,6 +6869,9 @@ SD_INIT_FUNC(CPU) #ifdef CONFIG_SCHED_MC SD_INIT_FUNC(MC) #endif +#ifdef CONFIG_SCHED_BOOK + SD_INIT_FUNC(BOOK) +#endif static int default_relax_domain_level = -1; @@ -6888,6 +6921,8 @@ static void __free_domain_allocs(struct s_data *d, enum s_alloc what, free_cpumask_var(d->tmpmask); /* fall through */ case sa_send_covered: free_cpumask_var(d->send_covered); /* fall through */ + case sa_this_book_map: + free_cpumask_var(d->this_book_map); /* fall through */ case sa_this_core_map: free_cpumask_var(d->this_core_map); /* fall through */ case sa_this_sibling_map: @@ -6934,8 +6969,10 @@ static enum s_alloc __visit_domain_allocation_hell(struct s_data *d, return sa_nodemask; if (!alloc_cpumask_var(&d->this_core_map, GFP_KERNEL)) return sa_this_sibling_map; - if (!alloc_cpumask_var(&d->send_covered, GFP_KERNEL)) + if (!alloc_cpumask_var(&d->this_book_map, GFP_KERNEL)) return sa_this_core_map; + if (!alloc_cpumask_var(&d->send_covered, GFP_KERNEL)) + return sa_this_book_map; if (!alloc_cpumask_var(&d->tmpmask, GFP_KERNEL)) return sa_send_covered; d->rd = alloc_rootdomain(); @@ -6993,6 +7030,23 @@ static struct sched_domain *__build_cpu_sched_domain(struct s_data *d, return sd; } +static struct sched_domain *__build_book_sched_domain(struct s_data *d, + const struct cpumask *cpu_map, struct sched_domain_attr *attr, + struct sched_domain *parent, int i) +{ + struct sched_domain *sd = parent; +#ifdef CONFIG_SCHED_BOOK + sd = &per_cpu(book_domains, i).sd; + SD_INIT(sd, BOOK); + set_domain_attribute(sd, attr); + cpumask_and(sched_domain_span(sd), cpu_map, cpu_book_mask(i)); + sd->parent = parent; + parent->child = sd; + cpu_to_book_group(i, cpu_map, &sd->groups, d->tmpmask); +#endif + return sd; +} + static struct sched_domain *__build_mc_sched_domain(struct s_data *d, const struct cpumask *cpu_map, struct sched_domain_attr *attr, struct sched_domain *parent, int i) @@ -7049,6 +7103,15 @@ static void build_sched_groups(struct s_data *d, enum sched_domain_level l, &cpu_to_core_group, d->send_covered, d->tmpmask); break; +#endif +#ifdef CONFIG_SCHED_BOOK + case SD_LV_BOOK: /* set up book groups */ + cpumask_and(d->this_book_map, cpu_map, cpu_book_mask(cpu)); + if (cpu == cpumask_first(d->this_book_map)) + init_sched_build_groups(d->this_book_map, cpu_map, + &cpu_to_book_group, + d->send_covered, d->tmpmask); + break; #endif case SD_LV_CPU: /* set up physical groups */ cpumask_and(d->nodemask, cpumask_of_node(cpu), cpu_map); @@ -7097,12 +7160,14 @@ static int __build_sched_domains(const struct cpumask *cpu_map, sd = __build_numa_sched_domains(&d, cpu_map, attr, i); sd = __build_cpu_sched_domain(&d, cpu_map, attr, sd, i); + sd = __build_book_sched_domain(&d, cpu_map, attr, sd, i); sd = __build_mc_sched_domain(&d, cpu_map, attr, sd, i); sd = __build_smt_sched_domain(&d, cpu_map, attr, sd, i); } for_each_cpu(i, cpu_map) { build_sched_groups(&d, SD_LV_SIBLING, cpu_map, i); + build_sched_groups(&d, SD_LV_BOOK, cpu_map, i); build_sched_groups(&d, SD_LV_MC, cpu_map, i); } @@ -7133,6 +7198,12 @@ static int __build_sched_domains(const struct cpumask *cpu_map, init_sched_groups_power(i, sd); } #endif +#ifdef CONFIG_SCHED_BOOK + for_each_cpu(i, cpu_map) { + sd = &per_cpu(book_domains, i).sd; + init_sched_groups_power(i, sd); + } +#endif for_each_cpu(i, cpu_map) { sd = &per_cpu(phys_domains, i).sd; @@ -7158,6 +7229,8 @@ static int __build_sched_domains(const struct cpumask *cpu_map, sd = &per_cpu(cpu_domains, i).sd; #elif defined(CONFIG_SCHED_MC) sd = &per_cpu(core_domains, i).sd; +#elif defined(CONFIG_SCHED_BOOK) + sd = &per_cpu(book_domains, i).sd; #else sd = &per_cpu(phys_domains, i).sd; #endif -- cgit v1.2.3 From 51b0fe39549a04858001922919ab355dee9bdfcf Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Fri, 11 Jun 2010 13:35:57 +0200 Subject: perf: Deconstify struct pmu sed -ie 's/const struct pmu\>/struct pmu/g' `git grep -l "const struct pmu\>"` Signed-off-by: Peter Zijlstra Cc: paulus Cc: stephane eranian Cc: Robert Richter Cc: Will Deacon Cc: Paul Mundt Cc: Frederic Weisbecker Cc: Cyrill Gorcunov Cc: Lin Ming Cc: Yanmin Cc: Deng-Cheng Zhu Cc: David Miller Cc: Michael Cree LKML-Reference: Signed-off-by: Ingo Molnar --- kernel/perf_event.c | 26 +++++++++++++------------- 1 file changed, 13 insertions(+), 13 deletions(-) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index 2d74f31220ad..fb46fd13f31f 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -75,7 +75,7 @@ static DEFINE_SPINLOCK(perf_resource_lock); /* * Architecture provided APIs - weak aliases: */ -extern __weak const struct pmu *hw_perf_event_init(struct perf_event *event) +extern __weak struct pmu *hw_perf_event_init(struct perf_event *event) { return NULL; } @@ -691,7 +691,7 @@ group_sched_in(struct perf_event *group_event, struct perf_event_context *ctx) { struct perf_event *event, *partial_group = NULL; - const struct pmu *pmu = group_event->pmu; + struct pmu *pmu = group_event->pmu; bool txn = false; if (group_event->state == PERF_EVENT_STATE_OFF) @@ -4501,7 +4501,7 @@ static int perf_swevent_int(struct perf_event *event) return 0; } -static const struct pmu perf_ops_generic = { +static struct pmu perf_ops_generic = { .enable = perf_swevent_enable, .disable = perf_swevent_disable, .start = perf_swevent_int, @@ -4614,7 +4614,7 @@ static void cpu_clock_perf_event_read(struct perf_event *event) cpu_clock_perf_event_update(event); } -static const struct pmu perf_ops_cpu_clock = { +static struct pmu perf_ops_cpu_clock = { .enable = cpu_clock_perf_event_enable, .disable = cpu_clock_perf_event_disable, .read = cpu_clock_perf_event_read, @@ -4671,7 +4671,7 @@ static void task_clock_perf_event_read(struct perf_event *event) task_clock_perf_event_update(event, time); } -static const struct pmu perf_ops_task_clock = { +static struct pmu perf_ops_task_clock = { .enable = task_clock_perf_event_enable, .disable = task_clock_perf_event_disable, .read = task_clock_perf_event_read, @@ -4785,7 +4785,7 @@ static int swevent_hlist_get(struct perf_event *event) #ifdef CONFIG_EVENT_TRACING -static const struct pmu perf_ops_tracepoint = { +static struct pmu perf_ops_tracepoint = { .enable = perf_trace_enable, .disable = perf_trace_disable, .start = perf_swevent_int, @@ -4849,7 +4849,7 @@ static void tp_perf_event_destroy(struct perf_event *event) perf_trace_destroy(event); } -static const struct pmu *tp_perf_event_init(struct perf_event *event) +static struct pmu *tp_perf_event_init(struct perf_event *event) { int err; @@ -4896,7 +4896,7 @@ static void perf_event_free_filter(struct perf_event *event) #else -static const struct pmu *tp_perf_event_init(struct perf_event *event) +static struct pmu *tp_perf_event_init(struct perf_event *event) { return NULL; } @@ -4918,7 +4918,7 @@ static void bp_perf_event_destroy(struct perf_event *event) release_bp_slot(event); } -static const struct pmu *bp_perf_event_init(struct perf_event *bp) +static struct pmu *bp_perf_event_init(struct perf_event *bp) { int err; @@ -4942,7 +4942,7 @@ void perf_bp_event(struct perf_event *bp, void *data) perf_swevent_add(bp, 1, 1, &sample, regs); } #else -static const struct pmu *bp_perf_event_init(struct perf_event *bp) +static struct pmu *bp_perf_event_init(struct perf_event *bp) { return NULL; } @@ -4964,9 +4964,9 @@ static void sw_perf_event_destroy(struct perf_event *event) swevent_hlist_put(event); } -static const struct pmu *sw_perf_event_init(struct perf_event *event) +static struct pmu *sw_perf_event_init(struct perf_event *event) { - const struct pmu *pmu = NULL; + struct pmu *pmu = NULL; u64 event_id = event->attr.config; /* @@ -5028,7 +5028,7 @@ perf_event_alloc(struct perf_event_attr *attr, perf_overflow_handler_t overflow_handler, gfp_t gfpflags) { - const struct pmu *pmu; + struct pmu *pmu; struct perf_event *event; struct hw_perf_event *hwc; long err; -- cgit v1.2.3 From b0a873ebbf87bf38bf70b5e39a7cadc96099fa13 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Fri, 11 Jun 2010 13:35:08 +0200 Subject: perf: Register PMU implementations Simple registration interface for struct pmu, this provides the infrastructure for removing all the weak functions. Signed-off-by: Peter Zijlstra Cc: paulus Cc: stephane eranian Cc: Robert Richter Cc: Will Deacon Cc: Paul Mundt Cc: Frederic Weisbecker Cc: Cyrill Gorcunov Cc: Lin Ming Cc: Yanmin Cc: Deng-Cheng Zhu Cc: David Miller Cc: Michael Cree LKML-Reference: Signed-off-by: Ingo Molnar --- kernel/hw_breakpoint.c | 35 ++- kernel/perf_event.c | 588 ++++++++++++++++++++++++------------------------- 2 files changed, 320 insertions(+), 303 deletions(-) (limited to 'kernel') diff --git a/kernel/hw_breakpoint.c b/kernel/hw_breakpoint.c index d71a987fd2bf..e9c5cfa1fd20 100644 --- a/kernel/hw_breakpoint.c +++ b/kernel/hw_breakpoint.c @@ -565,6 +565,34 @@ static struct notifier_block hw_breakpoint_exceptions_nb = { .priority = 0x7fffffff }; +static void bp_perf_event_destroy(struct perf_event *event) +{ + release_bp_slot(event); +} + +static int hw_breakpoint_event_init(struct perf_event *bp) +{ + int err; + + if (bp->attr.type != PERF_TYPE_BREAKPOINT) + return -ENOENT; + + err = register_perf_hw_breakpoint(bp); + if (err) + return err; + + bp->destroy = bp_perf_event_destroy; + + return 0; +} + +static struct pmu perf_breakpoint = { + .event_init = hw_breakpoint_event_init, + .enable = arch_install_hw_breakpoint, + .disable = arch_uninstall_hw_breakpoint, + .read = hw_breakpoint_pmu_read, +}; + static int __init init_hw_breakpoint(void) { unsigned int **task_bp_pinned; @@ -586,6 +614,8 @@ static int __init init_hw_breakpoint(void) constraints_initialized = 1; + perf_pmu_register(&perf_breakpoint); + return register_die_notifier(&hw_breakpoint_exceptions_nb); err_alloc: @@ -601,8 +631,3 @@ static int __init init_hw_breakpoint(void) core_initcall(init_hw_breakpoint); -struct pmu perf_ops_bp = { - .enable = arch_install_hw_breakpoint, - .disable = arch_uninstall_hw_breakpoint, - .read = hw_breakpoint_pmu_read, -}; diff --git a/kernel/perf_event.c b/kernel/perf_event.c index fb46fd13f31f..288ce43de57c 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -31,7 +31,6 @@ #include #include #include -#include #include @@ -72,14 +71,6 @@ static atomic64_t perf_event_id; */ static DEFINE_SPINLOCK(perf_resource_lock); -/* - * Architecture provided APIs - weak aliases: - */ -extern __weak struct pmu *hw_perf_event_init(struct perf_event *event) -{ - return NULL; -} - void __weak hw_perf_disable(void) { barrier(); } void __weak hw_perf_enable(void) { barrier(); } @@ -4501,182 +4492,6 @@ static int perf_swevent_int(struct perf_event *event) return 0; } -static struct pmu perf_ops_generic = { - .enable = perf_swevent_enable, - .disable = perf_swevent_disable, - .start = perf_swevent_int, - .stop = perf_swevent_void, - .read = perf_swevent_read, - .unthrottle = perf_swevent_void, /* hwc->interrupts already reset */ -}; - -/* - * hrtimer based swevent callback - */ - -static enum hrtimer_restart perf_swevent_hrtimer(struct hrtimer *hrtimer) -{ - enum hrtimer_restart ret = HRTIMER_RESTART; - struct perf_sample_data data; - struct pt_regs *regs; - struct perf_event *event; - u64 period; - - event = container_of(hrtimer, struct perf_event, hw.hrtimer); - event->pmu->read(event); - - perf_sample_data_init(&data, 0); - data.period = event->hw.last_period; - regs = get_irq_regs(); - - if (regs && !perf_exclude_event(event, regs)) { - if (!(event->attr.exclude_idle && current->pid == 0)) - if (perf_event_overflow(event, 0, &data, regs)) - ret = HRTIMER_NORESTART; - } - - period = max_t(u64, 10000, event->hw.sample_period); - hrtimer_forward_now(hrtimer, ns_to_ktime(period)); - - return ret; -} - -static void perf_swevent_start_hrtimer(struct perf_event *event) -{ - struct hw_perf_event *hwc = &event->hw; - - hrtimer_init(&hwc->hrtimer, CLOCK_MONOTONIC, HRTIMER_MODE_REL); - hwc->hrtimer.function = perf_swevent_hrtimer; - if (hwc->sample_period) { - u64 period; - - if (hwc->remaining) { - if (hwc->remaining < 0) - period = 10000; - else - period = hwc->remaining; - hwc->remaining = 0; - } else { - period = max_t(u64, 10000, hwc->sample_period); - } - __hrtimer_start_range_ns(&hwc->hrtimer, - ns_to_ktime(period), 0, - HRTIMER_MODE_REL, 0); - } -} - -static void perf_swevent_cancel_hrtimer(struct perf_event *event) -{ - struct hw_perf_event *hwc = &event->hw; - - if (hwc->sample_period) { - ktime_t remaining = hrtimer_get_remaining(&hwc->hrtimer); - hwc->remaining = ktime_to_ns(remaining); - - hrtimer_cancel(&hwc->hrtimer); - } -} - -/* - * Software event: cpu wall time clock - */ - -static void cpu_clock_perf_event_update(struct perf_event *event) -{ - int cpu = raw_smp_processor_id(); - s64 prev; - u64 now; - - now = cpu_clock(cpu); - prev = local64_xchg(&event->hw.prev_count, now); - local64_add(now - prev, &event->count); -} - -static int cpu_clock_perf_event_enable(struct perf_event *event) -{ - struct hw_perf_event *hwc = &event->hw; - int cpu = raw_smp_processor_id(); - - local64_set(&hwc->prev_count, cpu_clock(cpu)); - perf_swevent_start_hrtimer(event); - - return 0; -} - -static void cpu_clock_perf_event_disable(struct perf_event *event) -{ - perf_swevent_cancel_hrtimer(event); - cpu_clock_perf_event_update(event); -} - -static void cpu_clock_perf_event_read(struct perf_event *event) -{ - cpu_clock_perf_event_update(event); -} - -static struct pmu perf_ops_cpu_clock = { - .enable = cpu_clock_perf_event_enable, - .disable = cpu_clock_perf_event_disable, - .read = cpu_clock_perf_event_read, -}; - -/* - * Software event: task time clock - */ - -static void task_clock_perf_event_update(struct perf_event *event, u64 now) -{ - u64 prev; - s64 delta; - - prev = local64_xchg(&event->hw.prev_count, now); - delta = now - prev; - local64_add(delta, &event->count); -} - -static int task_clock_perf_event_enable(struct perf_event *event) -{ - struct hw_perf_event *hwc = &event->hw; - u64 now; - - now = event->ctx->time; - - local64_set(&hwc->prev_count, now); - - perf_swevent_start_hrtimer(event); - - return 0; -} - -static void task_clock_perf_event_disable(struct perf_event *event) -{ - perf_swevent_cancel_hrtimer(event); - task_clock_perf_event_update(event, event->ctx->time); - -} - -static void task_clock_perf_event_read(struct perf_event *event) -{ - u64 time; - - if (!in_nmi()) { - update_context_time(event->ctx); - time = event->ctx->time; - } else { - u64 now = perf_clock(); - u64 delta = now - event->ctx->timestamp; - time = event->ctx->time + delta; - } - - task_clock_perf_event_update(event, time); -} - -static struct pmu perf_ops_task_clock = { - .enable = task_clock_perf_event_enable, - .disable = task_clock_perf_event_disable, - .read = task_clock_perf_event_read, -}; - /* Deref the hlist from the update side */ static inline struct swevent_hlist * swevent_hlist_deref(struct perf_cpu_context *cpuctx) @@ -4783,17 +4598,63 @@ static int swevent_hlist_get(struct perf_event *event) return err; } -#ifdef CONFIG_EVENT_TRACING +atomic_t perf_swevent_enabled[PERF_COUNT_SW_MAX]; -static struct pmu perf_ops_tracepoint = { - .enable = perf_trace_enable, - .disable = perf_trace_disable, +static void sw_perf_event_destroy(struct perf_event *event) +{ + u64 event_id = event->attr.config; + + WARN_ON(event->parent); + + atomic_dec(&perf_swevent_enabled[event_id]); + swevent_hlist_put(event); +} + +static int perf_swevent_init(struct perf_event *event) +{ + int event_id = event->attr.config; + + if (event->attr.type != PERF_TYPE_SOFTWARE) + return -ENOENT; + + switch (event_id) { + case PERF_COUNT_SW_CPU_CLOCK: + case PERF_COUNT_SW_TASK_CLOCK: + return -ENOENT; + + default: + break; + } + + if (event_id > PERF_COUNT_SW_MAX) + return -ENOENT; + + if (!event->parent) { + int err; + + err = swevent_hlist_get(event); + if (err) + return err; + + atomic_inc(&perf_swevent_enabled[event_id]); + event->destroy = sw_perf_event_destroy; + } + + return 0; +} + +static struct pmu perf_swevent = { + .event_init = perf_swevent_init, + .enable = perf_swevent_enable, + .disable = perf_swevent_disable, .start = perf_swevent_int, .stop = perf_swevent_void, .read = perf_swevent_read, - .unthrottle = perf_swevent_void, + .unthrottle = perf_swevent_void, /* hwc->interrupts already reset */ }; +#ifdef CONFIG_EVENT_TRACING + static int perf_tp_filter_match(struct perf_event *event, struct perf_sample_data *data) { @@ -4849,10 +4710,13 @@ static void tp_perf_event_destroy(struct perf_event *event) perf_trace_destroy(event); } -static struct pmu *tp_perf_event_init(struct perf_event *event) +static int perf_tp_event_init(struct perf_event *event) { int err; + if (event->attr.type != PERF_TYPE_TRACEPOINT) + return -ENOENT; + /* * Raw tracepoint data is a severe data leak, only allow root to * have these. @@ -4860,15 +4724,30 @@ static struct pmu *tp_perf_event_init(struct perf_event *event) if ((event->attr.sample_type & PERF_SAMPLE_RAW) && perf_paranoid_tracepoint_raw() && !capable(CAP_SYS_ADMIN)) - return ERR_PTR(-EPERM); + return -EPERM; err = perf_trace_init(event); if (err) - return NULL; + return err; event->destroy = tp_perf_event_destroy; - return &perf_ops_tracepoint; + return 0; +} + +static struct pmu perf_tracepoint = { + .event_init = perf_tp_event_init, + .enable = perf_trace_enable, + .disable = perf_trace_disable, + .start = perf_swevent_int, + .stop = perf_swevent_void, + .read = perf_swevent_read, + .unthrottle = perf_swevent_void, +}; + +static inline void perf_tp_register(void) +{ + perf_pmu_register(&perf_tracepoint); } static int perf_event_set_filter(struct perf_event *event, void __user *arg) @@ -4896,9 +4775,8 @@ static void perf_event_free_filter(struct perf_event *event) #else -static struct pmu *tp_perf_event_init(struct perf_event *event) +static inline void perf_tp_register(void) { - return NULL; } static int perf_event_set_filter(struct perf_event *event, void __user *arg) @@ -4913,105 +4791,247 @@ static void perf_event_free_filter(struct perf_event *event) #endif /* CONFIG_EVENT_TRACING */ #ifdef CONFIG_HAVE_HW_BREAKPOINT -static void bp_perf_event_destroy(struct perf_event *event) +void perf_bp_event(struct perf_event *bp, void *data) { - release_bp_slot(event); + struct perf_sample_data sample; + struct pt_regs *regs = data; + + perf_sample_data_init(&sample, bp->attr.bp_addr); + + if (!perf_exclude_event(bp, regs)) + perf_swevent_add(bp, 1, 1, &sample, regs); } +#endif + +/* + * hrtimer based swevent callback + */ -static struct pmu *bp_perf_event_init(struct perf_event *bp) +static enum hrtimer_restart perf_swevent_hrtimer(struct hrtimer *hrtimer) { - int err; + enum hrtimer_restart ret = HRTIMER_RESTART; + struct perf_sample_data data; + struct pt_regs *regs; + struct perf_event *event; + u64 period; - err = register_perf_hw_breakpoint(bp); - if (err) - return ERR_PTR(err); + event = container_of(hrtimer, struct perf_event, hw.hrtimer); + event->pmu->read(event); + + perf_sample_data_init(&data, 0); + data.period = event->hw.last_period; + regs = get_irq_regs(); + + if (regs && !perf_exclude_event(event, regs)) { + if (!(event->attr.exclude_idle && current->pid == 0)) + if (perf_event_overflow(event, 0, &data, regs)) + ret = HRTIMER_NORESTART; + } - bp->destroy = bp_perf_event_destroy; + period = max_t(u64, 10000, event->hw.sample_period); + hrtimer_forward_now(hrtimer, ns_to_ktime(period)); - return &perf_ops_bp; + return ret; } -void perf_bp_event(struct perf_event *bp, void *data) +static void perf_swevent_start_hrtimer(struct perf_event *event) { - struct perf_sample_data sample; - struct pt_regs *regs = data; + struct hw_perf_event *hwc = &event->hw; - perf_sample_data_init(&sample, bp->attr.bp_addr); + hrtimer_init(&hwc->hrtimer, CLOCK_MONOTONIC, HRTIMER_MODE_REL); + hwc->hrtimer.function = perf_swevent_hrtimer; + if (hwc->sample_period) { + u64 period; - if (!perf_exclude_event(bp, regs)) - perf_swevent_add(bp, 1, 1, &sample, regs); + if (hwc->remaining) { + if (hwc->remaining < 0) + period = 10000; + else + period = hwc->remaining; + hwc->remaining = 0; + } else { + period = max_t(u64, 10000, hwc->sample_period); + } + __hrtimer_start_range_ns(&hwc->hrtimer, + ns_to_ktime(period), 0, + HRTIMER_MODE_REL, 0); + } } -#else -static struct pmu *bp_perf_event_init(struct perf_event *bp) + +static void perf_swevent_cancel_hrtimer(struct perf_event *event) { - return NULL; + struct hw_perf_event *hwc = &event->hw; + + if (hwc->sample_period) { + ktime_t remaining = hrtimer_get_remaining(&hwc->hrtimer); + hwc->remaining = ktime_to_ns(remaining); + + hrtimer_cancel(&hwc->hrtimer); + } } -void perf_bp_event(struct perf_event *bp, void *regs) +/* + * Software event: cpu wall time clock + */ + +static void cpu_clock_event_update(struct perf_event *event) { + int cpu = raw_smp_processor_id(); + s64 prev; + u64 now; + + now = cpu_clock(cpu); + prev = local64_xchg(&event->hw.prev_count, now); + local64_add(now - prev, &event->count); } -#endif -atomic_t perf_swevent_enabled[PERF_COUNT_SW_MAX]; +static int cpu_clock_event_enable(struct perf_event *event) +{ + struct hw_perf_event *hwc = &event->hw; + int cpu = raw_smp_processor_id(); -static void sw_perf_event_destroy(struct perf_event *event) + local64_set(&hwc->prev_count, cpu_clock(cpu)); + perf_swevent_start_hrtimer(event); + + return 0; +} + +static void cpu_clock_event_disable(struct perf_event *event) { - u64 event_id = event->attr.config; + perf_swevent_cancel_hrtimer(event); + cpu_clock_event_update(event); +} - WARN_ON(event->parent); +static void cpu_clock_event_read(struct perf_event *event) +{ + cpu_clock_event_update(event); +} - atomic_dec(&perf_swevent_enabled[event_id]); - swevent_hlist_put(event); +static int cpu_clock_event_init(struct perf_event *event) +{ + if (event->attr.type != PERF_TYPE_SOFTWARE) + return -ENOENT; + + if (event->attr.config != PERF_COUNT_SW_CPU_CLOCK) + return -ENOENT; + + return 0; } -static struct pmu *sw_perf_event_init(struct perf_event *event) +static struct pmu perf_cpu_clock = { + .event_init = cpu_clock_event_init, + .enable = cpu_clock_event_enable, + .disable = cpu_clock_event_disable, + .read = cpu_clock_event_read, +}; + +/* + * Software event: task time clock + */ + +static void task_clock_event_update(struct perf_event *event, u64 now) { - struct pmu *pmu = NULL; - u64 event_id = event->attr.config; + u64 prev; + s64 delta; - /* - * Software events (currently) can't in general distinguish - * between user, kernel and hypervisor events. - * However, context switches and cpu migrations are considered - * to be kernel events, and page faults are never hypervisor - * events. - */ - switch (event_id) { - case PERF_COUNT_SW_CPU_CLOCK: - pmu = &perf_ops_cpu_clock; + prev = local64_xchg(&event->hw.prev_count, now); + delta = now - prev; + local64_add(delta, &event->count); +} - break; - case PERF_COUNT_SW_TASK_CLOCK: - /* - * If the user instantiates this as a per-cpu event, - * use the cpu_clock event instead. - */ - if (event->ctx->task) - pmu = &perf_ops_task_clock; - else - pmu = &perf_ops_cpu_clock; +static int task_clock_event_enable(struct perf_event *event) +{ + struct hw_perf_event *hwc = &event->hw; + u64 now; - break; - case PERF_COUNT_SW_PAGE_FAULTS: - case PERF_COUNT_SW_PAGE_FAULTS_MIN: - case PERF_COUNT_SW_PAGE_FAULTS_MAJ: - case PERF_COUNT_SW_CONTEXT_SWITCHES: - case PERF_COUNT_SW_CPU_MIGRATIONS: - case PERF_COUNT_SW_ALIGNMENT_FAULTS: - case PERF_COUNT_SW_EMULATION_FAULTS: - if (!event->parent) { - int err; - - err = swevent_hlist_get(event); - if (err) - return ERR_PTR(err); + now = event->ctx->time; - atomic_inc(&perf_swevent_enabled[event_id]); - event->destroy = sw_perf_event_destroy; + local64_set(&hwc->prev_count, now); + + perf_swevent_start_hrtimer(event); + + return 0; +} + +static void task_clock_event_disable(struct perf_event *event) +{ + perf_swevent_cancel_hrtimer(event); + task_clock_event_update(event, event->ctx->time); + +} + +static void task_clock_event_read(struct perf_event *event) +{ + u64 time; + + if (!in_nmi()) { + update_context_time(event->ctx); + time = event->ctx->time; + } else { + u64 now = perf_clock(); + u64 delta = now - event->ctx->timestamp; + time = event->ctx->time + delta; + } + + task_clock_event_update(event, time); +} + +static int task_clock_event_init(struct perf_event *event) +{ + if (event->attr.type != PERF_TYPE_SOFTWARE) + return -ENOENT; + + if (event->attr.config != PERF_COUNT_SW_TASK_CLOCK) + return -ENOENT; + + return 0; +} + +static struct pmu perf_task_clock = { + .event_init = task_clock_event_init, + .enable = task_clock_event_enable, + .disable = task_clock_event_disable, + .read = task_clock_event_read, +}; + +static LIST_HEAD(pmus); +static DEFINE_MUTEX(pmus_lock); +static struct srcu_struct pmus_srcu; + +int perf_pmu_register(struct pmu *pmu) +{ + mutex_lock(&pmus_lock); + list_add_rcu(&pmu->entry, &pmus); + mutex_unlock(&pmus_lock); + + return 0; +} + +void perf_pmu_unregister(struct pmu *pmu) +{ + mutex_lock(&pmus_lock); + list_del_rcu(&pmu->entry); + mutex_unlock(&pmus_lock); + + synchronize_srcu(&pmus_srcu); +} + +struct pmu *perf_init_event(struct perf_event *event) +{ + struct pmu *pmu = NULL; + int idx; + + idx = srcu_read_lock(&pmus_srcu); + list_for_each_entry_rcu(pmu, &pmus, entry) { + int ret = pmu->event_init(event); + if (!ret) + break; + if (ret != -ENOENT) { + pmu = ERR_PTR(ret); + break; } - pmu = &perf_ops_generic; - break; } + srcu_read_unlock(&pmus_srcu, idx); return pmu; } @@ -5092,29 +5112,8 @@ perf_event_alloc(struct perf_event_attr *attr, if (attr->inherit && (attr->read_format & PERF_FORMAT_GROUP)) goto done; - switch (attr->type) { - case PERF_TYPE_RAW: - case PERF_TYPE_HARDWARE: - case PERF_TYPE_HW_CACHE: - pmu = hw_perf_event_init(event); - break; - - case PERF_TYPE_SOFTWARE: - pmu = sw_perf_event_init(event); - break; - - case PERF_TYPE_TRACEPOINT: - pmu = tp_perf_event_init(event); - break; + pmu = perf_init_event(event); - case PERF_TYPE_BREAKPOINT: - pmu = bp_perf_event_init(event); - break; - - - default: - break; - } done: err = 0; if (!pmu) @@ -5979,22 +5978,15 @@ perf_cpu_notify(struct notifier_block *self, unsigned long action, void *hcpu) return NOTIFY_OK; } -/* - * This has to have a higher priority than migration_notifier in sched.c. - */ -static struct notifier_block __cpuinitdata perf_cpu_nb = { - .notifier_call = perf_cpu_notify, - .priority = 20, -}; - void __init perf_event_init(void) { perf_event_init_all_cpus(); - perf_cpu_notify(&perf_cpu_nb, (unsigned long)CPU_UP_PREPARE, - (void *)(long)smp_processor_id()); - perf_cpu_notify(&perf_cpu_nb, (unsigned long)CPU_ONLINE, - (void *)(long)smp_processor_id()); - register_cpu_notifier(&perf_cpu_nb); + init_srcu_struct(&pmus_srcu); + perf_pmu_register(&perf_swevent); + perf_pmu_register(&perf_cpu_clock); + perf_pmu_register(&perf_task_clock); + perf_tp_register(); + perf_cpu_notifier(perf_cpu_notify); } static ssize_t perf_show_reserve_percpu(struct sysdev_class *class, -- cgit v1.2.3 From 9ed6060d286b1eb55974d09080f442f809408c42 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Fri, 11 Jun 2010 17:36:35 +0200 Subject: perf: Unindent labels Fixup random annoying style bits. Signed-off-by: Peter Zijlstra Cc: paulus LKML-Reference: Signed-off-by: Ingo Molnar --- kernel/perf_event.c | 43 ++++++++++++++++++++++++------------------- 1 file changed, 24 insertions(+), 19 deletions(-) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index 288ce43de57c..149ca18371b7 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -147,7 +147,7 @@ perf_lock_task_context(struct task_struct *task, unsigned long *flags) struct perf_event_context *ctx; rcu_read_lock(); - retry: +retry: ctx = rcu_dereference(task->perf_event_ctxp); if (ctx) { /* @@ -619,7 +619,7 @@ void perf_event_disable(struct perf_event *event) return; } - retry: +retry: task_oncpu_function_call(task, __perf_event_disable, event); raw_spin_lock_irq(&ctx->lock); @@ -849,7 +849,7 @@ static void __perf_install_in_context(void *info) if (!err && !ctx->task && cpuctx->max_pertask) cpuctx->max_pertask--; - unlock: +unlock: perf_enable(); raw_spin_unlock(&ctx->lock); @@ -922,10 +922,12 @@ static void __perf_event_mark_enabled(struct perf_event *event, event->state = PERF_EVENT_STATE_INACTIVE; event->tstamp_enabled = ctx->time - event->total_time_enabled; - list_for_each_entry(sub, &event->sibling_list, group_entry) - if (sub->state >= PERF_EVENT_STATE_INACTIVE) + list_for_each_entry(sub, &event->sibling_list, group_entry) { + if (sub->state >= PERF_EVENT_STATE_INACTIVE) { sub->tstamp_enabled = ctx->time - sub->total_time_enabled; + } + } } /* @@ -991,7 +993,7 @@ static void __perf_event_enable(void *info) } } - unlock: +unlock: raw_spin_unlock(&ctx->lock); } @@ -1032,7 +1034,7 @@ void perf_event_enable(struct perf_event *event) if (event->state == PERF_EVENT_STATE_ERROR) event->state = PERF_EVENT_STATE_OFF; - retry: +retry: raw_spin_unlock_irq(&ctx->lock); task_oncpu_function_call(task, __perf_event_enable, event); @@ -1052,7 +1054,7 @@ void perf_event_enable(struct perf_event *event) if (event->state == PERF_EVENT_STATE_OFF) __perf_event_mark_enabled(event, ctx); - out: +out: raw_spin_unlock_irq(&ctx->lock); } @@ -1092,17 +1094,19 @@ static void ctx_sched_out(struct perf_event_context *ctx, if (!ctx->nr_active) goto out_enable; - if (event_type & EVENT_PINNED) + if (event_type & EVENT_PINNED) { list_for_each_entry(event, &ctx->pinned_groups, group_entry) group_sched_out(event, cpuctx, ctx); + } - if (event_type & EVENT_FLEXIBLE) + if (event_type & EVENT_FLEXIBLE) { list_for_each_entry(event, &ctx->flexible_groups, group_entry) group_sched_out(event, cpuctx, ctx); + } out_enable: perf_enable(); - out: +out: raw_spin_unlock(&ctx->lock); } @@ -1341,9 +1345,10 @@ ctx_flexible_sched_in(struct perf_event_context *ctx, if (event->cpu != -1 && event->cpu != smp_processor_id()) continue; - if (group_can_go_on(event, cpuctx, can_add_hw)) + if (group_can_go_on(event, cpuctx, can_add_hw)) { if (group_sched_in(event, cpuctx, ctx)) can_add_hw = 0; + } } } @@ -1373,7 +1378,7 @@ ctx_sched_in(struct perf_event_context *ctx, ctx_flexible_sched_in(ctx, cpuctx); perf_enable(); - out: +out: raw_spin_unlock(&ctx->lock); } @@ -1714,7 +1719,7 @@ static void perf_event_enable_on_exec(struct task_struct *task) raw_spin_unlock(&ctx->lock); perf_event_task_sched_in(task); - out: +out: local_irq_restore(flags); } @@ -2053,7 +2058,7 @@ static struct perf_event_context *find_get_context(pid_t pid, int cpu) if (!ptrace_may_access(task, PTRACE_MODE_READ)) goto errout; - retry: +retry: ctx = perf_lock_task_context(task, &flags); if (ctx) { unclone_ctx(ctx); @@ -2081,7 +2086,7 @@ static struct perf_event_context *find_get_context(pid_t pid, int cpu) put_task_struct(task); return ctx; - errout: +errout: put_task_struct(task); return ERR_PTR(err); } @@ -3264,7 +3269,7 @@ again: if (handle->wakeup != local_read(&buffer->wakeup)) perf_output_wakeup(handle); - out: +out: preempt_enable(); } @@ -4562,7 +4567,7 @@ static int swevent_hlist_get_cpu(struct perf_event *event, int cpu) rcu_assign_pointer(cpuctx->swevent_hlist, hlist); } cpuctx->hlist_refcount++; - exit: +exit: mutex_unlock(&cpuctx->hlist_mutex); return err; @@ -4587,7 +4592,7 @@ static int swevent_hlist_get(struct perf_event *event) put_online_cpus(); return 0; - fail: +fail: for_each_possible_cpu(cpu) { if (cpu == failed_cpu) break; -- cgit v1.2.3 From 24cd7f54a0d47e1d5b3de29e2456bfbd2d8447b7 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Fri, 11 Jun 2010 17:32:03 +0200 Subject: perf: Reduce perf_disable() usage Since the current perf_disable() usage is only an optimization, remove it for now. This eases the removal of the __weak hw_perf_enable() interface. Signed-off-by: Peter Zijlstra Cc: paulus Cc: stephane eranian Cc: Robert Richter Cc: Will Deacon Cc: Paul Mundt Cc: Frederic Weisbecker Cc: Cyrill Gorcunov Cc: Lin Ming Cc: Yanmin Cc: Deng-Cheng Zhu Cc: David Miller Cc: Michael Cree LKML-Reference: Signed-off-by: Ingo Molnar --- kernel/perf_event.c | 37 +------------------------------------ 1 file changed, 1 insertion(+), 36 deletions(-) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index 149ca18371b7..9a98ce953561 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -478,11 +478,6 @@ static void __perf_event_remove_from_context(void *info) return; raw_spin_lock(&ctx->lock); - /* - * Protect the list operation against NMI by disabling the - * events on a global level. - */ - perf_disable(); event_sched_out(event, cpuctx, ctx); @@ -498,7 +493,6 @@ static void __perf_event_remove_from_context(void *info) perf_max_events - perf_reserved_percpu); } - perf_enable(); raw_spin_unlock(&ctx->lock); } @@ -803,12 +797,6 @@ static void __perf_install_in_context(void *info) ctx->is_active = 1; update_context_time(ctx); - /* - * Protect the list operation against NMI by disabling the - * events on a global level. NOP for non NMI based events. - */ - perf_disable(); - add_event_to_ctx(event, ctx); if (event->cpu != -1 && event->cpu != smp_processor_id()) @@ -850,8 +838,6 @@ static void __perf_install_in_context(void *info) cpuctx->max_pertask--; unlock: - perf_enable(); - raw_spin_unlock(&ctx->lock); } @@ -972,12 +958,10 @@ static void __perf_event_enable(void *info) if (!group_can_go_on(event, cpuctx, 1)) { err = -EEXIST; } else { - perf_disable(); if (event == leader) err = group_sched_in(event, cpuctx, ctx); else err = event_sched_in(event, cpuctx, ctx); - perf_enable(); } if (err) { @@ -1090,9 +1074,8 @@ static void ctx_sched_out(struct perf_event_context *ctx, goto out; update_context_time(ctx); - perf_disable(); if (!ctx->nr_active) - goto out_enable; + goto out; if (event_type & EVENT_PINNED) { list_for_each_entry(event, &ctx->pinned_groups, group_entry) @@ -1103,9 +1086,6 @@ static void ctx_sched_out(struct perf_event_context *ctx, list_for_each_entry(event, &ctx->flexible_groups, group_entry) group_sched_out(event, cpuctx, ctx); } - - out_enable: - perf_enable(); out: raw_spin_unlock(&ctx->lock); } @@ -1364,8 +1344,6 @@ ctx_sched_in(struct perf_event_context *ctx, ctx->timestamp = perf_clock(); - perf_disable(); - /* * First go through the list and put on any pinned groups * in order to give them the best chance of going on. @@ -1377,7 +1355,6 @@ ctx_sched_in(struct perf_event_context *ctx, if (event_type & EVENT_FLEXIBLE) ctx_flexible_sched_in(ctx, cpuctx); - perf_enable(); out: raw_spin_unlock(&ctx->lock); } @@ -1425,8 +1402,6 @@ void perf_event_task_sched_in(struct task_struct *task) if (cpuctx->task_ctx == ctx) return; - perf_disable(); - /* * We want to keep the following priority order: * cpu pinned (that don't need to move), task pinned, @@ -1439,8 +1414,6 @@ void perf_event_task_sched_in(struct task_struct *task) ctx_sched_in(ctx, cpuctx, EVENT_FLEXIBLE); cpuctx->task_ctx = ctx; - - perf_enable(); } #define MAX_INTERRUPTS (~0ULL) @@ -1555,11 +1528,9 @@ static void perf_adjust_period(struct perf_event *event, u64 nsec, u64 count) hwc->sample_period = sample_period; if (local64_read(&hwc->period_left) > 8*sample_period) { - perf_disable(); perf_event_stop(event); local64_set(&hwc->period_left, 0); perf_event_start(event); - perf_enable(); } } @@ -1588,15 +1559,12 @@ static void perf_ctx_adjust_freq(struct perf_event_context *ctx) */ if (interrupts == MAX_INTERRUPTS) { perf_log_throttle(event, 1); - perf_disable(); event->pmu->unthrottle(event); - perf_enable(); } if (!event->attr.freq || !event->attr.sample_freq) continue; - perf_disable(); event->pmu->read(event); now = local64_read(&event->count); delta = now - hwc->freq_count_stamp; @@ -1604,7 +1572,6 @@ static void perf_ctx_adjust_freq(struct perf_event_context *ctx) if (delta > 0) perf_adjust_period(event, TICK_NSEC, delta); - perf_enable(); } raw_spin_unlock(&ctx->lock); } @@ -1647,7 +1614,6 @@ void perf_event_task_tick(struct task_struct *curr) if (!rotate) return; - perf_disable(); cpu_ctx_sched_out(cpuctx, EVENT_FLEXIBLE); if (ctx) task_ctx_sched_out(ctx, EVENT_FLEXIBLE); @@ -1659,7 +1625,6 @@ void perf_event_task_tick(struct task_struct *curr) cpu_ctx_sched_in(cpuctx, EVENT_FLEXIBLE); if (ctx) task_ctx_sched_in(curr, EVENT_FLEXIBLE); - perf_enable(); } static int event_enable_on_exec(struct perf_event *event, -- cgit v1.2.3 From 33696fc0d141bbbcb12f75b69608ea83282e3117 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Mon, 14 Jun 2010 08:49:00 +0200 Subject: perf: Per PMU disable Changes perf_disable() into perf_pmu_disable(). Signed-off-by: Peter Zijlstra Cc: paulus Cc: stephane eranian Cc: Robert Richter Cc: Will Deacon Cc: Paul Mundt Cc: Frederic Weisbecker Cc: Cyrill Gorcunov Cc: Lin Ming Cc: Yanmin Cc: Deng-Cheng Zhu Cc: David Miller Cc: Michael Cree LKML-Reference: Signed-off-by: Ingo Molnar --- kernel/perf_event.c | 31 +++++++++++++++++++------------ 1 file changed, 19 insertions(+), 12 deletions(-) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index 9a98ce953561..5ed0c06765bb 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -71,23 +71,20 @@ static atomic64_t perf_event_id; */ static DEFINE_SPINLOCK(perf_resource_lock); -void __weak hw_perf_disable(void) { barrier(); } -void __weak hw_perf_enable(void) { barrier(); } - void __weak perf_event_print_debug(void) { } -static DEFINE_PER_CPU(int, perf_disable_count); - -void perf_disable(void) +void perf_pmu_disable(struct pmu *pmu) { - if (!__get_cpu_var(perf_disable_count)++) - hw_perf_disable(); + int *count = this_cpu_ptr(pmu->pmu_disable_count); + if (!(*count)++) + pmu->pmu_disable(pmu); } -void perf_enable(void) +void perf_pmu_enable(struct pmu *pmu) { - if (!--__get_cpu_var(perf_disable_count)) - hw_perf_enable(); + int *count = this_cpu_ptr(pmu->pmu_disable_count); + if (!--(*count)) + pmu->pmu_enable(pmu); } static void get_ctx(struct perf_event_context *ctx) @@ -4970,11 +4967,19 @@ static struct srcu_struct pmus_srcu; int perf_pmu_register(struct pmu *pmu) { + int ret; + mutex_lock(&pmus_lock); + ret = -ENOMEM; + pmu->pmu_disable_count = alloc_percpu(int); + if (!pmu->pmu_disable_count) + goto unlock; list_add_rcu(&pmu->entry, &pmus); + ret = 0; +unlock: mutex_unlock(&pmus_lock); - return 0; + return ret; } void perf_pmu_unregister(struct pmu *pmu) @@ -4984,6 +4989,8 @@ void perf_pmu_unregister(struct pmu *pmu) mutex_unlock(&pmus_lock); synchronize_srcu(&pmus_srcu); + + free_percpu(pmu->pmu_disable_count); } struct pmu *perf_init_event(struct perf_event *event) -- cgit v1.2.3 From ad5133b7030d04ce7701aa7cbe98f561347c79c2 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Tue, 15 Jun 2010 12:22:39 +0200 Subject: perf: Default PMU ops Provide default implementations for the pmu txn methods, this allows us to remove some conditional code. Signed-off-by: Peter Zijlstra Cc: paulus Cc: stephane eranian Cc: Robert Richter Cc: Will Deacon Cc: Paul Mundt Cc: Frederic Weisbecker Cc: Cyrill Gorcunov Cc: Lin Ming Cc: Yanmin Cc: Deng-Cheng Zhu Cc: David Miller Cc: Michael Cree LKML-Reference: Signed-off-by: Ingo Molnar --- kernel/perf_event.c | 64 +++++++++++++++++++++++++++++++++++++++++++---------- 1 file changed, 52 insertions(+), 12 deletions(-) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index 5ed0c06765bb..8ef4ba3bcb1f 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -674,21 +674,14 @@ group_sched_in(struct perf_event *group_event, { struct perf_event *event, *partial_group = NULL; struct pmu *pmu = group_event->pmu; - bool txn = false; if (group_event->state == PERF_EVENT_STATE_OFF) return 0; - /* Check if group transaction availabe */ - if (pmu->start_txn) - txn = true; - - if (txn) - pmu->start_txn(pmu); + pmu->start_txn(pmu); if (event_sched_in(group_event, cpuctx, ctx)) { - if (txn) - pmu->cancel_txn(pmu); + pmu->cancel_txn(pmu); return -EAGAIN; } @@ -702,7 +695,7 @@ group_sched_in(struct perf_event *group_event, } } - if (!txn || !pmu->commit_txn(pmu)) + if (!pmu->commit_txn(pmu)) return 0; group_error: @@ -717,8 +710,7 @@ group_error: } event_sched_out(group_event, cpuctx, ctx); - if (txn) - pmu->cancel_txn(pmu); + pmu->cancel_txn(pmu); return -EAGAIN; } @@ -4965,6 +4957,31 @@ static LIST_HEAD(pmus); static DEFINE_MUTEX(pmus_lock); static struct srcu_struct pmus_srcu; +static void perf_pmu_nop_void(struct pmu *pmu) +{ +} + +static int perf_pmu_nop_int(struct pmu *pmu) +{ + return 0; +} + +static void perf_pmu_start_txn(struct pmu *pmu) +{ + perf_pmu_disable(pmu); +} + +static int perf_pmu_commit_txn(struct pmu *pmu) +{ + perf_pmu_enable(pmu); + return 0; +} + +static void perf_pmu_cancel_txn(struct pmu *pmu) +{ + perf_pmu_enable(pmu); +} + int perf_pmu_register(struct pmu *pmu) { int ret; @@ -4974,6 +4991,29 @@ int perf_pmu_register(struct pmu *pmu) pmu->pmu_disable_count = alloc_percpu(int); if (!pmu->pmu_disable_count) goto unlock; + + if (!pmu->start_txn) { + if (pmu->pmu_enable) { + /* + * If we have pmu_enable/pmu_disable calls, install + * transaction stubs that use that to try and batch + * hardware accesses. + */ + pmu->start_txn = perf_pmu_start_txn; + pmu->commit_txn = perf_pmu_commit_txn; + pmu->cancel_txn = perf_pmu_cancel_txn; + } else { + pmu->start_txn = perf_pmu_nop_void; + pmu->commit_txn = perf_pmu_nop_int; + pmu->cancel_txn = perf_pmu_nop_void; + } + } + + if (!pmu->pmu_enable) { + pmu->pmu_enable = perf_pmu_nop_void; + pmu->pmu_disable = perf_pmu_nop_void; + } + list_add_rcu(&pmu->entry, &pmus); ret = 0; unlock: -- cgit v1.2.3 From fa407f35e0298d841e4088f95a7f9cf6e725c6d5 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Thu, 24 Jun 2010 12:35:12 +0200 Subject: perf: Shrink hw_perf_event Use hw_perf_event::period_left instead of hw_perf_event::remaining and win back 8 bytes. Signed-off-by: Peter Zijlstra Cc: paulus Cc: Frederic Weisbecker LKML-Reference: Signed-off-by: Ingo Molnar --- kernel/perf_event.c | 13 ++++++------- 1 file changed, 6 insertions(+), 7 deletions(-) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index 8ef4ba3bcb1f..1a6cdbf0d091 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -4800,14 +4800,13 @@ static void perf_swevent_start_hrtimer(struct perf_event *event) hrtimer_init(&hwc->hrtimer, CLOCK_MONOTONIC, HRTIMER_MODE_REL); hwc->hrtimer.function = perf_swevent_hrtimer; if (hwc->sample_period) { - u64 period; + s64 period = local64_read(&hwc->period_left); - if (hwc->remaining) { - if (hwc->remaining < 0) + if (period) { + if (period < 0) period = 10000; - else - period = hwc->remaining; - hwc->remaining = 0; + + local64_set(&hwc->period_left, 0); } else { period = max_t(u64, 10000, hwc->sample_period); } @@ -4823,7 +4822,7 @@ static void perf_swevent_cancel_hrtimer(struct perf_event *event) if (hwc->sample_period) { ktime_t remaining = hrtimer_get_remaining(&hwc->hrtimer); - hwc->remaining = ktime_to_ns(remaining); + local64_set(&hwc->period_left, ktime_to_ns(remaining)); hrtimer_cancel(&hwc->hrtimer); } -- cgit v1.2.3 From a4eaf7f14675cb512d69f0c928055e73d0c6d252 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Wed, 16 Jun 2010 14:37:10 +0200 Subject: perf: Rework the PMU methods Replace pmu::{enable,disable,start,stop,unthrottle} with pmu::{add,del,start,stop}, all of which take a flags argument. The new interface extends the capability to stop a counter while keeping it scheduled on the PMU. We replace the throttled state with the generic stopped state. This also allows us to efficiently stop/start counters over certain code paths (like IRQ handlers). It also allows scheduling a counter without it starting, allowing for a generic frozen state (useful for rotating stopped counters). The stopped state is implemented in two different ways, depending on how the architecture implemented the throttled state: 1) We disable the counter: a) the pmu has per-counter enable bits, we flip that b) we program a NOP event, preserving the counter state 2) We store the counter state and ignore all read/overflow events Signed-off-by: Peter Zijlstra Cc: paulus Cc: stephane eranian Cc: Robert Richter Cc: Will Deacon Cc: Paul Mundt Cc: Frederic Weisbecker Cc: Cyrill Gorcunov Cc: Lin Ming Cc: Yanmin Cc: Deng-Cheng Zhu Cc: David Miller Cc: Michael Cree LKML-Reference: Signed-off-by: Ingo Molnar --- kernel/hw_breakpoint.c | 29 ++++++++- kernel/perf_event.c | 140 +++++++++++++++++++++------------------- kernel/trace/trace_event_perf.c | 7 +- 3 files changed, 104 insertions(+), 72 deletions(-) (limited to 'kernel') diff --git a/kernel/hw_breakpoint.c b/kernel/hw_breakpoint.c index e9c5cfa1fd20..6f150095cafe 100644 --- a/kernel/hw_breakpoint.c +++ b/kernel/hw_breakpoint.c @@ -586,10 +586,35 @@ static int hw_breakpoint_event_init(struct perf_event *bp) return 0; } +static int hw_breakpoint_add(struct perf_event *bp, int flags) +{ + if (!(flags & PERF_EF_START)) + bp->hw.state = PERF_HES_STOPPED; + + return arch_install_hw_breakpoint(bp); +} + +static void hw_breakpoint_del(struct perf_event *bp, int flags) +{ + arch_uninstall_hw_breakpoint(bp); +} + +static void hw_breakpoint_start(struct perf_event *bp, int flags) +{ + bp->hw.state = 0; +} + +static void hw_breakpoint_stop(struct perf_event *bp, int flags) +{ + bp->hw.state = PERF_HES_STOPPED; +} + static struct pmu perf_breakpoint = { .event_init = hw_breakpoint_event_init, - .enable = arch_install_hw_breakpoint, - .disable = arch_uninstall_hw_breakpoint, + .add = hw_breakpoint_add, + .del = hw_breakpoint_del, + .start = hw_breakpoint_start, + .stop = hw_breakpoint_stop, .read = hw_breakpoint_pmu_read, }; diff --git a/kernel/perf_event.c b/kernel/perf_event.c index 1a6cdbf0d091..3bace4fd0355 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -424,7 +424,7 @@ event_sched_out(struct perf_event *event, event->state = PERF_EVENT_STATE_OFF; } event->tstamp_stopped = ctx->time; - event->pmu->disable(event); + event->pmu->del(event, 0); event->oncpu = -1; if (!is_software_event(event)) @@ -649,7 +649,7 @@ event_sched_in(struct perf_event *event, */ smp_wmb(); - if (event->pmu->enable(event)) { + if (event->pmu->add(event, PERF_EF_START)) { event->state = PERF_EVENT_STATE_INACTIVE; event->oncpu = -1; return -EAGAIN; @@ -1482,22 +1482,6 @@ do { \ return div64_u64(dividend, divisor); } -static void perf_event_stop(struct perf_event *event) -{ - if (!event->pmu->stop) - return event->pmu->disable(event); - - return event->pmu->stop(event); -} - -static int perf_event_start(struct perf_event *event) -{ - if (!event->pmu->start) - return event->pmu->enable(event); - - return event->pmu->start(event); -} - static void perf_adjust_period(struct perf_event *event, u64 nsec, u64 count) { struct hw_perf_event *hwc = &event->hw; @@ -1517,9 +1501,9 @@ static void perf_adjust_period(struct perf_event *event, u64 nsec, u64 count) hwc->sample_period = sample_period; if (local64_read(&hwc->period_left) > 8*sample_period) { - perf_event_stop(event); + event->pmu->stop(event, PERF_EF_UPDATE); local64_set(&hwc->period_left, 0); - perf_event_start(event); + event->pmu->start(event, PERF_EF_RELOAD); } } @@ -1548,7 +1532,7 @@ static void perf_ctx_adjust_freq(struct perf_event_context *ctx) */ if (interrupts == MAX_INTERRUPTS) { perf_log_throttle(event, 1); - event->pmu->unthrottle(event); + event->pmu->start(event, 0); } if (!event->attr.freq || !event->attr.sample_freq) @@ -2506,6 +2490,9 @@ int perf_event_task_disable(void) static int perf_event_index(struct perf_event *event) { + if (event->hw.state & PERF_HES_STOPPED) + return 0; + if (event->state != PERF_EVENT_STATE_ACTIVE) return 0; @@ -4120,8 +4107,6 @@ static int __perf_event_overflow(struct perf_event *event, int nmi, struct hw_perf_event *hwc = &event->hw; int ret = 0; - throttle = (throttle && event->pmu->unthrottle != NULL); - if (!throttle) { hwc->interrupts++; } else { @@ -4246,7 +4231,7 @@ static void perf_swevent_overflow(struct perf_event *event, u64 overflow, } } -static void perf_swevent_add(struct perf_event *event, u64 nr, +static void perf_swevent_event(struct perf_event *event, u64 nr, int nmi, struct perf_sample_data *data, struct pt_regs *regs) { @@ -4272,6 +4257,9 @@ static void perf_swevent_add(struct perf_event *event, u64 nr, static int perf_exclude_event(struct perf_event *event, struct pt_regs *regs) { + if (event->hw.state & PERF_HES_STOPPED) + return 0; + if (regs) { if (event->attr.exclude_user && user_mode(regs)) return 1; @@ -4371,7 +4359,7 @@ static void do_perf_sw_event(enum perf_type_id type, u32 event_id, hlist_for_each_entry_rcu(event, node, head, hlist_entry) { if (perf_swevent_match(event, type, event_id, data, regs)) - perf_swevent_add(event, nr, nmi, data, regs); + perf_swevent_event(event, nr, nmi, data, regs); } end: rcu_read_unlock(); @@ -4415,7 +4403,7 @@ static void perf_swevent_read(struct perf_event *event) { } -static int perf_swevent_enable(struct perf_event *event) +static int perf_swevent_add(struct perf_event *event, int flags) { struct hw_perf_event *hwc = &event->hw; struct perf_cpu_context *cpuctx; @@ -4428,6 +4416,8 @@ static int perf_swevent_enable(struct perf_event *event) perf_swevent_set_period(event); } + hwc->state = !(flags & PERF_EF_START); + head = find_swevent_head(cpuctx, event); if (WARN_ON_ONCE(!head)) return -EINVAL; @@ -4437,18 +4427,19 @@ static int perf_swevent_enable(struct perf_event *event) return 0; } -static void perf_swevent_disable(struct perf_event *event) +static void perf_swevent_del(struct perf_event *event, int flags) { hlist_del_rcu(&event->hlist_entry); } -static void perf_swevent_void(struct perf_event *event) +static void perf_swevent_start(struct perf_event *event, int flags) { + event->hw.state = 0; } -static int perf_swevent_int(struct perf_event *event) +static void perf_swevent_stop(struct perf_event *event, int flags) { - return 0; + event->hw.state = PERF_HES_STOPPED; } /* Deref the hlist from the update side */ @@ -4604,12 +4595,11 @@ static int perf_swevent_init(struct perf_event *event) static struct pmu perf_swevent = { .event_init = perf_swevent_init, - .enable = perf_swevent_enable, - .disable = perf_swevent_disable, - .start = perf_swevent_int, - .stop = perf_swevent_void, + .add = perf_swevent_add, + .del = perf_swevent_del, + .start = perf_swevent_start, + .stop = perf_swevent_stop, .read = perf_swevent_read, - .unthrottle = perf_swevent_void, /* hwc->interrupts already reset */ }; #ifdef CONFIG_EVENT_TRACING @@ -4657,7 +4647,7 @@ void perf_tp_event(u64 addr, u64 count, void *record, int entry_size, hlist_for_each_entry_rcu(event, node, head, hlist_entry) { if (perf_tp_event_match(event, &data, regs)) - perf_swevent_add(event, count, 1, &data, regs); + perf_swevent_event(event, count, 1, &data, regs); } perf_swevent_put_recursion_context(rctx); @@ -4696,12 +4686,11 @@ static int perf_tp_event_init(struct perf_event *event) static struct pmu perf_tracepoint = { .event_init = perf_tp_event_init, - .enable = perf_trace_enable, - .disable = perf_trace_disable, - .start = perf_swevent_int, - .stop = perf_swevent_void, + .add = perf_trace_add, + .del = perf_trace_del, + .start = perf_swevent_start, + .stop = perf_swevent_stop, .read = perf_swevent_read, - .unthrottle = perf_swevent_void, }; static inline void perf_tp_register(void) @@ -4757,8 +4746,8 @@ void perf_bp_event(struct perf_event *bp, void *data) perf_sample_data_init(&sample, bp->attr.bp_addr); - if (!perf_exclude_event(bp, regs)) - perf_swevent_add(bp, 1, 1, &sample, regs); + if (!bp->hw.state && !perf_exclude_event(bp, regs)) + perf_swevent_event(bp, 1, 1, &sample, regs); } #endif @@ -4834,32 +4823,39 @@ static void perf_swevent_cancel_hrtimer(struct perf_event *event) static void cpu_clock_event_update(struct perf_event *event) { - int cpu = raw_smp_processor_id(); s64 prev; u64 now; - now = cpu_clock(cpu); + now = local_clock(); prev = local64_xchg(&event->hw.prev_count, now); local64_add(now - prev, &event->count); } -static int cpu_clock_event_enable(struct perf_event *event) +static void cpu_clock_event_start(struct perf_event *event, int flags) { - struct hw_perf_event *hwc = &event->hw; - int cpu = raw_smp_processor_id(); - - local64_set(&hwc->prev_count, cpu_clock(cpu)); + local64_set(&event->hw.prev_count, local_clock()); perf_swevent_start_hrtimer(event); - - return 0; } -static void cpu_clock_event_disable(struct perf_event *event) +static void cpu_clock_event_stop(struct perf_event *event, int flags) { perf_swevent_cancel_hrtimer(event); cpu_clock_event_update(event); } +static int cpu_clock_event_add(struct perf_event *event, int flags) +{ + if (flags & PERF_EF_START) + cpu_clock_event_start(event, flags); + + return 0; +} + +static void cpu_clock_event_del(struct perf_event *event, int flags) +{ + cpu_clock_event_stop(event, flags); +} + static void cpu_clock_event_read(struct perf_event *event) { cpu_clock_event_update(event); @@ -4878,8 +4874,10 @@ static int cpu_clock_event_init(struct perf_event *event) static struct pmu perf_cpu_clock = { .event_init = cpu_clock_event_init, - .enable = cpu_clock_event_enable, - .disable = cpu_clock_event_disable, + .add = cpu_clock_event_add, + .del = cpu_clock_event_del, + .start = cpu_clock_event_start, + .stop = cpu_clock_event_stop, .read = cpu_clock_event_read, }; @@ -4897,25 +4895,29 @@ static void task_clock_event_update(struct perf_event *event, u64 now) local64_add(delta, &event->count); } -static int task_clock_event_enable(struct perf_event *event) +static void task_clock_event_start(struct perf_event *event, int flags) { - struct hw_perf_event *hwc = &event->hw; - u64 now; - - now = event->ctx->time; - - local64_set(&hwc->prev_count, now); - + local64_set(&event->hw.prev_count, event->ctx->time); perf_swevent_start_hrtimer(event); - - return 0; } -static void task_clock_event_disable(struct perf_event *event) +static void task_clock_event_stop(struct perf_event *event, int flags) { perf_swevent_cancel_hrtimer(event); task_clock_event_update(event, event->ctx->time); +} + +static int task_clock_event_add(struct perf_event *event, int flags) +{ + if (flags & PERF_EF_START) + task_clock_event_start(event, flags); + return 0; +} + +static void task_clock_event_del(struct perf_event *event, int flags) +{ + task_clock_event_stop(event, PERF_EF_UPDATE); } static void task_clock_event_read(struct perf_event *event) @@ -4947,8 +4949,10 @@ static int task_clock_event_init(struct perf_event *event) static struct pmu perf_task_clock = { .event_init = task_clock_event_init, - .enable = task_clock_event_enable, - .disable = task_clock_event_disable, + .add = task_clock_event_add, + .del = task_clock_event_del, + .start = task_clock_event_start, + .stop = task_clock_event_stop, .read = task_clock_event_read, }; diff --git a/kernel/trace/trace_event_perf.c b/kernel/trace/trace_event_perf.c index f3bbcd1c90c8..39c059ca670e 100644 --- a/kernel/trace/trace_event_perf.c +++ b/kernel/trace/trace_event_perf.c @@ -101,7 +101,7 @@ int perf_trace_init(struct perf_event *p_event) return ret; } -int perf_trace_enable(struct perf_event *p_event) +int perf_trace_add(struct perf_event *p_event, int flags) { struct ftrace_event_call *tp_event = p_event->tp_event; struct hlist_head __percpu *pcpu_list; @@ -111,13 +111,16 @@ int perf_trace_enable(struct perf_event *p_event) if (WARN_ON_ONCE(!pcpu_list)) return -EINVAL; + if (!(flags & PERF_EF_START)) + p_event->hw.state = PERF_HES_STOPPED; + list = this_cpu_ptr(pcpu_list); hlist_add_head_rcu(&p_event->hlist_entry, list); return 0; } -void perf_trace_disable(struct perf_event *p_event) +void perf_trace_del(struct perf_event *p_event, int flags) { hlist_del_rcu(&p_event->hlist_entry); } -- cgit v1.2.3 From 15ac9a395a753cb28c674e7ea80386ffdff21785 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Mon, 6 Sep 2010 15:51:45 +0200 Subject: perf: Remove the sysfs bits Neither the overcommit nor the reservation sysfs parameter were actually working, remove them as they'll only get in the way. Signed-off-by: Peter Zijlstra Cc: paulus LKML-Reference: Signed-off-by: Ingo Molnar --- kernel/perf_event.c | 124 ---------------------------------------------------- 1 file changed, 124 deletions(-) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index 3bace4fd0355..8462e69409ae 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -39,10 +39,6 @@ */ static DEFINE_PER_CPU(struct perf_cpu_context, perf_cpu_context); -int perf_max_events __read_mostly = 1; -static int perf_reserved_percpu __read_mostly; -static int perf_overcommit __read_mostly = 1; - static atomic_t nr_events __read_mostly; static atomic_t nr_mmap_events __read_mostly; static atomic_t nr_comm_events __read_mostly; @@ -66,11 +62,6 @@ int sysctl_perf_event_sample_rate __read_mostly = 100000; static atomic64_t perf_event_id; -/* - * Lock for (sysadmin-configurable) event reservations: - */ -static DEFINE_SPINLOCK(perf_resource_lock); - void __weak perf_event_print_debug(void) { } void perf_pmu_disable(struct pmu *pmu) @@ -480,16 +471,6 @@ static void __perf_event_remove_from_context(void *info) list_del_event(event, ctx); - if (!ctx->task) { - /* - * Allow more per task events with respect to the - * reservation: - */ - cpuctx->max_pertask = - min(perf_max_events - ctx->nr_events, - perf_max_events - perf_reserved_percpu); - } - raw_spin_unlock(&ctx->lock); } @@ -823,9 +804,6 @@ static void __perf_install_in_context(void *info) } } - if (!err && !ctx->task && cpuctx->max_pertask) - cpuctx->max_pertask--; - unlock: raw_spin_unlock(&ctx->lock); } @@ -5930,10 +5908,6 @@ static void __cpuinit perf_event_init_cpu(int cpu) cpuctx = &per_cpu(perf_cpu_context, cpu); - spin_lock(&perf_resource_lock); - cpuctx->max_pertask = perf_max_events - perf_reserved_percpu; - spin_unlock(&perf_resource_lock); - mutex_lock(&cpuctx->hlist_mutex); if (cpuctx->hlist_refcount > 0) { struct swevent_hlist *hlist; @@ -6008,101 +5982,3 @@ void __init perf_event_init(void) perf_tp_register(); perf_cpu_notifier(perf_cpu_notify); } - -static ssize_t perf_show_reserve_percpu(struct sysdev_class *class, - struct sysdev_class_attribute *attr, - char *buf) -{ - return sprintf(buf, "%d\n", perf_reserved_percpu); -} - -static ssize_t -perf_set_reserve_percpu(struct sysdev_class *class, - struct sysdev_class_attribute *attr, - const char *buf, - size_t count) -{ - struct perf_cpu_context *cpuctx; - unsigned long val; - int err, cpu, mpt; - - err = strict_strtoul(buf, 10, &val); - if (err) - return err; - if (val > perf_max_events) - return -EINVAL; - - spin_lock(&perf_resource_lock); - perf_reserved_percpu = val; - for_each_online_cpu(cpu) { - cpuctx = &per_cpu(perf_cpu_context, cpu); - raw_spin_lock_irq(&cpuctx->ctx.lock); - mpt = min(perf_max_events - cpuctx->ctx.nr_events, - perf_max_events - perf_reserved_percpu); - cpuctx->max_pertask = mpt; - raw_spin_unlock_irq(&cpuctx->ctx.lock); - } - spin_unlock(&perf_resource_lock); - - return count; -} - -static ssize_t perf_show_overcommit(struct sysdev_class *class, - struct sysdev_class_attribute *attr, - char *buf) -{ - return sprintf(buf, "%d\n", perf_overcommit); -} - -static ssize_t -perf_set_overcommit(struct sysdev_class *class, - struct sysdev_class_attribute *attr, - const char *buf, size_t count) -{ - unsigned long val; - int err; - - err = strict_strtoul(buf, 10, &val); - if (err) - return err; - if (val > 1) - return -EINVAL; - - spin_lock(&perf_resource_lock); - perf_overcommit = val; - spin_unlock(&perf_resource_lock); - - return count; -} - -static SYSDEV_CLASS_ATTR( - reserve_percpu, - 0644, - perf_show_reserve_percpu, - perf_set_reserve_percpu - ); - -static SYSDEV_CLASS_ATTR( - overcommit, - 0644, - perf_show_overcommit, - perf_set_overcommit - ); - -static struct attribute *perfclass_attrs[] = { - &attr_reserve_percpu.attr, - &attr_overcommit.attr, - NULL -}; - -static struct attribute_group perfclass_attr_group = { - .attrs = perfclass_attrs, - .name = "perf_events", -}; - -static int __init perf_event_sysfs_init(void) -{ - return sysfs_create_group(&cpu_sysdev_class.kset.kobj, - &perfclass_attr_group); -} -device_initcall(perf_event_sysfs_init); -- cgit v1.2.3 From c3f00c70276d8ae82578c8b773e2db657f69a478 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Wed, 18 Aug 2010 14:37:15 +0200 Subject: perf: Separate find_get_context() from event initialization Separate find_get_context() from the event allocation and initialization so that we may make find_get_context() depend on the event pmu in a later patch. Signed-off-by: Peter Zijlstra Cc: paulus Cc: stephane eranian Cc: Robert Richter Cc: Frederic Weisbecker Cc: Lin Ming Cc: Yanmin LKML-Reference: Signed-off-by: Ingo Molnar --- kernel/perf_event.c | 73 +++++++++++++++++++++++++---------------------------- 1 file changed, 35 insertions(+), 38 deletions(-) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index 8462e69409ae..a3c86a8335c4 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -827,6 +827,8 @@ perf_install_in_context(struct perf_event_context *ctx, { struct task_struct *task = ctx->task; + event->ctx = ctx; + if (!task) { /* * Per cpu events are installed via an smp call and @@ -5038,20 +5040,17 @@ struct pmu *perf_init_event(struct perf_event *event) * Allocate and initialize a event structure */ static struct perf_event * -perf_event_alloc(struct perf_event_attr *attr, - int cpu, - struct perf_event_context *ctx, +perf_event_alloc(struct perf_event_attr *attr, int cpu, struct perf_event *group_leader, struct perf_event *parent_event, - perf_overflow_handler_t overflow_handler, - gfp_t gfpflags) + perf_overflow_handler_t overflow_handler) { struct pmu *pmu; struct perf_event *event; struct hw_perf_event *hwc; long err; - event = kzalloc(sizeof(*event), gfpflags); + event = kzalloc(sizeof(*event), GFP_KERNEL); if (!event) return ERR_PTR(-ENOMEM); @@ -5076,7 +5075,6 @@ perf_event_alloc(struct perf_event_attr *attr, event->attr = *attr; event->group_leader = group_leader; event->pmu = NULL; - event->ctx = ctx; event->oncpu = -1; event->parent = parent_event; @@ -5321,20 +5319,26 @@ SYSCALL_DEFINE5(perf_event_open, if (event_fd < 0) return event_fd; + event = perf_event_alloc(&attr, cpu, group_leader, NULL, NULL); + if (IS_ERR(event)) { + err = PTR_ERR(event); + goto err_fd; + } + /* * Get the target context (task or percpu): */ ctx = find_get_context(pid, cpu); if (IS_ERR(ctx)) { err = PTR_ERR(ctx); - goto err_fd; + goto err_alloc; } if (group_fd != -1) { group_leader = perf_fget_light(group_fd, &fput_needed); if (IS_ERR(group_leader)) { err = PTR_ERR(group_leader); - goto err_put_context; + goto err_context; } group_file = group_leader->filp; if (flags & PERF_FLAG_FD_OUTPUT) @@ -5354,37 +5358,30 @@ SYSCALL_DEFINE5(perf_event_open, * becoming part of another group-sibling): */ if (group_leader->group_leader != group_leader) - goto err_put_context; + goto err_context; /* * Do not allow to attach to a group in a different * task or CPU context: */ if (group_leader->ctx != ctx) - goto err_put_context; + goto err_context; /* * Only a group leader can be exclusive or pinned */ if (attr.exclusive || attr.pinned) - goto err_put_context; - } - - event = perf_event_alloc(&attr, cpu, ctx, group_leader, - NULL, NULL, GFP_KERNEL); - if (IS_ERR(event)) { - err = PTR_ERR(event); - goto err_put_context; + goto err_context; } if (output_event) { err = perf_event_set_output(event, output_event); if (err) - goto err_free_put_context; + goto err_context; } event_file = anon_inode_getfile("[perf_event]", &perf_fops, event, O_RDWR); if (IS_ERR(event_file)) { err = PTR_ERR(event_file); - goto err_free_put_context; + goto err_context; } event->filp = event_file; @@ -5410,11 +5407,11 @@ SYSCALL_DEFINE5(perf_event_open, fd_install(event_fd, event_file); return event_fd; -err_free_put_context: - free_event(event); -err_put_context: +err_context: fput_light(group_file, fput_needed); put_ctx(ctx); +err_alloc: + free_event(event); err_fd: put_unused_fd(event_fd); return err; @@ -5432,25 +5429,24 @@ perf_event_create_kernel_counter(struct perf_event_attr *attr, int cpu, pid_t pid, perf_overflow_handler_t overflow_handler) { - struct perf_event *event; struct perf_event_context *ctx; + struct perf_event *event; int err; /* * Get the target context (task or percpu): */ + event = perf_event_alloc(attr, cpu, NULL, NULL, overflow_handler); + if (IS_ERR(event)) { + err = PTR_ERR(event); + goto err; + } + ctx = find_get_context(pid, cpu); if (IS_ERR(ctx)) { err = PTR_ERR(ctx); - goto err_exit; - } - - event = perf_event_alloc(attr, cpu, ctx, NULL, - NULL, overflow_handler, GFP_KERNEL); - if (IS_ERR(event)) { - err = PTR_ERR(event); - goto err_put_context; + goto err_free; } event->filp = NULL; @@ -5468,9 +5464,9 @@ perf_event_create_kernel_counter(struct perf_event_attr *attr, int cpu, return event; - err_put_context: - put_ctx(ctx); - err_exit: +err_free: + free_event(event); +err: return ERR_PTR(err); } EXPORT_SYMBOL_GPL(perf_event_create_kernel_counter); @@ -5498,9 +5494,9 @@ inherit_event(struct perf_event *parent_event, parent_event = parent_event->parent; child_event = perf_event_alloc(&parent_event->attr, - parent_event->cpu, child_ctx, + parent_event->cpu, group_leader, parent_event, - NULL, GFP_KERNEL); + NULL); if (IS_ERR(child_event)) return child_event; get_ctx(child_ctx); @@ -5525,6 +5521,7 @@ inherit_event(struct perf_event *parent_event, local64_set(&hwc->period_left, sample_period); } + child_event->ctx = child_ctx; child_event->overflow_handler = parent_event->overflow_handler; /* -- cgit v1.2.3 From b28ab83c595e767f2028276b7398d17f2253cec0 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Mon, 6 Sep 2010 14:48:15 +0200 Subject: perf: Remove the swevent hash-table from the cpu context Separate the swevent hash-table from the cpu_context bits in preparation for per pmu cpu contexts. This keeps the swevent hash a global entity. Signed-off-by: Peter Zijlstra Cc: paulus Cc: stephane eranian Cc: Robert Richter Cc: Frederic Weisbecker Cc: Lin Ming Cc: Yanmin LKML-Reference: Signed-off-by: Ingo Molnar --- kernel/perf_event.c | 104 +++++++++++++++++++++++++++++----------------------- 1 file changed, 58 insertions(+), 46 deletions(-) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index a3c86a8335c4..2c47ed6c4f26 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -4154,6 +4154,17 @@ int perf_event_overflow(struct perf_event *event, int nmi, * Generic software event infrastructure */ +struct swevent_htable { + struct swevent_hlist *swevent_hlist; + struct mutex hlist_mutex; + int hlist_refcount; + + /* Recursion avoidance in each contexts */ + int recursion[PERF_NR_CONTEXTS]; +}; + +static DEFINE_PER_CPU(struct swevent_htable, swevent_htable); + /* * We directly increment event->count and keep a second value in * event->hw.period_left to count intervals. This period event @@ -4286,11 +4297,11 @@ __find_swevent_head(struct swevent_hlist *hlist, u64 type, u32 event_id) /* For the read side: events when they trigger */ static inline struct hlist_head * -find_swevent_head_rcu(struct perf_cpu_context *ctx, u64 type, u32 event_id) +find_swevent_head_rcu(struct swevent_htable *swhash, u64 type, u32 event_id) { struct swevent_hlist *hlist; - hlist = rcu_dereference(ctx->swevent_hlist); + hlist = rcu_dereference(swhash->swevent_hlist); if (!hlist) return NULL; @@ -4299,7 +4310,7 @@ find_swevent_head_rcu(struct perf_cpu_context *ctx, u64 type, u32 event_id) /* For the event head insertion and removal in the hlist */ static inline struct hlist_head * -find_swevent_head(struct perf_cpu_context *ctx, struct perf_event *event) +find_swevent_head(struct swevent_htable *swhash, struct perf_event *event) { struct swevent_hlist *hlist; u32 event_id = event->attr.config; @@ -4310,7 +4321,7 @@ find_swevent_head(struct perf_cpu_context *ctx, struct perf_event *event) * and release. Which makes the protected version suitable here. * The context lock guarantees that. */ - hlist = rcu_dereference_protected(ctx->swevent_hlist, + hlist = rcu_dereference_protected(swhash->swevent_hlist, lockdep_is_held(&event->ctx->lock)); if (!hlist) return NULL; @@ -4323,17 +4334,13 @@ static void do_perf_sw_event(enum perf_type_id type, u32 event_id, struct perf_sample_data *data, struct pt_regs *regs) { - struct perf_cpu_context *cpuctx; + struct swevent_htable *swhash = &__get_cpu_var(swevent_htable); struct perf_event *event; struct hlist_node *node; struct hlist_head *head; - cpuctx = &__get_cpu_var(perf_cpu_context); - rcu_read_lock(); - - head = find_swevent_head_rcu(cpuctx, type, event_id); - + head = find_swevent_head_rcu(swhash, type, event_id); if (!head) goto end; @@ -4347,17 +4354,17 @@ end: int perf_swevent_get_recursion_context(void) { - struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context); + struct swevent_htable *swhash = &__get_cpu_var(swevent_htable); - return get_recursion_context(cpuctx->recursion); + return get_recursion_context(swhash->recursion); } EXPORT_SYMBOL_GPL(perf_swevent_get_recursion_context); void inline perf_swevent_put_recursion_context(int rctx) { - struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context); + struct swevent_htable *swhash = &__get_cpu_var(swevent_htable); - put_recursion_context(cpuctx->recursion, rctx); + put_recursion_context(swhash->recursion, rctx); } void __perf_sw_event(u32 event_id, u64 nr, int nmi, @@ -4385,12 +4392,10 @@ static void perf_swevent_read(struct perf_event *event) static int perf_swevent_add(struct perf_event *event, int flags) { + struct swevent_htable *swhash = &__get_cpu_var(swevent_htable); struct hw_perf_event *hwc = &event->hw; - struct perf_cpu_context *cpuctx; struct hlist_head *head; - cpuctx = &__get_cpu_var(perf_cpu_context); - if (hwc->sample_period) { hwc->last_period = hwc->sample_period; perf_swevent_set_period(event); @@ -4398,7 +4403,7 @@ static int perf_swevent_add(struct perf_event *event, int flags) hwc->state = !(flags & PERF_EF_START); - head = find_swevent_head(cpuctx, event); + head = find_swevent_head(swhash, event); if (WARN_ON_ONCE(!head)) return -EINVAL; @@ -4424,10 +4429,10 @@ static void perf_swevent_stop(struct perf_event *event, int flags) /* Deref the hlist from the update side */ static inline struct swevent_hlist * -swevent_hlist_deref(struct perf_cpu_context *cpuctx) +swevent_hlist_deref(struct swevent_htable *swhash) { - return rcu_dereference_protected(cpuctx->swevent_hlist, - lockdep_is_held(&cpuctx->hlist_mutex)); + return rcu_dereference_protected(swhash->swevent_hlist, + lockdep_is_held(&swhash->hlist_mutex)); } static void swevent_hlist_release_rcu(struct rcu_head *rcu_head) @@ -4438,27 +4443,27 @@ static void swevent_hlist_release_rcu(struct rcu_head *rcu_head) kfree(hlist); } -static void swevent_hlist_release(struct perf_cpu_context *cpuctx) +static void swevent_hlist_release(struct swevent_htable *swhash) { - struct swevent_hlist *hlist = swevent_hlist_deref(cpuctx); + struct swevent_hlist *hlist = swevent_hlist_deref(swhash); if (!hlist) return; - rcu_assign_pointer(cpuctx->swevent_hlist, NULL); + rcu_assign_pointer(swhash->swevent_hlist, NULL); call_rcu(&hlist->rcu_head, swevent_hlist_release_rcu); } static void swevent_hlist_put_cpu(struct perf_event *event, int cpu) { - struct perf_cpu_context *cpuctx = &per_cpu(perf_cpu_context, cpu); + struct swevent_htable *swhash = &per_cpu(swevent_htable, cpu); - mutex_lock(&cpuctx->hlist_mutex); + mutex_lock(&swhash->hlist_mutex); - if (!--cpuctx->hlist_refcount) - swevent_hlist_release(cpuctx); + if (!--swhash->hlist_refcount) + swevent_hlist_release(swhash); - mutex_unlock(&cpuctx->hlist_mutex); + mutex_unlock(&swhash->hlist_mutex); } static void swevent_hlist_put(struct perf_event *event) @@ -4476,12 +4481,12 @@ static void swevent_hlist_put(struct perf_event *event) static int swevent_hlist_get_cpu(struct perf_event *event, int cpu) { - struct perf_cpu_context *cpuctx = &per_cpu(perf_cpu_context, cpu); + struct swevent_htable *swhash = &per_cpu(swevent_htable, cpu); int err = 0; - mutex_lock(&cpuctx->hlist_mutex); + mutex_lock(&swhash->hlist_mutex); - if (!swevent_hlist_deref(cpuctx) && cpu_online(cpu)) { + if (!swevent_hlist_deref(swhash) && cpu_online(cpu)) { struct swevent_hlist *hlist; hlist = kzalloc(sizeof(*hlist), GFP_KERNEL); @@ -4489,11 +4494,11 @@ static int swevent_hlist_get_cpu(struct perf_event *event, int cpu) err = -ENOMEM; goto exit; } - rcu_assign_pointer(cpuctx->swevent_hlist, hlist); + rcu_assign_pointer(swhash->swevent_hlist, hlist); } - cpuctx->hlist_refcount++; + swhash->hlist_refcount++; exit: - mutex_unlock(&cpuctx->hlist_mutex); + mutex_unlock(&swhash->hlist_mutex); return err; } @@ -5889,12 +5894,15 @@ int perf_event_init_task(struct task_struct *child) static void __init perf_event_init_all_cpus(void) { - int cpu; struct perf_cpu_context *cpuctx; + struct swevent_htable *swhash; + int cpu; for_each_possible_cpu(cpu) { + swhash = &per_cpu(swevent_htable, cpu); + mutex_init(&swhash->hlist_mutex); + cpuctx = &per_cpu(perf_cpu_context, cpu); - mutex_init(&cpuctx->hlist_mutex); __perf_event_init_context(&cpuctx->ctx, NULL); } } @@ -5902,18 +5910,21 @@ static void __init perf_event_init_all_cpus(void) static void __cpuinit perf_event_init_cpu(int cpu) { struct perf_cpu_context *cpuctx; + struct swevent_htable *swhash; cpuctx = &per_cpu(perf_cpu_context, cpu); - mutex_lock(&cpuctx->hlist_mutex); - if (cpuctx->hlist_refcount > 0) { + swhash = &per_cpu(swevent_htable, cpu); + + mutex_lock(&swhash->hlist_mutex); + if (swhash->hlist_refcount > 0) { struct swevent_hlist *hlist; - hlist = kzalloc(sizeof(*hlist), GFP_KERNEL); - WARN_ON_ONCE(!hlist); - rcu_assign_pointer(cpuctx->swevent_hlist, hlist); + hlist = kzalloc_node(sizeof(*hlist), GFP_KERNEL, cpu_to_node(cpu)); + WARN_ON(!hlist); + rcu_assign_pointer(swhash->swevent_hlist, hlist); } - mutex_unlock(&cpuctx->hlist_mutex); + mutex_unlock(&swhash->hlist_mutex); } #ifdef CONFIG_HOTPLUG_CPU @@ -5931,11 +5942,12 @@ static void __perf_event_exit_cpu(void *info) static void perf_event_exit_cpu(int cpu) { struct perf_cpu_context *cpuctx = &per_cpu(perf_cpu_context, cpu); + struct swevent_htable *swhash = &per_cpu(swevent_htable, cpu); struct perf_event_context *ctx = &cpuctx->ctx; - mutex_lock(&cpuctx->hlist_mutex); - swevent_hlist_release(cpuctx); - mutex_unlock(&cpuctx->hlist_mutex); + mutex_lock(&swhash->hlist_mutex); + swevent_hlist_release(swhash); + mutex_unlock(&swhash->hlist_mutex); mutex_lock(&ctx->mutex); smp_call_function_single(cpu, __perf_event_exit_cpu, NULL, 1); -- cgit v1.2.3 From b5ab4cd563e7ab49b27957704112a8ecade54e1f Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Mon, 6 Sep 2010 16:32:21 +0200 Subject: perf: Per cpu-context rotation timer Give each cpu-context its own timer so that it is a self contained entity, this eases the way for per-pmu-per-cpu contexts as well as provides the basic infrastructure to allow different rotation times per pmu. Things to look at: - folding the tick and these TICK_NSEC timers - separate task context rotation Signed-off-by: Peter Zijlstra Cc: paulus Cc: stephane eranian Cc: Robert Richter Cc: Frederic Weisbecker Cc: Lin Ming Cc: Yanmin LKML-Reference: Signed-off-by: Ingo Molnar --- kernel/perf_event.c | 80 +++++++++++++++++++++++++++++++++++++++++------------ kernel/sched.c | 2 -- 2 files changed, 63 insertions(+), 19 deletions(-) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index 2c47ed6c4f26..d75e4c8727f9 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -78,6 +78,25 @@ void perf_pmu_enable(struct pmu *pmu) pmu->pmu_enable(pmu); } +static void perf_pmu_rotate_start(void) +{ + struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context); + + if (hrtimer_active(&cpuctx->timer)) + return; + + __hrtimer_start_range_ns(&cpuctx->timer, + ns_to_ktime(cpuctx->timer_interval), 0, + HRTIMER_MODE_REL_PINNED, 0); +} + +static void perf_pmu_rotate_stop(void) +{ + struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context); + + hrtimer_cancel(&cpuctx->timer); +} + static void get_ctx(struct perf_event_context *ctx) { WARN_ON(!atomic_inc_not_zero(&ctx->refcount)); @@ -281,6 +300,8 @@ list_add_event(struct perf_event *event, struct perf_event_context *ctx) } list_add_rcu(&event->event_entry, &ctx->event_list); + if (!ctx->nr_events) + perf_pmu_rotate_start(); ctx->nr_events++; if (event->attr.inherit_stat) ctx->nr_stat++; @@ -1383,6 +1404,12 @@ void perf_event_task_sched_in(struct task_struct *task) ctx_sched_in(ctx, cpuctx, EVENT_FLEXIBLE); cpuctx->task_ctx = ctx; + + /* + * Since these rotations are per-cpu, we need to ensure the + * cpu-context we got scheduled on is actually rotating. + */ + perf_pmu_rotate_start(); } #define MAX_INTERRUPTS (~0ULL) @@ -1487,7 +1514,7 @@ static void perf_adjust_period(struct perf_event *event, u64 nsec, u64 count) } } -static void perf_ctx_adjust_freq(struct perf_event_context *ctx) +static void perf_ctx_adjust_freq(struct perf_event_context *ctx, u64 period) { struct perf_event *event; struct hw_perf_event *hwc; @@ -1524,7 +1551,7 @@ static void perf_ctx_adjust_freq(struct perf_event_context *ctx) hwc->freq_count_stamp = now; if (delta > 0) - perf_adjust_period(event, TICK_NSEC, delta); + perf_adjust_period(event, period, delta); } raw_spin_unlock(&ctx->lock); } @@ -1542,30 +1569,39 @@ static void rotate_ctx(struct perf_event_context *ctx) raw_spin_unlock(&ctx->lock); } -void perf_event_task_tick(struct task_struct *curr) +/* + * Cannot race with ->pmu_rotate_start() because this is ran from hardirq + * context, and ->pmu_rotate_start() is called with irqs disabled (both are + * cpu affine, so there are no SMP races). + */ +static enum hrtimer_restart perf_event_context_tick(struct hrtimer *timer) { + enum hrtimer_restart restart = HRTIMER_NORESTART; struct perf_cpu_context *cpuctx; struct perf_event_context *ctx; int rotate = 0; - if (!atomic_read(&nr_events)) - return; + cpuctx = container_of(timer, struct perf_cpu_context, timer); - cpuctx = &__get_cpu_var(perf_cpu_context); - if (cpuctx->ctx.nr_events && - cpuctx->ctx.nr_events != cpuctx->ctx.nr_active) - rotate = 1; + if (cpuctx->ctx.nr_events) { + restart = HRTIMER_RESTART; + if (cpuctx->ctx.nr_events != cpuctx->ctx.nr_active) + rotate = 1; + } - ctx = curr->perf_event_ctxp; - if (ctx && ctx->nr_events && ctx->nr_events != ctx->nr_active) - rotate = 1; + ctx = current->perf_event_ctxp; + if (ctx && ctx->nr_events) { + restart = HRTIMER_RESTART; + if (ctx->nr_events != ctx->nr_active) + rotate = 1; + } - perf_ctx_adjust_freq(&cpuctx->ctx); + perf_ctx_adjust_freq(&cpuctx->ctx, cpuctx->timer_interval); if (ctx) - perf_ctx_adjust_freq(ctx); + perf_ctx_adjust_freq(ctx, cpuctx->timer_interval); if (!rotate) - return; + goto done; cpu_ctx_sched_out(cpuctx, EVENT_FLEXIBLE); if (ctx) @@ -1577,7 +1613,12 @@ void perf_event_task_tick(struct task_struct *curr) cpu_ctx_sched_in(cpuctx, EVENT_FLEXIBLE); if (ctx) - task_ctx_sched_in(curr, EVENT_FLEXIBLE); + task_ctx_sched_in(current, EVENT_FLEXIBLE); + +done: + hrtimer_forward_now(timer, ns_to_ktime(cpuctx->timer_interval)); + + return restart; } static int event_enable_on_exec(struct perf_event *event, @@ -4786,7 +4827,7 @@ static void perf_swevent_start_hrtimer(struct perf_event *event) } __hrtimer_start_range_ns(&hwc->hrtimer, ns_to_ktime(period), 0, - HRTIMER_MODE_REL, 0); + HRTIMER_MODE_REL_PINNED, 0); } } @@ -5904,6 +5945,9 @@ static void __init perf_event_init_all_cpus(void) cpuctx = &per_cpu(perf_cpu_context, cpu); __perf_event_init_context(&cpuctx->ctx, NULL); + cpuctx->timer_interval = TICK_NSEC; + hrtimer_init(&cpuctx->timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL); + cpuctx->timer.function = perf_event_context_tick; } } @@ -5934,6 +5978,8 @@ static void __perf_event_exit_cpu(void *info) struct perf_event_context *ctx = &cpuctx->ctx; struct perf_event *event, *tmp; + perf_pmu_rotate_stop(); + list_for_each_entry_safe(event, tmp, &ctx->pinned_groups, group_entry) __perf_event_remove_from_context(event); list_for_each_entry_safe(event, tmp, &ctx->flexible_groups, group_entry) diff --git a/kernel/sched.c b/kernel/sched.c index 09b574e7f4df..66a02ba83c01 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -3578,8 +3578,6 @@ void scheduler_tick(void) curr->sched_class->task_tick(rq, curr, 0); raw_spin_unlock(&rq->lock); - perf_event_task_tick(curr); - #ifdef CONFIG_SMP rq->idle_at_tick = idle_cpu(cpu); trigger_load_balance(rq, cpu); -- cgit v1.2.3 From 108b02cfce04ee90b0a07ee0b104baffd39f5934 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Mon, 6 Sep 2010 14:32:03 +0200 Subject: perf: Per-pmu-per-cpu contexts Allocate per-cpu contexts per pmu. Signed-off-by: Peter Zijlstra Cc: paulus Cc: stephane eranian Cc: Robert Richter Cc: Frederic Weisbecker Cc: Lin Ming Cc: Yanmin LKML-Reference: Signed-off-by: Ingo Molnar --- kernel/perf_event.c | 178 ++++++++++++++++++++++++++++++++-------------------- 1 file changed, 110 insertions(+), 68 deletions(-) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index d75e4c8727f9..8ca6e690ffe3 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -34,16 +34,15 @@ #include -/* - * Each CPU has a list of per CPU events: - */ -static DEFINE_PER_CPU(struct perf_cpu_context, perf_cpu_context); - static atomic_t nr_events __read_mostly; static atomic_t nr_mmap_events __read_mostly; static atomic_t nr_comm_events __read_mostly; static atomic_t nr_task_events __read_mostly; +static LIST_HEAD(pmus); +static DEFINE_MUTEX(pmus_lock); +static struct srcu_struct pmus_srcu; + /* * perf event paranoia level: * -1 - not paranoid at all @@ -78,9 +77,9 @@ void perf_pmu_enable(struct pmu *pmu) pmu->pmu_enable(pmu); } -static void perf_pmu_rotate_start(void) +static void perf_pmu_rotate_start(struct pmu *pmu) { - struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context); + struct perf_cpu_context *cpuctx = this_cpu_ptr(pmu->pmu_cpu_context); if (hrtimer_active(&cpuctx->timer)) return; @@ -90,9 +89,9 @@ static void perf_pmu_rotate_start(void) HRTIMER_MODE_REL_PINNED, 0); } -static void perf_pmu_rotate_stop(void) +static void perf_pmu_rotate_stop(struct pmu *pmu) { - struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context); + struct perf_cpu_context *cpuctx = this_cpu_ptr(pmu->pmu_cpu_context); hrtimer_cancel(&cpuctx->timer); } @@ -301,7 +300,7 @@ list_add_event(struct perf_event *event, struct perf_event_context *ctx) list_add_rcu(&event->event_entry, &ctx->event_list); if (!ctx->nr_events) - perf_pmu_rotate_start(); + perf_pmu_rotate_start(ctx->pmu); ctx->nr_events++; if (event->attr.inherit_stat) ctx->nr_stat++; @@ -466,6 +465,12 @@ group_sched_out(struct perf_event *group_event, cpuctx->exclusive = 0; } +static inline struct perf_cpu_context * +__get_cpu_context(struct perf_event_context *ctx) +{ + return this_cpu_ptr(ctx->pmu->pmu_cpu_context); +} + /* * Cross CPU call to remove a performance event * @@ -474,9 +479,9 @@ group_sched_out(struct perf_event *group_event, */ static void __perf_event_remove_from_context(void *info) { - struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context); struct perf_event *event = info; struct perf_event_context *ctx = event->ctx; + struct perf_cpu_context *cpuctx = __get_cpu_context(ctx); /* * If this is a task context, we need to check whether it is @@ -556,8 +561,8 @@ retry: static void __perf_event_disable(void *info) { struct perf_event *event = info; - struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context); struct perf_event_context *ctx = event->ctx; + struct perf_cpu_context *cpuctx = __get_cpu_context(ctx); /* * If this is a per-task event, need to check whether this @@ -765,10 +770,10 @@ static void add_event_to_ctx(struct perf_event *event, */ static void __perf_install_in_context(void *info) { - struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context); struct perf_event *event = info; struct perf_event_context *ctx = event->ctx; struct perf_event *leader = event->group_leader; + struct perf_cpu_context *cpuctx = __get_cpu_context(ctx); int err; /* @@ -912,9 +917,9 @@ static void __perf_event_mark_enabled(struct perf_event *event, static void __perf_event_enable(void *info) { struct perf_event *event = info; - struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context); struct perf_event_context *ctx = event->ctx; struct perf_event *leader = event->group_leader; + struct perf_cpu_context *cpuctx = __get_cpu_context(ctx); int err; /* @@ -1188,15 +1193,19 @@ static void perf_event_sync_stat(struct perf_event_context *ctx, void perf_event_task_sched_out(struct task_struct *task, struct task_struct *next) { - struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context); struct perf_event_context *ctx = task->perf_event_ctxp; struct perf_event_context *next_ctx; struct perf_event_context *parent; + struct perf_cpu_context *cpuctx; int do_switch = 1; perf_sw_event(PERF_COUNT_SW_CONTEXT_SWITCHES, 1, 1, NULL, 0); - if (likely(!ctx || !cpuctx->task_ctx)) + if (likely(!ctx)) + return; + + cpuctx = __get_cpu_context(ctx); + if (!cpuctx->task_ctx) return; rcu_read_lock(); @@ -1242,7 +1251,7 @@ void perf_event_task_sched_out(struct task_struct *task, static void task_ctx_sched_out(struct perf_event_context *ctx, enum event_type_t event_type) { - struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context); + struct perf_cpu_context *cpuctx = __get_cpu_context(ctx); if (!cpuctx->task_ctx) return; @@ -1360,8 +1369,8 @@ static void cpu_ctx_sched_in(struct perf_cpu_context *cpuctx, static void task_ctx_sched_in(struct task_struct *task, enum event_type_t event_type) { - struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context); struct perf_event_context *ctx = task->perf_event_ctxp; + struct perf_cpu_context *cpuctx = __get_cpu_context(ctx); if (likely(!ctx)) return; @@ -1383,12 +1392,13 @@ static void task_ctx_sched_in(struct task_struct *task, */ void perf_event_task_sched_in(struct task_struct *task) { - struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context); struct perf_event_context *ctx = task->perf_event_ctxp; + struct perf_cpu_context *cpuctx; if (likely(!ctx)) return; + cpuctx = __get_cpu_context(ctx); if (cpuctx->task_ctx == ctx) return; @@ -1409,7 +1419,7 @@ void perf_event_task_sched_in(struct task_struct *task) * Since these rotations are per-cpu, we need to ensure the * cpu-context we got scheduled on is actually rotating. */ - perf_pmu_rotate_start(); + perf_pmu_rotate_start(ctx->pmu); } #define MAX_INTERRUPTS (~0ULL) @@ -1687,9 +1697,9 @@ out: */ static void __perf_event_read(void *info) { - struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context); struct perf_event *event = info; struct perf_event_context *ctx = event->ctx; + struct perf_cpu_context *cpuctx = __get_cpu_context(ctx); /* * If this is a task context, we need to check whether it is @@ -1962,7 +1972,8 @@ __perf_event_init_context(struct perf_event_context *ctx, ctx->task = task; } -static struct perf_event_context *find_get_context(pid_t pid, int cpu) +static struct perf_event_context * +find_get_context(struct pmu *pmu, pid_t pid, int cpu) { struct perf_event_context *ctx; struct perf_cpu_context *cpuctx; @@ -1986,7 +1997,7 @@ static struct perf_event_context *find_get_context(pid_t pid, int cpu) if (!cpu_online(cpu)) return ERR_PTR(-ENODEV); - cpuctx = &per_cpu(perf_cpu_context, cpu); + cpuctx = per_cpu_ptr(pmu->pmu_cpu_context, cpu); ctx = &cpuctx->ctx; get_ctx(ctx); @@ -2030,6 +2041,7 @@ retry: if (!ctx) goto errout; __perf_event_init_context(ctx, task); + ctx->pmu = pmu; get_ctx(ctx); if (cmpxchg(&task->perf_event_ctxp, NULL, ctx)) { /* @@ -3745,18 +3757,20 @@ static void perf_event_task_ctx(struct perf_event_context *ctx, static void perf_event_task_event(struct perf_task_event *task_event) { - struct perf_cpu_context *cpuctx; struct perf_event_context *ctx = task_event->task_ctx; + struct perf_cpu_context *cpuctx; + struct pmu *pmu; - rcu_read_lock(); - cpuctx = &get_cpu_var(perf_cpu_context); - perf_event_task_ctx(&cpuctx->ctx, task_event); + rcu_read_lock_sched(); + list_for_each_entry_rcu(pmu, &pmus, entry) { + cpuctx = this_cpu_ptr(pmu->pmu_cpu_context); + perf_event_task_ctx(&cpuctx->ctx, task_event); + } if (!ctx) ctx = rcu_dereference(current->perf_event_ctxp); if (ctx) perf_event_task_ctx(ctx, task_event); - put_cpu_var(perf_cpu_context); - rcu_read_unlock(); + rcu_read_unlock_sched(); } static void perf_event_task(struct task_struct *task, @@ -3861,6 +3875,7 @@ static void perf_event_comm_event(struct perf_comm_event *comm_event) struct perf_cpu_context *cpuctx; struct perf_event_context *ctx; unsigned int size; + struct pmu *pmu; char comm[TASK_COMM_LEN]; memset(comm, 0, sizeof(comm)); @@ -3872,14 +3887,15 @@ static void perf_event_comm_event(struct perf_comm_event *comm_event) comm_event->event_id.header.size = sizeof(comm_event->event_id) + size; - rcu_read_lock(); - cpuctx = &get_cpu_var(perf_cpu_context); - perf_event_comm_ctx(&cpuctx->ctx, comm_event); + rcu_read_lock_sched(); + list_for_each_entry_rcu(pmu, &pmus, entry) { + cpuctx = this_cpu_ptr(pmu->pmu_cpu_context); + perf_event_comm_ctx(&cpuctx->ctx, comm_event); + } ctx = rcu_dereference(current->perf_event_ctxp); if (ctx) perf_event_comm_ctx(ctx, comm_event); - put_cpu_var(perf_cpu_context); - rcu_read_unlock(); + rcu_read_unlock_sched(); } void perf_event_comm(struct task_struct *task) @@ -3989,6 +4005,7 @@ static void perf_event_mmap_event(struct perf_mmap_event *mmap_event) char tmp[16]; char *buf = NULL; const char *name; + struct pmu *pmu; memset(tmp, 0, sizeof(tmp)); @@ -4040,14 +4057,16 @@ got_name: mmap_event->event_id.header.size = sizeof(mmap_event->event_id) + size; - rcu_read_lock(); - cpuctx = &get_cpu_var(perf_cpu_context); - perf_event_mmap_ctx(&cpuctx->ctx, mmap_event, vma->vm_flags & VM_EXEC); + rcu_read_lock_sched(); + list_for_each_entry_rcu(pmu, &pmus, entry) { + cpuctx = this_cpu_ptr(pmu->pmu_cpu_context); + perf_event_mmap_ctx(&cpuctx->ctx, mmap_event, + vma->vm_flags & VM_EXEC); + } ctx = rcu_dereference(current->perf_event_ctxp); if (ctx) perf_event_mmap_ctx(ctx, mmap_event, vma->vm_flags & VM_EXEC); - put_cpu_var(perf_cpu_context); - rcu_read_unlock(); + rcu_read_unlock_sched(); kfree(buf); } @@ -4982,10 +5001,6 @@ static struct pmu perf_task_clock = { .read = task_clock_event_read, }; -static LIST_HEAD(pmus); -static DEFINE_MUTEX(pmus_lock); -static struct srcu_struct pmus_srcu; - static void perf_pmu_nop_void(struct pmu *pmu) { } @@ -5013,7 +5028,7 @@ static void perf_pmu_cancel_txn(struct pmu *pmu) int perf_pmu_register(struct pmu *pmu) { - int ret; + int cpu, ret; mutex_lock(&pmus_lock); ret = -ENOMEM; @@ -5021,6 +5036,21 @@ int perf_pmu_register(struct pmu *pmu) if (!pmu->pmu_disable_count) goto unlock; + pmu->pmu_cpu_context = alloc_percpu(struct perf_cpu_context); + if (!pmu->pmu_cpu_context) + goto free_pdc; + + for_each_possible_cpu(cpu) { + struct perf_cpu_context *cpuctx; + + cpuctx = per_cpu_ptr(pmu->pmu_cpu_context, cpu); + __perf_event_init_context(&cpuctx->ctx, NULL); + cpuctx->ctx.pmu = pmu; + cpuctx->timer_interval = TICK_NSEC; + hrtimer_init(&cpuctx->timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL); + cpuctx->timer.function = perf_event_context_tick; + } + if (!pmu->start_txn) { if (pmu->pmu_enable) { /* @@ -5049,6 +5079,10 @@ unlock: mutex_unlock(&pmus_lock); return ret; + +free_pdc: + free_percpu(pmu->pmu_disable_count); + goto unlock; } void perf_pmu_unregister(struct pmu *pmu) @@ -5057,9 +5091,14 @@ void perf_pmu_unregister(struct pmu *pmu) list_del_rcu(&pmu->entry); mutex_unlock(&pmus_lock); + /* + * We use the pmu list either under SRCU or preempt_disable, + * synchronize_srcu() implies synchronize_sched() so we're good. + */ synchronize_srcu(&pmus_srcu); free_percpu(pmu->pmu_disable_count); + free_percpu(pmu->pmu_cpu_context); } struct pmu *perf_init_event(struct perf_event *event) @@ -5374,7 +5413,7 @@ SYSCALL_DEFINE5(perf_event_open, /* * Get the target context (task or percpu): */ - ctx = find_get_context(pid, cpu); + ctx = find_get_context(event->pmu, pid, cpu); if (IS_ERR(ctx)) { err = PTR_ERR(ctx); goto err_alloc; @@ -5489,7 +5528,7 @@ perf_event_create_kernel_counter(struct perf_event_attr *attr, int cpu, goto err; } - ctx = find_get_context(pid, cpu); + ctx = find_get_context(event->pmu, pid, cpu); if (IS_ERR(ctx)) { err = PTR_ERR(ctx); goto err_free; @@ -5833,6 +5872,7 @@ inherit_task_group(struct perf_event *event, struct task_struct *parent, return -ENOMEM; __perf_event_init_context(child_ctx, child); + child_ctx->pmu = event->pmu; child->perf_event_ctxp = child_ctx; get_task_struct(child); } @@ -5935,30 +5975,18 @@ int perf_event_init_task(struct task_struct *child) static void __init perf_event_init_all_cpus(void) { - struct perf_cpu_context *cpuctx; struct swevent_htable *swhash; int cpu; for_each_possible_cpu(cpu) { swhash = &per_cpu(swevent_htable, cpu); mutex_init(&swhash->hlist_mutex); - - cpuctx = &per_cpu(perf_cpu_context, cpu); - __perf_event_init_context(&cpuctx->ctx, NULL); - cpuctx->timer_interval = TICK_NSEC; - hrtimer_init(&cpuctx->timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL); - cpuctx->timer.function = perf_event_context_tick; } } static void __cpuinit perf_event_init_cpu(int cpu) { - struct perf_cpu_context *cpuctx; - struct swevent_htable *swhash; - - cpuctx = &per_cpu(perf_cpu_context, cpu); - - swhash = &per_cpu(swevent_htable, cpu); + struct swevent_htable *swhash = &per_cpu(swevent_htable, cpu); mutex_lock(&swhash->hlist_mutex); if (swhash->hlist_refcount > 0) { @@ -5972,32 +6000,46 @@ static void __cpuinit perf_event_init_cpu(int cpu) } #ifdef CONFIG_HOTPLUG_CPU -static void __perf_event_exit_cpu(void *info) +static void __perf_event_exit_context(void *__info) { - struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context); - struct perf_event_context *ctx = &cpuctx->ctx; + struct perf_event_context *ctx = __info; struct perf_event *event, *tmp; - perf_pmu_rotate_stop(); + perf_pmu_rotate_stop(ctx->pmu); list_for_each_entry_safe(event, tmp, &ctx->pinned_groups, group_entry) __perf_event_remove_from_context(event); list_for_each_entry_safe(event, tmp, &ctx->flexible_groups, group_entry) __perf_event_remove_from_context(event); } + +static void perf_event_exit_cpu_context(int cpu) +{ + struct perf_event_context *ctx; + struct pmu *pmu; + int idx; + + idx = srcu_read_lock(&pmus_srcu); + list_for_each_entry_rcu(pmu, &pmus, entry) { + ctx = &this_cpu_ptr(pmu->pmu_cpu_context)->ctx; + + mutex_lock(&ctx->mutex); + smp_call_function_single(cpu, __perf_event_exit_context, ctx, 1); + mutex_unlock(&ctx->mutex); + } + srcu_read_unlock(&pmus_srcu, idx); + +} + static void perf_event_exit_cpu(int cpu) { - struct perf_cpu_context *cpuctx = &per_cpu(perf_cpu_context, cpu); struct swevent_htable *swhash = &per_cpu(swevent_htable, cpu); - struct perf_event_context *ctx = &cpuctx->ctx; mutex_lock(&swhash->hlist_mutex); swevent_hlist_release(swhash); mutex_unlock(&swhash->hlist_mutex); - mutex_lock(&ctx->mutex); - smp_call_function_single(cpu, __perf_event_exit_cpu, NULL, 1); - mutex_unlock(&ctx->mutex); + perf_event_exit_cpu_context(cpu); } #else static inline void perf_event_exit_cpu(int cpu) { } -- cgit v1.2.3 From 97dee4f3206622f31396dede2b5ddb8670458f56 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Tue, 7 Sep 2010 15:35:33 +0200 Subject: perf: Move some code around Move all inherit code near each other. Signed-off-by: Peter Zijlstra Cc: paulus Cc: stephane eranian Cc: Robert Richter Cc: Frederic Weisbecker Cc: Lin Ming Cc: Yanmin LKML-Reference: Signed-off-by: Ingo Molnar --- kernel/perf_event.c | 200 ++++++++++++++++++++++++++-------------------------- 1 file changed, 100 insertions(+), 100 deletions(-) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index 8ca6e690ffe3..dae0e2f30293 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -5556,106 +5556,6 @@ err: } EXPORT_SYMBOL_GPL(perf_event_create_kernel_counter); -/* - * inherit a event from parent task to child task: - */ -static struct perf_event * -inherit_event(struct perf_event *parent_event, - struct task_struct *parent, - struct perf_event_context *parent_ctx, - struct task_struct *child, - struct perf_event *group_leader, - struct perf_event_context *child_ctx) -{ - struct perf_event *child_event; - - /* - * Instead of creating recursive hierarchies of events, - * we link inherited events back to the original parent, - * which has a filp for sure, which we use as the reference - * count: - */ - if (parent_event->parent) - parent_event = parent_event->parent; - - child_event = perf_event_alloc(&parent_event->attr, - parent_event->cpu, - group_leader, parent_event, - NULL); - if (IS_ERR(child_event)) - return child_event; - get_ctx(child_ctx); - - /* - * Make the child state follow the state of the parent event, - * not its attr.disabled bit. We hold the parent's mutex, - * so we won't race with perf_event_{en, dis}able_family. - */ - if (parent_event->state >= PERF_EVENT_STATE_INACTIVE) - child_event->state = PERF_EVENT_STATE_INACTIVE; - else - child_event->state = PERF_EVENT_STATE_OFF; - - if (parent_event->attr.freq) { - u64 sample_period = parent_event->hw.sample_period; - struct hw_perf_event *hwc = &child_event->hw; - - hwc->sample_period = sample_period; - hwc->last_period = sample_period; - - local64_set(&hwc->period_left, sample_period); - } - - child_event->ctx = child_ctx; - child_event->overflow_handler = parent_event->overflow_handler; - - /* - * Link it up in the child's context: - */ - add_event_to_ctx(child_event, child_ctx); - - /* - * Get a reference to the parent filp - we will fput it - * when the child event exits. This is safe to do because - * we are in the parent and we know that the filp still - * exists and has a nonzero count: - */ - atomic_long_inc(&parent_event->filp->f_count); - - /* - * Link this into the parent event's child list - */ - WARN_ON_ONCE(parent_event->ctx->parent_ctx); - mutex_lock(&parent_event->child_mutex); - list_add_tail(&child_event->child_list, &parent_event->child_list); - mutex_unlock(&parent_event->child_mutex); - - return child_event; -} - -static int inherit_group(struct perf_event *parent_event, - struct task_struct *parent, - struct perf_event_context *parent_ctx, - struct task_struct *child, - struct perf_event_context *child_ctx) -{ - struct perf_event *leader; - struct perf_event *sub; - struct perf_event *child_ctr; - - leader = inherit_event(parent_event, parent, parent_ctx, - child, NULL, child_ctx); - if (IS_ERR(leader)) - return PTR_ERR(leader); - list_for_each_entry(sub, &parent_event->sibling_list, group_entry) { - child_ctr = inherit_event(sub, parent, parent_ctx, - child, leader, child_ctx); - if (IS_ERR(child_ctr)) - return PTR_ERR(child_ctr); - } - return 0; -} - static void sync_child_event(struct perf_event *child_event, struct task_struct *child) { @@ -5844,6 +5744,106 @@ again: put_ctx(ctx); } +/* + * inherit a event from parent task to child task: + */ +static struct perf_event * +inherit_event(struct perf_event *parent_event, + struct task_struct *parent, + struct perf_event_context *parent_ctx, + struct task_struct *child, + struct perf_event *group_leader, + struct perf_event_context *child_ctx) +{ + struct perf_event *child_event; + + /* + * Instead of creating recursive hierarchies of events, + * we link inherited events back to the original parent, + * which has a filp for sure, which we use as the reference + * count: + */ + if (parent_event->parent) + parent_event = parent_event->parent; + + child_event = perf_event_alloc(&parent_event->attr, + parent_event->cpu, + group_leader, parent_event, + NULL); + if (IS_ERR(child_event)) + return child_event; + get_ctx(child_ctx); + + /* + * Make the child state follow the state of the parent event, + * not its attr.disabled bit. We hold the parent's mutex, + * so we won't race with perf_event_{en, dis}able_family. + */ + if (parent_event->state >= PERF_EVENT_STATE_INACTIVE) + child_event->state = PERF_EVENT_STATE_INACTIVE; + else + child_event->state = PERF_EVENT_STATE_OFF; + + if (parent_event->attr.freq) { + u64 sample_period = parent_event->hw.sample_period; + struct hw_perf_event *hwc = &child_event->hw; + + hwc->sample_period = sample_period; + hwc->last_period = sample_period; + + local64_set(&hwc->period_left, sample_period); + } + + child_event->ctx = child_ctx; + child_event->overflow_handler = parent_event->overflow_handler; + + /* + * Link it up in the child's context: + */ + add_event_to_ctx(child_event, child_ctx); + + /* + * Get a reference to the parent filp - we will fput it + * when the child event exits. This is safe to do because + * we are in the parent and we know that the filp still + * exists and has a nonzero count: + */ + atomic_long_inc(&parent_event->filp->f_count); + + /* + * Link this into the parent event's child list + */ + WARN_ON_ONCE(parent_event->ctx->parent_ctx); + mutex_lock(&parent_event->child_mutex); + list_add_tail(&child_event->child_list, &parent_event->child_list); + mutex_unlock(&parent_event->child_mutex); + + return child_event; +} + +static int inherit_group(struct perf_event *parent_event, + struct task_struct *parent, + struct perf_event_context *parent_ctx, + struct task_struct *child, + struct perf_event_context *child_ctx) +{ + struct perf_event *leader; + struct perf_event *sub; + struct perf_event *child_ctr; + + leader = inherit_event(parent_event, parent, parent_ctx, + child, NULL, child_ctx); + if (IS_ERR(leader)) + return PTR_ERR(leader); + list_for_each_entry(sub, &parent_event->sibling_list, group_entry) { + child_ctr = inherit_event(sub, parent, parent_ctx, + child, leader, child_ctx); + if (IS_ERR(child_ctr)) + return PTR_ERR(child_ctr); + } + return 0; +} + static int inherit_task_group(struct perf_event *event, struct task_struct *parent, struct perf_event_context *parent_ctx, -- cgit v1.2.3 From eb184479874238393ac186c4e054d24311c34aaa Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Tue, 7 Sep 2010 15:55:13 +0200 Subject: perf: Clean up perf_event_context allocation Unify the two perf_event_context allocation sites. Signed-off-by: Peter Zijlstra Cc: paulus Cc: stephane eranian Cc: Robert Richter Cc: Frederic Weisbecker Cc: Lin Ming Cc: Yanmin LKML-Reference: Signed-off-by: Ingo Molnar --- kernel/perf_event.c | 41 ++++++++++++++++++++++++++--------------- 1 file changed, 26 insertions(+), 15 deletions(-) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index dae0e2f30293..13d98d756347 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -1959,9 +1959,7 @@ exit_put: /* * Initialize the perf_event context in a task_struct: */ -static void -__perf_event_init_context(struct perf_event_context *ctx, - struct task_struct *task) +static void __perf_event_init_context(struct perf_event_context *ctx) { raw_spin_lock_init(&ctx->lock); mutex_init(&ctx->mutex); @@ -1969,7 +1967,25 @@ __perf_event_init_context(struct perf_event_context *ctx, INIT_LIST_HEAD(&ctx->flexible_groups); INIT_LIST_HEAD(&ctx->event_list); atomic_set(&ctx->refcount, 1); - ctx->task = task; +} + +static struct perf_event_context * +alloc_perf_context(struct pmu *pmu, struct task_struct *task) +{ + struct perf_event_context *ctx; + + ctx = kzalloc(sizeof(struct perf_event_context), GFP_KERNEL); + if (!ctx) + return NULL; + + __perf_event_init_context(ctx); + if (task) { + ctx->task = task; + get_task_struct(task); + } + ctx->pmu = pmu; + + return ctx; } static struct perf_event_context * @@ -2036,22 +2052,22 @@ retry: } if (!ctx) { - ctx = kzalloc(sizeof(struct perf_event_context), GFP_KERNEL); + ctx = alloc_perf_context(pmu, task); err = -ENOMEM; if (!ctx) goto errout; - __perf_event_init_context(ctx, task); - ctx->pmu = pmu; + get_ctx(ctx); + if (cmpxchg(&task->perf_event_ctxp, NULL, ctx)) { /* * We raced with some other task; use * the context they set. */ + put_task_struct(task); kfree(ctx); goto retry; } - get_task_struct(task); } put_task_struct(task); @@ -5044,7 +5060,7 @@ int perf_pmu_register(struct pmu *pmu) struct perf_cpu_context *cpuctx; cpuctx = per_cpu_ptr(pmu->pmu_cpu_context, cpu); - __perf_event_init_context(&cpuctx->ctx, NULL); + __perf_event_init_context(&cpuctx->ctx); cpuctx->ctx.pmu = pmu; cpuctx->timer_interval = TICK_NSEC; hrtimer_init(&cpuctx->timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL); @@ -5866,15 +5882,11 @@ inherit_task_group(struct perf_event *event, struct task_struct *parent, * child. */ - child_ctx = kzalloc(sizeof(struct perf_event_context), - GFP_KERNEL); + child_ctx = alloc_perf_context(event->pmu, child); if (!child_ctx) return -ENOMEM; - __perf_event_init_context(child_ctx, child); - child_ctx->pmu = event->pmu; child->perf_event_ctxp = child_ctx; - get_task_struct(child); } ret = inherit_group(event, parent, parent_ctx, @@ -5886,7 +5898,6 @@ inherit_task_group(struct perf_event *event, struct task_struct *parent, return ret; } - /* * Initialize the perf_event context in task_struct */ -- cgit v1.2.3 From 8dc85d547285668e509f86c177bcd4ea055bcaaf Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Thu, 2 Sep 2010 16:50:03 +0200 Subject: perf: Multiple task contexts Provide the infrastructure for multiple task contexts. A more flexible approach would have resulted in more pointer chases in the scheduling hot-paths. This approach has the limitation of a static number of task contexts. Since I expect most external PMUs to be system wide, or at least node wide (as per the intel uncore unit) they won't actually need a task context. Signed-off-by: Peter Zijlstra Cc: paulus Cc: stephane eranian Cc: Robert Richter Cc: Frederic Weisbecker Cc: Lin Ming Cc: Yanmin LKML-Reference: Signed-off-by: Ingo Molnar --- kernel/perf_event.c | 336 ++++++++++++++++++++++++++++++++++++---------------- 1 file changed, 231 insertions(+), 105 deletions(-) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index 13d98d756347..7223ea875861 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -148,13 +148,13 @@ static u64 primary_event_id(struct perf_event *event) * the context could get moved to another task. */ static struct perf_event_context * -perf_lock_task_context(struct task_struct *task, unsigned long *flags) +perf_lock_task_context(struct task_struct *task, int ctxn, unsigned long *flags) { struct perf_event_context *ctx; rcu_read_lock(); retry: - ctx = rcu_dereference(task->perf_event_ctxp); + ctx = rcu_dereference(task->perf_event_ctxp[ctxn]); if (ctx) { /* * If this context is a clone of another, it might @@ -167,7 +167,7 @@ retry: * can't get swapped on us any more. */ raw_spin_lock_irqsave(&ctx->lock, *flags); - if (ctx != rcu_dereference(task->perf_event_ctxp)) { + if (ctx != rcu_dereference(task->perf_event_ctxp[ctxn])) { raw_spin_unlock_irqrestore(&ctx->lock, *flags); goto retry; } @@ -186,12 +186,13 @@ retry: * can't get swapped to another task. This also increments its * reference count so that the context can't get freed. */ -static struct perf_event_context *perf_pin_task_context(struct task_struct *task) +static struct perf_event_context * +perf_pin_task_context(struct task_struct *task, int ctxn) { struct perf_event_context *ctx; unsigned long flags; - ctx = perf_lock_task_context(task, &flags); + ctx = perf_lock_task_context(task, ctxn, &flags); if (ctx) { ++ctx->pin_count; raw_spin_unlock_irqrestore(&ctx->lock, flags); @@ -1179,28 +1180,15 @@ static void perf_event_sync_stat(struct perf_event_context *ctx, } } -/* - * Called from scheduler to remove the events of the current task, - * with interrupts disabled. - * - * We stop each event and update the event value in event->count. - * - * This does not protect us against NMI, but disable() - * sets the disabled bit in the control field of event _before_ - * accessing the event control register. If a NMI hits, then it will - * not restart the event. - */ -void perf_event_task_sched_out(struct task_struct *task, - struct task_struct *next) +void perf_event_context_sched_out(struct task_struct *task, int ctxn, + struct task_struct *next) { - struct perf_event_context *ctx = task->perf_event_ctxp; + struct perf_event_context *ctx = task->perf_event_ctxp[ctxn]; struct perf_event_context *next_ctx; struct perf_event_context *parent; struct perf_cpu_context *cpuctx; int do_switch = 1; - perf_sw_event(PERF_COUNT_SW_CONTEXT_SWITCHES, 1, 1, NULL, 0); - if (likely(!ctx)) return; @@ -1210,7 +1198,7 @@ void perf_event_task_sched_out(struct task_struct *task, rcu_read_lock(); parent = rcu_dereference(ctx->parent_ctx); - next_ctx = next->perf_event_ctxp; + next_ctx = next->perf_event_ctxp[ctxn]; if (parent && next_ctx && rcu_dereference(next_ctx->parent_ctx) == parent) { /* @@ -1229,8 +1217,8 @@ void perf_event_task_sched_out(struct task_struct *task, * XXX do we need a memory barrier of sorts * wrt to rcu_dereference() of perf_event_ctxp */ - task->perf_event_ctxp = next_ctx; - next->perf_event_ctxp = ctx; + task->perf_event_ctxp[ctxn] = next_ctx; + next->perf_event_ctxp[ctxn] = ctx; ctx->task = next; next_ctx->task = task; do_switch = 0; @@ -1248,6 +1236,31 @@ void perf_event_task_sched_out(struct task_struct *task, } } +#define for_each_task_context_nr(ctxn) \ + for ((ctxn) = 0; (ctxn) < perf_nr_task_contexts; (ctxn)++) + +/* + * Called from scheduler to remove the events of the current task, + * with interrupts disabled. + * + * We stop each event and update the event value in event->count. + * + * This does not protect us against NMI, but disable() + * sets the disabled bit in the control field of event _before_ + * accessing the event control register. If a NMI hits, then it will + * not restart the event. + */ +void perf_event_task_sched_out(struct task_struct *task, + struct task_struct *next) +{ + int ctxn; + + perf_sw_event(PERF_COUNT_SW_CONTEXT_SWITCHES, 1, 1, NULL, 0); + + for_each_task_context_nr(ctxn) + perf_event_context_sched_out(task, ctxn, next); +} + static void task_ctx_sched_out(struct perf_event_context *ctx, enum event_type_t event_type) { @@ -1366,38 +1379,23 @@ static void cpu_ctx_sched_in(struct perf_cpu_context *cpuctx, ctx_sched_in(ctx, cpuctx, event_type); } -static void task_ctx_sched_in(struct task_struct *task, +static void task_ctx_sched_in(struct perf_event_context *ctx, enum event_type_t event_type) { - struct perf_event_context *ctx = task->perf_event_ctxp; - struct perf_cpu_context *cpuctx = __get_cpu_context(ctx); + struct perf_cpu_context *cpuctx; - if (likely(!ctx)) - return; + cpuctx = __get_cpu_context(ctx); if (cpuctx->task_ctx == ctx) return; + ctx_sched_in(ctx, cpuctx, event_type); cpuctx->task_ctx = ctx; } -/* - * Called from scheduler to add the events of the current task - * with interrupts disabled. - * - * We restore the event value and then enable it. - * - * This does not protect us against NMI, but enable() - * sets the enabled bit in the control field of event _before_ - * accessing the event control register. If a NMI hits, then it will - * keep the event running. - */ -void perf_event_task_sched_in(struct task_struct *task) + +void perf_event_context_sched_in(struct perf_event_context *ctx) { - struct perf_event_context *ctx = task->perf_event_ctxp; struct perf_cpu_context *cpuctx; - if (likely(!ctx)) - return; - cpuctx = __get_cpu_context(ctx); if (cpuctx->task_ctx == ctx) return; @@ -1422,6 +1420,31 @@ void perf_event_task_sched_in(struct task_struct *task) perf_pmu_rotate_start(ctx->pmu); } +/* + * Called from scheduler to add the events of the current task + * with interrupts disabled. + * + * We restore the event value and then enable it. + * + * This does not protect us against NMI, but enable() + * sets the enabled bit in the control field of event _before_ + * accessing the event control register. If a NMI hits, then it will + * keep the event running. + */ +void perf_event_task_sched_in(struct task_struct *task) +{ + struct perf_event_context *ctx; + int ctxn; + + for_each_task_context_nr(ctxn) { + ctx = task->perf_event_ctxp[ctxn]; + if (likely(!ctx)) + continue; + + perf_event_context_sched_in(ctx); + } +} + #define MAX_INTERRUPTS (~0ULL) static void perf_log_throttle(struct perf_event *event, int enable); @@ -1588,7 +1611,7 @@ static enum hrtimer_restart perf_event_context_tick(struct hrtimer *timer) { enum hrtimer_restart restart = HRTIMER_NORESTART; struct perf_cpu_context *cpuctx; - struct perf_event_context *ctx; + struct perf_event_context *ctx = NULL; int rotate = 0; cpuctx = container_of(timer, struct perf_cpu_context, timer); @@ -1599,7 +1622,7 @@ static enum hrtimer_restart perf_event_context_tick(struct hrtimer *timer) rotate = 1; } - ctx = current->perf_event_ctxp; + ctx = cpuctx->task_ctx; if (ctx && ctx->nr_events) { restart = HRTIMER_RESTART; if (ctx->nr_events != ctx->nr_active) @@ -1623,7 +1646,7 @@ static enum hrtimer_restart perf_event_context_tick(struct hrtimer *timer) cpu_ctx_sched_in(cpuctx, EVENT_FLEXIBLE); if (ctx) - task_ctx_sched_in(current, EVENT_FLEXIBLE); + task_ctx_sched_in(ctx, EVENT_FLEXIBLE); done: hrtimer_forward_now(timer, ns_to_ktime(cpuctx->timer_interval)); @@ -1650,20 +1673,18 @@ static int event_enable_on_exec(struct perf_event *event, * Enable all of a task's events that have been marked enable-on-exec. * This expects task == current. */ -static void perf_event_enable_on_exec(struct task_struct *task) +static void perf_event_enable_on_exec(struct perf_event_context *ctx) { - struct perf_event_context *ctx; struct perf_event *event; unsigned long flags; int enabled = 0; int ret; local_irq_save(flags); - ctx = task->perf_event_ctxp; if (!ctx || !ctx->nr_events) goto out; - __perf_event_task_sched_out(ctx); + task_ctx_sched_out(ctx, EVENT_ALL); raw_spin_lock(&ctx->lock); @@ -1687,7 +1708,7 @@ static void perf_event_enable_on_exec(struct task_struct *task) raw_spin_unlock(&ctx->lock); - perf_event_task_sched_in(task); + perf_event_context_sched_in(ctx); out: local_irq_restore(flags); } @@ -1995,7 +2016,7 @@ find_get_context(struct pmu *pmu, pid_t pid, int cpu) struct perf_cpu_context *cpuctx; struct task_struct *task; unsigned long flags; - int err; + int ctxn, err; if (pid == -1 && cpu != -1) { /* Must be root to operate on a CPU event: */ @@ -2044,8 +2065,13 @@ find_get_context(struct pmu *pmu, pid_t pid, int cpu) if (!ptrace_may_access(task, PTRACE_MODE_READ)) goto errout; + err = -EINVAL; + ctxn = pmu->task_ctx_nr; + if (ctxn < 0) + goto errout; + retry: - ctx = perf_lock_task_context(task, &flags); + ctx = perf_lock_task_context(task, ctxn, &flags); if (ctx) { unclone_ctx(ctx); raw_spin_unlock_irqrestore(&ctx->lock, flags); @@ -2059,7 +2085,7 @@ retry: get_ctx(ctx); - if (cmpxchg(&task->perf_event_ctxp, NULL, ctx)) { + if (cmpxchg(&task->perf_event_ctxp[ctxn], NULL, ctx)) { /* * We raced with some other task; use * the context they set. @@ -3773,19 +3799,26 @@ static void perf_event_task_ctx(struct perf_event_context *ctx, static void perf_event_task_event(struct perf_task_event *task_event) { - struct perf_event_context *ctx = task_event->task_ctx; struct perf_cpu_context *cpuctx; + struct perf_event_context *ctx; struct pmu *pmu; + int ctxn; rcu_read_lock_sched(); list_for_each_entry_rcu(pmu, &pmus, entry) { cpuctx = this_cpu_ptr(pmu->pmu_cpu_context); perf_event_task_ctx(&cpuctx->ctx, task_event); + + ctx = task_event->task_ctx; + if (!ctx) { + ctxn = pmu->task_ctx_nr; + if (ctxn < 0) + continue; + ctx = rcu_dereference(current->perf_event_ctxp[ctxn]); + } + if (ctx) + perf_event_task_ctx(ctx, task_event); } - if (!ctx) - ctx = rcu_dereference(current->perf_event_ctxp); - if (ctx) - perf_event_task_ctx(ctx, task_event); rcu_read_unlock_sched(); } @@ -3890,9 +3923,10 @@ static void perf_event_comm_event(struct perf_comm_event *comm_event) { struct perf_cpu_context *cpuctx; struct perf_event_context *ctx; + char comm[TASK_COMM_LEN]; unsigned int size; struct pmu *pmu; - char comm[TASK_COMM_LEN]; + int ctxn; memset(comm, 0, sizeof(comm)); strlcpy(comm, comm_event->task->comm, sizeof(comm)); @@ -3907,19 +3941,31 @@ static void perf_event_comm_event(struct perf_comm_event *comm_event) list_for_each_entry_rcu(pmu, &pmus, entry) { cpuctx = this_cpu_ptr(pmu->pmu_cpu_context); perf_event_comm_ctx(&cpuctx->ctx, comm_event); + + ctxn = pmu->task_ctx_nr; + if (ctxn < 0) + continue; + + ctx = rcu_dereference(current->perf_event_ctxp[ctxn]); + if (ctx) + perf_event_comm_ctx(ctx, comm_event); } - ctx = rcu_dereference(current->perf_event_ctxp); - if (ctx) - perf_event_comm_ctx(ctx, comm_event); rcu_read_unlock_sched(); } void perf_event_comm(struct task_struct *task) { struct perf_comm_event comm_event; + struct perf_event_context *ctx; + int ctxn; - if (task->perf_event_ctxp) - perf_event_enable_on_exec(task); + for_each_task_context_nr(ctxn) { + ctx = task->perf_event_ctxp[ctxn]; + if (!ctx) + continue; + + perf_event_enable_on_exec(ctx); + } if (!atomic_read(&nr_comm_events)) return; @@ -4022,6 +4068,7 @@ static void perf_event_mmap_event(struct perf_mmap_event *mmap_event) char *buf = NULL; const char *name; struct pmu *pmu; + int ctxn; memset(tmp, 0, sizeof(tmp)); @@ -4078,10 +4125,17 @@ got_name: cpuctx = this_cpu_ptr(pmu->pmu_cpu_context); perf_event_mmap_ctx(&cpuctx->ctx, mmap_event, vma->vm_flags & VM_EXEC); + + ctxn = pmu->task_ctx_nr; + if (ctxn < 0) + continue; + + ctx = rcu_dereference(current->perf_event_ctxp[ctxn]); + if (ctx) { + perf_event_mmap_ctx(ctx, mmap_event, + vma->vm_flags & VM_EXEC); + } } - ctx = rcu_dereference(current->perf_event_ctxp); - if (ctx) - perf_event_mmap_ctx(ctx, mmap_event, vma->vm_flags & VM_EXEC); rcu_read_unlock_sched(); kfree(buf); @@ -5042,6 +5096,43 @@ static void perf_pmu_cancel_txn(struct pmu *pmu) perf_pmu_enable(pmu); } +/* + * Ensures all contexts with the same task_ctx_nr have the same + * pmu_cpu_context too. + */ +static void *find_pmu_context(int ctxn) +{ + struct pmu *pmu; + + if (ctxn < 0) + return NULL; + + list_for_each_entry(pmu, &pmus, entry) { + if (pmu->task_ctx_nr == ctxn) + return pmu->pmu_cpu_context; + } + + return NULL; +} + +static void free_pmu_context(void * __percpu cpu_context) +{ + struct pmu *pmu; + + mutex_lock(&pmus_lock); + /* + * Like a real lame refcount. + */ + list_for_each_entry(pmu, &pmus, entry) { + if (pmu->pmu_cpu_context == cpu_context) + goto out; + } + + free_percpu(cpu_context); +out: + mutex_unlock(&pmus_lock); +} + int perf_pmu_register(struct pmu *pmu) { int cpu, ret; @@ -5052,6 +5143,10 @@ int perf_pmu_register(struct pmu *pmu) if (!pmu->pmu_disable_count) goto unlock; + pmu->pmu_cpu_context = find_pmu_context(pmu->task_ctx_nr); + if (pmu->pmu_cpu_context) + goto got_cpu_context; + pmu->pmu_cpu_context = alloc_percpu(struct perf_cpu_context); if (!pmu->pmu_cpu_context) goto free_pdc; @@ -5067,6 +5162,7 @@ int perf_pmu_register(struct pmu *pmu) cpuctx->timer.function = perf_event_context_tick; } +got_cpu_context: if (!pmu->start_txn) { if (pmu->pmu_enable) { /* @@ -5114,7 +5210,7 @@ void perf_pmu_unregister(struct pmu *pmu) synchronize_srcu(&pmus_srcu); free_percpu(pmu->pmu_disable_count); - free_percpu(pmu->pmu_cpu_context); + free_pmu_context(pmu->pmu_cpu_context); } struct pmu *perf_init_event(struct perf_event *event) @@ -5628,16 +5724,13 @@ __perf_event_exit_task(struct perf_event *child_event, } } -/* - * When a child task exits, feed back event values to parent events. - */ -void perf_event_exit_task(struct task_struct *child) +static void perf_event_exit_task_context(struct task_struct *child, int ctxn) { struct perf_event *child_event, *tmp; struct perf_event_context *child_ctx; unsigned long flags; - if (likely(!child->perf_event_ctxp)) { + if (likely(!child->perf_event_ctxp[ctxn])) { perf_event_task(child, NULL, 0); return; } @@ -5649,7 +5742,7 @@ void perf_event_exit_task(struct task_struct *child) * scheduled, so we are now safe from rescheduling changing * our context. */ - child_ctx = child->perf_event_ctxp; + child_ctx = child->perf_event_ctxp[ctxn]; __perf_event_task_sched_out(child_ctx); /* @@ -5658,7 +5751,7 @@ void perf_event_exit_task(struct task_struct *child) * incremented the context's refcount before we do put_ctx below. */ raw_spin_lock(&child_ctx->lock); - child->perf_event_ctxp = NULL; + child->perf_event_ctxp[ctxn] = NULL; /* * If this context is a clone; unclone it so it can't get * swapped to another process while we're removing all @@ -5711,6 +5804,17 @@ again: put_ctx(child_ctx); } +/* + * When a child task exits, feed back event values to parent events. + */ +void perf_event_exit_task(struct task_struct *child) +{ + int ctxn; + + for_each_task_context_nr(ctxn) + perf_event_exit_task_context(child, ctxn); +} + static void perf_free_event(struct perf_event *event, struct perf_event_context *ctx) { @@ -5732,32 +5836,37 @@ static void perf_free_event(struct perf_event *event, /* * free an unexposed, unused context as created by inheritance by - * init_task below, used by fork() in case of fail. + * perf_event_init_task below, used by fork() in case of fail. */ void perf_event_free_task(struct task_struct *task) { - struct perf_event_context *ctx = task->perf_event_ctxp; + struct perf_event_context *ctx; struct perf_event *event, *tmp; + int ctxn; - if (!ctx) - return; + for_each_task_context_nr(ctxn) { + ctx = task->perf_event_ctxp[ctxn]; + if (!ctx) + continue; - mutex_lock(&ctx->mutex); + mutex_lock(&ctx->mutex); again: - list_for_each_entry_safe(event, tmp, &ctx->pinned_groups, group_entry) - perf_free_event(event, ctx); + list_for_each_entry_safe(event, tmp, &ctx->pinned_groups, + group_entry) + perf_free_event(event, ctx); - list_for_each_entry_safe(event, tmp, &ctx->flexible_groups, - group_entry) - perf_free_event(event, ctx); + list_for_each_entry_safe(event, tmp, &ctx->flexible_groups, + group_entry) + perf_free_event(event, ctx); - if (!list_empty(&ctx->pinned_groups) || - !list_empty(&ctx->flexible_groups)) - goto again; + if (!list_empty(&ctx->pinned_groups) || + !list_empty(&ctx->flexible_groups)) + goto again; - mutex_unlock(&ctx->mutex); + mutex_unlock(&ctx->mutex); - put_ctx(ctx); + put_ctx(ctx); + } } /* @@ -5863,17 +5972,18 @@ static int inherit_group(struct perf_event *parent_event, static int inherit_task_group(struct perf_event *event, struct task_struct *parent, struct perf_event_context *parent_ctx, - struct task_struct *child, + struct task_struct *child, int ctxn, int *inherited_all) { int ret; - struct perf_event_context *child_ctx = child->perf_event_ctxp; + struct perf_event_context *child_ctx; if (!event->attr.inherit) { *inherited_all = 0; return 0; } + child_ctx = child->perf_event_ctxp[ctxn]; if (!child_ctx) { /* * This is executed from the parent task context, so @@ -5886,7 +5996,7 @@ inherit_task_group(struct perf_event *event, struct task_struct *parent, if (!child_ctx) return -ENOMEM; - child->perf_event_ctxp = child_ctx; + child->perf_event_ctxp[ctxn] = child_ctx; } ret = inherit_group(event, parent, parent_ctx, @@ -5901,7 +6011,7 @@ inherit_task_group(struct perf_event *event, struct task_struct *parent, /* * Initialize the perf_event context in task_struct */ -int perf_event_init_task(struct task_struct *child) +int perf_event_init_context(struct task_struct *child, int ctxn) { struct perf_event_context *child_ctx, *parent_ctx; struct perf_event_context *cloned_ctx; @@ -5910,19 +6020,19 @@ int perf_event_init_task(struct task_struct *child) int inherited_all = 1; int ret = 0; - child->perf_event_ctxp = NULL; + child->perf_event_ctxp[ctxn] = NULL; mutex_init(&child->perf_event_mutex); INIT_LIST_HEAD(&child->perf_event_list); - if (likely(!parent->perf_event_ctxp)) + if (likely(!parent->perf_event_ctxp[ctxn])) return 0; /* * If the parent's context is a clone, pin it so it won't get * swapped under us. */ - parent_ctx = perf_pin_task_context(parent); + parent_ctx = perf_pin_task_context(parent, ctxn); /* * No need to check if parent_ctx != NULL here; since we saw @@ -5942,20 +6052,20 @@ int perf_event_init_task(struct task_struct *child) * the list, not manipulating it: */ list_for_each_entry(event, &parent_ctx->pinned_groups, group_entry) { - ret = inherit_task_group(event, parent, parent_ctx, child, - &inherited_all); + ret = inherit_task_group(event, parent, parent_ctx, + child, ctxn, &inherited_all); if (ret) break; } list_for_each_entry(event, &parent_ctx->flexible_groups, group_entry) { - ret = inherit_task_group(event, parent, parent_ctx, child, - &inherited_all); + ret = inherit_task_group(event, parent, parent_ctx, + child, ctxn, &inherited_all); if (ret) break; } - child_ctx = child->perf_event_ctxp; + child_ctx = child->perf_event_ctxp[ctxn]; if (child_ctx && inherited_all) { /* @@ -5984,6 +6094,22 @@ int perf_event_init_task(struct task_struct *child) return ret; } +/* + * Initialize the perf_event context in task_struct + */ +int perf_event_init_task(struct task_struct *child) +{ + int ctxn, ret; + + for_each_task_context_nr(ctxn) { + ret = perf_event_init_context(child, ctxn); + if (ret) + return ret; + } + + return 0; +} + static void __init perf_event_init_all_cpus(void) { struct swevent_htable *swhash; -- cgit v1.2.3 From 89a1e18731959e9953fae15ddc1a983eb15a4f19 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Tue, 7 Sep 2010 17:34:50 +0200 Subject: perf: Provide a separate task context for swevents Since software events are always schedulable, mixing them up with hardware events (who are not) can lead to funny scheduling oddities. Giving them their own context solves this. Signed-off-by: Peter Zijlstra Cc: paulus Cc: stephane eranian Cc: Robert Richter Cc: Frederic Weisbecker Cc: Lin Ming Cc: Yanmin LKML-Reference: Signed-off-by: Ingo Molnar --- kernel/hw_breakpoint.c | 2 ++ kernel/perf_event.c | 40 +++++++++++++++++++++++++++++----------- 2 files changed, 31 insertions(+), 11 deletions(-) (limited to 'kernel') diff --git a/kernel/hw_breakpoint.c b/kernel/hw_breakpoint.c index 6f150095cafe..3b2aaffb65f0 100644 --- a/kernel/hw_breakpoint.c +++ b/kernel/hw_breakpoint.c @@ -610,6 +610,8 @@ static void hw_breakpoint_stop(struct perf_event *bp, int flags) } static struct pmu perf_breakpoint = { + .task_ctx_nr = perf_sw_context, /* could eventually get its own */ + .event_init = hw_breakpoint_event_init, .add = hw_breakpoint_add, .del = hw_breakpoint_del, diff --git a/kernel/perf_event.c b/kernel/perf_event.c index 7223ea875861..357ee8d5e8ae 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -4709,6 +4709,8 @@ static int perf_swevent_init(struct perf_event *event) } static struct pmu perf_swevent = { + .task_ctx_nr = perf_sw_context, + .event_init = perf_swevent_init, .add = perf_swevent_add, .del = perf_swevent_del, @@ -4800,6 +4802,8 @@ static int perf_tp_event_init(struct perf_event *event) } static struct pmu perf_tracepoint = { + .task_ctx_nr = perf_sw_context, + .event_init = perf_tp_event_init, .add = perf_trace_add, .del = perf_trace_del, @@ -4988,6 +4992,8 @@ static int cpu_clock_event_init(struct perf_event *event) } static struct pmu perf_cpu_clock = { + .task_ctx_nr = perf_sw_context, + .event_init = cpu_clock_event_init, .add = cpu_clock_event_add, .del = cpu_clock_event_del, @@ -5063,6 +5069,8 @@ static int task_clock_event_init(struct perf_event *event) } static struct pmu perf_task_clock = { + .task_ctx_nr = perf_sw_context, + .event_init = task_clock_event_init, .add = task_clock_event_add, .del = task_clock_event_del, @@ -5490,6 +5498,7 @@ SYSCALL_DEFINE5(perf_event_open, struct perf_event_context *ctx; struct file *event_file = NULL; struct file *group_file = NULL; + struct pmu *pmu; int event_fd; int fput_needed = 0; int err; @@ -5522,20 +5531,11 @@ SYSCALL_DEFINE5(perf_event_open, goto err_fd; } - /* - * Get the target context (task or percpu): - */ - ctx = find_get_context(event->pmu, pid, cpu); - if (IS_ERR(ctx)) { - err = PTR_ERR(ctx); - goto err_alloc; - } - if (group_fd != -1) { group_leader = perf_fget_light(group_fd, &fput_needed); if (IS_ERR(group_leader)) { err = PTR_ERR(group_leader); - goto err_context; + goto err_alloc; } group_file = group_leader->filp; if (flags & PERF_FLAG_FD_OUTPUT) @@ -5544,6 +5544,23 @@ SYSCALL_DEFINE5(perf_event_open, group_leader = NULL; } + /* + * Special case software events and allow them to be part of + * any hardware group. + */ + pmu = event->pmu; + if ((pmu->task_ctx_nr == perf_sw_context) && group_leader) + pmu = group_leader->pmu; + + /* + * Get the target context (task or percpu): + */ + ctx = find_get_context(pmu, pid, cpu); + if (IS_ERR(ctx)) { + err = PTR_ERR(ctx); + goto err_group_fd; + } + /* * Look up the group leader (we will attach this event to it): */ @@ -5605,8 +5622,9 @@ SYSCALL_DEFINE5(perf_event_open, return event_fd; err_context: - fput_light(group_file, fput_needed); put_ctx(ctx); +err_group_fd: + fput_light(group_file, fput_needed); err_alloc: free_event(event); err_fd: -- cgit v1.2.3 From 1b9a644fece117cfa5474a2388d6b89d1baf8ddf Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Tue, 7 Sep 2010 18:32:22 +0200 Subject: perf: Optimize context ops Assuming we don't mix events of different pmus onto a single context (with the exeption of software events inside a hardware group) we can now assume that all events on a particular context belong to the same pmu, hence we can disable the pmu for the entire context operations. This reduces the amount of hardware writes. The exception for swevents comes from the fact that the sw pmu disable is a nop. Signed-off-by: Peter Zijlstra Cc: paulus Cc: stephane eranian Cc: Robert Richter Cc: Frederic Weisbecker Cc: Lin Ming Cc: Yanmin LKML-Reference: Signed-off-by: Ingo Molnar --- kernel/perf_event.c | 6 ++++++ 1 file changed, 6 insertions(+) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index 357ee8d5e8ae..9819a69a61a1 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -1065,6 +1065,7 @@ static void ctx_sched_out(struct perf_event_context *ctx, struct perf_event *event; raw_spin_lock(&ctx->lock); + perf_pmu_disable(ctx->pmu); ctx->is_active = 0; if (likely(!ctx->nr_events)) goto out; @@ -1083,6 +1084,7 @@ static void ctx_sched_out(struct perf_event_context *ctx, group_sched_out(event, cpuctx, ctx); } out: + perf_pmu_enable(ctx->pmu); raw_spin_unlock(&ctx->lock); } @@ -1400,6 +1402,7 @@ void perf_event_context_sched_in(struct perf_event_context *ctx) if (cpuctx->task_ctx == ctx) return; + perf_pmu_disable(ctx->pmu); /* * We want to keep the following priority order: * cpu pinned (that don't need to move), task pinned, @@ -1418,6 +1421,7 @@ void perf_event_context_sched_in(struct perf_event_context *ctx) * cpu-context we got scheduled on is actually rotating. */ perf_pmu_rotate_start(ctx->pmu); + perf_pmu_enable(ctx->pmu); } /* @@ -1629,6 +1633,7 @@ static enum hrtimer_restart perf_event_context_tick(struct hrtimer *timer) rotate = 1; } + perf_pmu_disable(cpuctx->ctx.pmu); perf_ctx_adjust_freq(&cpuctx->ctx, cpuctx->timer_interval); if (ctx) perf_ctx_adjust_freq(ctx, cpuctx->timer_interval); @@ -1649,6 +1654,7 @@ static enum hrtimer_restart perf_event_context_tick(struct hrtimer *timer) task_ctx_sched_in(ctx, EVENT_FLEXIBLE); done: + perf_pmu_enable(cpuctx->ctx.pmu); hrtimer_forward_now(timer, ns_to_ktime(cpuctx->timer_interval)); return restart; -- cgit v1.2.3 From 8af3c153baf95374eff20a37f00c59a295b52756 Mon Sep 17 00:00:00 2001 From: Miroslav Lichvar Date: Tue, 7 Sep 2010 16:43:46 +0200 Subject: ntp: Clamp PLL update interval Clamp update interval to reduce PLL gain with low sampling rate (e.g. intermittent network connection) to avoid instability. The clamp roughly corresponds to the loop time constant, it's 8 * poll interval for SHIFT_PLL 2 and 32 * poll interval for SHIFT_PLL 4. This gives good results without affecting the gain in normal conditions where ntpd skips only up to seven consecutive samples. Signed-off-by: Miroslav Lichvar Acked-by: john stultz LKML-Reference: <1283870626-9472-1-git-send-email-mlichvar@redhat.com> Signed-off-by: Thomas Gleixner --- kernel/time/ntp.c | 14 +++++++++++--- 1 file changed, 11 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/time/ntp.c b/kernel/time/ntp.c index c63116863a80..d2321891538f 100644 --- a/kernel/time/ntp.c +++ b/kernel/time/ntp.c @@ -149,10 +149,18 @@ static void ntp_update_offset(long offset) time_reftime = get_seconds(); offset64 = offset; - freq_adj = (offset64 * secs) << - (NTP_SCALE_SHIFT - 2 * (SHIFT_PLL + 2 + time_constant)); + freq_adj = ntp_update_offset_fll(offset64, secs); - freq_adj += ntp_update_offset_fll(offset64, secs); + /* + * Clamp update interval to reduce PLL gain with low + * sampling rate (e.g. intermittent network connection) + * to avoid instability. + */ + if (unlikely(secs > 1 << (SHIFT_PLL + 1 + time_constant))) + secs = 1 << (SHIFT_PLL + 1 + time_constant); + + freq_adj += (offset64 * secs) << + (NTP_SCALE_SHIFT - 2 * (SHIFT_PLL + 2 + time_constant)); freq_adj = min(freq_adj + time_freq, MAXFREQ_SCALED); -- cgit v1.2.3 From 4e231c7962ce711c7d8c2a4dc23ecd1e8fc28363 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Thu, 9 Sep 2010 21:01:59 +0200 Subject: perf: Fix up delayed_put_task_struct() I missed a perf_event_ctxp user when converting it to an array. Pull this last user into perf_event.c as well and fix it up. Signed-off-by: Peter Zijlstra LKML-Reference: Signed-off-by: Ingo Molnar --- kernel/exit.c | 4 +--- kernel/perf_event.c | 8 ++++++++ 2 files changed, 9 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/exit.c b/kernel/exit.c index 03120229db28..e2bdf37f9fde 100644 --- a/kernel/exit.c +++ b/kernel/exit.c @@ -149,9 +149,7 @@ static void delayed_put_task_struct(struct rcu_head *rhp) { struct task_struct *tsk = container_of(rhp, struct task_struct, rcu); -#ifdef CONFIG_PERF_EVENTS - WARN_ON_ONCE(tsk->perf_event_ctxp); -#endif + perf_event_delayed_put(tsk); trace_sched_process_free(tsk); put_task_struct(tsk); } diff --git a/kernel/perf_event.c b/kernel/perf_event.c index 9819a69a61a1..eaf1c5de6dcc 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -5893,6 +5893,14 @@ again: } } +void perf_event_delayed_put(struct task_struct *task) +{ + int ctxn; + + for_each_task_context_nr(ctxn) + WARN_ON_ONCE(task->perf_event_ctxp[ctxn]); +} + /* * inherit a event from parent task to child task: */ -- cgit v1.2.3 From 85a0fdfd0f967507f3903e8419bc7e408f5a59de Mon Sep 17 00:00:00 2001 From: Peter Oberparleiter Date: Thu, 9 Sep 2010 16:37:35 -0700 Subject: gcov: fix null-pointer dereference for certain module types The gcov-kernel infrastructure expects that each object file is loaded only once. This may not be true, e.g. when loading multiple kernel modules which are linked to the same object file. As a result, loading such kernel modules will result in incorrect gcov results while unloading will cause a null-pointer dereference. This patch fixes these problems by changing the gcov-kernel infrastructure so that multiple profiling data sets can be associated with one debugfs entry. It applies to 2.6.36-rc1. Signed-off-by: Peter Oberparleiter Reported-by: Werner Spies Cc: Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/gcov/fs.c | 244 ++++++++++++++++++++++++++++++++++++++++--------------- 1 file changed, 180 insertions(+), 64 deletions(-) (limited to 'kernel') diff --git a/kernel/gcov/fs.c b/kernel/gcov/fs.c index ef3c3f88a7a3..f83972b16564 100644 --- a/kernel/gcov/fs.c +++ b/kernel/gcov/fs.c @@ -33,10 +33,11 @@ * @children: child nodes * @all: list head for list of all nodes * @parent: parent node - * @info: associated profiling data structure if not a directory - * @ghost: when an object file containing profiling data is unloaded we keep a - * copy of the profiling data here to allow collecting coverage data - * for cleanup code. Such a node is called a "ghost". + * @loaded_info: array of pointers to profiling data sets for loaded object + * files. + * @num_loaded: number of profiling data sets for loaded object files. + * @unloaded_info: accumulated copy of profiling data sets for unloaded + * object files. Used only when gcov_persist=1. * @dentry: main debugfs entry, either a directory or data file * @links: associated symbolic links * @name: data file basename @@ -51,10 +52,11 @@ struct gcov_node { struct list_head children; struct list_head all; struct gcov_node *parent; - struct gcov_info *info; - struct gcov_info *ghost; + struct gcov_info **loaded_info; + struct gcov_info *unloaded_info; struct dentry *dentry; struct dentry **links; + int num_loaded; char name[0]; }; @@ -136,16 +138,37 @@ static const struct seq_operations gcov_seq_ops = { }; /* - * Return the profiling data set for a given node. This can either be the - * original profiling data structure or a duplicate (also called "ghost") - * in case the associated object file has been unloaded. + * Return a profiling data set associated with the given node. This is + * either a data set for a loaded object file or a data set copy in case + * all associated object files have been unloaded. */ static struct gcov_info *get_node_info(struct gcov_node *node) { - if (node->info) - return node->info; + if (node->num_loaded > 0) + return node->loaded_info[0]; - return node->ghost; + return node->unloaded_info; +} + +/* + * Return a newly allocated profiling data set which contains the sum of + * all profiling data associated with the given node. + */ +static struct gcov_info *get_accumulated_info(struct gcov_node *node) +{ + struct gcov_info *info; + int i = 0; + + if (node->unloaded_info) + info = gcov_info_dup(node->unloaded_info); + else + info = gcov_info_dup(node->loaded_info[i++]); + if (!info) + return NULL; + for (; i < node->num_loaded; i++) + gcov_info_add(info, node->loaded_info[i]); + + return info; } /* @@ -163,9 +186,10 @@ static int gcov_seq_open(struct inode *inode, struct file *file) mutex_lock(&node_lock); /* * Read from a profiling data copy to minimize reference tracking - * complexity and concurrent access. + * complexity and concurrent access and to keep accumulating multiple + * profiling data sets associated with one node simple. */ - info = gcov_info_dup(get_node_info(node)); + info = get_accumulated_info(node); if (!info) goto out_unlock; iter = gcov_iter_new(info); @@ -225,12 +249,25 @@ static struct gcov_node *get_node_by_name(const char *name) return NULL; } +/* + * Reset all profiling data associated with the specified node. + */ +static void reset_node(struct gcov_node *node) +{ + int i; + + if (node->unloaded_info) + gcov_info_reset(node->unloaded_info); + for (i = 0; i < node->num_loaded; i++) + gcov_info_reset(node->loaded_info[i]); +} + static void remove_node(struct gcov_node *node); /* * write() implementation for gcov data files. Reset profiling data for the - * associated file. If the object file has been unloaded (i.e. this is - * a "ghost" node), remove the debug fs node as well. + * corresponding file. If all associated object files have been unloaded, + * remove the debug fs node as well. */ static ssize_t gcov_seq_write(struct file *file, const char __user *addr, size_t len, loff_t *pos) @@ -245,10 +282,10 @@ static ssize_t gcov_seq_write(struct file *file, const char __user *addr, node = get_node_by_name(info->filename); if (node) { /* Reset counts or remove node for unloaded modules. */ - if (node->ghost) + if (node->num_loaded == 0) remove_node(node); else - gcov_info_reset(node->info); + reset_node(node); } /* Reset counts for open file. */ gcov_info_reset(info); @@ -378,7 +415,10 @@ static void init_node(struct gcov_node *node, struct gcov_info *info, INIT_LIST_HEAD(&node->list); INIT_LIST_HEAD(&node->children); INIT_LIST_HEAD(&node->all); - node->info = info; + if (node->loaded_info) { + node->loaded_info[0] = info; + node->num_loaded = 1; + } node->parent = parent; if (name) strcpy(node->name, name); @@ -394,9 +434,13 @@ static struct gcov_node *new_node(struct gcov_node *parent, struct gcov_node *node; node = kzalloc(sizeof(struct gcov_node) + strlen(name) + 1, GFP_KERNEL); - if (!node) { - pr_warning("out of memory\n"); - return NULL; + if (!node) + goto err_nomem; + if (info) { + node->loaded_info = kcalloc(1, sizeof(struct gcov_info *), + GFP_KERNEL); + if (!node->loaded_info) + goto err_nomem; } init_node(node, info, name, parent); /* Differentiate between gcov data file nodes and directory nodes. */ @@ -416,6 +460,11 @@ static struct gcov_node *new_node(struct gcov_node *parent, list_add(&node->all, &all_head); return node; + +err_nomem: + kfree(node); + pr_warning("out of memory\n"); + return NULL; } /* Remove symbolic links associated with node. */ @@ -441,8 +490,9 @@ static void release_node(struct gcov_node *node) list_del(&node->all); debugfs_remove(node->dentry); remove_links(node); - if (node->ghost) - gcov_info_free(node->ghost); + kfree(node->loaded_info); + if (node->unloaded_info) + gcov_info_free(node->unloaded_info); kfree(node); } @@ -477,7 +527,7 @@ static struct gcov_node *get_child_by_name(struct gcov_node *parent, /* * write() implementation for reset file. Reset all profiling data to zero - * and remove ghost nodes. + * and remove nodes for which all associated object files are unloaded. */ static ssize_t reset_write(struct file *file, const char __user *addr, size_t len, loff_t *pos) @@ -487,8 +537,8 @@ static ssize_t reset_write(struct file *file, const char __user *addr, mutex_lock(&node_lock); restart: list_for_each_entry(node, &all_head, all) { - if (node->info) - gcov_info_reset(node->info); + if (node->num_loaded > 0) + reset_node(node); else if (list_empty(&node->children)) { remove_node(node); /* Several nodes may have gone - restart loop. */ @@ -564,37 +614,115 @@ err_remove: } /* - * The profiling data set associated with this node is being unloaded. Store a - * copy of the profiling data and turn this node into a "ghost". + * Associate a profiling data set with an existing node. Needs to be called + * with node_lock held. */ -static int ghost_node(struct gcov_node *node) +static void add_info(struct gcov_node *node, struct gcov_info *info) { - node->ghost = gcov_info_dup(node->info); - if (!node->ghost) { - pr_warning("could not save data for '%s' (out of memory)\n", - node->info->filename); - return -ENOMEM; + struct gcov_info **loaded_info; + int num = node->num_loaded; + + /* + * Prepare new array. This is done first to simplify cleanup in + * case the new data set is incompatible, the node only contains + * unloaded data sets and there's not enough memory for the array. + */ + loaded_info = kcalloc(num + 1, sizeof(struct gcov_info *), GFP_KERNEL); + if (!loaded_info) { + pr_warning("could not add '%s' (out of memory)\n", + info->filename); + return; + } + memcpy(loaded_info, node->loaded_info, + num * sizeof(struct gcov_info *)); + loaded_info[num] = info; + /* Check if the new data set is compatible. */ + if (num == 0) { + /* + * A module was unloaded, modified and reloaded. The new + * data set replaces the copy of the last one. + */ + if (!gcov_info_is_compatible(node->unloaded_info, info)) { + pr_warning("discarding saved data for %s " + "(incompatible version)\n", info->filename); + gcov_info_free(node->unloaded_info); + node->unloaded_info = NULL; + } + } else { + /* + * Two different versions of the same object file are loaded. + * The initial one takes precedence. + */ + if (!gcov_info_is_compatible(node->loaded_info[0], info)) { + pr_warning("could not add '%s' (incompatible " + "version)\n", info->filename); + kfree(loaded_info); + return; + } } - node->info = NULL; + /* Overwrite previous array. */ + kfree(node->loaded_info); + node->loaded_info = loaded_info; + node->num_loaded = num + 1; +} - return 0; +/* + * Return the index of a profiling data set associated with a node. + */ +static int get_info_index(struct gcov_node *node, struct gcov_info *info) +{ + int i; + + for (i = 0; i < node->num_loaded; i++) { + if (node->loaded_info[i] == info) + return i; + } + return -ENOENT; } /* - * Profiling data for this node has been loaded again. Add profiling data - * from previous instantiation and turn this node into a regular node. + * Save the data of a profiling data set which is being unloaded. */ -static void revive_node(struct gcov_node *node, struct gcov_info *info) +static void save_info(struct gcov_node *node, struct gcov_info *info) { - if (gcov_info_is_compatible(node->ghost, info)) - gcov_info_add(info, node->ghost); + if (node->unloaded_info) + gcov_info_add(node->unloaded_info, info); else { - pr_warning("discarding saved data for '%s' (version changed)\n", + node->unloaded_info = gcov_info_dup(info); + if (!node->unloaded_info) { + pr_warning("could not save data for '%s' " + "(out of memory)\n", info->filename); + } + } +} + +/* + * Disassociate a profiling data set from a node. Needs to be called with + * node_lock held. + */ +static void remove_info(struct gcov_node *node, struct gcov_info *info) +{ + int i; + + i = get_info_index(node, info); + if (i < 0) { + pr_warning("could not remove '%s' (not found)\n", info->filename); + return; } - gcov_info_free(node->ghost); - node->ghost = NULL; - node->info = info; + if (gcov_persist) + save_info(node, info); + /* Shrink array. */ + node->loaded_info[i] = node->loaded_info[node->num_loaded - 1]; + node->num_loaded--; + if (node->num_loaded > 0) + return; + /* Last loaded data set was removed. */ + kfree(node->loaded_info); + node->loaded_info = NULL; + node->num_loaded = 0; + if (!node->unloaded_info) + remove_node(node); } /* @@ -609,30 +737,18 @@ void gcov_event(enum gcov_action action, struct gcov_info *info) node = get_node_by_name(info->filename); switch (action) { case GCOV_ADD: - /* Add new node or revive ghost. */ - if (!node) { + if (node) + add_info(node, info); + else add_node(info); - break; - } - if (gcov_persist) - revive_node(node, info); - else { - pr_warning("could not add '%s' (already exists)\n", - info->filename); - } break; case GCOV_REMOVE: - /* Remove node or turn into ghost. */ - if (!node) { + if (node) + remove_info(node, info); + else { pr_warning("could not remove '%s' (not found)\n", info->filename); - break; } - if (gcov_persist) { - if (!ghost_node(node)) - break; - } - remove_node(node); break; } mutex_unlock(&node_lock); -- cgit v1.2.3 From 31583bb0cf6cc40f2a468a4d2f3b9cbefd24f891 Mon Sep 17 00:00:00 2001 From: "Michael S. Tsirkin" Date: Thu, 9 Sep 2010 16:37:37 -0700 Subject: cgroups: fix API thinko Add cgroup_attach_task_all() The existing cgroup_attach_task_current_cg() API is called by a thread to attach another thread to all of its cgroups; this is unsuitable for cases where a privileged task wants to attach itself to the cgroups of a less privileged one, since the call must be made from the context of the target task. This patch adds a more generic cgroup_attach_task_all() API that allows both the source task and to-be-moved task to be specified. cgroup_attach_task_current_cg() becomes a specialization of the more generic new function. [menage@google.com: rewrote changelog] [akpm@linux-foundation.org: address reviewer comments] Signed-off-by: Michael S. Tsirkin Tested-by: Alex Williamson Acked-by: Paul Menage Cc: Li Zefan Cc: Ben Blum Cc: Sridhar Samudrala Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/cgroup.c | 13 +++++++------ 1 file changed, 7 insertions(+), 6 deletions(-) (limited to 'kernel') diff --git a/kernel/cgroup.c b/kernel/cgroup.c index 192f88c5b0f9..c9483d8f6140 100644 --- a/kernel/cgroup.c +++ b/kernel/cgroup.c @@ -1791,19 +1791,20 @@ out: } /** - * cgroup_attach_task_current_cg - attach task 'tsk' to current task's cgroup + * cgroup_attach_task_all - attach task 'tsk' to all cgroups of task 'from' + * @from: attach to all cgroups of a given task * @tsk: the task to be attached */ -int cgroup_attach_task_current_cg(struct task_struct *tsk) +int cgroup_attach_task_all(struct task_struct *from, struct task_struct *tsk) { struct cgroupfs_root *root; - struct cgroup *cur_cg; int retval = 0; cgroup_lock(); for_each_active_root(root) { - cur_cg = task_cgroup_from_root(current, root); - retval = cgroup_attach_task(cur_cg, tsk); + struct cgroup *from_cg = task_cgroup_from_root(from, root); + + retval = cgroup_attach_task(from_cg, tsk); if (retval) break; } @@ -1811,7 +1812,7 @@ int cgroup_attach_task_current_cg(struct task_struct *tsk) return retval; } -EXPORT_SYMBOL_GPL(cgroup_attach_task_current_cg); +EXPORT_SYMBOL_GPL(cgroup_attach_task_all); /* * Attach task with pid 'pid' to cgroup 'cgrp'. Call with cgroup_mutex -- cgit v1.2.3 From 1c24de60e50fb19b94d94225458da17c720f0729 Mon Sep 17 00:00:00 2001 From: Jerome Marchand Date: Thu, 9 Sep 2010 16:37:59 -0700 Subject: kernel/groups.c: fix integer overflow in groups_search gid_t is a unsigned int. If group_info contains a gid greater than MAX_INT, groups_search() function may look on the wrong side of the search tree. This solves some unfair "permission denied" problems. Signed-off-by: Jerome Marchand Cc: Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/groups.c | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/groups.c b/kernel/groups.c index 53b1916c9492..253dc0f35cf4 100644 --- a/kernel/groups.c +++ b/kernel/groups.c @@ -143,10 +143,9 @@ int groups_search(const struct group_info *group_info, gid_t grp) right = group_info->ngroups; while (left < right) { unsigned int mid = (left+right)/2; - int cmp = grp - GROUP_AT(group_info, mid); - if (cmp > 0) + if (grp > GROUP_AT(group_info, mid)) left = mid + 1; - else if (cmp < 0) + else if (grp < GROUP_AT(group_info, mid)) right = mid; else return 1; -- cgit v1.2.3 From 910321ea817a202ff70fac666e37e2c8e2f88823 Mon Sep 17 00:00:00 2001 From: Hugh Dickins Date: Thu, 9 Sep 2010 16:38:07 -0700 Subject: swap: revert special hibernation allocation Please revert 2.6.36-rc commit d2997b1042ec150616c1963b5e5e919ffd0b0ebf "hibernation: freeze swap at hibernation". It complicated matters by adding a second swap allocation path, just for hibernation; without in any way fixing the issue that it was intended to address - page reclaim after fixing the hibernation image might free swap from a page already imaged as swapcache, letting its swap be reallocated to store a different page of the image: resulting in data corruption if the imaged page were freed as clean then swapped back in. Pages freed to si->swap_map were still in danger of being reallocated by the alternative allocation path. I guess it inadvertently fixed slow SSD swap allocation for hibernation, as reported by Nigel Cunningham: by missing out the discards that occur on the usual swap allocation path; but that was unintentional, and needs a separate fix. Signed-off-by: Hugh Dickins Cc: KAMEZAWA Hiroyuki Cc: KOSAKI Motohiro Cc: "Rafael J. Wysocki" Cc: Ondrej Zary Cc: Andrea Gelmini Cc: Balbir Singh Cc: Andrea Arcangeli Cc: Nigel Cunningham Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/power/hibernate.c | 1 - kernel/power/snapshot.c | 1 - kernel/power/swap.c | 6 +++--- 3 files changed, 3 insertions(+), 5 deletions(-) (limited to 'kernel') diff --git a/kernel/power/hibernate.c b/kernel/power/hibernate.c index c77963938bca..8dc31e02ae12 100644 --- a/kernel/power/hibernate.c +++ b/kernel/power/hibernate.c @@ -338,7 +338,6 @@ int hibernation_snapshot(int platform_mode) goto Close; suspend_console(); - hibernation_freeze_swap(); saved_mask = clear_gfp_allowed_mask(GFP_IOFS); error = dpm_suspend_start(PMSG_FREEZE); if (error) diff --git a/kernel/power/snapshot.c b/kernel/power/snapshot.c index 5e7edfb05e66..f6cd6faf84fd 100644 --- a/kernel/power/snapshot.c +++ b/kernel/power/snapshot.c @@ -1086,7 +1086,6 @@ void swsusp_free(void) buffer = NULL; alloc_normal = 0; alloc_highmem = 0; - hibernation_thaw_swap(); } /* Helper functions used for the shrinking of memory. */ diff --git a/kernel/power/swap.c b/kernel/power/swap.c index 5d0059eed3e4..e6a5bdf61a37 100644 --- a/kernel/power/swap.c +++ b/kernel/power/swap.c @@ -136,10 +136,10 @@ sector_t alloc_swapdev_block(int swap) { unsigned long offset; - offset = swp_offset(get_swap_for_hibernation(swap)); + offset = swp_offset(get_swap_page_of_type(swap)); if (offset) { if (swsusp_extents_insert(offset)) - swap_free_for_hibernation(swp_entry(swap, offset)); + swap_free(swp_entry(swap, offset)); else return swapdev_block(swap, offset); } @@ -163,7 +163,7 @@ void free_all_swap_pages(int swap) ext = container_of(node, struct swsusp_extent, node); rb_erase(node, &swsusp_extents); for (offset = ext->start; offset <= ext->end; offset++) - swap_free_for_hibernation(swp_entry(swap, offset)); + swap_free(swp_entry(swap, offset)); kfree(ext); } -- cgit v1.2.3 From df09162550fbb53354f0c88e85b5d0e6129ee9cc Mon Sep 17 00:00:00 2001 From: Chris Wright Date: Thu, 9 Sep 2010 16:34:59 -0700 Subject: tracing: t_start: reset FTRACE_ITER_HASH in case of seek/pread Be sure to avoid entering t_show() with FTRACE_ITER_HASH set without having properly started the iterator to iterate the hash. This case is degenerate and, as discovered by Robert Swiecki, can cause t_hash_show() to misuse a pointer. This causes a NULL ptr deref with possible security implications. Tracked as CVE-2010-3079. Cc: Robert Swiecki Cc: Eugene Teo Cc: Signed-off-by: Chris Wright Signed-off-by: Steven Rostedt --- kernel/trace/ftrace.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'kernel') diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index 83a16e9ee518..fa7ece649fe1 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -1510,6 +1510,8 @@ static void *t_start(struct seq_file *m, loff_t *pos) if (*pos > 0) return t_hash_start(m, pos); iter->flags |= FTRACE_ITER_PRINTALL; + /* reset in case of seek/pread */ + iter->flags &= ~FTRACE_ITER_HASH; return iter; } -- cgit v1.2.3 From cee010ec5211b96f33c5c2208f5c14ebb04b634a Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Fri, 10 Sep 2010 12:51:54 +0200 Subject: perf: Ensure we call add_event_to_ctx() with the right locks held Even though we call it from the inherit path, where the child is not yet accessible, we need to hold ctx->lock, add_event_to_ctx() assumes IRQs are disabled. Signed-off-by: Peter Zijlstra LKML-Reference: Signed-off-by: Ingo Molnar --- kernel/perf_event.c | 3 +++ 1 file changed, 3 insertions(+) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index eaf1c5de6dcc..f395fb4d9b74 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -5913,6 +5913,7 @@ inherit_event(struct perf_event *parent_event, struct perf_event_context *child_ctx) { struct perf_event *child_event; + unsigned long flags; /* * Instead of creating recursive hierarchies of events, @@ -5957,7 +5958,9 @@ inherit_event(struct perf_event *parent_event, /* * Link it up in the child's context: */ + raw_spin_lock_irqsave(&child_ctx->lock, flags); add_event_to_ctx(child_event, child_ctx); + raw_spin_unlock_irqrestore(&child_ctx->lock, flags); /* * Get a reference to the parent filp - we will fput it -- cgit v1.2.3 From 27c379f7f89a4d558c685b5d89b5ba2fe79ae701 Mon Sep 17 00:00:00 2001 From: Heiko Carstens Date: Fri, 10 Sep 2010 13:47:29 +0200 Subject: generic-ipi: Fix deadlock in __smp_call_function_single Just got my 6 way machine to a state where cpu 0 is in an endless loop within __smp_call_function_single. All other cpus are idle. The call trace on cpu 0 looks like this: __smp_call_function_single scheduler_tick update_process_times tick_sched_timer __run_hrtimer hrtimer_interrupt clock_comparator_work do_extint ext_int_handler ----> timer irq cpu_idle __smp_call_function_single() got called from nohz_balancer_kick() (inlined) with the remote cpu being 1, wait being 0 and the per cpu variable remote_sched_softirq_cb (call_single_data) of the current cpu (0). Then it loops forever when it tries to grab the lock of the call_single_data, since it is already locked and enqueued on cpu 0. My theory how this could have happened: for some reason the scheduler decided to call __smp_call_function_single() on it's own cpu, and sends an IPI to itself. The interrupt stays pending since IRQs are disabled. If then the hypervisor schedules the cpu away it might happen that upon rescheduling both the IPI and the timer IRQ are pending. If then interrupts are enabled again it depends which one gets scheduled first. If the timer interrupt gets delivered first we end up with the local deadlock as seen in the calltrace above. Let's make __smp_call_function_single() check if the target cpu is the current cpu and execute the function immediately just like smp_call_function_single does. That should prevent at least the scenario described here. It might also be that the scheduler is not supposed to call __smp_call_function_single with the remote cpu being the current cpu, but that is a different issue. Signed-off-by: Heiko Carstens Acked-by: Peter Zijlstra Acked-by: Jens Axboe Cc: Venkatesh Pallipadi Cc: Suresh Siddha LKML-Reference: <20100910114729.GB2827@osiris.boeblingen.de.ibm.com> Signed-off-by: Ingo Molnar --- kernel/smp.c | 17 ++++++++++++++--- 1 file changed, 14 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/smp.c b/kernel/smp.c index 75c970c715d3..ed6aacfcb7ef 100644 --- a/kernel/smp.c +++ b/kernel/smp.c @@ -365,9 +365,10 @@ call: EXPORT_SYMBOL_GPL(smp_call_function_any); /** - * __smp_call_function_single(): Run a function on another CPU + * __smp_call_function_single(): Run a function on a specific CPU * @cpu: The CPU to run on. * @data: Pre-allocated and setup data structure + * @wait: If true, wait until function has completed on specified CPU. * * Like smp_call_function_single(), but allow caller to pass in a * pre-allocated data structure. Useful for embedding @data inside @@ -376,8 +377,10 @@ EXPORT_SYMBOL_GPL(smp_call_function_any); void __smp_call_function_single(int cpu, struct call_single_data *data, int wait) { - csd_lock(data); + unsigned int this_cpu; + unsigned long flags; + this_cpu = get_cpu(); /* * Can deadlock when called with interrupts disabled. * We allow cpu's that are not yet online though, as no one else can @@ -387,7 +390,15 @@ void __smp_call_function_single(int cpu, struct call_single_data *data, WARN_ON_ONCE(cpu_online(smp_processor_id()) && wait && irqs_disabled() && !oops_in_progress); - generic_exec_single(cpu, data, wait); + if (cpu == this_cpu) { + local_irq_save(flags); + data->func(data->info); + local_irq_restore(flags); + } else { + csd_lock(data); + generic_exec_single(cpu, data, wait); + } + put_cpu(); } /** -- cgit v1.2.3 From e5f4d3394a52ac351f1a479fe136d92fa5228eff Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Fri, 10 Sep 2010 17:38:06 +0200 Subject: perf: Fix perf_init_event() We ought to return -ENOENT when non of the registered PMUs recognise the requested event. This fixes a boot crash that occurs if no PMU is available but the NMI watchdog tries to register an event. Reported-by: Ingo Molnar Signed-off-by: Peter Zijlstra LKML-Reference: Signed-off-by: Ingo Molnar --- kernel/perf_event.c | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index f395fb4d9b74..f29b52576ec1 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -5236,12 +5236,15 @@ struct pmu *perf_init_event(struct perf_event *event) list_for_each_entry_rcu(pmu, &pmus, entry) { int ret = pmu->event_init(event); if (!ret) - break; + goto unlock; + if (ret != -ENOENT) { pmu = ERR_PTR(ret); - break; + goto unlock; } } + pmu = ERR_PTR(-ENOENT); +unlock: srcu_read_unlock(&pmus_srcu, idx); return pmu; -- cgit v1.2.3 From 0109c2c48d062a04685638926a35ed20153fedc8 Mon Sep 17 00:00:00 2001 From: mark gross Date: Thu, 9 Sep 2010 23:20:09 +0200 Subject: PM QoS: Correct pr_debug() misuse and improve parameter checks Correct some pr_debug() misuse and add a stronger parameter check to pm_qos_write() for the ASCII hex value case. Thanks to Dan Carpenter for pointing out the problem! Signed-off-by: mark gross Signed-off-by: Rafael J. Wysocki --- kernel/pm_qos_params.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/pm_qos_params.c b/kernel/pm_qos_params.c index b7e4c362361b..645e541a45f6 100644 --- a/kernel/pm_qos_params.c +++ b/kernel/pm_qos_params.c @@ -389,10 +389,12 @@ static ssize_t pm_qos_power_write(struct file *filp, const char __user *buf, } else if (count == 11) { /* len('0x12345678/0') */ if (copy_from_user(ascii_value, buf, 11)) return -EFAULT; + if (strlen(ascii_value) != 10) + return -EINVAL; x = sscanf(ascii_value, "%x", &value); if (x != 1) return -EINVAL; - pr_debug(KERN_ERR "%s, %d, 0x%x\n", ascii_value, x, value); + pr_debug("%s, %d, 0x%x\n", ascii_value, x, value); } else return -EINVAL; -- cgit v1.2.3 From 6715045ddc7472a22be5e49d4047d2d89b391f45 Mon Sep 17 00:00:00 2001 From: "Rafael J. Wysocki" Date: Sat, 11 Sep 2010 20:58:27 +0200 Subject: PM / Hibernate: Avoid hitting OOM during preallocation of memory There is a problem in hibernate_preallocate_memory() that it calls preallocate_image_memory() with an argument that may be greater than the total number of available non-highmem memory pages. If that's the case, the OOM condition is guaranteed to trigger, which in turn can cause significant slowdown to occur during hibernation. To avoid that, make preallocate_image_memory() adjust its argument before calling preallocate_image_pages(), so that the total number of saveable non-highem pages left is not less than the minimum size of a hibernation image. Change hibernate_preallocate_memory() to try to allocate from highmem if the number of pages allocated by preallocate_image_memory() is too low. Modify free_unnecessary_pages() to take all possible memory allocation patterns into account. Reported-by: KOSAKI Motohiro Signed-off-by: Rafael J. Wysocki Tested-by: M. Vefa Bicakci --- kernel/power/snapshot.c | 85 +++++++++++++++++++++++++++++++++++++------------ 1 file changed, 65 insertions(+), 20 deletions(-) (limited to 'kernel') diff --git a/kernel/power/snapshot.c b/kernel/power/snapshot.c index 5e7edfb05e66..5209b39e6982 100644 --- a/kernel/power/snapshot.c +++ b/kernel/power/snapshot.c @@ -1122,9 +1122,19 @@ static unsigned long preallocate_image_pages(unsigned long nr_pages, gfp_t mask) return nr_alloc; } -static unsigned long preallocate_image_memory(unsigned long nr_pages) +static unsigned long preallocate_image_memory(unsigned long nr_pages, + unsigned long avail_normal) { - return preallocate_image_pages(nr_pages, GFP_IMAGE); + unsigned long alloc; + + if (avail_normal <= alloc_normal) + return 0; + + alloc = avail_normal - alloc_normal; + if (nr_pages < alloc) + alloc = nr_pages; + + return preallocate_image_pages(alloc, GFP_IMAGE); } #ifdef CONFIG_HIGHMEM @@ -1170,15 +1180,22 @@ static inline unsigned long preallocate_highmem_fraction(unsigned long nr_pages, */ static void free_unnecessary_pages(void) { - unsigned long save_highmem, to_free_normal, to_free_highmem; + unsigned long save, to_free_normal, to_free_highmem; - to_free_normal = alloc_normal - count_data_pages(); - save_highmem = count_highmem_pages(); - if (alloc_highmem > save_highmem) { - to_free_highmem = alloc_highmem - save_highmem; + save = count_data_pages(); + if (alloc_normal >= save) { + to_free_normal = alloc_normal - save; + save = 0; + } else { + to_free_normal = 0; + save -= alloc_normal; + } + save += count_highmem_pages(); + if (alloc_highmem >= save) { + to_free_highmem = alloc_highmem - save; } else { to_free_highmem = 0; - to_free_normal -= save_highmem - alloc_highmem; + to_free_normal -= save - alloc_highmem; } memory_bm_position_reset(©_bm); @@ -1259,7 +1276,7 @@ int hibernate_preallocate_memory(void) { struct zone *zone; unsigned long saveable, size, max_size, count, highmem, pages = 0; - unsigned long alloc, save_highmem, pages_highmem; + unsigned long alloc, save_highmem, pages_highmem, avail_normal; struct timeval start, stop; int error; @@ -1296,6 +1313,7 @@ int hibernate_preallocate_memory(void) else count += zone_page_state(zone, NR_FREE_PAGES); } + avail_normal = count; count += highmem; count -= totalreserve_pages; @@ -1310,12 +1328,21 @@ int hibernate_preallocate_memory(void) */ if (size >= saveable) { pages = preallocate_image_highmem(save_highmem); - pages += preallocate_image_memory(saveable - pages); + pages += preallocate_image_memory(saveable - pages, avail_normal); goto out; } /* Estimate the minimum size of the image. */ pages = minimum_image_size(saveable); + /* + * To avoid excessive pressure on the normal zone, leave room in it to + * accommodate an image of the minimum size (unless it's already too + * small, in which case don't preallocate pages from it at all). + */ + if (avail_normal > pages) + avail_normal -= pages; + else + avail_normal = 0; if (size < pages) size = min_t(unsigned long, pages, max_size); @@ -1336,16 +1363,34 @@ int hibernate_preallocate_memory(void) */ pages_highmem = preallocate_image_highmem(highmem / 2); alloc = (count - max_size) - pages_highmem; - pages = preallocate_image_memory(alloc); - if (pages < alloc) - goto err_out; - size = max_size - size; - alloc = size; - size = preallocate_highmem_fraction(size, highmem, count); - pages_highmem += size; - alloc -= size; - pages += preallocate_image_memory(alloc); - pages += pages_highmem; + pages = preallocate_image_memory(alloc, avail_normal); + if (pages < alloc) { + /* We have exhausted non-highmem pages, try highmem. */ + alloc -= pages; + pages += pages_highmem; + pages_highmem = preallocate_image_highmem(alloc); + if (pages_highmem < alloc) + goto err_out; + pages += pages_highmem; + /* + * size is the desired number of saveable pages to leave in + * memory, so try to preallocate (all memory - size) pages. + */ + alloc = (count - pages) - size; + pages += preallocate_image_highmem(alloc); + } else { + /* + * There are approximately max_size saveable pages at this point + * and we want to reduce this number down to size. + */ + alloc = max_size - size; + size = preallocate_highmem_fraction(alloc, highmem, count); + pages_highmem += size; + alloc -= size; + size = preallocate_image_memory(alloc, avail_normal); + pages_highmem += preallocate_image_highmem(alloc - size); + pages += pages_highmem + size; + } /* * We only need as many page frames for the image as there are saveable -- cgit v1.2.3 From c54fce6eff197d9c57c97afbf6c9722ce434fc8f Mon Sep 17 00:00:00 2001 From: Tejun Heo Date: Fri, 10 Sep 2010 16:51:36 +0200 Subject: workqueue: add documentation Update copyright notice and add Documentation/workqueue.txt. Randy Dunlap, Dave Chinner: misc fixes. Signed-off-by: Tejun Heo Reviewed-By: Florian Mickler Cc: Ingo Molnar Cc: Christoph Lameter Cc: Randy Dunlap Cc: Dave Chinner --- kernel/workqueue.c | 27 +++++++++++++++++---------- 1 file changed, 17 insertions(+), 10 deletions(-) (limited to 'kernel') diff --git a/kernel/workqueue.c b/kernel/workqueue.c index 727f24e563ae..f77afd939229 100644 --- a/kernel/workqueue.c +++ b/kernel/workqueue.c @@ -1,19 +1,26 @@ /* - * linux/kernel/workqueue.c + * kernel/workqueue.c - generic async execution with shared worker pool * - * Generic mechanism for defining kernel helper threads for running - * arbitrary tasks in process context. + * Copyright (C) 2002 Ingo Molnar * - * Started by Ingo Molnar, Copyright (C) 2002 + * Derived from the taskqueue/keventd code by: + * David Woodhouse + * Andrew Morton + * Kai Petzke + * Theodore Ts'o * - * Derived from the taskqueue/keventd code by: + * Made to use alloc_percpu by Christoph Lameter. * - * David Woodhouse - * Andrew Morton - * Kai Petzke - * Theodore Ts'o + * Copyright (C) 2010 SUSE Linux Products GmbH + * Copyright (C) 2010 Tejun Heo * - * Made to use alloc_percpu by Christoph Lameter. + * This is the generic async execution mechanism. Work items as are + * executed in process context. The worker pool is shared and + * automatically managed. There is one worker pool for each CPU and + * one extra for works which are better served by workers which are + * not bound to any specific CPU. + * + * Please read Documentation/workqueue.txt for details. */ #include -- cgit v1.2.3 From cde8e88498c8de69271fcb6d4dd974979368fa67 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Mon, 13 Sep 2010 11:06:55 +0200 Subject: perf: Sanitize the RCU logic Simplify things and simply synchronize against two RCU variants for PMU unregister -- we don't care about performance, its module unload if anything. Reported-by: Frederic Weisbecker Signed-off-by: Peter Zijlstra Cc: Paul E. McKenney LKML-Reference: Signed-off-by: Ingo Molnar --- kernel/perf_event.c | 17 +++++++++-------- 1 file changed, 9 insertions(+), 8 deletions(-) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index f29b52576ec1..bc46bff69620 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -3810,7 +3810,7 @@ static void perf_event_task_event(struct perf_task_event *task_event) struct pmu *pmu; int ctxn; - rcu_read_lock_sched(); + rcu_read_lock(); list_for_each_entry_rcu(pmu, &pmus, entry) { cpuctx = this_cpu_ptr(pmu->pmu_cpu_context); perf_event_task_ctx(&cpuctx->ctx, task_event); @@ -3825,7 +3825,7 @@ static void perf_event_task_event(struct perf_task_event *task_event) if (ctx) perf_event_task_ctx(ctx, task_event); } - rcu_read_unlock_sched(); + rcu_read_unlock(); } static void perf_event_task(struct task_struct *task, @@ -3943,7 +3943,7 @@ static void perf_event_comm_event(struct perf_comm_event *comm_event) comm_event->event_id.header.size = sizeof(comm_event->event_id) + size; - rcu_read_lock_sched(); + rcu_read_lock(); list_for_each_entry_rcu(pmu, &pmus, entry) { cpuctx = this_cpu_ptr(pmu->pmu_cpu_context); perf_event_comm_ctx(&cpuctx->ctx, comm_event); @@ -3956,7 +3956,7 @@ static void perf_event_comm_event(struct perf_comm_event *comm_event) if (ctx) perf_event_comm_ctx(ctx, comm_event); } - rcu_read_unlock_sched(); + rcu_read_unlock(); } void perf_event_comm(struct task_struct *task) @@ -4126,7 +4126,7 @@ got_name: mmap_event->event_id.header.size = sizeof(mmap_event->event_id) + size; - rcu_read_lock_sched(); + rcu_read_lock(); list_for_each_entry_rcu(pmu, &pmus, entry) { cpuctx = this_cpu_ptr(pmu->pmu_cpu_context); perf_event_mmap_ctx(&cpuctx->ctx, mmap_event, @@ -4142,7 +4142,7 @@ got_name: vma->vm_flags & VM_EXEC); } } - rcu_read_unlock_sched(); + rcu_read_unlock(); kfree(buf); } @@ -5218,10 +5218,11 @@ void perf_pmu_unregister(struct pmu *pmu) mutex_unlock(&pmus_lock); /* - * We use the pmu list either under SRCU or preempt_disable, - * synchronize_srcu() implies synchronize_sched() so we're good. + * We dereference the pmu list under both SRCU and regular RCU, so + * synchronize against both of those. */ synchronize_srcu(&pmus_srcu); + synchronize_rcu(); free_percpu(pmu->pmu_disable_count); free_pmu_context(pmu->pmu_cpu_context); -- cgit v1.2.3 From 0c67b40872326a5340cab51d79a192a5fbaeb484 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Mon, 13 Sep 2010 11:15:58 +0200 Subject: perf: Fix free_event() With the context rework stuff we can actually end up freeing an event before it gets attached to a context. Reported-by: Cyrill Gorcunov Signed-off-by: Peter Zijlstra LKML-Reference: Signed-off-by: Ingo Molnar --- kernel/perf_event.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index bc46bff69620..440f9ca067b2 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -2150,7 +2150,9 @@ static void free_event(struct perf_event *event) if (event->destroy) event->destroy(event); - put_ctx(event->ctx); + if (event->ctx) + put_ctx(event->ctx); + call_rcu(&event->rcu_head, free_event_rcu); } -- cgit v1.2.3 From 0bf377bbb0bea6130f35613491887cc622e42a8b Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Sun, 12 Sep 2010 08:14:52 +0200 Subject: sched: Improve latencies under load by decreasing minimum scheduling granularity MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Mathieu reported bad latencies with make -j10 kind of kbuild workloads - which is mostly caused by us scheduling with a too coarse granularity. Reduce the minimum granularity some more, to make sure we can meet the latency target. I got the following results (make -j10 kbuild load, average of 3 runs): vanilla: maximum latency: 38278.9 µs average latency: 7730.1 µs patched: maximum latency: 22702.1 µs average latency: 6684.8 µs Mathieu also measured it: | | * wakeup-latency.c (SIGEV_THREAD) with make -j10 | | - Mainline 2.6.35.2 kernel | | maximum latency: 45762.1 µs | average latency: 7348.6 µs | | - With only Peter's smaller min_gran (shown below): | | maximum latency: 29100.6 µs | average latency: 6684.1 µs | Reported-by: Mathieu Desnoyers Reported-by: Linus Torvalds Acked-by: Mathieu Desnoyers Suggested-by: Peter Zijlstra Acked-by: Peter Zijlstra LKML-Reference: Signed-off-by: Ingo Molnar --- kernel/sched_fair.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c index 9b5b4f86b742..a171138a9402 100644 --- a/kernel/sched_fair.c +++ b/kernel/sched_fair.c @@ -54,13 +54,13 @@ enum sched_tunable_scaling sysctl_sched_tunable_scaling * Minimal preemption granularity for CPU-bound tasks: * (default: 2 msec * (1 + ilog(ncpus)), units: nanoseconds) */ -unsigned int sysctl_sched_min_granularity = 2000000ULL; -unsigned int normalized_sysctl_sched_min_granularity = 2000000ULL; +unsigned int sysctl_sched_min_granularity = 750000ULL; +unsigned int normalized_sysctl_sched_min_granularity = 750000ULL; /* * is kept at sysctl_sched_latency / sysctl_sched_min_granularity */ -static unsigned int sched_nr_latency = 3; +static unsigned int sched_nr_latency = 8; /* * After fork, child runs first. If set to 0 (default) then -- cgit v1.2.3 From 7740191cd909b75d75685fb08a5d1f54b8a9d28b Mon Sep 17 00:00:00 2001 From: Mathieu Desnoyers Date: Mon, 13 Sep 2010 17:47:00 -0400 Subject: sched: Fix string comparison in /proc/sched_features Fix incorrect handling of the following case: INTERACTIVE INTERACTIVE_SOMETHING_ELSE The comparison only checks up to each element's length. Changelog since v1: - Embellish using some Rostedtisms. [ mingo: ^^ == smaller and cleaner ] Signed-off-by: Mathieu Desnoyers Reviewed-by: Steven Rostedt Cc: Cc: Peter Zijlstra Cc: Tony Lindgren LKML-Reference: <20100913214700.GB16118@Krystal> Signed-off-by: Ingo Molnar --- kernel/sched.c | 7 +++---- 1 file changed, 3 insertions(+), 4 deletions(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index 26f83e2f1534..b40b82e33590 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -721,7 +721,7 @@ sched_feat_write(struct file *filp, const char __user *ubuf, size_t cnt, loff_t *ppos) { char buf[64]; - char *cmp = buf; + char *cmp; int neg = 0; int i; @@ -732,6 +732,7 @@ sched_feat_write(struct file *filp, const char __user *ubuf, return -EFAULT; buf[cnt] = 0; + cmp = strstrip(buf); if (strncmp(buf, "NO_", 3) == 0) { neg = 1; @@ -739,9 +740,7 @@ sched_feat_write(struct file *filp, const char __user *ubuf, } for (i = 0; sched_feat_names[i]; i++) { - int len = strlen(sched_feat_names[i]); - - if (strncmp(cmp, sched_feat_names[i], len) == 0) { + if (strcmp(cmp, sched_feat_names[i]) == 0) { if (neg) sysctl_sched_features &= ~(1UL << i); else -- cgit v1.2.3 From 2bccfffd1538f3523847583213567e2f7ce00926 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Thu, 9 Sep 2010 08:43:22 -0400 Subject: tracing: Do not reset *pos in set_ftrace_filter After the filtered functions are read, the probed functions are read from the hash in set_ftrace_filter. When the hashed probed functions are read, the *pos passed in is reset. Instead of modifying the pos given to the read function, just record the pos where the filtered functions ended and subtract from that. Signed-off-by: Steven Rostedt --- kernel/trace/ftrace.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index fa7ece649fe1..585ea27025b1 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -1368,6 +1368,7 @@ enum { #define FTRACE_BUFF_MAX (KSYM_SYMBOL_LEN+4) /* room for wildcards */ struct ftrace_iterator { + loff_t func_pos; struct ftrace_page *pg; int hidx; int idx; @@ -1418,12 +1419,15 @@ static void *t_hash_start(struct seq_file *m, loff_t *pos) loff_t l; if (!(iter->flags & FTRACE_ITER_HASH)) - *pos = 0; + iter->func_pos = *pos; + + if (iter->func_pos > *pos) + return NULL; iter->flags |= FTRACE_ITER_HASH; iter->hidx = 0; - for (l = 0; l <= *pos; ) { + for (l = 0; l <= (*pos - iter->func_pos); ) { p = t_hash_next(m, p, &l); if (!p) break; -- cgit v1.2.3 From 4aeb69672d011fac5c8df671f3ca89f7987c104e Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Thu, 9 Sep 2010 10:00:28 -0400 Subject: tracing: Replace typecasted void pointer in set_ftrace_filter code The set_ftrace_filter uses seq_file and reads from two lists. The pointer returned by t_next() can either be of type struct dyn_ftrace or struct ftrace_func_probe. If there is a bug (there was one) the wrong pointer may be used and the reference can cause an oops. This patch makes t_next() and friends only return the iterator structure which now has a pointer of type struct dyn_ftrace and struct ftrace_func_probe. The t_show() can now test if the pointer is NULL or not and if the pointer exists, it is guaranteed to be of the correct type. Now if there's a bug, only wrong data will be shown but not an oops. Cc: Chris Wright Signed-off-by: Steven Rostedt --- kernel/trace/ftrace.c | 67 +++++++++++++++++++++++++++++++++++---------------- 1 file changed, 46 insertions(+), 21 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index 585ea27025b1..c8db0dbb984e 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -1368,25 +1368,29 @@ enum { #define FTRACE_BUFF_MAX (KSYM_SYMBOL_LEN+4) /* room for wildcards */ struct ftrace_iterator { - loff_t func_pos; - struct ftrace_page *pg; - int hidx; - int idx; - unsigned flags; - struct trace_parser parser; + loff_t func_pos; + struct ftrace_page *pg; + struct dyn_ftrace *func; + struct ftrace_func_probe *probe; + struct trace_parser parser; + int hidx; + int idx; + unsigned flags; }; static void * -t_hash_next(struct seq_file *m, void *v, loff_t *pos) +t_hash_next(struct seq_file *m, loff_t *pos) { struct ftrace_iterator *iter = m->private; - struct hlist_node *hnd = v; + struct hlist_node *hnd = NULL; struct hlist_head *hhd; WARN_ON(!(iter->flags & FTRACE_ITER_HASH)); (*pos)++; + if (iter->probe) + hnd = &iter->probe->node; retry: if (iter->hidx >= FTRACE_FUNC_HASHSIZE) return NULL; @@ -1409,7 +1413,12 @@ t_hash_next(struct seq_file *m, void *v, loff_t *pos) } } - return hnd; + if (WARN_ON_ONCE(!hnd)) + return NULL; + + iter->probe = hlist_entry(hnd, struct ftrace_func_probe, node); + + return iter; } static void *t_hash_start(struct seq_file *m, loff_t *pos) @@ -1428,19 +1437,24 @@ static void *t_hash_start(struct seq_file *m, loff_t *pos) iter->hidx = 0; for (l = 0; l <= (*pos - iter->func_pos); ) { - p = t_hash_next(m, p, &l); + p = t_hash_next(m, &l); if (!p) break; } - return p; + if (!p) + return NULL; + + return iter; } -static int t_hash_show(struct seq_file *m, void *v) +static int +t_hash_show(struct seq_file *m, struct ftrace_iterator *iter) { struct ftrace_func_probe *rec; - struct hlist_node *hnd = v; - rec = hlist_entry(hnd, struct ftrace_func_probe, node); + rec = iter->probe; + if (WARN_ON_ONCE(!rec)) + return -EIO; if (rec->ops->print) return rec->ops->print(m, rec->ip, rec->ops, rec->data); @@ -1461,7 +1475,7 @@ t_next(struct seq_file *m, void *v, loff_t *pos) struct dyn_ftrace *rec = NULL; if (iter->flags & FTRACE_ITER_HASH) - return t_hash_next(m, v, pos); + return t_hash_next(m, pos); (*pos)++; @@ -1495,7 +1509,12 @@ t_next(struct seq_file *m, void *v, loff_t *pos) } } - return rec; + if (!rec) + return NULL; + + iter->func = rec; + + return iter; } static void *t_start(struct seq_file *m, loff_t *pos) @@ -1530,10 +1549,14 @@ static void *t_start(struct seq_file *m, loff_t *pos) break; } - if (!p && iter->flags & FTRACE_ITER_FILTER) - return t_hash_start(m, pos); + if (!p) { + if (iter->flags & FTRACE_ITER_FILTER) + return t_hash_start(m, pos); - return p; + return NULL; + } + + return iter; } static void t_stop(struct seq_file *m, void *p) @@ -1544,16 +1567,18 @@ static void t_stop(struct seq_file *m, void *p) static int t_show(struct seq_file *m, void *v) { struct ftrace_iterator *iter = m->private; - struct dyn_ftrace *rec = v; + struct dyn_ftrace *rec; if (iter->flags & FTRACE_ITER_HASH) - return t_hash_show(m, v); + return t_hash_show(m, iter); if (iter->flags & FTRACE_ITER_PRINTALL) { seq_printf(m, "#### all functions enabled ####\n"); return 0; } + rec = iter->func; + if (!rec) return 0; -- cgit v1.2.3 From 98c4fd046f07156ca6055677e8f03d4280be16c1 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Fri, 10 Sep 2010 11:47:43 -0400 Subject: tracing: Keep track of set_ftrace_filter position and allow lseek again This patch keeps track of the index within the elements of set_ftrace_filter and if the position goes backwards, it nicely resets and starts from the beginning again. This allows for lseek and pread to work properly now. Signed-off-by: Steven Rostedt --- kernel/trace/ftrace.c | 34 ++++++++++++++++++++++++++-------- 1 file changed, 26 insertions(+), 8 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index c8db0dbb984e..2d51166b93fe 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -1368,6 +1368,7 @@ enum { #define FTRACE_BUFF_MAX (KSYM_SYMBOL_LEN+4) /* room for wildcards */ struct ftrace_iterator { + loff_t pos; loff_t func_pos; struct ftrace_page *pg; struct dyn_ftrace *func; @@ -1385,9 +1386,8 @@ t_hash_next(struct seq_file *m, loff_t *pos) struct hlist_node *hnd = NULL; struct hlist_head *hhd; - WARN_ON(!(iter->flags & FTRACE_ITER_HASH)); - (*pos)++; + iter->pos = *pos; if (iter->probe) hnd = &iter->probe->node; @@ -1427,14 +1427,9 @@ static void *t_hash_start(struct seq_file *m, loff_t *pos) void *p = NULL; loff_t l; - if (!(iter->flags & FTRACE_ITER_HASH)) - iter->func_pos = *pos; - if (iter->func_pos > *pos) return NULL; - iter->flags |= FTRACE_ITER_HASH; - iter->hidx = 0; for (l = 0; l <= (*pos - iter->func_pos); ) { p = t_hash_next(m, &l); @@ -1444,6 +1439,9 @@ static void *t_hash_start(struct seq_file *m, loff_t *pos) if (!p) return NULL; + /* Only set this if we have an item */ + iter->flags |= FTRACE_ITER_HASH; + return iter; } @@ -1478,6 +1476,8 @@ t_next(struct seq_file *m, void *v, loff_t *pos) return t_hash_next(m, pos); (*pos)++; + iter->pos = *pos; + iter->func_pos = *pos; if (iter->flags & FTRACE_ITER_PRINTALL) return NULL; @@ -1517,6 +1517,13 @@ t_next(struct seq_file *m, void *v, loff_t *pos) return iter; } +static void reset_iter_read(struct ftrace_iterator *iter) +{ + iter->pos = 0; + iter->func_pos = 0; + iter->flags &= ~(FTRACE_ITER_PRINTALL & FTRACE_ITER_HASH); +} + static void *t_start(struct seq_file *m, loff_t *pos) { struct ftrace_iterator *iter = m->private; @@ -1524,6 +1531,12 @@ static void *t_start(struct seq_file *m, loff_t *pos) loff_t l; mutex_lock(&ftrace_lock); + /* + * If an lseek was done, then reset and start from beginning. + */ + if (*pos < iter->pos) + reset_iter_read(iter); + /* * For set_ftrace_filter reading, if we have the filter * off, we can short cut and just print out that all @@ -1541,6 +1554,11 @@ static void *t_start(struct seq_file *m, loff_t *pos) if (iter->flags & FTRACE_ITER_HASH) return t_hash_start(m, pos); + /* + * Unfortunately, we need to restart at ftrace_pages_start + * every time we let go of the ftrace_mutex. This is because + * those pointers can change without the lock. + */ iter->pg = ftrace_pages_start; iter->idx = 0; for (l = 0; l <= *pos; ) { @@ -2447,7 +2465,7 @@ static const struct file_operations ftrace_filter_fops = { .open = ftrace_filter_open, .read = seq_read, .write = ftrace_filter_write, - .llseek = no_llseek, + .llseek = ftrace_regex_lseek, .release = ftrace_filter_release, }; -- cgit v1.2.3 From 57c072c7113f54f9512624d6c665db6184448782 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Tue, 14 Sep 2010 11:21:11 -0400 Subject: tracing: Fix reading of set_ftrace_filter across lists If we do: # cd /sys/kernel/debug # echo 'do_IRQ:traceon schedule:traceon sys_write:traceon' > \ set_ftrace_filter # cat set_ftrace_filter We get the following output: #### all functions enabled #### sys_write:traceon:unlimited schedule:traceon:unlimited do_IRQ:traceon:unlimited This outputs two lists. One is the fact that all functions are currently enabled for function tracing, the other has three probed functions, which happen to have 'traceon' as their commands. Currently, when reading the first list (functions enabled) the seq_file code will receive a "NULL" from the t_next() function causing it to exit early. This makes "read()" from userspace stop reading the code at this boarder. Although read is allowed to do this, some (broken) applications might consider this an end of file and stop early. This patch adds the start of the second list to t_next() when it finishes the first list. It is a simple change and gives the set_ftrace_filter file nicer reading ability. Signed-off-by: Steven Rostedt --- kernel/trace/ftrace.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index 2d51166b93fe..1884cf5bc110 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -1477,10 +1477,9 @@ t_next(struct seq_file *m, void *v, loff_t *pos) (*pos)++; iter->pos = *pos; - iter->func_pos = *pos; if (iter->flags & FTRACE_ITER_PRINTALL) - return NULL; + return t_hash_start(m, pos); retry: if (iter->idx >= iter->pg->index) { @@ -1510,8 +1509,9 @@ t_next(struct seq_file *m, void *v, loff_t *pos) } if (!rec) - return NULL; + return t_hash_start(m, pos); + iter->func_pos = *pos; iter->func = rec; return iter; -- cgit v1.2.3 From c41d68a513c71e35a14f66d71782d27a79a81ea6 Mon Sep 17 00:00:00 2001 From: "H. Peter Anvin" Date: Tue, 7 Sep 2010 16:16:18 -0700 Subject: compat: Make compat_alloc_user_space() incorporate the access_ok() compat_alloc_user_space() expects the caller to independently call access_ok() to verify the returned area. A missing call could introduce problems on some architectures. This patch incorporates the access_ok() check into compat_alloc_user_space() and also adds a sanity check on the length. The existing compat_alloc_user_space() implementations are renamed arch_compat_alloc_user_space() and are used as part of the implementation of the new global function. This patch assumes NULL will cause __get_user()/__put_user() to either fail or access userspace on all architectures. This should be followed by checking the return value of compat_access_user_space() for NULL in the callers, at which time the access_ok() in the callers can also be removed. Reported-by: Ben Hawkes Signed-off-by: H. Peter Anvin Acked-by: Benjamin Herrenschmidt Acked-by: Chris Metcalf Acked-by: David S. Miller Acked-by: Ingo Molnar Acked-by: Thomas Gleixner Acked-by: Tony Luck Cc: Andrew Morton Cc: Arnd Bergmann Cc: Fenghua Yu Cc: H. Peter Anvin Cc: Heiko Carstens Cc: Helge Deller Cc: James Bottomley Cc: Kyle McMartin Cc: Martin Schwidefsky Cc: Paul Mackerras Cc: Ralf Baechle Cc: --- kernel/compat.c | 21 +++++++++++++++++++++ 1 file changed, 21 insertions(+) (limited to 'kernel') diff --git a/kernel/compat.c b/kernel/compat.c index e167efce8423..c9e2ec0b34a8 100644 --- a/kernel/compat.c +++ b/kernel/compat.c @@ -1126,3 +1126,24 @@ compat_sys_sysinfo(struct compat_sysinfo __user *info) return 0; } + +/* + * Allocate user-space memory for the duration of a single system call, + * in order to marshall parameters inside a compat thunk. + */ +void __user *compat_alloc_user_space(unsigned long len) +{ + void __user *ptr; + + /* If len would occupy more than half of the entire compat space... */ + if (unlikely(len > (((compat_uptr_t)~0) >> 1))) + return NULL; + + ptr = arch_compat_alloc_user_space(len); + + if (unlikely(!access_ok(VERIFY_WRITE, ptr, len))) + return NULL; + + return ptr; +} +EXPORT_SYMBOL_GPL(compat_alloc_user_space); -- cgit v1.2.3 From 2bd16212b8eb86f9574e78d6605a5ba9e9aa8c4e Mon Sep 17 00:00:00 2001 From: Jiri Olsa Date: Tue, 7 Sep 2010 16:53:44 +0200 Subject: tracing: Add funcgraph-irq option for function graph tracer. It's handy to be able to disable the irq related output and not to have to jump over each irq related code, when you have no interrest in it. The option is by default enabled, so there's no change to current behaviour. It affects only the final output, so all the irq related data stay in the ring buffer. Signed-off-by: Jiri Olsa LKML-Reference: <20100907145344.GC1912@jolsa.brq.redhat.com> Signed-off-by: Steven Rostedt --- kernel/trace/trace_functions_graph.c | 101 ++++++++++++++++++++++++++++++++++- 1 file changed, 100 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/trace/trace_functions_graph.c b/kernel/trace/trace_functions_graph.c index c93bcb248638..8674750a5ece 100644 --- a/kernel/trace/trace_functions_graph.c +++ b/kernel/trace/trace_functions_graph.c @@ -18,6 +18,7 @@ struct fgraph_cpu_data { pid_t last_pid; int depth; + int depth_irq; int ignore; unsigned long enter_funcs[FTRACE_RETFUNC_DEPTH]; }; @@ -41,6 +42,7 @@ struct fgraph_data { #define TRACE_GRAPH_PRINT_PROC 0x8 #define TRACE_GRAPH_PRINT_DURATION 0x10 #define TRACE_GRAPH_PRINT_ABS_TIME 0x20 +#define TRACE_GRAPH_PRINT_IRQS 0x40 static struct tracer_opt trace_opts[] = { /* Display overruns? (for self-debug purpose) */ @@ -55,13 +57,15 @@ static struct tracer_opt trace_opts[] = { { TRACER_OPT(funcgraph-duration, TRACE_GRAPH_PRINT_DURATION) }, /* Display absolute time of an entry */ { TRACER_OPT(funcgraph-abstime, TRACE_GRAPH_PRINT_ABS_TIME) }, + /* Display interrupts */ + { TRACER_OPT(funcgraph-irqs, TRACE_GRAPH_PRINT_IRQS) }, { } /* Empty entry */ }; static struct tracer_flags tracer_flags = { /* Don't display overruns and proc by default */ .val = TRACE_GRAPH_PRINT_CPU | TRACE_GRAPH_PRINT_OVERHEAD | - TRACE_GRAPH_PRINT_DURATION, + TRACE_GRAPH_PRINT_DURATION | TRACE_GRAPH_PRINT_IRQS, .opts = trace_opts }; @@ -855,6 +859,92 @@ print_graph_prologue(struct trace_iterator *iter, struct trace_seq *s, return 0; } +/* + * Entry check for irq code + * + * returns 1 if + * - we are inside irq code + * - we just extered irq code + * + * retunns 0 if + * - funcgraph-interrupts option is set + * - we are not inside irq code + */ +static int +check_irq_entry(struct trace_iterator *iter, u32 flags, + unsigned long addr, int depth) +{ + int cpu = iter->cpu; + struct fgraph_data *data = iter->private; + int *depth_irq = &(per_cpu_ptr(data->cpu_data, cpu)->depth_irq); + + if (flags & TRACE_GRAPH_PRINT_IRQS) + return 0; + + /* + * We are inside the irq code + */ + if (*depth_irq >= 0) + return 1; + + if ((addr < (unsigned long)__irqentry_text_start) || + (addr >= (unsigned long)__irqentry_text_end)) + return 0; + + /* + * We are entering irq code. + */ + *depth_irq = depth; + return 1; +} + +/* + * Return check for irq code + * + * returns 1 if + * - we are inside irq code + * - we just left irq code + * + * returns 0 if + * - funcgraph-interrupts option is set + * - we are not inside irq code + */ +static int +check_irq_return(struct trace_iterator *iter, u32 flags, int depth) +{ + int cpu = iter->cpu; + struct fgraph_data *data = iter->private; + int *depth_irq = &(per_cpu_ptr(data->cpu_data, cpu)->depth_irq); + + if (flags & TRACE_GRAPH_PRINT_IRQS) + return 0; + + /* + * We are not inside the irq code. + */ + if (*depth_irq == -1) + return 0; + + /* + * We are inside the irq code, and this is returning entry. + * Let's not trace it and clear the entry depth, since + * we are out of irq code. + * + * This condition ensures that we 'leave the irq code' once + * we are out of the entry depth. Thus protecting us from + * the RETURN entry loss. + */ + if (*depth_irq >= depth) { + *depth_irq = -1; + return 1; + } + + /* + * We are inside the irq code, and this is not the entry. + */ + return 1; +} + static enum print_line_t print_graph_entry(struct ftrace_graph_ent_entry *field, struct trace_seq *s, struct trace_iterator *iter, u32 flags) @@ -865,6 +955,9 @@ print_graph_entry(struct ftrace_graph_ent_entry *field, struct trace_seq *s, static enum print_line_t ret; int cpu = iter->cpu; + if (check_irq_entry(iter, flags, call->func, call->depth)) + return TRACE_TYPE_HANDLED; + if (print_graph_prologue(iter, s, TRACE_GRAPH_ENT, call->func, flags)) return TRACE_TYPE_PARTIAL_LINE; @@ -902,6 +995,9 @@ print_graph_return(struct ftrace_graph_ret *trace, struct trace_seq *s, int ret; int i; + if (check_irq_return(iter, flags, trace->depth)) + return TRACE_TYPE_HANDLED; + if (data) { struct fgraph_cpu_data *cpu_data; int cpu = iter->cpu; @@ -1210,9 +1306,12 @@ void graph_trace_open(struct trace_iterator *iter) pid_t *pid = &(per_cpu_ptr(data->cpu_data, cpu)->last_pid); int *depth = &(per_cpu_ptr(data->cpu_data, cpu)->depth); int *ignore = &(per_cpu_ptr(data->cpu_data, cpu)->ignore); + int *depth_irq = &(per_cpu_ptr(data->cpu_data, cpu)->depth_irq); + *pid = -1; *depth = 0; *ignore = 0; + *depth_irq = -1; } iter->private = data; -- cgit v1.2.3 From b304d0441a4118fadd4c3f16e4dc600c271030b5 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Tue, 14 Sep 2010 18:58:33 -0400 Subject: tracing: Do not trace in irq when funcgraph-irq option is zero When the function graph tracer funcgraph-irq option is zero, disable tracing in IRQs. This makes the option have two effects. 1) When reading the trace file, do not display the functions that happen in interrupt context (when detected) 2) [*new*] When recording a trace, skip those that are detected to be in interrupt by the 'in_irq()' function Note, in_irq() is updated at irq_enter() and irq_exit(). There are still functions that are recorded by the function graph tracer that is in interrupt context but outside the irq_enter/exit() routines. Signed-off-by: Steven Rostedt --- kernel/trace/trace_functions_graph.c | 23 ++++++++++++++++++++++- 1 file changed, 22 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/trace/trace_functions_graph.c b/kernel/trace/trace_functions_graph.c index 8674750a5ece..02c708ae0d42 100644 --- a/kernel/trace/trace_functions_graph.c +++ b/kernel/trace/trace_functions_graph.c @@ -15,6 +15,9 @@ #include "trace.h" #include "trace_output.h" +/* When set, irq functions will be ignored */ +static int ftrace_graph_skip_irqs; + struct fgraph_cpu_data { pid_t last_pid; int depth; @@ -208,6 +211,14 @@ int __trace_graph_entry(struct trace_array *tr, return 1; } +static inline int ftrace_graph_ignore_irqs(void) +{ + if (!ftrace_graph_skip_irqs) + return 0; + + return in_irq(); +} + int trace_graph_entry(struct ftrace_graph_ent *trace) { struct trace_array *tr = graph_array; @@ -222,7 +233,8 @@ int trace_graph_entry(struct ftrace_graph_ent *trace) return 0; /* trace it when it is-nested-in or is a function enabled. */ - if (!(trace->depth || ftrace_graph_addr(trace->func))) + if (!(trace->depth || ftrace_graph_addr(trace->func)) || + ftrace_graph_ignore_irqs()) return 0; local_irq_save(flags); @@ -1334,6 +1346,14 @@ void graph_trace_close(struct trace_iterator *iter) } } +static int func_graph_set_flag(u32 old_flags, u32 bit, int set) +{ + if (bit == TRACE_GRAPH_PRINT_IRQS) + ftrace_graph_skip_irqs = !set; + + return 0; +} + static struct trace_event_functions graph_functions = { .trace = print_graph_function_event, }; @@ -1360,6 +1380,7 @@ static struct tracer graph_trace __read_mostly = { .print_line = print_graph_function, .print_header = print_graph_headers, .flags = &tracer_flags, + .set_flag = func_graph_set_flag, #ifdef CONFIG_FTRACE_SELFTEST .selftest = trace_selftest_startup_function_graph, #endif -- cgit v1.2.3 From 79e406d7b00ab2b261ae32a59f266fd3b7af6f29 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Tue, 14 Sep 2010 22:19:46 -0400 Subject: tracing: Remove leftover FTRACE_ENABLE/DISABLE_MCOUNT enums The enums for FTRACE_ENABLE_MCOUNT and FTRACE_DISABLE_MCOUNT were used as commands to ftrace_run_update_code(). But these commands were used by the old nasty ftrace daemon that has long been slain. This is a clean up patch to remove the references to these enums and simplify the code a little. Reported-by: Wu Zhangjin Signed-off-by: Steven Rostedt --- kernel/trace/ftrace.c | 18 ++++-------------- 1 file changed, 4 insertions(+), 14 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index 83a16e9ee518..20aff3f1c719 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -884,10 +884,8 @@ enum { FTRACE_ENABLE_CALLS = (1 << 0), FTRACE_DISABLE_CALLS = (1 << 1), FTRACE_UPDATE_TRACE_FUNC = (1 << 2), - FTRACE_ENABLE_MCOUNT = (1 << 3), - FTRACE_DISABLE_MCOUNT = (1 << 4), - FTRACE_START_FUNC_RET = (1 << 5), - FTRACE_STOP_FUNC_RET = (1 << 6), + FTRACE_START_FUNC_RET = (1 << 3), + FTRACE_STOP_FUNC_RET = (1 << 4), }; static int ftrace_filtered; @@ -1226,8 +1224,6 @@ static void ftrace_shutdown(int command) static void ftrace_startup_sysctl(void) { - int command = FTRACE_ENABLE_MCOUNT; - if (unlikely(ftrace_disabled)) return; @@ -1235,23 +1231,17 @@ static void ftrace_startup_sysctl(void) saved_ftrace_func = NULL; /* ftrace_start_up is true if we want ftrace running */ if (ftrace_start_up) - command |= FTRACE_ENABLE_CALLS; - - ftrace_run_update_code(command); + ftrace_run_update_code(FTRACE_ENABLE_CALLS); } static void ftrace_shutdown_sysctl(void) { - int command = FTRACE_DISABLE_MCOUNT; - if (unlikely(ftrace_disabled)) return; /* ftrace_start_up is true if ftrace is running */ if (ftrace_start_up) - command |= FTRACE_DISABLE_CALLS; - - ftrace_run_update_code(command); + ftrace_run_update_code(FTRACE_DISABLE_CALLS); } static cycle_t ftrace_update_time; -- cgit v1.2.3 From e75e863dd5c7d96b91ebbd241da5328fc38a78cc Mon Sep 17 00:00:00 2001 From: Stanislaw Gruszka Date: Tue, 14 Sep 2010 16:35:14 +0200 Subject: sched: Fix user time incorrectly accounted as system time on 32-bit We have 32-bit variable overflow possibility when multiply in task_times() and thread_group_times() functions. When the overflow happens then the scaled utime value becomes erroneously small and the scaled stime becomes i erroneously big. Reported here: https://bugzilla.redhat.com/show_bug.cgi?id=633037 https://bugzilla.kernel.org/show_bug.cgi?id=16559 Reported-by: Michael Chapman Reported-by: Ciriaco Garcia de Celis Signed-off-by: Stanislaw Gruszka Signed-off-by: Peter Zijlstra Cc: Hidetoshi Seto Cc: # 2.6.32.19+ (partially) and 2.6.33+ LKML-Reference: <20100914143513.GB8415@redhat.com> Signed-off-by: Ingo Molnar --- kernel/sched.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index ed09d4f2a69c..dc85ceb90832 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -3513,9 +3513,9 @@ void task_times(struct task_struct *p, cputime_t *ut, cputime_t *st) rtime = nsecs_to_cputime(p->se.sum_exec_runtime); if (total) { - u64 temp; + u64 temp = rtime; - temp = (u64)(rtime * utime); + temp *= utime; do_div(temp, total); utime = (cputime_t)temp; } else @@ -3546,9 +3546,9 @@ void thread_group_times(struct task_struct *p, cputime_t *ut, cputime_t *st) rtime = nsecs_to_cputime(cputime.sum_exec_runtime); if (total) { - u64 temp; + u64 temp = rtime; - temp = (u64)(rtime * cputime.utime); + temp *= cputime.utime; do_div(temp, total); utime = (cputime_t)temp; } else -- cgit v1.2.3 From d9ca07a05ce1c42ac9717e54eaea4546a3a80978 Mon Sep 17 00:00:00 2001 From: Stephane Eranian Date: Tue, 14 Sep 2010 15:34:01 +0200 Subject: watchdog: Avoid kernel crash when disabling watchdog In case you boot with the watchdog disabled, i.e., nowatchdog, then, if you try to disable it via /proc/sys/kernel/watchdog, you get a kernel crash. The reason is that you are trying to cancel a hrtimer which has never been initialized. This patch fixes this by skipping execution of watchdog_disable_all_cpus() when the watchdog is marked disabled from boot. Signed-off-by: Stephane Eranian Signed-off-by: Peter Zijlstra LKML-Reference: <4c8f7a23.cae9d80a.2c11.0bb4@mx.google.com> Signed-off-by: Ingo Molnar --- kernel/watchdog.c | 3 +++ 1 file changed, 3 insertions(+) (limited to 'kernel') diff --git a/kernel/watchdog.c b/kernel/watchdog.c index fa71aebda4ff..89eadbb9cefe 100644 --- a/kernel/watchdog.c +++ b/kernel/watchdog.c @@ -473,6 +473,9 @@ static void watchdog_disable_all_cpus(void) { int cpu; + if (no_watchdog) + return; + for_each_online_cpu(cpu) watchdog_disable(cpu); -- cgit v1.2.3 From d958077d007d98125766d11e82da2fd6497b91d6 Mon Sep 17 00:00:00 2001 From: Matt Helsley Date: Mon, 13 Sep 2010 13:01:18 -0700 Subject: hw breakpoints: Fix pid namespace bug Hardware breakpoints can't be registered within pid namespaces because tsk->pid is passed rather than the pid in the current namespace. (See https://bugzilla.kernel.org/show_bug.cgi?id=17281 ) This is a quick fix demonstrating the problem but is not the best method of solving the problem since passing pids internally is not the best way to avoid pid namespace bugs. Subsequent patches will show a better solution. Much thanks to Frederic Weisbecker for doing the bulk of the work finding this bug. Signed-off-by: Matt Helsley Signed-off-by: Peter Zijlstra Cc: Robin Green Cc: Prasad Cc: Arnaldo Carvalho de Melo Cc: Steven Rostedt Cc: Will Deacon Cc: Mahesh Salgaonkar LKML-Reference: Signed-off-by: Ingo Molnar --- kernel/hw_breakpoint.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/hw_breakpoint.c b/kernel/hw_breakpoint.c index 3b2aaffb65f0..6122f02cfedf 100644 --- a/kernel/hw_breakpoint.c +++ b/kernel/hw_breakpoint.c @@ -433,7 +433,8 @@ register_user_hw_breakpoint(struct perf_event_attr *attr, perf_overflow_handler_t triggered, struct task_struct *tsk) { - return perf_event_create_kernel_counter(attr, -1, tsk->pid, triggered); + return perf_event_create_kernel_counter(attr, -1, task_pid_vnr(tsk), + triggered); } EXPORT_SYMBOL_GPL(register_user_hw_breakpoint); -- cgit v1.2.3 From 2ebd4ffb6d0cb877787b1e42be8485820158857e Mon Sep 17 00:00:00 2001 From: Matt Helsley Date: Mon, 13 Sep 2010 13:01:19 -0700 Subject: perf events: Split out task search into helper Split out the code which searches for non-exiting tasks into its own helper. Creating this helper not only makes the code slightly more readable it prepares to move the search out of find_get_context() in a subsequent commit. Signed-off-by: Matt Helsley Signed-off-by: Peter Zijlstra Cc: Robin Green Cc: Prasad Cc: Arnaldo Carvalho de Melo Cc: Steven Rostedt Cc: Will Deacon Cc: Mahesh Salgaonkar LKML-Reference: <561205417b450b8a4bf7488374541d64b4690431.1284407762.git.matthltc@us.ibm.com> Signed-off-by: Ingo Molnar --- kernel/perf_event.c | 63 ++++++++++++++++++++++++++++++++++------------------- 1 file changed, 40 insertions(+), 23 deletions(-) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index 440f9ca067b2..3f5309db72f1 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -2015,6 +2015,43 @@ alloc_perf_context(struct pmu *pmu, struct task_struct *task) return ctx; } +static struct task_struct * +find_lively_task_by_vpid(pid_t vpid) +{ + struct task_struct *task; + int err; + + rcu_read_lock(); + if (!vpid) + task = current; + else + task = find_task_by_vpid(vpid); + if (task) + get_task_struct(task); + rcu_read_unlock(); + + if (!task) + return ERR_PTR(-ESRCH); + + /* + * Can't attach events to a dying task. + */ + err = -ESRCH; + if (task->flags & PF_EXITING) + goto errout; + + /* Reuse ptrace permission checks for now. */ + err = -EACCES; + if (!ptrace_may_access(task, PTRACE_MODE_READ)) + goto errout; + + return task; +errout: + put_task_struct(task); + return ERR_PTR(err); + +} + static struct perf_event_context * find_get_context(struct pmu *pmu, pid_t pid, int cpu) { @@ -2047,29 +2084,9 @@ find_get_context(struct pmu *pmu, pid_t pid, int cpu) return ctx; } - rcu_read_lock(); - if (!pid) - task = current; - else - task = find_task_by_vpid(pid); - if (task) - get_task_struct(task); - rcu_read_unlock(); - - if (!task) - return ERR_PTR(-ESRCH); - - /* - * Can't attach events to a dying task. - */ - err = -ESRCH; - if (task->flags & PF_EXITING) - goto errout; - - /* Reuse ptrace permission checks for now. */ - err = -EACCES; - if (!ptrace_may_access(task, PTRACE_MODE_READ)) - goto errout; + task = find_lively_task_by_vpid(pid); + if (IS_ERR(task)) + return (void*)task; err = -EINVAL; ctxn = pmu->task_ctx_nr; -- cgit v1.2.3 From 38a81da2205f94e8a2a834b51a6b99c91fc7c2e8 Mon Sep 17 00:00:00 2001 From: Matt Helsley Date: Mon, 13 Sep 2010 13:01:20 -0700 Subject: perf events: Clean up pid passing The kernel perf event creation path shouldn't use find_task_by_vpid() because a vpid exists in a specific namespace. find_task_by_vpid() uses current's pid namespace which isn't always the correct namespace to use for the vpid in all the places perf_event_create_kernel_counter() (and thus find_get_context()) is called. The goal is to clean up pid namespace handling and prevent bugs like: https://bugzilla.kernel.org/show_bug.cgi?id=17281 Instead of using pids switch find_get_context() to use task struct pointers directly. The syscall is responsible for resolving the pid to a task struct. This moves the pid namespace resolution into the syscall much like every other syscall that takes pid parameters. Signed-off-by: Matt Helsley Signed-off-by: Peter Zijlstra Cc: Robin Green Cc: Prasad Cc: Arnaldo Carvalho de Melo Cc: Steven Rostedt Cc: Will Deacon Cc: Mahesh Salgaonkar LKML-Reference: Signed-off-by: Ingo Molnar --- kernel/hw_breakpoint.c | 5 ++--- kernel/perf_event.c | 21 ++++++++++----------- kernel/watchdog.c | 2 +- 3 files changed, 13 insertions(+), 15 deletions(-) (limited to 'kernel') diff --git a/kernel/hw_breakpoint.c b/kernel/hw_breakpoint.c index 6122f02cfedf..3b714e839c10 100644 --- a/kernel/hw_breakpoint.c +++ b/kernel/hw_breakpoint.c @@ -433,8 +433,7 @@ register_user_hw_breakpoint(struct perf_event_attr *attr, perf_overflow_handler_t triggered, struct task_struct *tsk) { - return perf_event_create_kernel_counter(attr, -1, task_pid_vnr(tsk), - triggered); + return perf_event_create_kernel_counter(attr, -1, tsk, triggered); } EXPORT_SYMBOL_GPL(register_user_hw_breakpoint); @@ -516,7 +515,7 @@ register_wide_hw_breakpoint(struct perf_event_attr *attr, get_online_cpus(); for_each_online_cpu(cpu) { pevent = per_cpu_ptr(cpu_events, cpu); - bp = perf_event_create_kernel_counter(attr, cpu, -1, triggered); + bp = perf_event_create_kernel_counter(attr, cpu, NULL, triggered); *pevent = bp; diff --git a/kernel/perf_event.c b/kernel/perf_event.c index 3f5309db72f1..86f394e15d53 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -2053,15 +2053,14 @@ errout: } static struct perf_event_context * -find_get_context(struct pmu *pmu, pid_t pid, int cpu) +find_get_context(struct pmu *pmu, struct task_struct *task, int cpu) { struct perf_event_context *ctx; struct perf_cpu_context *cpuctx; - struct task_struct *task; unsigned long flags; int ctxn, err; - if (pid == -1 && cpu != -1) { + if (!task && cpu != -1) { /* Must be root to operate on a CPU event: */ if (perf_paranoid_cpu() && !capable(CAP_SYS_ADMIN)) return ERR_PTR(-EACCES); @@ -2084,10 +2083,6 @@ find_get_context(struct pmu *pmu, pid_t pid, int cpu) return ctx; } - task = find_lively_task_by_vpid(pid); - if (IS_ERR(task)) - return (void*)task; - err = -EINVAL; ctxn = pmu->task_ctx_nr; if (ctxn < 0) @@ -5527,6 +5522,7 @@ SYSCALL_DEFINE5(perf_event_open, struct perf_event_context *ctx; struct file *event_file = NULL; struct file *group_file = NULL; + struct task_struct *task = NULL; struct pmu *pmu; int event_fd; int fput_needed = 0; @@ -5581,10 +5577,13 @@ SYSCALL_DEFINE5(perf_event_open, if ((pmu->task_ctx_nr == perf_sw_context) && group_leader) pmu = group_leader->pmu; + if (pid != -1) + task = find_lively_task_by_vpid(pid); + /* * Get the target context (task or percpu): */ - ctx = find_get_context(pmu, pid, cpu); + ctx = find_get_context(pmu, task, cpu); if (IS_ERR(ctx)) { err = PTR_ERR(ctx); goto err_group_fd; @@ -5666,11 +5665,11 @@ err_fd: * * @attr: attributes of the counter to create * @cpu: cpu in which the counter is bound - * @pid: task to profile + * @task: task to profile (NULL for percpu) */ struct perf_event * perf_event_create_kernel_counter(struct perf_event_attr *attr, int cpu, - pid_t pid, + struct task_struct *task, perf_overflow_handler_t overflow_handler) { struct perf_event_context *ctx; @@ -5687,7 +5686,7 @@ perf_event_create_kernel_counter(struct perf_event_attr *attr, int cpu, goto err; } - ctx = find_get_context(event->pmu, pid, cpu); + ctx = find_get_context(event->pmu, task, cpu); if (IS_ERR(ctx)) { err = PTR_ERR(ctx); goto err_free; diff --git a/kernel/watchdog.c b/kernel/watchdog.c index 89eadbb9cefe..dc8e16824b51 100644 --- a/kernel/watchdog.c +++ b/kernel/watchdog.c @@ -358,7 +358,7 @@ static int watchdog_nmi_enable(int cpu) /* Try to register using hardware perf events */ wd_attr = &wd_hw_attr; wd_attr->sample_period = hw_nmi_get_sample_period(); - event = perf_event_create_kernel_counter(wd_attr, cpu, -1, watchdog_overflow_callback); + event = perf_event_create_kernel_counter(wd_attr, cpu, NULL, watchdog_overflow_callback); if (!IS_ERR(event)) { printk(KERN_INFO "NMI watchdog enabled, takes one hw-pmu counter.\n"); goto out_save; -- cgit v1.2.3 From edbaadbe42b0b790618ec49d29626223529d8195 Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Wed, 15 Sep 2010 10:04:26 +0900 Subject: kprobes: Remove redundant address check Remove call to kernel_text_address() in register_jprobes() because it is called right after in register_kprobe(). Signed-off-by: Namhyung Kim Acked-by: Masami Hiramatsu LKML-Reference: <1284512670-2369-2-git-send-email-namhyung@gmail.com> Signed-off-by: Ingo Molnar --- kernel/kprobes.c | 13 +++++-------- 1 file changed, 5 insertions(+), 8 deletions(-) (limited to 'kernel') diff --git a/kernel/kprobes.c b/kernel/kprobes.c index 282035f3ae96..8f967016cef0 100644 --- a/kernel/kprobes.c +++ b/kernel/kprobes.c @@ -1343,14 +1343,11 @@ int __kprobes register_jprobes(struct jprobe **jps, int num) jp = jps[i]; addr = arch_deref_entry_point(jp->entry); - if (!kernel_text_address(addr)) - ret = -EINVAL; - else { - /* Todo: Verify probepoint is a function entry point */ - jp->kp.pre_handler = setjmp_pre_handler; - jp->kp.break_handler = longjmp_break_handler; - ret = register_kprobe(&jp->kp); - } + /* Todo: Verify probepoint is a function entry point */ + jp->kp.pre_handler = setjmp_pre_handler; + jp->kp.break_handler = longjmp_break_handler; + ret = register_kprobe(&jp->kp); + if (ret < 0) { if (i > 0) unregister_jprobes(jps, i); -- cgit v1.2.3 From 05662bdb64c746079de7ac4dc4fb4caa5e8e119f Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Wed, 15 Sep 2010 10:04:27 +0900 Subject: kprobes: Verify jprobe entry point Verify jprobe's entry point is a function entry point using kallsyms' offset value. Signed-off-by: Namhyung Kim Acked-by: Masami Hiramatsu LKML-Reference: <1284512670-2369-3-git-send-email-namhyung@gmail.com> Signed-off-by: Ingo Molnar --- kernel/kprobes.c | 14 +++++++++----- 1 file changed, 9 insertions(+), 5 deletions(-) (limited to 'kernel') diff --git a/kernel/kprobes.c b/kernel/kprobes.c index 8f967016cef0..1b0dbe067077 100644 --- a/kernel/kprobes.c +++ b/kernel/kprobes.c @@ -1339,14 +1339,18 @@ int __kprobes register_jprobes(struct jprobe **jps, int num) if (num <= 0) return -EINVAL; for (i = 0; i < num; i++) { - unsigned long addr; + unsigned long addr, offset; jp = jps[i]; addr = arch_deref_entry_point(jp->entry); - /* Todo: Verify probepoint is a function entry point */ - jp->kp.pre_handler = setjmp_pre_handler; - jp->kp.break_handler = longjmp_break_handler; - ret = register_kprobe(&jp->kp); + /* Verify probepoint is a function entry point */ + if (kallsyms_lookup_size_offset(addr, NULL, &offset) && + offset == 0) { + jp->kp.pre_handler = setjmp_pre_handler; + jp->kp.break_handler = longjmp_break_handler; + ret = register_kprobe(&jp->kp); + } else + ret = -EINVAL; if (ret < 0) { if (i > 0) -- cgit v1.2.3 From 6376b2297502e72255b7eb2893c6044ad5a7b5f4 Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Wed, 15 Sep 2010 10:04:28 +0900 Subject: kprobes: Make functions static Make following (internal) functions static to make sparse happier :-) * get_optimized_kprobe: only called from static functions * kretprobe_table_unlock: _lock function is static * kprobes_optinsn_template_holder: never called but holding asm code Signed-off-by: Namhyung Kim Acked-by: Masami Hiramatsu LKML-Reference: <1284512670-2369-4-git-send-email-namhyung@gmail.com> Signed-off-by: Ingo Molnar --- kernel/kprobes.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/kprobes.c b/kernel/kprobes.c index 1b0dbe067077..c53aad5d7e5e 100644 --- a/kernel/kprobes.c +++ b/kernel/kprobes.c @@ -399,7 +399,7 @@ static inline int kprobe_optready(struct kprobe *p) * Return an optimized kprobe whose optimizing code replaces * instructions including addr (exclude breakpoint). */ -struct kprobe *__kprobes get_optimized_kprobe(unsigned long addr) +static struct kprobe *__kprobes get_optimized_kprobe(unsigned long addr) { int i; struct kprobe *p = NULL; @@ -857,7 +857,8 @@ void __kprobes kretprobe_hash_unlock(struct task_struct *tsk, spin_unlock_irqrestore(hlist_lock, *flags); } -void __kprobes kretprobe_table_unlock(unsigned long hash, unsigned long *flags) +static void __kprobes kretprobe_table_unlock(unsigned long hash, + unsigned long *flags) { spinlock_t *hlist_lock = kretprobe_table_lock_ptr(hash); spin_unlock_irqrestore(hlist_lock, *flags); -- cgit v1.2.3 From 635c17c2b2b4e5cd34f5dcba19d751b4e58533c2 Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Wed, 15 Sep 2010 10:04:30 +0900 Subject: kprobes: Add sparse context annotations This removes following warnings when build with C=1 warning: context imbalance in 'kretprobe_hash_lock' - wrong count at exit warning: context imbalance in 'kretprobe_table_lock' - wrong count at exit warning: context imbalance in 'kretprobe_hash_unlock' - unexpected unlock warning: context imbalance in 'kretprobe_table_unlock' - unexpected unlock Signed-off-by: Namhyung Kim Acked-by: Masami Hiramatsu LKML-Reference: <1284512670-2369-6-git-send-email-namhyung@gmail.com> Signed-off-by: Ingo Molnar --- kernel/kprobes.c | 4 ++++ 1 file changed, 4 insertions(+) (limited to 'kernel') diff --git a/kernel/kprobes.c b/kernel/kprobes.c index c53aad5d7e5e..6dd5359e1f0e 100644 --- a/kernel/kprobes.c +++ b/kernel/kprobes.c @@ -831,6 +831,7 @@ void __kprobes recycle_rp_inst(struct kretprobe_instance *ri, void __kprobes kretprobe_hash_lock(struct task_struct *tsk, struct hlist_head **head, unsigned long *flags) +__acquires(hlist_lock) { unsigned long hash = hash_ptr(tsk, KPROBE_HASH_BITS); spinlock_t *hlist_lock; @@ -842,6 +843,7 @@ void __kprobes kretprobe_hash_lock(struct task_struct *tsk, static void __kprobes kretprobe_table_lock(unsigned long hash, unsigned long *flags) +__acquires(hlist_lock) { spinlock_t *hlist_lock = kretprobe_table_lock_ptr(hash); spin_lock_irqsave(hlist_lock, *flags); @@ -849,6 +851,7 @@ static void __kprobes kretprobe_table_lock(unsigned long hash, void __kprobes kretprobe_hash_unlock(struct task_struct *tsk, unsigned long *flags) +__releases(hlist_lock) { unsigned long hash = hash_ptr(tsk, KPROBE_HASH_BITS); spinlock_t *hlist_lock; @@ -859,6 +862,7 @@ void __kprobes kretprobe_hash_unlock(struct task_struct *tsk, static void __kprobes kretprobe_table_unlock(unsigned long hash, unsigned long *flags) +__releases(hlist_lock) { spinlock_t *hlist_lock = kretprobe_table_lock_ptr(hash); spin_unlock_irqrestore(hlist_lock, *flags); -- cgit v1.2.3 From 31915ab4cbf507aadab40847cf9989da5e88b090 Mon Sep 17 00:00:00 2001 From: Heiko Carstens Date: Thu, 16 Sep 2010 14:42:25 +0200 Subject: sched: Remove branch hints within context_switch() With 710390d9 "sched: Optimize branch hint in context_switch()" the branch hint logic within context_switch() got inversed. In fact the hints "if (likely(!mm))" and "if (likely(!prev->mm))" mean that it is likely that the previous and next task are kernel threads. That assumption is certainly counter intuitive, but Tim has shown that at least with his workload this is true. Nevertheless the truth is: it depends on the current workload. So just remove the annotations which also improves readability. Reported-by: Tim Blechmann Signed-off-by: Heiko Carstens Cc: Peter Zijlstra Cc: Mike Galbraith LKML-Reference: <20100916124225.GA2209@osiris.boeblingen.de.ibm.com> Signed-off-by: Ingo Molnar --- kernel/sched.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index b40b82e33590..16a1129f51ec 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -2845,14 +2845,14 @@ context_switch(struct rq *rq, struct task_struct *prev, */ arch_start_context_switch(prev); - if (likely(!mm)) { + if (!mm) { next->active_mm = oldmm; atomic_inc(&oldmm->mm_count); enter_lazy_tlb(oldmm, next); } else switch_mm(oldmm, mm, next); - if (likely(!prev->mm)) { + if (!prev->mm) { prev->active_mm = NULL; rq->prev_mm = oldmm; } -- cgit v1.2.3 From 068e35eee9ef98eb4cab55181977e24995d273be Mon Sep 17 00:00:00 2001 From: Matt Helsley Date: Mon, 13 Sep 2010 13:01:18 -0700 Subject: hw breakpoints: Fix pid namespace bug Hardware breakpoints can't be registered within pid namespaces because tsk->pid is passed rather than the pid in the current namespace. (See https://bugzilla.kernel.org/show_bug.cgi?id=17281 ) This is a quick fix demonstrating the problem but is not the best method of solving the problem since passing pids internally is not the best way to avoid pid namespace bugs. Subsequent patches will show a better solution. Much thanks to Frederic Weisbecker for doing the bulk of the work finding this bug. Reported-by: Robin Green Signed-off-by: Matt Helsley Signed-off-by: Peter Zijlstra Cc: Prasad Cc: Arnaldo Carvalho de Melo Cc: Steven Rostedt Cc: Will Deacon Cc: Mahesh Salgaonkar Cc: 2.6.33-2.6.35 LKML-Reference: Signed-off-by: Ingo Molnar Signed-off-by: Frederic Weisbecker --- kernel/hw_breakpoint.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/hw_breakpoint.c b/kernel/hw_breakpoint.c index d71a987fd2bf..c7c2aed9e2dc 100644 --- a/kernel/hw_breakpoint.c +++ b/kernel/hw_breakpoint.c @@ -433,7 +433,8 @@ register_user_hw_breakpoint(struct perf_event_attr *attr, perf_overflow_handler_t triggered, struct task_struct *tsk) { - return perf_event_create_kernel_counter(attr, -1, tsk->pid, triggered); + return perf_event_create_kernel_counter(attr, -1, task_pid_vnr(tsk), + triggered); } EXPORT_SYMBOL_GPL(register_user_hw_breakpoint); -- cgit v1.2.3 From d14b12d7adbf214f33eb59f800b5c3d5ed9268e8 Mon Sep 17 00:00:00 2001 From: Stephane Eranian Date: Fri, 17 Sep 2010 11:28:47 +0200 Subject: perf_events: Fix broken event grouping Events were not grouped anymore. The reason was that in perf_event_open(), the field event->group_leader was initialized before the function looked up the group_fd to find the event leader. This patch fixes this by reordering the code correctly. Signed-off-by: Stephane Eranian Signed-off-by: Peter Zijlstra Cc: Robert Richter LKML-Reference: <20100917093009.360420946@chello.nl> Signed-off-by: Ingo Molnar --- kernel/perf_event.c | 15 +++++++-------- 1 file changed, 7 insertions(+), 8 deletions(-) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index 86f394e15d53..ce95617f5d2c 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -5550,17 +5550,11 @@ SYSCALL_DEFINE5(perf_event_open, if (event_fd < 0) return event_fd; - event = perf_event_alloc(&attr, cpu, group_leader, NULL, NULL); - if (IS_ERR(event)) { - err = PTR_ERR(event); - goto err_fd; - } - if (group_fd != -1) { group_leader = perf_fget_light(group_fd, &fput_needed); if (IS_ERR(group_leader)) { err = PTR_ERR(group_leader); - goto err_alloc; + goto err_fd; } group_file = group_leader->filp; if (flags & PERF_FLAG_FD_OUTPUT) @@ -5569,6 +5563,12 @@ SYSCALL_DEFINE5(perf_event_open, group_leader = NULL; } + event = perf_event_alloc(&attr, cpu, group_leader, NULL, NULL); + if (IS_ERR(event)) { + err = PTR_ERR(event); + goto err_fd; + } + /* * Special case software events and allow them to be part of * any hardware group. @@ -5653,7 +5653,6 @@ err_context: put_ctx(ctx); err_group_fd: fput_light(group_file, fput_needed); -err_alloc: free_event(event); err_fd: put_unused_fd(event_fd); -- cgit v1.2.3 From b04243ef7006cda301819f54ee7ce0a3632489e3 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Fri, 17 Sep 2010 11:28:48 +0200 Subject: perf: Complete software pmu grouping Aside from allowing software events into a !software group, allow adding !software events to pure software groups. Once we've moved the software group and attached the first !software event, the group will no longer be a pure software group and hence no longer be eligible for movement, at which point the straight ctx comparison is correct again. Signed-off-by: Peter Zijlstra Cc: Stephane Eranian Cc: Robert Richter Cc: Paul Mackerras LKML-Reference: <20100917093009.410784731@chello.nl> Signed-off-by: Ingo Molnar --- kernel/perf_event.c | 65 ++++++++++++++++++++++++++++++++++++++++++++++++----- 1 file changed, 60 insertions(+), 5 deletions(-) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index ce95617f5d2c..6d7eef5f3c41 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -5184,6 +5184,7 @@ int perf_pmu_register(struct pmu *pmu) cpuctx = per_cpu_ptr(pmu->pmu_cpu_context, cpu); __perf_event_init_context(&cpuctx->ctx); + cpuctx->ctx.type = cpu_context; cpuctx->ctx.pmu = pmu; cpuctx->timer_interval = TICK_NSEC; hrtimer_init(&cpuctx->timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL); @@ -5517,7 +5518,8 @@ SYSCALL_DEFINE5(perf_event_open, struct perf_event_attr __user *, attr_uptr, pid_t, pid, int, cpu, int, group_fd, unsigned long, flags) { - struct perf_event *event, *group_leader = NULL, *output_event = NULL; + struct perf_event *group_leader = NULL, *output_event = NULL; + struct perf_event *event, *sibling; struct perf_event_attr attr; struct perf_event_context *ctx; struct file *event_file = NULL; @@ -5525,6 +5527,7 @@ SYSCALL_DEFINE5(perf_event_open, struct task_struct *task = NULL; struct pmu *pmu; int event_fd; + int move_group = 0; int fput_needed = 0; int err; @@ -5574,8 +5577,29 @@ SYSCALL_DEFINE5(perf_event_open, * any hardware group. */ pmu = event->pmu; - if ((pmu->task_ctx_nr == perf_sw_context) && group_leader) - pmu = group_leader->pmu; + + if (group_leader && + (is_software_event(event) != is_software_event(group_leader))) { + if (is_software_event(event)) { + /* + * If event and group_leader are not both a software + * event, and event is, then group leader is not. + * + * Allow the addition of software events to !software + * groups, this is safe because software events never + * fail to schedule. + */ + pmu = group_leader->pmu; + } else if (is_software_event(group_leader) && + (group_leader->group_flags & PERF_GROUP_SOFTWARE)) { + /* + * In case the group is a pure software group, and we + * try to add a hardware event, move the whole group to + * the hardware context. + */ + move_group = 1; + } + } if (pid != -1) task = find_lively_task_by_vpid(pid); @@ -5605,8 +5629,14 @@ SYSCALL_DEFINE5(perf_event_open, * Do not allow to attach to a group in a different * task or CPU context: */ - if (group_leader->ctx != ctx) - goto err_context; + if (move_group) { + if (group_leader->ctx->type != ctx->type) + goto err_context; + } else { + if (group_leader->ctx != ctx) + goto err_context; + } + /* * Only a group leader can be exclusive or pinned */ @@ -5626,9 +5656,34 @@ SYSCALL_DEFINE5(perf_event_open, goto err_context; } + if (move_group) { + struct perf_event_context *gctx = group_leader->ctx; + + mutex_lock(&gctx->mutex); + perf_event_remove_from_context(group_leader); + list_for_each_entry(sibling, &group_leader->sibling_list, + group_entry) { + perf_event_remove_from_context(sibling); + put_ctx(gctx); + } + mutex_unlock(&gctx->mutex); + put_ctx(gctx); + } + event->filp = event_file; WARN_ON_ONCE(ctx->parent_ctx); mutex_lock(&ctx->mutex); + + if (move_group) { + perf_install_in_context(ctx, group_leader, cpu); + get_ctx(ctx); + list_for_each_entry(sibling, &group_leader->sibling_list, + group_entry) { + perf_install_in_context(ctx, sibling, cpu); + get_ctx(ctx); + } + } + perf_install_in_context(ctx, event, cpu); ++ctx->generation; mutex_unlock(&ctx->mutex); -- cgit v1.2.3 From 917bdd1c9b7b0f4c22f2504c2f0c1074c8ab9df7 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Fri, 17 Sep 2010 11:28:49 +0200 Subject: perf: Fix perf_event_exit_cpu_context() Use the right cpu-context.. spotted by preempt warning on hot-unplug Signed-off-by: Peter Zijlstra Cc: Stephane Eranian Cc: Robert Richter LKML-Reference: <20100917093009.461794357@chello.nl> Signed-off-by: Ingo Molnar --- kernel/perf_event.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index 6d7eef5f3c41..27332e5f51a7 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -6269,14 +6269,13 @@ static void perf_event_exit_cpu_context(int cpu) idx = srcu_read_lock(&pmus_srcu); list_for_each_entry_rcu(pmu, &pmus, entry) { - ctx = &this_cpu_ptr(pmu->pmu_cpu_context)->ctx; + ctx = &per_cpu_ptr(pmu->pmu_cpu_context, cpu)->ctx; mutex_lock(&ctx->mutex); smp_call_function_single(cpu, __perf_event_exit_context, ctx, 1); mutex_unlock(&ctx->mutex); } srcu_read_unlock(&pmus_srcu, idx); - } static void perf_event_exit_cpu(int cpu) -- cgit v1.2.3 From e9d2b064149ff7ef4acbc65a1b9374ac8b218d3e Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Fri, 17 Sep 2010 11:28:50 +0200 Subject: perf: Undo the per cpu-context timer stuff Revert the timer per cpu-context timers because of unfortunate nohz interaction. Fixing that would have been somewhat ugly, so go back to driving things from the regular tick. Provide a jiffies interval feature for people who want slower rotations. Signed-off-by: Peter Zijlstra Cc: Stephane Eranian Cc: Robert Richter Cc: Yinghai Lu LKML-Reference: <20100917093009.519845633@chello.nl> Signed-off-by: Ingo Molnar --- kernel/perf_event.c | 79 +++++++++++++++++++++++++++++++++-------------------- kernel/sched.c | 2 ++ 2 files changed, 51 insertions(+), 30 deletions(-) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index 27332e5f51a7..baae1367e945 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -77,23 +77,22 @@ void perf_pmu_enable(struct pmu *pmu) pmu->pmu_enable(pmu); } +static DEFINE_PER_CPU(struct list_head, rotation_list); + +/* + * perf_pmu_rotate_start() and perf_rotate_context() are fully serialized + * because they're strictly cpu affine and rotate_start is called with IRQs + * disabled, while rotate_context is called from IRQ context. + */ static void perf_pmu_rotate_start(struct pmu *pmu) { struct perf_cpu_context *cpuctx = this_cpu_ptr(pmu->pmu_cpu_context); + struct list_head *head = &__get_cpu_var(rotation_list); - if (hrtimer_active(&cpuctx->timer)) - return; + WARN_ON(!irqs_disabled()); - __hrtimer_start_range_ns(&cpuctx->timer, - ns_to_ktime(cpuctx->timer_interval), 0, - HRTIMER_MODE_REL_PINNED, 0); -} - -static void perf_pmu_rotate_stop(struct pmu *pmu) -{ - struct perf_cpu_context *cpuctx = this_cpu_ptr(pmu->pmu_cpu_context); - - hrtimer_cancel(&cpuctx->timer); + if (list_empty(&cpuctx->rotation_list)) + list_add(&cpuctx->rotation_list, head); } static void get_ctx(struct perf_event_context *ctx) @@ -1607,36 +1606,33 @@ static void rotate_ctx(struct perf_event_context *ctx) } /* - * Cannot race with ->pmu_rotate_start() because this is ran from hardirq - * context, and ->pmu_rotate_start() is called with irqs disabled (both are - * cpu affine, so there are no SMP races). + * perf_pmu_rotate_start() and perf_rotate_context() are fully serialized + * because they're strictly cpu affine and rotate_start is called with IRQs + * disabled, while rotate_context is called from IRQ context. */ -static enum hrtimer_restart perf_event_context_tick(struct hrtimer *timer) +static void perf_rotate_context(struct perf_cpu_context *cpuctx) { - enum hrtimer_restart restart = HRTIMER_NORESTART; - struct perf_cpu_context *cpuctx; + u64 interval = (u64)cpuctx->jiffies_interval * TICK_NSEC; struct perf_event_context *ctx = NULL; - int rotate = 0; - - cpuctx = container_of(timer, struct perf_cpu_context, timer); + int rotate = 0, remove = 1; if (cpuctx->ctx.nr_events) { - restart = HRTIMER_RESTART; + remove = 0; if (cpuctx->ctx.nr_events != cpuctx->ctx.nr_active) rotate = 1; } ctx = cpuctx->task_ctx; if (ctx && ctx->nr_events) { - restart = HRTIMER_RESTART; + remove = 0; if (ctx->nr_events != ctx->nr_active) rotate = 1; } perf_pmu_disable(cpuctx->ctx.pmu); - perf_ctx_adjust_freq(&cpuctx->ctx, cpuctx->timer_interval); + perf_ctx_adjust_freq(&cpuctx->ctx, interval); if (ctx) - perf_ctx_adjust_freq(ctx, cpuctx->timer_interval); + perf_ctx_adjust_freq(ctx, interval); if (!rotate) goto done; @@ -1654,10 +1650,24 @@ static enum hrtimer_restart perf_event_context_tick(struct hrtimer *timer) task_ctx_sched_in(ctx, EVENT_FLEXIBLE); done: + if (remove) + list_del_init(&cpuctx->rotation_list); + perf_pmu_enable(cpuctx->ctx.pmu); - hrtimer_forward_now(timer, ns_to_ktime(cpuctx->timer_interval)); +} + +void perf_event_task_tick(void) +{ + struct list_head *head = &__get_cpu_var(rotation_list); + struct perf_cpu_context *cpuctx, *tmp; - return restart; + WARN_ON(!irqs_disabled()); + + list_for_each_entry_safe(cpuctx, tmp, head, rotation_list) { + if (cpuctx->jiffies_interval == 1 || + !(jiffies % cpuctx->jiffies_interval)) + perf_rotate_context(cpuctx); + } } static int event_enable_on_exec(struct perf_event *event, @@ -5186,9 +5196,8 @@ int perf_pmu_register(struct pmu *pmu) __perf_event_init_context(&cpuctx->ctx); cpuctx->ctx.type = cpu_context; cpuctx->ctx.pmu = pmu; - cpuctx->timer_interval = TICK_NSEC; - hrtimer_init(&cpuctx->timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL); - cpuctx->timer.function = perf_event_context_tick; + cpuctx->jiffies_interval = 1; + INIT_LIST_HEAD(&cpuctx->rotation_list); } got_cpu_context: @@ -6229,6 +6238,7 @@ static void __init perf_event_init_all_cpus(void) for_each_possible_cpu(cpu) { swhash = &per_cpu(swevent_htable, cpu); mutex_init(&swhash->hlist_mutex); + INIT_LIST_HEAD(&per_cpu(rotation_list, cpu)); } } @@ -6248,6 +6258,15 @@ static void __cpuinit perf_event_init_cpu(int cpu) } #ifdef CONFIG_HOTPLUG_CPU +static void perf_pmu_rotate_stop(struct pmu *pmu) +{ + struct perf_cpu_context *cpuctx = this_cpu_ptr(pmu->pmu_cpu_context); + + WARN_ON(!irqs_disabled()); + + list_del_init(&cpuctx->rotation_list); +} + static void __perf_event_exit_context(void *__info) { struct perf_event_context *ctx = __info; diff --git a/kernel/sched.c b/kernel/sched.c index 1c3ea7a55b7b..794819eab9ca 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -3584,6 +3584,8 @@ void scheduler_tick(void) curr->sched_class->task_tick(rq, curr, 0); raw_spin_unlock(&rq->lock); + perf_event_task_tick(); + #ifdef CONFIG_SMP rq->idle_at_tick = idle_cpu(cpu); trigger_load_balance(rq, cpu); -- cgit v1.2.3 From 1dcc41bb037533839753df983d31778b30b67d93 Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Tue, 14 Sep 2010 21:43:46 +0900 Subject: futex: Change 3rd arg of fetch_robust_entry() to unsigned int* Sparse complains: kernel/futex.c:2495:59: warning: incorrect type in argument 3 (different signedness) Make 3rd argument of fetch_robust_entry() 'unsigned int'. Signed-off-by: Namhyung Kim Cc: Peter Zijlstra Cc: Darren Hart LKML-Reference: <1284468228-8723-1-git-send-email-namhyung@gmail.com> Signed-off-by: Thomas Gleixner --- kernel/futex.c | 2 +- kernel/futex_compat.c | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/futex.c b/kernel/futex.c index 6a3a5fa1526d..464de2751ff9 100644 --- a/kernel/futex.c +++ b/kernel/futex.c @@ -2458,7 +2458,7 @@ retry: */ static inline int fetch_robust_entry(struct robust_list __user **entry, struct robust_list __user * __user *head, - int *pi) + unsigned int *pi) { unsigned long uentry; diff --git a/kernel/futex_compat.c b/kernel/futex_compat.c index d49afb2395e5..06da4dfc339b 100644 --- a/kernel/futex_compat.c +++ b/kernel/futex_compat.c @@ -19,7 +19,7 @@ */ static inline int fetch_robust_entry(compat_uptr_t *uentry, struct robust_list __user **entry, - compat_uptr_t __user *head, int *pi) + compat_uptr_t __user *head, unsigned int *pi) { if (get_user(*uentry, head)) return -EFAULT; -- cgit v1.2.3 From a3c74c52570c0c4ac90c9a0216de800c39089ba7 Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Tue, 14 Sep 2010 21:43:47 +0900 Subject: futex: Mark restart_block.futex.uaddr[2] __user @uaddr and @uaddr2 fields in restart_block.futex are user pointers. Add __user and remove unnecessary casts. Signed-off-by: Namhyung Kim Cc: Peter Zijlstra Cc: Darren Hart LKML-Reference: <1284468228-8723-2-git-send-email-namhyung@gmail.com> Signed-off-by: Thomas Gleixner --- kernel/futex.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/futex.c b/kernel/futex.c index 464de2751ff9..45e448a5e440 100644 --- a/kernel/futex.c +++ b/kernel/futex.c @@ -1843,7 +1843,7 @@ retry: restart = ¤t_thread_info()->restart_block; restart->fn = futex_wait_restart; - restart->futex.uaddr = (u32 *)uaddr; + restart->futex.uaddr = uaddr; restart->futex.val = val; restart->futex.time = abs_time->tv64; restart->futex.bitset = bitset; @@ -1869,7 +1869,7 @@ out: static long futex_wait_restart(struct restart_block *restart) { - u32 __user *uaddr = (u32 __user *)restart->futex.uaddr; + u32 __user *uaddr = restart->futex.uaddr; int fshared = 0; ktime_t t, *tp = NULL; -- cgit v1.2.3 From 15e408cd6ccc3f4f453d87ccd5bc7a84d59feb96 Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Tue, 14 Sep 2010 21:43:48 +0900 Subject: futex: Add lock context annotations queue_lock/unlock/me() and unqueue_me_pi() grab/release spinlocks but are missing proper annotations. Add them. Signed-off-by: Namhyung Kim Cc: Peter Zijlstra Cc: Darren Hart LKML-Reference: <1284468228-8723-3-git-send-email-namhyung@gmail.com> Signed-off-by: Thomas Gleixner --- kernel/futex.c | 4 ++++ 1 file changed, 4 insertions(+) (limited to 'kernel') diff --git a/kernel/futex.c b/kernel/futex.c index 45e448a5e440..92a31d4cd564 100644 --- a/kernel/futex.c +++ b/kernel/futex.c @@ -1360,6 +1360,7 @@ out: /* The key must be already stored in q->key. */ static inline struct futex_hash_bucket *queue_lock(struct futex_q *q) + __acquires(&hb->lock) { struct futex_hash_bucket *hb; @@ -1373,6 +1374,7 @@ static inline struct futex_hash_bucket *queue_lock(struct futex_q *q) static inline void queue_unlock(struct futex_q *q, struct futex_hash_bucket *hb) + __releases(&hb->lock) { spin_unlock(&hb->lock); drop_futex_key_refs(&q->key); @@ -1391,6 +1393,7 @@ queue_unlock(struct futex_q *q, struct futex_hash_bucket *hb) * an example). */ static inline void queue_me(struct futex_q *q, struct futex_hash_bucket *hb) + __releases(&hb->lock) { int prio; @@ -1471,6 +1474,7 @@ retry: * and dropped here. */ static void unqueue_me_pi(struct futex_q *q) + __releases(q->lock_ptr) { WARN_ON(plist_node_empty(&q->list)); plist_del(&q->list, &q->list.plist); -- cgit v1.2.3 From 401a8d048eadfbe1b1c1bf53d3b614fcc894c61a Mon Sep 17 00:00:00 2001 From: Tejun Heo Date: Thu, 16 Sep 2010 10:36:00 +0200 Subject: workqueue: cleanup flush/cancel functions Make the following cleanup changes. * Relocate flush/cancel function prototypes and definitions. * Relocate wait_on_cpu_work() and wait_on_work() before try_to_grab_pending(). These will be used to implement flush_work_sync(). * Make all flush/cancel functions return bool instead of int. * Update wait_on_cpu_work() and wait_on_work() to return %true if they actually waited. * Add / update comments. This patch doesn't cause any functional changes. Signed-off-by: Tejun Heo --- kernel/workqueue.c | 175 ++++++++++++++++++++++++++++------------------------- 1 file changed, 94 insertions(+), 81 deletions(-) (limited to 'kernel') diff --git a/kernel/workqueue.c b/kernel/workqueue.c index f77afd939229..1240b9d94b03 100644 --- a/kernel/workqueue.c +++ b/kernel/workqueue.c @@ -2327,16 +2327,24 @@ out_unlock: EXPORT_SYMBOL_GPL(flush_workqueue); /** - * flush_work - block until a work_struct's callback has terminated - * @work: the work which is to be flushed + * flush_work - wait for a work to finish executing the last queueing instance + * @work: the work to flush * - * Returns false if @work has already terminated. + * Wait until @work has finished execution. This function considers + * only the last queueing instance of @work. If @work has been + * enqueued across different CPUs on a non-reentrant workqueue or on + * multiple workqueues, @work might still be executing on return on + * some of the CPUs from earlier queueing. * - * It is expected that, prior to calling flush_work(), the caller has - * arranged for the work to not be requeued, otherwise it doesn't make - * sense to use this function. + * If @work was queued only on a non-reentrant, ordered or unbound + * workqueue, @work is guaranteed to be idle on return if it hasn't + * been requeued since flush started. + * + * RETURNS: + * %true if flush_work() waited for the work to finish execution, + * %false if it was already idle. */ -int flush_work(struct work_struct *work) +bool flush_work(struct work_struct *work) { struct worker *worker = NULL; struct global_cwq *gcwq; @@ -2374,13 +2382,49 @@ int flush_work(struct work_struct *work) wait_for_completion(&barr.done); destroy_work_on_stack(&barr.work); - return 1; + return true; already_gone: spin_unlock_irq(&gcwq->lock); - return 0; + return false; } EXPORT_SYMBOL_GPL(flush_work); +static bool wait_on_cpu_work(struct global_cwq *gcwq, struct work_struct *work) +{ + struct wq_barrier barr; + struct worker *worker; + + spin_lock_irq(&gcwq->lock); + + worker = find_worker_executing_work(gcwq, work); + if (unlikely(worker)) + insert_wq_barrier(worker->current_cwq, &barr, work, worker); + + spin_unlock_irq(&gcwq->lock); + + if (unlikely(worker)) { + wait_for_completion(&barr.done); + destroy_work_on_stack(&barr.work); + return true; + } else + return false; +} + +static bool wait_on_work(struct work_struct *work) +{ + bool ret = false; + int cpu; + + might_sleep(); + + lock_map_acquire(&work->lockdep_map); + lock_map_release(&work->lockdep_map); + + for_each_gcwq_cpu(cpu) + ret |= wait_on_cpu_work(get_gcwq(cpu), work); + return ret; +} + /* * Upon a successful return (>= 0), the caller "owns" WORK_STRUCT_PENDING bit, * so this work can't be re-armed in any way. @@ -2423,39 +2467,7 @@ static int try_to_grab_pending(struct work_struct *work) return ret; } -static void wait_on_cpu_work(struct global_cwq *gcwq, struct work_struct *work) -{ - struct wq_barrier barr; - struct worker *worker; - - spin_lock_irq(&gcwq->lock); - - worker = find_worker_executing_work(gcwq, work); - if (unlikely(worker)) - insert_wq_barrier(worker->current_cwq, &barr, work, worker); - - spin_unlock_irq(&gcwq->lock); - - if (unlikely(worker)) { - wait_for_completion(&barr.done); - destroy_work_on_stack(&barr.work); - } -} - -static void wait_on_work(struct work_struct *work) -{ - int cpu; - - might_sleep(); - - lock_map_acquire(&work->lockdep_map); - lock_map_release(&work->lockdep_map); - - for_each_gcwq_cpu(cpu) - wait_on_cpu_work(get_gcwq(cpu), work); -} - -static int __cancel_work_timer(struct work_struct *work, +static bool __cancel_work_timer(struct work_struct *work, struct timer_list* timer) { int ret; @@ -2472,42 +2484,60 @@ static int __cancel_work_timer(struct work_struct *work, } /** - * cancel_work_sync - block until a work_struct's callback has terminated - * @work: the work which is to be flushed - * - * Returns true if @work was pending. + * cancel_work_sync - cancel a work and wait for it to finish + * @work: the work to cancel * - * cancel_work_sync() will cancel the work if it is queued. If the work's - * callback appears to be running, cancel_work_sync() will block until it - * has completed. + * Cancel @work and wait for its execution to finish. This function + * can be used even if the work re-queues itself or migrates to + * another workqueue. On return from this function, @work is + * guaranteed to be not pending or executing on any CPU. * - * It is possible to use this function if the work re-queues itself. It can - * cancel the work even if it migrates to another workqueue, however in that - * case it only guarantees that work->func() has completed on the last queued - * workqueue. - * - * cancel_work_sync(&delayed_work->work) should be used only if ->timer is not - * pending, otherwise it goes into a busy-wait loop until the timer expires. + * cancel_work_sync(&delayed_work->work) must not be used for + * delayed_work's. Use cancel_delayed_work_sync() instead. * - * The caller must ensure that workqueue_struct on which this work was last + * The caller must ensure that the workqueue on which @work was last * queued can't be destroyed before this function returns. + * + * RETURNS: + * %true if @work was pending, %false otherwise. */ -int cancel_work_sync(struct work_struct *work) +bool cancel_work_sync(struct work_struct *work) { return __cancel_work_timer(work, NULL); } EXPORT_SYMBOL_GPL(cancel_work_sync); /** - * cancel_delayed_work_sync - reliably kill off a delayed work. - * @dwork: the delayed work struct + * flush_delayed_work - wait for a dwork to finish executing the last queueing + * @dwork: the delayed work to flush + * + * Delayed timer is cancelled and the pending work is queued for + * immediate execution. Like flush_work(), this function only + * considers the last queueing instance of @dwork. + * + * RETURNS: + * %true if flush_work() waited for the work to finish execution, + * %false if it was already idle. + */ +bool flush_delayed_work(struct delayed_work *dwork) +{ + if (del_timer_sync(&dwork->timer)) + __queue_work(raw_smp_processor_id(), + get_work_cwq(&dwork->work)->wq, &dwork->work); + return flush_work(&dwork->work); +} +EXPORT_SYMBOL(flush_delayed_work); + +/** + * cancel_delayed_work_sync - cancel a delayed work and wait for it to finish + * @dwork: the delayed work cancel * - * Returns true if @dwork was pending. + * This is cancel_work_sync() for delayed works. * - * It is possible to use this function if @dwork rearms itself via queue_work() - * or queue_delayed_work(). See also the comment for cancel_work_sync(). + * RETURNS: + * %true if @dwork was pending, %false otherwise. */ -int cancel_delayed_work_sync(struct delayed_work *dwork) +bool cancel_delayed_work_sync(struct delayed_work *dwork) { return __cancel_work_timer(&dwork->work, &dwork->timer); } @@ -2558,23 +2588,6 @@ int schedule_delayed_work(struct delayed_work *dwork, } EXPORT_SYMBOL(schedule_delayed_work); -/** - * flush_delayed_work - block until a dwork_struct's callback has terminated - * @dwork: the delayed work which is to be flushed - * - * Any timeout is cancelled, and any pending work is run immediately. - */ -void flush_delayed_work(struct delayed_work *dwork) -{ - if (del_timer_sync(&dwork->timer)) { - __queue_work(get_cpu(), get_work_cwq(&dwork->work)->wq, - &dwork->work); - put_cpu(); - } - flush_work(&dwork->work); -} -EXPORT_SYMBOL(flush_delayed_work); - /** * schedule_delayed_work_on - queue work in global workqueue on CPU after delay * @cpu: cpu to use -- cgit v1.2.3 From baf59022c37d43f202e62d5130e4bac5e825b426 Mon Sep 17 00:00:00 2001 From: Tejun Heo Date: Thu, 16 Sep 2010 10:42:16 +0200 Subject: workqueue: factor out start_flush_work() Factor out start_flush_work() from flush_work(). start_flush_work() has @wait_executing argument which controls whether the barrier is queued only if the work is pending or also if executing. As flush_work() needs to wait for execution too, it uses %true. This commit doesn't cause any behavior difference. start_flush_work() will be used to implement flush_work_sync(). Signed-off-by: Tejun Heo --- kernel/workqueue.c | 64 +++++++++++++++++++++++++++++++----------------------- 1 file changed, 37 insertions(+), 27 deletions(-) (limited to 'kernel') diff --git a/kernel/workqueue.c b/kernel/workqueue.c index 1240b9d94b03..33d31d768706 100644 --- a/kernel/workqueue.c +++ b/kernel/workqueue.c @@ -2326,35 +2326,17 @@ out_unlock: } EXPORT_SYMBOL_GPL(flush_workqueue); -/** - * flush_work - wait for a work to finish executing the last queueing instance - * @work: the work to flush - * - * Wait until @work has finished execution. This function considers - * only the last queueing instance of @work. If @work has been - * enqueued across different CPUs on a non-reentrant workqueue or on - * multiple workqueues, @work might still be executing on return on - * some of the CPUs from earlier queueing. - * - * If @work was queued only on a non-reentrant, ordered or unbound - * workqueue, @work is guaranteed to be idle on return if it hasn't - * been requeued since flush started. - * - * RETURNS: - * %true if flush_work() waited for the work to finish execution, - * %false if it was already idle. - */ -bool flush_work(struct work_struct *work) +static bool start_flush_work(struct work_struct *work, struct wq_barrier *barr, + bool wait_executing) { struct worker *worker = NULL; struct global_cwq *gcwq; struct cpu_workqueue_struct *cwq; - struct wq_barrier barr; might_sleep(); gcwq = get_work_gcwq(work); if (!gcwq) - return 0; + return false; spin_lock_irq(&gcwq->lock); if (!list_empty(&work->entry)) { @@ -2367,26 +2349,54 @@ bool flush_work(struct work_struct *work) cwq = get_work_cwq(work); if (unlikely(!cwq || gcwq != cwq->gcwq)) goto already_gone; - } else { + } else if (wait_executing) { worker = find_worker_executing_work(gcwq, work); if (!worker) goto already_gone; cwq = worker->current_cwq; - } + } else + goto already_gone; - insert_wq_barrier(cwq, &barr, work, worker); + insert_wq_barrier(cwq, barr, work, worker); spin_unlock_irq(&gcwq->lock); lock_map_acquire(&cwq->wq->lockdep_map); lock_map_release(&cwq->wq->lockdep_map); - - wait_for_completion(&barr.done); - destroy_work_on_stack(&barr.work); return true; already_gone: spin_unlock_irq(&gcwq->lock); return false; } + +/** + * flush_work - wait for a work to finish executing the last queueing instance + * @work: the work to flush + * + * Wait until @work has finished execution. This function considers + * only the last queueing instance of @work. If @work has been + * enqueued across different CPUs on a non-reentrant workqueue or on + * multiple workqueues, @work might still be executing on return on + * some of the CPUs from earlier queueing. + * + * If @work was queued only on a non-reentrant, ordered or unbound + * workqueue, @work is guaranteed to be idle on return if it hasn't + * been requeued since flush started. + * + * RETURNS: + * %true if flush_work() waited for the work to finish execution, + * %false if it was already idle. + */ +bool flush_work(struct work_struct *work) +{ + struct wq_barrier barr; + + if (start_flush_work(work, &barr, true)) { + wait_for_completion(&barr.done); + destroy_work_on_stack(&barr.work); + return true; + } else + return false; +} EXPORT_SYMBOL_GPL(flush_work); static bool wait_on_cpu_work(struct global_cwq *gcwq, struct work_struct *work) -- cgit v1.2.3 From 09383498c5d35262e643bfdbae84826177a3c624 Mon Sep 17 00:00:00 2001 From: Tejun Heo Date: Thu, 16 Sep 2010 10:48:29 +0200 Subject: workqueue: implement flush[_delayed]_work_sync() Implement flush[_delayed]_work_sync(). These are flush functions which also make sure no CPU is still executing the target work from earlier queueing instances. These are similar to cancel[_delayed]_work_sync() except that the target work item is flushed instead of cancelled. Signed-off-by: Tejun Heo --- kernel/workqueue.c | 56 ++++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 56 insertions(+) (limited to 'kernel') diff --git a/kernel/workqueue.c b/kernel/workqueue.c index 33d31d768706..19e4bc15ee99 100644 --- a/kernel/workqueue.c +++ b/kernel/workqueue.c @@ -2435,6 +2435,41 @@ static bool wait_on_work(struct work_struct *work) return ret; } +/** + * flush_work_sync - wait until a work has finished execution + * @work: the work to flush + * + * Wait until @work has finished execution. On return, it's + * guaranteed that all queueing instances of @work which happened + * before this function is called are finished. In other words, if + * @work hasn't been requeued since this function was called, @work is + * guaranteed to be idle on return. + * + * RETURNS: + * %true if flush_work_sync() waited for the work to finish execution, + * %false if it was already idle. + */ +bool flush_work_sync(struct work_struct *work) +{ + struct wq_barrier barr; + bool pending, waited; + + /* we'll wait for executions separately, queue barr only if pending */ + pending = start_flush_work(work, &barr, false); + + /* wait for executions to finish */ + waited = wait_on_work(work); + + /* wait for the pending one */ + if (pending) { + wait_for_completion(&barr.done); + destroy_work_on_stack(&barr.work); + } + + return pending || waited; +} +EXPORT_SYMBOL_GPL(flush_work_sync); + /* * Upon a successful return (>= 0), the caller "owns" WORK_STRUCT_PENDING bit, * so this work can't be re-armed in any way. @@ -2538,6 +2573,27 @@ bool flush_delayed_work(struct delayed_work *dwork) } EXPORT_SYMBOL(flush_delayed_work); +/** + * flush_delayed_work_sync - wait for a dwork to finish + * @dwork: the delayed work to flush + * + * Delayed timer is cancelled and the pending work is queued for + * execution immediately. Other than timer handling, its behavior + * is identical to flush_work_sync(). + * + * RETURNS: + * %true if flush_work_sync() waited for the work to finish execution, + * %false if it was already idle. + */ +bool flush_delayed_work_sync(struct delayed_work *dwork) +{ + if (del_timer_sync(&dwork->timer)) + __queue_work(raw_smp_processor_id(), + get_work_cwq(&dwork->work)->wq, &dwork->work); + return flush_work_sync(&dwork->work); +} +EXPORT_SYMBOL(flush_delayed_work_sync); + /** * cancel_delayed_work_sync - cancel a delayed work and wait for it to finish * @dwork: the delayed work cancel -- cgit v1.2.3 From f6c3f1686e7ec1dd8725a9a3dcb857dfd0c7a5bf Mon Sep 17 00:00:00 2001 From: Suresh Siddha Date: Mon, 13 Sep 2010 11:02:21 -0700 Subject: sched: Fix nohz balance kick There's a situation where the nohz balancer will try to wake itself: cpu-x is idle which is also ilb_cpu got a scheduler tick during idle and the nohz_kick_needed() in trigger_load_balance() checks for rq_x->nr_running which might not be zero (because of someone waking a task on this rq etc) and this leads to the situation of the cpu-x sending a kick to itself. And this can cause a lockup. Avoid this by not marking ourself eligible for kicking. Signed-off-by: Suresh Siddha Signed-off-by: Peter Zijlstra LKML-Reference: <1284400941.2684.19.camel@sbsiddha-MOBL3.sc.intel.com> Signed-off-by: Ingo Molnar --- kernel/sched_fair.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c index a171138a9402..db3f674ca49d 100644 --- a/kernel/sched_fair.c +++ b/kernel/sched_fair.c @@ -3630,7 +3630,7 @@ static inline int nohz_kick_needed(struct rq *rq, int cpu) if (time_before(now, nohz.next_balance)) return 0; - if (!rq->nr_running) + if (rq->idle_at_tick) return 0; first_pick_cpu = atomic_read(&nohz.first_pick_cpu); -- cgit v1.2.3 From 41945f6ccf1e86f87fddf6b32db9cf431c05fb54 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Thu, 16 Sep 2010 19:17:24 +0200 Subject: perf: Avoid RCU vs preemption assumptions The per-pmu per-cpu context patch converted things from get_cpu_var() to this_cpu_ptr(), but that only works if rcu_read_lock() actually disables preemption, and since there is no such guarantee, we need to fix that. Use the newly introduced {get,put}_cpu_ptr(). Signed-off-by: Peter Zijlstra Cc: Tejun Heo LKML-Reference: <20100917093009.308453028@chello.nl> Signed-off-by: Ingo Molnar --- kernel/perf_event.c | 18 ++++++++++++------ 1 file changed, 12 insertions(+), 6 deletions(-) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index baae1367e945..c16158c77dfd 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -3836,18 +3836,20 @@ static void perf_event_task_event(struct perf_task_event *task_event) rcu_read_lock(); list_for_each_entry_rcu(pmu, &pmus, entry) { - cpuctx = this_cpu_ptr(pmu->pmu_cpu_context); + cpuctx = get_cpu_ptr(pmu->pmu_cpu_context); perf_event_task_ctx(&cpuctx->ctx, task_event); ctx = task_event->task_ctx; if (!ctx) { ctxn = pmu->task_ctx_nr; if (ctxn < 0) - continue; + goto next; ctx = rcu_dereference(current->perf_event_ctxp[ctxn]); } if (ctx) perf_event_task_ctx(ctx, task_event); +next: + put_cpu_ptr(pmu->pmu_cpu_context); } rcu_read_unlock(); } @@ -3969,16 +3971,18 @@ static void perf_event_comm_event(struct perf_comm_event *comm_event) rcu_read_lock(); list_for_each_entry_rcu(pmu, &pmus, entry) { - cpuctx = this_cpu_ptr(pmu->pmu_cpu_context); + cpuctx = get_cpu_ptr(pmu->pmu_cpu_context); perf_event_comm_ctx(&cpuctx->ctx, comm_event); ctxn = pmu->task_ctx_nr; if (ctxn < 0) - continue; + goto next; ctx = rcu_dereference(current->perf_event_ctxp[ctxn]); if (ctx) perf_event_comm_ctx(ctx, comm_event); +next: + put_cpu_ptr(pmu->pmu_cpu_context); } rcu_read_unlock(); } @@ -4152,19 +4156,21 @@ got_name: rcu_read_lock(); list_for_each_entry_rcu(pmu, &pmus, entry) { - cpuctx = this_cpu_ptr(pmu->pmu_cpu_context); + cpuctx = get_cpu_ptr(pmu->pmu_cpu_context); perf_event_mmap_ctx(&cpuctx->ctx, mmap_event, vma->vm_flags & VM_EXEC); ctxn = pmu->task_ctx_nr; if (ctxn < 0) - continue; + goto next; ctx = rcu_dereference(current->perf_event_ctxp[ctxn]); if (ctx) { perf_event_mmap_ctx(ctx, mmap_event, vma->vm_flags & VM_EXEC); } +next: + put_cpu_ptr(pmu->pmu_cpu_context); } rcu_read_unlock(); -- cgit v1.2.3 From 58b26c4c025778c09c7a1438ff185080e11b7d0a Mon Sep 17 00:00:00 2001 From: Venkatesh Pallipadi Date: Fri, 10 Sep 2010 18:19:17 -0700 Subject: sched: Increment cache_nice_tries only on periodic lb scheduler uses cache_nice_tries as an indicator to do cache_hot and active load balance, when normal load balance fails. Currently, this value is changed on any failed load balance attempt. That ends up being not so nice to workloads that enter/exit idle often, as they do more frequent new_idle balance and that pretty soon results in cache hot tasks being pulled in. Making the cache_nice_tries ignore failed new_idle balance seems to make better sense. With that only the failed load balance in periodic load balance gets accounted and the rate of accumulation of cache_nice_tries will not depend on idle entry/exit (short running sleep-wakeup kind of tasks). This reduces movement of cache_hot tasks. schedstat diff (after-before) excerpt from a workload that has frequent and short wakeup-idle pattern (:2 in cpu col below refers to NEWIDLE idx) This snapshot was across ~400 seconds. Without this change: domainstats: domain0 cpu cnt bln fld imb gain hgain nobusyq nobusyg 0:2 306487 219575 73167 110069413 44583 19070 1172 218403 1:2 292139 194853 81421 120893383 50745 21902 1259 193594 2:2 283166 174607 91359 129699642 54931 23688 1287 173320 3:2 273998 161788 93991 132757146 57122 24351 1366 160422 4:2 289851 215692 62190 83398383 36377 13680 851 214841 5:2 316312 222146 77605 117582154 49948 20281 988 221158 6:2 297172 195596 83623 122133390 52801 21301 929 194667 7:2 283391 178078 86378 126622761 55122 22239 928 177150 8:2 297655 210359 72995 110246694 45798 19777 1125 209234 9:2 297357 202011 79363 119753474 50953 22088 1089 200922 10:2 278797 178703 83180 122514385 52969 22726 1128 177575 11:2 272661 167669 86978 127342327 55857 24342 1195 166474 12:2 293039 204031 73211 110282059 47285 19651 948 203083 13:2 289502 196762 76803 114712942 49339 20547 1016 195746 14:2 264446 169609 78292 115715605 50459 21017 982 168627 15:2 260968 163660 80142 116811793 51483 21281 1064 162596 With this change: domainstats: domain0 cpu cnt bln fld imb gain hgain nobusyq nobusyg 0:2 272347 187380 77455 105420270 24975 1 953 186427 1:2 267276 172360 86234 116242264 28087 6 1028 171332 2:2 259769 156777 93281 123243134 30555 1 1043 155734 3:2 250870 143129 97627 127370868 32026 6 1188 141941 4:2 248422 177116 64096 78261112 22202 2 757 176359 5:2 275595 180683 84950 116075022 29400 6 778 179905 6:2 262418 162609 88944 119256898 31056 4 817 161792 7:2 252204 147946 92646 122388300 32879 4 824 147122 8:2 262335 172239 81631 110477214 26599 4 864 171375 9:2 261563 164775 88016 117203621 28331 3 849 163926 10:2 243389 140949 93379 121353071 29585 2 909 140040 11:2 242795 134651 98310 124768957 30895 2 1016 133635 12:2 255234 166622 79843 104696912 26483 4 746 165876 13:2 244944 151595 83855 109808099 27787 3 801 150794 14:2 241301 140982 89935 116954383 30403 6 845 140137 15:2 232271 128564 92821 119185207 31207 4 1416 127148 Signed-off-by: Venkatesh Pallipadi Signed-off-by: Peter Zijlstra LKML-Reference: <1284167957-3675-1-git-send-email-venki@google.com> Signed-off-by: Ingo Molnar --- kernel/sched_fair.c | 9 ++++++++- 1 file changed, 8 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c index a171138a9402..aa16cf1eb8fe 100644 --- a/kernel/sched_fair.c +++ b/kernel/sched_fair.c @@ -3031,7 +3031,14 @@ redo: if (!ld_moved) { schedstat_inc(sd, lb_failed[idle]); - sd->nr_balance_failed++; + /* + * Increment the failure counter only on periodic balance. + * We do not want newidle balance, which can be very + * frequent, pollute the failure counter causing + * excessive cache_hot migrations and active balances. + */ + if (idle != CPU_NEWLY_IDLE) + sd->nr_balance_failed++; if (need_active_balance(sd, sd_idle, idle, cpu_of(busiest), this_cpu)) { -- cgit v1.2.3 From 43fa5460fe60dea5c610490a1d263415419c60f6 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 20 Sep 2010 22:40:03 -0400 Subject: sched: Try not to migrate higher priority RT tasks When first working on the RT scheduler design, we concentrated on keeping all CPUs running RT tasks instead of having multiple RT tasks on a single CPU waiting for the migration thread to move them. Instead we take a more proactive stance and push or pull RT tasks from one CPU to another on wakeup or scheduling. When an RT task wakes up on a CPU that is running another RT task, instead of preempting it and killing the cache of the running RT task, we look to see if we can migrate the RT task that is waking up, even if the RT task waking up is of higher priority. This may sound a bit odd, but RT tasks should be limited in migration by the user anyway. But in practice, people do not do this, which causes high prio RT tasks to bounce around the CPUs. This becomes even worse when we have priority inheritance, because a high prio task can block on a lower prio task and boost its priority. When the lower prio task wakes up the high prio task, if it happens to be on the same CPU it will migrate off of it. But in reality, the above does not happen much either, because the wake up of the lower prio task, which has already been boosted, if it was on the same CPU as the higher prio task, it would then migrate off of it. But anyway, we do not want to migrate them either. To examine the scheduling, I created a test program and examined it under kernelshark. The test program created CPU * 2 threads, where each thread had a different priority. The program takes different options. The options used in this change log was to have priority inheritance mutexes or not. All threads did the following loop: static void grab_lock(long id, int iter, int l) { ftrace_write("thread %ld iter %d, taking lock %d\n", id, iter, l); pthread_mutex_lock(&locks[l]); ftrace_write("thread %ld iter %d, took lock %d\n", id, iter, l); busy_loop(nr_tasks - id); ftrace_write("thread %ld iter %d, unlock lock %d\n", id, iter, l); pthread_mutex_unlock(&locks[l]); } void *start_task(void *id) { [...] while (!done) { for (l = 0; l < nr_locks; l++) { grab_lock(id, i, l); ftrace_write("thread %ld iter %d sleeping\n", id, i); ms_sleep(id); } i++; } [...] } The busy_loop(ms) keeps the CPU spinning for ms milliseconds. The ms_sleep(ms) sleeps for ms milliseconds. The ftrace_write() writes to the ftrace buffer to help analyze via ftrace. The higher the id, the higher the prio, the shorter it does the busy loop, but the longer it spins. This is usually the case with RT tasks, the lower priority tasks usually run longer than higher priority tasks. At the end of the test, it records the number of loops each thread took, as well as the number of voluntary preemptions, non-voluntary preemptions, and number of migrations each thread took, taking the information from /proc/$$/sched and /proc/$$/status. Running this on a 4 CPU processor, the results without changes to the kernel looked like this: Task vol nonvol migrated iterations ---- --- ------ -------- ---------- 0: 53 3220 1470 98 1: 562 773 724 98 2: 752 933 1375 98 3: 749 39 697 98 4: 758 5 515 98 5: 764 2 679 99 6: 761 2 535 99 7: 757 3 346 99 total: 5156 4977 6341 787 Each thread regardless of priority migrated a few hundred times. The higher priority tasks, were a little better but still took quite an impact. By letting higher priority tasks bump the lower prio task from the CPU, things changed a bit: Task vol nonvol migrated iterations ---- --- ------ -------- ---------- 0: 37 2835 1937 98 1: 666 1821 1865 98 2: 654 1003 1385 98 3: 664 635 973 99 4: 698 197 352 99 5: 703 101 159 99 6: 708 1 75 99 7: 713 1 2 99 total: 4843 6594 6748 789 The total # of migrations did not change (several runs showed the difference all within the noise). But we now see a dramatic improvement to the higher priority tasks. (kernelshark showed that the watchdog timer bumped the highest priority task to give it the 2 count. This was actually consistent with every run). Notice that the # of iterations did not change either. The above was with priority inheritance mutexes. That is, when the higher prority task blocked on a lower priority task, the lower priority task would inherit the higher priority task (which shows why task 6 was bumped so many times). When not using priority inheritance mutexes, the current kernel shows this: Task vol nonvol migrated iterations ---- --- ------ -------- ---------- 0: 56 3101 1892 95 1: 594 713 937 95 2: 625 188 618 95 3: 628 4 491 96 4: 640 7 468 96 5: 631 2 501 96 6: 641 1 466 96 7: 643 2 497 96 total: 4458 4018 5870 765 Not much changed with or without priority inheritance mutexes. But if we let the high priority task bump lower priority tasks on wakeup we see: Task vol nonvol migrated iterations ---- --- ------ -------- ---------- 0: 115 3439 2782 98 1: 633 1354 1583 99 2: 652 919 1218 99 3: 645 713 934 99 4: 690 3 3 99 5: 694 1 4 99 6: 720 3 4 99 7: 747 0 1 100 Which shows a even bigger change. The big difference between task 3 and task 4 is because we have only 4 CPUs on the machine, causing the 4 highest prio tasks to always have preference. Although I did not measure cache misses, and I'm sure there would be little to measure since the test was not data intensive, I could imagine large improvements for higher priority tasks when dealing with lower priority tasks. Thus, I'm satisfied with making the change and agreeing with what Gregory Haskins argued a few years ago when we first had this discussion. One final note. All tasks in the above tests were RT tasks. Any RT task will always preempt a non RT task that is running on the CPU the RT task wants to run on. Signed-off-by: Steven Rostedt Signed-off-by: Peter Zijlstra Cc: Gregory Haskins LKML-Reference: <20100921024138.605460343@goodmis.org> Signed-off-by: Ingo Molnar --- kernel/sched_rt.c | 22 ++++++++++++---------- 1 file changed, 12 insertions(+), 10 deletions(-) (limited to 'kernel') diff --git a/kernel/sched_rt.c b/kernel/sched_rt.c index d10c80ebb67a..6a02b38ab653 100644 --- a/kernel/sched_rt.c +++ b/kernel/sched_rt.c @@ -960,18 +960,18 @@ select_task_rq_rt(struct rq *rq, struct task_struct *p, int sd_flag, int flags) * runqueue. Otherwise simply start this RT task * on its current runqueue. * - * We want to avoid overloading runqueues. Even if - * the RT task is of higher priority than the current RT task. - * RT tasks behave differently than other tasks. If - * one gets preempted, we try to push it off to another queue. - * So trying to keep a preempting RT task on the same - * cache hot CPU will force the running RT task to - * a cold CPU. So we waste all the cache for the lower - * RT task in hopes of saving some of a RT task - * that is just being woken and probably will have - * cold cache anyway. + * We want to avoid overloading runqueues. If the woken + * task is a higher priority, then it will stay on this CPU + * and the lower prio task should be moved to another CPU. + * Even though this will probably make the lower prio task + * lose its cache, we do not want to bounce a higher task + * around just because it gave up its CPU, perhaps for a + * lock? + * + * For equal prio tasks, we just let the scheduler sort it out. */ if (unlikely(rt_task(rq->curr)) && + rq->curr->prio < p->prio && (p->rt.nr_cpus_allowed > 1)) { int cpu = find_lowest_rq(p); @@ -1491,6 +1491,8 @@ static void task_woken_rt(struct rq *rq, struct task_struct *p) if (!task_running(rq, p) && !test_tsk_need_resched(rq->curr) && has_pushable_tasks(rq) && + rt_task(rq->curr) && + rq->curr->prio < p->prio && p->rt.nr_cpus_allowed > 1) push_rt_tasks(rq); } -- cgit v1.2.3 From b3bc211cfe7d5fe94b310480d78e00bea96fbf2a Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 20 Sep 2010 22:40:04 -0400 Subject: sched: Give CPU bound RT tasks preference If a high priority task is waking up on a CPU that is running a lower priority task that is bound to a CPU, see if we can move the high RT task to another CPU first. Note, if all other CPUs are running higher priority tasks than the CPU bounded current task, then it will be preempted regardless. Signed-off-by: Steven Rostedt Signed-off-by: Peter Zijlstra Cc: Gregory Haskins LKML-Reference: <20100921024138.888922071@goodmis.org> Signed-off-by: Ingo Molnar --- kernel/sched_rt.c | 8 +++++--- 1 file changed, 5 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/sched_rt.c b/kernel/sched_rt.c index 6a02b38ab653..baef30f08405 100644 --- a/kernel/sched_rt.c +++ b/kernel/sched_rt.c @@ -971,7 +971,8 @@ select_task_rq_rt(struct rq *rq, struct task_struct *p, int sd_flag, int flags) * For equal prio tasks, we just let the scheduler sort it out. */ if (unlikely(rt_task(rq->curr)) && - rq->curr->prio < p->prio && + (rq->curr->rt.nr_cpus_allowed < 2 || + rq->curr->prio < p->prio) && (p->rt.nr_cpus_allowed > 1)) { int cpu = find_lowest_rq(p); @@ -1491,9 +1492,10 @@ static void task_woken_rt(struct rq *rq, struct task_struct *p) if (!task_running(rq, p) && !test_tsk_need_resched(rq->curr) && has_pushable_tasks(rq) && + p->rt.nr_cpus_allowed > 1 && rt_task(rq->curr) && - rq->curr->prio < p->prio && - p->rt.nr_cpus_allowed > 1) + (rq->curr->rt.nr_cpus_allowed < 2 || + rq->curr->prio < p->prio)) push_rt_tasks(rq); } -- cgit v1.2.3 From a8027073eb127cd207070891374b5c54c2ce3d23 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 20 Sep 2010 15:13:34 -0400 Subject: tracing/sched: Add sched_pi_setprio tracepoint Add a tracepoint that shows the priority of a task being boosted via priority inheritance. Cc: Gregory Haskins Acked-by: Peter Zijlstra Signed-off-by: Steven Rostedt --- kernel/sched.c | 1 + 1 file changed, 1 insertion(+) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index 9ca8ad05950b..4ad473814350 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -4355,6 +4355,7 @@ void rt_mutex_setprio(struct task_struct *p, int prio) rq = task_rq_lock(p, &flags); + trace_sched_pi_setprio(p, prio); oldprio = p->prio; prev_class = p->sched_class; on_rq = p->se.on_rq; -- cgit v1.2.3 From 676cb02dc32adef13d9efb5ea52079e4ede1e3ec Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Mon, 20 Jul 2009 23:33:49 +0200 Subject: softirqs: Make wakeup_softirqd static No users outside of kernel/softirq.c Signed-off-by: Thomas Gleixner --- kernel/softirq.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/softirq.c b/kernel/softirq.c index 07b4f1b1a73a..80f6e3bd1d2a 100644 --- a/kernel/softirq.c +++ b/kernel/softirq.c @@ -67,7 +67,7 @@ char *softirq_to_name[NR_SOFTIRQS] = { * to the pending events, so lets the scheduler to balance * the softirq load for us. */ -void wakeup_softirqd(void) +static void wakeup_softirqd(void) { /* Interrupts are disabled: no need to stop preemption */ struct task_struct *tsk = __get_cpu_var(ksoftirqd); -- cgit v1.2.3 From bf5438fca2950b03c21ad868090cc1a8fcd49536 Mon Sep 17 00:00:00 2001 From: Jason Baron Date: Fri, 17 Sep 2010 11:09:00 -0400 Subject: jump label: Base patch for jump label base patch to implement 'jump labeling'. Based on a new 'asm goto' inline assembly gcc mechanism, we can now branch to labels from an 'asm goto' statment. This allows us to create a 'no-op' fastpath, which can subsequently be patched with a jump to the slowpath code. This is useful for code which might be rarely used, but which we'd like to be able to call, if needed. Tracepoints are the current usecase that these are being implemented for. Acked-by: David S. Miller Signed-off-by: Jason Baron LKML-Reference: [ cleaned up some formating ] Signed-off-by: Steven Rostedt --- kernel/Makefile | 2 +- kernel/jump_label.c | 346 ++++++++++++++++++++++++++++++++++++++++++++++++++++ kernel/kprobes.c | 1 + kernel/module.c | 6 + 4 files changed, 354 insertions(+), 1 deletion(-) create mode 100644 kernel/jump_label.c (limited to 'kernel') diff --git a/kernel/Makefile b/kernel/Makefile index 0b72d1a74be0..d52b473c99a1 100644 --- a/kernel/Makefile +++ b/kernel/Makefile @@ -10,7 +10,7 @@ obj-y = sched.o fork.o exec_domain.o panic.o printk.o \ kthread.o wait.o kfifo.o sys_ni.o posix-cpu-timers.o mutex.o \ hrtimer.o rwsem.o nsproxy.o srcu.o semaphore.o \ notifier.o ksysfs.o pm_qos_params.o sched_clock.o cred.o \ - async.o range.o + async.o range.o jump_label.o obj-$(CONFIG_HAVE_EARLY_RES) += early_res.o obj-y += groups.o diff --git a/kernel/jump_label.c b/kernel/jump_label.c new file mode 100644 index 000000000000..460fd40112b3 --- /dev/null +++ b/kernel/jump_label.c @@ -0,0 +1,346 @@ +/* + * jump label support + * + * Copyright (C) 2009 Jason Baron + * + */ +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#ifdef HAVE_JUMP_LABEL + +#define JUMP_LABEL_HASH_BITS 6 +#define JUMP_LABEL_TABLE_SIZE (1 << JUMP_LABEL_HASH_BITS) +static struct hlist_head jump_label_table[JUMP_LABEL_TABLE_SIZE]; + +/* mutex to protect coming/going of the the jump_label table */ +static DEFINE_MUTEX(jump_label_mutex); + +struct jump_label_entry { + struct hlist_node hlist; + struct jump_entry *table; + int nr_entries; + /* hang modules off here */ + struct hlist_head modules; + unsigned long key; +}; + +struct jump_label_module_entry { + struct hlist_node hlist; + struct jump_entry *table; + int nr_entries; + struct module *mod; +}; + +static int jump_label_cmp(const void *a, const void *b) +{ + const struct jump_entry *jea = a; + const struct jump_entry *jeb = b; + + if (jea->key < jeb->key) + return -1; + + if (jea->key > jeb->key) + return 1; + + return 0; +} + +static void +sort_jump_label_entries(struct jump_entry *start, struct jump_entry *stop) +{ + unsigned long size; + + size = (((unsigned long)stop - (unsigned long)start) + / sizeof(struct jump_entry)); + sort(start, size, sizeof(struct jump_entry), jump_label_cmp, NULL); +} + +static struct jump_label_entry *get_jump_label_entry(jump_label_t key) +{ + struct hlist_head *head; + struct hlist_node *node; + struct jump_label_entry *e; + u32 hash = jhash((void *)&key, sizeof(jump_label_t), 0); + + head = &jump_label_table[hash & (JUMP_LABEL_TABLE_SIZE - 1)]; + hlist_for_each_entry(e, node, head, hlist) { + if (key == e->key) + return e; + } + return NULL; +} + +static struct jump_label_entry * +add_jump_label_entry(jump_label_t key, int nr_entries, struct jump_entry *table) +{ + struct hlist_head *head; + struct jump_label_entry *e; + u32 hash; + + e = get_jump_label_entry(key); + if (e) + return ERR_PTR(-EEXIST); + + e = kmalloc(sizeof(struct jump_label_entry), GFP_KERNEL); + if (!e) + return ERR_PTR(-ENOMEM); + + hash = jhash((void *)&key, sizeof(jump_label_t), 0); + head = &jump_label_table[hash & (JUMP_LABEL_TABLE_SIZE - 1)]; + e->key = key; + e->table = table; + e->nr_entries = nr_entries; + INIT_HLIST_HEAD(&(e->modules)); + hlist_add_head(&e->hlist, head); + return e; +} + +static int +build_jump_label_hashtable(struct jump_entry *start, struct jump_entry *stop) +{ + struct jump_entry *iter, *iter_begin; + struct jump_label_entry *entry; + int count; + + sort_jump_label_entries(start, stop); + iter = start; + while (iter < stop) { + entry = get_jump_label_entry(iter->key); + if (!entry) { + iter_begin = iter; + count = 0; + while ((iter < stop) && + (iter->key == iter_begin->key)) { + iter++; + count++; + } + entry = add_jump_label_entry(iter_begin->key, + count, iter_begin); + if (IS_ERR(entry)) + return PTR_ERR(entry); + } else { + WARN_ONCE(1, KERN_ERR "build_jump_hashtable: unexpected entry!\n"); + return -1; + } + } + return 0; +} + +/*** + * jump_label_update - update jump label text + * @key - key value associated with a a jump label + * @type - enum set to JUMP_LABEL_ENABLE or JUMP_LABEL_DISABLE + * + * Will enable/disable the jump for jump label @key, depending on the + * value of @type. + * + */ + +void jump_label_update(unsigned long key, enum jump_label_type type) +{ + struct jump_entry *iter; + struct jump_label_entry *entry; + struct hlist_node *module_node; + struct jump_label_module_entry *e_module; + int count; + + mutex_lock(&jump_label_mutex); + entry = get_jump_label_entry((jump_label_t)key); + if (entry) { + count = entry->nr_entries; + iter = entry->table; + while (count--) { + if (kernel_text_address(iter->code)) + arch_jump_label_transform(iter, type); + iter++; + } + /* eanble/disable jump labels in modules */ + hlist_for_each_entry(e_module, module_node, &(entry->modules), + hlist) { + count = e_module->nr_entries; + iter = e_module->table; + while (count--) { + if (kernel_text_address(iter->code)) + arch_jump_label_transform(iter, type); + iter++; + } + } + } + mutex_unlock(&jump_label_mutex); +} + +static __init int init_jump_label(void) +{ + int ret; + struct jump_entry *iter_start = __start___jump_table; + struct jump_entry *iter_stop = __stop___jump_table; + struct jump_entry *iter; + + mutex_lock(&jump_label_mutex); + ret = build_jump_label_hashtable(__start___jump_table, + __stop___jump_table); + iter = iter_start; + while (iter < iter_stop) { + arch_jump_label_text_poke_early(iter->code); + iter++; + } + mutex_unlock(&jump_label_mutex); + return ret; +} +early_initcall(init_jump_label); + +#ifdef CONFIG_MODULES + +static struct jump_label_module_entry * +add_jump_label_module_entry(struct jump_label_entry *entry, + struct jump_entry *iter_begin, + int count, struct module *mod) +{ + struct jump_label_module_entry *e; + + e = kmalloc(sizeof(struct jump_label_module_entry), GFP_KERNEL); + if (!e) + return ERR_PTR(-ENOMEM); + e->mod = mod; + e->nr_entries = count; + e->table = iter_begin; + hlist_add_head(&e->hlist, &entry->modules); + return e; +} + +static int add_jump_label_module(struct module *mod) +{ + struct jump_entry *iter, *iter_begin; + struct jump_label_entry *entry; + struct jump_label_module_entry *module_entry; + int count; + + /* if the module doesn't have jump label entries, just return */ + if (!mod->num_jump_entries) + return 0; + + sort_jump_label_entries(mod->jump_entries, + mod->jump_entries + mod->num_jump_entries); + iter = mod->jump_entries; + while (iter < mod->jump_entries + mod->num_jump_entries) { + entry = get_jump_label_entry(iter->key); + iter_begin = iter; + count = 0; + while ((iter < mod->jump_entries + mod->num_jump_entries) && + (iter->key == iter_begin->key)) { + iter++; + count++; + } + if (!entry) { + entry = add_jump_label_entry(iter_begin->key, 0, NULL); + if (IS_ERR(entry)) + return PTR_ERR(entry); + } + module_entry = add_jump_label_module_entry(entry, iter_begin, + count, mod); + if (IS_ERR(module_entry)) + return PTR_ERR(module_entry); + } + return 0; +} + +static void remove_jump_label_module(struct module *mod) +{ + struct hlist_head *head; + struct hlist_node *node, *node_next, *module_node, *module_node_next; + struct jump_label_entry *e; + struct jump_label_module_entry *e_module; + int i; + + /* if the module doesn't have jump label entries, just return */ + if (!mod->num_jump_entries) + return; + + for (i = 0; i < JUMP_LABEL_TABLE_SIZE; i++) { + head = &jump_label_table[i]; + hlist_for_each_entry_safe(e, node, node_next, head, hlist) { + hlist_for_each_entry_safe(e_module, module_node, + module_node_next, + &(e->modules), hlist) { + if (e_module->mod == mod) { + hlist_del(&e_module->hlist); + kfree(e_module); + } + } + if (hlist_empty(&e->modules) && (e->nr_entries == 0)) { + hlist_del(&e->hlist); + kfree(e); + } + } + } +} + +static int +jump_label_module_notify(struct notifier_block *self, unsigned long val, + void *data) +{ + struct module *mod = data; + int ret = 0; + + switch (val) { + case MODULE_STATE_COMING: + mutex_lock(&jump_label_mutex); + ret = add_jump_label_module(mod); + if (ret) + remove_jump_label_module(mod); + mutex_unlock(&jump_label_mutex); + break; + case MODULE_STATE_GOING: + mutex_lock(&jump_label_mutex); + remove_jump_label_module(mod); + mutex_unlock(&jump_label_mutex); + break; + } + return ret; +} + +/*** + * apply_jump_label_nops - patch module jump labels with arch_get_jump_label_nop() + * @mod: module to patch + * + * Allow for run-time selection of the optimal nops. Before the module + * loads patch these with arch_get_jump_label_nop(), which is specified by + * the arch specific jump label code. + */ +void jump_label_apply_nops(struct module *mod) +{ + struct jump_entry *iter; + + /* if the module doesn't have jump label entries, just return */ + if (!mod->num_jump_entries) + return; + + iter = mod->jump_entries; + while (iter < mod->jump_entries + mod->num_jump_entries) { + arch_jump_label_text_poke_early(iter->code); + iter++; + } +} + +struct notifier_block jump_label_module_nb = { + .notifier_call = jump_label_module_notify, + .priority = 0, +}; + +static __init int init_jump_label_module(void) +{ + return register_module_notifier(&jump_label_module_nb); +} +early_initcall(init_jump_label_module); + +#endif /* CONFIG_MODULES */ + +#endif diff --git a/kernel/kprobes.c b/kernel/kprobes.c index 6dd5359e1f0e..18904e42a918 100644 --- a/kernel/kprobes.c +++ b/kernel/kprobes.c @@ -47,6 +47,7 @@ #include #include #include +#include #include #include diff --git a/kernel/module.c b/kernel/module.c index d0b5f8db11b4..eba134157ef6 100644 --- a/kernel/module.c +++ b/kernel/module.c @@ -55,6 +55,7 @@ #include #include #include +#include #define CREATE_TRACE_POINTS #include @@ -2308,6 +2309,11 @@ static void find_module_sections(struct module *mod, struct load_info *info) sizeof(*mod->tracepoints), &mod->num_tracepoints); #endif +#ifdef HAVE_JUMP_LABEL + mod->jump_entries = section_objs(info, "__jump_table", + sizeof(*mod->jump_entries), + &mod->num_jump_entries); +#endif #ifdef CONFIG_EVENT_TRACING mod->trace_events = section_objs(info, "_ftrace_events", sizeof(*mod->trace_events), -- cgit v1.2.3 From e0cf0cd49632552f063fb3ae58691946da45fb2e Mon Sep 17 00:00:00 2001 From: Jason Baron Date: Fri, 17 Sep 2010 11:09:04 -0400 Subject: jump label: Initialize workqueue tracepoints *before* they are registered Initialize the workqueue data structures *before* they are registered so that they are ready for callbacks. Signed-off-by: Jason Baron LKML-Reference: Signed-off-by: Steven Rostedt --- kernel/trace/trace_workqueue.c | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace_workqueue.c b/kernel/trace/trace_workqueue.c index a7cc3793baf6..209b379a4721 100644 --- a/kernel/trace/trace_workqueue.c +++ b/kernel/trace/trace_workqueue.c @@ -263,6 +263,11 @@ int __init trace_workqueue_early_init(void) { int ret, cpu; + for_each_possible_cpu(cpu) { + spin_lock_init(&workqueue_cpu_stat(cpu)->lock); + INIT_LIST_HEAD(&workqueue_cpu_stat(cpu)->list); + } + ret = register_trace_workqueue_insertion(probe_workqueue_insertion, NULL); if (ret) goto out; @@ -279,11 +284,6 @@ int __init trace_workqueue_early_init(void) if (ret) goto no_creation; - for_each_possible_cpu(cpu) { - spin_lock_init(&workqueue_cpu_stat(cpu)->lock); - INIT_LIST_HEAD(&workqueue_cpu_stat(cpu)->list); - } - return 0; no_creation: -- cgit v1.2.3 From 4c3ef6d79328c0e23ade60cbfc8d496123a6855c Mon Sep 17 00:00:00 2001 From: Jason Baron Date: Fri, 17 Sep 2010 11:09:08 -0400 Subject: jump label: Add jump_label_text_reserved() to reserve jump points Add a jump_label_text_reserved(void *start, void *end), so that other pieces of code that want to modify kernel text, can first verify that jump label has not reserved the instruction. Acked-by: Masami Hiramatsu Signed-off-by: Jason Baron LKML-Reference: <06236663a3a7b1c1f13576bb9eccb6d9c17b7bfe.1284733808.git.jbaron@redhat.com> Signed-off-by: Steven Rostedt --- kernel/jump_label.c | 83 +++++++++++++++++++++++++++++++++++++++++++++++++++++ kernel/kprobes.c | 3 +- 2 files changed, 85 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/jump_label.c b/kernel/jump_label.c index 460fd40112b3..7be868bf25c6 100644 --- a/kernel/jump_label.c +++ b/kernel/jump_label.c @@ -177,6 +177,89 @@ void jump_label_update(unsigned long key, enum jump_label_type type) mutex_unlock(&jump_label_mutex); } +static int addr_conflict(struct jump_entry *entry, void *start, void *end) +{ + if (entry->code <= (unsigned long)end && + entry->code + JUMP_LABEL_NOP_SIZE > (unsigned long)start) + return 1; + + return 0; +} + +#ifdef CONFIG_MODULES + +static int module_conflict(void *start, void *end) +{ + struct hlist_head *head; + struct hlist_node *node, *node_next, *module_node, *module_node_next; + struct jump_label_entry *e; + struct jump_label_module_entry *e_module; + struct jump_entry *iter; + int i, count; + int conflict = 0; + + for (i = 0; i < JUMP_LABEL_TABLE_SIZE; i++) { + head = &jump_label_table[i]; + hlist_for_each_entry_safe(e, node, node_next, head, hlist) { + hlist_for_each_entry_safe(e_module, module_node, + module_node_next, + &(e->modules), hlist) { + count = e_module->nr_entries; + iter = e_module->table; + while (count--) { + if (addr_conflict(iter, start, end)) { + conflict = 1; + goto out; + } + iter++; + } + } + } + } +out: + return conflict; +} + +#endif + +/*** + * jump_label_text_reserved - check if addr range is reserved + * @start: start text addr + * @end: end text addr + * + * checks if the text addr located between @start and @end + * overlaps with any of the jump label patch addresses. Code + * that wants to modify kernel text should first verify that + * it does not overlap with any of the jump label addresses. + * + * returns 1 if there is an overlap, 0 otherwise + */ +int jump_label_text_reserved(void *start, void *end) +{ + struct jump_entry *iter; + struct jump_entry *iter_start = __start___jump_table; + struct jump_entry *iter_stop = __start___jump_table; + int conflict = 0; + + mutex_lock(&jump_label_mutex); + iter = iter_start; + while (iter < iter_stop) { + if (addr_conflict(iter, start, end)) { + conflict = 1; + goto out; + } + iter++; + } + + /* now check modules */ +#ifdef CONFIG_MODULES + conflict = module_conflict(start, end); +#endif +out: + mutex_unlock(&jump_label_mutex); + return conflict; +} + static __init int init_jump_label(void) { int ret; diff --git a/kernel/kprobes.c b/kernel/kprobes.c index 18904e42a918..ec4210c6501e 100644 --- a/kernel/kprobes.c +++ b/kernel/kprobes.c @@ -1147,7 +1147,8 @@ int __kprobes register_kprobe(struct kprobe *p) preempt_disable(); if (!kernel_text_address((unsigned long) p->addr) || in_kprobes_functions((unsigned long) p->addr) || - ftrace_text_reserved(p->addr, p->addr)) { + ftrace_text_reserved(p->addr, p->addr) || + jump_label_text_reserved(p->addr, p->addr)) { preempt_enable(); return -EINVAL; } -- cgit v1.2.3 From 8f7b50c514206211cc282a4247f7b12f18dee674 Mon Sep 17 00:00:00 2001 From: Jason Baron Date: Fri, 17 Sep 2010 11:09:13 -0400 Subject: jump label: Tracepoint support for jump labels Make use of the jump label infrastructure for tracepoints. Signed-off-by: Jason Baron LKML-Reference: Signed-off-by: Steven Rostedt --- kernel/tracepoint.c | 14 ++++++++++++-- 1 file changed, 12 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/tracepoint.c b/kernel/tracepoint.c index c77f3eceea25..d6073a50a6ca 100644 --- a/kernel/tracepoint.c +++ b/kernel/tracepoint.c @@ -25,6 +25,7 @@ #include #include #include +#include extern struct tracepoint __start___tracepoints[]; extern struct tracepoint __stop___tracepoints[]; @@ -263,7 +264,13 @@ static void set_tracepoint(struct tracepoint_entry **entry, * is used. */ rcu_assign_pointer(elem->funcs, (*entry)->funcs); - elem->state = active; + if (!elem->state && active) { + enable_jump_label(&elem->state); + elem->state = active; + } else if (elem->state && !active) { + disable_jump_label(&elem->state); + elem->state = active; + } } /* @@ -277,7 +284,10 @@ static void disable_tracepoint(struct tracepoint *elem) if (elem->unregfunc && elem->state) elem->unregfunc(); - elem->state = 0; + if (elem->state) { + disable_jump_label(&elem->state); + elem->state = 0; + } rcu_assign_pointer(elem->funcs, NULL); } -- cgit v1.2.3 From a247c3a97a0216b18a46243eda26081f1928ec37 Mon Sep 17 00:00:00 2001 From: Andrea Arcangeli Date: Wed, 22 Sep 2010 13:05:12 -0700 Subject: rmap: fix walk during fork The below bug in fork led to the rmap walk finding the parent huge-pmd twice instead of just once, because the anon_vma_chain objects of the child vma still point to the vma->vm_mm of the parent. The patch fixes it by making the rmap walk accurate during fork. It's not a big deal normally but it worth being accurate considering the cost is the same. Signed-off-by: Andrea Arcangeli Acked-by: Johannes Weiner Acked-by: Rik van Riel Acked-by: Hugh Dickins Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/fork.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/fork.c b/kernel/fork.c index b7e9d60a675d..c445f8cc408d 100644 --- a/kernel/fork.c +++ b/kernel/fork.c @@ -356,10 +356,10 @@ static int dup_mmap(struct mm_struct *mm, struct mm_struct *oldmm) if (IS_ERR(pol)) goto fail_nomem_policy; vma_set_policy(tmp, pol); + tmp->vm_mm = mm; if (anon_vma_fork(tmp, mpnt)) goto fail_nomem_anon_vma_fork; tmp->vm_flags &= ~VM_LOCKED; - tmp->vm_mm = mm; tmp->vm_next = tmp->vm_prev = NULL; file = tmp->vm_file; if (file) { -- cgit v1.2.3 From 1ef21199a5d98702755cfa0df39b8b5112c423d6 Mon Sep 17 00:00:00 2001 From: Stephen Rothwell Date: Fri, 3 Sep 2010 13:19:04 +1000 Subject: powerpc: define a compat_sys_recv cond_syscall Signed-off-by: Stephen Rothwell Signed-off-by: Benjamin Herrenschmidt --- kernel/sys_ni.c | 1 + 1 file changed, 1 insertion(+) (limited to 'kernel') diff --git a/kernel/sys_ni.c b/kernel/sys_ni.c index bad369ec5403..c782fe9924c7 100644 --- a/kernel/sys_ni.c +++ b/kernel/sys_ni.c @@ -50,6 +50,7 @@ cond_syscall(compat_sys_sendmsg); cond_syscall(sys_recvmsg); cond_syscall(sys_recvmmsg); cond_syscall(compat_sys_recvmsg); +cond_syscall(compat_sys_recv); cond_syscall(compat_sys_recvfrom); cond_syscall(compat_sys_recvmmsg); cond_syscall(sys_socketcall); -- cgit v1.2.3 From 99a51792dbb7e6e39b2a4ebcfe202f1dcc7354c4 Mon Sep 17 00:00:00 2001 From: Joe Perches Date: Sat, 4 Sep 2010 18:52:50 -0700 Subject: kernel/pm_qos_params.c: Remove unnecessary casts of private_data Signed-off-by: Joe Perches Signed-off-by: Jiri Kosina --- kernel/pm_qos_params.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/pm_qos_params.c b/kernel/pm_qos_params.c index 996a4dec5f96..49829c5b4b47 100644 --- a/kernel/pm_qos_params.c +++ b/kernel/pm_qos_params.c @@ -394,7 +394,7 @@ static ssize_t pm_qos_power_write(struct file *filp, const char __user *buf, } else return -EINVAL; - pm_qos_req = (struct pm_qos_request_list *)filp->private_data; + pm_qos_req = filp->private_data; pm_qos_update_request(pm_qos_req, value); return count; -- cgit v1.2.3 From 829f8ed2c963df7c23d1c644db6c4387eb1601fa Mon Sep 17 00:00:00 2001 From: Christian Dietrich Date: Mon, 6 Sep 2010 16:37:12 +0200 Subject: kernel: Remove undead ifdef CONFIG_DEBUG_LOCK_ALLOC The CONFIG_DEBUG_LOCK_ALLOC ifdef isn't necessary at this point, because it is checked in an outer ifdef level already and has no effect here. Signed-off-by: Christian Dietrich Signed-off-by: Paul E. McKenney --- kernel/srcu.c | 2 -- 1 file changed, 2 deletions(-) (limited to 'kernel') diff --git a/kernel/srcu.c b/kernel/srcu.c index 2980da3fd509..c71e07500536 100644 --- a/kernel/srcu.c +++ b/kernel/srcu.c @@ -46,11 +46,9 @@ static int init_srcu_struct_fields(struct srcu_struct *sp) int __init_srcu_struct(struct srcu_struct *sp, const char *name, struct lock_class_key *key) { -#ifdef CONFIG_DEBUG_LOCK_ALLOC /* Don't re-initialize a lock while it is held. */ debug_check_no_locks_freed((void *)sp, sizeof(*sp)); lockdep_init_map(&sp->dep_map, name, key, 0); -#endif /* #ifdef CONFIG_DEBUG_LOCK_ALLOC */ return init_srcu_struct_fields(sp); } EXPORT_SYMBOL_GPL(__init_srcu_struct); -- cgit v1.2.3 From 0ddea0ead2ced99eaaaddff2beb755381e5c89f8 Mon Sep 17 00:00:00 2001 From: "Paul E. McKenney" Date: Sun, 19 Sep 2010 21:06:14 -0700 Subject: rcu: fix sparse errors in rcutorture.c Add the sparse __rcu address-space identifier and make a couple of variables static. Signed-off-by: Paul E. McKenney --- kernel/rcutorture.c | 11 +++++++---- 1 file changed, 7 insertions(+), 4 deletions(-) (limited to 'kernel') diff --git a/kernel/rcutorture.c b/kernel/rcutorture.c index 729710273dcb..9d8e8fb2515f 100644 --- a/kernel/rcutorture.c +++ b/kernel/rcutorture.c @@ -120,7 +120,7 @@ struct rcu_torture { }; static LIST_HEAD(rcu_torture_freelist); -static struct rcu_torture *rcu_torture_current; +static struct rcu_torture __rcu *rcu_torture_current; static long rcu_torture_current_version; static struct rcu_torture rcu_tortures[10 * RCU_TORTURE_PIPE_LEN]; static DEFINE_SPINLOCK(rcu_torture_lock); @@ -153,8 +153,10 @@ int rcutorture_runnable = RCUTORTURE_RUNNABLE_INIT; #define FULLSTOP_SHUTDOWN 1 /* System shutdown with rcutorture running. */ #define FULLSTOP_RMMOD 2 /* Normal rmmod of rcutorture. */ static int fullstop = FULLSTOP_RMMOD; -DEFINE_MUTEX(fullstop_mutex); /* Protect fullstop transitions and spawning */ - /* of kthreads. */ +/* + * Protect fullstop transitions and spawning of kthreads. + */ +static DEFINE_MUTEX(fullstop_mutex); /* * Detect and respond to a system shutdown. @@ -737,7 +739,8 @@ rcu_torture_writer(void *arg) continue; rp->rtort_pipe_count = 0; udelay(rcu_random(&rand) & 0x3ff); - old_rp = rcu_torture_current; + old_rp = rcu_dereference_check(rcu_torture_current, + current == writer_task); rp->rtort_mbtest = 1; rcu_assign_pointer(rcu_torture_current, rp); smp_wmb(); /* Mods to old_rp must follow rcu_assign_pointer() */ -- cgit v1.2.3 From 269dcc1c2ec25864308ee03a3fa26ea819d9f5d0 Mon Sep 17 00:00:00 2001 From: "Paul E. McKenney" Date: Tue, 7 Sep 2010 14:23:09 -0700 Subject: rcu: Add tracing data to support queueing models The current tracing data is not sufficient to deduce the average time that a callback spends waiting for a grace period to end. Add three per-CPU counters recording the number of callbacks invoked (ci), the number of callbacks orphaned (co), and the number of callbacks adopted (ca). Given the existing callback queue length (ql), the average wait time in absence of CPU hotplug operations is ql/ci. The units of wait time will be in terms of the duration over which ci was measured. In the presence of CPU hotplug operations, there is room for argument, but ql/(ci-co+ca) won't steer you too far wrong. Also fixes a typo called out by Lucas De Marchi . Signed-off-by: Paul E. McKenney --- kernel/rcutree.c | 3 +++ kernel/rcutree.h | 3 +++ kernel/rcutree_trace.c | 10 +++++++--- 3 files changed, 13 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/rcutree.c b/kernel/rcutree.c index 42140a860bb9..e75073504a31 100644 --- a/kernel/rcutree.c +++ b/kernel/rcutree.c @@ -1004,6 +1004,7 @@ static void rcu_send_cbs_to_orphanage(struct rcu_state *rsp) for (i = 0; i < RCU_NEXT_SIZE; i++) rdp->nxttail[i] = &rdp->nxtlist; rsp->orphan_qlen += rdp->qlen; + rdp->n_cbs_orphaned += rdp->qlen; rdp->qlen = 0; raw_spin_unlock(&rsp->onofflock); /* irqs remain disabled. */ } @@ -1025,6 +1026,7 @@ static void rcu_adopt_orphan_cbs(struct rcu_state *rsp) *rdp->nxttail[RCU_NEXT_TAIL] = rsp->orphan_cbs_list; rdp->nxttail[RCU_NEXT_TAIL] = rsp->orphan_cbs_tail; rdp->qlen += rsp->orphan_qlen; + rdp->n_cbs_adopted += rsp->orphan_qlen; rsp->orphan_cbs_list = NULL; rsp->orphan_cbs_tail = &rsp->orphan_cbs_list; rsp->orphan_qlen = 0; @@ -1156,6 +1158,7 @@ static void rcu_do_batch(struct rcu_state *rsp, struct rcu_data *rdp) /* Update count, and requeue any remaining callbacks. */ rdp->qlen -= count; + rdp->n_cbs_invoked += count; if (list != NULL) { *tail = rdp->nxtlist; rdp->nxtlist = list; diff --git a/kernel/rcutree.h b/kernel/rcutree.h index 7918ba61873f..91d4170c5c13 100644 --- a/kernel/rcutree.h +++ b/kernel/rcutree.h @@ -202,6 +202,9 @@ struct rcu_data { long qlen; /* # of queued callbacks */ long qlen_last_fqs_check; /* qlen at last check for QS forcing */ + unsigned long n_cbs_invoked; /* count of RCU cbs invoked. */ + unsigned long n_cbs_orphaned; /* RCU cbs sent to orphanage. */ + unsigned long n_cbs_adopted; /* RCU cbs adopted from orphanage. */ unsigned long n_force_qs_snap; /* did other CPU force QS recently? */ long blimit; /* Upper limit on a processed batch */ diff --git a/kernel/rcutree_trace.c b/kernel/rcutree_trace.c index 458e032a3a30..d15430b9d122 100644 --- a/kernel/rcutree_trace.c +++ b/kernel/rcutree_trace.c @@ -64,7 +64,9 @@ static void print_one_rcu_data(struct seq_file *m, struct rcu_data *rdp) rdp->dynticks_fqs); #endif /* #ifdef CONFIG_NO_HZ */ seq_printf(m, " of=%lu ri=%lu", rdp->offline_fqs, rdp->resched_ipi); - seq_printf(m, " ql=%ld b=%ld\n", rdp->qlen, rdp->blimit); + seq_printf(m, " ql=%ld b=%ld", rdp->qlen, rdp->blimit); + seq_printf(m, " ci=%lu co=%lu ca=%lu\n", + rdp->n_cbs_invoked, rdp->n_cbs_orphaned, rdp->n_cbs_adopted); } #define PRINT_RCU_DATA(name, func, m) \ @@ -119,7 +121,9 @@ static void print_one_rcu_data_csv(struct seq_file *m, struct rcu_data *rdp) rdp->dynticks_fqs); #endif /* #ifdef CONFIG_NO_HZ */ seq_printf(m, ",%lu,%lu", rdp->offline_fqs, rdp->resched_ipi); - seq_printf(m, ",%ld,%ld\n", rdp->qlen, rdp->blimit); + seq_printf(m, ",%ld,%ld", rdp->qlen, rdp->blimit); + seq_printf(m, ",%lu,%lu,%lu\n", + rdp->n_cbs_invoked, rdp->n_cbs_orphaned, rdp->n_cbs_adopted); } static int show_rcudata_csv(struct seq_file *m, void *unused) @@ -128,7 +132,7 @@ static int show_rcudata_csv(struct seq_file *m, void *unused) #ifdef CONFIG_NO_HZ seq_puts(m, "\"dt\",\"dt nesting\",\"dn\",\"df\","); #endif /* #ifdef CONFIG_NO_HZ */ - seq_puts(m, "\"of\",\"ri\",\"ql\",\"b\"\n"); + seq_puts(m, "\"of\",\"ri\",\"ql\",\"b\",\"ci\",\"co\",\"ca\"\n"); #ifdef CONFIG_TREE_PREEMPT_RCU seq_puts(m, "\"rcu_preempt:\"\n"); PRINT_RCU_DATA(rcu_preempt_data, print_one_rcu_data_csv, m); -- cgit v1.2.3 From d1ea13c6e2cce0106531852daaa93dd97aec9580 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Thu, 23 Sep 2010 18:40:07 +0200 Subject: genirq: Cleanup irq_chip->typename leftovers 3 years transition phase is enough. Cleanup the last users and remove the cruft. Signed-off-by: Thomas Gleixner Cc: Leo Chen Cc: Hirokazu Takata Cc: Chris Metcalf Cc: Jeff Dike Cc: Chris Zankel --- kernel/irq/chip.c | 2 -- 1 file changed, 2 deletions(-) (limited to 'kernel') diff --git a/kernel/irq/chip.c b/kernel/irq/chip.c index b7091d5ca2f8..4ea775cc60f0 100644 --- a/kernel/irq/chip.c +++ b/kernel/irq/chip.c @@ -344,8 +344,6 @@ void irq_chip_set_defaults(struct irq_chip *chip) if (!chip->shutdown) chip->shutdown = chip->disable != default_disable ? chip->disable : default_shutdown; - if (!chip->name) - chip->name = chip->typename; if (!chip->end) chip->end = dummy_irq_chip.end; } -- cgit v1.2.3 From 399f1e30ac17b77d383444aff480c7390f5adf2a Mon Sep 17 00:00:00 2001 From: "Ira W. Snyder" Date: Thu, 30 Sep 2010 15:15:27 -0700 Subject: kfifo: fix scatterlist usage The kfifo_dma family of functions use sg_mark_end() on the last element in their scatterlist. This forces use of a fresh scatterlist for each DMA operation, which makes recycling a single scatterlist impossible. Change the behavior of the kfifo_dma functions to match the usage of the dma_map_sg function. This means that users must respect the returned nents value. The sample code is updated to reflect the change. This bug is trivial to cause: call kfifo_dma_in_prepare() such that it prepares a scatterlist with a single entry comprising the whole fifo. This is the case when you map the entirety of a newly created empty fifo. This causes the setup_sgl() function to mark the first scatterlist entry as the end of the chain, no matter what comes after it. Afterwards, add and remove some data from the fifo such that another call to kfifo_dma_in_prepare() will create two scatterlist entries. It returns nents=2. However, due to the previous sg_mark_end() call, sg_is_last() will now return true for the first scatterlist element. This causes the sample code to print a single scatterlist element when it should print two. By removing the call to sg_mark_end(), we make the API as similar as possible to the DMA mapping API. All users are required to respect the returned nents. Signed-off-by: Ira W. Snyder Cc: Stefani Seibold Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/kfifo.c | 2 -- 1 file changed, 2 deletions(-) (limited to 'kernel') diff --git a/kernel/kfifo.c b/kernel/kfifo.c index 6b5580c57644..01a0700e873f 100644 --- a/kernel/kfifo.c +++ b/kernel/kfifo.c @@ -365,8 +365,6 @@ static unsigned int setup_sgl(struct __kfifo *fifo, struct scatterlist *sgl, n = setup_sgl_buf(sgl, fifo->data + off, nents, l); n += setup_sgl_buf(sgl + n, fifo->data, nents - n, len - l); - if (n) - sg_mark_end(sgl + n - 1); return n; } -- cgit v1.2.3 From d9817ebeeef16e01487549312c68540ca8f1561b Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Mon, 27 Sep 2010 12:45:59 +0000 Subject: genirq: Provide Kconfig The generic irq Kconfig options are copied around all archs. Provide a generic Kconfig file which can be included. Signed-off-by: Thomas Gleixner Cc: Peter Zijlstra LKML-Reference: <20100927121843.217333624@linutronix.de> Reviewed-by: H. Peter Anvin Reviewed-by: Ingo Molnar --- kernel/irq/Kconfig | 54 ++++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 54 insertions(+) create mode 100644 kernel/irq/Kconfig (limited to 'kernel') diff --git a/kernel/irq/Kconfig b/kernel/irq/Kconfig new file mode 100644 index 000000000000..e0fc6cd78aa0 --- /dev/null +++ b/kernel/irq/Kconfig @@ -0,0 +1,54 @@ +config HAVE_GENERIC_HARDIRQS + def_bool n + +if HAVE_GENERIC_HARDIRQS +menu "IRQ subsystem" +# +# Interrupt subsystem related configuration options +# +config GENERIC_HARDIRQS + def_bool y + +config GENERIC_HARDIRQS_NO__DO_IRQ + def_bool y + +# Options selectable by the architecture code +config HAVE_SPARSE_IRQ + def_bool n + +config GENERIC_IRQ_PROBE + def_bool n + +config GENERIC_PENDING_IRQ + def_bool n + +if SPARSE_IRQ && NUMA +config NUMA_IRQ_DESC + def_bool n +endif + +config AUTO_IRQ_AFFINITY + def_bool n + +config IRQ_PER_CPU + def_bool n + +config HARDIRQS_SW_RESEND + def_bool n + +config SPARSE_IRQ + bool "Support sparse irq numbering" + depends on HAVE_SPARSE_IRQ + ---help--- + + Sparse irq numbering is useful for distro kernels that want + to define a high CONFIG_NR_CPUS value but still want to have + low kernel memory footprint on smaller machines. + + ( Sparse irqs can also be beneficial on NUMA boxes, as they spread + out the interrupt descriptors in a more NUMA-friendly way. ) + + If you don't know what to do here, say N. + +endmenu +endif -- cgit v1.2.3 From ff7dcd44dd446db2c3e13bdedf2d52b8e0127f16 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Mon, 27 Sep 2010 12:44:25 +0000 Subject: genirq: Create irq_data Low level chip functions need access to irq_desc->handler_data, irq_desc->chip_data and irq_desc->msi_desc. We hand down the irq number to the low level functions, so they need to lookup irq_desc. With sparse irq this means a radix tree lookup. We could hand down irq_desc itself, but low level chip functions have no need to fiddle with it directly and we want to restrict access to irq_desc further. Preparatory patch for new chip functions. Note, that the ugly anon union/struct is there to avoid a full tree wide clean up for now. This is not going to last 3 years like __do_IRQ() Signed-off-by: Thomas Gleixner Cc: Peter Zijlstra LKML-Reference: <20100927121841.645542300@linutronix.de> Reviewed-by: H. Peter Anvin Reviewed-by: Ingo Molnar --- kernel/irq/handle.c | 39 +++++++++++++++++++-------------------- 1 file changed, 19 insertions(+), 20 deletions(-) (limited to 'kernel') diff --git a/kernel/irq/handle.c b/kernel/irq/handle.c index 27e5c6911223..099d4fc368c3 100644 --- a/kernel/irq/handle.c +++ b/kernel/irq/handle.c @@ -75,12 +75,10 @@ EXPORT_SYMBOL_GPL(nr_irqs); #ifdef CONFIG_SPARSE_IRQ static struct irq_desc irq_desc_init = { - .irq = -1, - .status = IRQ_DISABLED, - .chip = &no_irq_chip, - .handle_irq = handle_bad_irq, - .depth = 1, - .lock = __RAW_SPIN_LOCK_UNLOCKED(irq_desc_init.lock), + .status = IRQ_DISABLED, + .handle_irq = handle_bad_irq, + .depth = 1, + .lock = __RAW_SPIN_LOCK_UNLOCKED(irq_desc_init.lock), }; void __ref init_kstat_irqs(struct irq_desc *desc, int node, int nr) @@ -105,7 +103,7 @@ static void init_one_irq_desc(int irq, struct irq_desc *desc, int node) memcpy(desc, &irq_desc_init, sizeof(struct irq_desc)); raw_spin_lock_init(&desc->lock); - desc->irq = irq; + desc->irq_data.irq = irq; #ifdef CONFIG_SMP desc->node = node; #endif @@ -151,12 +149,10 @@ void replace_irq_desc(unsigned int irq, struct irq_desc *desc) static struct irq_desc irq_desc_legacy[NR_IRQS_LEGACY] __cacheline_aligned_in_smp = { [0 ... NR_IRQS_LEGACY-1] = { - .irq = -1, - .status = IRQ_DISABLED, - .chip = &no_irq_chip, - .handle_irq = handle_bad_irq, - .depth = 1, - .lock = __RAW_SPIN_LOCK_UNLOCKED(irq_desc_init.lock), + .status = IRQ_DISABLED, + .handle_irq = handle_bad_irq, + .depth = 1, + .lock = __RAW_SPIN_LOCK_UNLOCKED(irq_desc_init.lock), } }; @@ -183,8 +179,11 @@ int __init early_irq_init(void) kstat_irqs_legacy = kzalloc_node(NR_IRQS_LEGACY * nr_cpu_ids * sizeof(int), GFP_NOWAIT, node); + irq_desc_init.irq_data.chip = &no_irq_chip; + for (i = 0; i < legacy_count; i++) { - desc[i].irq = i; + desc[i].irq_data.irq = i; + desc[i].irq_data.chip = &no_irq_chip; #ifdef CONFIG_SMP desc[i].node = node; #endif @@ -241,11 +240,10 @@ out_unlock: struct irq_desc irq_desc[NR_IRQS] __cacheline_aligned_in_smp = { [0 ... NR_IRQS-1] = { - .status = IRQ_DISABLED, - .chip = &no_irq_chip, - .handle_irq = handle_bad_irq, - .depth = 1, - .lock = __RAW_SPIN_LOCK_UNLOCKED(irq_desc->lock), + .status = IRQ_DISABLED, + .handle_irq = handle_bad_irq, + .depth = 1, + .lock = __RAW_SPIN_LOCK_UNLOCKED(irq_desc->lock), } }; @@ -264,7 +262,8 @@ int __init early_irq_init(void) count = ARRAY_SIZE(irq_desc); for (i = 0; i < count; i++) { - desc[i].irq = i; + desc[i].irq_data.irq = i; + desc[i].irq_data.chip = &no_irq_chip; alloc_desc_masks(&desc[i], 0, true); init_desc_masks(&desc[i]); desc[i].kstat_irqs = kstat_irqs_all[i]; -- cgit v1.2.3 From 6b8ff3120c758340505dddf08ad685ebb841d5d5 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Fri, 1 Oct 2010 12:58:38 +0200 Subject: genirq: Convert core code to irq_data Convert all references in the core code to orq, chip, handler_data, chip_data, msi_desc, affinity to irq_data.* Signed-off-by: Thomas Gleixner Reviewed-by: Ingo Molnar --- kernel/irq/autoprobe.c | 14 ++++----- kernel/irq/chip.c | 78 +++++++++++++++++++++++------------------------ kernel/irq/handle.c | 16 +++++----- kernel/irq/internals.h | 12 ++++---- kernel/irq/manage.c | 54 ++++++++++++++++---------------- kernel/irq/migration.c | 10 +++--- kernel/irq/numa_migrate.c | 8 ++--- kernel/irq/proc.c | 8 ++--- kernel/irq/resend.c | 5 +-- kernel/irq/spurious.c | 6 ++-- 10 files changed, 106 insertions(+), 105 deletions(-) (limited to 'kernel') diff --git a/kernel/irq/autoprobe.c b/kernel/irq/autoprobe.c index 2295a31ef110..f9bf9b228033 100644 --- a/kernel/irq/autoprobe.c +++ b/kernel/irq/autoprobe.c @@ -57,9 +57,9 @@ unsigned long probe_irq_on(void) * Some chips need to know about probing in * progress: */ - if (desc->chip->set_type) - desc->chip->set_type(i, IRQ_TYPE_PROBE); - desc->chip->startup(i); + if (desc->irq_data.chip->set_type) + desc->irq_data.chip->set_type(i, IRQ_TYPE_PROBE); + desc->irq_data.chip->startup(i); } raw_spin_unlock_irq(&desc->lock); } @@ -76,7 +76,7 @@ unsigned long probe_irq_on(void) raw_spin_lock_irq(&desc->lock); if (!desc->action && !(desc->status & IRQ_NOPROBE)) { desc->status |= IRQ_AUTODETECT | IRQ_WAITING; - if (desc->chip->startup(i)) + if (desc->irq_data.chip->startup(i)) desc->status |= IRQ_PENDING; } raw_spin_unlock_irq(&desc->lock); @@ -98,7 +98,7 @@ unsigned long probe_irq_on(void) /* It triggered already - consider it spurious. */ if (!(status & IRQ_WAITING)) { desc->status = status & ~IRQ_AUTODETECT; - desc->chip->shutdown(i); + desc->irq_data.chip->shutdown(i); } else if (i < 32) mask |= 1 << i; @@ -137,7 +137,7 @@ unsigned int probe_irq_mask(unsigned long val) mask |= 1 << i; desc->status = status & ~IRQ_AUTODETECT; - desc->chip->shutdown(i); + desc->irq_data.chip->shutdown(i); } raw_spin_unlock_irq(&desc->lock); } @@ -181,7 +181,7 @@ int probe_irq_off(unsigned long val) nr_of_irqs++; } desc->status = status & ~IRQ_AUTODETECT; - desc->chip->shutdown(i); + desc->irq_data.chip->shutdown(i); } raw_spin_unlock_irq(&desc->lock); } diff --git a/kernel/irq/chip.c b/kernel/irq/chip.c index 4ea775cc60f0..e0e93ff10afd 100644 --- a/kernel/irq/chip.c +++ b/kernel/irq/chip.c @@ -32,18 +32,18 @@ static void dynamic_irq_init_x(unsigned int irq, bool keep_chip_data) /* Ensure we don't have left over values from a previous use of this irq */ raw_spin_lock_irqsave(&desc->lock, flags); desc->status = IRQ_DISABLED; - desc->chip = &no_irq_chip; + desc->irq_data.chip = &no_irq_chip; desc->handle_irq = handle_bad_irq; desc->depth = 1; - desc->msi_desc = NULL; - desc->handler_data = NULL; + desc->irq_data.msi_desc = NULL; + desc->irq_data.handler_data = NULL; if (!keep_chip_data) - desc->chip_data = NULL; + desc->irq_data.chip_data = NULL; desc->action = NULL; desc->irq_count = 0; desc->irqs_unhandled = 0; #ifdef CONFIG_SMP - cpumask_setall(desc->affinity); + cpumask_setall(desc->irq_data.affinity); #ifdef CONFIG_GENERIC_PENDING_IRQ cpumask_clear(desc->pending_mask); #endif @@ -64,7 +64,7 @@ void dynamic_irq_init(unsigned int irq) * dynamic_irq_init_keep_chip_data - initialize a dynamically allocated irq * @irq: irq number to initialize * - * does not set irq_to_desc(irq)->chip_data to NULL + * does not set irq_to_desc(irq)->irq_data.chip_data to NULL */ void dynamic_irq_init_keep_chip_data(unsigned int irq) { @@ -88,12 +88,12 @@ static void dynamic_irq_cleanup_x(unsigned int irq, bool keep_chip_data) irq); return; } - desc->msi_desc = NULL; - desc->handler_data = NULL; + desc->irq_data.msi_desc = NULL; + desc->irq_data.handler_data = NULL; if (!keep_chip_data) - desc->chip_data = NULL; + desc->irq_data.chip_data = NULL; desc->handle_irq = handle_bad_irq; - desc->chip = &no_irq_chip; + desc->irq_data.chip = &no_irq_chip; desc->name = NULL; clear_kstat_irqs(desc); raw_spin_unlock_irqrestore(&desc->lock, flags); @@ -112,7 +112,7 @@ void dynamic_irq_cleanup(unsigned int irq) * dynamic_irq_cleanup_keep_chip_data - cleanup a dynamically allocated irq * @irq: irq number to initialize * - * does not set irq_to_desc(irq)->chip_data to NULL + * does not set irq_to_desc(irq)->irq_data.chip_data to NULL */ void dynamic_irq_cleanup_keep_chip_data(unsigned int irq) { @@ -140,7 +140,7 @@ int set_irq_chip(unsigned int irq, struct irq_chip *chip) raw_spin_lock_irqsave(&desc->lock, flags); irq_chip_set_defaults(chip); - desc->chip = chip; + desc->irq_data.chip = chip; raw_spin_unlock_irqrestore(&desc->lock, flags); return 0; @@ -193,7 +193,7 @@ int set_irq_data(unsigned int irq, void *data) } raw_spin_lock_irqsave(&desc->lock, flags); - desc->handler_data = data; + desc->irq_data.handler_data = data; raw_spin_unlock_irqrestore(&desc->lock, flags); return 0; } @@ -218,7 +218,7 @@ int set_irq_msi(unsigned int irq, struct msi_desc *entry) } raw_spin_lock_irqsave(&desc->lock, flags); - desc->msi_desc = entry; + desc->irq_data.msi_desc = entry; if (entry) entry->irq = irq; raw_spin_unlock_irqrestore(&desc->lock, flags); @@ -243,13 +243,13 @@ int set_irq_chip_data(unsigned int irq, void *data) return -EINVAL; } - if (!desc->chip) { + if (!desc->irq_data.chip) { printk(KERN_ERR "BUG: bad set_irq_chip_data(IRQ#%d)\n", irq); return -EINVAL; } raw_spin_lock_irqsave(&desc->lock, flags); - desc->chip_data = data; + desc->irq_data.chip_data = data; raw_spin_unlock_irqrestore(&desc->lock, flags); return 0; @@ -291,7 +291,7 @@ static void default_enable(unsigned int irq) { struct irq_desc *desc = irq_to_desc(irq); - desc->chip->unmask(irq); + desc->irq_data.chip->unmask(irq); desc->status &= ~IRQ_MASKED; } @@ -309,7 +309,7 @@ static unsigned int default_startup(unsigned int irq) { struct irq_desc *desc = irq_to_desc(irq); - desc->chip->enable(irq); + desc->irq_data.chip->enable(irq); return 0; } @@ -320,7 +320,7 @@ static void default_shutdown(unsigned int irq) { struct irq_desc *desc = irq_to_desc(irq); - desc->chip->mask(irq); + desc->irq_data.chip->mask(irq); desc->status |= IRQ_MASKED; } @@ -350,28 +350,28 @@ void irq_chip_set_defaults(struct irq_chip *chip) static inline void mask_ack_irq(struct irq_desc *desc, int irq) { - if (desc->chip->mask_ack) - desc->chip->mask_ack(irq); + if (desc->irq_data.chip->mask_ack) + desc->irq_data.chip->mask_ack(irq); else { - desc->chip->mask(irq); - if (desc->chip->ack) - desc->chip->ack(irq); + desc->irq_data.chip->mask(irq); + if (desc->irq_data.chip->ack) + desc->irq_data.chip->ack(irq); } desc->status |= IRQ_MASKED; } static inline void mask_irq(struct irq_desc *desc, int irq) { - if (desc->chip->mask) { - desc->chip->mask(irq); + if (desc->irq_data.chip->mask) { + desc->irq_data.chip->mask(irq); desc->status |= IRQ_MASKED; } } static inline void unmask_irq(struct irq_desc *desc, int irq) { - if (desc->chip->unmask) { - desc->chip->unmask(irq); + if (desc->irq_data.chip->unmask) { + desc->irq_data.chip->unmask(irq); desc->status &= ~IRQ_MASKED; } } @@ -552,7 +552,7 @@ handle_fasteoi_irq(unsigned int irq, struct irq_desc *desc) raw_spin_lock(&desc->lock); desc->status &= ~IRQ_INPROGRESS; out: - desc->chip->eoi(irq); + desc->irq_data.chip->eoi(irq); raw_spin_unlock(&desc->lock); } @@ -594,8 +594,8 @@ handle_edge_irq(unsigned int irq, struct irq_desc *desc) kstat_incr_irqs_this_cpu(irq, desc); /* Start handling the irq */ - if (desc->chip->ack) - desc->chip->ack(irq); + if (desc->irq_data.chip->ack) + desc->irq_data.chip->ack(irq); /* Mark the IRQ currently in progress.*/ desc->status |= IRQ_INPROGRESS; @@ -648,15 +648,15 @@ handle_percpu_irq(unsigned int irq, struct irq_desc *desc) kstat_incr_irqs_this_cpu(irq, desc); - if (desc->chip->ack) - desc->chip->ack(irq); + if (desc->irq_data.chip->ack) + desc->irq_data.chip->ack(irq); action_ret = handle_IRQ_event(irq, desc->action); if (!noirqdebug) note_interrupt(irq, desc, action_ret); - if (desc->chip->eoi) - desc->chip->eoi(irq); + if (desc->irq_data.chip->eoi) + desc->irq_data.chip->eoi(irq); } void @@ -674,7 +674,7 @@ __set_irq_handler(unsigned int irq, irq_flow_handler_t handle, int is_chained, if (!handle) handle = handle_bad_irq; - else if (desc->chip == &no_irq_chip) { + else if (desc->irq_data.chip == &no_irq_chip) { printk(KERN_WARNING "Trying to install %sinterrupt handler " "for IRQ%d\n", is_chained ? "chained " : "", irq); /* @@ -684,7 +684,7 @@ __set_irq_handler(unsigned int irq, irq_flow_handler_t handle, int is_chained, * prevent us to setup the interrupt at all. Switch it to * dummy_irq_chip for easy transition. */ - desc->chip = &dummy_irq_chip; + desc->irq_data.chip = &dummy_irq_chip; } chip_bus_lock(irq, desc); @@ -692,7 +692,7 @@ __set_irq_handler(unsigned int irq, irq_flow_handler_t handle, int is_chained, /* Uninstall? */ if (handle == handle_bad_irq) { - if (desc->chip != &no_irq_chip) + if (desc->irq_data.chip != &no_irq_chip) mask_ack_irq(desc, irq); desc->status |= IRQ_DISABLED; desc->depth = 1; @@ -704,7 +704,7 @@ __set_irq_handler(unsigned int irq, irq_flow_handler_t handle, int is_chained, desc->status &= ~IRQ_DISABLED; desc->status |= IRQ_NOREQUEST | IRQ_NOPROBE; desc->depth = 0; - desc->chip->startup(irq); + desc->irq_data.chip->startup(irq); } raw_spin_unlock_irqrestore(&desc->lock, flags); chip_bus_sync_unlock(irq, desc); diff --git a/kernel/irq/handle.c b/kernel/irq/handle.c index 099d4fc368c3..fc27d76e83ef 100644 --- a/kernel/irq/handle.c +++ b/kernel/irq/handle.c @@ -105,7 +105,7 @@ static void init_one_irq_desc(int irq, struct irq_desc *desc, int node) raw_spin_lock_init(&desc->lock); desc->irq_data.irq = irq; #ifdef CONFIG_SMP - desc->node = node; + desc->irq_data.node = node; #endif lockdep_set_class(&desc->lock, &irq_desc_lock_class); init_kstat_irqs(desc, node, nr_cpu_ids); @@ -185,7 +185,7 @@ int __init early_irq_init(void) desc[i].irq_data.irq = i; desc[i].irq_data.chip = &no_irq_chip; #ifdef CONFIG_SMP - desc[i].node = node; + desc[i].irq_data.node = node; #endif desc[i].kstat_irqs = kstat_irqs_legacy + i * nr_cpu_ids; lockdep_set_class(&desc[i].lock, &irq_desc_lock_class); @@ -456,20 +456,20 @@ unsigned int __do_IRQ(unsigned int irq) /* * No locking required for CPU-local interrupts: */ - if (desc->chip->ack) - desc->chip->ack(irq); + if (desc->irq_data.chip->ack) + desc->irq_data.chip->ack(irq); if (likely(!(desc->status & IRQ_DISABLED))) { action_ret = handle_IRQ_event(irq, desc->action); if (!noirqdebug) note_interrupt(irq, desc, action_ret); } - desc->chip->end(irq); + desc->irq_data.chip->end(irq); return 1; } raw_spin_lock(&desc->lock); - if (desc->chip->ack) - desc->chip->ack(irq); + if (desc->irq_data.chip->ack) + desc->irq_data.chip->ack(irq); /* * REPLAY is when Linux resends an IRQ that was dropped earlier * WAITING is used by probe to mark irqs that are being tested @@ -529,7 +529,7 @@ out: * The ->end() handler has to deal with interrupts which got * disabled while the handler was running. */ - desc->chip->end(irq); + desc->irq_data.chip->end(irq); raw_spin_unlock(&desc->lock); return 1; diff --git a/kernel/irq/internals.h b/kernel/irq/internals.h index c63f3bc88f0b..a805a00cfd28 100644 --- a/kernel/irq/internals.h +++ b/kernel/irq/internals.h @@ -43,14 +43,14 @@ extern void irq_set_thread_affinity(struct irq_desc *desc); /* Inline functions for support of irq chips on slow busses */ static inline void chip_bus_lock(unsigned int irq, struct irq_desc *desc) { - if (unlikely(desc->chip->bus_lock)) - desc->chip->bus_lock(irq); + if (unlikely(desc->irq_data.chip->bus_lock)) + desc->irq_data.chip->bus_lock(irq); } static inline void chip_bus_sync_unlock(unsigned int irq, struct irq_desc *desc) { - if (unlikely(desc->chip->bus_sync_unlock)) - desc->chip->bus_sync_unlock(irq); + if (unlikely(desc->irq_data.chip->bus_sync_unlock)) + desc->irq_data.chip->bus_sync_unlock(irq); } /* @@ -67,8 +67,8 @@ static inline void print_irq_desc(unsigned int irq, struct irq_desc *desc) irq, desc, desc->depth, desc->irq_count, desc->irqs_unhandled); printk("->handle_irq(): %p, ", desc->handle_irq); print_symbol("%s\n", (unsigned long)desc->handle_irq); - printk("->chip(): %p, ", desc->chip); - print_symbol("%s\n", (unsigned long)desc->chip); + printk("->irq_data.chip(): %p, ", desc->irq_data.chip); + print_symbol("%s\n", (unsigned long)desc->irq_data.chip); printk("->action(): %p\n", desc->action); if (desc->action) { printk("->action->handler(): %p, ", desc->action->handler); diff --git a/kernel/irq/manage.c b/kernel/irq/manage.c index c3003e9d91a3..4dfb19521d9f 100644 --- a/kernel/irq/manage.c +++ b/kernel/irq/manage.c @@ -73,8 +73,8 @@ int irq_can_set_affinity(unsigned int irq) { struct irq_desc *desc = irq_to_desc(irq); - if (CHECK_IRQ_PER_CPU(desc->status) || !desc->chip || - !desc->chip->set_affinity) + if (CHECK_IRQ_PER_CPU(desc->status) || !desc->irq_data.chip || + !desc->irq_data.chip->set_affinity) return 0; return 1; @@ -111,15 +111,15 @@ int irq_set_affinity(unsigned int irq, const struct cpumask *cpumask) struct irq_desc *desc = irq_to_desc(irq); unsigned long flags; - if (!desc->chip->set_affinity) + if (!desc->irq_data.chip->set_affinity) return -EINVAL; raw_spin_lock_irqsave(&desc->lock, flags); #ifdef CONFIG_GENERIC_PENDING_IRQ if (desc->status & IRQ_MOVE_PCNTXT) { - if (!desc->chip->set_affinity(irq, cpumask)) { - cpumask_copy(desc->affinity, cpumask); + if (!desc->irq_data.chip->set_affinity(irq, cpumask)) { + cpumask_copy(desc->irq_data.affinity, cpumask); irq_set_thread_affinity(desc); } } @@ -128,8 +128,8 @@ int irq_set_affinity(unsigned int irq, const struct cpumask *cpumask) cpumask_copy(desc->pending_mask, cpumask); } #else - if (!desc->chip->set_affinity(irq, cpumask)) { - cpumask_copy(desc->affinity, cpumask); + if (!desc->irq_data.chip->set_affinity(irq, cpumask)) { + cpumask_copy(desc->irq_data.affinity, cpumask); irq_set_thread_affinity(desc); } #endif @@ -168,16 +168,16 @@ static int setup_affinity(unsigned int irq, struct irq_desc *desc) * one of the targets is online. */ if (desc->status & (IRQ_AFFINITY_SET | IRQ_NO_BALANCING)) { - if (cpumask_any_and(desc->affinity, cpu_online_mask) + if (cpumask_any_and(desc->irq_data.affinity, cpu_online_mask) < nr_cpu_ids) goto set_affinity; else desc->status &= ~IRQ_AFFINITY_SET; } - cpumask_and(desc->affinity, cpu_online_mask, irq_default_affinity); + cpumask_and(desc->irq_data.affinity, cpu_online_mask, irq_default_affinity); set_affinity: - desc->chip->set_affinity(irq, desc->affinity); + desc->irq_data.chip->set_affinity(irq, desc->irq_data.affinity); return 0; } @@ -223,7 +223,7 @@ void __disable_irq(struct irq_desc *desc, unsigned int irq, bool suspend) if (!desc->depth++) { desc->status |= IRQ_DISABLED; - desc->chip->disable(irq); + desc->irq_data.chip->disable(irq); } } @@ -313,7 +313,7 @@ void __enable_irq(struct irq_desc *desc, unsigned int irq, bool resume) * IRQ line is re-enabled. * * This function may be called from IRQ context only when - * desc->chip->bus_lock and desc->chip->bus_sync_unlock are NULL ! + * desc->irq_data.chip->bus_lock and desc->chip->bus_sync_unlock are NULL ! */ void enable_irq(unsigned int irq) { @@ -336,8 +336,8 @@ static int set_irq_wake_real(unsigned int irq, unsigned int on) struct irq_desc *desc = irq_to_desc(irq); int ret = -ENXIO; - if (desc->chip->set_wake) - ret = desc->chip->set_wake(irq, on); + if (desc->irq_data.chip->set_wake) + ret = desc->irq_data.chip->set_wake(irq, on); return ret; } @@ -432,7 +432,7 @@ int __irq_set_trigger(struct irq_desc *desc, unsigned int irq, unsigned long flags) { int ret; - struct irq_chip *chip = desc->chip; + struct irq_chip *chip = desc->irq_data.chip; if (!chip || !chip->set_type) { /* @@ -457,8 +457,8 @@ int __irq_set_trigger(struct irq_desc *desc, unsigned int irq, desc->status &= ~(IRQ_LEVEL | IRQ_TYPE_SENSE_MASK); desc->status |= flags; - if (chip != desc->chip) - irq_chip_set_defaults(desc->chip); + if (chip != desc->irq_data.chip) + irq_chip_set_defaults(desc->irq_data.chip); } return ret; @@ -528,7 +528,7 @@ again: if (!(desc->status & IRQ_DISABLED) && (desc->status & IRQ_MASKED)) { desc->status &= ~IRQ_MASKED; - desc->chip->unmask(irq); + desc->irq_data.chip->unmask(irq); } raw_spin_unlock_irq(&desc->lock); chip_bus_sync_unlock(irq, desc); @@ -556,7 +556,7 @@ irq_thread_check_affinity(struct irq_desc *desc, struct irqaction *action) } raw_spin_lock_irq(&desc->lock); - cpumask_copy(mask, desc->affinity); + cpumask_copy(mask, desc->irq_data.affinity); raw_spin_unlock_irq(&desc->lock); set_cpus_allowed_ptr(current, mask); @@ -657,7 +657,7 @@ __setup_irq(unsigned int irq, struct irq_desc *desc, struct irqaction *new) if (!desc) return -EINVAL; - if (desc->chip == &no_irq_chip) + if (desc->irq_data.chip == &no_irq_chip) return -ENOSYS; /* * Some drivers like serial.c use request_irq() heavily, @@ -752,7 +752,7 @@ __setup_irq(unsigned int irq, struct irq_desc *desc, struct irqaction *new) } if (!shared) { - irq_chip_set_defaults(desc->chip); + irq_chip_set_defaults(desc->irq_data.chip); init_waitqueue_head(&desc->wait_for_threads); @@ -779,7 +779,7 @@ __setup_irq(unsigned int irq, struct irq_desc *desc, struct irqaction *new) if (!(desc->status & IRQ_NOAUTOEN)) { desc->depth = 0; desc->status &= ~IRQ_DISABLED; - desc->chip->startup(irq); + desc->irq_data.chip->startup(irq); } else /* Undo nested disables: */ desc->depth = 1; @@ -912,17 +912,17 @@ static struct irqaction *__free_irq(unsigned int irq, void *dev_id) /* Currently used only by UML, might disappear one day: */ #ifdef CONFIG_IRQ_RELEASE_METHOD - if (desc->chip->release) - desc->chip->release(irq, dev_id); + if (desc->irq_data.chip->release) + desc->irq_data.chip->release(irq, dev_id); #endif /* If this was the last handler, shut down the IRQ line: */ if (!desc->action) { desc->status |= IRQ_DISABLED; - if (desc->chip->shutdown) - desc->chip->shutdown(irq); + if (desc->irq_data.chip->shutdown) + desc->irq_data.chip->shutdown(irq); else - desc->chip->disable(irq); + desc->irq_data.chip->disable(irq); } #ifdef CONFIG_SMP diff --git a/kernel/irq/migration.c b/kernel/irq/migration.c index 241962280836..f923c37e651a 100644 --- a/kernel/irq/migration.c +++ b/kernel/irq/migration.c @@ -24,7 +24,7 @@ void move_masked_irq(int irq) if (unlikely(cpumask_empty(desc->pending_mask))) return; - if (!desc->chip->set_affinity) + if (!desc->irq_data.chip->set_affinity) return; assert_raw_spin_locked(&desc->lock); @@ -43,8 +43,8 @@ void move_masked_irq(int irq) */ if (likely(cpumask_any_and(desc->pending_mask, cpu_online_mask) < nr_cpu_ids)) - if (!desc->chip->set_affinity(irq, desc->pending_mask)) { - cpumask_copy(desc->affinity, desc->pending_mask); + if (!desc->irq_data.chip->set_affinity(irq, desc->pending_mask)) { + cpumask_copy(desc->irq_data.affinity, desc->pending_mask); irq_set_thread_affinity(desc); } @@ -61,8 +61,8 @@ void move_native_irq(int irq) if (unlikely(desc->status & IRQ_DISABLED)) return; - desc->chip->mask(irq); + desc->irq_data.chip->mask(irq); move_masked_irq(irq); - desc->chip->unmask(irq); + desc->irq_data.chip->unmask(irq); } diff --git a/kernel/irq/numa_migrate.c b/kernel/irq/numa_migrate.c index 65d3845665ac..e7f1f16402c1 100644 --- a/kernel/irq/numa_migrate.c +++ b/kernel/irq/numa_migrate.c @@ -44,7 +44,7 @@ static bool init_copy_one_irq_desc(int irq, struct irq_desc *old_desc, return false; } raw_spin_lock_init(&desc->lock); - desc->node = node; + desc->irq_data.node = node; lockdep_set_class(&desc->lock, &irq_desc_lock_class); init_copy_kstat_irqs(old_desc, desc, node, nr_cpu_ids); init_copy_desc_masks(old_desc, desc); @@ -66,7 +66,7 @@ static struct irq_desc *__real_move_irq_desc(struct irq_desc *old_desc, unsigned int irq; unsigned long flags; - irq = old_desc->irq; + irq = old_desc->irq_data.irq; raw_spin_lock_irqsave(&sparse_irq_lock, flags); @@ -109,10 +109,10 @@ out_unlock: struct irq_desc *move_irq_desc(struct irq_desc *desc, int node) { /* those static or target node is -1, do not move them */ - if (desc->irq < NR_IRQS_LEGACY || node == -1) + if (desc->irq_data.irq < NR_IRQS_LEGACY || node == -1) return desc; - if (desc->node != node) + if (desc->irq_data.node != node) desc = __real_move_irq_desc(desc, node); return desc; diff --git a/kernel/irq/proc.c b/kernel/irq/proc.c index 09a2ee540bd2..9b0da94b5b2b 100644 --- a/kernel/irq/proc.c +++ b/kernel/irq/proc.c @@ -21,7 +21,7 @@ static struct proc_dir_entry *root_irq_dir; static int irq_affinity_proc_show(struct seq_file *m, void *v) { struct irq_desc *desc = irq_to_desc((long)m->private); - const struct cpumask *mask = desc->affinity; + const struct cpumask *mask = desc->irq_data.affinity; #ifdef CONFIG_GENERIC_PENDING_IRQ if (desc->status & IRQ_MOVE_PENDING) @@ -65,7 +65,7 @@ static ssize_t irq_affinity_proc_write(struct file *file, cpumask_var_t new_value; int err; - if (!irq_to_desc(irq)->chip->set_affinity || no_irq_affinity || + if (!irq_to_desc(irq)->irq_data.chip->set_affinity || no_irq_affinity || irq_balancing_disabled(irq)) return -EIO; @@ -185,7 +185,7 @@ static int irq_node_proc_show(struct seq_file *m, void *v) { struct irq_desc *desc = irq_to_desc((long) m->private); - seq_printf(m, "%d\n", desc->node); + seq_printf(m, "%d\n", desc->irq_data.node); return 0; } @@ -269,7 +269,7 @@ void register_irq_proc(unsigned int irq, struct irq_desc *desc) { char name [MAX_NAMELEN]; - if (!root_irq_dir || (desc->chip == &no_irq_chip) || desc->dir) + if (!root_irq_dir || (desc->irq_data.chip == &no_irq_chip) || desc->dir) return; memset(name, 0, MAX_NAMELEN); diff --git a/kernel/irq/resend.c b/kernel/irq/resend.c index 090c3763f3a2..47c56a097928 100644 --- a/kernel/irq/resend.c +++ b/kernel/irq/resend.c @@ -60,7 +60,7 @@ void check_irq_resend(struct irq_desc *desc, unsigned int irq) /* * Make sure the interrupt is enabled, before resending it: */ - desc->chip->enable(irq); + desc->irq_data.chip->enable(irq); /* * We do not resend level type interrupts. Level type @@ -70,7 +70,8 @@ void check_irq_resend(struct irq_desc *desc, unsigned int irq) if ((status & (IRQ_LEVEL | IRQ_PENDING | IRQ_REPLAY)) == IRQ_PENDING) { desc->status = (status & ~IRQ_PENDING) | IRQ_REPLAY; - if (!desc->chip->retrigger || !desc->chip->retrigger(irq)) { + if (!desc->irq_data.chip->retrigger || + !desc->irq_data.chip->retrigger(irq)) { #ifdef CONFIG_HARDIRQS_SW_RESEND /* Set it pending and activate the softirq: */ set_bit(irq, irqs_resend); diff --git a/kernel/irq/spurious.c b/kernel/irq/spurious.c index 89fb90ae534f..36c2c9289e2b 100644 --- a/kernel/irq/spurious.c +++ b/kernel/irq/spurious.c @@ -78,8 +78,8 @@ static int try_one_irq(int irq, struct irq_desc *desc) * If we did actual work for the real IRQ line we must let the * IRQ controller clean up too */ - if (work && desc->chip && desc->chip->end) - desc->chip->end(irq); + if (work && desc->irq_data.chip && desc->irq_data.chip->end) + desc->irq_data.chip->end(irq); raw_spin_unlock(&desc->lock); return ok; @@ -254,7 +254,7 @@ void note_interrupt(unsigned int irq, struct irq_desc *desc, printk(KERN_EMERG "Disabling IRQ #%d\n", irq); desc->status |= IRQ_DISABLED | IRQ_SPURIOUS_DISABLED; desc->depth++; - desc->chip->disable(irq); + desc->irq_data.chip->disable(irq); mod_timer(&poll_spurious_irq_timer, jiffies + POLL_SPURIOUS_IRQ_INTERVAL); -- cgit v1.2.3 From a77c4635915021c646cc017f22239e66d1aab4d5 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Fri, 1 Oct 2010 14:44:58 +0200 Subject: genirq: Add new functions to dummy chips The compat functions go away when the core code is converted. Signed-off-by: Thomas Gleixner Reviewed-by: Ingo Molnar --- kernel/irq/handle.c | 59 +++++++++++++++++++++++++++++++++++--------------- kernel/irq/internals.h | 2 ++ 2 files changed, 44 insertions(+), 17 deletions(-) (limited to 'kernel') diff --git a/kernel/irq/handle.c b/kernel/irq/handle.c index fc27d76e83ef..adca5b4b40d8 100644 --- a/kernel/irq/handle.c +++ b/kernel/irq/handle.c @@ -291,7 +291,15 @@ void clear_kstat_irqs(struct irq_desc *desc) * What should we do if we get a hw irq event on an illegal vector? * Each architecture has to answer this themself. */ -static void ack_bad(unsigned int irq) +static void ack_bad(struct irq_data *data) +{ + struct irq_desc *desc = irq_data_to_desc(data); + + print_irq_desc(data->irq, desc); + ack_bad_irq(data->irq); +} + +static void compat_ack_bad(unsigned int irq) { struct irq_desc *desc = irq_to_desc(irq); @@ -302,11 +310,16 @@ static void ack_bad(unsigned int irq) /* * NOP functions */ -static void noop(unsigned int irq) +static void noop(struct irq_data *data) { } + +static unsigned int noop_ret(struct irq_data *data) { + return 0; } -static unsigned int noop_ret(unsigned int irq) +static void compat_noop(unsigned int irq) { } + +static unsigned int compat_noop_ret(unsigned int irq) { return 0; } @@ -316,12 +329,17 @@ static unsigned int noop_ret(unsigned int irq) */ struct irq_chip no_irq_chip = { .name = "none", - .startup = noop_ret, - .shutdown = noop, - .enable = noop, - .disable = noop, - .ack = ack_bad, - .end = noop, + .irq_startup = noop_ret, + .irq_shutdown = noop, + .irq_enable = noop, + .irq_disable = noop, + .irq_ack = ack_bad, + .startup = compat_noop_ret, + .shutdown = compat_noop, + .enable = compat_noop, + .disable = compat_noop, + .ack = compat_ack_bad, + .end = compat_noop, }; /* @@ -330,14 +348,21 @@ struct irq_chip no_irq_chip = { */ struct irq_chip dummy_irq_chip = { .name = "dummy", - .startup = noop_ret, - .shutdown = noop, - .enable = noop, - .disable = noop, - .ack = noop, - .mask = noop, - .unmask = noop, - .end = noop, + .irq_startup = noop_ret, + .irq_shutdown = noop, + .irq_enable = noop, + .irq_disable = noop, + .irq_ack = noop, + .irq_mask = noop, + .irq_unmask = noop, + .startup = compat_noop_ret, + .shutdown = compat_noop, + .enable = compat_noop, + .disable = compat_noop, + .ack = compat_noop, + .mask = compat_noop, + .unmask = compat_noop, + .end = compat_noop, }; /* diff --git a/kernel/irq/internals.h b/kernel/irq/internals.h index a805a00cfd28..562fc7eeabec 100644 --- a/kernel/irq/internals.h +++ b/kernel/irq/internals.h @@ -4,6 +4,8 @@ extern int noirqdebug; +#define irq_data_to_desc(data) container_of(data, struct irq_desc, irq_data) + /* Set default functions for irq_chip structures: */ extern void irq_chip_set_defaults(struct irq_chip *chip); -- cgit v1.2.3 From 3876ec9ef3775d062345b3760d3271ecb8cd3fea Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Mon, 27 Sep 2010 12:44:35 +0000 Subject: genirq: Provide compat handling for bus_lock/bus_sync_unlock Wrap the old chip functions for bus_lock/bus_sync_unlock until the migration is complete and the old chip functions are removed. Signed-off-by: Thomas Gleixner Cc: Peter Zijlstra LKML-Reference: <20100927121841.842536121@linutronix.de> Reviewed-by: H. Peter Anvin Reviewed-by: Ingo Molnar --- kernel/irq/chip.c | 20 ++++++++++++++++++-- kernel/irq/internals.h | 12 ++++++------ kernel/irq/manage.c | 22 +++++++++++----------- 3 files changed, 35 insertions(+), 19 deletions(-) (limited to 'kernel') diff --git a/kernel/irq/chip.c b/kernel/irq/chip.c index e0e93ff10afd..77e551d92239 100644 --- a/kernel/irq/chip.c +++ b/kernel/irq/chip.c @@ -324,6 +324,17 @@ static void default_shutdown(unsigned int irq) desc->status |= IRQ_MASKED; } +/* Temporary migration helpers */ +static void compat_bus_lock(struct irq_data *data) +{ + data->chip->bus_lock(data->irq); +} + +static void compat_bus_sync_unlock(struct irq_data *data) +{ + data->chip->bus_sync_unlock(data->irq); +} + /* * Fixup enable/disable function pointers */ @@ -346,6 +357,11 @@ void irq_chip_set_defaults(struct irq_chip *chip) chip->disable : default_shutdown; if (!chip->end) chip->end = dummy_irq_chip.end; + + if (chip->bus_lock) + chip->irq_bus_lock = compat_bus_lock; + if (chip->bus_sync_unlock) + chip->irq_bus_sync_unlock = compat_bus_sync_unlock; } static inline void mask_ack_irq(struct irq_desc *desc, int irq) @@ -687,7 +703,7 @@ __set_irq_handler(unsigned int irq, irq_flow_handler_t handle, int is_chained, desc->irq_data.chip = &dummy_irq_chip; } - chip_bus_lock(irq, desc); + chip_bus_lock(desc); raw_spin_lock_irqsave(&desc->lock, flags); /* Uninstall? */ @@ -707,7 +723,7 @@ __set_irq_handler(unsigned int irq, irq_flow_handler_t handle, int is_chained, desc->irq_data.chip->startup(irq); } raw_spin_unlock_irqrestore(&desc->lock, flags); - chip_bus_sync_unlock(irq, desc); + chip_bus_sync_unlock(desc); } EXPORT_SYMBOL_GPL(__set_irq_handler); diff --git a/kernel/irq/internals.h b/kernel/irq/internals.h index 562fc7eeabec..ecafbfee5b12 100644 --- a/kernel/irq/internals.h +++ b/kernel/irq/internals.h @@ -43,16 +43,16 @@ extern int irq_select_affinity_usr(unsigned int irq); extern void irq_set_thread_affinity(struct irq_desc *desc); /* Inline functions for support of irq chips on slow busses */ -static inline void chip_bus_lock(unsigned int irq, struct irq_desc *desc) +static inline void chip_bus_lock(struct irq_desc *desc) { - if (unlikely(desc->irq_data.chip->bus_lock)) - desc->irq_data.chip->bus_lock(irq); + if (unlikely(desc->irq_data.chip->irq_bus_lock)) + desc->irq_data.chip->irq_bus_lock(&desc->irq_data); } -static inline void chip_bus_sync_unlock(unsigned int irq, struct irq_desc *desc) +static inline void chip_bus_sync_unlock(struct irq_desc *desc) { - if (unlikely(desc->irq_data.chip->bus_sync_unlock)) - desc->irq_data.chip->bus_sync_unlock(irq); + if (unlikely(desc->irq_data.chip->irq_bus_sync_unlock)) + desc->irq_data.chip->irq_bus_sync_unlock(&desc->irq_data); } /* diff --git a/kernel/irq/manage.c b/kernel/irq/manage.c index 4dfb19521d9f..dfb02ff7d2ef 100644 --- a/kernel/irq/manage.c +++ b/kernel/irq/manage.c @@ -246,11 +246,11 @@ void disable_irq_nosync(unsigned int irq) if (!desc) return; - chip_bus_lock(irq, desc); + chip_bus_lock(desc); raw_spin_lock_irqsave(&desc->lock, flags); __disable_irq(desc, irq, false); raw_spin_unlock_irqrestore(&desc->lock, flags); - chip_bus_sync_unlock(irq, desc); + chip_bus_sync_unlock(desc); } EXPORT_SYMBOL(disable_irq_nosync); @@ -323,11 +323,11 @@ void enable_irq(unsigned int irq) if (!desc) return; - chip_bus_lock(irq, desc); + chip_bus_lock(desc); raw_spin_lock_irqsave(&desc->lock, flags); __enable_irq(desc, irq, false); raw_spin_unlock_irqrestore(&desc->lock, flags); - chip_bus_sync_unlock(irq, desc); + chip_bus_sync_unlock(desc); } EXPORT_SYMBOL(enable_irq); @@ -507,7 +507,7 @@ static int irq_wait_for_interrupt(struct irqaction *action) static void irq_finalize_oneshot(unsigned int irq, struct irq_desc *desc) { again: - chip_bus_lock(irq, desc); + chip_bus_lock(desc); raw_spin_lock_irq(&desc->lock); /* @@ -521,7 +521,7 @@ again: */ if (unlikely(desc->status & IRQ_INPROGRESS)) { raw_spin_unlock_irq(&desc->lock); - chip_bus_sync_unlock(irq, desc); + chip_bus_sync_unlock(desc); cpu_relax(); goto again; } @@ -531,7 +531,7 @@ again: desc->irq_data.chip->unmask(irq); } raw_spin_unlock_irq(&desc->lock); - chip_bus_sync_unlock(irq, desc); + chip_bus_sync_unlock(desc); } #ifdef CONFIG_SMP @@ -997,9 +997,9 @@ void free_irq(unsigned int irq, void *dev_id) if (!desc) return; - chip_bus_lock(irq, desc); + chip_bus_lock(desc); kfree(__free_irq(irq, dev_id)); - chip_bus_sync_unlock(irq, desc); + chip_bus_sync_unlock(desc); } EXPORT_SYMBOL(free_irq); @@ -1086,9 +1086,9 @@ int request_threaded_irq(unsigned int irq, irq_handler_t handler, action->name = devname; action->dev_id = dev_id; - chip_bus_lock(irq, desc); + chip_bus_lock(desc); retval = __setup_irq(irq, desc, action); - chip_bus_sync_unlock(irq, desc); + chip_bus_sync_unlock(desc); if (retval) kfree(action); -- cgit v1.2.3 From e2c0f8ff0fc26959952fbfa89f732fef928df77f Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Mon, 27 Sep 2010 12:44:42 +0000 Subject: genirq: Provide compat handling for chip->mask() Wrap the old chip function mask() until the migration is complete and the old chip functions are removed. Signed-off-by: Thomas Gleixner Cc: Peter Zijlstra LKML-Reference: <20100927121841.940355859@linutronix.de> Reviewed-by: H. Peter Anvin Reviewed-by: Ingo Molnar --- kernel/irq/chip.c | 22 +++++++++++++++------- kernel/irq/handle.c | 1 - kernel/irq/migration.c | 2 +- 3 files changed, 16 insertions(+), 9 deletions(-) (limited to 'kernel') diff --git a/kernel/irq/chip.c b/kernel/irq/chip.c index 77e551d92239..c041270bfe50 100644 --- a/kernel/irq/chip.c +++ b/kernel/irq/chip.c @@ -320,11 +320,16 @@ static void default_shutdown(unsigned int irq) { struct irq_desc *desc = irq_to_desc(irq); - desc->irq_data.chip->mask(irq); + desc->irq_data.chip->irq_mask(&desc->irq_data); desc->status |= IRQ_MASKED; } /* Temporary migration helpers */ +static void compat_irq_mask(struct irq_data *data) +{ + data->chip->mask(data->irq); +} + static void compat_bus_lock(struct irq_data *data) { data->chip->bus_lock(data->irq); @@ -362,6 +367,9 @@ void irq_chip_set_defaults(struct irq_chip *chip) chip->irq_bus_lock = compat_bus_lock; if (chip->bus_sync_unlock) chip->irq_bus_sync_unlock = compat_bus_sync_unlock; + + if (chip->mask) + chip->irq_mask = compat_irq_mask; } static inline void mask_ack_irq(struct irq_desc *desc, int irq) @@ -369,17 +377,17 @@ static inline void mask_ack_irq(struct irq_desc *desc, int irq) if (desc->irq_data.chip->mask_ack) desc->irq_data.chip->mask_ack(irq); else { - desc->irq_data.chip->mask(irq); + desc->irq_data.chip->irq_mask(&desc->irq_data); if (desc->irq_data.chip->ack) desc->irq_data.chip->ack(irq); } desc->status |= IRQ_MASKED; } -static inline void mask_irq(struct irq_desc *desc, int irq) +static inline void mask_irq(struct irq_desc *desc) { - if (desc->irq_data.chip->mask) { - desc->irq_data.chip->mask(irq); + if (desc->irq_data.chip->irq_mask) { + desc->irq_data.chip->irq_mask(&desc->irq_data); desc->status |= IRQ_MASKED; } } @@ -553,7 +561,7 @@ handle_fasteoi_irq(unsigned int irq, struct irq_desc *desc) action = desc->action; if (unlikely(!action || (desc->status & IRQ_DISABLED))) { desc->status |= IRQ_PENDING; - mask_irq(desc, irq); + mask_irq(desc); goto out; } @@ -621,7 +629,7 @@ handle_edge_irq(unsigned int irq, struct irq_desc *desc) irqreturn_t action_ret; if (unlikely(!action)) { - mask_irq(desc, irq); + mask_irq(desc); goto out_unlock; } diff --git a/kernel/irq/handle.c b/kernel/irq/handle.c index adca5b4b40d8..3b160ac236b0 100644 --- a/kernel/irq/handle.c +++ b/kernel/irq/handle.c @@ -360,7 +360,6 @@ struct irq_chip dummy_irq_chip = { .enable = compat_noop, .disable = compat_noop, .ack = compat_noop, - .mask = compat_noop, .unmask = compat_noop, .end = compat_noop, }; diff --git a/kernel/irq/migration.c b/kernel/irq/migration.c index f923c37e651a..b165ec26b757 100644 --- a/kernel/irq/migration.c +++ b/kernel/irq/migration.c @@ -61,7 +61,7 @@ void move_native_irq(int irq) if (unlikely(desc->status & IRQ_DISABLED)) return; - desc->irq_data.chip->mask(irq); + desc->irq_data.chip->irq_mask(&desc->irq_data); move_masked_irq(irq); desc->irq_data.chip->unmask(irq); } -- cgit v1.2.3 From 0eda58b7f3a30c9a13d83db1cfaab00e1c452055 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Mon, 27 Sep 2010 12:44:44 +0000 Subject: genirq: Provide compat handling for chip->unmask() Wrap the old chip function unmask() until the migration is complete and the old chip functions are removed. Signed-off-by: Thomas Gleixner Cc: Peter Zijlstra LKML-Reference: <20100927121842.043608928@linutronix.de> Reviewed-by: H. Peter Anvin Reviewed-by: Ingo Molnar --- kernel/irq/chip.c | 20 ++++++++++++++------ kernel/irq/handle.c | 1 - kernel/irq/manage.c | 2 +- kernel/irq/migration.c | 2 +- 4 files changed, 16 insertions(+), 9 deletions(-) (limited to 'kernel') diff --git a/kernel/irq/chip.c b/kernel/irq/chip.c index c041270bfe50..dbdb59a42093 100644 --- a/kernel/irq/chip.c +++ b/kernel/irq/chip.c @@ -291,7 +291,7 @@ static void default_enable(unsigned int irq) { struct irq_desc *desc = irq_to_desc(irq); - desc->irq_data.chip->unmask(irq); + desc->irq_data.chip->irq_unmask(&desc->irq_data); desc->status &= ~IRQ_MASKED; } @@ -330,6 +330,11 @@ static void compat_irq_mask(struct irq_data *data) data->chip->mask(data->irq); } +static void compat_irq_unmask(struct irq_data *data) +{ + data->chip->unmask(data->irq); +} + static void compat_bus_lock(struct irq_data *data) { data->chip->bus_lock(data->irq); @@ -370,6 +375,9 @@ void irq_chip_set_defaults(struct irq_chip *chip) if (chip->mask) chip->irq_mask = compat_irq_mask; + + if (chip->unmask) + chip->irq_unmask = compat_irq_unmask; } static inline void mask_ack_irq(struct irq_desc *desc, int irq) @@ -392,10 +400,10 @@ static inline void mask_irq(struct irq_desc *desc) } } -static inline void unmask_irq(struct irq_desc *desc, int irq) +static inline void unmask_irq(struct irq_desc *desc) { - if (desc->irq_data.chip->unmask) { - desc->irq_data.chip->unmask(irq); + if (desc->irq_data.chip->irq_unmask) { + desc->irq_data.chip->irq_unmask(&desc->irq_data); desc->status &= ~IRQ_MASKED; } } @@ -524,7 +532,7 @@ handle_level_irq(unsigned int irq, struct irq_desc *desc) desc->status &= ~IRQ_INPROGRESS; if (!(desc->status & (IRQ_DISABLED | IRQ_ONESHOT))) - unmask_irq(desc, irq); + unmask_irq(desc); out_unlock: raw_spin_unlock(&desc->lock); } @@ -641,7 +649,7 @@ handle_edge_irq(unsigned int irq, struct irq_desc *desc) if (unlikely((desc->status & (IRQ_PENDING | IRQ_MASKED | IRQ_DISABLED)) == (IRQ_PENDING | IRQ_MASKED))) { - unmask_irq(desc, irq); + unmask_irq(desc); } desc->status &= ~IRQ_PENDING; diff --git a/kernel/irq/handle.c b/kernel/irq/handle.c index 3b160ac236b0..f334c8c59dd2 100644 --- a/kernel/irq/handle.c +++ b/kernel/irq/handle.c @@ -360,7 +360,6 @@ struct irq_chip dummy_irq_chip = { .enable = compat_noop, .disable = compat_noop, .ack = compat_noop, - .unmask = compat_noop, .end = compat_noop, }; diff --git a/kernel/irq/manage.c b/kernel/irq/manage.c index dfb02ff7d2ef..b3986bce64ff 100644 --- a/kernel/irq/manage.c +++ b/kernel/irq/manage.c @@ -528,7 +528,7 @@ again: if (!(desc->status & IRQ_DISABLED) && (desc->status & IRQ_MASKED)) { desc->status &= ~IRQ_MASKED; - desc->irq_data.chip->unmask(irq); + desc->irq_data.chip->irq_unmask(&desc->irq_data); } raw_spin_unlock_irq(&desc->lock); chip_bus_sync_unlock(desc); diff --git a/kernel/irq/migration.c b/kernel/irq/migration.c index b165ec26b757..7888e5d5575a 100644 --- a/kernel/irq/migration.c +++ b/kernel/irq/migration.c @@ -63,6 +63,6 @@ void move_native_irq(int irq) desc->irq_data.chip->irq_mask(&desc->irq_data); move_masked_irq(irq); - desc->irq_data.chip->unmask(irq); + desc->irq_data.chip->irq_unmask(&desc->irq_data); } -- cgit v1.2.3 From 22a49163e90dd7088f801dd54e25b04e1f337e9b Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Mon, 27 Sep 2010 12:44:47 +0000 Subject: genirq: Provide compat handling for chip->ack() Wrap the old chip function ack() until the migration is complete and the old chip functions are removed. Signed-off-by: Thomas Gleixner Cc: Peter Zijlstra LKML-Reference: <20100927121842.142624725@linutronix.de> Reviewed-by: H. Peter Anvin Reviewed-by: Ingo Molnar --- kernel/irq/chip.c | 20 ++++++++++++-------- kernel/irq/handle.c | 10 ---------- 2 files changed, 12 insertions(+), 18 deletions(-) (limited to 'kernel') diff --git a/kernel/irq/chip.c b/kernel/irq/chip.c index dbdb59a42093..864c3abdf8f4 100644 --- a/kernel/irq/chip.c +++ b/kernel/irq/chip.c @@ -335,6 +335,11 @@ static void compat_irq_unmask(struct irq_data *data) data->chip->unmask(data->irq); } +static void compat_irq_ack(struct irq_data *data) +{ + data->chip->ack(data->irq); +} + static void compat_bus_lock(struct irq_data *data) { data->chip->bus_lock(data->irq); @@ -372,12 +377,12 @@ void irq_chip_set_defaults(struct irq_chip *chip) chip->irq_bus_lock = compat_bus_lock; if (chip->bus_sync_unlock) chip->irq_bus_sync_unlock = compat_bus_sync_unlock; - if (chip->mask) chip->irq_mask = compat_irq_mask; - if (chip->unmask) chip->irq_unmask = compat_irq_unmask; + if (chip->ack) + chip->irq_ack = compat_irq_ack; } static inline void mask_ack_irq(struct irq_desc *desc, int irq) @@ -386,8 +391,8 @@ static inline void mask_ack_irq(struct irq_desc *desc, int irq) desc->irq_data.chip->mask_ack(irq); else { desc->irq_data.chip->irq_mask(&desc->irq_data); - if (desc->irq_data.chip->ack) - desc->irq_data.chip->ack(irq); + if (desc->irq_data.chip->irq_ack) + desc->irq_data.chip->irq_ack(&desc->irq_data); } desc->status |= IRQ_MASKED; } @@ -626,8 +631,7 @@ handle_edge_irq(unsigned int irq, struct irq_desc *desc) kstat_incr_irqs_this_cpu(irq, desc); /* Start handling the irq */ - if (desc->irq_data.chip->ack) - desc->irq_data.chip->ack(irq); + desc->irq_data.chip->irq_ack(&desc->irq_data); /* Mark the IRQ currently in progress.*/ desc->status |= IRQ_INPROGRESS; @@ -680,8 +684,8 @@ handle_percpu_irq(unsigned int irq, struct irq_desc *desc) kstat_incr_irqs_this_cpu(irq, desc); - if (desc->irq_data.chip->ack) - desc->irq_data.chip->ack(irq); + if (desc->irq_data.chip->irq_ack) + desc->irq_data.chip->irq_ack(&desc->irq_data); action_ret = handle_IRQ_event(irq, desc->action); if (!noirqdebug) diff --git a/kernel/irq/handle.c b/kernel/irq/handle.c index f334c8c59dd2..9ba7aece0e4d 100644 --- a/kernel/irq/handle.c +++ b/kernel/irq/handle.c @@ -299,14 +299,6 @@ static void ack_bad(struct irq_data *data) ack_bad_irq(data->irq); } -static void compat_ack_bad(unsigned int irq) -{ - struct irq_desc *desc = irq_to_desc(irq); - - print_irq_desc(irq, desc); - ack_bad_irq(irq); -} - /* * NOP functions */ @@ -338,7 +330,6 @@ struct irq_chip no_irq_chip = { .shutdown = compat_noop, .enable = compat_noop, .disable = compat_noop, - .ack = compat_ack_bad, .end = compat_noop, }; @@ -359,7 +350,6 @@ struct irq_chip dummy_irq_chip = { .shutdown = compat_noop, .enable = compat_noop, .disable = compat_noop, - .ack = compat_noop, .end = compat_noop, }; -- cgit v1.2.3 From 9205e31d1af0f725e71bbae10d199c6b9e8d6dd8 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Mon, 27 Sep 2010 12:44:50 +0000 Subject: genirq: Provide compat handling for chip->mask_ack() Wrap the old chip function mask_ack() until the migration is complete and the old chip functions are removed. Signed-off-by: Thomas Gleixner Cc: Peter Zijlstra LKML-Reference: <20100927121842.240806983@linutronix.de> Reviewed-by: H. Peter Anvin Reviewed-by: Ingo Molnar --- kernel/irq/chip.c | 19 +++++++++++++------ 1 file changed, 13 insertions(+), 6 deletions(-) (limited to 'kernel') diff --git a/kernel/irq/chip.c b/kernel/irq/chip.c index 864c3abdf8f4..09c1a4493440 100644 --- a/kernel/irq/chip.c +++ b/kernel/irq/chip.c @@ -340,6 +340,11 @@ static void compat_irq_ack(struct irq_data *data) data->chip->ack(data->irq); } +static void compat_irq_mask_ack(struct irq_data *data) +{ + data->chip->mask_ack(data->irq); +} + static void compat_bus_lock(struct irq_data *data) { data->chip->bus_lock(data->irq); @@ -383,12 +388,14 @@ void irq_chip_set_defaults(struct irq_chip *chip) chip->irq_unmask = compat_irq_unmask; if (chip->ack) chip->irq_ack = compat_irq_ack; + if (chip->mask_ack) + chip->irq_mask_ack = compat_irq_mask_ack; } -static inline void mask_ack_irq(struct irq_desc *desc, int irq) +static inline void mask_ack_irq(struct irq_desc *desc) { - if (desc->irq_data.chip->mask_ack) - desc->irq_data.chip->mask_ack(irq); + if (desc->irq_data.chip->irq_mask_ack) + desc->irq_data.chip->irq_mask_ack(&desc->irq_data); else { desc->irq_data.chip->irq_mask(&desc->irq_data); if (desc->irq_data.chip->irq_ack) @@ -511,7 +518,7 @@ handle_level_irq(unsigned int irq, struct irq_desc *desc) irqreturn_t action_ret; raw_spin_lock(&desc->lock); - mask_ack_irq(desc, irq); + mask_ack_irq(desc); if (unlikely(desc->status & IRQ_INPROGRESS)) goto out_unlock; @@ -625,7 +632,7 @@ handle_edge_irq(unsigned int irq, struct irq_desc *desc) if (unlikely((desc->status & (IRQ_INPROGRESS | IRQ_DISABLED)) || !desc->action)) { desc->status |= (IRQ_PENDING | IRQ_MASKED); - mask_ack_irq(desc, irq); + mask_ack_irq(desc); goto out_unlock; } kstat_incr_irqs_this_cpu(irq, desc); @@ -729,7 +736,7 @@ __set_irq_handler(unsigned int irq, irq_flow_handler_t handle, int is_chained, /* Uninstall? */ if (handle == handle_bad_irq) { if (desc->irq_data.chip != &no_irq_chip) - mask_ack_irq(desc, irq); + mask_ack_irq(desc); desc->status |= IRQ_DISABLED; desc->depth = 1; } -- cgit v1.2.3 From 0c5c15572ac096001f52d37b416f2a4be9aebb80 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Mon, 27 Sep 2010 12:44:53 +0000 Subject: genirq: Provide compat handling for chip->eoi() Wrap the old chip function eoi() until the migration is complete and the old chip functions are removed. Signed-off-by: Thomas Gleixner Cc: Peter Zijlstra LKML-Reference: <20100927121842.339657617@linutronix.de> Reviewed-by: H. Peter Anvin Reviewed-by: Ingo Molnar --- kernel/irq/chip.c | 13 ++++++++++--- 1 file changed, 10 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/irq/chip.c b/kernel/irq/chip.c index 09c1a4493440..c8648a83b80a 100644 --- a/kernel/irq/chip.c +++ b/kernel/irq/chip.c @@ -345,6 +345,11 @@ static void compat_irq_mask_ack(struct irq_data *data) data->chip->mask_ack(data->irq); } +static void compat_irq_eoi(struct irq_data *data) +{ + data->chip->eoi(data->irq); +} + static void compat_bus_lock(struct irq_data *data) { data->chip->bus_lock(data->irq); @@ -390,6 +395,8 @@ void irq_chip_set_defaults(struct irq_chip *chip) chip->irq_ack = compat_irq_ack; if (chip->mask_ack) chip->irq_mask_ack = compat_irq_mask_ack; + if (chip->eoi) + chip->irq_eoi = compat_irq_eoi; } static inline void mask_ack_irq(struct irq_desc *desc) @@ -596,7 +603,7 @@ handle_fasteoi_irq(unsigned int irq, struct irq_desc *desc) raw_spin_lock(&desc->lock); desc->status &= ~IRQ_INPROGRESS; out: - desc->irq_data.chip->eoi(irq); + desc->irq_data.chip->irq_eoi(&desc->irq_data); raw_spin_unlock(&desc->lock); } @@ -698,8 +705,8 @@ handle_percpu_irq(unsigned int irq, struct irq_desc *desc) if (!noirqdebug) note_interrupt(irq, desc, action_ret); - if (desc->irq_data.chip->eoi) - desc->irq_data.chip->eoi(irq); + if (desc->irq_data.chip->irq_eoi) + desc->irq_data.chip->irq_eoi(&desc->irq_data); } void -- cgit v1.2.3 From c5f756344c390f629243b4a28c2bd198fdfd7ee9 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Mon, 27 Sep 2010 12:44:56 +0000 Subject: genirq: Provide compat handling for chip->enable() Wrap the old chip function enable() until the migration is complete and the old chip functions are removed. Signed-off-by: Thomas Gleixner Cc: Peter Zijlstra LKML-Reference: <20100927121842.437159182@linutronix.de> Reviewed-by: H. Peter Anvin Reviewed-by: Ingo Molnar --- kernel/irq/chip.c | 25 ++++++++++++++++++++----- kernel/irq/handle.c | 2 -- kernel/irq/resend.c | 2 +- 3 files changed, 21 insertions(+), 8 deletions(-) (limited to 'kernel') diff --git a/kernel/irq/chip.c b/kernel/irq/chip.c index c8648a83b80a..a95b47831269 100644 --- a/kernel/irq/chip.c +++ b/kernel/irq/chip.c @@ -287,9 +287,9 @@ EXPORT_SYMBOL_GPL(set_irq_nested_thread); /* * default enable function */ -static void default_enable(unsigned int irq) +static void default_enable(struct irq_data *data) { - struct irq_desc *desc = irq_to_desc(irq); + struct irq_desc *desc = irq_data_to_desc(data); desc->irq_data.chip->irq_unmask(&desc->irq_data); desc->status &= ~IRQ_MASKED; @@ -309,7 +309,7 @@ static unsigned int default_startup(unsigned int irq) { struct irq_desc *desc = irq_to_desc(irq); - desc->irq_data.chip->enable(irq); + desc->irq_data.chip->irq_enable(&desc->irq_data); return 0; } @@ -350,6 +350,11 @@ static void compat_irq_eoi(struct irq_data *data) data->chip->eoi(data->irq); } +static void compat_irq_enable(struct irq_data *data) +{ + data->chip->enable(data->irq); +} + static void compat_bus_lock(struct irq_data *data) { data->chip->bus_lock(data->irq); @@ -365,8 +370,18 @@ static void compat_bus_sync_unlock(struct irq_data *data) */ void irq_chip_set_defaults(struct irq_chip *chip) { - if (!chip->enable) - chip->enable = default_enable; + /* + * Compat fixup functions need to be before we set the + * defaults for enable/disable/startup/shutdown + */ + if (chip->enable) + chip->irq_enable = compat_irq_enable; + + /* + * The real defaults + */ + if (!chip->irq_enable) + chip->irq_enable = default_enable; if (!chip->disable) chip->disable = default_disable; if (!chip->startup) diff --git a/kernel/irq/handle.c b/kernel/irq/handle.c index 9ba7aece0e4d..ac8c749dfee5 100644 --- a/kernel/irq/handle.c +++ b/kernel/irq/handle.c @@ -328,7 +328,6 @@ struct irq_chip no_irq_chip = { .irq_ack = ack_bad, .startup = compat_noop_ret, .shutdown = compat_noop, - .enable = compat_noop, .disable = compat_noop, .end = compat_noop, }; @@ -348,7 +347,6 @@ struct irq_chip dummy_irq_chip = { .irq_unmask = noop, .startup = compat_noop_ret, .shutdown = compat_noop, - .enable = compat_noop, .disable = compat_noop, .end = compat_noop, }; diff --git a/kernel/irq/resend.c b/kernel/irq/resend.c index 47c56a097928..a798a2328f8a 100644 --- a/kernel/irq/resend.c +++ b/kernel/irq/resend.c @@ -60,7 +60,7 @@ void check_irq_resend(struct irq_desc *desc, unsigned int irq) /* * Make sure the interrupt is enabled, before resending it: */ - desc->irq_data.chip->enable(irq); + desc->irq_data.chip->irq_enable(&desc->irq_data); /* * We do not resend level type interrupts. Level type -- cgit v1.2.3 From bc310dda41be6439364c8f3b9fe7c9d743d22b1c Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Mon, 27 Sep 2010 12:45:02 +0000 Subject: genirq: Provide compat handling for chip->disable()/shutdown() Wrap the old chip functions disable() and shutdown() until the migration is complete and the old chip functions are removed. Signed-off-by: Thomas Gleixner Cc: Peter Zijlstra LKML-Reference: <20100927121842.532070631@linutronix.de> Reviewed-by: H. Peter Anvin Reviewed-by: Ingo Molnar --- kernel/irq/autoprobe.c | 6 +++--- kernel/irq/chip.c | 37 +++++++++++++++++++++++++++---------- kernel/irq/handle.c | 4 ---- kernel/irq/manage.c | 8 ++++---- kernel/irq/spurious.c | 2 +- 5 files changed, 35 insertions(+), 22 deletions(-) (limited to 'kernel') diff --git a/kernel/irq/autoprobe.c b/kernel/irq/autoprobe.c index f9bf9b228033..95806a45be78 100644 --- a/kernel/irq/autoprobe.c +++ b/kernel/irq/autoprobe.c @@ -98,7 +98,7 @@ unsigned long probe_irq_on(void) /* It triggered already - consider it spurious. */ if (!(status & IRQ_WAITING)) { desc->status = status & ~IRQ_AUTODETECT; - desc->irq_data.chip->shutdown(i); + desc->irq_data.chip->irq_shutdown(&desc->irq_data); } else if (i < 32) mask |= 1 << i; @@ -137,7 +137,7 @@ unsigned int probe_irq_mask(unsigned long val) mask |= 1 << i; desc->status = status & ~IRQ_AUTODETECT; - desc->irq_data.chip->shutdown(i); + desc->irq_data.chip->irq_shutdown(&desc->irq_data); } raw_spin_unlock_irq(&desc->lock); } @@ -181,7 +181,7 @@ int probe_irq_off(unsigned long val) nr_of_irqs++; } desc->status = status & ~IRQ_AUTODETECT; - desc->irq_data.chip->shutdown(i); + desc->irq_data.chip->irq_shutdown(&desc->irq_data); } raw_spin_unlock_irq(&desc->lock); } diff --git a/kernel/irq/chip.c b/kernel/irq/chip.c index a95b47831269..b8a47f0a26cc 100644 --- a/kernel/irq/chip.c +++ b/kernel/irq/chip.c @@ -298,7 +298,7 @@ static void default_enable(struct irq_data *data) /* * default disable function */ -static void default_disable(unsigned int irq) +static void default_disable(struct irq_data *data) { } @@ -316,9 +316,9 @@ static unsigned int default_startup(unsigned int irq) /* * default shutdown function */ -static void default_shutdown(unsigned int irq) +static void default_shutdown(struct irq_data *data) { - struct irq_desc *desc = irq_to_desc(irq); + struct irq_desc *desc = irq_data_to_desc(data); desc->irq_data.chip->irq_mask(&desc->irq_data); desc->status |= IRQ_MASKED; @@ -355,6 +355,16 @@ static void compat_irq_enable(struct irq_data *data) data->chip->enable(data->irq); } +static void compat_irq_disable(struct irq_data *data) +{ + data->chip->disable(data->irq); +} + +static void compat_irq_shutdown(struct irq_data *data) +{ + data->chip->shutdown(data->irq); +} + static void compat_bus_lock(struct irq_data *data) { data->chip->bus_lock(data->irq); @@ -376,28 +386,35 @@ void irq_chip_set_defaults(struct irq_chip *chip) */ if (chip->enable) chip->irq_enable = compat_irq_enable; + if (chip->disable) + chip->irq_disable = compat_irq_disable; + if (chip->shutdown) + chip->irq_shutdown = compat_irq_shutdown; /* * The real defaults */ if (!chip->irq_enable) chip->irq_enable = default_enable; - if (!chip->disable) - chip->disable = default_disable; + if (!chip->irq_disable) + chip->irq_disable = default_disable; if (!chip->startup) chip->startup = default_startup; /* - * We use chip->disable, when the user provided its own. When - * we have default_disable set for chip->disable, then we need + * We use chip->irq_disable, when the user provided its own. When + * we have default_disable set for chip->irq_disable, then we need * to use default_shutdown, otherwise the irq line is not * disabled on free_irq(): */ - if (!chip->shutdown) - chip->shutdown = chip->disable != default_disable ? - chip->disable : default_shutdown; + if (!chip->irq_shutdown) + chip->irq_shutdown = chip->irq_disable != default_disable ? + chip->irq_disable : default_shutdown; if (!chip->end) chip->end = dummy_irq_chip.end; + /* + * Now fix up the remaining compat handlers + */ if (chip->bus_lock) chip->irq_bus_lock = compat_bus_lock; if (chip->bus_sync_unlock) diff --git a/kernel/irq/handle.c b/kernel/irq/handle.c index ac8c749dfee5..60e25c46eb55 100644 --- a/kernel/irq/handle.c +++ b/kernel/irq/handle.c @@ -327,8 +327,6 @@ struct irq_chip no_irq_chip = { .irq_disable = noop, .irq_ack = ack_bad, .startup = compat_noop_ret, - .shutdown = compat_noop, - .disable = compat_noop, .end = compat_noop, }; @@ -346,8 +344,6 @@ struct irq_chip dummy_irq_chip = { .irq_mask = noop, .irq_unmask = noop, .startup = compat_noop_ret, - .shutdown = compat_noop, - .disable = compat_noop, .end = compat_noop, }; diff --git a/kernel/irq/manage.c b/kernel/irq/manage.c index b3986bce64ff..f3f36f6af9a1 100644 --- a/kernel/irq/manage.c +++ b/kernel/irq/manage.c @@ -223,7 +223,7 @@ void __disable_irq(struct irq_desc *desc, unsigned int irq, bool suspend) if (!desc->depth++) { desc->status |= IRQ_DISABLED; - desc->irq_data.chip->disable(irq); + desc->irq_data.chip->irq_disable(&desc->irq_data); } } @@ -919,10 +919,10 @@ static struct irqaction *__free_irq(unsigned int irq, void *dev_id) /* If this was the last handler, shut down the IRQ line: */ if (!desc->action) { desc->status |= IRQ_DISABLED; - if (desc->irq_data.chip->shutdown) - desc->irq_data.chip->shutdown(irq); + if (desc->irq_data.chip->irq_shutdown) + desc->irq_data.chip->irq_shutdown(&desc->irq_data); else - desc->irq_data.chip->disable(irq); + desc->irq_data.chip->irq_disable(&desc->irq_data); } #ifdef CONFIG_SMP diff --git a/kernel/irq/spurious.c b/kernel/irq/spurious.c index 36c2c9289e2b..9ee704d3a23c 100644 --- a/kernel/irq/spurious.c +++ b/kernel/irq/spurious.c @@ -254,7 +254,7 @@ void note_interrupt(unsigned int irq, struct irq_desc *desc, printk(KERN_EMERG "Disabling IRQ #%d\n", irq); desc->status |= IRQ_DISABLED | IRQ_SPURIOUS_DISABLED; desc->depth++; - desc->irq_data.chip->disable(irq); + desc->irq_data.chip->irq_disable(&desc->irq_data); mod_timer(&poll_spurious_irq_timer, jiffies + POLL_SPURIOUS_IRQ_INTERVAL); -- cgit v1.2.3 From 37e12df709f09eac17314d79a52190ac46746e33 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Mon, 27 Sep 2010 12:45:38 +0000 Subject: genirq: Provide compat handling for chip->startup() Wrap the old chip function startup() until the migration is complete and the old chip functions are removed. Signed-off-by: Thomas Gleixner Cc: Peter Zijlstra LKML-Reference: <20100927121842.635152961@linutronix.de> Reviewed-by: H. Peter Anvin Reviewed-by: Ingo Molnar --- kernel/irq/autoprobe.c | 4 ++-- kernel/irq/chip.c | 19 +++++++++++++------ kernel/irq/handle.c | 7 ------- kernel/irq/manage.c | 2 +- 4 files changed, 16 insertions(+), 16 deletions(-) (limited to 'kernel') diff --git a/kernel/irq/autoprobe.c b/kernel/irq/autoprobe.c index 95806a45be78..7a468254e533 100644 --- a/kernel/irq/autoprobe.c +++ b/kernel/irq/autoprobe.c @@ -59,7 +59,7 @@ unsigned long probe_irq_on(void) */ if (desc->irq_data.chip->set_type) desc->irq_data.chip->set_type(i, IRQ_TYPE_PROBE); - desc->irq_data.chip->startup(i); + desc->irq_data.chip->irq_startup(&desc->irq_data); } raw_spin_unlock_irq(&desc->lock); } @@ -76,7 +76,7 @@ unsigned long probe_irq_on(void) raw_spin_lock_irq(&desc->lock); if (!desc->action && !(desc->status & IRQ_NOPROBE)) { desc->status |= IRQ_AUTODETECT | IRQ_WAITING; - if (desc->irq_data.chip->startup(i)) + if (desc->irq_data.chip->irq_startup(&desc->irq_data)) desc->status |= IRQ_PENDING; } raw_spin_unlock_irq(&desc->lock); diff --git a/kernel/irq/chip.c b/kernel/irq/chip.c index b8a47f0a26cc..cce85f0734b0 100644 --- a/kernel/irq/chip.c +++ b/kernel/irq/chip.c @@ -305,11 +305,11 @@ static void default_disable(struct irq_data *data) /* * default startup function */ -static unsigned int default_startup(unsigned int irq) +static unsigned int default_startup(struct irq_data *data) { - struct irq_desc *desc = irq_to_desc(irq); + struct irq_desc *desc = irq_data_to_desc(data); - desc->irq_data.chip->irq_enable(&desc->irq_data); + desc->irq_data.chip->irq_enable(data); return 0; } @@ -365,6 +365,11 @@ static void compat_irq_shutdown(struct irq_data *data) data->chip->shutdown(data->irq); } +static unsigned int compat_irq_startup(struct irq_data *data) +{ + return data->chip->startup(data->irq); +} + static void compat_bus_lock(struct irq_data *data) { data->chip->bus_lock(data->irq); @@ -390,6 +395,8 @@ void irq_chip_set_defaults(struct irq_chip *chip) chip->irq_disable = compat_irq_disable; if (chip->shutdown) chip->irq_shutdown = compat_irq_shutdown; + if (chip->startup) + chip->irq_startup = compat_irq_startup; /* * The real defaults @@ -398,8 +405,8 @@ void irq_chip_set_defaults(struct irq_chip *chip) chip->irq_enable = default_enable; if (!chip->irq_disable) chip->irq_disable = default_disable; - if (!chip->startup) - chip->startup = default_startup; + if (!chip->irq_startup) + chip->irq_startup = default_startup; /* * We use chip->irq_disable, when the user provided its own. When * we have default_disable set for chip->irq_disable, then we need @@ -786,7 +793,7 @@ __set_irq_handler(unsigned int irq, irq_flow_handler_t handle, int is_chained, desc->status &= ~IRQ_DISABLED; desc->status |= IRQ_NOREQUEST | IRQ_NOPROBE; desc->depth = 0; - desc->irq_data.chip->startup(irq); + desc->irq_data.chip->irq_startup(&desc->irq_data); } raw_spin_unlock_irqrestore(&desc->lock, flags); chip_bus_sync_unlock(desc); diff --git a/kernel/irq/handle.c b/kernel/irq/handle.c index 60e25c46eb55..8d0697f892a2 100644 --- a/kernel/irq/handle.c +++ b/kernel/irq/handle.c @@ -311,11 +311,6 @@ static unsigned int noop_ret(struct irq_data *data) static void compat_noop(unsigned int irq) { } -static unsigned int compat_noop_ret(unsigned int irq) -{ - return 0; -} - /* * Generic no controller implementation */ @@ -326,7 +321,6 @@ struct irq_chip no_irq_chip = { .irq_enable = noop, .irq_disable = noop, .irq_ack = ack_bad, - .startup = compat_noop_ret, .end = compat_noop, }; @@ -343,7 +337,6 @@ struct irq_chip dummy_irq_chip = { .irq_ack = noop, .irq_mask = noop, .irq_unmask = noop, - .startup = compat_noop_ret, .end = compat_noop, }; diff --git a/kernel/irq/manage.c b/kernel/irq/manage.c index f3f36f6af9a1..31d7678e0269 100644 --- a/kernel/irq/manage.c +++ b/kernel/irq/manage.c @@ -779,7 +779,7 @@ __setup_irq(unsigned int irq, struct irq_desc *desc, struct irqaction *new) if (!(desc->status & IRQ_NOAUTOEN)) { desc->depth = 0; desc->status &= ~IRQ_DISABLED; - desc->irq_data.chip->startup(irq); + desc->irq_data.chip->irq_startup(&desc->irq_data); } else /* Undo nested disables: */ desc->depth = 1; -- cgit v1.2.3 From c96b3b3c448592a0b87ef20306deb8b1fb4878c7 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Mon, 27 Sep 2010 12:45:41 +0000 Subject: genirq: Provide compat handling for chip->set_affinity() Wrap the old chip function set_affinity() until the migration is complete and the old chip functions are removed. Signed-off-by: Thomas Gleixner Cc: Peter Zijlstra LKML-Reference: <20100927121842.732894108@linutronix.de> Reviewed-by: H. Peter Anvin Reviewed-by: Ingo Molnar --- kernel/irq/chip.c | 8 ++++++++ kernel/irq/manage.c | 11 ++++++----- kernel/irq/migration.c | 6 ++++-- kernel/irq/proc.c | 2 +- 4 files changed, 19 insertions(+), 8 deletions(-) (limited to 'kernel') diff --git a/kernel/irq/chip.c b/kernel/irq/chip.c index cce85f0734b0..df51792d9fd3 100644 --- a/kernel/irq/chip.c +++ b/kernel/irq/chip.c @@ -370,6 +370,12 @@ static unsigned int compat_irq_startup(struct irq_data *data) return data->chip->startup(data->irq); } +static int compat_irq_set_affinity(struct irq_data *data, + const struct cpumask *dest, bool force) +{ + return data->chip->set_affinity(data->irq, dest); +} + static void compat_bus_lock(struct irq_data *data) { data->chip->bus_lock(data->irq); @@ -436,6 +442,8 @@ void irq_chip_set_defaults(struct irq_chip *chip) chip->irq_mask_ack = compat_irq_mask_ack; if (chip->eoi) chip->irq_eoi = compat_irq_eoi; + if (chip->set_affinity) + chip->irq_set_affinity = compat_irq_set_affinity; } static inline void mask_ack_irq(struct irq_desc *desc) diff --git a/kernel/irq/manage.c b/kernel/irq/manage.c index 31d7678e0269..305a60ff756b 100644 --- a/kernel/irq/manage.c +++ b/kernel/irq/manage.c @@ -74,7 +74,7 @@ int irq_can_set_affinity(unsigned int irq) struct irq_desc *desc = irq_to_desc(irq); if (CHECK_IRQ_PER_CPU(desc->status) || !desc->irq_data.chip || - !desc->irq_data.chip->set_affinity) + !desc->irq_data.chip->irq_set_affinity) return 0; return 1; @@ -109,16 +109,17 @@ void irq_set_thread_affinity(struct irq_desc *desc) int irq_set_affinity(unsigned int irq, const struct cpumask *cpumask) { struct irq_desc *desc = irq_to_desc(irq); + struct irq_chip *chip = desc->irq_data.chip; unsigned long flags; - if (!desc->irq_data.chip->set_affinity) + if (!chip->irq_set_affinity) return -EINVAL; raw_spin_lock_irqsave(&desc->lock, flags); #ifdef CONFIG_GENERIC_PENDING_IRQ if (desc->status & IRQ_MOVE_PCNTXT) { - if (!desc->irq_data.chip->set_affinity(irq, cpumask)) { + if (!chip->irq_set_affinity(&desc->irq_data, cpumask, false)) { cpumask_copy(desc->irq_data.affinity, cpumask); irq_set_thread_affinity(desc); } @@ -128,7 +129,7 @@ int irq_set_affinity(unsigned int irq, const struct cpumask *cpumask) cpumask_copy(desc->pending_mask, cpumask); } #else - if (!desc->irq_data.chip->set_affinity(irq, cpumask)) { + if (!chip->irq_set_affinity(&desc->irq_data, cpumask, false)) { cpumask_copy(desc->irq_data.affinity, cpumask); irq_set_thread_affinity(desc); } @@ -177,7 +178,7 @@ static int setup_affinity(unsigned int irq, struct irq_desc *desc) cpumask_and(desc->irq_data.affinity, cpu_online_mask, irq_default_affinity); set_affinity: - desc->irq_data.chip->set_affinity(irq, desc->irq_data.affinity); + desc->irq_data.chip->irq_set_affinity(&desc->irq_data, desc->irq_data.affinity, false); return 0; } diff --git a/kernel/irq/migration.c b/kernel/irq/migration.c index 7888e5d5575a..1d2541940480 100644 --- a/kernel/irq/migration.c +++ b/kernel/irq/migration.c @@ -7,6 +7,7 @@ void move_masked_irq(int irq) { struct irq_desc *desc = irq_to_desc(irq); + struct irq_chip *chip = desc->irq_data.chip; if (likely(!(desc->status & IRQ_MOVE_PENDING))) return; @@ -24,7 +25,7 @@ void move_masked_irq(int irq) if (unlikely(cpumask_empty(desc->pending_mask))) return; - if (!desc->irq_data.chip->set_affinity) + if (!chip->irq_set_affinity) return; assert_raw_spin_locked(&desc->lock); @@ -43,7 +44,8 @@ void move_masked_irq(int irq) */ if (likely(cpumask_any_and(desc->pending_mask, cpu_online_mask) < nr_cpu_ids)) - if (!desc->irq_data.chip->set_affinity(irq, desc->pending_mask)) { + if (!chip->irq_set_affinity(&desc->irq_data, + desc->pending_mask, false)) { cpumask_copy(desc->irq_data.affinity, desc->pending_mask); irq_set_thread_affinity(desc); } diff --git a/kernel/irq/proc.c b/kernel/irq/proc.c index 9b0da94b5b2b..d9fddf918b41 100644 --- a/kernel/irq/proc.c +++ b/kernel/irq/proc.c @@ -65,7 +65,7 @@ static ssize_t irq_affinity_proc_write(struct file *file, cpumask_var_t new_value; int err; - if (!irq_to_desc(irq)->irq_data.chip->set_affinity || no_irq_affinity || + if (!irq_to_desc(irq)->irq_data.chip->irq_set_affinity || no_irq_affinity || irq_balancing_disabled(irq)) return -EIO; -- cgit v1.2.3 From b2ba2c30033c10cca2454f8b44bf98f5249e61c6 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Mon, 27 Sep 2010 12:45:47 +0000 Subject: genirq: Provide compat handling for chip->set_type() Wrap the old chip function set_type() until the migration is complete and the old chip functions are removed. Signed-off-by: Thomas Gleixner Cc: Peter Zijlstra LKML-Reference: <20100927121842.832261548@linutronix.de> Reviewed-by: H. Peter Anvin Reviewed-by: Ingo Molnar --- kernel/irq/autoprobe.c | 5 +++-- kernel/irq/chip.c | 7 +++++++ kernel/irq/manage.c | 10 +++++----- 3 files changed, 15 insertions(+), 7 deletions(-) (limited to 'kernel') diff --git a/kernel/irq/autoprobe.c b/kernel/irq/autoprobe.c index 7a468254e533..505798f86c36 100644 --- a/kernel/irq/autoprobe.c +++ b/kernel/irq/autoprobe.c @@ -57,8 +57,9 @@ unsigned long probe_irq_on(void) * Some chips need to know about probing in * progress: */ - if (desc->irq_data.chip->set_type) - desc->irq_data.chip->set_type(i, IRQ_TYPE_PROBE); + if (desc->irq_data.chip->irq_set_type) + desc->irq_data.chip->irq_set_type(&desc->irq_data, + IRQ_TYPE_PROBE); desc->irq_data.chip->irq_startup(&desc->irq_data); } raw_spin_unlock_irq(&desc->lock); diff --git a/kernel/irq/chip.c b/kernel/irq/chip.c index df51792d9fd3..b7dd02a99c80 100644 --- a/kernel/irq/chip.c +++ b/kernel/irq/chip.c @@ -376,6 +376,11 @@ static int compat_irq_set_affinity(struct irq_data *data, return data->chip->set_affinity(data->irq, dest); } +static int compat_irq_set_type(struct irq_data *data, unsigned int type) +{ + return data->chip->set_type(data->irq, type); +} + static void compat_bus_lock(struct irq_data *data) { data->chip->bus_lock(data->irq); @@ -444,6 +449,8 @@ void irq_chip_set_defaults(struct irq_chip *chip) chip->irq_eoi = compat_irq_eoi; if (chip->set_affinity) chip->irq_set_affinity = compat_irq_set_affinity; + if (chip->set_type) + chip->irq_set_type = compat_irq_set_type; } static inline void mask_ack_irq(struct irq_desc *desc) diff --git a/kernel/irq/manage.c b/kernel/irq/manage.c index 305a60ff756b..3618362b3d8d 100644 --- a/kernel/irq/manage.c +++ b/kernel/irq/manage.c @@ -430,12 +430,12 @@ void compat_irq_chip_set_default_handler(struct irq_desc *desc) } int __irq_set_trigger(struct irq_desc *desc, unsigned int irq, - unsigned long flags) + unsigned long flags) { int ret; struct irq_chip *chip = desc->irq_data.chip; - if (!chip || !chip->set_type) { + if (!chip || !chip->irq_set_type) { /* * IRQF_TRIGGER_* but the PIC does not support multiple * flow-types? @@ -446,11 +446,11 @@ int __irq_set_trigger(struct irq_desc *desc, unsigned int irq, } /* caller masked out all except trigger mode flags */ - ret = chip->set_type(irq, flags); + ret = chip->irq_set_type(&desc->irq_data, flags); if (ret) - pr_err("setting trigger mode %d for irq %u failed (%pF)\n", - (int)flags, irq, chip->set_type); + pr_err("setting trigger mode %lu for irq %u failed (%pF)\n", + flags, irq, chip->irq_set_type); else { if (flags & (IRQ_TYPE_LEVEL_LOW | IRQ_TYPE_LEVEL_HIGH)) flags |= IRQ_LEVEL; -- cgit v1.2.3 From 2f7e99bb9be6a2d8d7b808dc86037710cc8b7bf1 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Mon, 27 Sep 2010 12:45:50 +0000 Subject: genirq: Provide compat handling for chip->set_wake() Wrap the old chip function set_wake() until the migration is complete and the old chip functions are removed. Signed-off-by: Thomas Gleixner Cc: Peter Zijlstra LKML-Reference: <20100927121842.927527393@linutronix.de> Reviewed-by: H. Peter Anvin Reviewed-by: Ingo Molnar --- kernel/irq/chip.c | 7 +++++++ kernel/irq/manage.c | 4 ++-- 2 files changed, 9 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/irq/chip.c b/kernel/irq/chip.c index b7dd02a99c80..8775dd39ab3d 100644 --- a/kernel/irq/chip.c +++ b/kernel/irq/chip.c @@ -381,6 +381,11 @@ static int compat_irq_set_type(struct irq_data *data, unsigned int type) return data->chip->set_type(data->irq, type); } +static int compat_irq_set_wake(struct irq_data *data, unsigned int on) +{ + return data->chip->set_wake(data->irq, on); +} + static void compat_bus_lock(struct irq_data *data) { data->chip->bus_lock(data->irq); @@ -451,6 +456,8 @@ void irq_chip_set_defaults(struct irq_chip *chip) chip->irq_set_affinity = compat_irq_set_affinity; if (chip->set_type) chip->irq_set_type = compat_irq_set_type; + if (chip->set_wake) + chip->irq_set_wake = compat_irq_set_wake; } static inline void mask_ack_irq(struct irq_desc *desc) diff --git a/kernel/irq/manage.c b/kernel/irq/manage.c index 3618362b3d8d..644e8d5fa367 100644 --- a/kernel/irq/manage.c +++ b/kernel/irq/manage.c @@ -337,8 +337,8 @@ static int set_irq_wake_real(unsigned int irq, unsigned int on) struct irq_desc *desc = irq_to_desc(irq); int ret = -ENXIO; - if (desc->irq_data.chip->set_wake) - ret = desc->irq_data.chip->set_wake(irq, on); + if (desc->irq_data.chip->irq_set_wake) + ret = desc->irq_data.chip->irq_set_wake(&desc->irq_data, on); return ret; } -- cgit v1.2.3 From 21e2b8c62cca8f7dbec0c8c131ca1637e4a5670f Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Mon, 27 Sep 2010 12:45:53 +0000 Subject: genirq: Provide compat handling for chip->retrigger() Wrap the old chip function retrigger() until the migration is complete and the old chip functions are removed. Signed-off-by: Thomas Gleixner Cc: Peter Zijlstra LKML-Reference: <20100927121843.025801092@linutronix.de> Reviewed-by: H. Peter Anvin Reviewed-by: Ingo Molnar --- kernel/irq/chip.c | 7 +++++++ kernel/irq/resend.c | 4 ++-- 2 files changed, 9 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/irq/chip.c b/kernel/irq/chip.c index 8775dd39ab3d..f2c4d28c508a 100644 --- a/kernel/irq/chip.c +++ b/kernel/irq/chip.c @@ -386,6 +386,11 @@ static int compat_irq_set_wake(struct irq_data *data, unsigned int on) return data->chip->set_wake(data->irq, on); } +static int compat_irq_retrigger(struct irq_data *data) +{ + return data->chip->retrigger(data->irq); +} + static void compat_bus_lock(struct irq_data *data) { data->chip->bus_lock(data->irq); @@ -458,6 +463,8 @@ void irq_chip_set_defaults(struct irq_chip *chip) chip->irq_set_type = compat_irq_set_type; if (chip->set_wake) chip->irq_set_wake = compat_irq_set_wake; + if (chip->retrigger) + chip->irq_retrigger = compat_irq_retrigger; } static inline void mask_ack_irq(struct irq_desc *desc) diff --git a/kernel/irq/resend.c b/kernel/irq/resend.c index a798a2328f8a..891115a929aa 100644 --- a/kernel/irq/resend.c +++ b/kernel/irq/resend.c @@ -70,8 +70,8 @@ void check_irq_resend(struct irq_desc *desc, unsigned int irq) if ((status & (IRQ_LEVEL | IRQ_PENDING | IRQ_REPLAY)) == IRQ_PENDING) { desc->status = (status & ~IRQ_PENDING) | IRQ_REPLAY; - if (!desc->irq_data.chip->retrigger || - !desc->irq_data.chip->retrigger(irq)) { + if (!desc->irq_data.chip->irq_retrigger || + !desc->irq_data.chip->irq_retrigger(&desc->irq_data)) { #ifdef CONFIG_HARDIRQS_SW_RESEND /* Set it pending and activate the softirq: */ set_bit(irq, irqs_resend); -- cgit v1.2.3 From 540804b5c52065a87d826f7714b18a3ec0b269f9 Mon Sep 17 00:00:00 2001 From: Stephane Eranian Date: Mon, 4 Oct 2010 12:00:02 +0200 Subject: perf_events: Fix invalid pointer when pid is invalid This patch fixes an error in perf_event_open() when the pid provided by the user is invalid. find_lively_task_by_vpid() does not return NULL on error but an error code. Without the fix the error code was silently passed to find_get_context() which would eventually cause a invalid pointer dereference. Signed-off-by: Stephane Eranian Cc: peterz@infradead.org Cc: paulus@samba.org Cc: davem@davemloft.net Cc: fweisbec@gmail.com Cc: perfmon2-devel@lists.sf.net Cc: eranian@gmail.com Cc: robert.richter@amd.com LKML-Reference: <4ca9a5d1.e8e9d80a.3dbb.ffff8f2e@mx.google.com> Signed-off-by: Ingo Molnar --- kernel/perf_event.c | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index c16158c77dfd..64507eaa2d9e 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -5616,8 +5616,13 @@ SYSCALL_DEFINE5(perf_event_open, } } - if (pid != -1) + if (pid != -1) { task = find_lively_task_by_vpid(pid); + if (IS_ERR(task)) { + err = PTR_ERR(task); + goto err_group_fd; + } + } /* * Get the target context (task or percpu): -- cgit v1.2.3 From bd151412263a67b5321e9dd1d5b4bf6d96fdebf3 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Fri, 1 Oct 2010 15:17:14 +0200 Subject: genirq: Provide config option to disable deprecated code This option covers now the old chip functions and the irq_desc data fields which are moving to struct irq_data. More stuff will follow. Pretty handy for testing a conversion, whether something broke or not. Signed-off-by: Thomas Gleixner Reviewed-by: Ingo Molnar --- kernel/irq/Kconfig | 4 ++++ kernel/irq/chip.c | 8 +++++++- kernel/irq/handle.c | 9 +++++++-- kernel/irq/internals.h | 10 ++++++++++ kernel/irq/spurious.c | 6 ++++-- 5 files changed, 32 insertions(+), 5 deletions(-) (limited to 'kernel') diff --git a/kernel/irq/Kconfig b/kernel/irq/Kconfig index e0fc6cd78aa0..a42c0191d71a 100644 --- a/kernel/irq/Kconfig +++ b/kernel/irq/Kconfig @@ -12,6 +12,10 @@ config GENERIC_HARDIRQS config GENERIC_HARDIRQS_NO__DO_IRQ def_bool y +# Select this to disable the deprecated stuff +config GENERIC_HARDIRQS_NO_DEPRECATED + def_bool n + # Options selectable by the architecture code config HAVE_SPARSE_IRQ def_bool n diff --git a/kernel/irq/chip.c b/kernel/irq/chip.c index f2c4d28c508a..323547983f15 100644 --- a/kernel/irq/chip.c +++ b/kernel/irq/chip.c @@ -324,6 +324,7 @@ static void default_shutdown(struct irq_data *data) desc->status |= IRQ_MASKED; } +#ifndef CONFIG_GENERIC_HARDIRQS_NO_DEPRECATED /* Temporary migration helpers */ static void compat_irq_mask(struct irq_data *data) { @@ -400,12 +401,14 @@ static void compat_bus_sync_unlock(struct irq_data *data) { data->chip->bus_sync_unlock(data->irq); } +#endif /* * Fixup enable/disable function pointers */ void irq_chip_set_defaults(struct irq_chip *chip) { +#ifndef CONFIG_GENERIC_HARDIRQS_NO_DEPRECATED /* * Compat fixup functions need to be before we set the * defaults for enable/disable/startup/shutdown @@ -418,7 +421,7 @@ void irq_chip_set_defaults(struct irq_chip *chip) chip->irq_shutdown = compat_irq_shutdown; if (chip->startup) chip->irq_startup = compat_irq_startup; - +#endif /* * The real defaults */ @@ -437,6 +440,8 @@ void irq_chip_set_defaults(struct irq_chip *chip) if (!chip->irq_shutdown) chip->irq_shutdown = chip->irq_disable != default_disable ? chip->irq_disable : default_shutdown; + +#ifndef CONFIG_GENERIC_HARDIRQS_NO_DEPRECATED if (!chip->end) chip->end = dummy_irq_chip.end; @@ -465,6 +470,7 @@ void irq_chip_set_defaults(struct irq_chip *chip) chip->irq_set_wake = compat_irq_set_wake; if (chip->retrigger) chip->irq_retrigger = compat_irq_retrigger; +#endif } static inline void mask_ack_irq(struct irq_desc *desc) diff --git a/kernel/irq/handle.c b/kernel/irq/handle.c index 8d0697f892a2..3fcef37154a1 100644 --- a/kernel/irq/handle.c +++ b/kernel/irq/handle.c @@ -309,7 +309,12 @@ static unsigned int noop_ret(struct irq_data *data) return 0; } +#ifndef CONFIG_GENERIC_HARDIRQS_NO_DEPRECATED static void compat_noop(unsigned int irq) { } +#define END_INIT .end = compat_noop +#else +#define END_INIT +#endif /* * Generic no controller implementation @@ -321,7 +326,7 @@ struct irq_chip no_irq_chip = { .irq_enable = noop, .irq_disable = noop, .irq_ack = ack_bad, - .end = compat_noop, + END_INIT }; /* @@ -337,7 +342,7 @@ struct irq_chip dummy_irq_chip = { .irq_ack = noop, .irq_mask = noop, .irq_unmask = noop, - .end = compat_noop, + END_INIT }; /* diff --git a/kernel/irq/internals.h b/kernel/irq/internals.h index ecafbfee5b12..b905f0ab1bb2 100644 --- a/kernel/irq/internals.h +++ b/kernel/irq/internals.h @@ -42,6 +42,16 @@ extern int irq_select_affinity_usr(unsigned int irq); extern void irq_set_thread_affinity(struct irq_desc *desc); +#ifndef CONFIG_GENERIC_HARDIRQS_NO_DEPRECATED +static inline void irq_end(unsigned int irq, struct irq_desc *desc) +{ + if (desc->irq_data.chip && desc->irq_data.chip->end) + desc->irq_data.chip->end(irq); +} +#else +static inline void irq_end(unsigned int irq, struct irq_desc *desc) { } +#endif + /* Inline functions for support of irq chips on slow busses */ static inline void chip_bus_lock(struct irq_desc *desc) { diff --git a/kernel/irq/spurious.c b/kernel/irq/spurious.c index 9ee704d3a23c..3089d3b9d5f3 100644 --- a/kernel/irq/spurious.c +++ b/kernel/irq/spurious.c @@ -14,6 +14,8 @@ #include #include +#include "internals.h" + static int irqfixup __read_mostly; #define POLL_SPURIOUS_IRQ_INTERVAL (HZ/10) @@ -78,8 +80,8 @@ static int try_one_irq(int irq, struct irq_desc *desc) * If we did actual work for the real IRQ line we must let the * IRQ controller clean up too */ - if (work && desc->irq_data.chip && desc->irq_data.chip->end) - desc->irq_data.chip->end(irq); + if (work) + irq_end(irq, desc); raw_spin_unlock(&desc->lock); return ok; -- cgit v1.2.3 From db71922217a214e5c9268448e537b54fc1f301ea Mon Sep 17 00:00:00 2001 From: Jan Blunck Date: Sun, 15 Aug 2010 22:51:10 +0200 Subject: BKL: Explicitly add BKL around get_sb/fill_super This patch is a preparation necessary to remove the BKL from do_new_mount(). It explicitly adds calls to lock_kernel()/unlock_kernel() around get_sb/fill_super operations for filesystems that still uses the BKL. I've read through all the code formerly covered by the BKL inside do_kern_mount() and have satisfied myself that it doesn't need the BKL any more. do_kern_mount() is already called without the BKL when mounting the rootfs and in nfsctl. do_kern_mount() calls vfs_kern_mount(), which is called from various places without BKL: simple_pin_fs(), nfs_do_clone_mount() through nfs_follow_mountpoint(), afs_mntpt_do_automount() through afs_mntpt_follow_link(). Both later functions are actually the filesystems follow_link inode operation. vfs_kern_mount() is calling the specified get_sb function and lets the filesystem do its job by calling the given fill_super function. Therefore I think it is safe to push down the BKL from the VFS to the low-level filesystems get_sb/fill_super operation. [arnd: do not add the BKL to those file systems that already don't use it elsewhere] Signed-off-by: Jan Blunck Signed-off-by: Arnd Bergmann Cc: Matthew Wilcox Cc: Christoph Hellwig --- kernel/cgroup.c | 4 ++++ 1 file changed, 4 insertions(+) (limited to 'kernel') diff --git a/kernel/cgroup.c b/kernel/cgroup.c index c9483d8f6140..a7ba3bccadc5 100644 --- a/kernel/cgroup.c +++ b/kernel/cgroup.c @@ -1430,6 +1430,8 @@ static int cgroup_get_sb(struct file_system_type *fs_type, struct super_block *sb; struct cgroupfs_root *new_root; + lock_kernel(); + /* First find the desired set of subsystems */ mutex_lock(&cgroup_mutex); ret = parse_cgroupfs_options(data, &opts); @@ -1559,6 +1561,7 @@ static int cgroup_get_sb(struct file_system_type *fs_type, simple_set_mnt(mnt, sb); kfree(opts.release_agent); kfree(opts.name); + unlock_kernel(); return 0; drop_new_super: @@ -1568,6 +1571,7 @@ static int cgroup_get_sb(struct file_system_type *fs_type, out_err: kfree(opts.release_agent); kfree(opts.name); + unlock_kernel(); return ret; } -- cgit v1.2.3 From 38d018dba3f725b969f196550d92a6ec1c092428 Mon Sep 17 00:00:00 2001 From: Jan Blunck Date: Wed, 24 Feb 2010 13:25:34 +0100 Subject: BKL: Remove BKL from cgroup The BKL is only used in remount_fs and get_sb that are both protected by the superblocks s_umount rw_semaphore. Therefore it is safe to remove the BKL entirely. Signed-off-by: Jan Blunck Signed-off-by: Arnd Bergmann --- kernel/cgroup.c | 8 -------- 1 file changed, 8 deletions(-) (limited to 'kernel') diff --git a/kernel/cgroup.c b/kernel/cgroup.c index a7ba3bccadc5..304d27759949 100644 --- a/kernel/cgroup.c +++ b/kernel/cgroup.c @@ -52,7 +52,6 @@ #include #include #include -#include #include #include #include /* TODO: replace with more sophisticated array */ @@ -1222,7 +1221,6 @@ static int cgroup_remount(struct super_block *sb, int *flags, char *data) struct cgroup *cgrp = &root->top_cgroup; struct cgroup_sb_opts opts; - lock_kernel(); mutex_lock(&cgrp->dentry->d_inode->i_mutex); mutex_lock(&cgroup_mutex); @@ -1255,7 +1253,6 @@ static int cgroup_remount(struct super_block *sb, int *flags, char *data) kfree(opts.name); mutex_unlock(&cgroup_mutex); mutex_unlock(&cgrp->dentry->d_inode->i_mutex); - unlock_kernel(); return ret; } @@ -1430,8 +1427,6 @@ static int cgroup_get_sb(struct file_system_type *fs_type, struct super_block *sb; struct cgroupfs_root *new_root; - lock_kernel(); - /* First find the desired set of subsystems */ mutex_lock(&cgroup_mutex); ret = parse_cgroupfs_options(data, &opts); @@ -1561,7 +1556,6 @@ static int cgroup_get_sb(struct file_system_type *fs_type, simple_set_mnt(mnt, sb); kfree(opts.release_agent); kfree(opts.name); - unlock_kernel(); return 0; drop_new_super: @@ -1571,8 +1565,6 @@ static int cgroup_get_sb(struct file_system_type *fs_type, out_err: kfree(opts.release_agent); kfree(opts.name); - unlock_kernel(); - return ret; } -- cgit v1.2.3 From 97bd234701b2b39a0e749c1fe0e44f1d14c94292 Mon Sep 17 00:00:00 2001 From: Tejun Heo Date: Tue, 5 Oct 2010 10:41:14 +0200 Subject: workqueue: prepare for more tracepoints Define workqueue_work event class and use it for workqueue_execute_end trace point. Also, move trace/events/workqueue.h include downwards such that all struct definitions are visible to it. This is to prepare for more tracepoints and doesn't cause any functional change. Signed-off-by: Tejun Heo Cc: Frederic Weisbecker --- kernel/workqueue.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/workqueue.c b/kernel/workqueue.c index 19e4bc15ee99..026f778e879b 100644 --- a/kernel/workqueue.c +++ b/kernel/workqueue.c @@ -42,9 +42,6 @@ #include #include -#define CREATE_TRACE_POINTS -#include - #include "workqueue_sched.h" enum { @@ -257,6 +254,9 @@ EXPORT_SYMBOL_GPL(system_long_wq); EXPORT_SYMBOL_GPL(system_nrt_wq); EXPORT_SYMBOL_GPL(system_unbound_wq); +#define CREATE_TRACE_POINTS +#include + #define for_each_busy_worker(worker, i, pos, gcwq) \ for (i = 0; i < BUSY_WORKER_HASH_SIZE; i++) \ hlist_for_each_entry(worker, pos, &gcwq->busy_hash[i], hentry) -- cgit v1.2.3 From cdadf0097cdca06c497ffaeb5982e028c6e4ed38 Mon Sep 17 00:00:00 2001 From: Tejun Heo Date: Tue, 5 Oct 2010 10:49:55 +0200 Subject: workqueue: add queue_work and activate_work trace points These two tracepoints allow tracking when and how a work is queued and activated. This patch is based on Frederic's patch to add queue_work trace point. Signed-off-by: Tejun Heo Cc: Frederic Weisbecker --- kernel/workqueue.c | 3 +++ 1 file changed, 3 insertions(+) (limited to 'kernel') diff --git a/kernel/workqueue.c b/kernel/workqueue.c index 026f778e879b..cb2ccfbed0c6 100644 --- a/kernel/workqueue.c +++ b/kernel/workqueue.c @@ -997,6 +997,7 @@ static void __queue_work(unsigned int cpu, struct workqueue_struct *wq, /* gcwq determined, get cwq and queue */ cwq = get_cwq(gcwq->cpu, wq); + trace_workqueue_queue_work(cpu, cwq, work); BUG_ON(!list_empty(&work->entry)); @@ -1004,6 +1005,7 @@ static void __queue_work(unsigned int cpu, struct workqueue_struct *wq, work_flags = work_color_to_flags(cwq->work_color); if (likely(cwq->nr_active < cwq->max_active)) { + trace_workqueue_activate_work(work); cwq->nr_active++; worklist = gcwq_determine_ins_pos(gcwq, cwq); } else { @@ -1679,6 +1681,7 @@ static void cwq_activate_first_delayed(struct cpu_workqueue_struct *cwq) struct work_struct, entry); struct list_head *pos = gcwq_determine_ins_pos(cwq->gcwq, cwq); + trace_workqueue_activate_work(work); move_linked_works(work, pos, NULL); __clear_bit(WORK_STRUCT_DELAYED_BIT, work_data_bits(work)); cwq->nr_active++; -- cgit v1.2.3 From 5336377d6225959624146629ce3fc88ee8ecda3d Mon Sep 17 00:00:00 2001 From: Linus Torvalds Date: Tue, 5 Oct 2010 11:29:27 -0700 Subject: modules: Fix module_bug_list list corruption race With all the recent module loading cleanups, we've minimized the code that sits under module_mutex, fixing various deadlocks and making it possible to do most of the module loading in parallel. However, that whole conversion totally missed the rather obscure code that adds a new module to the list for BUG() handling. That code was doubly obscure because (a) the code itself lives in lib/bugs.c (for dubious reasons) and (b) it gets called from the architecture-specific "module_finalize()" rather than from generic code. Calling it from arch-specific code makes no sense what-so-ever to begin with, and is now actively wrong since that code isn't protected by the module loading lock any more. So this commit moves the "module_bug_{finalize,cleanup}()" calls away from the arch-specific code, and into the generic code - and in the process protects it with the module_mutex so that the list operations are now safe. Future fixups: - move the module list handling code into kernel/module.c where it belongs. - get rid of 'module_bug_list' and just use the regular list of modules (called 'modules' - imagine that) that we already create and maintain for other reasons. Reported-and-tested-by: Thomas Gleixner Cc: Rusty Russell Cc: Adrian Bunk Cc: Andrew Morton Cc: stable@kernel.org Signed-off-by: Linus Torvalds --- kernel/module.c | 4 ++++ 1 file changed, 4 insertions(+) (limited to 'kernel') diff --git a/kernel/module.c b/kernel/module.c index d0b5f8db11b4..ccd641991842 100644 --- a/kernel/module.c +++ b/kernel/module.c @@ -1537,6 +1537,7 @@ static int __unlink_module(void *_mod) { struct module *mod = _mod; list_del(&mod->list); + module_bug_cleanup(mod); return 0; } @@ -2625,6 +2626,7 @@ static struct module *load_module(void __user *umod, if (err < 0) goto ddebug; + module_bug_finalize(info.hdr, info.sechdrs, mod); list_add_rcu(&mod->list, &modules); mutex_unlock(&module_mutex); @@ -2650,6 +2652,8 @@ static struct module *load_module(void __user *umod, mutex_lock(&module_mutex); /* Unlink carefully: kallsyms could be walking list. */ list_del_rcu(&mod->list); + module_bug_cleanup(mod); + ddebug: if (!mod->taints) dynamic_debug_remove(info.debug); -- cgit v1.2.3 From 773e3f93577ffb493fb7c39b1a6ecf39b5748e87 Mon Sep 17 00:00:00 2001 From: "Paul E. McKenney" Date: Tue, 5 Oct 2010 14:03:02 -0700 Subject: rcu: move check from rcu_dereference_bh to rcu_read_lock_bh_held As suggested by Linus, push the irqs_disabled() down to the rcu_read_lock_bh_held() level so that all callers get the benefit of the correct check. Signed-off-by: Paul E. McKenney --- kernel/rcupdate.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/rcupdate.c b/kernel/rcupdate.c index 4d169835fb36..0af1dc70fece 100644 --- a/kernel/rcupdate.c +++ b/kernel/rcupdate.c @@ -86,7 +86,7 @@ int rcu_read_lock_bh_held(void) { if (!debug_lockdep_rcu_enabled()) return 1; - return in_softirq(); + return in_softirq() || irqs_disabled(); } EXPORT_SYMBOL_GPL(rcu_read_lock_bh_held); -- cgit v1.2.3 From a337fdac7a5622d1e6547f4b476c14dfe5a2c892 Mon Sep 17 00:00:00 2001 From: Andi Kleen Date: Mon, 27 Sep 2010 20:32:19 +0200 Subject: HWPOISON: Copy si_addr_lsb to user The original hwpoison code added a new siginfo field si_addr_lsb to pass the granuality of the fault address to user space. Unfortunately this field was never copied to user space. Fix this here. I added explicit checks for the MCEERR codes to avoid having to patch all potential callers to initialize the field. Signed-off-by: Andi Kleen --- kernel/signal.c | 8 ++++++++ 1 file changed, 8 insertions(+) (limited to 'kernel') diff --git a/kernel/signal.c b/kernel/signal.c index bded65187780..919562c3d6b7 100644 --- a/kernel/signal.c +++ b/kernel/signal.c @@ -2214,6 +2214,14 @@ int copy_siginfo_to_user(siginfo_t __user *to, siginfo_t *from) err |= __put_user(from->si_addr, &to->si_addr); #ifdef __ARCH_SI_TRAPNO err |= __put_user(from->si_trapno, &to->si_trapno); +#endif +#ifdef BUS_MCEERR_AO + /* + * Other callers might not initialize the si_lsb field, + * so check explicitely for the right codes here. + */ + if (from->si_code == BUS_MCEERR_AR || from->si_code == BUS_MCEERR_AO) + err |= __put_user(from->si_addr_lsb, &to->si_addr_lsb); #endif break; case __SI_CHLD: -- cgit v1.2.3 From b0a0f667a349247bd7f05f806b662a25653822bc Mon Sep 17 00:00:00 2001 From: "Paul E. McKenney" Date: Wed, 6 Oct 2010 17:32:51 -0700 Subject: sched: suppress RCU lockdep splat in task_fork_fair > =================================================== > [ INFO: suspicious rcu_dereference_check() usage. ] > --------------------------------------------------- > /home/greearb/git/linux.wireless-testing/kernel/sched.c:618 invoked rcu_dereference_check() without protection! > > other info that might help us debug this: > > rcu_scheduler_active = 1, debug_locks = 1 > 1 lock held by ifup/23517: > #0: (&rq->lock){-.-.-.}, at: [] task_fork_fair+0x3b/0x108 > > stack backtrace: > Pid: 23517, comm: ifup Not tainted 2.6.36-rc6-wl+ #5 > Call Trace: > [] ? printk+0xf/0x16 > [] lockdep_rcu_dereference+0x74/0x7d > [] task_group+0x6d/0x79 > [] set_task_rq+0xe/0x57 > [] task_fork_fair+0x57/0x108 > [] sched_fork+0x82/0xf9 > [] copy_process+0x569/0xe8e > [] do_fork+0x118/0x262 > [] ? do_page_fault+0x16a/0x2cf > [] ? up_read+0x16/0x2a > [] sys_clone+0x1b/0x20 > [] ptregs_clone+0x15/0x30 > [] ? sysenter_do_call+0x12/0x38 Here a newly created task is having its runqueue assigned. The new task is not yet on the tasklist, so cannot go away. This is therefore a false positive, suppress with an RCU read-side critical section. Reported-by: Ben Greear Tested-by: Ben Greear Date: Tue, 28 Sep 2010 16:32:43 +0800 Subject: rcu: using ACCESS_ONCE() to observe the jiffies_stall/rnp->qsmask value Using ACCESS_ONCE() to observe the jiffies_stall/rnp->qsmask value due to the caller didn't hold the root_rcu/rnp node's lock. Although use without ACCESS_ONCE() is safe due to the value loaded being used but once, the ACCESS_ONCE() is a good documentation aid -- the variables are being loaded without the services of a lock. Signed-off-by: Dongdong Deng CC: Dipankar Sarma CC: Paul E. McKenney Signed-off-by: Paul E. McKenney --- kernel/rcutree.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/rcutree.c b/kernel/rcutree.c index e75073504a31..ccdc04c47981 100644 --- a/kernel/rcutree.c +++ b/kernel/rcutree.c @@ -545,9 +545,9 @@ static void check_cpu_stall(struct rcu_state *rsp, struct rcu_data *rdp) if (rcu_cpu_stall_suppress) return; - delta = jiffies - rsp->jiffies_stall; + delta = jiffies - ACCESS_ONCE(rsp->jiffies_stall); rnp = rdp->mynode; - if ((rnp->qsmask & rdp->grpmask) && delta >= 0) { + if ((ACCESS_ONCE(rnp->qsmask) & rdp->grpmask) && delta >= 0) { /* We haven't checked in, so go dump stack. */ print_cpu_stall(rsp); -- cgit v1.2.3 From 6506cf6ce68d78a5470a8360c965dafe8e4b78e3 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Thu, 16 Sep 2010 17:50:31 +0200 Subject: sched: fix RCU lockdep splat from task_group() This addresses the following RCU lockdep splat: [0.051203] CPU0: AMD QEMU Virtual CPU version 0.12.4 stepping 03 [0.052999] lockdep: fixing up alternatives. [0.054105] [0.054106] =================================================== [0.054999] [ INFO: suspicious rcu_dereference_check() usage. ] [0.054999] --------------------------------------------------- [0.054999] kernel/sched.c:616 invoked rcu_dereference_check() without protection! [0.054999] [0.054999] other info that might help us debug this: [0.054999] [0.054999] [0.054999] rcu_scheduler_active = 1, debug_locks = 1 [0.054999] 3 locks held by swapper/1: [0.054999] #0: (cpu_add_remove_lock){+.+.+.}, at: [] cpu_up+0x42/0x6a [0.054999] #1: (cpu_hotplug.lock){+.+.+.}, at: [] cpu_hotplug_begin+0x2a/0x51 [0.054999] #2: (&rq->lock){-.-...}, at: [] init_idle+0x2f/0x113 [0.054999] [0.054999] stack backtrace: [0.054999] Pid: 1, comm: swapper Not tainted 2.6.35 #1 [0.054999] Call Trace: [0.054999] [] lockdep_rcu_dereference+0x9b/0xa3 [0.054999] [] task_group+0x7b/0x8a [0.054999] [] set_task_rq+0x13/0x40 [0.054999] [] init_idle+0xd2/0x113 [0.054999] [] fork_idle+0xb8/0xc7 [0.054999] [] ? mark_held_locks+0x4d/0x6b [0.054999] [] do_fork_idle+0x17/0x2b [0.054999] [] native_cpu_up+0x1c1/0x724 [0.054999] [] ? do_fork_idle+0x0/0x2b [0.054999] [] _cpu_up+0xac/0x127 [0.054999] [] cpu_up+0x55/0x6a [0.054999] [] kernel_init+0xe1/0x1ff [0.054999] [] kernel_thread_helper+0x4/0x10 [0.054999] [] ? restore_args+0x0/0x30 [0.054999] [] ? kernel_init+0x0/0x1ff [0.054999] [] ? kernel_thread_helper+0x0/0x10 [0.056074] Booting Node 0, Processors #1lockdep: fixing up alternatives. [0.130045] #2lockdep: fixing up alternatives. [0.203089] #3 Ok. [0.275286] Brought up 4 CPUs [0.276005] Total of 4 processors activated (16017.17 BogoMIPS). The cgroup_subsys_state structures referenced by idle tasks are never freed, because the idle tasks should be part of the root cgroup, which is not removable. The problem is that while we do in-fact hold rq->lock, the newly spawned idle thread's cpu is not yet set to the correct cpu so the lockdep check in task_group(): lockdep_is_held(&task_rq(p)->lock) will fail. But this is a chicken and egg problem. Setting the CPU's runqueue requires that the CPU's runqueue already be set. ;-) So insert an RCU read-side critical section to avoid the complaint. Signed-off-by: Peter Zijlstra Signed-off-by: Paul E. McKenney --- kernel/sched.c | 12 ++++++++++++ 1 file changed, 12 insertions(+) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index dc85ceb90832..ae8f75a5ceb4 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -5337,7 +5337,19 @@ void __cpuinit init_idle(struct task_struct *idle, int cpu) idle->se.exec_start = sched_clock(); cpumask_copy(&idle->cpus_allowed, cpumask_of(cpu)); + /* + * We're having a chicken and egg problem, even though we are + * holding rq->lock, the cpu isn't yet set to this cpu so the + * lockdep check in task_group() will fail. + * + * Similar case to sched_fork(). / Alternatively we could + * use task_rq_lock() here and obtain the other rq->lock. + * + * Silence PROVE_RCU + */ + rcu_read_lock(); __set_task_cpu(idle, cpu); + rcu_read_unlock(); rq->curr = rq->idle = idle; #if defined(CONFIG_SMP) && defined(__ARCH_WANT_UNLOCKED_CTXSW) -- cgit v1.2.3 From 27b3d80a7b6adcf069b5e869e4efcc3a79f88a91 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Thu, 7 Oct 2010 12:59:29 -0700 Subject: sysctl: fix min/max handling in __do_proc_doulongvec_minmax() When proc_doulongvec_minmax() is used with an array of longs, and no min/max check requested (.extra1 or .extra2 being NULL), we dereference a NULL pointer for the second element of the array. Noticed while doing some changes in network stack for the "16TB problem" Fix is to not change min & max pointers in __do_proc_doulongvec_minmax(), so that all elements of the vector share an unique min/max limit, like proc_dointvec_minmax(). [akpm@linux-foundation.org: coding-style fixes] Signed-off-by: Eric Dumazet Cc: "Eric W. Biederman" Cc: Americo Wang Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/sysctl.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/sysctl.c b/kernel/sysctl.c index f88552c6d227..3a45c224770f 100644 --- a/kernel/sysctl.c +++ b/kernel/sysctl.c @@ -2485,7 +2485,7 @@ static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int kbuf[left] = 0; } - for (; left && vleft--; i++, min++, max++, first=0) { + for (; left && vleft--; i++, first = 0) { unsigned long val; if (write) { -- cgit v1.2.3 From 30310045dd20a286cf3800f063f79b468e132fb1 Mon Sep 17 00:00:00 2001 From: Tejun Heo Date: Mon, 11 Oct 2010 11:51:57 +0200 Subject: workqueue: fix HIGHPRI handling in keep_working() The policy function keep_working() didn't check GCWQ_HIGHPRI_PENDING and could return %false with highpri work pending. This could lead to late execution of a highpri work which was delayed due to @max_active throttling if other works are actively consuming CPU cycles. For example, the following could happen. 1. Work W0 which burns CPU cycles. 2. Two works W1 and W2 are queued to a highpri wq w/ @max_active of 1. 3. W1 starts executing and W2 is put to delayed queue. W0 and W1 are both runnable. 4. W1 finishes which puts W2 to pending queue but keep_working() incorrectly returns %false and the worker goes to sleep. 5. W0 finishes and W2 starts execution. With this patch applied, W2 starts execution as soon as W1 finishes. Signed-off-by: Tejun Heo --- kernel/workqueue.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/workqueue.c b/kernel/workqueue.c index cb2ccfbed0c6..b57a8babdec3 100644 --- a/kernel/workqueue.c +++ b/kernel/workqueue.c @@ -604,7 +604,9 @@ static bool keep_working(struct global_cwq *gcwq) { atomic_t *nr_running = get_gcwq_nr_running(gcwq->cpu); - return !list_empty(&gcwq->worklist) && atomic_read(nr_running) <= 1; + return !list_empty(&gcwq->worklist) && + (atomic_read(nr_running) <= 1 || + gcwq->flags & GCWQ_HIGHPRI_PENDING); } /* Do we need a new worker? Called from manager. */ -- cgit v1.2.3 From 6370a6ad3b53df90b4700977f7718118a2cd524a Mon Sep 17 00:00:00 2001 From: Tejun Heo Date: Mon, 11 Oct 2010 15:12:27 +0200 Subject: workqueue: add and use WQ_MEM_RECLAIM flag Add WQ_MEM_RECLAIM flag which currently maps to WQ_RESCUER, mark WQ_RESCUER as internal and replace all external WQ_RESCUER usages to WQ_MEM_RECLAIM. This makes the API users express the intent of the workqueue instead of indicating the internal mechanism used to guarantee forward progress. This is also to make it cleaner to add more semantics to WQ_MEM_RECLAIM. For example, if deemed necessary, memory reclaim workqueues can be made highpri. This patch doesn't introduce any functional change. Signed-off-by: Tejun Heo Cc: Jeff Garzik Cc: Dave Chinner Cc: Steven Whitehouse --- kernel/workqueue.c | 7 +++++++ 1 file changed, 7 insertions(+) (limited to 'kernel') diff --git a/kernel/workqueue.c b/kernel/workqueue.c index b57a8babdec3..2c6871cbcbee 100644 --- a/kernel/workqueue.c +++ b/kernel/workqueue.c @@ -2847,6 +2847,13 @@ struct workqueue_struct *__alloc_workqueue_key(const char *name, struct workqueue_struct *wq; unsigned int cpu; + /* + * Workqueues which may be used during memory reclaim should + * have a rescuer to guarantee forward progress. + */ + if (flags & WQ_MEM_RECLAIM) + flags |= WQ_RESCUER; + /* * Unbound workqueues aren't concurrency managed and should be * dispatched to workers immediately. -- cgit v1.2.3 From 84c7991059c9c4530cc911137c5bf508a41ed129 Mon Sep 17 00:00:00 2001 From: Matt Fleming Date: Sun, 3 Oct 2010 21:41:13 +0100 Subject: perf: New helper function for pmu name Introduce perf_pmu_name() helper function that returns the name of the pmu. This gives us a generic way to get the name of a pmu regardless of how an architecture identifies it internally. Signed-off-by: Matt Fleming Acked-by: Peter Zijlstra Acked-by: Paul Mundt Signed-off-by: Robert Richter --- kernel/perf_event.c | 5 +++++ 1 file changed, 5 insertions(+) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index 403d1804b198..e2534691db0d 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -85,6 +85,11 @@ void __weak hw_perf_enable(void) { barrier(); } void __weak perf_event_print_debug(void) { } +extern __weak const char *perf_pmu_name(void) +{ + return "pmu"; +} + static DEFINE_PER_CPU(int, perf_disable_count); void perf_disable(void) -- cgit v1.2.3 From ad0cf3478de8677f720ee06393b3147819568d6a Mon Sep 17 00:00:00 2001 From: John Blackwood Date: Tue, 28 Sep 2010 18:03:11 -0400 Subject: perf: Fix incorrect copy_from_user() usage perf events: repair incorrect use of copy_from_user This makes the perf_event_period() return 0 instead of -EFAULT on success. Signed-off-by: John Blackwood Signed-off-by: Joe Korty Acked-by: Peter Zijlstra LKML-Reference: <20100928220311.GA18145@tsunami.ccur.com> Signed-off-by: Ingo Molnar --- kernel/perf_event.c | 4 +--- 1 file changed, 1 insertion(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index db5b56064687..b98bed3d8182 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -2202,15 +2202,13 @@ static void perf_event_for_each(struct perf_event *event, static int perf_event_period(struct perf_event *event, u64 __user *arg) { struct perf_event_context *ctx = event->ctx; - unsigned long size; int ret = 0; u64 value; if (!event->attr.sample_period) return -EINVAL; - size = copy_from_user(&value, arg, sizeof(value)); - if (size != sizeof(value)) + if (copy_from_user(&value, arg, sizeof(value))) return -EFAULT; if (!value) -- cgit v1.2.3 From e144710b302525de5b90b9c3ba43562458d8957f Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Fri, 1 Oct 2010 16:03:45 +0200 Subject: genirq: Distangle irq.h Move irq_desc and internal functions out of irq.h Signed-off-by: Thomas Gleixner Reviewed-by: Ingo Molnar --- kernel/irq/internals.h | 100 +++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 100 insertions(+) (limited to 'kernel') diff --git a/kernel/irq/internals.h b/kernel/irq/internals.h index b905f0ab1bb2..e281e45fbb55 100644 --- a/kernel/irq/internals.h +++ b/kernel/irq/internals.h @@ -1,6 +1,7 @@ /* * IRQ subsystem internal functions and variables: */ +#include extern int noirqdebug; @@ -22,6 +23,9 @@ extern void init_kstat_irqs(struct irq_desc *desc, int node, int nr); extern void clear_kstat_irqs(struct irq_desc *desc); extern raw_spinlock_t sparse_irq_lock; +/* Resending of interrupts :*/ +void check_irq_resend(struct irq_desc *desc, unsigned int irq); + #ifdef CONFIG_SPARSE_IRQ void replace_irq_desc(unsigned int irq, struct irq_desc *desc); #endif @@ -105,3 +109,99 @@ static inline void print_irq_desc(unsigned int irq, struct irq_desc *desc) #undef P +/* Stuff below will be cleaned up after the sparse allocator is done */ + +#ifdef CONFIG_SMP +/** + * alloc_desc_masks - allocate cpumasks for irq_desc + * @desc: pointer to irq_desc struct + * @node: node which will be handling the cpumasks + * @boot: true if need bootmem + * + * Allocates affinity and pending_mask cpumask if required. + * Returns true if successful (or not required). + */ +static inline bool alloc_desc_masks(struct irq_desc *desc, int node, + bool boot) +{ + gfp_t gfp = GFP_ATOMIC; + + if (boot) + gfp = GFP_NOWAIT; + +#ifdef CONFIG_CPUMASK_OFFSTACK + if (!alloc_cpumask_var_node(&desc->irq_data.affinity, gfp, node)) + return false; + +#ifdef CONFIG_GENERIC_PENDING_IRQ + if (!alloc_cpumask_var_node(&desc->pending_mask, gfp, node)) { + free_cpumask_var(desc->irq_data.affinity); + return false; + } +#endif +#endif + return true; +} + +static inline void init_desc_masks(struct irq_desc *desc) +{ + cpumask_setall(desc->irq_data.affinity); +#ifdef CONFIG_GENERIC_PENDING_IRQ + cpumask_clear(desc->pending_mask); +#endif +} + +/** + * init_copy_desc_masks - copy cpumasks for irq_desc + * @old_desc: pointer to old irq_desc struct + * @new_desc: pointer to new irq_desc struct + * + * Insures affinity and pending_masks are copied to new irq_desc. + * If !CONFIG_CPUMASKS_OFFSTACK the cpumasks are embedded in the + * irq_desc struct so the copy is redundant. + */ + +static inline void init_copy_desc_masks(struct irq_desc *old_desc, + struct irq_desc *new_desc) +{ +#ifdef CONFIG_CPUMASK_OFFSTACK + cpumask_copy(new_desc->irq_data.affinity, old_desc->irq_data.affinity); + +#ifdef CONFIG_GENERIC_PENDING_IRQ + cpumask_copy(new_desc->pending_mask, old_desc->pending_mask); +#endif +#endif +} + +static inline void free_desc_masks(struct irq_desc *old_desc, + struct irq_desc *new_desc) +{ + free_cpumask_var(old_desc->irq_data.affinity); + +#ifdef CONFIG_GENERIC_PENDING_IRQ + free_cpumask_var(old_desc->pending_mask); +#endif +} + +#else /* !CONFIG_SMP */ + +static inline bool alloc_desc_masks(struct irq_desc *desc, int node, + bool boot) +{ + return true; +} + +static inline void init_desc_masks(struct irq_desc *desc) +{ +} + +static inline void init_copy_desc_masks(struct irq_desc *old_desc, + struct irq_desc *new_desc) +{ +} + +static inline void free_desc_masks(struct irq_desc *old_desc, + struct irq_desc *new_desc) +{ +} +#endif /* CONFIG_SMP */ -- cgit v1.2.3 From 442471848f5abb55b99cba1229301655f67492b4 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 28 Sep 2010 10:40:18 +0200 Subject: genirq: Provide status modifier Provide a irq_desc.status modifier function to cleanup the direct access to irq_desc in arch and driver code. Signed-off-by: Thomas Gleixner Reviewed-by: Ingo Molnar --- kernel/irq/chip.c | 26 +++++++------------------- 1 file changed, 7 insertions(+), 19 deletions(-) (limited to 'kernel') diff --git a/kernel/irq/chip.c b/kernel/irq/chip.c index 323547983f15..2b1f6906b824 100644 --- a/kernel/irq/chip.c +++ b/kernel/irq/chip.c @@ -851,32 +851,20 @@ set_irq_chip_and_handler_name(unsigned int irq, struct irq_chip *chip, __set_irq_handler(irq, handle, 0, name); } -void set_irq_noprobe(unsigned int irq) +void irq_modify_status(unsigned int irq, unsigned long clr, unsigned long set) { struct irq_desc *desc = irq_to_desc(irq); unsigned long flags; - if (!desc) { - printk(KERN_ERR "Trying to mark IRQ%d non-probeable\n", irq); + if (!desc) return; - } - - raw_spin_lock_irqsave(&desc->lock, flags); - desc->status |= IRQ_NOPROBE; - raw_spin_unlock_irqrestore(&desc->lock, flags); -} -void set_irq_probe(unsigned int irq) -{ - struct irq_desc *desc = irq_to_desc(irq); - unsigned long flags; - - if (!desc) { - printk(KERN_ERR "Trying to mark IRQ%d probeable\n", irq); - return; - } + /* Sanitize flags */ + set &= IRQF_MODIFY_MASK; + clr &= IRQF_MODIFY_MASK; raw_spin_lock_irqsave(&desc->lock, flags); - desc->status &= ~IRQ_NOPROBE; + desc->status &= ~clr; + desc->status |= set; raw_spin_unlock_irqrestore(&desc->lock, flags); } -- cgit v1.2.3 From f303a6dd127b5ec6de90d1cd79ed19820c7e9658 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 28 Sep 2010 17:34:01 +0200 Subject: genirq: Sanitize irq_data accessors Get the data structure from the core and provide inline wrappers to access the irq_data members. Provide accessor inlines for irq_data as well. Signed-off-by: Thomas Gleixner Reviewed-by: Ingo Molnar --- kernel/irq/chip.c | 8 ++++++++ 1 file changed, 8 insertions(+) (limited to 'kernel') diff --git a/kernel/irq/chip.c b/kernel/irq/chip.c index 2b1f6906b824..659be326c8e8 100644 --- a/kernel/irq/chip.c +++ b/kernel/irq/chip.c @@ -256,6 +256,14 @@ int set_irq_chip_data(unsigned int irq, void *data) } EXPORT_SYMBOL(set_irq_chip_data); +struct irq_data *irq_get_irq_data(unsigned int irq) +{ + struct irq_desc *desc = irq_to_desc(irq); + + return desc ? &desc->irq_data : NULL; +} +EXPORT_SYMBOL_GPL(irq_get_irq_data); + /** * set_irq_nested_thread - Set/Reset the IRQ_NESTED_THREAD flag of an irq * -- cgit v1.2.3 From 3795de236d67a05994a1a12759db9d4dd9ffc42c Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Wed, 22 Sep 2010 17:09:43 +0200 Subject: genirq: Distangle kernel/irq/handle.c kernel/irq/handle.c has become a dumpground for random code in random order. Split out the irq descriptor management and the dummy irq_chip implementation into separate files. Cleanup the include maze while at it. No code change. Signed-off-by: Thomas Gleixner Reviewed-by: Ingo Molnar --- kernel/irq/Makefile | 2 +- kernel/irq/dummychip.c | 68 ++++++++++ kernel/irq/handle.c | 333 +------------------------------------------------ kernel/irq/irqdesc.c | 269 +++++++++++++++++++++++++++++++++++++++ 4 files changed, 340 insertions(+), 332 deletions(-) create mode 100644 kernel/irq/dummychip.c create mode 100644 kernel/irq/irqdesc.c (limited to 'kernel') diff --git a/kernel/irq/Makefile b/kernel/irq/Makefile index 7d047808419d..1eaab0da56db 100644 --- a/kernel/irq/Makefile +++ b/kernel/irq/Makefile @@ -1,5 +1,5 @@ -obj-y := handle.o manage.o spurious.o resend.o chip.o devres.o +obj-y := irqdesc.o handle.o manage.o spurious.o resend.o chip.o dummychip.o devres.o obj-$(CONFIG_GENERIC_IRQ_PROBE) += autoprobe.o obj-$(CONFIG_PROC_FS) += proc.o obj-$(CONFIG_GENERIC_PENDING_IRQ) += migration.o diff --git a/kernel/irq/dummychip.c b/kernel/irq/dummychip.c new file mode 100644 index 000000000000..918dea9de9ea --- /dev/null +++ b/kernel/irq/dummychip.c @@ -0,0 +1,68 @@ +/* + * Copyright (C) 1992, 1998-2006 Linus Torvalds, Ingo Molnar + * Copyright (C) 2005-2006, Thomas Gleixner, Russell King + * + * This file contains the dummy interrupt chip implementation + */ +#include +#include + +#include "internals.h" + +/* + * What should we do if we get a hw irq event on an illegal vector? + * Each architecture has to answer this themself. + */ +static void ack_bad(struct irq_data *data) +{ + struct irq_desc *desc = irq_data_to_desc(data); + + print_irq_desc(data->irq, desc); + ack_bad_irq(data->irq); +} + +/* + * NOP functions + */ +static void noop(struct irq_data *data) { } + +static unsigned int noop_ret(struct irq_data *data) +{ + return 0; +} + +#ifndef CONFIG_GENERIC_HARDIRQS_NO_CRUFT +static void compat_noop(unsigned int irq) { } +#define END_INIT .end = compat_noop +#else +#define END_INIT +#endif + +/* + * Generic no controller implementation + */ +struct irq_chip no_irq_chip = { + .name = "none", + .irq_startup = noop_ret, + .irq_shutdown = noop, + .irq_enable = noop, + .irq_disable = noop, + .irq_ack = ack_bad, + END_INIT +}; + +/* + * Generic dummy implementation which can be used for + * real dumb interrupt sources + */ +struct irq_chip dummy_irq_chip = { + .name = "dummy", + .irq_startup = noop_ret, + .irq_shutdown = noop, + .irq_enable = noop, + .irq_disable = noop, + .irq_ack = noop, + .irq_mask = noop, + .irq_unmask = noop, + END_INIT +}; diff --git a/kernel/irq/handle.c b/kernel/irq/handle.c index 3fcef37154a1..e2347eb63306 100644 --- a/kernel/irq/handle.c +++ b/kernel/irq/handle.c @@ -11,24 +11,15 @@ */ #include -#include -#include -#include #include +#include #include #include -#include -#include -#include + #include #include "internals.h" -/* - * lockdep: we want to handle all irq_desc locks as a single lock-class: - */ -struct lock_class_key irq_desc_lock_class; - /** * handle_bad_irq - handle spurious and unhandled irqs * @irq: the interrupt number @@ -43,308 +34,6 @@ void handle_bad_irq(unsigned int irq, struct irq_desc *desc) ack_bad_irq(irq); } -#if defined(CONFIG_SMP) && defined(CONFIG_GENERIC_HARDIRQS) -static void __init init_irq_default_affinity(void) -{ - alloc_cpumask_var(&irq_default_affinity, GFP_NOWAIT); - cpumask_setall(irq_default_affinity); -} -#else -static void __init init_irq_default_affinity(void) -{ -} -#endif - -/* - * Linux has a controller-independent interrupt architecture. - * Every controller has a 'controller-template', that is used - * by the main code to do the right thing. Each driver-visible - * interrupt source is transparently wired to the appropriate - * controller. Thus drivers need not be aware of the - * interrupt-controller. - * - * The code is designed to be easily extended with new/different - * interrupt controllers, without having to do assembly magic or - * having to touch the generic code. - * - * Controller mappings for all interrupt sources: - */ -int nr_irqs = NR_IRQS; -EXPORT_SYMBOL_GPL(nr_irqs); - -#ifdef CONFIG_SPARSE_IRQ - -static struct irq_desc irq_desc_init = { - .status = IRQ_DISABLED, - .handle_irq = handle_bad_irq, - .depth = 1, - .lock = __RAW_SPIN_LOCK_UNLOCKED(irq_desc_init.lock), -}; - -void __ref init_kstat_irqs(struct irq_desc *desc, int node, int nr) -{ - void *ptr; - - ptr = kzalloc_node(nr * sizeof(*desc->kstat_irqs), - GFP_ATOMIC, node); - - /* - * don't overwite if can not get new one - * init_copy_kstat_irqs() could still use old one - */ - if (ptr) { - printk(KERN_DEBUG " alloc kstat_irqs on node %d\n", node); - desc->kstat_irqs = ptr; - } -} - -static void init_one_irq_desc(int irq, struct irq_desc *desc, int node) -{ - memcpy(desc, &irq_desc_init, sizeof(struct irq_desc)); - - raw_spin_lock_init(&desc->lock); - desc->irq_data.irq = irq; -#ifdef CONFIG_SMP - desc->irq_data.node = node; -#endif - lockdep_set_class(&desc->lock, &irq_desc_lock_class); - init_kstat_irqs(desc, node, nr_cpu_ids); - if (!desc->kstat_irqs) { - printk(KERN_ERR "can not alloc kstat_irqs\n"); - BUG_ON(1); - } - if (!alloc_desc_masks(desc, node, false)) { - printk(KERN_ERR "can not alloc irq_desc cpumasks\n"); - BUG_ON(1); - } - init_desc_masks(desc); - arch_init_chip_data(desc, node); -} - -/* - * Protect the sparse_irqs: - */ -DEFINE_RAW_SPINLOCK(sparse_irq_lock); - -static RADIX_TREE(irq_desc_tree, GFP_ATOMIC); - -static void set_irq_desc(unsigned int irq, struct irq_desc *desc) -{ - radix_tree_insert(&irq_desc_tree, irq, desc); -} - -struct irq_desc *irq_to_desc(unsigned int irq) -{ - return radix_tree_lookup(&irq_desc_tree, irq); -} - -void replace_irq_desc(unsigned int irq, struct irq_desc *desc) -{ - void **ptr; - - ptr = radix_tree_lookup_slot(&irq_desc_tree, irq); - if (ptr) - radix_tree_replace_slot(ptr, desc); -} - -static struct irq_desc irq_desc_legacy[NR_IRQS_LEGACY] __cacheline_aligned_in_smp = { - [0 ... NR_IRQS_LEGACY-1] = { - .status = IRQ_DISABLED, - .handle_irq = handle_bad_irq, - .depth = 1, - .lock = __RAW_SPIN_LOCK_UNLOCKED(irq_desc_init.lock), - } -}; - -static unsigned int *kstat_irqs_legacy; - -int __init early_irq_init(void) -{ - struct irq_desc *desc; - int legacy_count; - int node; - int i; - - init_irq_default_affinity(); - - /* initialize nr_irqs based on nr_cpu_ids */ - arch_probe_nr_irqs(); - printk(KERN_INFO "NR_IRQS:%d nr_irqs:%d\n", NR_IRQS, nr_irqs); - - desc = irq_desc_legacy; - legacy_count = ARRAY_SIZE(irq_desc_legacy); - node = first_online_node; - - /* allocate based on nr_cpu_ids */ - kstat_irqs_legacy = kzalloc_node(NR_IRQS_LEGACY * nr_cpu_ids * - sizeof(int), GFP_NOWAIT, node); - - irq_desc_init.irq_data.chip = &no_irq_chip; - - for (i = 0; i < legacy_count; i++) { - desc[i].irq_data.irq = i; - desc[i].irq_data.chip = &no_irq_chip; -#ifdef CONFIG_SMP - desc[i].irq_data.node = node; -#endif - desc[i].kstat_irqs = kstat_irqs_legacy + i * nr_cpu_ids; - lockdep_set_class(&desc[i].lock, &irq_desc_lock_class); - alloc_desc_masks(&desc[i], node, true); - init_desc_masks(&desc[i]); - set_irq_desc(i, &desc[i]); - } - - return arch_early_irq_init(); -} - -struct irq_desc * __ref irq_to_desc_alloc_node(unsigned int irq, int node) -{ - struct irq_desc *desc; - unsigned long flags; - - if (irq >= nr_irqs) { - WARN(1, "irq (%d) >= nr_irqs (%d) in irq_to_desc_alloc\n", - irq, nr_irqs); - return NULL; - } - - desc = irq_to_desc(irq); - if (desc) - return desc; - - raw_spin_lock_irqsave(&sparse_irq_lock, flags); - - /* We have to check it to avoid races with another CPU */ - desc = irq_to_desc(irq); - if (desc) - goto out_unlock; - - desc = kzalloc_node(sizeof(*desc), GFP_ATOMIC, node); - - printk(KERN_DEBUG " alloc irq_desc for %d on node %d\n", irq, node); - if (!desc) { - printk(KERN_ERR "can not alloc irq_desc\n"); - BUG_ON(1); - } - init_one_irq_desc(irq, desc, node); - - set_irq_desc(irq, desc); - -out_unlock: - raw_spin_unlock_irqrestore(&sparse_irq_lock, flags); - - return desc; -} - -#else /* !CONFIG_SPARSE_IRQ */ - -struct irq_desc irq_desc[NR_IRQS] __cacheline_aligned_in_smp = { - [0 ... NR_IRQS-1] = { - .status = IRQ_DISABLED, - .handle_irq = handle_bad_irq, - .depth = 1, - .lock = __RAW_SPIN_LOCK_UNLOCKED(irq_desc->lock), - } -}; - -static unsigned int kstat_irqs_all[NR_IRQS][NR_CPUS]; -int __init early_irq_init(void) -{ - struct irq_desc *desc; - int count; - int i; - - init_irq_default_affinity(); - - printk(KERN_INFO "NR_IRQS:%d\n", NR_IRQS); - - desc = irq_desc; - count = ARRAY_SIZE(irq_desc); - - for (i = 0; i < count; i++) { - desc[i].irq_data.irq = i; - desc[i].irq_data.chip = &no_irq_chip; - alloc_desc_masks(&desc[i], 0, true); - init_desc_masks(&desc[i]); - desc[i].kstat_irqs = kstat_irqs_all[i]; - } - return arch_early_irq_init(); -} - -struct irq_desc *irq_to_desc(unsigned int irq) -{ - return (irq < NR_IRQS) ? irq_desc + irq : NULL; -} - -struct irq_desc *irq_to_desc_alloc_node(unsigned int irq, int node) -{ - return irq_to_desc(irq); -} -#endif /* !CONFIG_SPARSE_IRQ */ - -void clear_kstat_irqs(struct irq_desc *desc) -{ - memset(desc->kstat_irqs, 0, nr_cpu_ids * sizeof(*(desc->kstat_irqs))); -} - -/* - * What should we do if we get a hw irq event on an illegal vector? - * Each architecture has to answer this themself. - */ -static void ack_bad(struct irq_data *data) -{ - struct irq_desc *desc = irq_data_to_desc(data); - - print_irq_desc(data->irq, desc); - ack_bad_irq(data->irq); -} - -/* - * NOP functions - */ -static void noop(struct irq_data *data) { } - -static unsigned int noop_ret(struct irq_data *data) -{ - return 0; -} - -#ifndef CONFIG_GENERIC_HARDIRQS_NO_DEPRECATED -static void compat_noop(unsigned int irq) { } -#define END_INIT .end = compat_noop -#else -#define END_INIT -#endif - -/* - * Generic no controller implementation - */ -struct irq_chip no_irq_chip = { - .name = "none", - .irq_startup = noop_ret, - .irq_shutdown = noop, - .irq_enable = noop, - .irq_disable = noop, - .irq_ack = ack_bad, - END_INIT -}; - -/* - * Generic dummy implementation which can be used for - * real dumb interrupt sources - */ -struct irq_chip dummy_irq_chip = { - .name = "dummy", - .irq_startup = noop_ret, - .irq_shutdown = noop, - .irq_enable = noop, - .irq_disable = noop, - .irq_ack = noop, - .irq_mask = noop, - .irq_unmask = noop, - END_INIT -}; - /* * Special, empty irq handler: */ @@ -540,21 +229,3 @@ out: return 1; } #endif - -void early_init_irq_lock_class(void) -{ - struct irq_desc *desc; - int i; - - for_each_irq_desc(i, desc) { - lockdep_set_class(&desc->lock, &irq_desc_lock_class); - } -} - -unsigned int kstat_irqs_cpu(unsigned int irq, int cpu) -{ - struct irq_desc *desc = irq_to_desc(irq); - return desc ? desc->kstat_irqs[cpu] : 0; -} -EXPORT_SYMBOL(kstat_irqs_cpu); - diff --git a/kernel/irq/irqdesc.c b/kernel/irq/irqdesc.c new file mode 100644 index 000000000000..fbf8cfa00510 --- /dev/null +++ b/kernel/irq/irqdesc.c @@ -0,0 +1,269 @@ +/* + * Copyright (C) 1992, 1998-2006 Linus Torvalds, Ingo Molnar + * Copyright (C) 2005-2006, Thomas Gleixner, Russell King + * + * This file contains the interrupt descriptor management code + * + * Detailed information is available in Documentation/DocBook/genericirq + * + */ +#include +#include +#include +#include +#include +#include + +#include "internals.h" + +/* + * lockdep: we want to handle all irq_desc locks as a single lock-class: + */ +struct lock_class_key irq_desc_lock_class; + +#if defined(CONFIG_SMP) && defined(CONFIG_GENERIC_HARDIRQS) +static void __init init_irq_default_affinity(void) +{ + alloc_cpumask_var(&irq_default_affinity, GFP_NOWAIT); + cpumask_setall(irq_default_affinity); +} +#else +static void __init init_irq_default_affinity(void) +{ +} +#endif + +int nr_irqs = NR_IRQS; +EXPORT_SYMBOL_GPL(nr_irqs); + +#ifdef CONFIG_SPARSE_IRQ + +static struct irq_desc irq_desc_init = { + .status = IRQ_DISABLED, + .handle_irq = handle_bad_irq, + .depth = 1, + .lock = __RAW_SPIN_LOCK_UNLOCKED(irq_desc_init.lock), +}; + +void __ref init_kstat_irqs(struct irq_desc *desc, int node, int nr) +{ + void *ptr; + + ptr = kzalloc_node(nr * sizeof(*desc->kstat_irqs), + GFP_ATOMIC, node); + + /* + * don't overwite if can not get new one + * init_copy_kstat_irqs() could still use old one + */ + if (ptr) { + printk(KERN_DEBUG " alloc kstat_irqs on node %d\n", node); + desc->kstat_irqs = ptr; + } +} + +static void init_one_irq_desc(int irq, struct irq_desc *desc, int node) +{ + memcpy(desc, &irq_desc_init, sizeof(struct irq_desc)); + + raw_spin_lock_init(&desc->lock); + desc->irq_data.irq = irq; +#ifdef CONFIG_SMP + desc->irq_data.node = node; +#endif + lockdep_set_class(&desc->lock, &irq_desc_lock_class); + init_kstat_irqs(desc, node, nr_cpu_ids); + if (!desc->kstat_irqs) { + printk(KERN_ERR "can not alloc kstat_irqs\n"); + BUG_ON(1); + } + if (!alloc_desc_masks(desc, node, false)) { + printk(KERN_ERR "can not alloc irq_desc cpumasks\n"); + BUG_ON(1); + } + init_desc_masks(desc); + arch_init_chip_data(desc, node); +} + +/* + * Protect the sparse_irqs: + */ +DEFINE_RAW_SPINLOCK(sparse_irq_lock); + +static RADIX_TREE(irq_desc_tree, GFP_ATOMIC); + +static void set_irq_desc(unsigned int irq, struct irq_desc *desc) +{ + radix_tree_insert(&irq_desc_tree, irq, desc); +} + +struct irq_desc *irq_to_desc(unsigned int irq) +{ + return radix_tree_lookup(&irq_desc_tree, irq); +} + +void replace_irq_desc(unsigned int irq, struct irq_desc *desc) +{ + void **ptr; + + ptr = radix_tree_lookup_slot(&irq_desc_tree, irq); + if (ptr) + radix_tree_replace_slot(ptr, desc); +} + +static struct irq_desc irq_desc_legacy[NR_IRQS_LEGACY] __cacheline_aligned_in_smp = { + [0 ... NR_IRQS_LEGACY-1] = { + .status = IRQ_DISABLED, + .handle_irq = handle_bad_irq, + .depth = 1, + .lock = __RAW_SPIN_LOCK_UNLOCKED(irq_desc_init.lock), + } +}; + +static unsigned int *kstat_irqs_legacy; + +int __init early_irq_init(void) +{ + struct irq_desc *desc; + int legacy_count; + int node; + int i; + + init_irq_default_affinity(); + + /* initialize nr_irqs based on nr_cpu_ids */ + arch_probe_nr_irqs(); + printk(KERN_INFO "NR_IRQS:%d nr_irqs:%d\n", NR_IRQS, nr_irqs); + + desc = irq_desc_legacy; + legacy_count = ARRAY_SIZE(irq_desc_legacy); + node = first_online_node; + + /* allocate based on nr_cpu_ids */ + kstat_irqs_legacy = kzalloc_node(NR_IRQS_LEGACY * nr_cpu_ids * + sizeof(int), GFP_NOWAIT, node); + + irq_desc_init.irq_data.chip = &no_irq_chip; + + for (i = 0; i < legacy_count; i++) { + desc[i].irq_data.irq = i; + desc[i].irq_data.chip = &no_irq_chip; +#ifdef CONFIG_SMP + desc[i].irq_data.node = node; +#endif + desc[i].kstat_irqs = kstat_irqs_legacy + i * nr_cpu_ids; + lockdep_set_class(&desc[i].lock, &irq_desc_lock_class); + alloc_desc_masks(&desc[i], node, true); + init_desc_masks(&desc[i]); + set_irq_desc(i, &desc[i]); + } + + return arch_early_irq_init(); +} + +struct irq_desc * __ref irq_to_desc_alloc_node(unsigned int irq, int node) +{ + struct irq_desc *desc; + unsigned long flags; + + if (irq >= nr_irqs) { + WARN(1, "irq (%d) >= nr_irqs (%d) in irq_to_desc_alloc\n", + irq, nr_irqs); + return NULL; + } + + desc = irq_to_desc(irq); + if (desc) + return desc; + + raw_spin_lock_irqsave(&sparse_irq_lock, flags); + + /* We have to check it to avoid races with another CPU */ + desc = irq_to_desc(irq); + if (desc) + goto out_unlock; + + desc = kzalloc_node(sizeof(*desc), GFP_ATOMIC, node); + + printk(KERN_DEBUG " alloc irq_desc for %d on node %d\n", irq, node); + if (!desc) { + printk(KERN_ERR "can not alloc irq_desc\n"); + BUG_ON(1); + } + init_one_irq_desc(irq, desc, node); + + set_irq_desc(irq, desc); + +out_unlock: + raw_spin_unlock_irqrestore(&sparse_irq_lock, flags); + + return desc; +} + +#else /* !CONFIG_SPARSE_IRQ */ + +struct irq_desc irq_desc[NR_IRQS] __cacheline_aligned_in_smp = { + [0 ... NR_IRQS-1] = { + .status = IRQ_DISABLED, + .handle_irq = handle_bad_irq, + .depth = 1, + .lock = __RAW_SPIN_LOCK_UNLOCKED(irq_desc->lock), + } +}; + +static unsigned int kstat_irqs_all[NR_IRQS][NR_CPUS]; +int __init early_irq_init(void) +{ + struct irq_desc *desc; + int count; + int i; + + init_irq_default_affinity(); + + printk(KERN_INFO "NR_IRQS:%d\n", NR_IRQS); + + desc = irq_desc; + count = ARRAY_SIZE(irq_desc); + + for (i = 0; i < count; i++) { + desc[i].irq_data.irq = i; + desc[i].irq_data.chip = &no_irq_chip; + alloc_desc_masks(&desc[i], 0, true); + init_desc_masks(&desc[i]); + desc[i].kstat_irqs = kstat_irqs_all[i]; + } + return arch_early_irq_init(); +} + +struct irq_desc *irq_to_desc(unsigned int irq) +{ + return (irq < NR_IRQS) ? irq_desc + irq : NULL; +} + +struct irq_desc *irq_to_desc_alloc_node(unsigned int irq, int node) +{ + return irq_to_desc(irq); +} +#endif /* !CONFIG_SPARSE_IRQ */ + +void clear_kstat_irqs(struct irq_desc *desc) +{ + memset(desc->kstat_irqs, 0, nr_cpu_ids * sizeof(*(desc->kstat_irqs))); +} + +void early_init_irq_lock_class(void) +{ + struct irq_desc *desc; + int i; + + for_each_irq_desc(i, desc) { + lockdep_set_class(&desc->lock, &irq_desc_lock_class); + } +} + +unsigned int kstat_irqs_cpu(unsigned int irq, int cpu) +{ + struct irq_desc *desc = irq_to_desc(irq); + return desc ? desc->kstat_irqs[cpu] : 0; +} +EXPORT_SYMBOL(kstat_irqs_cpu); -- cgit v1.2.3 From 154cd387cdf0e5566ce523cbddf92dd2a062dfd6 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Wed, 22 Sep 2010 15:58:45 +0200 Subject: genirq: Remove early_init_irq_lock_class() early_init_irq_lock_class() is called way before anything touches the irq descriptors. In case of SPARSE_IRQ=y this is a NOP operation because the radix tree is empty at this point. For the SPARSE_IRQ=n case it's sufficient to set the lock class in early_init_irq(). Signed-off-by: Thomas Gleixner Reviewed-by: Ingo Molnar --- kernel/irq/irqdesc.c | 11 +---------- 1 file changed, 1 insertion(+), 10 deletions(-) (limited to 'kernel') diff --git a/kernel/irq/irqdesc.c b/kernel/irq/irqdesc.c index fbf8cfa00510..0a7a0908afbc 100644 --- a/kernel/irq/irqdesc.c +++ b/kernel/irq/irqdesc.c @@ -231,6 +231,7 @@ int __init early_irq_init(void) alloc_desc_masks(&desc[i], 0, true); init_desc_masks(&desc[i]); desc[i].kstat_irqs = kstat_irqs_all[i]; + lockdep_set_class(&desc[i].lock, &irq_desc_lock_class); } return arch_early_irq_init(); } @@ -251,16 +252,6 @@ void clear_kstat_irqs(struct irq_desc *desc) memset(desc->kstat_irqs, 0, nr_cpu_ids * sizeof(*(desc->kstat_irqs))); } -void early_init_irq_lock_class(void) -{ - struct irq_desc *desc; - int i; - - for_each_irq_desc(i, desc) { - lockdep_set_class(&desc->lock, &irq_desc_lock_class); - } -} - unsigned int kstat_irqs_cpu(unsigned int irq, int cpu) { struct irq_desc *desc = irq_to_desc(irq); -- cgit v1.2.3 From d895f51ebb54cefe367bda135fcf2cd734d51d03 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Mon, 27 Sep 2010 17:45:49 +0200 Subject: genirq: Remove export of kstat_irqs_cpu The statistics accessor is only used by proc/stats and show_interrupts(). Both are compiled in. Signed-off-by: Thomas Gleixner Reviewed-by: Ingo Molnar --- kernel/irq/irqdesc.c | 1 - 1 file changed, 1 deletion(-) (limited to 'kernel') diff --git a/kernel/irq/irqdesc.c b/kernel/irq/irqdesc.c index 0a7a0908afbc..78ff426a6cb7 100644 --- a/kernel/irq/irqdesc.c +++ b/kernel/irq/irqdesc.c @@ -257,4 +257,3 @@ unsigned int kstat_irqs_cpu(unsigned int irq, int cpu) struct irq_desc *desc = irq_to_desc(irq); return desc ? desc->kstat_irqs[cpu] : 0; } -EXPORT_SYMBOL(kstat_irqs_cpu); -- cgit v1.2.3 From 1318a481fc37c503a901b96ae06b692ca2b21af5 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Mon, 27 Sep 2010 21:01:37 +0200 Subject: genirq: Provide default irq init flags Arch code sets it's own irq_desc.status flags right after boot and for dynamically allocated interrupts. That might involve iterating over a huge array. Allow ARCH_IRQ_INIT_FLAGS to set separate flags aside of IRQ_DISABLED which is the default. Signed-off-by: Thomas Gleixner Reviewed-by: Ingo Molnar --- kernel/irq/chip.c | 2 +- kernel/irq/irqdesc.c | 6 +++--- 2 files changed, 4 insertions(+), 4 deletions(-) (limited to 'kernel') diff --git a/kernel/irq/chip.c b/kernel/irq/chip.c index 659be326c8e8..3405761d6224 100644 --- a/kernel/irq/chip.c +++ b/kernel/irq/chip.c @@ -31,7 +31,7 @@ static void dynamic_irq_init_x(unsigned int irq, bool keep_chip_data) /* Ensure we don't have left over values from a previous use of this irq */ raw_spin_lock_irqsave(&desc->lock, flags); - desc->status = IRQ_DISABLED; + desc->status = IRQ_DEFAULT_INIT_FLAGS; desc->irq_data.chip = &no_irq_chip; desc->handle_irq = handle_bad_irq; desc->depth = 1; diff --git a/kernel/irq/irqdesc.c b/kernel/irq/irqdesc.c index 78ff426a6cb7..29963f99f24d 100644 --- a/kernel/irq/irqdesc.c +++ b/kernel/irq/irqdesc.c @@ -39,7 +39,7 @@ EXPORT_SYMBOL_GPL(nr_irqs); #ifdef CONFIG_SPARSE_IRQ static struct irq_desc irq_desc_init = { - .status = IRQ_DISABLED, + .status = IRQ_DEFAULT_INIT_FLAGS, .handle_irq = handle_bad_irq, .depth = 1, .lock = __RAW_SPIN_LOCK_UNLOCKED(irq_desc_init.lock), @@ -113,7 +113,7 @@ void replace_irq_desc(unsigned int irq, struct irq_desc *desc) static struct irq_desc irq_desc_legacy[NR_IRQS_LEGACY] __cacheline_aligned_in_smp = { [0 ... NR_IRQS_LEGACY-1] = { - .status = IRQ_DISABLED, + .status = IRQ_DEFAULT_INIT_FLAGS, .handle_irq = handle_bad_irq, .depth = 1, .lock = __RAW_SPIN_LOCK_UNLOCKED(irq_desc_init.lock), @@ -204,7 +204,7 @@ out_unlock: struct irq_desc irq_desc[NR_IRQS] __cacheline_aligned_in_smp = { [0 ... NR_IRQS-1] = { - .status = IRQ_DISABLED, + .status = IRQ_DEFAULT_INIT_FLAGS, .handle_irq = handle_bad_irq, .depth = 1, .lock = __RAW_SPIN_LOCK_UNLOCKED(irq_desc->lock), -- cgit v1.2.3 From 1f5a5b87f78fade3ae48dfd55e8765d1d622ea4e Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Mon, 27 Sep 2010 17:48:26 +0200 Subject: genirq: Implement a sane sparse_irq allocator The current sparse_irq allocator has several short comings due to failures in the design or the lack of it: - Requires iteration over the number of active irqs to find a free slot (Some architectures have grown their own workarounds for this) - Removal of entries is not possible - Racy between create_irq_nr and destroy_irq (plugged by horrible callbacks) - Migration of active irq descriptors is not possible - No bulk allocation of irq ranges - Sprinkeled irq_desc references all over the place outside of kernel/irq/ (The previous chip functions series is addressing this issue) Implement a sane allocator which fixes the above short comings (though migration of active descriptors needs a full tree wide cleanup of the direct and mostly unlocked access to irq_desc). The new allocator still uses a radix_tree, but uses a bitmap for keeping track of allocated irq numbers. That allows: - Fast lookup of a free slot - Allows the removal of descriptors - Prevents the create/destroy race - Bulk allocation of consecutive irq ranges - Basic design is ready for migration of life descriptors after further cleanups The bitmap is also used in the SPARSE_IRQ=n case for lookup and raceless (de)allocation of irq numbers. So it removes the requirement for looping through the descriptor array to find slots. Right now it uses sparse_irq_lock to protect the bitmap and the radix tree, but after cleaning up all users we should be able convert that to a mutex and to switch the radix_tree and decriptor allocations to GFP_KERNEL. [ Folded in a bugfix from Yinghai Lu ] Signed-off-by: Thomas Gleixner Reviewed-by: Ingo Molnar --- kernel/irq/irqdesc.c | 231 +++++++++++++++++++++++++++++++++++++++++++++++++-- 1 file changed, 223 insertions(+), 8 deletions(-) (limited to 'kernel') diff --git a/kernel/irq/irqdesc.c b/kernel/irq/irqdesc.c index 29963f99f24d..4eea48b4f576 100644 --- a/kernel/irq/irqdesc.c +++ b/kernel/irq/irqdesc.c @@ -13,6 +13,7 @@ #include #include #include +#include #include "internals.h" @@ -33,9 +34,54 @@ static void __init init_irq_default_affinity(void) } #endif +#ifdef CONFIG_SMP +static int alloc_masks(struct irq_desc *desc, gfp_t gfp, int node) +{ + if (!zalloc_cpumask_var_node(&desc->irq_data.affinity, gfp, node)) + return -ENOMEM; + +#ifdef CONFIG_GENERIC_PENDING_IRQ + if (!zalloc_cpumask_var_node(&desc->pending_mask, gfp, node)) { + free_cpumask_var(desc->irq_data.affinity); + return -ENOMEM; + } +#endif + return 0; +} + +static void desc_smp_init(struct irq_desc *desc, int node) +{ + desc->node = node; + cpumask_copy(desc->irq_data.affinity, irq_default_affinity); +} + +#else +static inline int +alloc_masks(struct irq_desc *desc, gfp_t gfp, int node) { return 0; } +static inline void desc_smp_init(struct irq_desc *desc, int node) { } +#endif + +static void desc_set_defaults(unsigned int irq, struct irq_desc *desc, int node) +{ + desc->irq_data.irq = irq; + desc->irq_data.chip = &no_irq_chip; + desc->irq_data.chip_data = NULL; + desc->irq_data.handler_data = NULL; + desc->irq_data.msi_desc = NULL; + desc->status = IRQ_DEFAULT_INIT_FLAGS; + desc->handle_irq = handle_bad_irq; + desc->depth = 1; + desc->name = NULL; + memset(desc->kstat_irqs, 0, nr_cpu_ids * sizeof(*(desc->kstat_irqs))); + desc_smp_init(desc, node); +} + int nr_irqs = NR_IRQS; EXPORT_SYMBOL_GPL(nr_irqs); +DEFINE_RAW_SPINLOCK(sparse_irq_lock); +static DECLARE_BITMAP(allocated_irqs, NR_IRQS); + #ifdef CONFIG_SPARSE_IRQ static struct irq_desc irq_desc_init = { @@ -85,14 +131,9 @@ static void init_one_irq_desc(int irq, struct irq_desc *desc, int node) arch_init_chip_data(desc, node); } -/* - * Protect the sparse_irqs: - */ -DEFINE_RAW_SPINLOCK(sparse_irq_lock); - static RADIX_TREE(irq_desc_tree, GFP_ATOMIC); -static void set_irq_desc(unsigned int irq, struct irq_desc *desc) +static void irq_insert_desc(unsigned int irq, struct irq_desc *desc) { radix_tree_insert(&irq_desc_tree, irq, desc); } @@ -111,6 +152,94 @@ void replace_irq_desc(unsigned int irq, struct irq_desc *desc) radix_tree_replace_slot(ptr, desc); } +static void delete_irq_desc(unsigned int irq) +{ + radix_tree_delete(&irq_desc_tree, irq); +} + +#ifdef CONFIG_SMP +static void free_masks(struct irq_desc *desc) +{ +#ifdef CONFIG_GENERIC_PENDING_IRQ + free_cpumask_var(desc->pending_mask); +#endif + free_cpumask_var(desc->affinity); +} +#else +static inline void free_masks(struct irq_desc *desc) { } +#endif + +static struct irq_desc *alloc_desc(int irq, int node) +{ + struct irq_desc *desc; + gfp_t gfp = GFP_KERNEL; + + desc = kzalloc_node(sizeof(*desc), gfp, node); + if (!desc) + return NULL; + /* allocate based on nr_cpu_ids */ + desc->kstat_irqs = kzalloc_node(nr_cpu_ids * sizeof(*desc->kstat_irqs), + gfp, node); + if (!desc->kstat_irqs) + goto err_desc; + + if (alloc_masks(desc, gfp, node)) + goto err_kstat; + + raw_spin_lock_init(&desc->lock); + lockdep_set_class(&desc->lock, &irq_desc_lock_class); + + desc_set_defaults(irq, desc, node); + + return desc; + +err_kstat: + kfree(desc->kstat_irqs); +err_desc: + kfree(desc); + return NULL; +} + +static void free_desc(unsigned int irq) +{ + struct irq_desc *desc = irq_to_desc(irq); + unsigned long flags; + + raw_spin_lock_irqsave(&sparse_irq_lock, flags); + delete_irq_desc(irq); + raw_spin_unlock_irqrestore(&sparse_irq_lock, flags); + + free_masks(desc); + kfree(desc->kstat_irqs); + kfree(desc); +} + +static int alloc_descs(unsigned int start, unsigned int cnt, int node) +{ + struct irq_desc *desc; + unsigned long flags; + int i; + + for (i = 0; i < cnt; i++) { + desc = alloc_desc(start + i, node); + if (!desc) + goto err; + raw_spin_lock_irqsave(&sparse_irq_lock, flags); + irq_insert_desc(start + i, desc); + raw_spin_unlock_irqrestore(&sparse_irq_lock, flags); + } + return start; + +err: + for (i--; i >= 0; i--) + free_desc(start + i); + + raw_spin_lock_irqsave(&sparse_irq_lock, flags); + bitmap_clear(allocated_irqs, start, cnt); + raw_spin_unlock_irqrestore(&sparse_irq_lock, flags); + return -ENOMEM; +} + static struct irq_desc irq_desc_legacy[NR_IRQS_LEGACY] __cacheline_aligned_in_smp = { [0 ... NR_IRQS_LEGACY-1] = { .status = IRQ_DEFAULT_INIT_FLAGS, @@ -155,7 +284,7 @@ int __init early_irq_init(void) lockdep_set_class(&desc[i].lock, &irq_desc_lock_class); alloc_desc_masks(&desc[i], node, true); init_desc_masks(&desc[i]); - set_irq_desc(i, &desc[i]); + irq_insert_desc(i, &desc[i]); } return arch_early_irq_init(); @@ -192,7 +321,7 @@ struct irq_desc * __ref irq_to_desc_alloc_node(unsigned int irq, int node) } init_one_irq_desc(irq, desc, node); - set_irq_desc(irq, desc); + irq_insert_desc(irq, desc); out_unlock: raw_spin_unlock_irqrestore(&sparse_irq_lock, flags); @@ -245,8 +374,94 @@ struct irq_desc *irq_to_desc_alloc_node(unsigned int irq, int node) { return irq_to_desc(irq); } + +#ifdef CONFIG_SMP +static inline int desc_node(struct irq_desc *desc) +{ + return desc->irq_data.node; +} +#else +static inline int desc_node(struct irq_desc *desc) { return 0; } +#endif + +static void free_desc(unsigned int irq) +{ + struct irq_desc *desc = irq_to_desc(irq); + unsigned long flags; + + raw_spin_lock_irqsave(&desc->lock, flags); + desc_set_defaults(irq, desc, desc_node(desc)); + raw_spin_unlock_irqrestore(&desc->lock, flags); +} + +static inline int alloc_descs(unsigned int start, unsigned int cnt, int node) +{ + return start; +} #endif /* !CONFIG_SPARSE_IRQ */ +/* Dynamic interrupt handling */ + +/** + * irq_free_descs - free irq descriptors + * @from: Start of descriptor range + * @cnt: Number of consecutive irqs to free + */ +void irq_free_descs(unsigned int from, unsigned int cnt) +{ + unsigned long flags; + int i; + + if (from >= nr_irqs || (from + cnt) > nr_irqs) + return; + + for (i = 0; i < cnt; i++) + free_desc(from + i); + + raw_spin_lock_irqsave(&sparse_irq_lock, flags); + bitmap_clear(allocated_irqs, from, cnt); + raw_spin_unlock_irqrestore(&sparse_irq_lock, flags); +} + +/** + * irq_alloc_descs - allocate and initialize a range of irq descriptors + * @irq: Allocate for specific irq number if irq >= 0 + * @from: Start the search from this irq number + * @cnt: Number of consecutive irqs to allocate. + * @node: Preferred node on which the irq descriptor should be allocated + * + * Returns the first irq number or error code + */ +int __ref +irq_alloc_descs(int irq, unsigned int from, unsigned int cnt, int node) +{ + unsigned long flags; + int start, ret; + + if (!cnt) + return -EINVAL; + + raw_spin_lock_irqsave(&sparse_irq_lock, flags); + + start = bitmap_find_next_zero_area(allocated_irqs, nr_irqs, from, cnt, 0); + ret = -EEXIST; + if (irq >=0 && start != irq) + goto err; + + ret = -ENOMEM; + if (start >= nr_irqs) + goto err; + + bitmap_set(allocated_irqs, start, cnt); + raw_spin_unlock_irqrestore(&sparse_irq_lock, flags); + return alloc_descs(start, cnt, node); + +err: + raw_spin_unlock_irqrestore(&sparse_irq_lock, flags); + return ret; +} + +/* Statistics access */ void clear_kstat_irqs(struct irq_desc *desc) { memset(desc->kstat_irqs, 0, nr_cpu_ids * sizeof(*(desc->kstat_irqs))); -- cgit v1.2.3 From 13bfe99e09123ef5edb6acb81ba337d2db600b53 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Thu, 30 Sep 2010 02:46:07 +0200 Subject: genirq: Prepare proc for real sparse irq support /proc/irq never removes any entries, but when irq descriptors can be freed for real this is necessary. Otherwise we'd reference a freed descriptor in /proc/irq/N Signed-off-by: Thomas Gleixner Reviewed-by: Ingo Molnar --- kernel/irq/internals.h | 2 ++ kernel/irq/irqdesc.c | 2 ++ kernel/irq/proc.c | 18 ++++++++++++++++++ 3 files changed, 22 insertions(+) (limited to 'kernel') diff --git a/kernel/irq/internals.h b/kernel/irq/internals.h index e281e45fbb55..8eb01e379ccc 100644 --- a/kernel/irq/internals.h +++ b/kernel/irq/internals.h @@ -32,10 +32,12 @@ void replace_irq_desc(unsigned int irq, struct irq_desc *desc); #ifdef CONFIG_PROC_FS extern void register_irq_proc(unsigned int irq, struct irq_desc *desc); +extern void unregister_irq_proc(unsigned int irq, struct irq_desc *desc); extern void register_handler_proc(unsigned int irq, struct irqaction *action); extern void unregister_handler_proc(unsigned int irq, struct irqaction *action); #else static inline void register_irq_proc(unsigned int irq, struct irq_desc *desc) { } +static inline void unregister_irq_proc(unsigned int irq, struct irq_desc *desc) { } static inline void register_handler_proc(unsigned int irq, struct irqaction *action) { } static inline void unregister_handler_proc(unsigned int irq, diff --git a/kernel/irq/irqdesc.c b/kernel/irq/irqdesc.c index 4eea48b4f576..6312a2c83971 100644 --- a/kernel/irq/irqdesc.c +++ b/kernel/irq/irqdesc.c @@ -205,6 +205,8 @@ static void free_desc(unsigned int irq) struct irq_desc *desc = irq_to_desc(irq); unsigned long flags; + unregister_irq_proc(irq, desc); + raw_spin_lock_irqsave(&sparse_irq_lock, flags); delete_irq_desc(irq); raw_spin_unlock_irqrestore(&sparse_irq_lock, flags); diff --git a/kernel/irq/proc.c b/kernel/irq/proc.c index d9fddf918b41..01b1d3a88983 100644 --- a/kernel/irq/proc.c +++ b/kernel/irq/proc.c @@ -297,6 +297,24 @@ void register_irq_proc(unsigned int irq, struct irq_desc *desc) &irq_spurious_proc_fops, (void *)(long)irq); } +void unregister_irq_proc(unsigned int irq, struct irq_desc *desc) +{ + char name [MAX_NAMELEN]; + + if (!root_irq_dir || !desc->dir) + return; +#ifdef CONFIG_SMP + remove_proc_entry("smp_affinity", desc->dir); + remove_proc_entry("affinity_hint", desc->dir); + remove_proc_entry("node", desc->dir); +#endif + remove_proc_entry("spurious", desc->dir); + + memset(name, 0, MAX_NAMELEN); + sprintf(name, "%u", irq); + remove_proc_entry(name, root_irq_dir); +} + #undef MAX_NAMELEN void unregister_handler_proc(unsigned int irq, struct irqaction *action) -- cgit v1.2.3 From a98d24b71b6e229965f18dc00d28dc71cb8fe324 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Thu, 30 Sep 2010 10:45:07 +0200 Subject: genirq: Implement sane enumeration Use the allocator bitmap to lookup active interrupts. Signed-off-by: Thomas Gleixner Reviewed-by: Ingo Molnar --- kernel/irq/irqdesc.c | 11 +++++++++++ 1 file changed, 11 insertions(+) (limited to 'kernel') diff --git a/kernel/irq/irqdesc.c b/kernel/irq/irqdesc.c index 6312a2c83971..2e7e94ef64da 100644 --- a/kernel/irq/irqdesc.c +++ b/kernel/irq/irqdesc.c @@ -463,6 +463,17 @@ err: return ret; } +/** + * irq_get_next_irq - get next allocated irq number + * @offset: where to start the search + * + * Returns next irq number after offset or nr_irqs if none is found. + */ +unsigned int irq_get_next_irq(unsigned int offset) +{ + return find_next_bit(allocated_irqs, nr_irqs, offset); +} + /* Statistics access */ void clear_kstat_irqs(struct irq_desc *desc) { -- cgit v1.2.3 From 06f6c3399e9f9ff6eafc200e80f9226c3cee0eaf Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 12 Oct 2010 12:31:46 +0200 Subject: genirq: Implement irq reservation Mark a range of interrupts as allocated. In the SPARSE_IRQ=n case we need this to update the bitmap for the legacy irqs so the enumerator via irq_get_next_irq() works. Signed-off-by: Thomas Gleixner --- kernel/irq/irqdesc.c | 26 ++++++++++++++++++++++++++ 1 file changed, 26 insertions(+) (limited to 'kernel') diff --git a/kernel/irq/irqdesc.c b/kernel/irq/irqdesc.c index 2e7e94ef64da..35d9052901b9 100644 --- a/kernel/irq/irqdesc.c +++ b/kernel/irq/irqdesc.c @@ -463,6 +463,32 @@ err: return ret; } +/** + * irq_reserve_irqs - mark irqs allocated + * @from: mark from irq number + * @cnt: number of irqs to mark + * + * Returns 0 on success or an appropriate error code + */ +int irq_reserve_irqs(unsigned int from, unsigned int cnt) +{ + unsigned long flags; + unsigned int start; + int ret = 0; + + if (!cnt || (from + cnt) > nr_irqs) + return -EINVAL; + + raw_spin_lock_irqsave(&sparse_irq_lock, flags); + start = bitmap_find_next_zero_area(allocated_irqs, nr_irqs, from, cnt, 0); + if (start == from) + bitmap_set(allocated_irqs, start, cnt); + else + ret = -EEXIST; + raw_spin_unlock_irqrestore(&sparse_irq_lock, flags); + return ret; +} + /** * irq_get_next_irq - get next allocated irq number * @offset: where to start the search -- cgit v1.2.3 From aa99ec0f3f26bf2bcd0fa5176de93598427f1e5e Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Mon, 27 Sep 2010 20:02:56 +0200 Subject: genirq: Use sane sparse allocator Make irq_to_desc_alloc_node() a wrapper around the new allocator. Signed-off-by: Thomas Gleixner Reviewed-by: Ingo Molnar --- kernel/irq/irqdesc.c | 129 ++++++++------------------------------------------- 1 file changed, 20 insertions(+), 109 deletions(-) (limited to 'kernel') diff --git a/kernel/irq/irqdesc.c b/kernel/irq/irqdesc.c index 35d9052901b9..7cbe4f93e2fb 100644 --- a/kernel/irq/irqdesc.c +++ b/kernel/irq/irqdesc.c @@ -51,7 +51,7 @@ static int alloc_masks(struct irq_desc *desc, gfp_t gfp, int node) static void desc_smp_init(struct irq_desc *desc, int node) { - desc->node = node; + desc->irq_data.node = node; cpumask_copy(desc->irq_data.affinity, irq_default_affinity); } @@ -84,13 +84,6 @@ static DECLARE_BITMAP(allocated_irqs, NR_IRQS); #ifdef CONFIG_SPARSE_IRQ -static struct irq_desc irq_desc_init = { - .status = IRQ_DEFAULT_INIT_FLAGS, - .handle_irq = handle_bad_irq, - .depth = 1, - .lock = __RAW_SPIN_LOCK_UNLOCKED(irq_desc_init.lock), -}; - void __ref init_kstat_irqs(struct irq_desc *desc, int node, int nr) { void *ptr; @@ -108,29 +101,6 @@ void __ref init_kstat_irqs(struct irq_desc *desc, int node, int nr) } } -static void init_one_irq_desc(int irq, struct irq_desc *desc, int node) -{ - memcpy(desc, &irq_desc_init, sizeof(struct irq_desc)); - - raw_spin_lock_init(&desc->lock); - desc->irq_data.irq = irq; -#ifdef CONFIG_SMP - desc->irq_data.node = node; -#endif - lockdep_set_class(&desc->lock, &irq_desc_lock_class); - init_kstat_irqs(desc, node, nr_cpu_ids); - if (!desc->kstat_irqs) { - printk(KERN_ERR "can not alloc kstat_irqs\n"); - BUG_ON(1); - } - if (!alloc_desc_masks(desc, node, false)) { - printk(KERN_ERR "can not alloc irq_desc cpumasks\n"); - BUG_ON(1); - } - init_desc_masks(desc); - arch_init_chip_data(desc, node); -} - static RADIX_TREE(irq_desc_tree, GFP_ATOMIC); static void irq_insert_desc(unsigned int irq, struct irq_desc *desc) @@ -171,8 +141,9 @@ static inline void free_masks(struct irq_desc *desc) { } static struct irq_desc *alloc_desc(int irq, int node) { + /* Temporary hack until we can switch to GFP_KERNEL */ + gfp_t gfp = gfp_allowed_mask == GFP_BOOT_MASK ? GFP_NOWAIT : GFP_ATOMIC; struct irq_desc *desc; - gfp_t gfp = GFP_KERNEL; desc = kzalloc_node(sizeof(*desc), gfp, node); if (!desc) @@ -226,6 +197,8 @@ static int alloc_descs(unsigned int start, unsigned int cnt, int node) desc = alloc_desc(start + i, node); if (!desc) goto err; + /* temporary until I fixed x86 madness */ + arch_init_chip_data(desc, node); raw_spin_lock_irqsave(&sparse_irq_lock, flags); irq_insert_desc(start + i, desc); raw_spin_unlock_irqrestore(&sparse_irq_lock, flags); @@ -242,23 +215,19 @@ err: return -ENOMEM; } -static struct irq_desc irq_desc_legacy[NR_IRQS_LEGACY] __cacheline_aligned_in_smp = { - [0 ... NR_IRQS_LEGACY-1] = { - .status = IRQ_DEFAULT_INIT_FLAGS, - .handle_irq = handle_bad_irq, - .depth = 1, - .lock = __RAW_SPIN_LOCK_UNLOCKED(irq_desc_init.lock), - } -}; +struct irq_desc * __ref irq_to_desc_alloc_node(unsigned int irq, int node) +{ + int res = irq_alloc_descs(irq, irq, 1, node); -static unsigned int *kstat_irqs_legacy; + if (res == -EEXIST || res == irq) + return irq_to_desc(irq); + return NULL; +} int __init early_irq_init(void) { + int i, node = first_online_node; struct irq_desc *desc; - int legacy_count; - int node; - int i; init_irq_default_affinity(); @@ -266,71 +235,14 @@ int __init early_irq_init(void) arch_probe_nr_irqs(); printk(KERN_INFO "NR_IRQS:%d nr_irqs:%d\n", NR_IRQS, nr_irqs); - desc = irq_desc_legacy; - legacy_count = ARRAY_SIZE(irq_desc_legacy); - node = first_online_node; - - /* allocate based on nr_cpu_ids */ - kstat_irqs_legacy = kzalloc_node(NR_IRQS_LEGACY * nr_cpu_ids * - sizeof(int), GFP_NOWAIT, node); - - irq_desc_init.irq_data.chip = &no_irq_chip; - - for (i = 0; i < legacy_count; i++) { - desc[i].irq_data.irq = i; - desc[i].irq_data.chip = &no_irq_chip; -#ifdef CONFIG_SMP - desc[i].irq_data.node = node; -#endif - desc[i].kstat_irqs = kstat_irqs_legacy + i * nr_cpu_ids; - lockdep_set_class(&desc[i].lock, &irq_desc_lock_class); - alloc_desc_masks(&desc[i], node, true); - init_desc_masks(&desc[i]); - irq_insert_desc(i, &desc[i]); + for (i = 0; i < NR_IRQS_LEGACY; i++) { + desc = alloc_desc(i, node); + set_bit(i, allocated_irqs); + irq_insert_desc(i, desc); } - return arch_early_irq_init(); } -struct irq_desc * __ref irq_to_desc_alloc_node(unsigned int irq, int node) -{ - struct irq_desc *desc; - unsigned long flags; - - if (irq >= nr_irqs) { - WARN(1, "irq (%d) >= nr_irqs (%d) in irq_to_desc_alloc\n", - irq, nr_irqs); - return NULL; - } - - desc = irq_to_desc(irq); - if (desc) - return desc; - - raw_spin_lock_irqsave(&sparse_irq_lock, flags); - - /* We have to check it to avoid races with another CPU */ - desc = irq_to_desc(irq); - if (desc) - goto out_unlock; - - desc = kzalloc_node(sizeof(*desc), GFP_ATOMIC, node); - - printk(KERN_DEBUG " alloc irq_desc for %d on node %d\n", irq, node); - if (!desc) { - printk(KERN_ERR "can not alloc irq_desc\n"); - BUG_ON(1); - } - init_one_irq_desc(irq, desc, node); - - irq_insert_desc(irq, desc); - -out_unlock: - raw_spin_unlock_irqrestore(&sparse_irq_lock, flags); - - return desc; -} - #else /* !CONFIG_SPARSE_IRQ */ struct irq_desc irq_desc[NR_IRQS] __cacheline_aligned_in_smp = { @@ -345,9 +257,8 @@ struct irq_desc irq_desc[NR_IRQS] __cacheline_aligned_in_smp = { static unsigned int kstat_irqs_all[NR_IRQS][NR_CPUS]; int __init early_irq_init(void) { + int count, i, node = first_online_node; struct irq_desc *desc; - int count; - int i; init_irq_default_affinity(); @@ -359,9 +270,9 @@ int __init early_irq_init(void) for (i = 0; i < count; i++) { desc[i].irq_data.irq = i; desc[i].irq_data.chip = &no_irq_chip; - alloc_desc_masks(&desc[i], 0, true); - init_desc_masks(&desc[i]); desc[i].kstat_irqs = kstat_irqs_all[i]; + alloc_masks(desc + i, GFP_KERNEL, node); + desc_smp_init(desc + i, node); lockdep_set_class(&desc[i].lock, &irq_desc_lock_class); } return arch_early_irq_init(); -- cgit v1.2.3 From b683de2b3cb17bb10fa6fd4af614dc75b5749fe0 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Mon, 27 Sep 2010 20:55:03 +0200 Subject: genirq: Query arch for number of early descriptors sparse irq sets up NR_IRQS_LEGACY irq descriptors and archs then go ahead and allocate more. Use the unused return value of arch_probe_nr_irqs() to let the architecture return the number of early allocations. Fix up all users. Signed-off-by: Thomas Gleixner Reviewed-by: Ingo Molnar --- kernel/irq/irqdesc.c | 10 +++++----- kernel/softirq.c | 4 +++- 2 files changed, 8 insertions(+), 6 deletions(-) (limited to 'kernel') diff --git a/kernel/irq/irqdesc.c b/kernel/irq/irqdesc.c index 7cbe4f93e2fb..a1fbd1d347af 100644 --- a/kernel/irq/irqdesc.c +++ b/kernel/irq/irqdesc.c @@ -226,16 +226,16 @@ struct irq_desc * __ref irq_to_desc_alloc_node(unsigned int irq, int node) int __init early_irq_init(void) { - int i, node = first_online_node; + int i, initcnt, node = first_online_node; struct irq_desc *desc; init_irq_default_affinity(); - /* initialize nr_irqs based on nr_cpu_ids */ - arch_probe_nr_irqs(); - printk(KERN_INFO "NR_IRQS:%d nr_irqs:%d\n", NR_IRQS, nr_irqs); + /* Let arch update nr_irqs and return the nr of preallocated irqs */ + initcnt = arch_probe_nr_irqs(); + printk(KERN_INFO "NR_IRQS:%d nr_irqs:%d %d\n", NR_IRQS, nr_irqs, initcnt); - for (i = 0; i < NR_IRQS_LEGACY; i++) { + for (i = 0; i < initcnt; i++) { desc = alloc_desc(i, node); set_bit(i, allocated_irqs); irq_insert_desc(i, desc); diff --git a/kernel/softirq.c b/kernel/softirq.c index 07b4f1b1a73a..14a7b80b2cce 100644 --- a/kernel/softirq.c +++ b/kernel/softirq.c @@ -886,9 +886,10 @@ int __init __weak early_irq_init(void) return 0; } +#ifdef CONFIG_GENERIC_HARDIRQS int __init __weak arch_probe_nr_irqs(void) { - return 0; + return NR_IRQS_LEGACY; } int __init __weak arch_early_irq_init(void) @@ -900,3 +901,4 @@ int __weak arch_init_chip_data(struct irq_desc *desc, int node) { return 0; } +#endif -- cgit v1.2.3 From b7d0d8258a9f71949b810e0f82a3d75088f4d364 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Wed, 29 Sep 2010 18:44:23 +0200 Subject: genirq: Remove arch_init_chip_data() This function should have not been there in the first place. Signed-off-by: Thomas Gleixner Reviewed-by: Ingo Molnar --- kernel/irq/irqdesc.c | 2 -- kernel/softirq.c | 5 ----- 2 files changed, 7 deletions(-) (limited to 'kernel') diff --git a/kernel/irq/irqdesc.c b/kernel/irq/irqdesc.c index a1fbd1d347af..6c71f8ea5d7d 100644 --- a/kernel/irq/irqdesc.c +++ b/kernel/irq/irqdesc.c @@ -197,8 +197,6 @@ static int alloc_descs(unsigned int start, unsigned int cnt, int node) desc = alloc_desc(start + i, node); if (!desc) goto err; - /* temporary until I fixed x86 madness */ - arch_init_chip_data(desc, node); raw_spin_lock_irqsave(&sparse_irq_lock, flags); irq_insert_desc(start + i, desc); raw_spin_unlock_irqrestore(&sparse_irq_lock, flags); diff --git a/kernel/softirq.c b/kernel/softirq.c index 14a7b80b2cce..d19b1c9aa7c5 100644 --- a/kernel/softirq.c +++ b/kernel/softirq.c @@ -896,9 +896,4 @@ int __init __weak arch_early_irq_init(void) { return 0; } - -int __weak arch_init_chip_data(struct irq_desc *desc, int node) -{ - return 0; -} #endif -- cgit v1.2.3 From b7b29338dc7111ed8bd4d6555d84afae13ebe752 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Wed, 29 Sep 2010 18:46:55 +0200 Subject: genirq: Sanitize dynamic irq handling Use the cleanup functions of the dynamic allocator. No need to have separate implementations. Signed-off-by: Thomas Gleixner Reviewed-by: Ingo Molnar --- kernel/irq/chip.c | 102 ------------------------------------------------- kernel/irq/internals.h | 1 - kernel/irq/irqdesc.c | 41 +++++++++++--------- 3 files changed, 23 insertions(+), 121 deletions(-) (limited to 'kernel') diff --git a/kernel/irq/chip.c b/kernel/irq/chip.c index 3405761d6224..baa5c4acad83 100644 --- a/kernel/irq/chip.c +++ b/kernel/irq/chip.c @@ -18,108 +18,6 @@ #include "internals.h" -static void dynamic_irq_init_x(unsigned int irq, bool keep_chip_data) -{ - struct irq_desc *desc; - unsigned long flags; - - desc = irq_to_desc(irq); - if (!desc) { - WARN(1, KERN_ERR "Trying to initialize invalid IRQ%d\n", irq); - return; - } - - /* Ensure we don't have left over values from a previous use of this irq */ - raw_spin_lock_irqsave(&desc->lock, flags); - desc->status = IRQ_DEFAULT_INIT_FLAGS; - desc->irq_data.chip = &no_irq_chip; - desc->handle_irq = handle_bad_irq; - desc->depth = 1; - desc->irq_data.msi_desc = NULL; - desc->irq_data.handler_data = NULL; - if (!keep_chip_data) - desc->irq_data.chip_data = NULL; - desc->action = NULL; - desc->irq_count = 0; - desc->irqs_unhandled = 0; -#ifdef CONFIG_SMP - cpumask_setall(desc->irq_data.affinity); -#ifdef CONFIG_GENERIC_PENDING_IRQ - cpumask_clear(desc->pending_mask); -#endif -#endif - raw_spin_unlock_irqrestore(&desc->lock, flags); -} - -/** - * dynamic_irq_init - initialize a dynamically allocated irq - * @irq: irq number to initialize - */ -void dynamic_irq_init(unsigned int irq) -{ - dynamic_irq_init_x(irq, false); -} - -/** - * dynamic_irq_init_keep_chip_data - initialize a dynamically allocated irq - * @irq: irq number to initialize - * - * does not set irq_to_desc(irq)->irq_data.chip_data to NULL - */ -void dynamic_irq_init_keep_chip_data(unsigned int irq) -{ - dynamic_irq_init_x(irq, true); -} - -static void dynamic_irq_cleanup_x(unsigned int irq, bool keep_chip_data) -{ - struct irq_desc *desc = irq_to_desc(irq); - unsigned long flags; - - if (!desc) { - WARN(1, KERN_ERR "Trying to cleanup invalid IRQ%d\n", irq); - return; - } - - raw_spin_lock_irqsave(&desc->lock, flags); - if (desc->action) { - raw_spin_unlock_irqrestore(&desc->lock, flags); - WARN(1, KERN_ERR "Destroying IRQ%d without calling free_irq\n", - irq); - return; - } - desc->irq_data.msi_desc = NULL; - desc->irq_data.handler_data = NULL; - if (!keep_chip_data) - desc->irq_data.chip_data = NULL; - desc->handle_irq = handle_bad_irq; - desc->irq_data.chip = &no_irq_chip; - desc->name = NULL; - clear_kstat_irqs(desc); - raw_spin_unlock_irqrestore(&desc->lock, flags); -} - -/** - * dynamic_irq_cleanup - cleanup a dynamically allocated irq - * @irq: irq number to initialize - */ -void dynamic_irq_cleanup(unsigned int irq) -{ - dynamic_irq_cleanup_x(irq, false); -} - -/** - * dynamic_irq_cleanup_keep_chip_data - cleanup a dynamically allocated irq - * @irq: irq number to initialize - * - * does not set irq_to_desc(irq)->irq_data.chip_data to NULL - */ -void dynamic_irq_cleanup_keep_chip_data(unsigned int irq) -{ - dynamic_irq_cleanup_x(irq, true); -} - - /** * set_irq_chip - set the irq chip for an irq * @irq: irq number diff --git a/kernel/irq/internals.h b/kernel/irq/internals.h index 8eb01e379ccc..f444203a772d 100644 --- a/kernel/irq/internals.h +++ b/kernel/irq/internals.h @@ -20,7 +20,6 @@ extern void __enable_irq(struct irq_desc *desc, unsigned int irq, bool resume); extern struct lock_class_key irq_desc_lock_class; extern void init_kstat_irqs(struct irq_desc *desc, int node, int nr); -extern void clear_kstat_irqs(struct irq_desc *desc); extern raw_spinlock_t sparse_irq_lock; /* Resending of interrupts :*/ diff --git a/kernel/irq/irqdesc.c b/kernel/irq/irqdesc.c index 6c71f8ea5d7d..c9d5a1c12874 100644 --- a/kernel/irq/irqdesc.c +++ b/kernel/irq/irqdesc.c @@ -53,12 +53,21 @@ static void desc_smp_init(struct irq_desc *desc, int node) { desc->irq_data.node = node; cpumask_copy(desc->irq_data.affinity, irq_default_affinity); +#ifdef CONFIG_GENERIC_PENDING_IRQ + cpumask_clear(desc->pending_mask); +#endif +} + +static inline int desc_node(struct irq_desc *desc) +{ + return desc->irq_data.node; } #else static inline int alloc_masks(struct irq_desc *desc, gfp_t gfp, int node) { return 0; } static inline void desc_smp_init(struct irq_desc *desc, int node) { } +static inline int desc_node(struct irq_desc *desc) { return 0; } #endif static void desc_set_defaults(unsigned int irq, struct irq_desc *desc, int node) @@ -71,6 +80,8 @@ static void desc_set_defaults(unsigned int irq, struct irq_desc *desc, int node) desc->status = IRQ_DEFAULT_INIT_FLAGS; desc->handle_irq = handle_bad_irq; desc->depth = 1; + desc->irq_count = 0; + desc->irqs_unhandled = 0; desc->name = NULL; memset(desc->kstat_irqs, 0, nr_cpu_ids * sizeof(*(desc->kstat_irqs))); desc_smp_init(desc, node); @@ -286,23 +297,9 @@ struct irq_desc *irq_to_desc_alloc_node(unsigned int irq, int node) return irq_to_desc(irq); } -#ifdef CONFIG_SMP -static inline int desc_node(struct irq_desc *desc) -{ - return desc->irq_data.node; -} -#else -static inline int desc_node(struct irq_desc *desc) { return 0; } -#endif - static void free_desc(unsigned int irq) { - struct irq_desc *desc = irq_to_desc(irq); - unsigned long flags; - - raw_spin_lock_irqsave(&desc->lock, flags); - desc_set_defaults(irq, desc, desc_node(desc)); - raw_spin_unlock_irqrestore(&desc->lock, flags); + dynamic_irq_cleanup(irq); } static inline int alloc_descs(unsigned int start, unsigned int cnt, int node) @@ -409,10 +406,18 @@ unsigned int irq_get_next_irq(unsigned int offset) return find_next_bit(allocated_irqs, nr_irqs, offset); } -/* Statistics access */ -void clear_kstat_irqs(struct irq_desc *desc) +/** + * dynamic_irq_cleanup - cleanup a dynamically allocated irq + * @irq: irq number to initialize + */ +void dynamic_irq_cleanup(unsigned int irq) { - memset(desc->kstat_irqs, 0, nr_cpu_ids * sizeof(*(desc->kstat_irqs))); + struct irq_desc *desc = irq_to_desc(irq); + unsigned long flags; + + raw_spin_lock_irqsave(&desc->lock, flags); + desc_set_defaults(irq, desc, desc_node(desc)); + raw_spin_unlock_irqrestore(&desc->lock, flags); } unsigned int kstat_irqs_cpu(unsigned int irq, int cpu) -- cgit v1.2.3 From 78f90d91f395cd0dc1ef3f21e0c5cd6fd50d202c Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Wed, 29 Sep 2010 17:18:47 +0200 Subject: genirq: Remove the now unused sparse irq leftovers The move_irq_desc() function was only used due to the problem that the allocator did not free the old descriptors. So the descriptors had to be moved in create_irq_nr(). That's history. The code would have never been able to move active interrupt descriptors on affinity settings. That can be done in a completely different way w/o all this horror. Remove all of it. Signed-off-by: Thomas Gleixner Reviewed-by: Ingo Molnar --- kernel/irq/Kconfig | 5 -- kernel/irq/Makefile | 1 - kernel/irq/internals.h | 102 --------------------------------------- kernel/irq/irqdesc.c | 30 +----------- kernel/irq/numa_migrate.c | 120 ---------------------------------------------- 5 files changed, 2 insertions(+), 256 deletions(-) delete mode 100644 kernel/irq/numa_migrate.c (limited to 'kernel') diff --git a/kernel/irq/Kconfig b/kernel/irq/Kconfig index a42c0191d71a..31d766bf5d2e 100644 --- a/kernel/irq/Kconfig +++ b/kernel/irq/Kconfig @@ -26,11 +26,6 @@ config GENERIC_IRQ_PROBE config GENERIC_PENDING_IRQ def_bool n -if SPARSE_IRQ && NUMA -config NUMA_IRQ_DESC - def_bool n -endif - config AUTO_IRQ_AFFINITY def_bool n diff --git a/kernel/irq/Makefile b/kernel/irq/Makefile index 1eaab0da56db..54329cd7b3ee 100644 --- a/kernel/irq/Makefile +++ b/kernel/irq/Makefile @@ -3,5 +3,4 @@ obj-y := irqdesc.o handle.o manage.o spurious.o resend.o chip.o dummychip.o devr obj-$(CONFIG_GENERIC_IRQ_PROBE) += autoprobe.o obj-$(CONFIG_PROC_FS) += proc.o obj-$(CONFIG_GENERIC_PENDING_IRQ) += migration.o -obj-$(CONFIG_NUMA_IRQ_DESC) += numa_migrate.o obj-$(CONFIG_PM_SLEEP) += pm.o diff --git a/kernel/irq/internals.h b/kernel/irq/internals.h index f444203a772d..4571ae7e085a 100644 --- a/kernel/irq/internals.h +++ b/kernel/irq/internals.h @@ -18,17 +18,11 @@ extern int __irq_set_trigger(struct irq_desc *desc, unsigned int irq, extern void __disable_irq(struct irq_desc *desc, unsigned int irq, bool susp); extern void __enable_irq(struct irq_desc *desc, unsigned int irq, bool resume); -extern struct lock_class_key irq_desc_lock_class; extern void init_kstat_irqs(struct irq_desc *desc, int node, int nr); -extern raw_spinlock_t sparse_irq_lock; /* Resending of interrupts :*/ void check_irq_resend(struct irq_desc *desc, unsigned int irq); -#ifdef CONFIG_SPARSE_IRQ -void replace_irq_desc(unsigned int irq, struct irq_desc *desc); -#endif - #ifdef CONFIG_PROC_FS extern void register_irq_proc(unsigned int irq, struct irq_desc *desc); extern void unregister_irq_proc(unsigned int irq, struct irq_desc *desc); @@ -110,99 +104,3 @@ static inline void print_irq_desc(unsigned int irq, struct irq_desc *desc) #undef P -/* Stuff below will be cleaned up after the sparse allocator is done */ - -#ifdef CONFIG_SMP -/** - * alloc_desc_masks - allocate cpumasks for irq_desc - * @desc: pointer to irq_desc struct - * @node: node which will be handling the cpumasks - * @boot: true if need bootmem - * - * Allocates affinity and pending_mask cpumask if required. - * Returns true if successful (or not required). - */ -static inline bool alloc_desc_masks(struct irq_desc *desc, int node, - bool boot) -{ - gfp_t gfp = GFP_ATOMIC; - - if (boot) - gfp = GFP_NOWAIT; - -#ifdef CONFIG_CPUMASK_OFFSTACK - if (!alloc_cpumask_var_node(&desc->irq_data.affinity, gfp, node)) - return false; - -#ifdef CONFIG_GENERIC_PENDING_IRQ - if (!alloc_cpumask_var_node(&desc->pending_mask, gfp, node)) { - free_cpumask_var(desc->irq_data.affinity); - return false; - } -#endif -#endif - return true; -} - -static inline void init_desc_masks(struct irq_desc *desc) -{ - cpumask_setall(desc->irq_data.affinity); -#ifdef CONFIG_GENERIC_PENDING_IRQ - cpumask_clear(desc->pending_mask); -#endif -} - -/** - * init_copy_desc_masks - copy cpumasks for irq_desc - * @old_desc: pointer to old irq_desc struct - * @new_desc: pointer to new irq_desc struct - * - * Insures affinity and pending_masks are copied to new irq_desc. - * If !CONFIG_CPUMASKS_OFFSTACK the cpumasks are embedded in the - * irq_desc struct so the copy is redundant. - */ - -static inline void init_copy_desc_masks(struct irq_desc *old_desc, - struct irq_desc *new_desc) -{ -#ifdef CONFIG_CPUMASK_OFFSTACK - cpumask_copy(new_desc->irq_data.affinity, old_desc->irq_data.affinity); - -#ifdef CONFIG_GENERIC_PENDING_IRQ - cpumask_copy(new_desc->pending_mask, old_desc->pending_mask); -#endif -#endif -} - -static inline void free_desc_masks(struct irq_desc *old_desc, - struct irq_desc *new_desc) -{ - free_cpumask_var(old_desc->irq_data.affinity); - -#ifdef CONFIG_GENERIC_PENDING_IRQ - free_cpumask_var(old_desc->pending_mask); -#endif -} - -#else /* !CONFIG_SMP */ - -static inline bool alloc_desc_masks(struct irq_desc *desc, int node, - bool boot) -{ - return true; -} - -static inline void init_desc_masks(struct irq_desc *desc) -{ -} - -static inline void init_copy_desc_masks(struct irq_desc *old_desc, - struct irq_desc *new_desc) -{ -} - -static inline void free_desc_masks(struct irq_desc *old_desc, - struct irq_desc *new_desc) -{ -} -#endif /* CONFIG_SMP */ diff --git a/kernel/irq/irqdesc.c b/kernel/irq/irqdesc.c index c9d5a1c12874..4f0b9c9d5c46 100644 --- a/kernel/irq/irqdesc.c +++ b/kernel/irq/irqdesc.c @@ -20,7 +20,7 @@ /* * lockdep: we want to handle all irq_desc locks as a single lock-class: */ -struct lock_class_key irq_desc_lock_class; +static struct lock_class_key irq_desc_lock_class; #if defined(CONFIG_SMP) && defined(CONFIG_GENERIC_HARDIRQS) static void __init init_irq_default_affinity(void) @@ -90,28 +90,11 @@ static void desc_set_defaults(unsigned int irq, struct irq_desc *desc, int node) int nr_irqs = NR_IRQS; EXPORT_SYMBOL_GPL(nr_irqs); -DEFINE_RAW_SPINLOCK(sparse_irq_lock); +static DEFINE_RAW_SPINLOCK(sparse_irq_lock); static DECLARE_BITMAP(allocated_irqs, NR_IRQS); #ifdef CONFIG_SPARSE_IRQ -void __ref init_kstat_irqs(struct irq_desc *desc, int node, int nr) -{ - void *ptr; - - ptr = kzalloc_node(nr * sizeof(*desc->kstat_irqs), - GFP_ATOMIC, node); - - /* - * don't overwite if can not get new one - * init_copy_kstat_irqs() could still use old one - */ - if (ptr) { - printk(KERN_DEBUG " alloc kstat_irqs on node %d\n", node); - desc->kstat_irqs = ptr; - } -} - static RADIX_TREE(irq_desc_tree, GFP_ATOMIC); static void irq_insert_desc(unsigned int irq, struct irq_desc *desc) @@ -124,15 +107,6 @@ struct irq_desc *irq_to_desc(unsigned int irq) return radix_tree_lookup(&irq_desc_tree, irq); } -void replace_irq_desc(unsigned int irq, struct irq_desc *desc) -{ - void **ptr; - - ptr = radix_tree_lookup_slot(&irq_desc_tree, irq); - if (ptr) - radix_tree_replace_slot(ptr, desc); -} - static void delete_irq_desc(unsigned int irq) { radix_tree_delete(&irq_desc_tree, irq); diff --git a/kernel/irq/numa_migrate.c b/kernel/irq/numa_migrate.c deleted file mode 100644 index e7f1f16402c1..000000000000 --- a/kernel/irq/numa_migrate.c +++ /dev/null @@ -1,120 +0,0 @@ -/* - * NUMA irq-desc migration code - * - * Migrate IRQ data structures (irq_desc, chip_data, etc.) over to - * the new "home node" of the IRQ. - */ - -#include -#include -#include -#include -#include -#include - -#include "internals.h" - -static void init_copy_kstat_irqs(struct irq_desc *old_desc, - struct irq_desc *desc, - int node, int nr) -{ - init_kstat_irqs(desc, node, nr); - - if (desc->kstat_irqs != old_desc->kstat_irqs) - memcpy(desc->kstat_irqs, old_desc->kstat_irqs, - nr * sizeof(*desc->kstat_irqs)); -} - -static void free_kstat_irqs(struct irq_desc *old_desc, struct irq_desc *desc) -{ - if (old_desc->kstat_irqs == desc->kstat_irqs) - return; - - kfree(old_desc->kstat_irqs); - old_desc->kstat_irqs = NULL; -} - -static bool init_copy_one_irq_desc(int irq, struct irq_desc *old_desc, - struct irq_desc *desc, int node) -{ - memcpy(desc, old_desc, sizeof(struct irq_desc)); - if (!alloc_desc_masks(desc, node, false)) { - printk(KERN_ERR "irq %d: can not get new irq_desc cpumask " - "for migration.\n", irq); - return false; - } - raw_spin_lock_init(&desc->lock); - desc->irq_data.node = node; - lockdep_set_class(&desc->lock, &irq_desc_lock_class); - init_copy_kstat_irqs(old_desc, desc, node, nr_cpu_ids); - init_copy_desc_masks(old_desc, desc); - arch_init_copy_chip_data(old_desc, desc, node); - return true; -} - -static void free_one_irq_desc(struct irq_desc *old_desc, struct irq_desc *desc) -{ - free_kstat_irqs(old_desc, desc); - free_desc_masks(old_desc, desc); - arch_free_chip_data(old_desc, desc); -} - -static struct irq_desc *__real_move_irq_desc(struct irq_desc *old_desc, - int node) -{ - struct irq_desc *desc; - unsigned int irq; - unsigned long flags; - - irq = old_desc->irq_data.irq; - - raw_spin_lock_irqsave(&sparse_irq_lock, flags); - - /* We have to check it to avoid races with another CPU */ - desc = irq_to_desc(irq); - - if (desc && old_desc != desc) - goto out_unlock; - - desc = kzalloc_node(sizeof(*desc), GFP_ATOMIC, node); - if (!desc) { - printk(KERN_ERR "irq %d: can not get new irq_desc " - "for migration.\n", irq); - /* still use old one */ - desc = old_desc; - goto out_unlock; - } - if (!init_copy_one_irq_desc(irq, old_desc, desc, node)) { - /* still use old one */ - kfree(desc); - desc = old_desc; - goto out_unlock; - } - - replace_irq_desc(irq, desc); - raw_spin_unlock_irqrestore(&sparse_irq_lock, flags); - - /* free the old one */ - free_one_irq_desc(old_desc, desc); - kfree(old_desc); - - return desc; - -out_unlock: - raw_spin_unlock_irqrestore(&sparse_irq_lock, flags); - - return desc; -} - -struct irq_desc *move_irq_desc(struct irq_desc *desc, int node) -{ - /* those static or target node is -1, do not move them */ - if (desc->irq_data.irq < NR_IRQS_LEGACY || node == -1) - return desc; - - if (desc->irq_data.node != node) - desc = __real_move_irq_desc(desc, node); - - return desc; -} - -- cgit v1.2.3 From a05a900a51c7622ebd8ddb41f14f8bf9db599d8d Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Fri, 8 Oct 2010 12:47:53 +0200 Subject: genirq: Make sparse_lock a mutex No callers from atomic regions. Signed-off-by: Thomas Gleixner Reviewed-by: Ingo Molnar --- kernel/irq/irqdesc.c | 33 ++++++++++++++------------------- 1 file changed, 14 insertions(+), 19 deletions(-) (limited to 'kernel') diff --git a/kernel/irq/irqdesc.c b/kernel/irq/irqdesc.c index 4f0b9c9d5c46..0e302f90d2ee 100644 --- a/kernel/irq/irqdesc.c +++ b/kernel/irq/irqdesc.c @@ -90,7 +90,7 @@ static void desc_set_defaults(unsigned int irq, struct irq_desc *desc, int node) int nr_irqs = NR_IRQS; EXPORT_SYMBOL_GPL(nr_irqs); -static DEFINE_RAW_SPINLOCK(sparse_irq_lock); +static DEFINE_MUTEX(sparse_irq_lock); static DECLARE_BITMAP(allocated_irqs, NR_IRQS); #ifdef CONFIG_SPARSE_IRQ @@ -159,13 +159,12 @@ err_desc: static void free_desc(unsigned int irq) { struct irq_desc *desc = irq_to_desc(irq); - unsigned long flags; unregister_irq_proc(irq, desc); - raw_spin_lock_irqsave(&sparse_irq_lock, flags); + mutex_lock(&sparse_irq_lock); delete_irq_desc(irq); - raw_spin_unlock_irqrestore(&sparse_irq_lock, flags); + mutex_unlock(&sparse_irq_lock); free_masks(desc); kfree(desc->kstat_irqs); @@ -175,16 +174,15 @@ static void free_desc(unsigned int irq) static int alloc_descs(unsigned int start, unsigned int cnt, int node) { struct irq_desc *desc; - unsigned long flags; int i; for (i = 0; i < cnt; i++) { desc = alloc_desc(start + i, node); if (!desc) goto err; - raw_spin_lock_irqsave(&sparse_irq_lock, flags); + mutex_lock(&sparse_irq_lock); irq_insert_desc(start + i, desc); - raw_spin_unlock_irqrestore(&sparse_irq_lock, flags); + mutex_unlock(&sparse_irq_lock); } return start; @@ -192,9 +190,9 @@ err: for (i--; i >= 0; i--) free_desc(start + i); - raw_spin_lock_irqsave(&sparse_irq_lock, flags); + mutex_lock(&sparse_irq_lock); bitmap_clear(allocated_irqs, start, cnt); - raw_spin_unlock_irqrestore(&sparse_irq_lock, flags); + mutex_unlock(&sparse_irq_lock); return -ENOMEM; } @@ -291,7 +289,6 @@ static inline int alloc_descs(unsigned int start, unsigned int cnt, int node) */ void irq_free_descs(unsigned int from, unsigned int cnt) { - unsigned long flags; int i; if (from >= nr_irqs || (from + cnt) > nr_irqs) @@ -300,9 +297,9 @@ void irq_free_descs(unsigned int from, unsigned int cnt) for (i = 0; i < cnt; i++) free_desc(from + i); - raw_spin_lock_irqsave(&sparse_irq_lock, flags); + mutex_lock(&sparse_irq_lock); bitmap_clear(allocated_irqs, from, cnt); - raw_spin_unlock_irqrestore(&sparse_irq_lock, flags); + mutex_unlock(&sparse_irq_lock); } /** @@ -317,13 +314,12 @@ void irq_free_descs(unsigned int from, unsigned int cnt) int __ref irq_alloc_descs(int irq, unsigned int from, unsigned int cnt, int node) { - unsigned long flags; int start, ret; if (!cnt) return -EINVAL; - raw_spin_lock_irqsave(&sparse_irq_lock, flags); + mutex_lock(&sparse_irq_lock); start = bitmap_find_next_zero_area(allocated_irqs, nr_irqs, from, cnt, 0); ret = -EEXIST; @@ -335,11 +331,11 @@ irq_alloc_descs(int irq, unsigned int from, unsigned int cnt, int node) goto err; bitmap_set(allocated_irqs, start, cnt); - raw_spin_unlock_irqrestore(&sparse_irq_lock, flags); + mutex_unlock(&sparse_irq_lock); return alloc_descs(start, cnt, node); err: - raw_spin_unlock_irqrestore(&sparse_irq_lock, flags); + mutex_unlock(&sparse_irq_lock); return ret; } @@ -352,20 +348,19 @@ err: */ int irq_reserve_irqs(unsigned int from, unsigned int cnt) { - unsigned long flags; unsigned int start; int ret = 0; if (!cnt || (from + cnt) > nr_irqs) return -EINVAL; - raw_spin_lock_irqsave(&sparse_irq_lock, flags); + mutex_lock(&sparse_irq_lock); start = bitmap_find_next_zero_area(allocated_irqs, nr_irqs, from, cnt, 0); if (start == from) bitmap_set(allocated_irqs, start, cnt); else ret = -EEXIST; - raw_spin_unlock_irqrestore(&sparse_irq_lock, flags); + mutex_unlock(&sparse_irq_lock); return ret; } -- cgit v1.2.3 From baa0d233afe765daa6dc01ff233aea8c5944f534 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 5 Oct 2010 15:14:35 +0200 Subject: genirq: Switch sparse_irq allocator to GFP_KERNEL The allocator functions are now called outside of preempt disabled regions. Switch to GFP_KERNEL. Signed-off-by: Thomas Gleixner Reviewed-by: Ingo Molnar --- kernel/irq/irqdesc.c | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/irq/irqdesc.c b/kernel/irq/irqdesc.c index 0e302f90d2ee..f6f660cef7af 100644 --- a/kernel/irq/irqdesc.c +++ b/kernel/irq/irqdesc.c @@ -95,7 +95,7 @@ static DECLARE_BITMAP(allocated_irqs, NR_IRQS); #ifdef CONFIG_SPARSE_IRQ -static RADIX_TREE(irq_desc_tree, GFP_ATOMIC); +static RADIX_TREE(irq_desc_tree, GFP_KERNEL); static void irq_insert_desc(unsigned int irq, struct irq_desc *desc) { @@ -126,9 +126,8 @@ static inline void free_masks(struct irq_desc *desc) { } static struct irq_desc *alloc_desc(int irq, int node) { - /* Temporary hack until we can switch to GFP_KERNEL */ - gfp_t gfp = gfp_allowed_mask == GFP_BOOT_MASK ? GFP_NOWAIT : GFP_ATOMIC; struct irq_desc *desc; + gfp_t gfp = GFP_KERNEL; desc = kzalloc_node(sizeof(*desc), gfp, node); if (!desc) -- cgit v1.2.3 From 5b8c4f23c54edda640a4850bc9b81dee70acb525 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 7 Sep 2010 14:33:43 +0000 Subject: printk: Make console_sem a semaphore not a pseudo mutex It needs to be investigated whether it can be replaced by a real mutex, but that needs more thought. Signed-off-by: Thomas Gleixner Cc: Peter Zijlstra Cc: Christoph Hellwig LKML-Reference: <20100907125057.179587334@linutronix.de> --- kernel/printk.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/printk.c b/kernel/printk.c index 8fe465ac008a..2531017795f6 100644 --- a/kernel/printk.c +++ b/kernel/printk.c @@ -85,7 +85,7 @@ EXPORT_SYMBOL(oops_in_progress); * provides serialisation for access to the entire console * driver system. */ -static DECLARE_MUTEX(console_sem); +static DEFINE_SEMAPHORE(console_sem); struct console *console_drivers; EXPORT_SYMBOL_GPL(console_drivers); @@ -556,7 +556,7 @@ static void zap_locks(void) /* If a crash is occurring, make sure we can't deadlock */ spin_lock_init(&logbuf_lock); /* And make sure that we print immediately */ - init_MUTEX(&console_sem); + sema_init(&console_sem, 1); } #if defined(CONFIG_PRINTK_TIME) -- cgit v1.2.3 From d01343244abdedd18303d0323b518ed9cdcb1988 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Tue, 12 Oct 2010 12:06:43 -0400 Subject: ring-buffer: Fix typo of time extends per page Time stamps for the ring buffer are created by the difference between two events. Each page of the ring buffer holds a full 64 bit timestamp. Each event has a 27 bit delta stamp from the last event. The unit of time is nanoseconds, so 27 bits can hold ~134 milliseconds. If two events happen more than 134 milliseconds apart, a time extend is inserted to add more bits for the delta. The time extend has 59 bits, which is good for ~18 years. Currently the time extend is committed separately from the event. If an event is discarded before it is committed, due to filtering, the time extend still exists. If all events are being filtered, then after ~134 milliseconds a new time extend will be added to the buffer. This can only happen till the end of the page. Since each page holds a full timestamp, there is no reason to add a time extend to the beginning of a page. Time extends can only fill a page that has actual data at the beginning, so there is no fear that time extends will fill more than a page without any data. When reading an event, a loop is made to skip over time extends since they are only used to maintain the time stamp and are never given to the caller. As a paranoid check to prevent the loop running forever, with the knowledge that time extends may only fill a page, a check is made that tests the iteration of the loop, and if the iteration is more than the number of time extends that can fit in a page a warning is printed and the ring buffer is disabled (all of ftrace is also disabled with it). There is another event type that is called a TIMESTAMP which can hold 64 bits of data in the theoretical case that two events happen 18 years apart. This code has not been implemented, but the name of this event exists, as well as the structure for it. The size of a TIMESTAMP is 16 bytes, where as a time extend is only 8 bytes. The macro used to calculate how many time extends can fit on a page used the TIMESTAMP size instead of the time extend size cutting the amount in half. The following test case can easily trigger the warning since we only need to have half the page filled with time extends to trigger the warning: # cd /sys/kernel/debug/tracing/ # echo function > current_tracer # echo 'common_pid < 0' > events/ftrace/function/filter # echo > trace # echo 1 > trace_marker # sleep 120 # cat trace Enabling the function tracer and then setting the filter to only trace functions where the process id is negative (no events), then clearing the trace buffer to ensure that we have nothing in the buffer, then write to trace_marker to add an event to the beginning of a page, sleep for 2 minutes (only 35 seconds is probably needed, but this guarantees the bug), and then finally reading the trace which will trigger the bug. This patch fixes the typo and prevents the false positive of that warning. Reported-by: Hans J. Koch Tested-by: Hans J. Koch Cc: Thomas Gleixner Cc: Stable Kernel Signed-off-by: Steven Rostedt --- kernel/trace/ring_buffer.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c index 492197e2f86c..bca96377fd4e 100644 --- a/kernel/trace/ring_buffer.c +++ b/kernel/trace/ring_buffer.c @@ -405,7 +405,7 @@ static inline int test_time_stamp(u64 delta) #define BUF_MAX_DATA_SIZE (BUF_PAGE_SIZE - (sizeof(u32) * 2)) /* Max number of timestamps that can fit on a page */ -#define RB_TIMESTAMPS_PER_PAGE (BUF_PAGE_SIZE / RB_LEN_TIME_STAMP) +#define RB_TIMESTAMPS_PER_PAGE (BUF_PAGE_SIZE / RB_LEN_TIME_EXTEND) int ring_buffer_print_page_header(struct trace_seq *s) { -- cgit v1.2.3 From c0a19ebc018222ffd1dd93af5b53d9efd779c19b Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 12 Oct 2010 21:58:27 +0200 Subject: genirq: Fix CONFIG_GENIRQ_NO_DEPRECATED=y build This option can be set to verify the full conversion to the new chip functions. Fix the fallout of the patch rework, so the core code compiles and works with it. Signed-off-by: Thomas Gleixner --- kernel/irq/dummychip.c | 2 +- kernel/irq/irqdesc.c | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/irq/dummychip.c b/kernel/irq/dummychip.c index 918dea9de9ea..20dc5474947e 100644 --- a/kernel/irq/dummychip.c +++ b/kernel/irq/dummychip.c @@ -31,7 +31,7 @@ static unsigned int noop_ret(struct irq_data *data) return 0; } -#ifndef CONFIG_GENERIC_HARDIRQS_NO_CRUFT +#ifndef CONFIG_GENERIC_HARDIRQS_NO_DEPRECATED static void compat_noop(unsigned int irq) { } #define END_INIT .end = compat_noop #else diff --git a/kernel/irq/irqdesc.c b/kernel/irq/irqdesc.c index f6f660cef7af..9d917ff72675 100644 --- a/kernel/irq/irqdesc.c +++ b/kernel/irq/irqdesc.c @@ -118,7 +118,7 @@ static void free_masks(struct irq_desc *desc) #ifdef CONFIG_GENERIC_PENDING_IRQ free_cpumask_var(desc->pending_mask); #endif - free_cpumask_var(desc->affinity); + free_cpumask_var(desc->irq_data.affinity); } #else static inline void free_masks(struct irq_desc *desc) { } -- cgit v1.2.3 From 14cae9bd2faf6d0d75702c2e107e75207bcdfec1 Mon Sep 17 00:00:00 2001 From: Borislav Petkov Date: Wed, 29 Sep 2010 10:08:23 +0200 Subject: tracing: Fix function-graph build warning on 32-bit MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Fix kernel/trace/trace_functions_graph.c: In function ‘trace_print_graph_duration’: kernel/trace/trace_functions_graph.c:652: warning: comparison of distinct pointer types lacks a cast when building 36-rc6 on a 32-bit due to the strict type check failing in the min() macro. Signed-off-by: Borislav Petkov Cc: Chase Douglas Cc: Steven Rostedt Cc: Ingo Molnar LKML-Reference: <20100929080823.GA13595@liondog.tnic> Signed-off-by: Frederic Weisbecker --- kernel/trace/trace_functions_graph.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace_functions_graph.c b/kernel/trace/trace_functions_graph.c index 02c708ae0d42..ef49e9370b25 100644 --- a/kernel/trace/trace_functions_graph.c +++ b/kernel/trace/trace_functions_graph.c @@ -665,8 +665,9 @@ trace_print_graph_duration(unsigned long long duration, struct trace_seq *s) /* Print nsecs (we don't want to exceed 7 numbers) */ if (len < 7) { - snprintf(nsecs_str, min(sizeof(nsecs_str), 8UL - len), "%03lu", - nsecs_rem); + size_t slen = min_t(size_t, sizeof(nsecs_str), 8UL - len); + + snprintf(nsecs_str, slen, "%03lu", nsecs_rem); ret = trace_seq_printf(s, ".%s", nsecs_str); if (!ret) return TRACE_TYPE_PARTIAL_LINE; -- cgit v1.2.3 From fd02e6f7ae085840d43d780149fcf95a614eca5e Mon Sep 17 00:00:00 2001 From: Masami Hiramatsu Date: Thu, 14 Oct 2010 12:10:24 +0900 Subject: kprobes: Fix selftest to clear flags field for reusing probes Fix selftest to clear flags field for reusing probes because the flags field can be modified by Kprobes. This also set NULL to kprobe.addr instead of 0. Signed-off-by: Masami Hiramatsu Cc: Rusty Russell Cc: Ananth N Mavinakayanahalli Cc: 2nddept-manager@sdl.hitachi.co.jp LKML-Reference: <20101014031024.4100.50107.stgit@ltc236.sdl.hitachi.co.jp> Signed-off-by: Ingo Molnar --- kernel/test_kprobes.c | 12 +++++++++--- 1 file changed, 9 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/test_kprobes.c b/kernel/test_kprobes.c index 4f104515a19b..f8b11a283171 100644 --- a/kernel/test_kprobes.c +++ b/kernel/test_kprobes.c @@ -115,7 +115,9 @@ static int test_kprobes(void) int ret; struct kprobe *kps[2] = {&kp, &kp2}; - kp.addr = 0; /* addr should be cleard for reusing kprobe. */ + /* addr and flags should be cleard for reusing kprobe. */ + kp.addr = NULL; + kp.flags = 0; ret = register_kprobes(kps, 2); if (ret < 0) { printk(KERN_ERR "Kprobe smoke test failed: " @@ -210,7 +212,9 @@ static int test_jprobes(void) int ret; struct jprobe *jps[2] = {&jp, &jp2}; - jp.kp.addr = 0; /* addr should be cleard for reusing kprobe. */ + /* addr and flags should be cleard for reusing kprobe. */ + jp.kp.addr = NULL; + jp.kp.flags = 0; ret = register_jprobes(jps, 2); if (ret < 0) { printk(KERN_ERR "Kprobe smoke test failed: " @@ -323,7 +327,9 @@ static int test_kretprobes(void) int ret; struct kretprobe *rps[2] = {&rp, &rp2}; - rp.kp.addr = 0; /* addr should be cleard for reusing kprobe. */ + /* addr and flags should be cleard for reusing kprobe. */ + rp.kp.addr = NULL; + rp.kp.flags = 0; ret = register_kretprobes(rps, 2); if (ret < 0) { printk(KERN_ERR "Kprobe smoke test failed: " -- cgit v1.2.3 From fb62db2ba943b1683f1d7181bb2988fce4c60870 Mon Sep 17 00:00:00 2001 From: Randy Dunlap Date: Wed, 13 Oct 2010 11:02:34 -0700 Subject: futex: Fix kernel-doc notation & typos Convert futex_requeue() function parameters to use @name kernel-doc notation and add @fshared & @cmpval to prevent kernel-doc warnings. Add @list to struct futex_q. Fix a few typos. Signed-off-by: Randy Dunlap Acked-by: Rusty Russell LKML-Reference: <20101013110234.89b06043.randy.dunlap@oracle.com> Signed-off-by: Ingo Molnar --- kernel/futex.c | 25 ++++++++++++++----------- 1 file changed, 14 insertions(+), 11 deletions(-) (limited to 'kernel') diff --git a/kernel/futex.c b/kernel/futex.c index 92a31d4cd564..9b9fda73ba2e 100644 --- a/kernel/futex.c +++ b/kernel/futex.c @@ -91,6 +91,7 @@ struct futex_pi_state { /** * struct futex_q - The hashed futex queue entry, one per waiting task + * @list: priority-sorted list of tasks waiting on this futex * @task: the task waiting on the futex * @lock_ptr: the hash bucket lock * @key: the key the futex is hashed on @@ -104,7 +105,7 @@ struct futex_pi_state { * * A futex_q has a woken state, just like tasks have TASK_RUNNING. * It is considered woken when plist_node_empty(&q->list) || q->lock_ptr == 0. - * The order of wakup is always to make the first condition true, then + * The order of wakeup is always to make the first condition true, then * the second. * * PI futexes are typically woken before they are removed from the hash list via @@ -295,7 +296,7 @@ void put_futex_key(int fshared, union futex_key *key) * Slow path to fixup the fault we just took in the atomic write * access to @uaddr. * - * We have no generic implementation of a non destructive write to the + * We have no generic implementation of a non-destructive write to the * user address. We know that we faulted in the atomic pagefault * disabled section so we can as well avoid the #PF overhead by * calling get_user_pages() right away. @@ -515,7 +516,7 @@ lookup_pi_state(u32 uval, struct futex_hash_bucket *hb, */ pi_state = this->pi_state; /* - * Userspace might have messed up non PI and PI futexes + * Userspace might have messed up non-PI and PI futexes */ if (unlikely(!pi_state)) return -EINVAL; @@ -736,8 +737,8 @@ static void wake_futex(struct futex_q *q) /* * We set q->lock_ptr = NULL _before_ we wake up the task. If - * a non futex wake up happens on another CPU then the task - * might exit and p would dereference a non existing task + * a non-futex wake up happens on another CPU then the task + * might exit and p would dereference a non-existing task * struct. Prevent this by holding a reference on p across the * wake up. */ @@ -1131,11 +1132,13 @@ static int futex_proxy_trylock_atomic(u32 __user *pifutex, /** * futex_requeue() - Requeue waiters from uaddr1 to uaddr2 - * uaddr1: source futex user address - * uaddr2: target futex user address - * nr_wake: number of waiters to wake (must be 1 for requeue_pi) - * nr_requeue: number of waiters to requeue (0-INT_MAX) - * requeue_pi: if we are attempting to requeue from a non-pi futex to a + * @uaddr1: source futex user address + * @fshared: 0 for a PROCESS_PRIVATE futex, 1 for PROCESS_SHARED + * @uaddr2: target futex user address + * @nr_wake: number of waiters to wake (must be 1 for requeue_pi) + * @nr_requeue: number of waiters to requeue (0-INT_MAX) + * @cmpval: @uaddr1 expected value (or %NULL) + * @requeue_pi: if we are attempting to requeue from a non-pi futex to a * pi futex (pi to pi requeue is not supported) * * Requeue waiters on uaddr1 to uaddr2. In the requeue_pi case, try to acquire @@ -2651,7 +2654,7 @@ static int __init futex_init(void) * of the complex code paths. Also we want to prevent * registration of robust lists in that case. NULL is * guaranteed to fault and we get -EFAULT on functional - * implementation, the non functional ones will return + * implementation, the non-functional ones will return * -ENOSYS. */ curval = cmpxchg_futex_value_locked(NULL, 0, 0); -- cgit v1.2.3 From 864616ee6785d9fac7a2cd80c01a2da89579f2e4 Mon Sep 17 00:00:00 2001 From: Takuya Yoshikawa Date: Thu, 14 Oct 2010 16:09:13 +0900 Subject: sched: Comment updates: fix default latency and granularity numbers Targeted preemption latency and minimal preemption granularity for CPU-bound tasks have been changed. This patch updates the comments about these values. Signed-off-by: Takuya Yoshikawa Cc: Peter Zijlstra Cc: Mike Galbraith LKML-Reference: <20101014160913.eb24fef4.yoshikawa.takuya@oss.ntt.co.jp> Signed-off-by: Ingo Molnar --- kernel/sched_fair.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c index 623e9aceef8f..bf87192e97fe 100644 --- a/kernel/sched_fair.c +++ b/kernel/sched_fair.c @@ -25,7 +25,7 @@ /* * Targeted preemption latency for CPU-bound tasks: - * (default: 5ms * (1 + ilog(ncpus)), units: nanoseconds) + * (default: 6ms * (1 + ilog(ncpus)), units: nanoseconds) * * NOTE: this latency value is not the same as the concept of * 'timeslice length' - timeslices in CFS are of variable length @@ -52,7 +52,7 @@ enum sched_tunable_scaling sysctl_sched_tunable_scaling /* * Minimal preemption granularity for CPU-bound tasks: - * (default: 2 msec * (1 + ilog(ncpus)), units: nanoseconds) + * (default: 0.75 msec * (1 + ilog(ncpus)), units: nanoseconds) */ unsigned int sysctl_sched_min_granularity = 750000ULL; unsigned int normalized_sysctl_sched_min_granularity = 750000ULL; -- cgit v1.2.3 From f13d4f979c518119bba5439dd2364d76d31dcd3f Mon Sep 17 00:00:00 2001 From: Salman Qazi Date: Tue, 12 Oct 2010 07:25:19 -0700 Subject: hrtimer: Preserve timer state in remove_hrtimer() The race is described as follows: CPU X CPU Y remove_hrtimer // state & QUEUED == 0 timer->state = CALLBACK unlock timer base timer->f(n) //very long hrtimer_start lock timer base remove_hrtimer // no effect hrtimer_enqueue timer->state = CALLBACK | QUEUED unlock timer base hrtimer_start lock timer base remove_hrtimer mode = INACTIVE // CALLBACK bit lost! switch_hrtimer_base CALLBACK bit not set: timer->base changes to a different CPU. lock this CPU's timer base The bug was introduced with commit ca109491f (hrtimer: removing all ur callback modes) in 2.6.29 [ tglx: Feed new state via local variable and add a comment. ] Signed-off-by: Salman Qazi Cc: akpm@linux-foundation.org Cc: Peter Zijlstra LKML-Reference: <20101012142351.8485.21823.stgit@dungbeetle.mtv.corp.google.com> Signed-off-by: Thomas Gleixner Cc: stable@kernel.org --- kernel/hrtimer.c | 13 +++++++++++-- 1 file changed, 11 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/hrtimer.c b/kernel/hrtimer.c index 1decafbb6b1a..72206cf5c6cf 100644 --- a/kernel/hrtimer.c +++ b/kernel/hrtimer.c @@ -931,6 +931,7 @@ static inline int remove_hrtimer(struct hrtimer *timer, struct hrtimer_clock_base *base) { if (hrtimer_is_queued(timer)) { + unsigned long state; int reprogram; /* @@ -944,8 +945,13 @@ remove_hrtimer(struct hrtimer *timer, struct hrtimer_clock_base *base) debug_deactivate(timer); timer_stats_hrtimer_clear_start_info(timer); reprogram = base->cpu_base == &__get_cpu_var(hrtimer_bases); - __remove_hrtimer(timer, base, HRTIMER_STATE_INACTIVE, - reprogram); + /* + * We must preserve the CALLBACK state flag here, + * otherwise we could move the timer base in + * switch_hrtimer_base. + */ + state = timer->state & HRTIMER_STATE_CALLBACK; + __remove_hrtimer(timer, base, state, reprogram); return 1; } return 0; @@ -1231,6 +1237,9 @@ static void __run_hrtimer(struct hrtimer *timer, ktime_t *now) BUG_ON(timer->state != HRTIMER_STATE_CALLBACK); enqueue_hrtimer(timer, base); } + + WARN_ON_ONCE(!(timer->state & HRTIMER_STATE_CALLBACK)); + timer->state &= ~HRTIMER_STATE_CALLBACK; } -- cgit v1.2.3 From 72441cb1fd77d092f09ddfac748955703884c9a7 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Wed, 13 Oct 2010 17:12:30 -0400 Subject: ftrace/x86: Add support for C version of recordmcount This patch adds the support for the C version of recordmcount and compile times show ~ 12% improvement. After verifying this works, other archs can add: HAVE_C_MCOUNT_RECORD in its Kconfig and it will use the C version of recordmcount instead of the perl version. Cc: Cc: Michal Marek Cc: linux-kbuild@vger.kernel.org Cc: John Reiser Signed-off-by: Steven Rostedt --- kernel/trace/Kconfig | 5 +++++ 1 file changed, 5 insertions(+) (limited to 'kernel') diff --git a/kernel/trace/Kconfig b/kernel/trace/Kconfig index 538501c6ea50..df00fbbaf609 100644 --- a/kernel/trace/Kconfig +++ b/kernel/trace/Kconfig @@ -49,6 +49,11 @@ config HAVE_SYSCALL_TRACEPOINTS help See Documentation/trace/ftrace-design.txt +config HAVE_C_MCOUNT_RECORD + bool + help + C version of recordmcount available? + config TRACER_MAX_TRACE bool -- cgit v1.2.3 From cf4db2597ae93b60efc0a7a4ec08690b75d629b1 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Thu, 14 Oct 2010 23:32:44 -0400 Subject: ftrace: Rename config option HAVE_C_MCOUNT_RECORD to HAVE_C_RECORDMCOUNT The config option used by archs to let the build system know that the C version of the recordmcount works for said arch is currently called HAVE_C_MCOUNT_RECORD which enables BUILD_C_RECORDMCOUNT. To be more consistent with the name that all archs may use, it has been renamed to HAVE_C_RECORDMCOUNT. This will be less confusing since we are building a C recordmcount and not a mcount_record. Suggested-by: Ingo Molnar Cc: Cc: Michal Marek Cc: linux-kbuild@vger.kernel.org Cc: John Reiser Signed-off-by: Steven Rostedt --- kernel/trace/Kconfig | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/trace/Kconfig b/kernel/trace/Kconfig index df00fbbaf609..e550d2eda1df 100644 --- a/kernel/trace/Kconfig +++ b/kernel/trace/Kconfig @@ -49,7 +49,7 @@ config HAVE_SYSCALL_TRACEPOINTS help See Documentation/trace/ftrace-design.txt -config HAVE_C_MCOUNT_RECORD +config HAVE_C_RECORDMCOUNT bool help C version of recordmcount available? -- cgit v1.2.3 From 6038f373a3dc1f1c26496e60b6c40b164716f07e Mon Sep 17 00:00:00 2001 From: Arnd Bergmann Date: Sun, 15 Aug 2010 18:52:59 +0200 Subject: llseek: automatically add .llseek fop All file_operations should get a .llseek operation so we can make nonseekable_open the default for future file operations without a .llseek pointer. The three cases that we can automatically detect are no_llseek, seq_lseek and default_llseek. For cases where we can we can automatically prove that the file offset is always ignored, we use noop_llseek, which maintains the current behavior of not returning an error from a seek. New drivers should normally not use noop_llseek but instead use no_llseek and call nonseekable_open at open time. Existing drivers can be converted to do the same when the maintainer knows for certain that no user code relies on calling seek on the device file. The generated code is often incorrectly indented and right now contains comments that clarify for each added line why a specific variant was chosen. In the version that gets submitted upstream, the comments will be gone and I will manually fix the indentation, because there does not seem to be a way to do that using coccinelle. Some amount of new code is currently sitting in linux-next that should get the same modifications, which I will do at the end of the merge window. Many thanks to Julia Lawall for helping me learn to write a semantic patch that does all this. ===== begin semantic patch ===== // This adds an llseek= method to all file operations, // as a preparation for making no_llseek the default. // // The rules are // - use no_llseek explicitly if we do nonseekable_open // - use seq_lseek for sequential files // - use default_llseek if we know we access f_pos // - use noop_llseek if we know we don't access f_pos, // but we still want to allow users to call lseek // @ open1 exists @ identifier nested_open; @@ nested_open(...) { <+... nonseekable_open(...) ...+> } @ open exists@ identifier open_f; identifier i, f; identifier open1.nested_open; @@ int open_f(struct inode *i, struct file *f) { <+... ( nonseekable_open(...) | nested_open(...) ) ...+> } @ read disable optional_qualifier exists @ identifier read_f; identifier f, p, s, off; type ssize_t, size_t, loff_t; expression E; identifier func; @@ ssize_t read_f(struct file *f, char *p, size_t s, loff_t *off) { <+... ( *off = E | *off += E | func(..., off, ...) | E = *off ) ...+> } @ read_no_fpos disable optional_qualifier exists @ identifier read_f; identifier f, p, s, off; type ssize_t, size_t, loff_t; @@ ssize_t read_f(struct file *f, char *p, size_t s, loff_t *off) { ... when != off } @ write @ identifier write_f; identifier f, p, s, off; type ssize_t, size_t, loff_t; expression E; identifier func; @@ ssize_t write_f(struct file *f, const char *p, size_t s, loff_t *off) { <+... ( *off = E | *off += E | func(..., off, ...) | E = *off ) ...+> } @ write_no_fpos @ identifier write_f; identifier f, p, s, off; type ssize_t, size_t, loff_t; @@ ssize_t write_f(struct file *f, const char *p, size_t s, loff_t *off) { ... when != off } @ fops0 @ identifier fops; @@ struct file_operations fops = { ... }; @ has_llseek depends on fops0 @ identifier fops0.fops; identifier llseek_f; @@ struct file_operations fops = { ... .llseek = llseek_f, ... }; @ has_read depends on fops0 @ identifier fops0.fops; identifier read_f; @@ struct file_operations fops = { ... .read = read_f, ... }; @ has_write depends on fops0 @ identifier fops0.fops; identifier write_f; @@ struct file_operations fops = { ... .write = write_f, ... }; @ has_open depends on fops0 @ identifier fops0.fops; identifier open_f; @@ struct file_operations fops = { ... .open = open_f, ... }; // use no_llseek if we call nonseekable_open //////////////////////////////////////////// @ nonseekable1 depends on !has_llseek && has_open @ identifier fops0.fops; identifier nso ~= "nonseekable_open"; @@ struct file_operations fops = { ... .open = nso, ... +.llseek = no_llseek, /* nonseekable */ }; @ nonseekable2 depends on !has_llseek @ identifier fops0.fops; identifier open.open_f; @@ struct file_operations fops = { ... .open = open_f, ... +.llseek = no_llseek, /* open uses nonseekable */ }; // use seq_lseek for sequential files ///////////////////////////////////// @ seq depends on !has_llseek @ identifier fops0.fops; identifier sr ~= "seq_read"; @@ struct file_operations fops = { ... .read = sr, ... +.llseek = seq_lseek, /* we have seq_read */ }; // use default_llseek if there is a readdir /////////////////////////////////////////// @ fops1 depends on !has_llseek && !nonseekable1 && !nonseekable2 && !seq @ identifier fops0.fops; identifier readdir_e; @@ // any other fop is used that changes pos struct file_operations fops = { ... .readdir = readdir_e, ... +.llseek = default_llseek, /* readdir is present */ }; // use default_llseek if at least one of read/write touches f_pos ///////////////////////////////////////////////////////////////// @ fops2 depends on !fops1 && !has_llseek && !nonseekable1 && !nonseekable2 && !seq @ identifier fops0.fops; identifier read.read_f; @@ // read fops use offset struct file_operations fops = { ... .read = read_f, ... +.llseek = default_llseek, /* read accesses f_pos */ }; @ fops3 depends on !fops1 && !fops2 && !has_llseek && !nonseekable1 && !nonseekable2 && !seq @ identifier fops0.fops; identifier write.write_f; @@ // write fops use offset struct file_operations fops = { ... .write = write_f, ... + .llseek = default_llseek, /* write accesses f_pos */ }; // Use noop_llseek if neither read nor write accesses f_pos /////////////////////////////////////////////////////////// @ fops4 depends on !fops1 && !fops2 && !fops3 && !has_llseek && !nonseekable1 && !nonseekable2 && !seq @ identifier fops0.fops; identifier read_no_fpos.read_f; identifier write_no_fpos.write_f; @@ // write fops use offset struct file_operations fops = { ... .write = write_f, .read = read_f, ... +.llseek = noop_llseek, /* read and write both use no f_pos */ }; @ depends on has_write && !has_read && !fops1 && !fops2 && !has_llseek && !nonseekable1 && !nonseekable2 && !seq @ identifier fops0.fops; identifier write_no_fpos.write_f; @@ struct file_operations fops = { ... .write = write_f, ... +.llseek = noop_llseek, /* write uses no f_pos */ }; @ depends on has_read && !has_write && !fops1 && !fops2 && !has_llseek && !nonseekable1 && !nonseekable2 && !seq @ identifier fops0.fops; identifier read_no_fpos.read_f; @@ struct file_operations fops = { ... .read = read_f, ... +.llseek = noop_llseek, /* read uses no f_pos */ }; @ depends on !has_read && !has_write && !fops1 && !fops2 && !has_llseek && !nonseekable1 && !nonseekable2 && !seq @ identifier fops0.fops; @@ struct file_operations fops = { ... +.llseek = noop_llseek, /* no read or write fn */ }; ===== End semantic patch ===== Signed-off-by: Arnd Bergmann Cc: Julia Lawall Cc: Christoph Hellwig --- kernel/configs.c | 1 + kernel/gcov/fs.c | 1 + kernel/kprobes.c | 1 + kernel/pm_qos_params.c | 1 + kernel/profile.c | 1 + kernel/trace/blktrace.c | 2 ++ kernel/trace/ftrace.c | 2 ++ kernel/trace/ring_buffer.c | 1 + kernel/trace/trace_events.c | 6 ++++++ kernel/trace/trace_stack.c | 1 + 10 files changed, 17 insertions(+) (limited to 'kernel') diff --git a/kernel/configs.c b/kernel/configs.c index abaee684ecbf..b4066b44a99d 100644 --- a/kernel/configs.c +++ b/kernel/configs.c @@ -66,6 +66,7 @@ ikconfig_read_current(struct file *file, char __user *buf, static const struct file_operations ikconfig_file_ops = { .owner = THIS_MODULE, .read = ikconfig_read_current, + .llseek = default_llseek, }; static int __init ikconfig_init(void) diff --git a/kernel/gcov/fs.c b/kernel/gcov/fs.c index f83972b16564..9bd0934f6c33 100644 --- a/kernel/gcov/fs.c +++ b/kernel/gcov/fs.c @@ -561,6 +561,7 @@ static ssize_t reset_read(struct file *file, char __user *addr, size_t len, static const struct file_operations gcov_reset_fops = { .write = reset_write, .read = reset_read, + .llseek = noop_llseek, }; /* diff --git a/kernel/kprobes.c b/kernel/kprobes.c index 282035f3ae96..8b5ff2655ae0 100644 --- a/kernel/kprobes.c +++ b/kernel/kprobes.c @@ -1992,6 +1992,7 @@ static ssize_t write_enabled_file_bool(struct file *file, static const struct file_operations fops_kp = { .read = read_enabled_file_bool, .write = write_enabled_file_bool, + .llseek = default_llseek, }; static int __kprobes debugfs_kprobe_init(void) diff --git a/kernel/pm_qos_params.c b/kernel/pm_qos_params.c index 645e541a45f6..a96b850ba08a 100644 --- a/kernel/pm_qos_params.c +++ b/kernel/pm_qos_params.c @@ -110,6 +110,7 @@ static const struct file_operations pm_qos_power_fops = { .write = pm_qos_power_write, .open = pm_qos_power_open, .release = pm_qos_power_release, + .llseek = noop_llseek, }; /* unlocked internal variant */ diff --git a/kernel/profile.c b/kernel/profile.c index b22a899934cc..66f841b7fbd3 100644 --- a/kernel/profile.c +++ b/kernel/profile.c @@ -555,6 +555,7 @@ static ssize_t write_profile(struct file *file, const char __user *buf, static const struct file_operations proc_profile_operations = { .read = read_profile, .write = write_profile, + .llseek = default_llseek, }; #ifdef CONFIG_SMP diff --git a/kernel/trace/blktrace.c b/kernel/trace/blktrace.c index 959f8d6c8cc1..2d5f3a757316 100644 --- a/kernel/trace/blktrace.c +++ b/kernel/trace/blktrace.c @@ -326,6 +326,7 @@ static const struct file_operations blk_dropped_fops = { .owner = THIS_MODULE, .open = blk_dropped_open, .read = blk_dropped_read, + .llseek = default_llseek, }; static int blk_msg_open(struct inode *inode, struct file *filp) @@ -365,6 +366,7 @@ static const struct file_operations blk_msg_fops = { .owner = THIS_MODULE, .open = blk_msg_open, .write = blk_msg_write, + .llseek = noop_llseek, }; /* diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index fa7ece649fe1..5e1ad4763090 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -800,6 +800,7 @@ static const struct file_operations ftrace_profile_fops = { .open = tracing_open_generic, .read = ftrace_profile_read, .write = ftrace_profile_write, + .llseek = default_llseek, }; /* used to initialize the real stat files */ @@ -2632,6 +2633,7 @@ static const struct file_operations ftrace_graph_fops = { .read = seq_read, .write = ftrace_graph_write, .release = ftrace_graph_release, + .llseek = seq_lseek, }; #endif /* CONFIG_FUNCTION_GRAPH_TRACER */ diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c index 492197e2f86c..3aea966d16de 100644 --- a/kernel/trace/ring_buffer.c +++ b/kernel/trace/ring_buffer.c @@ -3965,6 +3965,7 @@ static const struct file_operations rb_simple_fops = { .open = tracing_open_generic, .read = rb_simple_read, .write = rb_simple_write, + .llseek = default_llseek, }; diff --git a/kernel/trace/trace_events.c b/kernel/trace/trace_events.c index 4c758f146328..0369c5e09984 100644 --- a/kernel/trace/trace_events.c +++ b/kernel/trace/trace_events.c @@ -951,6 +951,7 @@ static const struct file_operations ftrace_enable_fops = { .open = tracing_open_generic, .read = event_enable_read, .write = event_enable_write, + .llseek = default_llseek, }; static const struct file_operations ftrace_event_format_fops = { @@ -963,29 +964,34 @@ static const struct file_operations ftrace_event_format_fops = { static const struct file_operations ftrace_event_id_fops = { .open = tracing_open_generic, .read = event_id_read, + .llseek = default_llseek, }; static const struct file_operations ftrace_event_filter_fops = { .open = tracing_open_generic, .read = event_filter_read, .write = event_filter_write, + .llseek = default_llseek, }; static const struct file_operations ftrace_subsystem_filter_fops = { .open = tracing_open_generic, .read = subsystem_filter_read, .write = subsystem_filter_write, + .llseek = default_llseek, }; static const struct file_operations ftrace_system_enable_fops = { .open = tracing_open_generic, .read = system_enable_read, .write = system_enable_write, + .llseek = default_llseek, }; static const struct file_operations ftrace_show_header_fops = { .open = tracing_open_generic, .read = show_header, + .llseek = default_llseek, }; static struct dentry *event_trace_events_dir(void) diff --git a/kernel/trace/trace_stack.c b/kernel/trace/trace_stack.c index a6b7e0e0f3eb..4c5dead0c239 100644 --- a/kernel/trace/trace_stack.c +++ b/kernel/trace/trace_stack.c @@ -195,6 +195,7 @@ static const struct file_operations stack_max_size_fops = { .open = tracing_open_generic, .read = stack_max_size_read, .write = stack_max_size_write, + .llseek = default_llseek, }; static void * -- cgit v1.2.3 From a9febbb4bd1302b6f01aa1203b0a804e4e5c9e25 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Fri, 15 Oct 2010 14:34:12 -0700 Subject: sysctl: min/max bounds are optional sysctl check complains with a WARN() when proc_doulongvec_minmax() or proc_doulongvec_ms_jiffies_minmax() are used by a vector of longs (with more than one element), with no min or max value specified. This is unexpected, given we had a bug on this min/max handling :) Reported-by: Jiri Slaby Signed-off-by: Eric Dumazet Cc: "Eric W. Biederman" Cc: David Miller Acked-by: WANG Cong Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/sysctl_check.c | 9 --------- 1 file changed, 9 deletions(-) (limited to 'kernel') diff --git a/kernel/sysctl_check.c b/kernel/sysctl_check.c index 04cdcf72c827..10b90d8a03c4 100644 --- a/kernel/sysctl_check.c +++ b/kernel/sysctl_check.c @@ -143,15 +143,6 @@ int sysctl_check_table(struct nsproxy *namespaces, struct ctl_table *table) if (!table->maxlen) set_fail(&fail, table, "No maxlen"); } - if ((table->proc_handler == proc_doulongvec_minmax) || - (table->proc_handler == proc_doulongvec_ms_jiffies_minmax)) { - if (table->maxlen > sizeof (unsigned long)) { - if (!table->extra1) - set_fail(&fail, table, "No min"); - if (!table->extra2) - set_fail(&fail, table, "No max"); - } - } #ifdef CONFIG_PROC_SYSCTL if (table->procname && !table->proc_handler) set_fail(&fail, table, "No proc_handler"); -- cgit v1.2.3 From f996fc9671d088bd5f52a70f18c64bfe3d0e418f Mon Sep 17 00:00:00 2001 From: Bojan Smojver Date: Thu, 9 Sep 2010 23:06:23 +0200 Subject: PM / Hibernate: Compress hibernation image with LZO Compress hibernation image with LZO in order to save on I/O and therefore time to hibernate/thaw. [rjw: Added hibernate=nocompress command line option instead of just nocompress which would be confusing, fixed a couple of compiler warnings, fixed kerneldoc comments, minor cleanups.] Signed-off-by: Bojan Smojver Signed-off-by: Rafael J. Wysocki --- kernel/power/Kconfig | 2 + kernel/power/hibernate.c | 13 +++ kernel/power/power.h | 1 + kernel/power/swap.c | 290 +++++++++++++++++++++++++++++++++++++++++++++-- 4 files changed, 299 insertions(+), 7 deletions(-) (limited to 'kernel') diff --git a/kernel/power/Kconfig b/kernel/power/Kconfig index ca6066a6952e..cb57eb99215f 100644 --- a/kernel/power/Kconfig +++ b/kernel/power/Kconfig @@ -137,6 +137,8 @@ config SUSPEND_FREEZER config HIBERNATION bool "Hibernation (aka 'suspend to disk')" depends on PM && SWAP && ARCH_HIBERNATION_POSSIBLE + select LZO_COMPRESS + select LZO_DECOMPRESS select SUSPEND_NVS if HAS_IOMEM ---help--- Enable the suspend to disk (STD) functionality, which is usually diff --git a/kernel/power/hibernate.c b/kernel/power/hibernate.c index 8dc31e02ae12..6c9c9dc48c75 100644 --- a/kernel/power/hibernate.c +++ b/kernel/power/hibernate.c @@ -29,6 +29,7 @@ #include "power.h" +static int nocompress = 0; static int noresume = 0; static char resume_file[256] = CONFIG_PM_STD_PARTITION; dev_t swsusp_resume_device; @@ -638,6 +639,8 @@ int hibernate(void) if (hibernation_mode == HIBERNATION_PLATFORM) flags |= SF_PLATFORM_MODE; + if (nocompress) + flags |= SF_NOCOMPRESS_MODE; pr_debug("PM: writing image.\n"); error = swsusp_write(flags); swsusp_free(); @@ -1004,6 +1007,15 @@ static int __init resume_offset_setup(char *str) return 1; } +static int __init hibernate_setup(char *str) +{ + if (!strncmp(str, "noresume", 8)) + noresume = 1; + else if (!strncmp(str, "nocompress", 10)) + nocompress = 1; + return 1; +} + static int __init noresume_setup(char *str) { noresume = 1; @@ -1013,3 +1025,4 @@ static int __init noresume_setup(char *str) __setup("noresume", noresume_setup); __setup("resume_offset=", resume_offset_setup); __setup("resume=", resume_setup); +__setup("hibernate=", hibernate_setup); diff --git a/kernel/power/power.h b/kernel/power/power.h index 006270fe382d..c7e42e47eb0b 100644 --- a/kernel/power/power.h +++ b/kernel/power/power.h @@ -134,6 +134,7 @@ extern int swsusp_swap_in_use(void); * the image header. */ #define SF_PLATFORM_MODE 1 +#define SF_NOCOMPRESS_MODE 2 /* kernel/power/hibernate.c */ extern int swsusp_check(void); diff --git a/kernel/power/swap.c b/kernel/power/swap.c index e6a5bdf61a37..3dc0552cddfb 100644 --- a/kernel/power/swap.c +++ b/kernel/power/swap.c @@ -24,6 +24,8 @@ #include #include #include +#include +#include #include "power.h" @@ -357,6 +359,18 @@ static int swap_writer_finish(struct swap_map_handle *handle, return error; } +/* We need to remember how much compressed data we need to read. */ +#define LZO_HEADER sizeof(size_t) + +/* Number of pages/bytes we'll compress at one time. */ +#define LZO_UNC_PAGES 32 +#define LZO_UNC_SIZE (LZO_UNC_PAGES * PAGE_SIZE) + +/* Number of pages/bytes we need for compressed data (worst case). */ +#define LZO_CMP_PAGES DIV_ROUND_UP(lzo1x_worst_compress(LZO_UNC_SIZE) + \ + LZO_HEADER, PAGE_SIZE) +#define LZO_CMP_SIZE (LZO_CMP_PAGES * PAGE_SIZE) + /** * save_image - save the suspend image data */ @@ -404,6 +418,137 @@ static int save_image(struct swap_map_handle *handle, return ret; } + +/** + * save_image_lzo - Save the suspend image data compressed with LZO. + * @handle: Swap mam handle to use for saving the image. + * @snapshot: Image to read data from. + * @nr_to_write: Number of pages to save. + */ +static int save_image_lzo(struct swap_map_handle *handle, + struct snapshot_handle *snapshot, + unsigned int nr_to_write) +{ + unsigned int m; + int ret = 0; + int nr_pages; + int err2; + struct bio *bio; + struct timeval start; + struct timeval stop; + size_t off, unc_len, cmp_len; + unsigned char *unc, *cmp, *wrk, *page; + + page = (void *)__get_free_page(__GFP_WAIT | __GFP_HIGH); + if (!page) { + printk(KERN_ERR "PM: Failed to allocate LZO page\n"); + return -ENOMEM; + } + + wrk = vmalloc(LZO1X_1_MEM_COMPRESS); + if (!wrk) { + printk(KERN_ERR "PM: Failed to allocate LZO workspace\n"); + free_page((unsigned long)page); + return -ENOMEM; + } + + unc = vmalloc(LZO_UNC_SIZE); + if (!unc) { + printk(KERN_ERR "PM: Failed to allocate LZO uncompressed\n"); + vfree(wrk); + free_page((unsigned long)page); + return -ENOMEM; + } + + cmp = vmalloc(LZO_CMP_SIZE); + if (!cmp) { + printk(KERN_ERR "PM: Failed to allocate LZO compressed\n"); + vfree(unc); + vfree(wrk); + free_page((unsigned long)page); + return -ENOMEM; + } + + printk(KERN_INFO + "PM: Compressing and saving image data (%u pages) ... ", + nr_to_write); + m = nr_to_write / 100; + if (!m) + m = 1; + nr_pages = 0; + bio = NULL; + do_gettimeofday(&start); + for (;;) { + for (off = 0; off < LZO_UNC_SIZE; off += PAGE_SIZE) { + ret = snapshot_read_next(snapshot); + if (ret < 0) + goto out_finish; + + if (!ret) + break; + + memcpy(unc + off, data_of(*snapshot), PAGE_SIZE); + + if (!(nr_pages % m)) + printk(KERN_CONT "\b\b\b\b%3d%%", nr_pages / m); + nr_pages++; + } + + if (!off) + break; + + unc_len = off; + ret = lzo1x_1_compress(unc, unc_len, + cmp + LZO_HEADER, &cmp_len, wrk); + if (ret < 0) { + printk(KERN_ERR "PM: LZO compression failed\n"); + break; + } + + if (unlikely(!cmp_len || + cmp_len > lzo1x_worst_compress(unc_len))) { + printk(KERN_ERR "PM: Invalid LZO compressed length\n"); + ret = -1; + break; + } + + *(size_t *)cmp = cmp_len; + + /* + * Given we are writing one page at a time to disk, we copy + * that much from the buffer, although the last bit will likely + * be smaller than full page. This is OK - we saved the length + * of the compressed data, so any garbage at the end will be + * discarded when we read it. + */ + for (off = 0; off < LZO_HEADER + cmp_len; off += PAGE_SIZE) { + memcpy(page, cmp + off, PAGE_SIZE); + + ret = swap_write_page(handle, page, &bio); + if (ret) + goto out_finish; + } + } + +out_finish: + err2 = hib_wait_on_bio_chain(&bio); + do_gettimeofday(&stop); + if (!ret) + ret = err2; + if (!ret) + printk(KERN_CONT "\b\b\b\bdone\n"); + else + printk(KERN_CONT "\n"); + swsusp_show_speed(&start, &stop, nr_to_write, "Wrote"); + + vfree(cmp); + vfree(unc); + vfree(wrk); + free_page((unsigned long)page); + + return ret; +} + /** * enough_swap - Make sure we have enough swap to save the image. * @@ -411,12 +556,16 @@ static int save_image(struct swap_map_handle *handle, * space avaiable from the resume partition. */ -static int enough_swap(unsigned int nr_pages) +static int enough_swap(unsigned int nr_pages, unsigned int flags) { unsigned int free_swap = count_swap_pages(root_swap, 1); + unsigned int required; pr_debug("PM: Free swap pages: %u\n", free_swap); - return free_swap > nr_pages + PAGES_FOR_IO; + + required = PAGES_FOR_IO + ((flags & SF_NOCOMPRESS_MODE) ? + nr_pages : (nr_pages * LZO_CMP_PAGES) / LZO_UNC_PAGES + 1); + return free_swap > required; } /** @@ -443,7 +592,7 @@ int swsusp_write(unsigned int flags) printk(KERN_ERR "PM: Cannot get swap writer\n"); return error; } - if (!enough_swap(pages)) { + if (!enough_swap(pages, flags)) { printk(KERN_ERR "PM: Not enough free swap\n"); error = -ENOSPC; goto out_finish; @@ -458,8 +607,11 @@ int swsusp_write(unsigned int flags) } header = (struct swsusp_info *)data_of(snapshot); error = swap_write_page(&handle, header, NULL); - if (!error) - error = save_image(&handle, &snapshot, pages - 1); + if (!error) { + error = (flags & SF_NOCOMPRESS_MODE) ? + save_image(&handle, &snapshot, pages - 1) : + save_image_lzo(&handle, &snapshot, pages - 1); + } out_finish: error = swap_writer_finish(&handle, flags, error); return error; @@ -589,6 +741,127 @@ static int load_image(struct swap_map_handle *handle, return error; } +/** + * load_image_lzo - Load compressed image data and decompress them with LZO. + * @handle: Swap map handle to use for loading data. + * @snapshot: Image to copy uncompressed data into. + * @nr_to_read: Number of pages to load. + */ +static int load_image_lzo(struct swap_map_handle *handle, + struct snapshot_handle *snapshot, + unsigned int nr_to_read) +{ + unsigned int m; + int error = 0; + struct timeval start; + struct timeval stop; + unsigned nr_pages; + size_t off, unc_len, cmp_len; + unsigned char *unc, *cmp, *page; + + page = (void *)__get_free_page(__GFP_WAIT | __GFP_HIGH); + if (!page) { + printk(KERN_ERR "PM: Failed to allocate LZO page\n"); + return -ENOMEM; + } + + unc = vmalloc(LZO_UNC_SIZE); + if (!unc) { + printk(KERN_ERR "PM: Failed to allocate LZO uncompressed\n"); + free_page((unsigned long)page); + return -ENOMEM; + } + + cmp = vmalloc(LZO_CMP_SIZE); + if (!cmp) { + printk(KERN_ERR "PM: Failed to allocate LZO compressed\n"); + vfree(unc); + free_page((unsigned long)page); + return -ENOMEM; + } + + printk(KERN_INFO + "PM: Loading and decompressing image data (%u pages) ... ", + nr_to_read); + m = nr_to_read / 100; + if (!m) + m = 1; + nr_pages = 0; + do_gettimeofday(&start); + + error = snapshot_write_next(snapshot); + if (error <= 0) + goto out_finish; + + for (;;) { + error = swap_read_page(handle, page, NULL); /* sync */ + if (error) + break; + + cmp_len = *(size_t *)page; + if (unlikely(!cmp_len || + cmp_len > lzo1x_worst_compress(LZO_UNC_SIZE))) { + printk(KERN_ERR "PM: Invalid LZO compressed length\n"); + error = -1; + break; + } + + memcpy(cmp, page, PAGE_SIZE); + for (off = PAGE_SIZE; off < LZO_HEADER + cmp_len; off += PAGE_SIZE) { + error = swap_read_page(handle, page, NULL); /* sync */ + if (error) + goto out_finish; + + memcpy(cmp + off, page, PAGE_SIZE); + } + + unc_len = LZO_UNC_SIZE; + error = lzo1x_decompress_safe(cmp + LZO_HEADER, cmp_len, + unc, &unc_len); + if (error < 0) { + printk(KERN_ERR "PM: LZO decompression failed\n"); + break; + } + + if (unlikely(!unc_len || + unc_len > LZO_UNC_SIZE || + unc_len & (PAGE_SIZE - 1))) { + printk(KERN_ERR "PM: Invalid LZO uncompressed length\n"); + error = -1; + break; + } + + for (off = 0; off < unc_len; off += PAGE_SIZE) { + memcpy(data_of(*snapshot), unc + off, PAGE_SIZE); + + if (!(nr_pages % m)) + printk("\b\b\b\b%3d%%", nr_pages / m); + nr_pages++; + + error = snapshot_write_next(snapshot); + if (error <= 0) + goto out_finish; + } + } + +out_finish: + do_gettimeofday(&stop); + if (!error) { + printk("\b\b\b\bdone\n"); + snapshot_write_finalize(snapshot); + if (!snapshot_image_loaded(snapshot)) + error = -ENODATA; + } else + printk("\n"); + swsusp_show_speed(&start, &stop, nr_to_read, "Read"); + + vfree(cmp); + vfree(unc); + free_page((unsigned long)page); + + return error; +} + /** * swsusp_read - read the hibernation image. * @flags_p: flags passed by the "frozen" kernel in the image header should @@ -612,8 +885,11 @@ int swsusp_read(unsigned int *flags_p) goto end; if (!error) error = swap_read_page(&handle, header, NULL); - if (!error) - error = load_image(&handle, &snapshot, header->pages - 1); + if (!error) { + error = (*flags_p & SF_NOCOMPRESS_MODE) ? + load_image(&handle, &snapshot, header->pages - 1) : + load_image_lzo(&handle, &snapshot, header->pages - 1); + } swap_reader_finish(&handle); end: if (!error) -- cgit v1.2.3 From ede890c2c069d611ece0e184103a6b9236ce416a Mon Sep 17 00:00:00 2001 From: "Rafael J. Wysocki" Date: Sun, 12 Sep 2010 21:40:01 +0200 Subject: PM: Fix unmet dependency warning from kconfig Fix the following build warning: warning: (PM_SLEEP_SMP && SMP && (ARCH_SUSPEND_POSSIBLE || \ ARCH_HIBERNATION_POSSIBLE) && PM_SLEEP) selects HOTPLUG_CPU which \ has unmet direct dependencies (SMP && HOTPLUG) by selecting HOTPLUG along with CPU_HOTPLUG. Signed-off-by: Rafael J. Wysocki Acked-by: Randy Dunlap --- kernel/power/Kconfig | 1 + 1 file changed, 1 insertion(+) (limited to 'kernel') diff --git a/kernel/power/Kconfig b/kernel/power/Kconfig index cb57eb99215f..e45894c696ee 100644 --- a/kernel/power/Kconfig +++ b/kernel/power/Kconfig @@ -86,6 +86,7 @@ config PM_SLEEP_SMP depends on SMP depends on ARCH_SUSPEND_POSSIBLE || ARCH_HIBERNATION_POSSIBLE depends on PM_SLEEP + select HOTPLUG select HOTPLUG_CPU default y -- cgit v1.2.3 From bcb5ba8b4e8a5ae14b27351bdf499dd4c3bcc944 Mon Sep 17 00:00:00 2001 From: "Rafael J. Wysocki" Date: Mon, 20 Sep 2010 19:44:17 +0200 Subject: PM / Runtime: Use alloc_workqueue() for creating the PM workqueue Although we need the PM workqueue to be freezable, we don't need it to be singlethread. Also, the number of concurrent work items running on a single CPU need not be constrained. For these reasons use alloc_workqueue() directly, with suitable arguments, instead of create_freezeable_workqueue(), to create the runtime PM workqueue. Signed-off-by: Rafael J. Wysocki Acked-by: Tejun Heo --- kernel/power/main.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/power/main.c b/kernel/power/main.c index 62b0bc6e4983..0a28d4db3597 100644 --- a/kernel/power/main.c +++ b/kernel/power/main.c @@ -308,7 +308,7 @@ EXPORT_SYMBOL_GPL(pm_wq); static int __init pm_start_workqueue(void) { - pm_wq = create_freezeable_workqueue("pm"); + pm_wq = alloc_workqueue("pm", WQ_FREEZEABLE, 0); return pm_wq ? 0 : -ENOMEM; } -- cgit v1.2.3 From 266f1a25eff5ff98c498d7754a419aacfd88f71c Mon Sep 17 00:00:00 2001 From: "Rafael J. Wysocki" Date: Mon, 20 Sep 2010 19:44:38 +0200 Subject: PM / Hibernate: Improve comments in hibernate_preallocate_memory() One comment in hibernate_preallocate_memory() is wrong, so fix it and add one more comment to clarify the meaning of the fixed one. Signed-off-by: Rafael J. Wysocki --- kernel/power/snapshot.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/power/snapshot.c b/kernel/power/snapshot.c index d3f795f01bbc..d9191b40cf6e 100644 --- a/kernel/power/snapshot.c +++ b/kernel/power/snapshot.c @@ -1318,12 +1318,14 @@ int hibernate_preallocate_memory(void) /* Compute the maximum number of saveable pages to leave in memory. */ max_size = (count - (size + PAGES_FOR_IO)) / 2 - 2 * SPARE_PAGES; + /* Compute the desired number of image pages specified by image_size. */ size = DIV_ROUND_UP(image_size, PAGE_SIZE); if (size > max_size) size = max_size; /* - * If the maximum is not less than the current number of saveable pages - * in memory, allocate page frames for the image and we're done. + * If the desired number of image pages is at least as large as the + * current number of saveable pages in memory, allocate page frames for + * the image and we're done. */ if (size >= saveable) { pages = preallocate_image_highmem(save_highmem); -- cgit v1.2.3 From ac5c24ec1e983313ef0015258fba6f630e54e7cf Mon Sep 17 00:00:00 2001 From: "Rafael J. Wysocki" Date: Mon, 20 Sep 2010 19:44:56 +0200 Subject: PM / Hibernate: Make default image size depend on total RAM size The default hibernation image size is currently hard coded and euqal to 500 MB, which is not a reasonable default on many contemporary systems. Make it equal 2/5 of the total RAM size (this is slightly below the maximum, i.e. 1/2 of the total RAM size, and seems to be generally suitable). Signed-off-by: Rafael J. Wysocki Tested-by: M. Vefa Bicakci --- kernel/power/main.c | 1 + kernel/power/power.h | 9 ++++++++- kernel/power/snapshot.c | 7 ++++++- 3 files changed, 15 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/power/main.c b/kernel/power/main.c index 0a28d4db3597..f06ad6eff37a 100644 --- a/kernel/power/main.c +++ b/kernel/power/main.c @@ -321,6 +321,7 @@ static int __init pm_init(void) int error = pm_start_workqueue(); if (error) return error; + hibernate_image_size_init(); power_kobj = kobject_create_and_add("power", NULL); if (!power_kobj) return -ENOMEM; diff --git a/kernel/power/power.h b/kernel/power/power.h index c7e42e47eb0b..03634be55f62 100644 --- a/kernel/power/power.h +++ b/kernel/power/power.h @@ -14,6 +14,9 @@ struct swsusp_info { } __attribute__((aligned(PAGE_SIZE))); #ifdef CONFIG_HIBERNATION +/* kernel/power/snapshot.c */ +extern void __init hibernate_image_size_init(void); + #ifdef CONFIG_ARCH_HIBERNATION_HEADER /* Maximum size of architecture specific data in a hibernation header */ #define MAX_ARCH_HEADER_SIZE (sizeof(struct new_utsname) + 4) @@ -49,7 +52,11 @@ static inline char *check_image_kernel(struct swsusp_info *info) extern int hibernation_snapshot(int platform_mode); extern int hibernation_restore(int platform_mode); extern int hibernation_platform_enter(void); -#endif + +#else /* !CONFIG_HIBERNATION */ + +static inline void hibernate_image_size_init(void) {} +#endif /* !CONFIG_HIBERNATION */ extern int pfn_is_nosave(unsigned long); diff --git a/kernel/power/snapshot.c b/kernel/power/snapshot.c index d9191b40cf6e..ac7eb109f196 100644 --- a/kernel/power/snapshot.c +++ b/kernel/power/snapshot.c @@ -46,7 +46,12 @@ static void swsusp_unset_page_forbidden(struct page *); * size will not exceed N bytes, but if that is impossible, it will * try to create the smallest image possible. */ -unsigned long image_size = 500 * 1024 * 1024; +unsigned long image_size; + +void __init hibernate_image_size_init(void) +{ + image_size = ((totalram_pages * 2) / 5) * PAGE_SIZE; +} /* List of PBEs needed for restoring the pages that were allocated before * the suspend and included in the suspend image, but have also been -- cgit v1.2.3 From 074037ec79bea73edf1b1ec72fef1010e83e3cc5 Mon Sep 17 00:00:00 2001 From: "Rafael J. Wysocki" Date: Wed, 22 Sep 2010 22:09:10 +0200 Subject: PM / Wakeup: Introduce wakeup source objects and event statistics (v3) Introduce struct wakeup_source for representing system wakeup sources within the kernel and for collecting statistics related to them. Make the recently introduced helper functions pm_wakeup_event(), pm_stay_awake() and pm_relax() use struct wakeup_source objects internally, so that wakeup statistics associated with wakeup devices can be collected and reported in a consistent way (the definition of pm_relax() is changed, which is harmless, because this function is not called directly by anyone yet). Introduce new wakeup-related sysfs device attributes in /sys/devices/.../power for reporting the device wakeup statistics. Change the global wakeup events counters event_count and events_in_progress into atomic variables, so that it is not necessary to acquire a global spinlock in pm_wakeup_event(), pm_stay_awake() and pm_relax(), which should allow us to avoid lock contention in these functions on SMP systems with many wakeup devices. Signed-off-by: Rafael J. Wysocki Acked-by: Greg Kroah-Hartman --- kernel/power/main.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) (limited to 'kernel') diff --git a/kernel/power/main.c b/kernel/power/main.c index f06ad6eff37a..6b12a0cf4d9f 100644 --- a/kernel/power/main.c +++ b/kernel/power/main.c @@ -237,18 +237,18 @@ static ssize_t wakeup_count_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf) { - unsigned long val; + unsigned int val; - return pm_get_wakeup_count(&val) ? sprintf(buf, "%lu\n", val) : -EINTR; + return pm_get_wakeup_count(&val) ? sprintf(buf, "%u\n", val) : -EINTR; } static ssize_t wakeup_count_store(struct kobject *kobj, struct kobj_attribute *attr, const char *buf, size_t n) { - unsigned long val; + unsigned int val; - if (sscanf(buf, "%lu", &val) == 1) { + if (sscanf(buf, "%u", &val) == 1) { if (pm_save_wakeup_count(val)) return n; } -- cgit v1.2.3 From d0941ead3fdd31aafff992d211bcefdbff1eaedb Mon Sep 17 00:00:00 2001 From: "Rafael J. Wysocki" Date: Tue, 28 Sep 2010 23:31:22 +0200 Subject: PM / Hibernate: Make some boot messages look less scary The hibernate resume code checks if there is an image to resume from on every boot and, if the kernel is built with CONFIG_PM_DEBUG set and the image is not present, it prints some scary messages suggesting there was a boot error of some sort. Apparently, some users are confused by them, so make them look less scary and adjust the other hibernate resume debug messages to match them. Signed-off-by: Rafael J. Wysocki --- kernel/power/hibernate.c | 12 ++++++------ kernel/power/swap.c | 4 ++-- 2 files changed, 8 insertions(+), 8 deletions(-) (limited to 'kernel') diff --git a/kernel/power/hibernate.c b/kernel/power/hibernate.c index 6c9c9dc48c75..657272e91d0a 100644 --- a/kernel/power/hibernate.c +++ b/kernel/power/hibernate.c @@ -708,7 +708,7 @@ static int software_resume(void) goto Unlock; } - pr_debug("PM: Checking image partition %s\n", resume_file); + pr_debug("PM: Checking hibernation image partition %s\n", resume_file); /* Check if the device is there */ swsusp_resume_device = name_to_dev_t(resume_file); @@ -733,10 +733,10 @@ static int software_resume(void) } Check_image: - pr_debug("PM: Resume from partition %d:%d\n", + pr_debug("PM: Hibernation image partition %d:%d present\n", MAJOR(swsusp_resume_device), MINOR(swsusp_resume_device)); - pr_debug("PM: Checking hibernation image.\n"); + pr_debug("PM: Looking for hibernation image.\n"); error = swsusp_check(); if (error) goto Unlock; @@ -768,14 +768,14 @@ static int software_resume(void) goto Done; } - pr_debug("PM: Reading hibernation image.\n"); + pr_debug("PM: Loading hibernation image.\n"); error = swsusp_read(&flags); swsusp_close(FMODE_READ); if (!error) hibernation_restore(flags & SF_PLATFORM_MODE); - printk(KERN_ERR "PM: Restore failed, recovering.\n"); + printk(KERN_ERR "PM: Failed to load hibernation image, recovering.\n"); swsusp_free(); thaw_processes(); Done: @@ -788,7 +788,7 @@ static int software_resume(void) /* For success case, the suspend path will release the lock */ Unlock: mutex_unlock(&pm_mutex); - pr_debug("PM: Resume from disk failed.\n"); + pr_debug("PM: Hibernation image not present or could not be loaded.\n"); return error; close_finish: swsusp_close(FMODE_READ); diff --git a/kernel/power/swap.c b/kernel/power/swap.c index 3dc0552cddfb..2dfa87869b49 100644 --- a/kernel/power/swap.c +++ b/kernel/power/swap.c @@ -929,13 +929,13 @@ put: if (error) blkdev_put(hib_resume_bdev, FMODE_READ); else - pr_debug("PM: Signature found, resuming\n"); + pr_debug("PM: Image signature found, resuming\n"); } else { error = PTR_ERR(hib_resume_bdev); } if (error) - pr_debug("PM: Error %d checking image file\n", error); + pr_debug("PM: Image not found (code %d)\n", error); return error; } -- cgit v1.2.3 From dbeeec5fe868f2e2e92fe94daa2c5a047240fdc4 Mon Sep 17 00:00:00 2001 From: "Rafael J. Wysocki" Date: Mon, 4 Oct 2010 22:07:32 +0200 Subject: PM: Allow wakeup events to abort freezing of tasks MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit If there is a wakeup event during the freezing of tasks, suspend or hibernation will fail anyway. Since try_to_freeze_tasks() can take up to 20 seconds to complete or fail, aborting it as soon as a wakeup event is detected improves the worst case wakeup latency. Based on a patch from Arve HjønnevÃ¥g. Signed-off-by: Rafael J. Wysocki Acked-by: Pavel Machek --- kernel/power/process.c | 11 +++++++++-- 1 file changed, 9 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/power/process.c b/kernel/power/process.c index 028a99598f49..e50b4c1b2a0f 100644 --- a/kernel/power/process.c +++ b/kernel/power/process.c @@ -40,6 +40,7 @@ static int try_to_freeze_tasks(bool sig_only) struct timeval start, end; u64 elapsed_csecs64; unsigned int elapsed_csecs; + bool wakeup = false; do_gettimeofday(&start); @@ -78,6 +79,11 @@ static int try_to_freeze_tasks(bool sig_only) if (!todo || time_after(jiffies, end_time)) break; + if (!pm_check_wakeup_events()) { + wakeup = true; + break; + } + /* * We need to retry, but first give the freezing tasks some * time to enter the regrigerator. @@ -97,8 +103,9 @@ static int try_to_freeze_tasks(bool sig_only) * but it cleans up leftover PF_FREEZE requests. */ printk("\n"); - printk(KERN_ERR "Freezing of tasks failed after %d.%02d seconds " + printk(KERN_ERR "Freezing of tasks %s after %d.%02d seconds " "(%d tasks refusing to freeze, wq_busy=%d):\n", + wakeup ? "aborted" : "failed", elapsed_csecs / 100, elapsed_csecs % 100, todo - wq_busy, wq_busy); @@ -107,7 +114,7 @@ static int try_to_freeze_tasks(bool sig_only) read_lock(&tasklist_lock); do_each_thread(g, p) { task_lock(p); - if (freezing(p) && !freezer_should_skip(p)) + if (!wakeup && freezing(p) && !freezer_should_skip(p)) sched_show_task(p); cancel_freezing(p); task_unlock(p); -- cgit v1.2.3 From 3624eb04c24861ab296842414f9752a393e68372 Mon Sep 17 00:00:00 2001 From: "Rafael J. Wysocki" Date: Mon, 4 Oct 2010 22:08:12 +0200 Subject: PM / Hibernate: Modify signature used to mark swap Since we are adding compression to the kernel's hibernate code, change signature used by it to mark swap spaces, so that earlier kernels don't attempt to restore compressed images they cannot handle. Signed-off-by: Rafael J. Wysocki Acked-by: Pavel Machek --- kernel/power/swap.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/power/swap.c b/kernel/power/swap.c index 2dfa87869b49..916eaa790399 100644 --- a/kernel/power/swap.c +++ b/kernel/power/swap.c @@ -29,7 +29,7 @@ #include "power.h" -#define SWSUSP_SIG "S1SUSPEND" +#define HIBERNATE_SIG "LINHIB0001" /* * The swap map is a data structure used for keeping track of each page @@ -195,7 +195,7 @@ static int mark_swapfiles(struct swap_map_handle *handle, unsigned int flags) if (!memcmp("SWAP-SPACE",swsusp_header->sig, 10) || !memcmp("SWAPSPACE2",swsusp_header->sig, 10)) { memcpy(swsusp_header->orig_sig,swsusp_header->sig, 10); - memcpy(swsusp_header->sig,SWSUSP_SIG, 10); + memcpy(swsusp_header->sig, HIBERNATE_SIG, 10); swsusp_header->image = handle->first_sector; swsusp_header->flags = flags; error = hib_bio_write_page(swsusp_resume_block, @@ -916,7 +916,7 @@ int swsusp_check(void) if (error) goto put; - if (!memcmp(SWSUSP_SIG, swsusp_header->sig, 10)) { + if (!memcmp(HIBERNATE_SIG, swsusp_header->sig, 10)) { memcpy(swsusp_header->sig, swsusp_header->orig_sig, 10); /* Reset swap signature now */ error = hib_bio_write_page(swsusp_resume_block, -- cgit v1.2.3 From d33ac60beaf2c7dee5cd90aba7c1eb385dd70937 Mon Sep 17 00:00:00 2001 From: James Hogan Date: Tue, 12 Oct 2010 00:00:25 +0200 Subject: PM: Add sysfs attr for rechecking dev hash from PM trace If the device which fails to resume is part of a loadable kernel module it won't be checked at startup against the magic number stored in the RTC. Add a read-only sysfs attribute /sys/power/pm_trace_dev_match which contains a list of newline separated devices (usually just the one) which currently match the last magic number. This allows the device which is failing to resume to be found after the modules are loaded again. Signed-off-by: James Hogan Signed-off-by: Rafael J. Wysocki --- kernel/power/main.c | 18 ++++++++++++++++++ 1 file changed, 18 insertions(+) (limited to 'kernel') diff --git a/kernel/power/main.c b/kernel/power/main.c index 6b12a0cf4d9f..7b5db6a8561e 100644 --- a/kernel/power/main.c +++ b/kernel/power/main.c @@ -281,12 +281,30 @@ pm_trace_store(struct kobject *kobj, struct kobj_attribute *attr, } power_attr(pm_trace); + +static ssize_t pm_trace_dev_match_show(struct kobject *kobj, + struct kobj_attribute *attr, + char *buf) +{ + return show_trace_dev_match(buf, PAGE_SIZE); +} + +static ssize_t +pm_trace_dev_match_store(struct kobject *kobj, struct kobj_attribute *attr, + const char *buf, size_t n) +{ + return -EINVAL; +} + +power_attr(pm_trace_dev_match); + #endif /* CONFIG_PM_TRACE */ static struct attribute * g[] = { &state_attr.attr, #ifdef CONFIG_PM_TRACE &pm_trace_attr.attr, + &pm_trace_dev_match_attr.attr, #endif #ifdef CONFIG_PM_SLEEP &pm_async_attr.attr, -- cgit v1.2.3 From e1f60b292ffd61151403327aa19ff7a1871820bd Mon Sep 17 00:00:00 2001 From: Nishanth Menon Date: Wed, 13 Oct 2010 00:13:10 +0200 Subject: PM: Introduce library for device-specific OPPs (v7) SoCs have a standard set of tuples consisting of frequency and voltage pairs that the device will support per voltage domain. These are called Operating Performance Points or OPPs. The actual definitions of OPP varies over silicon versions. For a specific domain, we can have a set of {frequency, voltage} pairs. As the kernel boots and more information is available, a default set of these are activated based on the precise nature of device. Further on operation, based on conditions prevailing in the system (such as temperature), some OPP availability may be temporarily controlled by the SoC frameworks. To implement an OPP, some sort of power management support is necessary hence this library depends on CONFIG_PM. Contributions include: Sanjeev Premi for the initial concept: http://patchwork.kernel.org/patch/50998/ Kevin Hilman for converting original design to device-based. Kevin Hilman and Paul Walmsey for cleaning up many of the function abstractions, improvements and data structure handling. Romit Dasgupta for using enums instead of opp pointers. Thara Gopinath, Eduardo Valentin and Vishwanath BS for fixes and cleanups. Linus Walleij for recommending this layer be made generic for usage in other architectures beyond OMAP and ARM. Mark Brown, Andrew Morton, Rafael J. Wysocki, Paul E. McKenney for valuable improvements. Discussions and comments from: http://marc.info/?l=linux-omap&m=126033945313269&w=2 http://marc.info/?l=linux-omap&m=125482970102327&w=2 http://marc.info/?t=125809247500002&r=1&w=2 http://marc.info/?l=linux-omap&m=126025973426007&w=2 http://marc.info/?t=128152609200064&r=1&w=2 http://marc.info/?t=128468723000002&r=1&w=2 incorporated. v1: http://marc.info/?t=128468723000002&r=1&w=2 Signed-off-by: Nishanth Menon Signed-off-by: Kevin Hilman Signed-off-by: Rafael J. Wysocki --- kernel/power/Kconfig | 14 ++++++++++++++ 1 file changed, 14 insertions(+) (limited to 'kernel') diff --git a/kernel/power/Kconfig b/kernel/power/Kconfig index e45894c696ee..29bff6117abc 100644 --- a/kernel/power/Kconfig +++ b/kernel/power/Kconfig @@ -245,3 +245,17 @@ config PM_OPS bool depends on PM_SLEEP || PM_RUNTIME default y + +config PM_OPP + bool "Operating Performance Point (OPP) Layer library" + depends on PM + ---help--- + SOCs have a standard set of tuples consisting of frequency and + voltage pairs that the device will support per voltage domain. This + is called Operating Performance Point or OPP. The actual definitions + of OPP varies over silicon within the same family of devices. + + OPP layer organizes the data internally using device pointers + representing individual voltage domains and provides SOC + implementations a ready to use framework to manage OPPs. + For more information, read -- cgit v1.2.3 From 907f27840985fe6a0c62e43cd4702c6e04b4bcc7 Mon Sep 17 00:00:00 2001 From: matt mooney Date: Mon, 27 Sep 2010 19:04:53 -0700 Subject: tracing/trivial: Remove cast from void* Unnecessary cast from void* in assignment. Signed-off-by: matt mooney Signed-off-by: Steven Rostedt --- kernel/trace/ftrace.c | 4 ++-- kernel/trace/trace.c | 2 +- 2 files changed, 3 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index 65fb077ea79c..ebd80d50c474 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -1638,8 +1638,8 @@ ftrace_failures_open(struct inode *inode, struct file *file) ret = ftrace_avail_open(inode, file); if (!ret) { - m = (struct seq_file *)file->private_data; - iter = (struct ftrace_iterator *)m->private; + m = file->private_data; + iter = m->private; iter->flags = FTRACE_ITER_FAILURES; } diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 9ec59f541156..001bcd2ccf4a 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -2196,7 +2196,7 @@ int tracing_open_generic(struct inode *inode, struct file *filp) static int tracing_release(struct inode *inode, struct file *file) { - struct seq_file *m = (struct seq_file *)file->private_data; + struct seq_file *m = file->private_data; struct trace_iterator *iter; int cpu; -- cgit v1.2.3 From a9d61173dc1cb63e660ae89e874e51ba4fd2f991 Mon Sep 17 00:00:00 2001 From: Jiri Olsa Date: Fri, 24 Sep 2010 17:41:02 +0200 Subject: tracing: Add proper check for irq_depth routines The check_irq_entry and check_irq_return could be called from graph event context. In such case there's no graph private data allocated. Adding checks to handle this case. Signed-off-by: Jiri Olsa LKML-Reference: <20100924154102.GB1818@jolsa.brq.redhat.com> [ Fixed some grammar in the comments ] Signed-off-by: Steven Rostedt --- kernel/trace/trace_functions_graph.c | 24 ++++++++++++++++++++---- 1 file changed, 20 insertions(+), 4 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace_functions_graph.c b/kernel/trace/trace_functions_graph.c index ef49e9370b25..4c58ccc6427c 100644 --- a/kernel/trace/trace_functions_graph.c +++ b/kernel/trace/trace_functions_graph.c @@ -888,12 +888,20 @@ check_irq_entry(struct trace_iterator *iter, u32 flags, unsigned long addr, int depth) { int cpu = iter->cpu; + int *depth_irq; struct fgraph_data *data = iter->private; - int *depth_irq = &(per_cpu_ptr(data->cpu_data, cpu)->depth_irq); - if (flags & TRACE_GRAPH_PRINT_IRQS) + /* + * If we are either displaying irqs, or we got called as + * a graph event and private data does not exist, + * then we bypass the irq check. + */ + if ((flags & TRACE_GRAPH_PRINT_IRQS) || + (!data)) return 0; + depth_irq = &(per_cpu_ptr(data->cpu_data, cpu)->depth_irq); + /* * We are inside the irq code */ @@ -926,12 +934,20 @@ static int check_irq_return(struct trace_iterator *iter, u32 flags, int depth) { int cpu = iter->cpu; + int *depth_irq; struct fgraph_data *data = iter->private; - int *depth_irq = &(per_cpu_ptr(data->cpu_data, cpu)->depth_irq); - if (flags & TRACE_GRAPH_PRINT_IRQS) + /* + * If we are either displaying irqs, or we got called as + * a graph event and private data does not exist, + * then we bypass the irq check. + */ + if ((flags & TRACE_GRAPH_PRINT_IRQS) || + (!data)) return 0; + depth_irq = &(per_cpu_ptr(data->cpu_data, cpu)->depth_irq); + /* * We are not inside the irq code. */ -- cgit v1.2.3 From 0a772620a2e21fb55a02f70fe38d4b5c3a5fbbbf Mon Sep 17 00:00:00 2001 From: Jiri Olsa Date: Thu, 23 Sep 2010 14:00:52 +0200 Subject: tracing: Make graph related irqs/preemptsoff functions global Move trace_graph_function() and print_graph_headers_flags() functions to the trace_function_graph.c to be globaly available. Signed-off-by: Jiri Olsa LKML-Reference: <1285243253-7372-3-git-send-email-jolsa@redhat.com> Signed-off-by: Steven Rostedt --- kernel/trace/trace.h | 4 +++ kernel/trace/trace_functions_graph.c | 63 ++++++++++++++++++++++++++++++++++-- kernel/trace/trace_irqsoff.c | 56 ++++---------------------------- 3 files changed, 71 insertions(+), 52 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h index d39b3c5454a5..9021f8c0c0c3 100644 --- a/kernel/trace/trace.h +++ b/kernel/trace/trace.h @@ -343,6 +343,10 @@ void trace_function(struct trace_array *tr, unsigned long ip, unsigned long parent_ip, unsigned long flags, int pc); +void trace_graph_function(struct trace_array *tr, + unsigned long ip, + unsigned long parent_ip, + unsigned long flags, int pc); void trace_default_header(struct seq_file *m); void print_trace_header(struct seq_file *m, struct trace_iterator *iter); int trace_empty(struct trace_iterator *iter); diff --git a/kernel/trace/trace_functions_graph.c b/kernel/trace/trace_functions_graph.c index 4c58ccc6427c..6f8fe28acba1 100644 --- a/kernel/trace/trace_functions_graph.c +++ b/kernel/trace/trace_functions_graph.c @@ -262,6 +262,35 @@ int trace_graph_thresh_entry(struct ftrace_graph_ent *trace) return trace_graph_entry(trace); } +static void +__trace_graph_function(struct trace_array *tr, + unsigned long ip, unsigned long flags, int pc) +{ + u64 time = trace_clock_local(); + struct ftrace_graph_ent ent = { + .func = ip, + .depth = 0, + }; + struct ftrace_graph_ret ret = { + .func = ip, + .depth = 0, + .calltime = time, + .rettime = time, + }; + + __trace_graph_entry(tr, &ent, flags, pc); + __trace_graph_return(tr, &ret, flags, pc); +} + +void +trace_graph_function(struct trace_array *tr, + unsigned long ip, unsigned long parent_ip, + unsigned long flags, int pc) +{ + __trace_graph_function(tr, parent_ip, flags, pc); + __trace_graph_function(tr, ip, flags, pc); +} + void __trace_graph_return(struct trace_array *tr, struct ftrace_graph_ret *trace, unsigned long flags, @@ -1179,7 +1208,7 @@ print_graph_comment(struct trace_seq *s, struct trace_entry *ent, enum print_line_t -print_graph_function_flags(struct trace_iterator *iter, u32 flags) +__print_graph_function_flags(struct trace_iterator *iter, u32 flags) { struct ftrace_graph_ent_entry *field; struct fgraph_data *data = iter->private; @@ -1242,7 +1271,18 @@ print_graph_function_flags(struct trace_iterator *iter, u32 flags) static enum print_line_t print_graph_function(struct trace_iterator *iter) { - return print_graph_function_flags(iter, tracer_flags.val); + return __print_graph_function_flags(iter, tracer_flags.val); +} + +enum print_line_t print_graph_function_flags(struct trace_iterator *iter, + u32 flags) +{ + if (trace_flags & TRACE_ITER_LATENCY_FMT) + flags |= TRACE_GRAPH_PRINT_DURATION; + else + flags |= TRACE_GRAPH_PRINT_ABS_TIME; + + return __print_graph_function_flags(iter, flags); } static enum print_line_t @@ -1274,7 +1314,7 @@ static void print_lat_header(struct seq_file *s, u32 flags) seq_printf(s, "#%.*s|||| / \n", size, spaces); } -void print_graph_headers_flags(struct seq_file *s, u32 flags) +static void __print_graph_headers_flags(struct seq_file *s, u32 flags) { int lat = trace_flags & TRACE_ITER_LATENCY_FMT; @@ -1315,6 +1355,23 @@ void print_graph_headers(struct seq_file *s) print_graph_headers_flags(s, tracer_flags.val); } +void print_graph_headers_flags(struct seq_file *s, u32 flags) +{ + struct trace_iterator *iter = s->private; + + if (trace_flags & TRACE_ITER_LATENCY_FMT) { + /* print nothing if the buffers are empty */ + if (trace_empty(iter)) + return; + + print_trace_header(s, iter); + flags |= TRACE_GRAPH_PRINT_DURATION; + } else + flags |= TRACE_GRAPH_PRINT_ABS_TIME; + + __print_graph_headers_flags(s, flags); +} + void graph_trace_open(struct trace_iterator *iter) { /* pid and depth on the last trace processed */ diff --git a/kernel/trace/trace_irqsoff.c b/kernel/trace/trace_irqsoff.c index 73a6b0601f2e..4047e98afcba 100644 --- a/kernel/trace/trace_irqsoff.c +++ b/kernel/trace/trace_irqsoff.c @@ -229,75 +229,33 @@ static void irqsoff_trace_close(struct trace_iterator *iter) static enum print_line_t irqsoff_print_line(struct trace_iterator *iter) { - u32 flags = GRAPH_TRACER_FLAGS; - - if (trace_flags & TRACE_ITER_LATENCY_FMT) - flags |= TRACE_GRAPH_PRINT_DURATION; - else - flags |= TRACE_GRAPH_PRINT_ABS_TIME; - /* * In graph mode call the graph tracer output function, * otherwise go with the TRACE_FN event handler */ if (is_graph()) - return print_graph_function_flags(iter, flags); + return print_graph_function_flags(iter, GRAPH_TRACER_FLAGS); return TRACE_TYPE_UNHANDLED; } static void irqsoff_print_header(struct seq_file *s) { - if (is_graph()) { - struct trace_iterator *iter = s->private; - u32 flags = GRAPH_TRACER_FLAGS; - - if (trace_flags & TRACE_ITER_LATENCY_FMT) { - /* print nothing if the buffers are empty */ - if (trace_empty(iter)) - return; - - print_trace_header(s, iter); - flags |= TRACE_GRAPH_PRINT_DURATION; - } else - flags |= TRACE_GRAPH_PRINT_ABS_TIME; - - print_graph_headers_flags(s, flags); - } else + if (is_graph()) + print_graph_headers_flags(s, GRAPH_TRACER_FLAGS); + else trace_default_header(s); } -static void -trace_graph_function(struct trace_array *tr, - unsigned long ip, unsigned long flags, int pc) -{ - u64 time = trace_clock_local(); - struct ftrace_graph_ent ent = { - .func = ip, - .depth = 0, - }; - struct ftrace_graph_ret ret = { - .func = ip, - .depth = 0, - .calltime = time, - .rettime = time, - }; - - __trace_graph_entry(tr, &ent, flags, pc); - __trace_graph_return(tr, &ret, flags, pc); -} - static void __trace_function(struct trace_array *tr, unsigned long ip, unsigned long parent_ip, unsigned long flags, int pc) { - if (!is_graph()) + if (is_graph()) + trace_graph_function(tr, ip, parent_ip, flags, pc); + else trace_function(tr, ip, parent_ip, flags, pc); - else { - trace_graph_function(tr, parent_ip, flags, pc); - trace_graph_function(tr, ip, flags, pc); - } } #else -- cgit v1.2.3 From 7495a5beaa22f190f4888aa8cbe4827c16575d0a Mon Sep 17 00:00:00 2001 From: Jiri Olsa Date: Thu, 23 Sep 2010 14:00:53 +0200 Subject: tracing: Graph support for wakeup tracer Add function graph support for wakeup latency tracer. The graph output is enabled by setting the 'display-graph' trace option. Signed-off-by: Jiri Olsa LKML-Reference: <1285243253-7372-4-git-send-email-jolsa@redhat.com> Signed-off-by: Steven Rostedt --- kernel/trace/trace_sched_wakeup.c | 231 ++++++++++++++++++++++++++++++++++++-- 1 file changed, 221 insertions(+), 10 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace_sched_wakeup.c b/kernel/trace/trace_sched_wakeup.c index 4086eae6e81b..033510dbb322 100644 --- a/kernel/trace/trace_sched_wakeup.c +++ b/kernel/trace/trace_sched_wakeup.c @@ -31,13 +31,33 @@ static int wakeup_rt; static arch_spinlock_t wakeup_lock = (arch_spinlock_t)__ARCH_SPIN_LOCK_UNLOCKED; +static void wakeup_reset(struct trace_array *tr); static void __wakeup_reset(struct trace_array *tr); +static int wakeup_graph_entry(struct ftrace_graph_ent *trace); +static void wakeup_graph_return(struct ftrace_graph_ret *trace); static int save_lat_flag; +#define TRACE_DISPLAY_GRAPH 1 + +static struct tracer_opt trace_opts[] = { +#ifdef CONFIG_FUNCTION_GRAPH_TRACER + /* display latency trace as call graph */ + { TRACER_OPT(display-graph, TRACE_DISPLAY_GRAPH) }, +#endif + { } /* Empty entry */ +}; + +static struct tracer_flags tracer_flags = { + .val = 0, + .opts = trace_opts, +}; + +#define is_graph() (tracer_flags.val & TRACE_DISPLAY_GRAPH) + #ifdef CONFIG_FUNCTION_TRACER /* - * irqsoff uses its own tracer function to keep the overhead down: + * wakeup uses its own tracer function to keep the overhead down: */ static void wakeup_tracer_call(unsigned long ip, unsigned long parent_ip) @@ -80,8 +100,191 @@ static struct ftrace_ops trace_ops __read_mostly = { .func = wakeup_tracer_call, }; + +static int start_func_tracer(int graph) +{ + int ret; + + if (!graph) + ret = register_ftrace_function(&trace_ops); + else + ret = register_ftrace_graph(&wakeup_graph_return, + &wakeup_graph_entry); + + if (!ret && tracing_is_enabled()) + tracer_enabled = 1; + else + tracer_enabled = 0; + + return ret; +} + +static void stop_func_tracer(int graph) +{ + tracer_enabled = 0; + + if (!graph) + unregister_ftrace_function(&trace_ops); + else + unregister_ftrace_graph(); +} + #endif /* CONFIG_FUNCTION_TRACER */ +#ifdef CONFIG_FUNCTION_GRAPH_TRACER +static int wakeup_set_flag(u32 old_flags, u32 bit, int set) +{ + + if (!(bit & TRACE_DISPLAY_GRAPH)) + return -EINVAL; + + if (!(is_graph() ^ set)) + return 0; + + stop_func_tracer(!set); + + wakeup_reset(wakeup_trace); + tracing_max_latency = 0; + + return start_func_tracer(set); +} + +static int wakeup_graph_entry(struct ftrace_graph_ent *trace) +{ + struct trace_array *tr = wakeup_trace; + struct trace_array_cpu *data; + unsigned long flags; + long disabled; + int cpu, pc, ret = 0; + + if (likely(!wakeup_task)) + return 0; + + pc = preempt_count(); + preempt_disable_notrace(); + + cpu = raw_smp_processor_id(); + if (cpu != wakeup_current_cpu) + goto out_enable; + + data = tr->data[cpu]; + disabled = atomic_inc_return(&data->disabled); + if (unlikely(disabled != 1)) + goto out; + + local_save_flags(flags); + ret = __trace_graph_entry(tr, trace, flags, pc); + +out: + atomic_dec(&data->disabled); + +out_enable: + preempt_enable_notrace(); + return ret; +} + +static void wakeup_graph_return(struct ftrace_graph_ret *trace) +{ + struct trace_array *tr = wakeup_trace; + struct trace_array_cpu *data; + unsigned long flags; + long disabled; + int cpu, pc; + + if (likely(!wakeup_task)) + return; + + pc = preempt_count(); + preempt_disable_notrace(); + + cpu = raw_smp_processor_id(); + if (cpu != wakeup_current_cpu) + goto out_enable; + + data = tr->data[cpu]; + disabled = atomic_inc_return(&data->disabled); + if (unlikely(disabled != 1)) + goto out; + + local_save_flags(flags); + __trace_graph_return(tr, trace, flags, pc); + +out: + atomic_dec(&data->disabled); + +out_enable: + preempt_enable_notrace(); + return; +} + +static void wakeup_trace_open(struct trace_iterator *iter) +{ + if (is_graph()) + graph_trace_open(iter); +} + +static void wakeup_trace_close(struct trace_iterator *iter) +{ + if (iter->private) + graph_trace_close(iter); +} + +#define GRAPH_TRACER_FLAGS (TRACE_GRAPH_PRINT_PROC) + +static enum print_line_t wakeup_print_line(struct trace_iterator *iter) +{ + /* + * In graph mode call the graph tracer output function, + * otherwise go with the TRACE_FN event handler + */ + if (is_graph()) + return print_graph_function_flags(iter, GRAPH_TRACER_FLAGS); + + return TRACE_TYPE_UNHANDLED; +} + +static void wakeup_print_header(struct seq_file *s) +{ + if (is_graph()) + print_graph_headers_flags(s, GRAPH_TRACER_FLAGS); + else + trace_default_header(s); +} + +static void +__trace_function(struct trace_array *tr, + unsigned long ip, unsigned long parent_ip, + unsigned long flags, int pc) +{ + if (is_graph()) + trace_graph_function(tr, ip, parent_ip, flags, pc); + else + trace_function(tr, ip, parent_ip, flags, pc); +} +#else +#define __trace_function trace_function + +static int wakeup_set_flag(u32 old_flags, u32 bit, int set) +{ + return -EINVAL; +} + +static int wakeup_graph_entry(struct ftrace_graph_ent *trace) +{ + return -1; +} + +static enum print_line_t wakeup_print_line(struct trace_iterator *iter) +{ + return TRACE_TYPE_UNHANDLED; +} + +static void wakeup_graph_return(struct ftrace_graph_ret *trace) { } +static void wakeup_print_header(struct seq_file *s) { } +static void wakeup_trace_open(struct trace_iterator *iter) { } +static void wakeup_trace_close(struct trace_iterator *iter) { } +#endif /* CONFIG_FUNCTION_GRAPH_TRACER */ + /* * Should this new latency be reported/recorded? */ @@ -152,7 +355,7 @@ probe_wakeup_sched_switch(void *ignore, /* The task we are waiting for is waking up */ data = wakeup_trace->data[wakeup_cpu]; - trace_function(wakeup_trace, CALLER_ADDR0, CALLER_ADDR1, flags, pc); + __trace_function(wakeup_trace, CALLER_ADDR0, CALLER_ADDR1, flags, pc); tracing_sched_switch_trace(wakeup_trace, prev, next, flags, pc); T0 = data->preempt_timestamp; @@ -252,7 +455,7 @@ probe_wakeup(void *ignore, struct task_struct *p, int success) * is not called by an assembly function (where as schedule is) * it should be safe to use it here. */ - trace_function(wakeup_trace, CALLER_ADDR1, CALLER_ADDR2, flags, pc); + __trace_function(wakeup_trace, CALLER_ADDR1, CALLER_ADDR2, flags, pc); out_locked: arch_spin_unlock(&wakeup_lock); @@ -303,12 +506,8 @@ static void start_wakeup_tracer(struct trace_array *tr) */ smp_wmb(); - register_ftrace_function(&trace_ops); - - if (tracing_is_enabled()) - tracer_enabled = 1; - else - tracer_enabled = 0; + if (start_func_tracer(is_graph())) + printk(KERN_ERR "failed to start wakeup tracer\n"); return; fail_deprobe_wake_new: @@ -320,7 +519,7 @@ fail_deprobe: static void stop_wakeup_tracer(struct trace_array *tr) { tracer_enabled = 0; - unregister_ftrace_function(&trace_ops); + stop_func_tracer(is_graph()); unregister_trace_sched_switch(probe_wakeup_sched_switch, NULL); unregister_trace_sched_wakeup_new(probe_wakeup, NULL); unregister_trace_sched_wakeup(probe_wakeup, NULL); @@ -379,9 +578,15 @@ static struct tracer wakeup_tracer __read_mostly = .start = wakeup_tracer_start, .stop = wakeup_tracer_stop, .print_max = 1, + .print_header = wakeup_print_header, + .print_line = wakeup_print_line, + .flags = &tracer_flags, + .set_flag = wakeup_set_flag, #ifdef CONFIG_FTRACE_SELFTEST .selftest = trace_selftest_startup_wakeup, #endif + .open = wakeup_trace_open, + .close = wakeup_trace_close, .use_max_tr = 1, }; @@ -394,9 +599,15 @@ static struct tracer wakeup_rt_tracer __read_mostly = .stop = wakeup_tracer_stop, .wait_pipe = poll_wait_pipe, .print_max = 1, + .print_header = wakeup_print_header, + .print_line = wakeup_print_line, + .flags = &tracer_flags, + .set_flag = wakeup_set_flag, #ifdef CONFIG_FTRACE_SELFTEST .selftest = trace_selftest_startup_wakeup, #endif + .open = wakeup_trace_open, + .close = wakeup_trace_close, .use_max_tr = 1, }; -- cgit v1.2.3 From 542181d3769d001c59cd17573dd4381e87d215f2 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Tue, 5 Oct 2010 16:38:49 -0400 Subject: tracing: Use one prologue for the wakeup tracer function tracers The wakeup tracer has three types of function tracers. Normal function tracer, function graph entry, and function graph return. Each of these use a complex dance to prevent recursion and whether to trace the data or not (depending on the wake_task variable). This patch moves the duplicate code into a single routine, to prevent future mistakes with modifying duplicate complex code. Cc: Jiri Olsa Signed-off-by: Steven Rostedt --- kernel/trace/trace_sched_wakeup.c | 102 +++++++++++++++++++------------------- 1 file changed, 50 insertions(+), 52 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace_sched_wakeup.c b/kernel/trace/trace_sched_wakeup.c index 033510dbb322..31689d2df7f3 100644 --- a/kernel/trace/trace_sched_wakeup.c +++ b/kernel/trace/trace_sched_wakeup.c @@ -56,43 +56,73 @@ static struct tracer_flags tracer_flags = { #define is_graph() (tracer_flags.val & TRACE_DISPLAY_GRAPH) #ifdef CONFIG_FUNCTION_TRACER + /* - * wakeup uses its own tracer function to keep the overhead down: + * Prologue for the wakeup function tracers. + * + * Returns 1 if it is OK to continue, and preemption + * is disabled and data->disabled is incremented. + * 0 if the trace is to be ignored, and preemption + * is not disabled and data->disabled is + * kept the same. + * + * Note, this function is also used outside this ifdef but + * inside the #ifdef of the function graph tracer below. + * This is OK, since the function graph tracer is + * dependent on the function tracer. */ -static void -wakeup_tracer_call(unsigned long ip, unsigned long parent_ip) +static int +func_prolog_preempt_disable(struct trace_array *tr, + struct trace_array_cpu **data, + int *pc) { - struct trace_array *tr = wakeup_trace; - struct trace_array_cpu *data; - unsigned long flags; long disabled; int cpu; - int pc; if (likely(!wakeup_task)) - return; + return 0; - pc = preempt_count(); + *pc = preempt_count(); preempt_disable_notrace(); cpu = raw_smp_processor_id(); if (cpu != wakeup_current_cpu) goto out_enable; - data = tr->data[cpu]; - disabled = atomic_inc_return(&data->disabled); + *data = tr->data[cpu]; + disabled = atomic_inc_return(&(*data)->disabled); if (unlikely(disabled != 1)) goto out; - local_irq_save(flags); + return 1; - trace_function(tr, ip, parent_ip, flags, pc); +out: + atomic_dec(&(*data)->disabled); + +out_enable: + preempt_enable_notrace(); + return 0; +} + +/* + * wakeup uses its own tracer function to keep the overhead down: + */ +static void +wakeup_tracer_call(unsigned long ip, unsigned long parent_ip) +{ + struct trace_array *tr = wakeup_trace; + struct trace_array_cpu *data; + unsigned long flags; + int pc; + + if (!func_prolog_preempt_disable(tr, &data, &pc)) + return; + local_irq_save(flags); + trace_function(tr, ip, parent_ip, flags, pc); local_irq_restore(flags); - out: atomic_dec(&data->disabled); - out_enable: preempt_enable_notrace(); } @@ -154,32 +184,16 @@ static int wakeup_graph_entry(struct ftrace_graph_ent *trace) struct trace_array *tr = wakeup_trace; struct trace_array_cpu *data; unsigned long flags; - long disabled; - int cpu, pc, ret = 0; + int pc, ret = 0; - if (likely(!wakeup_task)) + if (!func_prolog_preempt_disable(tr, &data, &pc)) return 0; - pc = preempt_count(); - preempt_disable_notrace(); - - cpu = raw_smp_processor_id(); - if (cpu != wakeup_current_cpu) - goto out_enable; - - data = tr->data[cpu]; - disabled = atomic_inc_return(&data->disabled); - if (unlikely(disabled != 1)) - goto out; - local_save_flags(flags); ret = __trace_graph_entry(tr, trace, flags, pc); - -out: atomic_dec(&data->disabled); - -out_enable: preempt_enable_notrace(); + return ret; } @@ -188,31 +202,15 @@ static void wakeup_graph_return(struct ftrace_graph_ret *trace) struct trace_array *tr = wakeup_trace; struct trace_array_cpu *data; unsigned long flags; - long disabled; - int cpu, pc; + int pc; - if (likely(!wakeup_task)) + if (!func_prolog_preempt_disable(tr, &data, &pc)) return; - pc = preempt_count(); - preempt_disable_notrace(); - - cpu = raw_smp_processor_id(); - if (cpu != wakeup_current_cpu) - goto out_enable; - - data = tr->data[cpu]; - disabled = atomic_inc_return(&data->disabled); - if (unlikely(disabled != 1)) - goto out; - local_save_flags(flags); __trace_graph_return(tr, trace, flags, pc); - -out: atomic_dec(&data->disabled); -out_enable: preempt_enable_notrace(); return; } -- cgit v1.2.3 From 5e6d2b9cfa3a6e7fe62fc0135bc1bd778f5db564 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Tue, 5 Oct 2010 19:41:43 -0400 Subject: tracing: Use one prologue for the preempt irqs off tracer function tracers The preempt and irqsoff tracers have three types of function tracers. Normal function tracer, function graph entry, and function graph return. Each of these use a complex dance to prevent recursion and whether to trace the data or not (depending if interrupts are enabled or not). This patch moves the duplicate code into a single routine, to prevent future mistakes with modifying duplicate complex code. Cc: Jiri Olsa Signed-off-by: Steven Rostedt --- kernel/trace/trace_irqsoff.c | 96 ++++++++++++++++++++++---------------------- 1 file changed, 48 insertions(+), 48 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace_irqsoff.c b/kernel/trace/trace_irqsoff.c index 4047e98afcba..5cf8c602b880 100644 --- a/kernel/trace/trace_irqsoff.c +++ b/kernel/trace/trace_irqsoff.c @@ -87,14 +87,22 @@ static __cacheline_aligned_in_smp unsigned long max_sequence; #ifdef CONFIG_FUNCTION_TRACER /* - * irqsoff uses its own tracer function to keep the overhead down: + * Prologue for the preempt and irqs off function tracers. + * + * Returns 1 if it is OK to continue, and data->disabled is + * incremented. + * 0 if the trace is to be ignored, and data->disabled + * is kept the same. + * + * Note, this function is also used outside this ifdef but + * inside the #ifdef of the function graph tracer below. + * This is OK, since the function graph tracer is + * dependent on the function tracer. */ -static void -irqsoff_tracer_call(unsigned long ip, unsigned long parent_ip) +static int func_prolog_dec(struct trace_array *tr, + struct trace_array_cpu **data, + unsigned long *flags) { - struct trace_array *tr = irqsoff_trace; - struct trace_array_cpu *data; - unsigned long flags; long disabled; int cpu; @@ -106,18 +114,38 @@ irqsoff_tracer_call(unsigned long ip, unsigned long parent_ip) */ cpu = raw_smp_processor_id(); if (likely(!per_cpu(tracing_cpu, cpu))) - return; + return 0; - local_save_flags(flags); + local_save_flags(*flags); /* slight chance to get a false positive on tracing_cpu */ - if (!irqs_disabled_flags(flags)) - return; + if (!irqs_disabled_flags(*flags)) + return 0; - data = tr->data[cpu]; - disabled = atomic_inc_return(&data->disabled); + *data = tr->data[cpu]; + disabled = atomic_inc_return(&(*data)->disabled); if (likely(disabled == 1)) - trace_function(tr, ip, parent_ip, flags, preempt_count()); + return 1; + + atomic_dec(&(*data)->disabled); + + return 0; +} + +/* + * irqsoff uses its own tracer function to keep the overhead down: + */ +static void +irqsoff_tracer_call(unsigned long ip, unsigned long parent_ip) +{ + struct trace_array *tr = irqsoff_trace; + struct trace_array_cpu *data; + unsigned long flags; + + if (!func_prolog_dec(tr, &data, &flags)) + return; + + trace_function(tr, ip, parent_ip, flags, preempt_count()); atomic_dec(&data->disabled); } @@ -155,30 +183,16 @@ static int irqsoff_graph_entry(struct ftrace_graph_ent *trace) struct trace_array *tr = irqsoff_trace; struct trace_array_cpu *data; unsigned long flags; - long disabled; int ret; - int cpu; int pc; - cpu = raw_smp_processor_id(); - if (likely(!per_cpu(tracing_cpu, cpu))) - return 0; - - local_save_flags(flags); - /* slight chance to get a false positive on tracing_cpu */ - if (!irqs_disabled_flags(flags)) + if (!func_prolog_dec(tr, &data, &flags)) return 0; - data = tr->data[cpu]; - disabled = atomic_inc_return(&data->disabled); - - if (likely(disabled == 1)) { - pc = preempt_count(); - ret = __trace_graph_entry(tr, trace, flags, pc); - } else - ret = 0; - + pc = preempt_count(); + ret = __trace_graph_entry(tr, trace, flags, pc); atomic_dec(&data->disabled); + return ret; } @@ -187,27 +201,13 @@ static void irqsoff_graph_return(struct ftrace_graph_ret *trace) struct trace_array *tr = irqsoff_trace; struct trace_array_cpu *data; unsigned long flags; - long disabled; - int cpu; int pc; - cpu = raw_smp_processor_id(); - if (likely(!per_cpu(tracing_cpu, cpu))) - return; - - local_save_flags(flags); - /* slight chance to get a false positive on tracing_cpu */ - if (!irqs_disabled_flags(flags)) + if (!func_prolog_dec(tr, &data, &flags)) return; - data = tr->data[cpu]; - disabled = atomic_inc_return(&data->disabled); - - if (likely(disabled == 1)) { - pc = preempt_count(); - __trace_graph_return(tr, trace, flags, pc); - } - + pc = preempt_count(); + __trace_graph_return(tr, trace, flags, pc); atomic_dec(&data->disabled); } -- cgit v1.2.3 From 78c89ba121221d9224a5747803d7fffe51cd6e44 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Tue, 5 Oct 2010 23:22:19 -0400 Subject: tracing: Remove parent recording in latency tracer graph options Even though the parent is recorded with the normal function tracing of the latency tracers (irqsoff and wakeup), the function graph recording is bogus. This is due to the function graph messing with the return stack. The latency tracers pass in as the parent CALLER_ADDR0, which works fine for plain function tracing. But this causes bogus output with the graph tracer: 3) -0 | d.s3. 0.000 us | return_to_handler(); 3) -0 | d.s3. 0.000 us | _raw_spin_unlock_irqrestore(); 3) -0 | d.s3. 0.000 us | return_to_handler(); 3) -0 | d.s3. 0.000 us | trace_hardirqs_on(); The "return_to_handle()" call is the trampoline of the function graph tracer, and is meaningless in this context. Cc: Jiri Olsa Signed-off-by: Steven Rostedt --- kernel/trace/trace_functions_graph.c | 1 - 1 file changed, 1 deletion(-) (limited to 'kernel') diff --git a/kernel/trace/trace_functions_graph.c b/kernel/trace/trace_functions_graph.c index 6f8fe28acba1..76b05980225c 100644 --- a/kernel/trace/trace_functions_graph.c +++ b/kernel/trace/trace_functions_graph.c @@ -287,7 +287,6 @@ trace_graph_function(struct trace_array *tr, unsigned long ip, unsigned long parent_ip, unsigned long flags, int pc) { - __trace_graph_function(tr, parent_ip, flags, pc); __trace_graph_function(tr, ip, flags, pc); } -- cgit v1.2.3 From 4924627423d5e286136ad2520f5be536345ae590 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Sun, 17 Oct 2010 21:46:10 +0200 Subject: sched: Unindent labels Labels should be on column 0. Signed-off-by: Peter Zijlstra LKML-Reference: Signed-off-by: Ingo Molnar --- kernel/sched.c | 12 ++++++------ kernel/sched_rt.c | 6 +++--- 2 files changed, 9 insertions(+), 9 deletions(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index 2111491f6424..7f522832250c 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -4891,7 +4891,7 @@ long sched_setaffinity(pid_t pid, const struct cpumask *in_mask) cpuset_cpus_allowed(p, cpus_allowed); cpumask_and(new_mask, in_mask, cpus_allowed); - again: +again: retval = set_cpus_allowed_ptr(p, new_mask); if (!retval) { @@ -8141,9 +8141,9 @@ int alloc_fair_sched_group(struct task_group *tg, struct task_group *parent) return 1; - err_free_rq: +err_free_rq: kfree(cfs_rq); - err: +err: return 0; } @@ -8231,9 +8231,9 @@ int alloc_rt_sched_group(struct task_group *tg, struct task_group *parent) return 1; - err_free_rq: +err_free_rq: kfree(rt_rq); - err: +err: return 0; } @@ -8591,7 +8591,7 @@ static int tg_set_bandwidth(struct task_group *tg, raw_spin_unlock(&rt_rq->rt_runtime_lock); } raw_spin_unlock_irq(&tg->rt_bandwidth.rt_runtime_lock); - unlock: +unlock: read_unlock(&tasklist_lock); mutex_unlock(&rt_constraints_mutex); diff --git a/kernel/sched_rt.c b/kernel/sched_rt.c index baef30f08405..ab77aa00b7b1 100644 --- a/kernel/sched_rt.c +++ b/kernel/sched_rt.c @@ -1140,7 +1140,7 @@ static struct task_struct *pick_next_highest_task_rt(struct rq *rq, int cpu) for_each_leaf_rt_rq(rt_rq, rq) { array = &rt_rq->active; idx = sched_find_first_bit(array->bitmap); - next_idx: +next_idx: if (idx >= MAX_RT_PRIO) continue; if (next && next->prio < idx) @@ -1316,7 +1316,7 @@ static int push_rt_task(struct rq *rq) if (!next_task) return 0; - retry: +retry: if (unlikely(next_task == rq->curr)) { WARN_ON(1); return 0; @@ -1464,7 +1464,7 @@ static int pull_rt_task(struct rq *this_rq) * but possible) */ } - skip: +skip: double_unlock_balance(this_rq, src_rq); } -- cgit v1.2.3 From 34f971f6f7988be4d014eec3e3526bee6d007ffa Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Wed, 22 Sep 2010 13:53:15 +0200 Subject: sched: Create special class for stop/migrate work In order to separate the stop/migrate work thread from the SCHED_FIFO implementation, create a special class for it that is of higher priority than SCHED_FIFO itself. This currently solves a problem where cpu-hotplug consumes so much cpu-time that the SCHED_FIFO class gets throttled, but has the bandwidth replenishment timer pending on the now dead cpu. It is also required for when we add the planned deadline scheduling class above SCHED_FIFO, as the stop/migrate thread still needs to transcent those tasks. Tested-by: Heiko Carstens Signed-off-by: Peter Zijlstra LKML-Reference: <1285165776.2275.1022.camel@laptop> Signed-off-by: Ingo Molnar --- kernel/sched.c | 54 ++++++++++++++++++++---- kernel/sched_stoptask.c | 108 ++++++++++++++++++++++++++++++++++++++++++++++++ kernel/stop_machine.c | 8 ++-- 3 files changed, 158 insertions(+), 12 deletions(-) create mode 100644 kernel/sched_stoptask.c (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index 7f522832250c..5f64fed56a44 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -486,7 +486,7 @@ struct rq { */ unsigned long nr_uninterruptible; - struct task_struct *curr, *idle; + struct task_struct *curr, *idle, *stop; unsigned long next_balance; struct mm_struct *prev_mm; @@ -1837,7 +1837,7 @@ static inline void __set_task_cpu(struct task_struct *p, unsigned int cpu) static const struct sched_class rt_sched_class; -#define sched_class_highest (&rt_sched_class) +#define sched_class_highest (&stop_sched_class) #define for_each_class(class) \ for (class = sched_class_highest; class; class = class->next) @@ -1917,10 +1917,41 @@ static void deactivate_task(struct rq *rq, struct task_struct *p, int flags) #include "sched_idletask.c" #include "sched_fair.c" #include "sched_rt.c" +#include "sched_stoptask.c" #ifdef CONFIG_SCHED_DEBUG # include "sched_debug.c" #endif +void sched_set_stop_task(int cpu, struct task_struct *stop) +{ + struct sched_param param = { .sched_priority = MAX_RT_PRIO - 1 }; + struct task_struct *old_stop = cpu_rq(cpu)->stop; + + if (stop) { + /* + * Make it appear like a SCHED_FIFO task, its something + * userspace knows about and won't get confused about. + * + * Also, it will make PI more or less work without too + * much confusion -- but then, stop work should not + * rely on PI working anyway. + */ + sched_setscheduler_nocheck(stop, SCHED_FIFO, ¶m); + + stop->sched_class = &stop_sched_class; + } + + cpu_rq(cpu)->stop = stop; + + if (old_stop) { + /* + * Reset it back to a normal scheduling class so that + * it can die in pieces. + */ + old_stop->sched_class = &rt_sched_class; + } +} + /* * __normal_prio - return the priority that is based on the static prio */ @@ -3720,17 +3751,13 @@ pick_next_task(struct rq *rq) return p; } - class = sched_class_highest; - for ( ; ; ) { + for_each_class(class) { p = class->pick_next_task(rq); if (p) return p; - /* - * Will never be NULL as the idle class always - * returns a non-NULL p: - */ - class = class->next; } + + BUG(); /* the idle class will always have a runnable task */ } /* @@ -4659,6 +4686,15 @@ recheck: */ rq = __task_rq_lock(p); + /* + * Changing the policy of the stop threads its a very bad idea + */ + if (p == rq->stop) { + __task_rq_unlock(rq); + raw_spin_unlock_irqrestore(&p->pi_lock, flags); + return -EINVAL; + } + #ifdef CONFIG_RT_GROUP_SCHED if (user) { /* diff --git a/kernel/sched_stoptask.c b/kernel/sched_stoptask.c new file mode 100644 index 000000000000..45bddc0c1048 --- /dev/null +++ b/kernel/sched_stoptask.c @@ -0,0 +1,108 @@ +/* + * stop-task scheduling class. + * + * The stop task is the highest priority task in the system, it preempts + * everything and will be preempted by nothing. + * + * See kernel/stop_machine.c + */ + +#ifdef CONFIG_SMP +static int +select_task_rq_stop(struct rq *rq, struct task_struct *p, + int sd_flag, int flags) +{ + return task_cpu(p); /* stop tasks as never migrate */ +} +#endif /* CONFIG_SMP */ + +static void +check_preempt_curr_stop(struct rq *rq, struct task_struct *p, int flags) +{ + resched_task(rq->curr); /* we preempt everything */ +} + +static struct task_struct *pick_next_task_stop(struct rq *rq) +{ + struct task_struct *stop = rq->stop; + + if (stop && stop->state == TASK_RUNNING) + return stop; + + return NULL; +} + +static void +enqueue_task_stop(struct rq *rq, struct task_struct *p, int flags) +{ +} + +static void +dequeue_task_stop(struct rq *rq, struct task_struct *p, int flags) +{ +} + +static void yield_task_stop(struct rq *rq) +{ + BUG(); /* the stop task should never yield, its pointless. */ +} + +static void put_prev_task_stop(struct rq *rq, struct task_struct *prev) +{ +} + +static void task_tick_stop(struct rq *rq, struct task_struct *curr, int queued) +{ +} + +static void set_curr_task_stop(struct rq *rq) +{ +} + +static void switched_to_stop(struct rq *rq, struct task_struct *p, + int running) +{ + BUG(); /* its impossible to change to this class */ +} + +static void prio_changed_stop(struct rq *rq, struct task_struct *p, + int oldprio, int running) +{ + BUG(); /* how!?, what priority? */ +} + +static unsigned int +get_rr_interval_stop(struct rq *rq, struct task_struct *task) +{ + return 0; +} + +/* + * Simple, special scheduling class for the per-CPU stop tasks: + */ +static const struct sched_class stop_sched_class = { + .next = &rt_sched_class, + + .enqueue_task = enqueue_task_stop, + .dequeue_task = dequeue_task_stop, + .yield_task = yield_task_stop, + + .check_preempt_curr = check_preempt_curr_stop, + + .pick_next_task = pick_next_task_stop, + .put_prev_task = put_prev_task_stop, + +#ifdef CONFIG_SMP + .select_task_rq = select_task_rq_stop, +#endif + + .set_curr_task = set_curr_task_stop, + .task_tick = task_tick_stop, + + .get_rr_interval = get_rr_interval_stop, + + .prio_changed = prio_changed_stop, + .switched_to = switched_to_stop, + + /* no .task_new for stop tasks */ +}; diff --git a/kernel/stop_machine.c b/kernel/stop_machine.c index 4372ccb25127..090c28812ce1 100644 --- a/kernel/stop_machine.c +++ b/kernel/stop_machine.c @@ -287,11 +287,12 @@ repeat: goto repeat; } +extern void sched_set_stop_task(int cpu, struct task_struct *stop); + /* manage stopper for a cpu, mostly lifted from sched migration thread mgmt */ static int __cpuinit cpu_stop_cpu_callback(struct notifier_block *nfb, unsigned long action, void *hcpu) { - struct sched_param param = { .sched_priority = MAX_RT_PRIO - 1 }; unsigned int cpu = (unsigned long)hcpu; struct cpu_stopper *stopper = &per_cpu(cpu_stopper, cpu); struct task_struct *p; @@ -304,13 +305,13 @@ static int __cpuinit cpu_stop_cpu_callback(struct notifier_block *nfb, cpu); if (IS_ERR(p)) return NOTIFY_BAD; - sched_setscheduler_nocheck(p, SCHED_FIFO, ¶m); get_task_struct(p); + kthread_bind(p, cpu); + sched_set_stop_task(cpu, p); stopper->thread = p; break; case CPU_ONLINE: - kthread_bind(stopper->thread, cpu); /* strictly unnecessary, as first user will wake it */ wake_up_process(stopper->thread); /* mark enabled */ @@ -325,6 +326,7 @@ static int __cpuinit cpu_stop_cpu_callback(struct notifier_block *nfb, { struct cpu_stop_work *work; + sched_set_stop_task(cpu, NULL); /* kill the stopper */ kthread_stop(stopper->thread); /* drain remaining works */ -- cgit v1.2.3 From 17bdcf949d03306b308c5fb694849cd35f119807 Mon Sep 17 00:00:00 2001 From: Linus Walleij Date: Mon, 11 Oct 2010 16:36:51 +0200 Subject: sched: Drop all load weight manipulation for RT tasks Load weights are for the CFS, they do not belong in the RT task. This makes all RT scheduling classes leave the CFS weights alone. This fixes a real bug as well: I noticed the following phonomena: a process elevated to SCHED_RR forks with SCHED_RESET_ON_FORK set, and the child is indeed SCHED_OTHER, and the niceval is indeed reset to 0. However the weight inserted by set_load_weight() remains at 0, giving the task insignificat priority. With this fix, the weight is reset to what the task had before being elevated to SCHED_RR/SCHED_FIFO. Cc: Lennart Poettering Cc: stable@kernel.org Signed-off-by: Linus Walleij Signed-off-by: Peter Zijlstra LKML-Reference: <1286807811-10568-1-git-send-email-linus.walleij@stericsson.com> Signed-off-by: Ingo Molnar --- kernel/sched.c | 6 ------ 1 file changed, 6 deletions(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index 5f64fed56a44..728081a7ef1c 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -1855,12 +1855,6 @@ static void dec_nr_running(struct rq *rq) static void set_load_weight(struct task_struct *p) { - if (task_has_rt_policy(p)) { - p->se.load.weight = 0; - p->se.load.inv_weight = WMULT_CONST; - return; - } - /* * SCHED_IDLE tasks get minimal weight: */ -- cgit v1.2.3 From 620162505e5d46bc4494b1761743e4b0b3bf8e16 Mon Sep 17 00:00:00 2001 From: Hitoshi Mitake Date: Tue, 5 Oct 2010 18:01:51 +0900 Subject: lockdep: Add improved subclass caching Current lockdep_map only caches one class with subclass == 0, and looks up hash table of classes when subclass != 0. It seems that this has no problem because the case of subclass != 0 is rare. But locks of struct rq are acquired with subclass == 1 when task migration is executed. Task migration is high frequent event, so I modified lockdep to cache subclasses. I measured the score of perf bench sched messaging. This patch has slightly but certain (order of milli seconds or 10 milli seconds) effect when lots of tasks are running. I'll show the result in the tail of this description. NR_LOCKDEP_CACHING_CLASSES specifies how many classes can be cached in the instances of lockdep_map. I discussed with Peter Zijlstra in LinuxCon Japan about this approach and he taught me that caching every subclasses(8) is cleary waste of memory. So number of cached classes should be configurable. === Score comparison of benchmarks === # "min" means best score, and "max" means worst score for i in `seq 1 10`; do ./perf bench -f simple sched messaging; done before: min: 0.565000, max: 0.583000, avg: 0.572500 after: min: 0.559000, max: 0.568000, avg: 0.563300 # with more processes for i in `seq 1 10`; do ./perf bench -f simple sched messaging -g 40; done before: min: 2.274000, max: 2.298000, avg: 2.286300 after: min: 2.242000, max: 2.270000, avg: 2.259700 Signed-off-by: Hitoshi Mitake Cc: Frederic Weisbecker Signed-off-by: Peter Zijlstra LKML-Reference: <1286269311-28336-2-git-send-email-mitake@dcl.info.waseda.ac.jp> Signed-off-by: Ingo Molnar --- kernel/lockdep.c | 25 ++++++++++++++++++------- 1 file changed, 18 insertions(+), 7 deletions(-) (limited to 'kernel') diff --git a/kernel/lockdep.c b/kernel/lockdep.c index 84baa71cfda5..bc4d32871f9a 100644 --- a/kernel/lockdep.c +++ b/kernel/lockdep.c @@ -774,7 +774,9 @@ out_unlock_set: raw_local_irq_restore(flags); if (!subclass || force) - lock->class_cache = class; + lock->class_cache[0] = class; + else if (subclass < NR_LOCKDEP_CACHING_CLASSES) + lock->class_cache[subclass] = class; if (DEBUG_LOCKS_WARN_ON(class->subclass != subclass)) return NULL; @@ -2679,7 +2681,11 @@ static int mark_lock(struct task_struct *curr, struct held_lock *this, void lockdep_init_map(struct lockdep_map *lock, const char *name, struct lock_class_key *key, int subclass) { - lock->class_cache = NULL; + int i; + + for (i = 0; i < NR_LOCKDEP_CACHING_CLASSES; i++) + lock->class_cache[i] = NULL; + #ifdef CONFIG_LOCK_STAT lock->cpu = raw_smp_processor_id(); #endif @@ -2750,10 +2756,10 @@ static int __lock_acquire(struct lockdep_map *lock, unsigned int subclass, if (lock->key == &__lockdep_no_validate__) check = 1; - if (!subclass) - class = lock->class_cache; + if (subclass < NR_LOCKDEP_CACHING_CLASSES) + class = lock->class_cache[subclass]; /* - * Not cached yet or subclass? + * Not cached? */ if (unlikely(!class)) { class = register_lock_class(lock, subclass, 0); @@ -2918,7 +2924,7 @@ static int match_held_lock(struct held_lock *hlock, struct lockdep_map *lock) return 1; if (hlock->references) { - struct lock_class *class = lock->class_cache; + struct lock_class *class = lock->class_cache[0]; if (!class) class = look_up_lock_class(lock, 0); @@ -3559,7 +3565,12 @@ void lockdep_reset_lock(struct lockdep_map *lock) if (list_empty(head)) continue; list_for_each_entry_safe(class, next, head, hash_entry) { - if (unlikely(class == lock->class_cache)) { + int match = 0; + + for (j = 0; j < NR_LOCKDEP_CACHING_CLASSES; j++) + match |= class == lock->class_cache[j]; + + if (unlikely(match)) { if (debug_locks_off_graph_unlock()) WARN_ON(1); goto out_restore; -- cgit v1.2.3 From 4ba053c04aece1f4734056f21b751eee47ea3fb1 Mon Sep 17 00:00:00 2001 From: Hitoshi Mitake Date: Wed, 13 Oct 2010 17:30:26 +0900 Subject: lockdep: Check the depth of subclass Current look_up_lock_class() doesn't check the parameter "subclass". This rarely rises problems because the main caller of this function, register_lock_class(), checks it. But register_lock_class() is not the only function which calls look_up_lock_class(). lock_set_class() and its callees also call it. And lock_set_class() doesn't check this parameter. This will rise problems when the the value of subclass is larger than MAX_LOCKDEP_SUBCLASSES. Because the address (used as the key of class) caliculated with too large subclass has a probability to point another key in different lock_class_key. Of course this problem depends on the memory layout and occurs with really low probability. Signed-off-by: Hitoshi Mitake Cc: Dmitry Torokhov Cc: Vojtech Pavlik Cc: Frederic Weisbecker Signed-off-by: Peter Zijlstra LKML-Reference: <1286958626-986-1-git-send-email-mitake@dcl.info.waseda.ac.jp> Signed-off-by: Ingo Molnar --- kernel/lockdep.c | 18 ++++++++++-------- 1 file changed, 10 insertions(+), 8 deletions(-) (limited to 'kernel') diff --git a/kernel/lockdep.c b/kernel/lockdep.c index bc4d32871f9a..42ba65dff7d9 100644 --- a/kernel/lockdep.c +++ b/kernel/lockdep.c @@ -639,6 +639,16 @@ look_up_lock_class(struct lockdep_map *lock, unsigned int subclass) } #endif + if (unlikely(subclass >= MAX_LOCKDEP_SUBCLASSES)) { + debug_locks_off(); + printk(KERN_ERR + "BUG: looking up invalid subclass: %u\n", subclass); + printk(KERN_ERR + "turning off the locking correctness validator.\n"); + dump_stack(); + return NULL; + } + /* * Static locks do not have their class-keys yet - for them the key * is the lock object itself: @@ -2745,14 +2755,6 @@ static int __lock_acquire(struct lockdep_map *lock, unsigned int subclass, if (DEBUG_LOCKS_WARN_ON(!irqs_disabled())) return 0; - if (unlikely(subclass >= MAX_LOCKDEP_SUBCLASSES)) { - debug_locks_off(); - printk("BUG: MAX_LOCKDEP_SUBCLASSES too low!\n"); - printk("turning off the locking correctness validator.\n"); - dump_stack(); - return 0; - } - if (lock->key == &__lockdep_no_validate__) check = 1; -- cgit v1.2.3 From c530ccd9a1864a44a7ff35826681229ce9f2357a Mon Sep 17 00:00:00 2001 From: Stephane Eranian Date: Fri, 15 Oct 2010 15:26:01 +0200 Subject: perf_events: Fix bogus context time tracking You can only call update_context_time() when the context is active, i.e., the thread it is attached to is still running. However, perf_event_read() can be called even when the context is inactive, e.g., user read() the counters. The call to update_context_time() must be conditioned on the status of the context, otherwise, bogus time_enabled, time_running may be returned. Here is an example on AMD64. The task program is an example from libpfm4. The -p prints deltas every 1s. $ task -p -e cpu_clk_unhalted sleep 5 2,266,610 cpu_clk_unhalted (0.00% scaling, ena=2,158,982, run=2,158,982) 0 cpu_clk_unhalted (0.00% scaling, ena=2,158,982, run=2,158,982) 0 cpu_clk_unhalted (0.00% scaling, ena=2,158,982, run=2,158,982) 0 cpu_clk_unhalted (0.00% scaling, ena=2,158,982, run=2,158,982) 0 cpu_clk_unhalted (0.00% scaling, ena=2,158,982, run=2,158,982) 5,242,358,071 cpu_clk_unhalted (99.95% scaling, ena=5,000,359,984, run=2,319,270) Whereas if you don't read deltas, e.g., no call to perf_event_read() until the process terminates: $ task -e cpu_clk_unhalted sleep 5 2,497,783 cpu_clk_unhalted (0.00% scaling, ena=2,376,899, run=2,376,899) Notice that time_enable, time_running are bogus in the first example causing bogus scaling. This patch fixes the problem, by conditionally calling update_context_time() in perf_event_read(). Signed-off-by: Stephane Eranian Signed-off-by: Peter Zijlstra Cc: stable@kernel.org LKML-Reference: <4cb856dc.51edd80a.5ae0.38fb@mx.google.com> Signed-off-by: Ingo Molnar --- kernel/perf_event.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index 1ec3916ffef0..e7eeba1794fd 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -1780,7 +1780,13 @@ static u64 perf_event_read(struct perf_event *event) unsigned long flags; raw_spin_lock_irqsave(&ctx->lock, flags); - update_context_time(ctx); + /* + * may read while context is not active + * (e.g., thread is blocked), in that case + * we cannot update context time + */ + if (ctx->is_active) + update_context_time(ctx); update_event_times(event); raw_spin_unlock_irqrestore(&ctx->lock, flags); } -- cgit v1.2.3 From 8e5fc1a7320baf6076391607515dceb61319b36a Mon Sep 17 00:00:00 2001 From: Stephane Eranian Date: Fri, 15 Oct 2010 16:54:01 +0200 Subject: perf_events: Fix transaction recovery in group_sched_in() The group_sched_in() function uses a transactional approach to schedule a group of events. In a group, either all events can be scheduled or none are. To schedule each event in, the function calls event_sched_in(). In case of error, event_sched_out() is called on each event in the group. The problem is that event_sched_out() does not completely cancel the effects of event_sched_in(). Furthermore event_sched_out() changes the state of the event as if it had run which is not true is this particular case. Those inconsistencies impact time tracking fields and may lead to events in a group not all reporting the same time_enabled and time_running values. This is demonstrated with the example below: $ task -eunhalted_core_cycles,baclears,baclears -e unhalted_core_cycles,baclears,baclears sleep 5 1946101 unhalted_core_cycles (32.85% scaling, ena=829181, run=556827) 11423 baclears (32.85% scaling, ena=829181, run=556827) 7671 baclears (0.00% scaling, ena=556827, run=556827) 2250443 unhalted_core_cycles (57.83% scaling, ena=962822, run=405995) 11705 baclears (57.83% scaling, ena=962822, run=405995) 11705 baclears (57.83% scaling, ena=962822, run=405995) Notice that in the first group, the last baclears event does not report the same timings as its siblings. This issue comes from the fact that tstamp_stopped is updated by event_sched_out() as if the event had actually run. To solve the issue, we must ensure that, in case of error, there is no change in the event state whatsoever. That means timings must remain as they were when entering group_sched_in(). To do this we defer updating tstamp_running until we know the transaction succeeded. Therefore, we have split event_sched_in() in two parts separating the update to tstamp_running. Similarly, in case of error, we do not want to update tstamp_stopped. Therefore, we have split event_sched_out() in two parts separating the update to tstamp_stopped. With this patch, we now get the following output: $ task -eunhalted_core_cycles,baclears,baclears -e unhalted_core_cycles,baclears,baclears sleep 5 2492050 unhalted_core_cycles (71.75% scaling, ena=1093330, run=308841) 11243 baclears (71.75% scaling, ena=1093330, run=308841) 11243 baclears (71.75% scaling, ena=1093330, run=308841) 1852746 unhalted_core_cycles (0.00% scaling, ena=784489, run=784489) 9253 baclears (0.00% scaling, ena=784489, run=784489) 9253 baclears (0.00% scaling, ena=784489, run=784489) Note that the uneven timing between groups is a side effect of the process spending most of its time sleeping, i.e., not enough event rotations (but that's a separate issue). Signed-off-by: Stephane Eranian Signed-off-by: Peter Zijlstra LKML-Reference: <4cb86b4c.41e9d80a.44e9.3e19@mx.google.com> Signed-off-by: Ingo Molnar --- kernel/perf_event.c | 76 ++++++++++++++++++++++++++++++++++++++++++++--------- 1 file changed, 63 insertions(+), 13 deletions(-) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index e7eeba1794fd..634f86a4b2f9 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -412,8 +412,8 @@ event_filter_match(struct perf_event *event) return event->cpu == -1 || event->cpu == smp_processor_id(); } -static void -event_sched_out(struct perf_event *event, +static int +__event_sched_out(struct perf_event *event, struct perf_cpu_context *cpuctx, struct perf_event_context *ctx) { @@ -432,14 +432,13 @@ event_sched_out(struct perf_event *event, } if (event->state != PERF_EVENT_STATE_ACTIVE) - return; + return 0; event->state = PERF_EVENT_STATE_INACTIVE; if (event->pending_disable) { event->pending_disable = 0; event->state = PERF_EVENT_STATE_OFF; } - event->tstamp_stopped = ctx->time; event->pmu->del(event, 0); event->oncpu = -1; @@ -448,6 +447,19 @@ event_sched_out(struct perf_event *event, ctx->nr_active--; if (event->attr.exclusive || !cpuctx->active_oncpu) cpuctx->exclusive = 0; + return 1; +} + +static void +event_sched_out(struct perf_event *event, + struct perf_cpu_context *cpuctx, + struct perf_event_context *ctx) +{ + int ret; + + ret = __event_sched_out(event, cpuctx, ctx); + if (ret) + event->tstamp_stopped = ctx->time; } static void @@ -647,7 +659,7 @@ retry: } static int -event_sched_in(struct perf_event *event, +__event_sched_in(struct perf_event *event, struct perf_cpu_context *cpuctx, struct perf_event_context *ctx) { @@ -667,8 +679,6 @@ event_sched_in(struct perf_event *event, return -EAGAIN; } - event->tstamp_running += ctx->time - event->tstamp_stopped; - if (!is_software_event(event)) cpuctx->active_oncpu++; ctx->nr_active++; @@ -679,6 +689,35 @@ event_sched_in(struct perf_event *event, return 0; } +static inline int +event_sched_in(struct perf_event *event, + struct perf_cpu_context *cpuctx, + struct perf_event_context *ctx) +{ + int ret = __event_sched_in(event, cpuctx, ctx); + if (ret) + return ret; + event->tstamp_running += ctx->time - event->tstamp_stopped; + return 0; +} + +static void +group_commit_event_sched_in(struct perf_event *group_event, + struct perf_cpu_context *cpuctx, + struct perf_event_context *ctx) +{ + struct perf_event *event; + u64 now = ctx->time; + + group_event->tstamp_running += now - group_event->tstamp_stopped; + /* + * Schedule in siblings as one group (if any): + */ + list_for_each_entry(event, &group_event->sibling_list, group_entry) { + event->tstamp_running += now - event->tstamp_stopped; + } +} + static int group_sched_in(struct perf_event *group_event, struct perf_cpu_context *cpuctx, @@ -692,7 +731,13 @@ group_sched_in(struct perf_event *group_event, pmu->start_txn(pmu); - if (event_sched_in(group_event, cpuctx, ctx)) { + /* + * use __event_sched_in() to delay updating tstamp_running + * until the transaction is committed. In case of failure + * we will keep an unmodified tstamp_running which is a + * requirement to get correct timing information + */ + if (__event_sched_in(group_event, cpuctx, ctx)) { pmu->cancel_txn(pmu); return -EAGAIN; } @@ -701,26 +746,31 @@ group_sched_in(struct perf_event *group_event, * Schedule in siblings as one group (if any): */ list_for_each_entry(event, &group_event->sibling_list, group_entry) { - if (event_sched_in(event, cpuctx, ctx)) { + if (__event_sched_in(event, cpuctx, ctx)) { partial_group = event; goto group_error; } } - if (!pmu->commit_txn(pmu)) + if (!pmu->commit_txn(pmu)) { + /* commit tstamp_running */ + group_commit_event_sched_in(group_event, cpuctx, ctx); return 0; - + } group_error: /* * Groups can be scheduled in as one unit only, so undo any * partial group before returning: + * + * use __event_sched_out() to avoid updating tstamp_stopped + * because the event never actually ran */ list_for_each_entry(event, &group_event->sibling_list, group_entry) { if (event == partial_group) break; - event_sched_out(event, cpuctx, ctx); + __event_sched_out(event, cpuctx, ctx); } - event_sched_out(group_event, cpuctx, ctx); + __event_sched_out(group_event, cpuctx, ctx); pmu->cancel_txn(pmu); -- cgit v1.2.3 From e360adbe29241a0194e10e20595360dd7b98a2b3 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Thu, 14 Oct 2010 14:01:34 +0800 Subject: irq_work: Add generic hardirq context callbacks Provide a mechanism that allows running code in IRQ context. It is most useful for NMI code that needs to interact with the rest of the system -- like wakeup a task to drain buffers. Perf currently has such a mechanism, so extract that and provide it as a generic feature, independent of perf so that others may also benefit. The IRQ context callback is generated through self-IPIs where possible, or on architectures like powerpc the decrementer (the built-in timer facility) is set to generate an interrupt immediately. Architectures that don't have anything like this get to do with a callback from the timer tick. These architectures can call irq_work_run() at the tail of any IRQ handlers that might enqueue such work (like the perf IRQ handler) to avoid undue latencies in processing the work. Signed-off-by: Peter Zijlstra Acked-by: Kyle McMartin Acked-by: Martin Schwidefsky [ various fixes ] Signed-off-by: Huang Ying LKML-Reference: <1287036094.7768.291.camel@yhuang-dev> Signed-off-by: Ingo Molnar --- kernel/Makefile | 2 + kernel/irq_work.c | 164 ++++++++++++++++++++++++++++++++++++++++++++++++++++ kernel/perf_event.c | 104 ++------------------------------- kernel/timer.c | 7 ++- 4 files changed, 176 insertions(+), 101 deletions(-) create mode 100644 kernel/irq_work.c (limited to 'kernel') diff --git a/kernel/Makefile b/kernel/Makefile index d52b473c99a1..4d9bf5f8531f 100644 --- a/kernel/Makefile +++ b/kernel/Makefile @@ -23,6 +23,7 @@ CFLAGS_REMOVE_rtmutex-debug.o = -pg CFLAGS_REMOVE_cgroup-debug.o = -pg CFLAGS_REMOVE_sched_clock.o = -pg CFLAGS_REMOVE_perf_event.o = -pg +CFLAGS_REMOVE_irq_work.o = -pg endif obj-$(CONFIG_FREEZER) += freezer.o @@ -100,6 +101,7 @@ obj-$(CONFIG_TRACING) += trace/ obj-$(CONFIG_X86_DS) += trace/ obj-$(CONFIG_RING_BUFFER) += trace/ obj-$(CONFIG_SMP) += sched_cpupri.o +obj-$(CONFIG_IRQ_WORK) += irq_work.o obj-$(CONFIG_PERF_EVENTS) += perf_event.o obj-$(CONFIG_HAVE_HW_BREAKPOINT) += hw_breakpoint.o obj-$(CONFIG_USER_RETURN_NOTIFIER) += user-return-notifier.o diff --git a/kernel/irq_work.c b/kernel/irq_work.c new file mode 100644 index 000000000000..f16763ff8481 --- /dev/null +++ b/kernel/irq_work.c @@ -0,0 +1,164 @@ +/* + * Copyright (C) 2010 Red Hat, Inc., Peter Zijlstra + * + * Provides a framework for enqueueing and running callbacks from hardirq + * context. The enqueueing is NMI-safe. + */ + +#include +#include +#include +#include + +/* + * An entry can be in one of four states: + * + * free NULL, 0 -> {claimed} : free to be used + * claimed NULL, 3 -> {pending} : claimed to be enqueued + * pending next, 3 -> {busy} : queued, pending callback + * busy NULL, 2 -> {free, claimed} : callback in progress, can be claimed + * + * We use the lower two bits of the next pointer to keep PENDING and BUSY + * flags. + */ + +#define IRQ_WORK_PENDING 1UL +#define IRQ_WORK_BUSY 2UL +#define IRQ_WORK_FLAGS 3UL + +static inline bool irq_work_is_set(struct irq_work *entry, int flags) +{ + return (unsigned long)entry->next & flags; +} + +static inline struct irq_work *irq_work_next(struct irq_work *entry) +{ + unsigned long next = (unsigned long)entry->next; + next &= ~IRQ_WORK_FLAGS; + return (struct irq_work *)next; +} + +static inline struct irq_work *next_flags(struct irq_work *entry, int flags) +{ + unsigned long next = (unsigned long)entry; + next |= flags; + return (struct irq_work *)next; +} + +static DEFINE_PER_CPU(struct irq_work *, irq_work_list); + +/* + * Claim the entry so that no one else will poke at it. + */ +static bool irq_work_claim(struct irq_work *entry) +{ + struct irq_work *next, *nflags; + + do { + next = entry->next; + if ((unsigned long)next & IRQ_WORK_PENDING) + return false; + nflags = next_flags(next, IRQ_WORK_FLAGS); + } while (cmpxchg(&entry->next, next, nflags) != next); + + return true; +} + + +void __weak arch_irq_work_raise(void) +{ + /* + * Lame architectures will get the timer tick callback + */ +} + +/* + * Queue the entry and raise the IPI if needed. + */ +static void __irq_work_queue(struct irq_work *entry) +{ + struct irq_work **head, *next; + + head = &get_cpu_var(irq_work_list); + + do { + next = *head; + /* Can assign non-atomic because we keep the flags set. */ + entry->next = next_flags(next, IRQ_WORK_FLAGS); + } while (cmpxchg(head, next, entry) != next); + + /* The list was empty, raise self-interrupt to start processing. */ + if (!irq_work_next(entry)) + arch_irq_work_raise(); + + put_cpu_var(irq_work_list); +} + +/* + * Enqueue the irq_work @entry, returns true on success, failure when the + * @entry was already enqueued by someone else. + * + * Can be re-enqueued while the callback is still in progress. + */ +bool irq_work_queue(struct irq_work *entry) +{ + if (!irq_work_claim(entry)) { + /* + * Already enqueued, can't do! + */ + return false; + } + + __irq_work_queue(entry); + return true; +} +EXPORT_SYMBOL_GPL(irq_work_queue); + +/* + * Run the irq_work entries on this cpu. Requires to be ran from hardirq + * context with local IRQs disabled. + */ +void irq_work_run(void) +{ + struct irq_work *list, **head; + + head = &__get_cpu_var(irq_work_list); + if (*head == NULL) + return; + + BUG_ON(!in_irq()); + BUG_ON(!irqs_disabled()); + + list = xchg(head, NULL); + while (list != NULL) { + struct irq_work *entry = list; + + list = irq_work_next(list); + + /* + * Clear the PENDING bit, after this point the @entry + * can be re-used. + */ + entry->next = next_flags(NULL, IRQ_WORK_BUSY); + entry->func(entry); + /* + * Clear the BUSY bit and return to the free state if + * no-one else claimed it meanwhile. + */ + cmpxchg(&entry->next, next_flags(NULL, IRQ_WORK_BUSY), NULL); + } +} +EXPORT_SYMBOL_GPL(irq_work_run); + +/* + * Synchronize against the irq_work @entry, ensures the entry is not + * currently in use. + */ +void irq_work_sync(struct irq_work *entry) +{ + WARN_ON_ONCE(irqs_disabled()); + + while (irq_work_is_set(entry, IRQ_WORK_BUSY)) + cpu_relax(); +} +EXPORT_SYMBOL_GPL(irq_work_sync); diff --git a/kernel/perf_event.c b/kernel/perf_event.c index 634f86a4b2f9..99b9700e74d0 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -2206,12 +2206,11 @@ static void free_event_rcu(struct rcu_head *head) kfree(event); } -static void perf_pending_sync(struct perf_event *event); static void perf_buffer_put(struct perf_buffer *buffer); static void free_event(struct perf_event *event) { - perf_pending_sync(event); + irq_work_sync(&event->pending); if (!event->parent) { atomic_dec(&nr_events); @@ -3162,16 +3161,7 @@ void perf_event_wakeup(struct perf_event *event) } } -/* - * Pending wakeups - * - * Handle the case where we need to wakeup up from NMI (or rq->lock) context. - * - * The NMI bit means we cannot possibly take locks. Therefore, maintain a - * single linked list and use cmpxchg() to add entries lockless. - */ - -static void perf_pending_event(struct perf_pending_entry *entry) +static void perf_pending_event(struct irq_work *entry) { struct perf_event *event = container_of(entry, struct perf_event, pending); @@ -3187,89 +3177,6 @@ static void perf_pending_event(struct perf_pending_entry *entry) } } -#define PENDING_TAIL ((struct perf_pending_entry *)-1UL) - -static DEFINE_PER_CPU(struct perf_pending_entry *, perf_pending_head) = { - PENDING_TAIL, -}; - -static void perf_pending_queue(struct perf_pending_entry *entry, - void (*func)(struct perf_pending_entry *)) -{ - struct perf_pending_entry **head; - - if (cmpxchg(&entry->next, NULL, PENDING_TAIL) != NULL) - return; - - entry->func = func; - - head = &get_cpu_var(perf_pending_head); - - do { - entry->next = *head; - } while (cmpxchg(head, entry->next, entry) != entry->next); - - set_perf_event_pending(); - - put_cpu_var(perf_pending_head); -} - -static int __perf_pending_run(void) -{ - struct perf_pending_entry *list; - int nr = 0; - - list = xchg(&__get_cpu_var(perf_pending_head), PENDING_TAIL); - while (list != PENDING_TAIL) { - void (*func)(struct perf_pending_entry *); - struct perf_pending_entry *entry = list; - - list = list->next; - - func = entry->func; - entry->next = NULL; - /* - * Ensure we observe the unqueue before we issue the wakeup, - * so that we won't be waiting forever. - * -- see perf_not_pending(). - */ - smp_wmb(); - - func(entry); - nr++; - } - - return nr; -} - -static inline int perf_not_pending(struct perf_event *event) -{ - /* - * If we flush on whatever cpu we run, there is a chance we don't - * need to wait. - */ - get_cpu(); - __perf_pending_run(); - put_cpu(); - - /* - * Ensure we see the proper queue state before going to sleep - * so that we do not miss the wakeup. -- see perf_pending_handle() - */ - smp_rmb(); - return event->pending.next == NULL; -} - -static void perf_pending_sync(struct perf_event *event) -{ - wait_event(event->waitq, perf_not_pending(event)); -} - -void perf_event_do_pending(void) -{ - __perf_pending_run(); -} - /* * We assume there is only KVM supporting the callbacks. * Later on, we might change it to a list if there is @@ -3319,8 +3226,7 @@ static void perf_output_wakeup(struct perf_output_handle *handle) if (handle->nmi) { handle->event->pending_wakeup = 1; - perf_pending_queue(&handle->event->pending, - perf_pending_event); + irq_work_queue(&handle->event->pending); } else perf_event_wakeup(handle->event); } @@ -4356,8 +4262,7 @@ static int __perf_event_overflow(struct perf_event *event, int nmi, event->pending_kill = POLL_HUP; if (nmi) { event->pending_disable = 1; - perf_pending_queue(&event->pending, - perf_pending_event); + irq_work_queue(&event->pending); } else perf_event_disable(event); } @@ -5374,6 +5279,7 @@ perf_event_alloc(struct perf_event_attr *attr, int cpu, INIT_LIST_HEAD(&event->event_entry); INIT_LIST_HEAD(&event->sibling_list); init_waitqueue_head(&event->waitq); + init_irq_work(&event->pending, perf_pending_event); mutex_init(&event->mmap_mutex); diff --git a/kernel/timer.c b/kernel/timer.c index 97bf05baade7..68a9ae7679b7 100644 --- a/kernel/timer.c +++ b/kernel/timer.c @@ -37,7 +37,7 @@ #include #include #include -#include +#include #include #include @@ -1279,7 +1279,10 @@ void update_process_times(int user_tick) run_local_timers(); rcu_check_callbacks(cpu, user_tick); printk_tick(); - perf_event_do_pending(); +#ifdef CONFIG_IRQ_WORK + if (in_irq()) + irq_work_run(); +#endif scheduler_tick(); run_posix_cpu_timers(p); } -- cgit v1.2.3 From 74c3337c2fc6389d3a57a622a936036b6db6b2e8 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Fri, 15 Oct 2010 11:40:29 +0200 Subject: perf: Fix group moving Matt found we trigger the WARN_ON_ONCE() in perf_group_attach() when we take the move_group path in perf_event_open(). Since we cannot de-construct the group (we rely on it to move the events), we have to simply ignore the double attach. The group state is context invariant and doesn't need changing. Reported-by: Matt Fleming Signed-off-by: Peter Zijlstra LKML-Reference: <1287135757.29097.1368.camel@twins> Signed-off-by: Ingo Molnar --- kernel/perf_event.c | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index 99b9700e74d0..346dc0e35a0a 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -315,7 +315,12 @@ static void perf_group_attach(struct perf_event *event) { struct perf_event *group_leader = event->group_leader; - WARN_ON_ONCE(event->attach_state & PERF_ATTACH_GROUP); + /* + * We can have double attach due to group movement in perf_event_open. + */ + if (event->attach_state & PERF_ATTACH_GROUP) + return; + event->attach_state |= PERF_ATTACH_GROUP; if (group_leader == event) -- cgit v1.2.3 From e7d0bc047548d76feee6b23f7d3d9da927189a50 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Thu, 14 Oct 2010 16:54:51 +0200 Subject: perf: Fix task refcount bugs Currently it looks like find_lively_task_by_vpid() takes a task ref and relies on find_get_context() to drop it. The problem is that perf_event_create_kernel_counter() shouldn't be dropping task refs. Signed-off-by: Peter Zijlstra Acked-by: Frederic Weisbecker Acked-by: Matt Helsley LKML-Reference: <20101014203625.278436085@chello.nl> Signed-off-by: Ingo Molnar --- kernel/perf_event.c | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index 346dc0e35a0a..f928878a1c17 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -2190,11 +2190,9 @@ retry: } } - put_task_struct(task); return ctx; errout: - put_task_struct(task); return ERR_PTR(err); } @@ -5602,7 +5600,7 @@ SYSCALL_DEFINE5(perf_event_open, ctx = find_get_context(pmu, task, cpu); if (IS_ERR(ctx)) { err = PTR_ERR(ctx); - goto err_group_fd; + goto err_task; } /* @@ -5698,6 +5696,9 @@ SYSCALL_DEFINE5(perf_event_open, err_context: put_ctx(ctx); +err_task: + if (task) + put_task_struct(task); err_group_fd: fput_light(group_file, fput_needed); free_event(event); -- cgit v1.2.3 From c6be5a5cb62592d9d661899a2aa78236eb00ffa5 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Thu, 14 Oct 2010 16:59:46 +0200 Subject: perf: Find task before event alloc So that we can pass the task pointer to the event allocation, so that we can use task associated data during event initialization. Signed-off-by: Peter Zijlstra LKML-Reference: <20101014203625.340789919@chello.nl> Signed-off-by: Ingo Molnar --- kernel/perf_event.c | 23 ++++++++++++----------- 1 file changed, 12 insertions(+), 11 deletions(-) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index f928878a1c17..b21d06aaef60 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -5551,10 +5551,18 @@ SYSCALL_DEFINE5(perf_event_open, group_leader = NULL; } + if (pid != -1) { + task = find_lively_task_by_vpid(pid); + if (IS_ERR(task)) { + err = PTR_ERR(task); + goto err_group_fd; + } + } + event = perf_event_alloc(&attr, cpu, group_leader, NULL, NULL); if (IS_ERR(event)) { err = PTR_ERR(event); - goto err_fd; + goto err_task; } /* @@ -5586,21 +5594,13 @@ SYSCALL_DEFINE5(perf_event_open, } } - if (pid != -1) { - task = find_lively_task_by_vpid(pid); - if (IS_ERR(task)) { - err = PTR_ERR(task); - goto err_group_fd; - } - } - /* * Get the target context (task or percpu): */ ctx = find_get_context(pmu, task, cpu); if (IS_ERR(ctx)) { err = PTR_ERR(ctx); - goto err_task; + goto err_alloc; } /* @@ -5696,12 +5696,13 @@ SYSCALL_DEFINE5(perf_event_open, err_context: put_ctx(ctx); +err_alloc: + free_event(event); err_task: if (task) put_task_struct(task); err_group_fd: fput_light(group_file, fput_needed); - free_event(event); err_fd: put_unused_fd(event_fd); return err; -- cgit v1.2.3 From d580ff8699e8811a9af37e9de4dea375401bdeec Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Thu, 14 Oct 2010 17:43:23 +0200 Subject: perf, hw_breakpoint: Fix crash in hw_breakpoint creation hw_breakpoint creation needs to account stuff per-task to ensure there is always sufficient hardware resources to back these things due to ptrace. With the perf per pmu context changes the event initialization no longer has access to the event context, for the simple reason that we need to first find the pmu (result of initialization) before we can find the context. This makes hw_breakpoints unhappy, because it can no longer do per task accounting, cure this by frobbing a task pointer in the event::hw bits for now... Signed-off-by: Peter Zijlstra Cc: Frederic Weisbecker LKML-Reference: <20101014203625.391543667@chello.nl> Signed-off-by: Ingo Molnar --- kernel/hw_breakpoint.c | 8 ++++---- kernel/perf_event.c | 23 ++++++++++++++++++----- 2 files changed, 22 insertions(+), 9 deletions(-) (limited to 'kernel') diff --git a/kernel/hw_breakpoint.c b/kernel/hw_breakpoint.c index 3b714e839c10..2c9120f0afca 100644 --- a/kernel/hw_breakpoint.c +++ b/kernel/hw_breakpoint.c @@ -113,12 +113,12 @@ static unsigned int max_task_bp_pinned(int cpu, enum bp_type_idx type) */ static int task_bp_pinned(struct perf_event *bp, enum bp_type_idx type) { - struct perf_event_context *ctx = bp->ctx; + struct task_struct *tsk = bp->hw.bp_target; struct perf_event *iter; int count = 0; list_for_each_entry(iter, &bp_task_head, hw.bp_list) { - if (iter->ctx == ctx && find_slot_idx(iter) == type) + if (iter->hw.bp_target == tsk && find_slot_idx(iter) == type) count += hw_breakpoint_weight(iter); } @@ -134,7 +134,7 @@ fetch_bp_busy_slots(struct bp_busy_slots *slots, struct perf_event *bp, enum bp_type_idx type) { int cpu = bp->cpu; - struct task_struct *tsk = bp->ctx->task; + struct task_struct *tsk = bp->hw.bp_target; if (cpu >= 0) { slots->pinned = per_cpu(nr_cpu_bp_pinned[type], cpu); @@ -213,7 +213,7 @@ toggle_bp_slot(struct perf_event *bp, bool enable, enum bp_type_idx type, int weight) { int cpu = bp->cpu; - struct task_struct *tsk = bp->ctx->task; + struct task_struct *tsk = bp->hw.bp_target; /* Pinned counter cpu profiling */ if (!tsk) { diff --git a/kernel/perf_event.c b/kernel/perf_event.c index b21d06aaef60..856e20baf13f 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -5255,9 +5255,10 @@ unlock: */ static struct perf_event * perf_event_alloc(struct perf_event_attr *attr, int cpu, - struct perf_event *group_leader, - struct perf_event *parent_event, - perf_overflow_handler_t overflow_handler) + struct task_struct *task, + struct perf_event *group_leader, + struct perf_event *parent_event, + perf_overflow_handler_t overflow_handler) { struct pmu *pmu; struct perf_event *event; @@ -5299,6 +5300,17 @@ perf_event_alloc(struct perf_event_attr *attr, int cpu, event->state = PERF_EVENT_STATE_INACTIVE; + if (task) { + event->attach_state = PERF_ATTACH_TASK; +#ifdef CONFIG_HAVE_HW_BREAKPOINT + /* + * hw_breakpoint is a bit difficult here.. + */ + if (attr->type == PERF_TYPE_BREAKPOINT) + event->hw.bp_target = task; +#endif + } + if (!overflow_handler && parent_event) overflow_handler = parent_event->overflow_handler; @@ -5559,7 +5571,7 @@ SYSCALL_DEFINE5(perf_event_open, } } - event = perf_event_alloc(&attr, cpu, group_leader, NULL, NULL); + event = perf_event_alloc(&attr, cpu, task, group_leader, NULL, NULL); if (IS_ERR(event)) { err = PTR_ERR(event); goto err_task; @@ -5728,7 +5740,7 @@ perf_event_create_kernel_counter(struct perf_event_attr *attr, int cpu, * Get the target context (task or percpu): */ - event = perf_event_alloc(attr, cpu, NULL, NULL, overflow_handler); + event = perf_event_alloc(attr, cpu, task, NULL, NULL, overflow_handler); if (IS_ERR(event)) { err = PTR_ERR(event); goto err; @@ -5996,6 +6008,7 @@ inherit_event(struct perf_event *parent_event, child_event = perf_event_alloc(&parent_event->attr, parent_event->cpu, + child, group_leader, parent_event, NULL); if (IS_ERR(child_event)) -- cgit v1.2.3 From 3b6e901f839f42afb40f614418df82c08b01320a Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Thu, 14 Oct 2010 21:10:38 +0200 Subject: jump_label: Use more consistent naming Now that there's still only a few users around, rename things to make them more consistent. Signed-off-by: Peter Zijlstra LKML-Reference: <20101014203625.448565169@chello.nl> Signed-off-by: Ingo Molnar --- kernel/tracepoint.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/tracepoint.c b/kernel/tracepoint.c index d6073a50a6ca..e95ee7f31d43 100644 --- a/kernel/tracepoint.c +++ b/kernel/tracepoint.c @@ -265,10 +265,10 @@ static void set_tracepoint(struct tracepoint_entry **entry, */ rcu_assign_pointer(elem->funcs, (*entry)->funcs); if (!elem->state && active) { - enable_jump_label(&elem->state); + jump_label_enable(&elem->state); elem->state = active; } else if (elem->state && !active) { - disable_jump_label(&elem->state); + jump_label_disable(&elem->state); elem->state = active; } } @@ -285,7 +285,7 @@ static void disable_tracepoint(struct tracepoint *elem) elem->unregfunc(); if (elem->state) { - disable_jump_label(&elem->state); + jump_label_disable(&elem->state); elem->state = 0; } rcu_assign_pointer(elem->funcs, NULL); -- cgit v1.2.3 From 82cd6def9806dcb6a325fb6abbc1d61388a15f6a Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Thu, 14 Oct 2010 17:57:23 +0200 Subject: perf: Use jump_labels to optimize the scheduler hooks Trades a call + conditional + ret for an unconditional jmp. Acked-by: Frederic Weisbecker Signed-off-by: Peter Zijlstra LKML-Reference: <20101014203625.501657727@chello.nl> Signed-off-by: Ingo Molnar --- kernel/perf_event.c | 24 +++++++++--------------- 1 file changed, 9 insertions(+), 15 deletions(-) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index 856e20baf13f..f7febb02ab97 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -34,7 +34,7 @@ #include -static atomic_t nr_events __read_mostly; +atomic_t perf_task_events __read_mostly; static atomic_t nr_mmap_events __read_mostly; static atomic_t nr_comm_events __read_mostly; static atomic_t nr_task_events __read_mostly; @@ -1311,8 +1311,8 @@ void perf_event_context_sched_out(struct task_struct *task, int ctxn, * accessing the event control register. If a NMI hits, then it will * not restart the event. */ -void perf_event_task_sched_out(struct task_struct *task, - struct task_struct *next) +void __perf_event_task_sched_out(struct task_struct *task, + struct task_struct *next) { int ctxn; @@ -1337,14 +1337,6 @@ static void task_ctx_sched_out(struct perf_event_context *ctx, cpuctx->task_ctx = NULL; } -/* - * Called with IRQs disabled - */ -static void __perf_event_task_sched_out(struct perf_event_context *ctx) -{ - task_ctx_sched_out(ctx, EVENT_ALL); -} - /* * Called with IRQs disabled */ @@ -1494,7 +1486,7 @@ void perf_event_context_sched_in(struct perf_event_context *ctx) * accessing the event control register. If a NMI hits, then it will * keep the event running. */ -void perf_event_task_sched_in(struct task_struct *task) +void __perf_event_task_sched_in(struct task_struct *task) { struct perf_event_context *ctx; int ctxn; @@ -2216,7 +2208,8 @@ static void free_event(struct perf_event *event) irq_work_sync(&event->pending); if (!event->parent) { - atomic_dec(&nr_events); + if (event->attach_state & PERF_ATTACH_TASK) + jump_label_dec(&perf_task_events); if (event->attr.mmap || event->attr.mmap_data) atomic_dec(&nr_mmap_events); if (event->attr.comm) @@ -5354,7 +5347,8 @@ done: event->pmu = pmu; if (!event->parent) { - atomic_inc(&nr_events); + if (event->attach_state & PERF_ATTACH_TASK) + jump_label_inc(&perf_task_events); if (event->attr.mmap || event->attr.mmap_data) atomic_inc(&nr_mmap_events); if (event->attr.comm) @@ -5849,7 +5843,7 @@ static void perf_event_exit_task_context(struct task_struct *child, int ctxn) * our context. */ child_ctx = child->perf_event_ctxp[ctxn]; - __perf_event_task_sched_out(child_ctx); + task_ctx_sched_out(child_ctx, EVENT_ALL); /* * Take the context lock here so that if find_get_context is -- cgit v1.2.3 From 7e54a5a0b655734326dc78c2b5efc1eb35497bb6 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Thu, 14 Oct 2010 22:32:45 +0200 Subject: perf: Optimize sw events Acked-by: Frederic Weisbecker Signed-off-by: Peter Zijlstra LKML-Reference: Signed-off-by: Ingo Molnar --- kernel/perf_event.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index f7febb02ab97..05ecf6f7c672 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -4669,7 +4669,7 @@ static void sw_perf_event_destroy(struct perf_event *event) WARN_ON(event->parent); - atomic_dec(&perf_swevent_enabled[event_id]); + jump_label_dec(&perf_swevent_enabled[event_id]); swevent_hlist_put(event); } @@ -4699,7 +4699,7 @@ static int perf_swevent_init(struct perf_event *event) if (err) return err; - atomic_inc(&perf_swevent_enabled[event_id]); + jump_label_inc(&perf_swevent_enabled[event_id]); event->destroy = sw_perf_event_destroy; } -- cgit v1.2.3 From ef8002f6848236de5adc613063ebeabddea8a6fb Mon Sep 17 00:00:00 2001 From: Nikhil Rao Date: Wed, 13 Oct 2010 12:09:35 -0700 Subject: sched: Do not consider SCHED_IDLE tasks to be cache hot This patch adds a check in task_hot to return if the task has SCHED_IDLE policy. SCHED_IDLE tasks have very low weight, and when run with regular workloads, are typically scheduled many milliseconds apart. There is no need to consider these tasks hot for load balancing. Signed-off-by: Nikhil Rao Signed-off-by: Peter Zijlstra LKML-Reference: <1287173550-30365-2-git-send-email-ncrao@google.com> Signed-off-by: Ingo Molnar --- kernel/sched.c | 3 +++ 1 file changed, 3 insertions(+) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index 728081a7ef1c..771b518e5f1f 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -2025,6 +2025,9 @@ task_hot(struct task_struct *p, u64 now, struct sched_domain *sd) if (p->sched_class != &fair_sched_class) return 0; + if (unlikely(p->policy == SCHED_IDLE)) + return 0; + /* * Buddy candidates are cache hot: */ -- cgit v1.2.3 From 2582f0eba54066b5e98ff2b27ef0cfa833b59f54 Mon Sep 17 00:00:00 2001 From: Nikhil Rao Date: Wed, 13 Oct 2010 12:09:36 -0700 Subject: sched: Set group_imb only a task can be pulled from the busiest cpu When cycling through sched groups to determine the busiest group, set group_imb only if the busiest cpu has more than 1 runnable task. This patch fixes the case where two cpus in a group have one runnable task each, but there is a large weight differential between these two tasks. The load balancer is unable to migrate any task from this group, and hence do not consider this group to be imbalanced. Signed-off-by: Nikhil Rao Signed-off-by: Peter Zijlstra LKML-Reference: <1286996978-7007-3-git-send-email-ncrao@google.com> [ small code readability edits ] Signed-off-by: Ingo Molnar --- kernel/sched_fair.c | 12 +++++++----- 1 file changed, 7 insertions(+), 5 deletions(-) (limited to 'kernel') diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c index bf87192e97fe..3656480e0f79 100644 --- a/kernel/sched_fair.c +++ b/kernel/sched_fair.c @@ -2378,7 +2378,7 @@ static inline void update_sg_lb_stats(struct sched_domain *sd, int local_group, const struct cpumask *cpus, int *balance, struct sg_lb_stats *sgs) { - unsigned long load, max_cpu_load, min_cpu_load; + unsigned long load, max_cpu_load, min_cpu_load, max_nr_running; int i; unsigned int balance_cpu = -1, first_idle_cpu = 0; unsigned long avg_load_per_task = 0; @@ -2389,6 +2389,7 @@ static inline void update_sg_lb_stats(struct sched_domain *sd, /* Tally up the load of all CPUs in the group */ max_cpu_load = 0; min_cpu_load = ~0UL; + max_nr_running = 0; for_each_cpu_and(i, sched_group_cpus(group), cpus) { struct rq *rq = cpu_rq(i); @@ -2406,8 +2407,10 @@ static inline void update_sg_lb_stats(struct sched_domain *sd, load = target_load(i, load_idx); } else { load = source_load(i, load_idx); - if (load > max_cpu_load) + if (load > max_cpu_load) { max_cpu_load = load; + max_nr_running = rq->nr_running; + } if (min_cpu_load > load) min_cpu_load = load; } @@ -2447,11 +2450,10 @@ static inline void update_sg_lb_stats(struct sched_domain *sd, if (sgs->sum_nr_running) avg_load_per_task = sgs->sum_weighted_load / sgs->sum_nr_running; - if ((max_cpu_load - min_cpu_load) > 2*avg_load_per_task) + if ((max_cpu_load - min_cpu_load) > 2*avg_load_per_task && max_nr_running > 1) sgs->group_imb = 1; - sgs->group_capacity = - DIV_ROUND_CLOSEST(group->cpu_power, SCHED_LOAD_SCALE); + sgs->group_capacity = DIV_ROUND_CLOSEST(group->cpu_power, SCHED_LOAD_SCALE); if (!sgs->group_capacity) sgs->group_capacity = fix_small_capacity(sd, group); } -- cgit v1.2.3 From fab476228ba37907ad75216d0fd9732ada9c119e Mon Sep 17 00:00:00 2001 From: Nikhil Rao Date: Fri, 15 Oct 2010 13:12:29 -0700 Subject: sched: Force balancing on newidle balance if local group has capacity This patch forces a load balance on a newly idle cpu when the local group has extra capacity and the busiest group does not have any. It improves system utilization when balancing tasks with a large weight differential. Under certain situations, such as a niced down task (i.e. nice = -15) in the presence of nr_cpus NICE0 tasks, the niced task lands on a sched group and kicks away other tasks because of its large weight. This leads to sub-optimal utilization of the machine. Even though the sched group has capacity, it does not pull tasks because sds.this_load >> sds.max_load, and f_b_g() returns NULL. With this patch, if the local group has extra capacity, we shortcut the checks in f_b_g() and try to pull a task over. A sched group has extra capacity if the group capacity is greater than the number of running tasks in that group. Thanks to Mike Galbraith for discussions leading to this patch and for the insight to reuse SD_NEWIDLE_BALANCE. Signed-off-by: Nikhil Rao Signed-off-by: Peter Zijlstra LKML-Reference: <1287173550-30365-4-git-send-email-ncrao@google.com> Signed-off-by: Ingo Molnar --- kernel/sched_fair.c | 28 +++++++++++++++++++++++++--- 1 file changed, 25 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c index 3656480e0f79..032b548be0fc 100644 --- a/kernel/sched_fair.c +++ b/kernel/sched_fair.c @@ -1764,6 +1764,10 @@ static void pull_task(struct rq *src_rq, struct task_struct *p, set_task_cpu(p, this_cpu); activate_task(this_rq, p, 0); check_preempt_curr(this_rq, p, 0); + + /* re-arm NEWIDLE balancing when moving tasks */ + src_rq->avg_idle = this_rq->avg_idle = 2*sysctl_sched_migration_cost; + this_rq->idle_stamp = 0; } /* @@ -2030,12 +2034,14 @@ struct sd_lb_stats { unsigned long this_load; unsigned long this_load_per_task; unsigned long this_nr_running; + unsigned long this_has_capacity; /* Statistics of the busiest group */ unsigned long max_load; unsigned long busiest_load_per_task; unsigned long busiest_nr_running; unsigned long busiest_group_capacity; + unsigned long busiest_has_capacity; int group_imb; /* Is there imbalance in this sd */ #if defined(CONFIG_SCHED_MC) || defined(CONFIG_SCHED_SMT) @@ -2058,6 +2064,7 @@ struct sg_lb_stats { unsigned long sum_weighted_load; /* Weighted load of group's tasks */ unsigned long group_capacity; int group_imb; /* Is there an imbalance in the group ? */ + int group_has_capacity; /* Is there extra capacity in the group? */ }; /** @@ -2456,6 +2463,9 @@ static inline void update_sg_lb_stats(struct sched_domain *sd, sgs->group_capacity = DIV_ROUND_CLOSEST(group->cpu_power, SCHED_LOAD_SCALE); if (!sgs->group_capacity) sgs->group_capacity = fix_small_capacity(sd, group); + + if (sgs->group_capacity > sgs->sum_nr_running) + sgs->group_has_capacity = 1; } /** @@ -2554,12 +2564,14 @@ static inline void update_sd_lb_stats(struct sched_domain *sd, int this_cpu, sds->this = sg; sds->this_nr_running = sgs.sum_nr_running; sds->this_load_per_task = sgs.sum_weighted_load; + sds->this_has_capacity = sgs.group_has_capacity; } else if (update_sd_pick_busiest(sd, sds, sg, &sgs, this_cpu)) { sds->max_load = sgs.avg_load; sds->busiest = sg; sds->busiest_nr_running = sgs.sum_nr_running; sds->busiest_group_capacity = sgs.group_capacity; sds->busiest_load_per_task = sgs.sum_weighted_load; + sds->busiest_has_capacity = sgs.group_has_capacity; sds->group_imb = sgs.group_imb; } @@ -2756,6 +2768,7 @@ static inline void calculate_imbalance(struct sd_lb_stats *sds, int this_cpu, return fix_small_imbalance(sds, this_cpu, imbalance); } + /******* find_busiest_group() helpers end here *********************/ /** @@ -2807,6 +2820,11 @@ find_busiest_group(struct sched_domain *sd, int this_cpu, * 4) This group is more busy than the avg busieness at this * sched_domain. * 5) The imbalance is within the specified limit. + * + * Note: when doing newidle balance, if the local group has excess + * capacity (i.e. nr_running < group_capacity) and the busiest group + * does not have any capacity, we force a load balance to pull tasks + * to the local group. In this case, we skip past checks 3, 4 and 5. */ if (!(*balance)) goto ret; @@ -2818,6 +2836,11 @@ find_busiest_group(struct sched_domain *sd, int this_cpu, if (!sds.busiest || sds.busiest_nr_running == 0) goto out_balanced; + /* SD_BALANCE_NEWIDLE trumps SMP nice when underutilized */ + if (idle == CPU_NEWLY_IDLE && sds.this_has_capacity && + !sds.busiest_has_capacity) + goto force_balance; + if (sds.this_load >= sds.max_load) goto out_balanced; @@ -2829,6 +2852,7 @@ find_busiest_group(struct sched_domain *sd, int this_cpu, if (100 * sds.max_load <= sd->imbalance_pct * sds.this_load) goto out_balanced; +force_balance: /* Looks like there is an imbalance. Compute it */ calculate_imbalance(&sds, this_cpu, imbalance); return sds.busiest; @@ -3162,10 +3186,8 @@ static void idle_balance(int this_cpu, struct rq *this_rq) interval = msecs_to_jiffies(sd->balance_interval); if (time_after(next_balance, sd->last_balance + interval)) next_balance = sd->last_balance + interval; - if (pulled_task) { - this_rq->idle_stamp = 0; + if (pulled_task) break; - } } raw_spin_lock(&this_rq->lock); -- cgit v1.2.3 From 75dd321d79d495a0ee579e6249ebc38ddbb2667f Mon Sep 17 00:00:00 2001 From: Nikhil Rao Date: Fri, 15 Oct 2010 13:12:30 -0700 Subject: sched: Drop group_capacity to 1 only if local group has extra capacity When SD_PREFER_SIBLING is set on a sched domain, drop group_capacity to 1 only if the local group has extra capacity. The extra check prevents the case where you always pull from the heaviest group when it is already under-utilized (possible with a large weight task outweighs the tasks on the system). For example, consider a 16-cpu quad-core quad-socket machine with MC and NUMA scheduling domains. Let's say we spawn 15 nice0 tasks and one nice-15 task, and each task is running on one core. In this case, we observe the following events when balancing at the NUMA domain: - find_busiest_group() will always pick the sched group containing the niced task to be the busiest group. - find_busiest_queue() will then always pick one of the cpus running the nice0 task (never picks the cpu with the nice -15 task since weighted_cpuload > imbalance). - The load balancer fails to migrate the task since it is the running task and increments sd->nr_balance_failed. - It repeats the above steps a few more times until sd->nr_balance_failed > 5, at which point it kicks off the active load balancer, wakes up the migration thread and kicks the nice 0 task off the cpu. The load balancer doesn't stop until we kick out all nice 0 tasks from the sched group, leaving you with 3 idle cpus and one cpu running the nice -15 task. When balancing at the NUMA domain, we drop sgs.group_capacity to 1 if the child domain (in this case MC) has SD_PREFER_SIBLING set. Subsequent load checks are not relevant because the niced task has a very large weight. In this patch, we add an extra condition to the "if(prefer_sibling)" check in update_sd_lb_stats(). We drop the capacity of a group only if the local group has extra capacity, ie. nr_running < group_capacity. This patch preserves the original intent of the prefer_siblings check (to spread tasks across the system in low utilization scenarios) and fixes the case above. It helps in the following ways: - In low utilization cases (where nr_tasks << nr_cpus), we still drop group_capacity down to 1 if we prefer siblings. - On very busy systems (where nr_tasks >> nr_cpus), sgs.nr_running will most likely be > sgs.group_capacity. - When balancing large weight tasks, if the local group does not have extra capacity, we do not pick the group with the niced task as the busiest group. This prevents failed balances, active migration and the under-utilization described above. Signed-off-by: Nikhil Rao Signed-off-by: Peter Zijlstra LKML-Reference: <1287173550-30365-5-git-send-email-ncrao@google.com> Signed-off-by: Ingo Molnar --- kernel/sched_fair.c | 9 +++++++-- 1 file changed, 7 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c index 032b548be0fc..f1c615ff39d6 100644 --- a/kernel/sched_fair.c +++ b/kernel/sched_fair.c @@ -2554,9 +2554,14 @@ static inline void update_sd_lb_stats(struct sched_domain *sd, int this_cpu, /* * In case the child domain prefers tasks go to siblings * first, lower the sg capacity to one so that we'll try - * and move all the excess tasks away. + * and move all the excess tasks away. We lower the capacity + * of a group only if the local group has the capacity to fit + * these excess tasks, i.e. nr_running < group_capacity. The + * extra check prevents the case where you always pull from the + * heaviest group when it is already under-utilized (possible + * with a large weight task outweighs the tasks on the system). */ - if (prefer_sibling) + if (prefer_sibling && !local_group && sds->this_has_capacity) sgs.group_capacity = min(sgs.group_capacity, 1UL); if (local_group) { -- cgit v1.2.3 From 75e1056f5c57050415b64cb761a3acc35d91f013 Mon Sep 17 00:00:00 2001 From: Venkatesh Pallipadi Date: Mon, 4 Oct 2010 17:03:16 -0700 Subject: sched: Fix softirq time accounting Peter Zijlstra found a bug in the way softirq time is accounted in VIRT_CPU_ACCOUNTING on this thread: http://lkml.indiana.edu/hypermail//linux/kernel/1009.2/01366.html The problem is, softirq processing uses local_bh_disable internally. There is no way, later in the flow, to differentiate between whether softirq is being processed or is it just that bh has been disabled. So, a hardirq when bh is disabled results in time being wrongly accounted as softirq. Looking at the code a bit more, the problem exists in !VIRT_CPU_ACCOUNTING as well. As account_system_time() in normal tick based accouting also uses softirq_count, which will be set even when not in softirq with bh disabled. Peter also suggested solution of using 2*SOFTIRQ_OFFSET as irq count for local_bh_{disable,enable} and using just SOFTIRQ_OFFSET while softirq processing. The patch below does that and adds API in_serving_softirq() which returns whether we are currently processing softirq or not. Also changes one of the usages of softirq_count in net/sched/cls_cgroup.c to in_serving_softirq. Looks like many usages of in_softirq really want in_serving_softirq. Those changes can be made individually on a case by case basis. Signed-off-by: Venkatesh Pallipadi Signed-off-by: Peter Zijlstra LKML-Reference: <1286237003-12406-2-git-send-email-venki@google.com> Signed-off-by: Ingo Molnar --- kernel/sched.c | 2 +- kernel/softirq.c | 51 ++++++++++++++++++++++++++++++++++----------------- 2 files changed, 35 insertions(+), 18 deletions(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index 771b518e5f1f..089be8adb074 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -3422,7 +3422,7 @@ void account_system_time(struct task_struct *p, int hardirq_offset, tmp = cputime_to_cputime64(cputime); if (hardirq_count() - hardirq_offset) cpustat->irq = cputime64_add(cpustat->irq, tmp); - else if (softirq_count()) + else if (in_serving_softirq()) cpustat->softirq = cputime64_add(cpustat->softirq, tmp); else cpustat->system = cputime64_add(cpustat->system, tmp); diff --git a/kernel/softirq.c b/kernel/softirq.c index 07b4f1b1a73a..988dfbe6bbe8 100644 --- a/kernel/softirq.c +++ b/kernel/softirq.c @@ -76,12 +76,22 @@ void wakeup_softirqd(void) wake_up_process(tsk); } +/* + * preempt_count and SOFTIRQ_OFFSET usage: + * - preempt_count is changed by SOFTIRQ_OFFSET on entering or leaving + * softirq processing. + * - preempt_count is changed by SOFTIRQ_DISABLE_OFFSET (= 2 * SOFTIRQ_OFFSET) + * on local_bh_disable or local_bh_enable. + * This lets us distinguish between whether we are currently processing + * softirq and whether we just have bh disabled. + */ + /* * This one is for softirq.c-internal use, * where hardirqs are disabled legitimately: */ #ifdef CONFIG_TRACE_IRQFLAGS -static void __local_bh_disable(unsigned long ip) +static void __local_bh_disable(unsigned long ip, unsigned int cnt) { unsigned long flags; @@ -95,32 +105,43 @@ static void __local_bh_disable(unsigned long ip) * We must manually increment preempt_count here and manually * call the trace_preempt_off later. */ - preempt_count() += SOFTIRQ_OFFSET; + preempt_count() += cnt; /* * Were softirqs turned off above: */ - if (softirq_count() == SOFTIRQ_OFFSET) + if (softirq_count() == cnt) trace_softirqs_off(ip); raw_local_irq_restore(flags); - if (preempt_count() == SOFTIRQ_OFFSET) + if (preempt_count() == cnt) trace_preempt_off(CALLER_ADDR0, get_parent_ip(CALLER_ADDR1)); } #else /* !CONFIG_TRACE_IRQFLAGS */ -static inline void __local_bh_disable(unsigned long ip) +static inline void __local_bh_disable(unsigned long ip, unsigned int cnt) { - add_preempt_count(SOFTIRQ_OFFSET); + add_preempt_count(cnt); barrier(); } #endif /* CONFIG_TRACE_IRQFLAGS */ void local_bh_disable(void) { - __local_bh_disable((unsigned long)__builtin_return_address(0)); + __local_bh_disable((unsigned long)__builtin_return_address(0), + SOFTIRQ_DISABLE_OFFSET); } EXPORT_SYMBOL(local_bh_disable); +static void __local_bh_enable(unsigned int cnt) +{ + WARN_ON_ONCE(in_irq()); + WARN_ON_ONCE(!irqs_disabled()); + + if (softirq_count() == cnt) + trace_softirqs_on((unsigned long)__builtin_return_address(0)); + sub_preempt_count(cnt); +} + /* * Special-case - softirqs can safely be enabled in * cond_resched_softirq(), or by __do_softirq(), @@ -128,12 +149,7 @@ EXPORT_SYMBOL(local_bh_disable); */ void _local_bh_enable(void) { - WARN_ON_ONCE(in_irq()); - WARN_ON_ONCE(!irqs_disabled()); - - if (softirq_count() == SOFTIRQ_OFFSET) - trace_softirqs_on((unsigned long)__builtin_return_address(0)); - sub_preempt_count(SOFTIRQ_OFFSET); + __local_bh_enable(SOFTIRQ_DISABLE_OFFSET); } EXPORT_SYMBOL(_local_bh_enable); @@ -147,13 +163,13 @@ static inline void _local_bh_enable_ip(unsigned long ip) /* * Are softirqs going to be turned on now: */ - if (softirq_count() == SOFTIRQ_OFFSET) + if (softirq_count() == SOFTIRQ_DISABLE_OFFSET) trace_softirqs_on(ip); /* * Keep preemption disabled until we are done with * softirq processing: */ - sub_preempt_count(SOFTIRQ_OFFSET - 1); + sub_preempt_count(SOFTIRQ_DISABLE_OFFSET - 1); if (unlikely(!in_interrupt() && local_softirq_pending())) do_softirq(); @@ -198,7 +214,8 @@ asmlinkage void __do_softirq(void) pending = local_softirq_pending(); account_system_vtime(current); - __local_bh_disable((unsigned long)__builtin_return_address(0)); + __local_bh_disable((unsigned long)__builtin_return_address(0), + SOFTIRQ_OFFSET); lockdep_softirq_enter(); cpu = smp_processor_id(); @@ -245,7 +262,7 @@ restart: lockdep_softirq_exit(); account_system_vtime(current); - _local_bh_enable(); + __local_bh_enable(SOFTIRQ_OFFSET); } #ifndef __ARCH_HAS_DO_SOFTIRQ -- cgit v1.2.3 From 6cdd5199daf0cb7b0fcc8dca941af08492612887 Mon Sep 17 00:00:00 2001 From: Venkatesh Pallipadi Date: Mon, 4 Oct 2010 17:03:18 -0700 Subject: sched: Add a PF flag for ksoftirqd identification To account softirq time cleanly in scheduler, we need to identify whether softirq is invoked in ksoftirqd context or softirq at hardirq tail context. Add PF_KSOFTIRQD for that purpose. As all PF flag bits are currently taken, create space by moving one of the infrequently used bits (PF_THREAD_BOUND) down in task_struct to be along with some other state fields. Signed-off-by: Venkatesh Pallipadi Signed-off-by: Peter Zijlstra LKML-Reference: <1286237003-12406-4-git-send-email-venki@google.com> Signed-off-by: Ingo Molnar --- kernel/softirq.c | 1 + 1 file changed, 1 insertion(+) (limited to 'kernel') diff --git a/kernel/softirq.c b/kernel/softirq.c index 988dfbe6bbe8..267f7b763ebb 100644 --- a/kernel/softirq.c +++ b/kernel/softirq.c @@ -713,6 +713,7 @@ static int run_ksoftirqd(void * __bind_cpu) { set_current_state(TASK_INTERRUPTIBLE); + current->flags |= PF_KSOFTIRQD; while (!kthread_should_stop()) { preempt_disable(); if (!local_softirq_pending()) { -- cgit v1.2.3 From b52bfee445d315549d41eacf2fa7c156e7d153d5 Mon Sep 17 00:00:00 2001 From: Venkatesh Pallipadi Date: Mon, 4 Oct 2010 17:03:19 -0700 Subject: sched: Add IRQ_TIME_ACCOUNTING, finer accounting of irq time s390/powerpc/ia64 have support for CONFIG_VIRT_CPU_ACCOUNTING which does the fine granularity accounting of user, system, hardirq, softirq times. Adding that option on archs like x86 will be challenging however, given the state of TSC reliability on various platforms and also the overhead it will add in syscall entry exit. Instead, add a lighter variant that only does finer accounting of hardirq and softirq times, providing precise irq times (instead of timer tick based samples). This accounting is added with a new config option CONFIG_IRQ_TIME_ACCOUNTING so that there won't be any overhead for users not interested in paying the perf penalty. This accounting is based on sched_clock, with the code being generic. So, other archs may find it useful as well. This patch just adds the core logic and does not enable this logic yet. Signed-off-by: Venkatesh Pallipadi Signed-off-by: Peter Zijlstra LKML-Reference: <1286237003-12406-5-git-send-email-venki@google.com> Signed-off-by: Ingo Molnar --- kernel/sched.c | 49 +++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 49 insertions(+) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index 089be8adb074..9b302e355791 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -1908,6 +1908,55 @@ static void deactivate_task(struct rq *rq, struct task_struct *p, int flags) dec_nr_running(rq); } +#ifdef CONFIG_IRQ_TIME_ACCOUNTING + +static DEFINE_PER_CPU(u64, cpu_hardirq_time); +static DEFINE_PER_CPU(u64, cpu_softirq_time); + +static DEFINE_PER_CPU(u64, irq_start_time); +static int sched_clock_irqtime; + +void enable_sched_clock_irqtime(void) +{ + sched_clock_irqtime = 1; +} + +void disable_sched_clock_irqtime(void) +{ + sched_clock_irqtime = 0; +} + +void account_system_vtime(struct task_struct *curr) +{ + unsigned long flags; + int cpu; + u64 now, delta; + + if (!sched_clock_irqtime) + return; + + local_irq_save(flags); + + now = sched_clock(); + cpu = smp_processor_id(); + delta = now - per_cpu(irq_start_time, cpu); + per_cpu(irq_start_time, cpu) = now; + /* + * We do not account for softirq time from ksoftirqd here. + * We want to continue accounting softirq time to ksoftirqd thread + * in that case, so as not to confuse scheduler with a special task + * that do not consume any time, but still wants to run. + */ + if (hardirq_count()) + per_cpu(cpu_hardirq_time, cpu) += delta; + else if (in_serving_softirq() && !(curr->flags & PF_KSOFTIRQD)) + per_cpu(cpu_softirq_time, cpu) += delta; + + local_irq_restore(flags); +} + +#endif + #include "sched_idletask.c" #include "sched_fair.c" #include "sched_rt.c" -- cgit v1.2.3 From 305e6835e05513406fa12820e40e4a8ecb63743c Mon Sep 17 00:00:00 2001 From: Venkatesh Pallipadi Date: Mon, 4 Oct 2010 17:03:21 -0700 Subject: sched: Do not account irq time to current task Scheduler accounts both softirq and interrupt processing times to the currently running task. This means, if the interrupt processing was for some other task in the system, then the current task ends up being penalized as it gets shorter runtime than otherwise. Change sched task accounting to acoount only actual task time from currently running task. Now update_curr(), modifies the delta_exec to depend on rq->clock_task. Note that this change only handles CONFIG_IRQ_TIME_ACCOUNTING case. We can extend this to CONFIG_VIRT_CPU_ACCOUNTING with minimal effort. But, thats for later. This change will impact scheduling behavior in interrupt heavy conditions. Tested on a 4-way system with eth0 handled by CPU 2 and a network heavy task (nc) running on CPU 3 (and no RSS/RFS). With that I have CPU 2 spending 75%+ of its time in irq processing. CPU 3 spending around 35% time running nc task. Now, if I run another CPU intensive task on CPU 2, without this change /proc//schedstat shows 100% of time accounted to this task. With this change, it rightly shows less than 25% accounted to this task as remaining time is actually spent on irq processing. Signed-off-by: Venkatesh Pallipadi Signed-off-by: Peter Zijlstra LKML-Reference: <1286237003-12406-7-git-send-email-venki@google.com> Signed-off-by: Ingo Molnar --- kernel/sched.c | 43 ++++++++++++++++++++++++++++++++++++++++--- kernel/sched_fair.c | 6 +++--- kernel/sched_rt.c | 8 ++++---- 3 files changed, 47 insertions(+), 10 deletions(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index 9b302e355791..9e01b7100ef6 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -491,6 +491,7 @@ struct rq { struct mm_struct *prev_mm; u64 clock; + u64 clock_task; atomic_t nr_iowait; @@ -641,10 +642,19 @@ static inline struct task_group *task_group(struct task_struct *p) #endif /* CONFIG_CGROUP_SCHED */ +static u64 irq_time_cpu(int cpu); + inline void update_rq_clock(struct rq *rq) { - if (!rq->skip_clock_update) - rq->clock = sched_clock_cpu(cpu_of(rq)); + if (!rq->skip_clock_update) { + int cpu = cpu_of(rq); + u64 irq_time; + + rq->clock = sched_clock_cpu(cpu); + irq_time = irq_time_cpu(cpu); + if (rq->clock - irq_time > rq->clock_task) + rq->clock_task = rq->clock - irq_time; + } } /* @@ -1910,6 +1920,18 @@ static void deactivate_task(struct rq *rq, struct task_struct *p, int flags) #ifdef CONFIG_IRQ_TIME_ACCOUNTING +/* + * There are no locks covering percpu hardirq/softirq time. + * They are only modified in account_system_vtime, on corresponding CPU + * with interrupts disabled. So, writes are safe. + * They are read and saved off onto struct rq in update_rq_clock(). + * This may result in other CPU reading this CPU's irq time and can + * race with irq/account_system_vtime on this CPU. We would either get old + * or new value (or semi updated value on 32 bit) with a side effect of + * accounting a slice of irq time to wrong task when irq is in progress + * while we read rq->clock. That is a worthy compromise in place of having + * locks on each irq in account_system_time. + */ static DEFINE_PER_CPU(u64, cpu_hardirq_time); static DEFINE_PER_CPU(u64, cpu_softirq_time); @@ -1926,6 +1948,14 @@ void disable_sched_clock_irqtime(void) sched_clock_irqtime = 0; } +static u64 irq_time_cpu(int cpu) +{ + if (!sched_clock_irqtime) + return 0; + + return per_cpu(cpu_softirq_time, cpu) + per_cpu(cpu_hardirq_time, cpu); +} + void account_system_vtime(struct task_struct *curr) { unsigned long flags; @@ -1955,6 +1985,13 @@ void account_system_vtime(struct task_struct *curr) local_irq_restore(flags); } +#else + +static u64 irq_time_cpu(int cpu) +{ + return 0; +} + #endif #include "sched_idletask.c" @@ -3322,7 +3359,7 @@ static u64 do_task_delta_exec(struct task_struct *p, struct rq *rq) if (task_current(rq, p)) { update_rq_clock(rq); - ns = rq->clock - p->se.exec_start; + ns = rq->clock_task - p->se.exec_start; if ((s64)ns < 0) ns = 0; } diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c index f1c615ff39d6..c358d4081b81 100644 --- a/kernel/sched_fair.c +++ b/kernel/sched_fair.c @@ -519,7 +519,7 @@ __update_curr(struct cfs_rq *cfs_rq, struct sched_entity *curr, static void update_curr(struct cfs_rq *cfs_rq) { struct sched_entity *curr = cfs_rq->curr; - u64 now = rq_of(cfs_rq)->clock; + u64 now = rq_of(cfs_rq)->clock_task; unsigned long delta_exec; if (unlikely(!curr)) @@ -602,7 +602,7 @@ update_stats_curr_start(struct cfs_rq *cfs_rq, struct sched_entity *se) /* * We are starting a new run period: */ - se->exec_start = rq_of(cfs_rq)->clock; + se->exec_start = rq_of(cfs_rq)->clock_task; } /************************************************** @@ -1802,7 +1802,7 @@ int can_migrate_task(struct task_struct *p, struct rq *rq, int this_cpu, * 2) too many balance attempts have failed. */ - tsk_cache_hot = task_hot(p, rq->clock, sd); + tsk_cache_hot = task_hot(p, rq->clock_task, sd); if (!tsk_cache_hot || sd->nr_balance_failed > sd->cache_nice_tries) { #ifdef CONFIG_SCHEDSTATS diff --git a/kernel/sched_rt.c b/kernel/sched_rt.c index ab77aa00b7b1..bea7d79f7e9c 100644 --- a/kernel/sched_rt.c +++ b/kernel/sched_rt.c @@ -609,7 +609,7 @@ static void update_curr_rt(struct rq *rq) if (!task_has_rt_policy(curr)) return; - delta_exec = rq->clock - curr->se.exec_start; + delta_exec = rq->clock_task - curr->se.exec_start; if (unlikely((s64)delta_exec < 0)) delta_exec = 0; @@ -618,7 +618,7 @@ static void update_curr_rt(struct rq *rq) curr->se.sum_exec_runtime += delta_exec; account_group_exec_runtime(curr, delta_exec); - curr->se.exec_start = rq->clock; + curr->se.exec_start = rq->clock_task; cpuacct_charge(curr, delta_exec); sched_rt_avg_update(rq, delta_exec); @@ -1075,7 +1075,7 @@ static struct task_struct *_pick_next_task_rt(struct rq *rq) } while (rt_rq); p = rt_task_of(rt_se); - p->se.exec_start = rq->clock; + p->se.exec_start = rq->clock_task; return p; } @@ -1713,7 +1713,7 @@ static void set_curr_task_rt(struct rq *rq) { struct task_struct *p = rq->curr; - p->se.exec_start = rq->clock; + p->se.exec_start = rq->clock_task; /* The running task is never eligible for pushing */ dequeue_pushable_task(rq, p); -- cgit v1.2.3 From aa483808516ca5cacfa0e5849691f64fec25828e Mon Sep 17 00:00:00 2001 From: Venkatesh Pallipadi Date: Mon, 4 Oct 2010 17:03:22 -0700 Subject: sched: Remove irq time from available CPU power The idea was suggested by Peter Zijlstra here: http://marc.info/?l=linux-kernel&m=127476934517534&w=2 irq time is technically not available to the tasks running on the CPU. This patch removes irq time from CPU power piggybacking on sched_rt_avg_update(). Tested this by keeping CPU X busy with a network intensive task having 75% oa a single CPU irq processing (hard+soft) on a 4-way system. And start seven cycle soakers on the system. Without this change, there will be two tasks on each CPU. With this change, there is a single task on irq busy CPU X and remaining 7 tasks are spread around among other 3 CPUs. Signed-off-by: Venkatesh Pallipadi Signed-off-by: Peter Zijlstra LKML-Reference: <1286237003-12406-8-git-send-email-venki@google.com> Signed-off-by: Ingo Molnar --- kernel/sched.c | 18 ++++++++++++++++++ kernel/sched_fair.c | 8 +++++++- kernel/sched_features.h | 5 +++++ 3 files changed, 30 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index 9e01b7100ef6..bff9ef537df0 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -519,6 +519,10 @@ struct rq { u64 avg_idle; #endif +#ifdef CONFIG_IRQ_TIME_ACCOUNTING + u64 prev_irq_time; +#endif + /* calc_load related fields */ unsigned long calc_load_update; long calc_load_active; @@ -643,6 +647,7 @@ static inline struct task_group *task_group(struct task_struct *p) #endif /* CONFIG_CGROUP_SCHED */ static u64 irq_time_cpu(int cpu); +static void sched_irq_time_avg_update(struct rq *rq, u64 irq_time); inline void update_rq_clock(struct rq *rq) { @@ -654,6 +659,8 @@ inline void update_rq_clock(struct rq *rq) irq_time = irq_time_cpu(cpu); if (rq->clock - irq_time > rq->clock_task) rq->clock_task = rq->clock - irq_time; + + sched_irq_time_avg_update(rq, irq_time); } } @@ -1985,6 +1992,15 @@ void account_system_vtime(struct task_struct *curr) local_irq_restore(flags); } +static void sched_irq_time_avg_update(struct rq *rq, u64 curr_irq_time) +{ + if (sched_clock_irqtime && sched_feat(NONIRQ_POWER)) { + u64 delta_irq = curr_irq_time - rq->prev_irq_time; + rq->prev_irq_time = curr_irq_time; + sched_rt_avg_update(rq, delta_irq); + } +} + #else static u64 irq_time_cpu(int cpu) @@ -1992,6 +2008,8 @@ static u64 irq_time_cpu(int cpu) return 0; } +static void sched_irq_time_avg_update(struct rq *rq, u64 curr_irq_time) { } + #endif #include "sched_idletask.c" diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c index c358d4081b81..74cccfae87a8 100644 --- a/kernel/sched_fair.c +++ b/kernel/sched_fair.c @@ -2275,7 +2275,13 @@ unsigned long scale_rt_power(int cpu) u64 total, available; total = sched_avg_period() + (rq->clock - rq->age_stamp); - available = total - rq->rt_avg; + + if (unlikely(total < rq->rt_avg)) { + /* Ensures that power won't end up being negative */ + available = 0; + } else { + available = total - rq->rt_avg; + } if (unlikely((s64)total < SCHED_LOAD_SCALE)) total = SCHED_LOAD_SCALE; diff --git a/kernel/sched_features.h b/kernel/sched_features.h index 83c66e8ad3ee..185f920ec1a2 100644 --- a/kernel/sched_features.h +++ b/kernel/sched_features.h @@ -61,3 +61,8 @@ SCHED_FEAT(ASYM_EFF_LOAD, 1) * release the lock. Decreases scheduling overhead. */ SCHED_FEAT(OWNER_SPIN, 1) + +/* + * Decrement CPU power based on irq activity + */ +SCHED_FEAT(NONIRQ_POWER, 1) -- cgit v1.2.3 From d267f87fb8179c6dba03d08b91952e81bc3723c7 Mon Sep 17 00:00:00 2001 From: Venkatesh Pallipadi Date: Mon, 4 Oct 2010 17:03:23 -0700 Subject: sched: Call tick_check_idle before __irq_enter When CPU is idle and on first interrupt, irq_enter calls tick_check_idle() to notify interruption from idle. But, there is a problem if this call is done after __irq_enter, as all routines in __irq_enter may find stale time due to yet to be done tick_check_idle. Specifically, trace calls in __irq_enter when they use global clock and also account_system_vtime change in this patch as it wants to use sched_clock_cpu() to do proper irq timing. But, tick_check_idle was moved after __irq_enter intentionally to prevent problem of unneeded ksoftirqd wakeups by the commit ee5f80a: irq: call __irq_enter() before calling the tick_idle_check Impact: avoid spurious ksoftirqd wakeups Moving tick_check_idle() before __irq_enter and wrapping it with local_bh_enable/disable would solve both the problems. Fixed-by: Yong Zhang Signed-off-by: Venkatesh Pallipadi Signed-off-by: Peter Zijlstra LKML-Reference: <1286237003-12406-9-git-send-email-venki@google.com> Signed-off-by: Ingo Molnar --- kernel/sched.c | 2 +- kernel/softirq.c | 12 +++++++++--- 2 files changed, 10 insertions(+), 4 deletions(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index bff9ef537df0..567f5cb9808c 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -1974,8 +1974,8 @@ void account_system_vtime(struct task_struct *curr) local_irq_save(flags); - now = sched_clock(); cpu = smp_processor_id(); + now = sched_clock_cpu(cpu); delta = now - per_cpu(irq_start_time, cpu); per_cpu(irq_start_time, cpu) = now; /* diff --git a/kernel/softirq.c b/kernel/softirq.c index 267f7b763ebb..79ee8f1fc0e7 100644 --- a/kernel/softirq.c +++ b/kernel/softirq.c @@ -296,10 +296,16 @@ void irq_enter(void) rcu_irq_enter(); if (idle_cpu(cpu) && !in_interrupt()) { - __irq_enter(); + /* + * Prevent raise_softirq from needlessly waking up ksoftirqd + * here, as softirq will be serviced on return from interrupt. + */ + local_bh_disable(); tick_check_idle(cpu); - } else - __irq_enter(); + _local_bh_enable(); + } + + __irq_enter(); } #ifdef __ARCH_IRQ_EXIT_IRQS_DISABLED -- cgit v1.2.3 From b7dadc38797584f6203386da1947ed5edf516646 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 18 Oct 2010 20:00:37 +0200 Subject: sched: Export account_system_vtime() KVM uses it for example: ERROR: "account_system_vtime" [arch/x86/kvm/kvm.ko] undefined! Cc: Venkatesh Pallipadi Cc: Peter Zijlstra LKML-Reference: <1286237003-12406-3-git-send-email-venki@google.com> Signed-off-by: Ingo Molnar --- kernel/sched.c | 1 + 1 file changed, 1 insertion(+) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index 567f5cb9808c..5998222f901c 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -1991,6 +1991,7 @@ void account_system_vtime(struct task_struct *curr) local_irq_restore(flags); } +EXPORT_SYMBOL_GPL(account_system_vtime); static void sched_irq_time_avg_update(struct rq *rq, u64 curr_irq_time) { -- cgit v1.2.3 From 31ddd871fc3db73e2024cb3eb3ee5051edf5a80f Mon Sep 17 00:00:00 2001 From: Tejun Heo Date: Tue, 19 Oct 2010 11:14:49 +0200 Subject: workqueue: Clarify that schedule_on_each_cpu is synchronous The documentation for schedule_on_each_cpu() states that it calls a function on each online CPU from keventd. This can easily be interpreted as an asyncronous call because the description does not mention that flush_work is called. Clarify that it is synchronous. tj: rephrased a bit Signed-off-by: Mel Gorman Reviewed-by: KOSAKI Motohiro Signed-off-by: Tejun Heo --- kernel/workqueue.c | 10 ++++++---- 1 file changed, 6 insertions(+), 4 deletions(-) (limited to 'kernel') diff --git a/kernel/workqueue.c b/kernel/workqueue.c index 2c6871cbcbee..eb5c1972443a 100644 --- a/kernel/workqueue.c +++ b/kernel/workqueue.c @@ -2676,13 +2676,15 @@ int schedule_delayed_work_on(int cpu, EXPORT_SYMBOL(schedule_delayed_work_on); /** - * schedule_on_each_cpu - call a function on each online CPU from keventd + * schedule_on_each_cpu - execute a function synchronously on each online CPU * @func: the function to call * - * Returns zero on success. - * Returns -ve errno on failure. - * + * schedule_on_each_cpu() executes @func on each online CPU using the + * system workqueue and blocks until all CPUs have completed. * schedule_on_each_cpu() is very slow. + * + * RETURNS: + * 0 on success, -errno on failure. */ int schedule_on_each_cpu(work_func_t func) { -- cgit v1.2.3 From daaae6b010ac0f60c9c35e481589966f9f1fcc22 Mon Sep 17 00:00:00 2001 From: Tejun Heo Date: Tue, 19 Oct 2010 11:28:15 +0200 Subject: workqueue: remove in_workqueue_context() Commit a25909a4 (lockdep: Add an in_workqueue_context() lockdep-based test function) added in_workqueue_context() but there hasn't been any in-kernel user and the lockdep annotation in workqueue is scheduled to change. Remove the unused function. Signed-off-by: Tejun Heo Cc: Paul E. McKenney --- kernel/workqueue.c | 15 --------------- 1 file changed, 15 deletions(-) (limited to 'kernel') diff --git a/kernel/workqueue.c b/kernel/workqueue.c index eb5c1972443a..30acdb74cc23 100644 --- a/kernel/workqueue.c +++ b/kernel/workqueue.c @@ -310,21 +310,6 @@ static inline int __next_wq_cpu(int cpu, const struct cpumask *mask, (cpu) < WORK_CPU_NONE; \ (cpu) = __next_wq_cpu((cpu), cpu_possible_mask, (wq))) -#ifdef CONFIG_LOCKDEP -/** - * in_workqueue_context() - in context of specified workqueue? - * @wq: the workqueue of interest - * - * Checks lockdep state to see if the current task is executing from - * within a workqueue item. This function exists only if lockdep is - * enabled. - */ -int in_workqueue_context(struct workqueue_struct *wq) -{ - return lock_is_held(&wq->lockdep_map); -} -#endif - #ifdef CONFIG_DEBUG_OBJECTS_WORK static struct debug_obj_descr work_debug_descr; -- cgit v1.2.3 From 0fc86c7bd924debd0bddee790ecc884604fdcc63 Mon Sep 17 00:00:00 2001 From: Arnd Bergmann Date: Sat, 11 Sep 2010 20:11:08 +0200 Subject: rtmutex-tester: make it build without BKL The big kernel lock is going away, so make sure that if it is disabled by Kconfig, we do not try to validate it, which would result in compile errors. Signed-off-by: Arnd Bergmann Cc: Thomas Gleixner Cc: Ingo Molnar Cc: Arjan van de Ven Cc: Andrew Morton --- kernel/rtmutex-tester.c | 6 ++++++ 1 file changed, 6 insertions(+) (limited to 'kernel') diff --git a/kernel/rtmutex-tester.c b/kernel/rtmutex-tester.c index a56f629b057a..66cb89bc5ef1 100644 --- a/kernel/rtmutex-tester.c +++ b/kernel/rtmutex-tester.c @@ -76,7 +76,9 @@ static int handle_op(struct test_thread_data *td, int lockwakeup) } if (!lockwakeup && td->bkl == 4) { +#ifdef CONFIG_LOCK_KERNEL unlock_kernel(); +#endif td->bkl = 0; } return 0; @@ -133,14 +135,18 @@ static int handle_op(struct test_thread_data *td, int lockwakeup) if (td->bkl) return 0; td->bkl = 1; +#ifdef CONFIG_LOCK_KERNEL lock_kernel(); +#endif td->bkl = 4; return 0; case RTTEST_UNLOCKBKL: if (td->bkl != 4) break; +#ifdef CONFIG_LOCK_KERNEL unlock_kernel(); +#endif td->bkl = 0; return 0; -- cgit v1.2.3 From 01b284f9b6d51cc3f3bcf3b49f16d2601d3ca22d Mon Sep 17 00:00:00 2001 From: Arnd Bergmann Date: Fri, 17 Sep 2010 20:39:22 +0200 Subject: blktrace: remove the big kernel lock According to Jens, this code does not need the BKL at all, it is sufficiently serialized by bd_mutex. Signed-off-by: Arnd Bergmann Cc: Jens Axboe Cc: Steven Rostedt --- kernel/trace/blktrace.c | 14 +++----------- 1 file changed, 3 insertions(+), 11 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/blktrace.c b/kernel/trace/blktrace.c index 959f8d6c8cc1..5328e8779d4d 100644 --- a/kernel/trace/blktrace.c +++ b/kernel/trace/blktrace.c @@ -23,7 +23,6 @@ #include #include #include -#include #include #include @@ -639,7 +638,6 @@ int blk_trace_ioctl(struct block_device *bdev, unsigned cmd, char __user *arg) if (!q) return -ENXIO; - lock_kernel(); mutex_lock(&bdev->bd_mutex); switch (cmd) { @@ -667,7 +665,6 @@ int blk_trace_ioctl(struct block_device *bdev, unsigned cmd, char __user *arg) } mutex_unlock(&bdev->bd_mutex); - unlock_kernel(); return ret; } @@ -1652,10 +1649,9 @@ static ssize_t sysfs_blk_trace_attr_show(struct device *dev, struct block_device *bdev; ssize_t ret = -ENXIO; - lock_kernel(); bdev = bdget(part_devt(p)); if (bdev == NULL) - goto out_unlock_kernel; + goto out; q = blk_trace_get_queue(bdev); if (q == NULL) @@ -1683,8 +1679,7 @@ out_unlock_bdev: mutex_unlock(&bdev->bd_mutex); out_bdput: bdput(bdev); -out_unlock_kernel: - unlock_kernel(); +out: return ret; } @@ -1714,11 +1709,10 @@ static ssize_t sysfs_blk_trace_attr_store(struct device *dev, ret = -ENXIO; - lock_kernel(); p = dev_to_part(dev); bdev = bdget(part_devt(p)); if (bdev == NULL) - goto out_unlock_kernel; + goto out; q = blk_trace_get_queue(bdev); if (q == NULL) @@ -1753,8 +1747,6 @@ out_unlock_bdev: mutex_unlock(&bdev->bd_mutex); out_bdput: bdput(bdev); -out_unlock_kernel: - unlock_kernel(); out: return ret ? ret : count; } -- cgit v1.2.3 From 7ada876a8703f23befbb20a7465a702ee39b1704 Mon Sep 17 00:00:00 2001 From: Darren Hart Date: Sun, 17 Oct 2010 08:35:04 -0700 Subject: futex: Fix errors in nested key ref-counting MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit futex_wait() is leaking key references due to futex_wait_setup() acquiring an additional reference via the queue_lock() routine. The nested key ref-counting has been masking bugs and complicating code analysis. queue_lock() is only called with a previously ref-counted key, so remove the additional ref-counting from the queue_(un)lock() functions. Also futex_wait_requeue_pi() drops one key reference too many in unqueue_me_pi(). Remove the key reference handling from unqueue_me_pi(). This was paired with a queue_lock() in futex_lock_pi(), so the count remains unchanged. Document remaining nested key ref-counting sites. Signed-off-by: Darren Hart Reported-and-tested-by: Matthieu Fertré Reported-by: Louis Rilling Cc: Peter Zijlstra Cc: Eric Dumazet Cc: John Kacur Cc: Rusty Russell LKML-Reference: <4CBB17A8.70401@linux.intel.com> Signed-off-by: Thomas Gleixner Cc: stable@kernel.org --- kernel/futex.c | 31 ++++++++++++++++--------------- 1 file changed, 16 insertions(+), 15 deletions(-) (limited to 'kernel') diff --git a/kernel/futex.c b/kernel/futex.c index 6a3a5fa1526d..e328f574c97c 100644 --- a/kernel/futex.c +++ b/kernel/futex.c @@ -1363,7 +1363,6 @@ static inline struct futex_hash_bucket *queue_lock(struct futex_q *q) { struct futex_hash_bucket *hb; - get_futex_key_refs(&q->key); hb = hash_futex(&q->key); q->lock_ptr = &hb->lock; @@ -1375,7 +1374,6 @@ static inline void queue_unlock(struct futex_q *q, struct futex_hash_bucket *hb) { spin_unlock(&hb->lock); - drop_futex_key_refs(&q->key); } /** @@ -1480,8 +1478,6 @@ static void unqueue_me_pi(struct futex_q *q) q->pi_state = NULL; spin_unlock(q->lock_ptr); - - drop_futex_key_refs(&q->key); } /* @@ -1812,7 +1808,10 @@ static int futex_wait(u32 __user *uaddr, int fshared, } retry: - /* Prepare to wait on uaddr. */ + /* + * Prepare to wait on uaddr. On success, holds hb lock and increments + * q.key refs. + */ ret = futex_wait_setup(uaddr, val, fshared, &q, &hb); if (ret) goto out; @@ -1822,24 +1821,23 @@ retry: /* If we were woken (and unqueued), we succeeded, whatever. */ ret = 0; + /* unqueue_me() drops q.key ref */ if (!unqueue_me(&q)) - goto out_put_key; + goto out; ret = -ETIMEDOUT; if (to && !to->task) - goto out_put_key; + goto out; /* * We expect signal_pending(current), but we might be the * victim of a spurious wakeup as well. */ - if (!signal_pending(current)) { - put_futex_key(fshared, &q.key); + if (!signal_pending(current)) goto retry; - } ret = -ERESTARTSYS; if (!abs_time) - goto out_put_key; + goto out; restart = ¤t_thread_info()->restart_block; restart->fn = futex_wait_restart; @@ -1856,8 +1854,6 @@ retry: ret = -ERESTART_RESTARTBLOCK; -out_put_key: - put_futex_key(fshared, &q.key); out: if (to) { hrtimer_cancel(&to->timer); @@ -2236,7 +2232,10 @@ static int futex_wait_requeue_pi(u32 __user *uaddr, int fshared, q.rt_waiter = &rt_waiter; q.requeue_pi_key = &key2; - /* Prepare to wait on uaddr. */ + /* + * Prepare to wait on uaddr. On success, increments q.key (key1) ref + * count. + */ ret = futex_wait_setup(uaddr, val, fshared, &q, &hb); if (ret) goto out_key2; @@ -2254,7 +2253,9 @@ static int futex_wait_requeue_pi(u32 __user *uaddr, int fshared, * In order for us to be here, we know our q.key == key2, and since * we took the hb->lock above, we also know that futex_requeue() has * completed and we no longer have to concern ourselves with a wakeup - * race with the atomic proxy lock acquition by the requeue code. + * race with the atomic proxy lock acquisition by the requeue code. The + * futex_requeue dropped our key1 reference and incremented our key2 + * reference count. */ /* Check if the requeue code acquired the second futex for us. */ -- cgit v1.2.3 From 7e40798f406fe73f9bac496a390daabd8768a8f7 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Tue, 19 Oct 2010 10:56:19 -0400 Subject: tracing: Fix compile issue for trace_sched_wakeup.c The function start_func_tracer() was incorrectly added in the #ifdef CONFIG_FUNCTION_TRACER condition, but is still used even when function tracing is not enabled. The calls to register_ftrace_function() and register_ftrace_graph() become nops (and their arguments are even ignored), thus there is no reason to hide start_func_tracer() when function tracing is not enabled. Reported-by: Ingo Molnar Signed-off-by: Steven Rostedt --- kernel/trace/trace_sched_wakeup.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace_sched_wakeup.c b/kernel/trace/trace_sched_wakeup.c index 31689d2df7f3..7319559ed59f 100644 --- a/kernel/trace/trace_sched_wakeup.c +++ b/kernel/trace/trace_sched_wakeup.c @@ -130,6 +130,7 @@ static struct ftrace_ops trace_ops __read_mostly = { .func = wakeup_tracer_call, }; +#endif /* CONFIG_FUNCTION_TRACER */ static int start_func_tracer(int graph) { @@ -159,8 +160,6 @@ static void stop_func_tracer(int graph) unregister_ftrace_graph(); } -#endif /* CONFIG_FUNCTION_TRACER */ - #ifdef CONFIG_FUNCTION_GRAPH_TRACER static int wakeup_set_flag(u32 old_flags, u32 bit, int set) { -- cgit v1.2.3 From 747e94ae3d1b4c9bf5380e569f614eb9040b79e7 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Fri, 8 Oct 2010 13:51:48 -0400 Subject: ring-buffer: Make write slow path out of line Gcc inlines the slow path of the ring buffer write which can hurt performance. This patch simply forces the slow path function rb_move_tail() to always be a function. The ring_buffer_benchmark module with reader_disabled=1 shows that this patch changes the time to record an event from 135 ns to 132 ns. (3 ns or 2.22% improvement) Signed-off-by: Steven Rostedt --- kernel/trace/ring_buffer.c | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c index bca96377fd4e..0b88df849a59 100644 --- a/kernel/trace/ring_buffer.c +++ b/kernel/trace/ring_buffer.c @@ -1823,7 +1823,10 @@ rb_reset_tail(struct ring_buffer_per_cpu *cpu_buffer, local_sub(length, &tail_page->write); } -static struct ring_buffer_event * +/* + * This is the slow path, force gcc not to inline it. + */ +static noinline struct ring_buffer_event * rb_move_tail(struct ring_buffer_per_cpu *cpu_buffer, unsigned long length, unsigned long tail, struct buffer_page *tail_page, u64 *ts) @@ -1943,7 +1946,7 @@ __rb_reserve_next(struct ring_buffer_per_cpu *cpu_buffer, tail = write - length; /* See if we shot pass the end of this buffer page */ - if (write > BUF_PAGE_SIZE) + if (unlikely(write > BUF_PAGE_SIZE)) return rb_move_tail(cpu_buffer, length, tail, tail_page, ts); -- cgit v1.2.3 From e8bc43e84fada397af1b677b07dbf26e6ac78fcc Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Wed, 20 Oct 2010 10:58:02 -0400 Subject: ring-buffer: Pass timestamp by value and not by reference The original code for the ring buffer had locations that modified the timestamp and that change was used by the callers. Now, the timestamp is not reused by the callers and there is no reason to pass it by reference. By changing the call to pass by value, lets gcc optimize the code a bit more where it can store the timestamp in a register and not worry about updating the reference. Signed-off-by: Steven Rostedt --- kernel/trace/ring_buffer.c | 20 ++++++++++---------- 1 file changed, 10 insertions(+), 10 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c index 0b88df849a59..c8ce6bde7fa4 100644 --- a/kernel/trace/ring_buffer.c +++ b/kernel/trace/ring_buffer.c @@ -1829,7 +1829,7 @@ rb_reset_tail(struct ring_buffer_per_cpu *cpu_buffer, static noinline struct ring_buffer_event * rb_move_tail(struct ring_buffer_per_cpu *cpu_buffer, unsigned long length, unsigned long tail, - struct buffer_page *tail_page, u64 *ts) + struct buffer_page *tail_page, u64 ts) { struct buffer_page *commit_page = cpu_buffer->commit_page; struct ring_buffer *buffer = cpu_buffer->buffer; @@ -1912,8 +1912,8 @@ rb_move_tail(struct ring_buffer_per_cpu *cpu_buffer, * Nested commits always have zero deltas, so * just reread the time stamp */ - *ts = rb_time_stamp(buffer); - next_page->page->time_stamp = *ts; + ts = rb_time_stamp(buffer); + next_page->page->time_stamp = ts; } out_again: @@ -1932,7 +1932,7 @@ rb_move_tail(struct ring_buffer_per_cpu *cpu_buffer, static struct ring_buffer_event * __rb_reserve_next(struct ring_buffer_per_cpu *cpu_buffer, - unsigned type, unsigned long length, u64 *ts) + unsigned type, unsigned long length, u64 ts) { struct buffer_page *tail_page; struct ring_buffer_event *event; @@ -1965,7 +1965,7 @@ __rb_reserve_next(struct ring_buffer_per_cpu *cpu_buffer, * its timestamp. */ if (!tail) - tail_page->page->time_stamp = *ts; + tail_page->page->time_stamp = ts; return event; } @@ -2008,7 +2008,7 @@ rb_try_to_discard(struct ring_buffer_per_cpu *cpu_buffer, static int rb_add_time_stamp(struct ring_buffer_per_cpu *cpu_buffer, - u64 *ts, u64 *delta) + u64 ts, u64 *delta) { struct ring_buffer_event *event; int ret; @@ -2016,7 +2016,7 @@ rb_add_time_stamp(struct ring_buffer_per_cpu *cpu_buffer, WARN_ONCE(*delta > (1ULL << 59), KERN_WARNING "Delta way too big! %llu ts=%llu write stamp = %llu\n", (unsigned long long)*delta, - (unsigned long long)*ts, + (unsigned long long)ts, (unsigned long long)cpu_buffer->write_stamp); /* @@ -2051,7 +2051,7 @@ rb_add_time_stamp(struct ring_buffer_per_cpu *cpu_buffer, event->array[0] = 0; } } - cpu_buffer->write_stamp = *ts; + cpu_buffer->write_stamp = ts; /* let the caller know this was the commit */ ret = 1; } else { @@ -2175,7 +2175,7 @@ rb_reserve_next_event(struct ring_buffer *buffer, delta = diff; if (unlikely(test_time_stamp(delta))) { - commit = rb_add_time_stamp(cpu_buffer, &ts, &delta); + commit = rb_add_time_stamp(cpu_buffer, ts, &delta); if (commit == -EBUSY) goto out_fail; @@ -2187,7 +2187,7 @@ rb_reserve_next_event(struct ring_buffer *buffer, } get_event: - event = __rb_reserve_next(cpu_buffer, 0, length, &ts); + event = __rb_reserve_next(cpu_buffer, 0, length, ts); if (unlikely(PTR_ERR(event) == -EAGAIN)) goto again; -- cgit v1.2.3 From f25106aeab7408394b9dd707e5ecf557e269c723 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Wed, 20 Oct 2010 12:40:12 -0400 Subject: ring-buffer: Pass delta by value and not by reference The delta between events is passed to the timestamp code by reference and the timestamp code will reset the value. But it can be reset from the caller. No need to pass it in by reference. By changing the call to pass by value, lets gcc optimize the code a bit more where it can store the delta in a register and not worry about updating the reference. Signed-off-by: Steven Rostedt --- kernel/trace/ring_buffer.c | 16 ++++++++-------- 1 file changed, 8 insertions(+), 8 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c index c8ce6bde7fa4..3af77cd47f21 100644 --- a/kernel/trace/ring_buffer.c +++ b/kernel/trace/ring_buffer.c @@ -2008,14 +2008,14 @@ rb_try_to_discard(struct ring_buffer_per_cpu *cpu_buffer, static int rb_add_time_stamp(struct ring_buffer_per_cpu *cpu_buffer, - u64 ts, u64 *delta) + u64 ts, u64 delta) { struct ring_buffer_event *event; int ret; - WARN_ONCE(*delta > (1ULL << 59), + WARN_ONCE(delta > (1ULL << 59), KERN_WARNING "Delta way too big! %llu ts=%llu write stamp = %llu\n", - (unsigned long long)*delta, + (unsigned long long)delta, (unsigned long long)ts, (unsigned long long)cpu_buffer->write_stamp); @@ -2041,8 +2041,8 @@ rb_add_time_stamp(struct ring_buffer_per_cpu *cpu_buffer, * and if we can't just make it zero. */ if (rb_event_index(event)) { - event->time_delta = *delta & TS_MASK; - event->array[0] = *delta >> TS_SHIFT; + event->time_delta = delta & TS_MASK; + event->array[0] = delta >> TS_SHIFT; } else { /* try to discard, since we do not need this */ if (!rb_try_to_discard(cpu_buffer, event)) { @@ -2064,8 +2064,6 @@ rb_add_time_stamp(struct ring_buffer_per_cpu *cpu_buffer, ret = 0; } - *delta = 0; - return ret; } @@ -2175,7 +2173,9 @@ rb_reserve_next_event(struct ring_buffer *buffer, delta = diff; if (unlikely(test_time_stamp(delta))) { - commit = rb_add_time_stamp(cpu_buffer, ts, &delta); + commit = rb_add_time_stamp(cpu_buffer, ts, delta); + delta = 0; + if (commit == -EBUSY) goto out_fail; -- cgit v1.2.3 From 69d1b839f7eee347e357b3f6cce7f630cc6ff93d Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Thu, 7 Oct 2010 18:18:05 -0400 Subject: ring-buffer: Bind time extend and data events together When the time between two timestamps is greater than 2^27 nanosecs (~134 ms) a time extend event is added that extends the time difference to 59 bits (~18 years). This is due to events only having a 27 bit field to store time. Currently this time extend is a separate event. We add it just before the event data that is being written to the buffer. But before the event data is committed, the event data can also be discarded (as with the case of filters). But because the time extend has already been committed, it will stay in the buffer. If lots of events are being filtered and no event is being written, then every 134ms a time extend can be added to the buffer without any data attached. To keep from filling the entire buffer with time extends, a time extend will never be the first event in a page because the page timestamp can be used. Time extends can only fill the rest of a page with some data at the beginning. This patch binds the time extend with the data. The difference here is that the time extend is not committed before the data is added. Instead, when a time extend is needed, the space reserved on the ring buffer is the time extend + the data event size. The time extend is added to the first part of the reserved block and the data is added to the second. The time extend event is passed back to the reserver, but since the reserver also uses a function to find the data portion of the reserved block, no changes to the ring buffer interface need to be made. When a commit is discarded, we now remove both the time extend and the event. With this approach no more than one time extend can be in the buffer in a row. Data must always follow a time extend. Thanks to Mathieu Desnoyers for suggesting this idea. Suggested-by: Mathieu Desnoyers Cc: Thomas Gleixner Signed-off-by: Steven Rostedt --- kernel/trace/ring_buffer.c | 266 ++++++++++++++++++++++++--------------------- 1 file changed, 142 insertions(+), 124 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c index 3af77cd47f21..f50f43107e93 100644 --- a/kernel/trace/ring_buffer.c +++ b/kernel/trace/ring_buffer.c @@ -224,6 +224,9 @@ enum { RB_LEN_TIME_STAMP = 16, }; +#define skip_time_extend(event) \ + ((struct ring_buffer_event *)((char *)event + RB_LEN_TIME_EXTEND)) + static inline int rb_null_event(struct ring_buffer_event *event) { return event->type_len == RINGBUF_TYPE_PADDING && !event->time_delta; @@ -248,8 +251,12 @@ rb_event_data_length(struct ring_buffer_event *event) return length + RB_EVNT_HDR_SIZE; } -/* inline for ring buffer fast paths */ -static unsigned +/* + * Return the length of the given event. Will return + * the length of the time extend if the event is a + * time extend. + */ +static inline unsigned rb_event_length(struct ring_buffer_event *event) { switch (event->type_len) { @@ -274,13 +281,41 @@ rb_event_length(struct ring_buffer_event *event) return 0; } +/* + * Return total length of time extend and data, + * or just the event length for all other events. + */ +static inline unsigned +rb_event_ts_length(struct ring_buffer_event *event) +{ + unsigned len = 0; + + if (event->type_len == RINGBUF_TYPE_TIME_EXTEND) { + /* time extends include the data event after it */ + len = RB_LEN_TIME_EXTEND; + event = skip_time_extend(event); + } + return len + rb_event_length(event); +} + /** * ring_buffer_event_length - return the length of the event * @event: the event to get the length of + * + * Returns the size of the data load of a data event. + * If the event is something other than a data event, it + * returns the size of the event itself. With the exception + * of a TIME EXTEND, where it still returns the size of the + * data load of the data event after it. */ unsigned ring_buffer_event_length(struct ring_buffer_event *event) { - unsigned length = rb_event_length(event); + unsigned length; + + if (event->type_len == RINGBUF_TYPE_TIME_EXTEND) + event = skip_time_extend(event); + + length = rb_event_length(event); if (event->type_len > RINGBUF_TYPE_DATA_TYPE_LEN_MAX) return length; length -= RB_EVNT_HDR_SIZE; @@ -294,6 +329,8 @@ EXPORT_SYMBOL_GPL(ring_buffer_event_length); static void * rb_event_data(struct ring_buffer_event *event) { + if (event->type_len == RINGBUF_TYPE_TIME_EXTEND) + event = skip_time_extend(event); BUG_ON(event->type_len > RINGBUF_TYPE_DATA_TYPE_LEN_MAX); /* If length is in len field, then array[0] has the data */ if (event->type_len) @@ -1546,6 +1583,25 @@ static void rb_inc_iter(struct ring_buffer_iter *iter) iter->head = 0; } +/* Slow path, do not inline */ +static noinline struct ring_buffer_event * +rb_add_time_stamp(struct ring_buffer_event *event, u64 delta) +{ + event->type_len = RINGBUF_TYPE_TIME_EXTEND; + + /* Not the first event on the page? */ + if (rb_event_index(event)) { + event->time_delta = delta & TS_MASK; + event->array[0] = delta >> TS_SHIFT; + } else { + /* nope, just zero it */ + event->time_delta = 0; + event->array[0] = 0; + } + + return skip_time_extend(event); +} + /** * ring_buffer_update_event - update event type and data * @event: the even to update @@ -1558,28 +1614,31 @@ static void rb_inc_iter(struct ring_buffer_iter *iter) * data field. */ static void -rb_update_event(struct ring_buffer_event *event, - unsigned type, unsigned length) +rb_update_event(struct ring_buffer_per_cpu *cpu_buffer, + struct ring_buffer_event *event, unsigned length, + int add_timestamp, u64 delta) { - event->type_len = type; - - switch (type) { - - case RINGBUF_TYPE_PADDING: - case RINGBUF_TYPE_TIME_EXTEND: - case RINGBUF_TYPE_TIME_STAMP: - break; + /* Only a commit updates the timestamp */ + if (unlikely(!rb_event_is_commit(cpu_buffer, event))) + delta = 0; - case 0: - length -= RB_EVNT_HDR_SIZE; - if (length > RB_MAX_SMALL_DATA || RB_FORCE_8BYTE_ALIGNMENT) - event->array[0] = length; - else - event->type_len = DIV_ROUND_UP(length, RB_ALIGNMENT); - break; - default: - BUG(); + /* + * If we need to add a timestamp, then we + * add it to the start of the resevered space. + */ + if (unlikely(add_timestamp)) { + event = rb_add_time_stamp(event, delta); + length -= RB_LEN_TIME_EXTEND; + delta = 0; } + + event->time_delta = delta; + length -= RB_EVNT_HDR_SIZE; + if (length > RB_MAX_SMALL_DATA || RB_FORCE_8BYTE_ALIGNMENT) { + event->type_len = 0; + event->array[0] = length; + } else + event->type_len = DIV_ROUND_UP(length, RB_ALIGNMENT); } /* @@ -1932,12 +1991,21 @@ rb_move_tail(struct ring_buffer_per_cpu *cpu_buffer, static struct ring_buffer_event * __rb_reserve_next(struct ring_buffer_per_cpu *cpu_buffer, - unsigned type, unsigned long length, u64 ts) + unsigned long length, u64 ts, + u64 delta, int add_timestamp) { struct buffer_page *tail_page; struct ring_buffer_event *event; unsigned long tail, write; + /* + * If the time delta since the last event is too big to + * hold in the time field of the event, then we append a + * TIME EXTEND event ahead of the data event. + */ + if (unlikely(add_timestamp)) + length += RB_LEN_TIME_EXTEND; + tail_page = cpu_buffer->tail_page; write = local_add_return(length, &tail_page->write); @@ -1954,11 +2022,9 @@ __rb_reserve_next(struct ring_buffer_per_cpu *cpu_buffer, event = __rb_page_index(tail_page, tail); kmemcheck_annotate_bitfield(event, bitfield); - rb_update_event(event, type, length); + rb_update_event(cpu_buffer, event, length, add_timestamp, delta); - /* The passed in type is zero for DATA */ - if (likely(!type)) - local_inc(&tail_page->entries); + local_inc(&tail_page->entries); /* * If this is the first commit on the page, then update @@ -1980,7 +2046,7 @@ rb_try_to_discard(struct ring_buffer_per_cpu *cpu_buffer, unsigned long addr; new_index = rb_event_index(event); - old_index = new_index + rb_event_length(event); + old_index = new_index + rb_event_ts_length(event); addr = (unsigned long)event; addr &= PAGE_MASK; @@ -2006,67 +2072,6 @@ rb_try_to_discard(struct ring_buffer_per_cpu *cpu_buffer, return 0; } -static int -rb_add_time_stamp(struct ring_buffer_per_cpu *cpu_buffer, - u64 ts, u64 delta) -{ - struct ring_buffer_event *event; - int ret; - - WARN_ONCE(delta > (1ULL << 59), - KERN_WARNING "Delta way too big! %llu ts=%llu write stamp = %llu\n", - (unsigned long long)delta, - (unsigned long long)ts, - (unsigned long long)cpu_buffer->write_stamp); - - /* - * The delta is too big, we to add a - * new timestamp. - */ - event = __rb_reserve_next(cpu_buffer, - RINGBUF_TYPE_TIME_EXTEND, - RB_LEN_TIME_EXTEND, - ts); - if (!event) - return -EBUSY; - - if (PTR_ERR(event) == -EAGAIN) - return -EAGAIN; - - /* Only a commited time event can update the write stamp */ - if (rb_event_is_commit(cpu_buffer, event)) { - /* - * If this is the first on the page, then it was - * updated with the page itself. Try to discard it - * and if we can't just make it zero. - */ - if (rb_event_index(event)) { - event->time_delta = delta & TS_MASK; - event->array[0] = delta >> TS_SHIFT; - } else { - /* try to discard, since we do not need this */ - if (!rb_try_to_discard(cpu_buffer, event)) { - /* nope, just zero it */ - event->time_delta = 0; - event->array[0] = 0; - } - } - cpu_buffer->write_stamp = ts; - /* let the caller know this was the commit */ - ret = 1; - } else { - /* Try to discard the event */ - if (!rb_try_to_discard(cpu_buffer, event)) { - /* Darn, this is just wasted space */ - event->time_delta = 0; - event->array[0] = 0; - } - ret = 0; - } - - return ret; -} - static void rb_start_commit(struct ring_buffer_per_cpu *cpu_buffer) { local_inc(&cpu_buffer->committing); @@ -2111,9 +2116,9 @@ rb_reserve_next_event(struct ring_buffer *buffer, unsigned long length) { struct ring_buffer_event *event; - u64 ts, delta = 0; - int commit = 0; + u64 ts, delta; int nr_loops = 0; + int add_timestamp; rb_start_commit(cpu_buffer); @@ -2134,6 +2139,9 @@ rb_reserve_next_event(struct ring_buffer *buffer, length = rb_calculate_event_length(length); again: + add_timestamp = 0; + delta = 0; + /* * We allow for interrupts to reenter here and do a trace. * If one does, it will cause this original code to loop @@ -2172,33 +2180,24 @@ rb_reserve_next_event(struct ring_buffer *buffer, delta = diff; if (unlikely(test_time_stamp(delta))) { - - commit = rb_add_time_stamp(cpu_buffer, ts, delta); - delta = 0; - - if (commit == -EBUSY) - goto out_fail; - - if (commit == -EAGAIN) - goto again; - - RB_WARN_ON(cpu_buffer, commit < 0); + WARN_ONCE(delta > (1ULL << 59), + KERN_WARNING "Delta way too big! %llu ts=%llu write stamp = %llu\n", + (unsigned long long)delta, + (unsigned long long)ts, + (unsigned long long)cpu_buffer->write_stamp); + add_timestamp = 1; } } get_event: - event = __rb_reserve_next(cpu_buffer, 0, length, ts); + event = __rb_reserve_next(cpu_buffer, length, ts, + delta, add_timestamp); if (unlikely(PTR_ERR(event) == -EAGAIN)) goto again; if (!event) goto out_fail; - if (!rb_event_is_commit(cpu_buffer, event)) - delta = 0; - - event->time_delta = delta; - return event; out_fail: @@ -2311,12 +2310,28 @@ static void rb_update_write_stamp(struct ring_buffer_per_cpu *cpu_buffer, struct ring_buffer_event *event) { + u64 delta; + /* * The event first in the commit queue updates the * time stamp. */ - if (rb_event_is_commit(cpu_buffer, event)) - cpu_buffer->write_stamp += event->time_delta; + if (rb_event_is_commit(cpu_buffer, event)) { + /* + * A commit event that is first on a page + * updates the write timestamp with the page stamp + */ + if (!rb_event_index(event)) + cpu_buffer->write_stamp = + cpu_buffer->commit_page->page->time_stamp; + else if (event->type_len == RINGBUF_TYPE_TIME_EXTEND) { + delta = event->array[0]; + delta <<= TS_SHIFT; + delta += event->time_delta; + cpu_buffer->write_stamp += delta; + } else + cpu_buffer->write_stamp += event->time_delta; + } } static void rb_commit(struct ring_buffer_per_cpu *cpu_buffer, @@ -2356,6 +2371,9 @@ EXPORT_SYMBOL_GPL(ring_buffer_unlock_commit); static inline void rb_event_discard(struct ring_buffer_event *event) { + if (event->type_len == RINGBUF_TYPE_TIME_EXTEND) + event = skip_time_extend(event); + /* array[0] holds the actual length for the discarded event */ event->array[0] = rb_event_data_length(event) - RB_EVNT_HDR_SIZE; event->type_len = RINGBUF_TYPE_PADDING; @@ -3043,12 +3061,12 @@ rb_buffer_peek(struct ring_buffer_per_cpu *cpu_buffer, u64 *ts, again: /* - * We repeat when a timestamp is encountered. It is possible - * to get multiple timestamps from an interrupt entering just - * as one timestamp is about to be written, or from discarded - * commits. The most that we can have is the number on a single page. + * We repeat when a time extend is encountered. + * Since the time extend is always attached to a data event, + * we should never loop more than once. + * (We never hit the following condition more than twice). */ - if (RB_WARN_ON(cpu_buffer, ++nr_loops > RB_TIMESTAMPS_PER_PAGE)) + if (RB_WARN_ON(cpu_buffer, ++nr_loops > 2)) return NULL; reader = rb_get_reader_page(cpu_buffer); @@ -3124,14 +3142,12 @@ rb_iter_peek(struct ring_buffer_iter *iter, u64 *ts) return NULL; /* - * We repeat when a timestamp is encountered. - * We can get multiple timestamps by nested interrupts or also - * if filtering is on (discarding commits). Since discarding - * commits can be frequent we can get a lot of timestamps. - * But we limit them by not adding timestamps if they begin - * at the start of a page. + * We repeat when a time extend is encountered. + * Since the time extend is always attached to a data event, + * we should never loop more than once. + * (We never hit the following condition more than twice). */ - if (RB_WARN_ON(cpu_buffer, ++nr_loops > RB_TIMESTAMPS_PER_PAGE)) + if (RB_WARN_ON(cpu_buffer, ++nr_loops > 2)) return NULL; if (rb_per_cpu_empty(cpu_buffer)) @@ -3829,7 +3845,8 @@ int ring_buffer_read_page(struct ring_buffer *buffer, if (len > (commit - read)) len = (commit - read); - size = rb_event_length(event); + /* Always keep the time extend and data together */ + size = rb_event_ts_length(event); if (len < size) goto out_unlock; @@ -3851,7 +3868,8 @@ int ring_buffer_read_page(struct ring_buffer *buffer, break; event = rb_reader_event(cpu_buffer); - size = rb_event_length(event); + /* Always keep the time extend and data together */ + size = rb_event_ts_length(event); } while (len > size); /* update bpage */ -- cgit v1.2.3 From 140ff89127c74b1b1c1b0152a36ea3720ccf6bc3 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Fri, 8 Oct 2010 10:50:30 -0400 Subject: ring-buffer: Remove condition to add timestamp in fast path There's a condition to check if we should add a time extend or not in the fast path. But this condition is racey (in the sense that we can add a unnecessary time extend, but nothing that can break anything). We later check if the time or event time delta should be zero or have real data in it (not racey), making this first check redundant. This check may help save space once in a while, but really is not worth the hassle to try to save some space that happens at most 134 ms at a time. Signed-off-by: Steven Rostedt --- kernel/trace/ring_buffer.c | 28 ++++++---------------------- 1 file changed, 6 insertions(+), 22 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c index f50f43107e93..d9f3e7a82137 100644 --- a/kernel/trace/ring_buffer.c +++ b/kernel/trace/ring_buffer.c @@ -2119,6 +2119,7 @@ rb_reserve_next_event(struct ring_buffer *buffer, u64 ts, delta; int nr_loops = 0; int add_timestamp; + u64 diff; rb_start_commit(cpu_buffer); @@ -2155,29 +2156,13 @@ rb_reserve_next_event(struct ring_buffer *buffer, goto out_fail; ts = rb_time_stamp(cpu_buffer->buffer); + diff = ts - cpu_buffer->write_stamp; - /* - * Only the first commit can update the timestamp. - * Yes there is a race here. If an interrupt comes in - * just after the conditional and it traces too, then it - * will also check the deltas. More than one timestamp may - * also be made. But only the entry that did the actual - * commit will be something other than zero. - */ - if (likely(cpu_buffer->tail_page == cpu_buffer->commit_page && - rb_page_write(cpu_buffer->tail_page) == - rb_commit_index(cpu_buffer))) { - u64 diff; - - diff = ts - cpu_buffer->write_stamp; - - /* make sure this diff is calculated here */ - barrier(); - - /* Did the write stamp get updated already? */ - if (unlikely(ts < cpu_buffer->write_stamp)) - goto get_event; + /* make sure this diff is calculated here */ + barrier(); + /* Did the write stamp get updated already? */ + if (likely(ts >= cpu_buffer->write_stamp)) { delta = diff; if (unlikely(test_time_stamp(delta))) { WARN_ONCE(delta > (1ULL << 59), @@ -2189,7 +2174,6 @@ rb_reserve_next_event(struct ring_buffer *buffer, } } - get_event: event = __rb_reserve_next(cpu_buffer, length, ts, delta, add_timestamp); if (unlikely(PTR_ERR(event) == -EAGAIN)) -- cgit v1.2.3 From d9abde2138e0a00a0d7e44676928efa0ef629d48 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Tue, 19 Oct 2010 13:17:08 -0400 Subject: ring-buffer: Micro-optimize with some strategic inlining By using inline and noinline, we are able to make the fast path of recording an event 4% faster. Signed-off-by: Steven Rostedt --- kernel/trace/ring_buffer.c | 23 +++++++++++++++-------- 1 file changed, 15 insertions(+), 8 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c index d9f3e7a82137..f5007d0d932d 100644 --- a/kernel/trace/ring_buffer.c +++ b/kernel/trace/ring_buffer.c @@ -2078,7 +2078,7 @@ static void rb_start_commit(struct ring_buffer_per_cpu *cpu_buffer) local_inc(&cpu_buffer->commits); } -static void rb_end_commit(struct ring_buffer_per_cpu *cpu_buffer) +static inline void rb_end_commit(struct ring_buffer_per_cpu *cpu_buffer) { unsigned long commits; @@ -2193,13 +2193,9 @@ rb_reserve_next_event(struct ring_buffer *buffer, #define TRACE_RECURSIVE_DEPTH 16 -static int trace_recursive_lock(void) +/* Keep this code out of the fast path cache */ +static noinline void trace_recursive_fail(void) { - current->trace_recursion++; - - if (likely(current->trace_recursion < TRACE_RECURSIVE_DEPTH)) - return 0; - /* Disable all tracing before we do anything else */ tracing_off_permanent(); @@ -2211,10 +2207,21 @@ static int trace_recursive_lock(void) in_nmi()); WARN_ON_ONCE(1); +} + +static inline int trace_recursive_lock(void) +{ + current->trace_recursion++; + + if (likely(current->trace_recursion < TRACE_RECURSIVE_DEPTH)) + return 0; + + trace_recursive_fail(); + return -1; } -static void trace_recursive_unlock(void) +static inline void trace_recursive_unlock(void) { WARN_ON_ONCE(!current->trace_recursion); -- cgit v1.2.3 From b8b2663bd7c9da04ac804659b9f617c199d0252c Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Tue, 19 Oct 2010 13:23:25 -0400 Subject: ring-buffer: Remove unused macro RB_TIMESTAMPS_PER_PAGE With the binding of time extends to events we no longer need to use the macro RB_TIMESTAMPS_PER_PAGE. Remove it. Signed-off-by: Steven Rostedt --- kernel/trace/ring_buffer.c | 3 --- 1 file changed, 3 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c index f5007d0d932d..ad25490f8b40 100644 --- a/kernel/trace/ring_buffer.c +++ b/kernel/trace/ring_buffer.c @@ -441,9 +441,6 @@ static inline int test_time_stamp(u64 delta) /* Max payload is BUF_PAGE_SIZE - header (8bytes) */ #define BUF_MAX_DATA_SIZE (BUF_PAGE_SIZE - (sizeof(u32) * 2)) -/* Max number of timestamps that can fit on a page */ -#define RB_TIMESTAMPS_PER_PAGE (BUF_PAGE_SIZE / RB_LEN_TIME_EXTEND) - int ring_buffer_print_page_header(struct trace_seq *s) { struct buffer_data_page field; -- cgit v1.2.3 From b0ae19811375031ae3b3fecc65b702a9c6e5cc28 Mon Sep 17 00:00:00 2001 From: KOSAKI Motohiro Date: Fri, 15 Oct 2010 04:21:18 +0900 Subject: security: remove unused parameter from security_task_setscheduler() All security modules shouldn't change sched_param parameter of security_task_setscheduler(). This is not only meaningless, but also make a harmful result if caller pass a static variable. This patch remove policy and sched_param parameter from security_task_setscheduler() becuase none of security module is using it. Cc: James Morris Signed-off-by: KOSAKI Motohiro Signed-off-by: James Morris --- kernel/cpuset.c | 4 ++-- kernel/sched.c | 4 ++-- 2 files changed, 4 insertions(+), 4 deletions(-) (limited to 'kernel') diff --git a/kernel/cpuset.c b/kernel/cpuset.c index b23c0979bbe7..51b143e2a07a 100644 --- a/kernel/cpuset.c +++ b/kernel/cpuset.c @@ -1397,7 +1397,7 @@ static int cpuset_can_attach(struct cgroup_subsys *ss, struct cgroup *cont, if (tsk->flags & PF_THREAD_BOUND) return -EINVAL; - ret = security_task_setscheduler(tsk, 0, NULL); + ret = security_task_setscheduler(tsk); if (ret) return ret; if (threadgroup) { @@ -1405,7 +1405,7 @@ static int cpuset_can_attach(struct cgroup_subsys *ss, struct cgroup *cont, rcu_read_lock(); list_for_each_entry_rcu(c, &tsk->thread_group, thread_group) { - ret = security_task_setscheduler(c, 0, NULL); + ret = security_task_setscheduler(c); if (ret) { rcu_read_unlock(); return ret; diff --git a/kernel/sched.c b/kernel/sched.c index dc85ceb90832..df6579d9b4df 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -4645,7 +4645,7 @@ recheck: } if (user) { - retval = security_task_setscheduler(p, policy, param); + retval = security_task_setscheduler(p); if (retval) return retval; } @@ -4887,7 +4887,7 @@ long sched_setaffinity(pid_t pid, const struct cpumask *in_mask) if (!check_same_owner(p) && !capable(CAP_SYS_NICE)) goto out_unlock; - retval = security_task_setscheduler(p, 0, NULL); + retval = security_task_setscheduler(p); if (retval) goto out_unlock; -- cgit v1.2.3 From dd49a38cf30944be27892c10b1c0e5b3fa73bcb2 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Wed, 20 Oct 2010 21:51:26 -0400 Subject: tracing: Do not limit the size of the number of CPU buffers The tracing per_cpu buffers were limited to 999 CPUs for a mear savings in stack space of a char array. Up the array to 30 characters which is more than enough to hold a 64 bit number. Reported-by: Robin Holt Suggested-by: Ingo Molnar Signed-off-by: Steven Rostedt --- kernel/trace/trace.c | 8 ++------ 1 file changed, 2 insertions(+), 6 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 001bcd2ccf4a..82d9b8106cd0 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -3996,13 +3996,9 @@ static void tracing_init_debugfs_percpu(long cpu) { struct dentry *d_percpu = tracing_dentry_percpu(); struct dentry *d_cpu; - /* strlen(cpu) + MAX(log10(cpu)) + '\0' */ - char cpu_dir[7]; + char cpu_dir[30]; /* 30 characters should be more than enough */ - if (cpu > 999 || cpu < 0) - return; - - sprintf(cpu_dir, "cpu%ld", cpu); + snprintf(cpu_dir, 30, "cpu%ld", cpu); d_cpu = debugfs_create_dir(cpu_dir, d_percpu); if (!d_cpu) { pr_warning("Could not create debugfs '%s' entry\n", cpu_dir); -- cgit v1.2.3 From f4bc6bb2d562703eafc895c37e7be20906de139d Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 19 Oct 2010 15:00:13 +0200 Subject: tracing: Cleanup the convoluted softirq tracepoints With the addition of trace_softirq_raise() the softirq tracepoint got even more convoluted. Why the tracepoints take two pointers to assign an integer is beyond my comprehension. But adding an extra case which treats the first pointer as an unsigned long when the second pointer is NULL including the back and forth type casting is just horrible. Convert the softirq tracepoints to take a single unsigned int argument for the softirq vector number and fix the call sites. Signed-off-by: Thomas Gleixner LKML-Reference: Acked-by: Peter Zijlstra Acked-by: mathieu.desnoyers@efficios.com Cc: Frederic Weisbecker Cc: Steven Rostedt --- kernel/softirq.c | 16 +++++++++------- 1 file changed, 9 insertions(+), 7 deletions(-) (limited to 'kernel') diff --git a/kernel/softirq.c b/kernel/softirq.c index 07b4f1b1a73a..b3cb1dc15795 100644 --- a/kernel/softirq.c +++ b/kernel/softirq.c @@ -212,18 +212,20 @@ restart: do { if (pending & 1) { + unsigned int vec_nr = h - softirq_vec; int prev_count = preempt_count(); - kstat_incr_softirqs_this_cpu(h - softirq_vec); - trace_softirq_entry(h, softirq_vec); + kstat_incr_softirqs_this_cpu(vec_nr); + + trace_softirq_entry(vec_nr); h->action(h); - trace_softirq_exit(h, softirq_vec); + trace_softirq_exit(vec_nr); if (unlikely(prev_count != preempt_count())) { - printk(KERN_ERR "huh, entered softirq %td %s %p" + printk(KERN_ERR "huh, entered softirq %u %s %p" "with preempt_count %08x," - " exited with %08x?\n", h - softirq_vec, - softirq_to_name[h - softirq_vec], - h->action, prev_count, preempt_count()); + " exited with %08x?\n", vec_nr, + softirq_to_name[vec_nr], h->action, + prev_count, preempt_count()); preempt_count() = prev_count; } -- cgit v1.2.3 From b2b5ce022acf5e9f52f7b78c5579994fdde191d4 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Fri, 15 Oct 2010 15:24:15 +0200 Subject: sched, cgroup: Fixup broken cgroup movement Dima noticed that we fail to correct the ->vruntime of sleeping tasks when we move them between cgroups. Reported-by: Dima Zavin Signed-off-by: Peter Zijlstra Tested-by: Mike Galbraith LKML-Reference: <1287150604.29097.1513.camel@twins> Signed-off-by: Ingo Molnar --- kernel/sched.c | 8 ++++---- kernel/sched_fair.c | 25 +++++++++++++++++++------ 2 files changed, 23 insertions(+), 10 deletions(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index 5998222f901c..3fe253e6a6e9 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -8498,12 +8498,12 @@ void sched_move_task(struct task_struct *tsk) if (unlikely(running)) tsk->sched_class->put_prev_task(rq, tsk); - set_task_rq(tsk, task_cpu(tsk)); - #ifdef CONFIG_FAIR_GROUP_SCHED - if (tsk->sched_class->moved_group) - tsk->sched_class->moved_group(tsk, on_rq); + if (tsk->sched_class->task_move_group) + tsk->sched_class->task_move_group(tsk, on_rq); + else #endif + set_task_rq(tsk, task_cpu(tsk)); if (unlikely(running)) tsk->sched_class->set_curr_task(rq); diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c index 74cccfae87a8..3acc2a487c18 100644 --- a/kernel/sched_fair.c +++ b/kernel/sched_fair.c @@ -3866,13 +3866,26 @@ static void set_curr_task_fair(struct rq *rq) } #ifdef CONFIG_FAIR_GROUP_SCHED -static void moved_group_fair(struct task_struct *p, int on_rq) +static void task_move_group_fair(struct task_struct *p, int on_rq) { - struct cfs_rq *cfs_rq = task_cfs_rq(p); - - update_curr(cfs_rq); + /* + * If the task was not on the rq at the time of this cgroup movement + * it must have been asleep, sleeping tasks keep their ->vruntime + * absolute on their old rq until wakeup (needed for the fair sleeper + * bonus in place_entity()). + * + * If it was on the rq, we've just 'preempted' it, which does convert + * ->vruntime to a relative base. + * + * Make sure both cases convert their relative position when migrating + * to another cgroup's rq. This does somewhat interfere with the + * fair sleeper stuff for the first placement, but who cares. + */ + if (!on_rq) + p->se.vruntime -= cfs_rq_of(&p->se)->min_vruntime; + set_task_rq(p, task_cpu(p)); if (!on_rq) - place_entity(cfs_rq, &p->se, 1); + p->se.vruntime += cfs_rq_of(&p->se)->min_vruntime; } #endif @@ -3924,7 +3937,7 @@ static const struct sched_class fair_sched_class = { .get_rr_interval = get_rr_interval_fair, #ifdef CONFIG_FAIR_GROUP_SCHED - .moved_group = moved_group_fair, + .task_move_group = task_move_group_fair, #endif }; -- cgit v1.2.3 From 9ffcfa6f1f63eeac15555b745c292eb9f59130f6 Mon Sep 17 00:00:00 2001 From: Stephane Eranian Date: Wed, 20 Oct 2010 15:25:01 +0200 Subject: perf_events: Revert: Fix transaction recovery in group_sched_in() This patch reverts commit 8e5fc1a (perf_events: Fix transaction recovery in group_sched_in()) because it had one flaw in case the group could never be scheduled. It would cause time_enabled to get negative. Signed-off-by: Stephane Eranian Signed-off-by: Peter Zijlstra LKML-Reference: <4cbeeeb7.0aefd80a.6e40.0e2f@mx.google.com> Signed-off-by: Ingo Molnar --- kernel/perf_event.c | 76 +++++++++-------------------------------------------- 1 file changed, 13 insertions(+), 63 deletions(-) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index f309e8014c78..39afdb07d758 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -417,8 +417,8 @@ event_filter_match(struct perf_event *event) return event->cpu == -1 || event->cpu == smp_processor_id(); } -static int -__event_sched_out(struct perf_event *event, +static void +event_sched_out(struct perf_event *event, struct perf_cpu_context *cpuctx, struct perf_event_context *ctx) { @@ -437,13 +437,14 @@ __event_sched_out(struct perf_event *event, } if (event->state != PERF_EVENT_STATE_ACTIVE) - return 0; + return; event->state = PERF_EVENT_STATE_INACTIVE; if (event->pending_disable) { event->pending_disable = 0; event->state = PERF_EVENT_STATE_OFF; } + event->tstamp_stopped = ctx->time; event->pmu->del(event, 0); event->oncpu = -1; @@ -452,19 +453,6 @@ __event_sched_out(struct perf_event *event, ctx->nr_active--; if (event->attr.exclusive || !cpuctx->active_oncpu) cpuctx->exclusive = 0; - return 1; -} - -static void -event_sched_out(struct perf_event *event, - struct perf_cpu_context *cpuctx, - struct perf_event_context *ctx) -{ - int ret; - - ret = __event_sched_out(event, cpuctx, ctx); - if (ret) - event->tstamp_stopped = ctx->time; } static void @@ -664,7 +652,7 @@ retry: } static int -__event_sched_in(struct perf_event *event, +event_sched_in(struct perf_event *event, struct perf_cpu_context *cpuctx, struct perf_event_context *ctx) { @@ -684,6 +672,8 @@ __event_sched_in(struct perf_event *event, return -EAGAIN; } + event->tstamp_running += ctx->time - event->tstamp_stopped; + if (!is_software_event(event)) cpuctx->active_oncpu++; ctx->nr_active++; @@ -694,35 +684,6 @@ __event_sched_in(struct perf_event *event, return 0; } -static inline int -event_sched_in(struct perf_event *event, - struct perf_cpu_context *cpuctx, - struct perf_event_context *ctx) -{ - int ret = __event_sched_in(event, cpuctx, ctx); - if (ret) - return ret; - event->tstamp_running += ctx->time - event->tstamp_stopped; - return 0; -} - -static void -group_commit_event_sched_in(struct perf_event *group_event, - struct perf_cpu_context *cpuctx, - struct perf_event_context *ctx) -{ - struct perf_event *event; - u64 now = ctx->time; - - group_event->tstamp_running += now - group_event->tstamp_stopped; - /* - * Schedule in siblings as one group (if any): - */ - list_for_each_entry(event, &group_event->sibling_list, group_entry) { - event->tstamp_running += now - event->tstamp_stopped; - } -} - static int group_sched_in(struct perf_event *group_event, struct perf_cpu_context *cpuctx, @@ -736,13 +697,7 @@ group_sched_in(struct perf_event *group_event, pmu->start_txn(pmu); - /* - * use __event_sched_in() to delay updating tstamp_running - * until the transaction is committed. In case of failure - * we will keep an unmodified tstamp_running which is a - * requirement to get correct timing information - */ - if (__event_sched_in(group_event, cpuctx, ctx)) { + if (event_sched_in(group_event, cpuctx, ctx)) { pmu->cancel_txn(pmu); return -EAGAIN; } @@ -751,31 +706,26 @@ group_sched_in(struct perf_event *group_event, * Schedule in siblings as one group (if any): */ list_for_each_entry(event, &group_event->sibling_list, group_entry) { - if (__event_sched_in(event, cpuctx, ctx)) { + if (event_sched_in(event, cpuctx, ctx)) { partial_group = event; goto group_error; } } - if (!pmu->commit_txn(pmu)) { - /* commit tstamp_running */ - group_commit_event_sched_in(group_event, cpuctx, ctx); + if (!pmu->commit_txn(pmu)) return 0; - } + group_error: /* * Groups can be scheduled in as one unit only, so undo any * partial group before returning: - * - * use __event_sched_out() to avoid updating tstamp_stopped - * because the event never actually ran */ list_for_each_entry(event, &group_event->sibling_list, group_entry) { if (event == partial_group) break; - __event_sched_out(event, cpuctx, ctx); + event_sched_out(event, cpuctx, ctx); } - __event_sched_out(group_event, cpuctx, ctx); + event_sched_out(group_event, cpuctx, ctx); pmu->cancel_txn(pmu); -- cgit v1.2.3 From d7842da470f244d258f21c5f72cd8388b3541d04 Mon Sep 17 00:00:00 2001 From: Stephane Eranian Date: Wed, 20 Oct 2010 15:25:01 +0200 Subject: perf_events: Fix for transaction recovery in group_sched_in() This new version (see commit 8e5fc1a) is much simpler and ensures that in case of error in group_sched_in() during event_sched_in(), the events up to the failed event go through regular event_sched_out(). But the failed event and the remaining events in the group have their timings adjusted as if they had also gone through event_sched_in() and event_sched_out(). This ensures timing uniformity across all events in a group. This also takes care of the tstamp_stopped problem in case the group could never be scheduled. The tstamp_stopped is updated as if the event had actually run. With this patch, the following now reports correct time_enabled, in case the NMI watchdog is active: $ task -e unhalted_core_cycles,instructions_retired,baclears,baclears noploop 1 noploop for 1 seconds 0 unhalted_core_cycles (100.00% scaling, ena=997,552,872, run=0) 0 instructions_retired (100.00% scaling, ena=997,552,872, run=0) 0 baclears (100.00% scaling, ena=997,552,872, run=0) 0 baclears (100.00% scaling, ena=997,552,872, run=0) And the older test case also works: $ task -einstructions_retired,baclears,baclears -e unhalted_core_cycles,baclears,baclears sleep 5 1680885 instructions_retired (69.39% scaling, ena=950756, run=291006) 10735 baclears (69.39% scaling, ena=950756, run=291006) 10735 baclears (69.39% scaling, ena=950756, run=291006) 0 unhalted_core_cycles (100.00% scaling, ena=817932, run=0) 0 baclears (100.00% scaling, ena=817932, run=0) 0 baclears (100.00% scaling, ena=817932, run=0) Signed-off-by: Stephane Eranian Signed-off-by: Peter Zijlstra LKML-Reference: <4cbeeebc.8ee7d80a.5a28.0d5f@mx.google.com> Signed-off-by: Ingo Molnar --- kernel/perf_event.c | 22 ++++++++++++++++++++-- 1 file changed, 20 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index 39afdb07d758..517d827f4982 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -691,6 +691,8 @@ group_sched_in(struct perf_event *group_event, { struct perf_event *event, *partial_group = NULL; struct pmu *pmu = group_event->pmu; + u64 now = ctx->time; + bool simulate = false; if (group_event->state == PERF_EVENT_STATE_OFF) return 0; @@ -719,11 +721,27 @@ group_error: /* * Groups can be scheduled in as one unit only, so undo any * partial group before returning: + * The events up to the failed event are scheduled out normally, + * tstamp_stopped will be updated. + * + * The failed events and the remaining siblings need to have + * their timings updated as if they had gone thru event_sched_in() + * and event_sched_out(). This is required to get consistent timings + * across the group. This also takes care of the case where the group + * could never be scheduled by ensuring tstamp_stopped is set to mark + * the time the event was actually stopped, such that time delta + * calculation in update_event_times() is correct. */ list_for_each_entry(event, &group_event->sibling_list, group_entry) { if (event == partial_group) - break; - event_sched_out(event, cpuctx, ctx); + simulate = true; + + if (simulate) { + event->tstamp_running += now - event->tstamp_stopped; + event->tstamp_stopped = now; + } else { + event_sched_out(event, cpuctx, ctx); + } } event_sched_out(group_event, cpuctx, ctx); -- cgit v1.2.3 From 2656c36699677238edc9ec1fea79039f1fddbcb6 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Fri, 22 Oct 2010 14:47:57 +0200 Subject: genirq: Warn if enable_irq is called before irq is set up The recent changes in the genirq core unearthed a bug in arch/um which called enable_irq() before the interrupt was set up. Warn and return instead of crashing the machine with a NULL pointer dereference. Signed-off-by: Thomas Gleixner Cc: Richard Weinberger --- kernel/irq/manage.c | 4 ++++ 1 file changed, 4 insertions(+) (limited to 'kernel') diff --git a/kernel/irq/manage.c b/kernel/irq/manage.c index 644e8d5fa367..5f92acc5f952 100644 --- a/kernel/irq/manage.c +++ b/kernel/irq/manage.c @@ -324,6 +324,10 @@ void enable_irq(unsigned int irq) if (!desc) return; + if (WARN(!desc->irq_data.chip || !desc->irq_data.chip->irq_enable, + KERN_ERR "enable_irq before setup/request_irq: irq %u\n", irq)) + return; + chip_bus_lock(desc); raw_spin_lock_irqsave(&desc->lock, flags); __enable_irq(desc, irq, false); -- cgit v1.2.3 From 16cdc628c3aed47d02205135b7e2f01e0064f566 Mon Sep 17 00:00:00 2001 From: Jason Wessel Date: Fri, 6 Aug 2010 11:47:14 -0500 Subject: debug_core: move all watch dog syncs to a single function Move the various clock and watch dog syncs to a single function in advance of adding another sync for the rcu stall detector. Signed-off-by: Jason Wessel --- kernel/debug/debug_core.c | 15 +++++++++------ 1 file changed, 9 insertions(+), 6 deletions(-) (limited to 'kernel') diff --git a/kernel/debug/debug_core.c b/kernel/debug/debug_core.c index de407c78178d..c812857d0b80 100644 --- a/kernel/debug/debug_core.c +++ b/kernel/debug/debug_core.c @@ -470,6 +470,12 @@ static void dbg_cpu_switch(int cpu, int next_cpu) kgdb_info[next_cpu].exception_state |= DCPU_NEXT_MASTER; } +static void dbg_touch_watchdogs(void) +{ + touch_softlockup_watchdog_sync(); + clocksource_touch_watchdog(); +} + static int kgdb_cpu_enter(struct kgdb_state *ks, struct pt_regs *regs) { unsigned long flags; @@ -523,8 +529,7 @@ return_normal: if (trace_on) tracing_on(); atomic_dec(&cpu_in_kgdb[cpu]); - touch_softlockup_watchdog_sync(); - clocksource_touch_watchdog(); + dbg_touch_watchdogs(); local_irq_restore(flags); return 0; } @@ -541,8 +546,7 @@ return_normal: (kgdb_info[cpu].task && kgdb_info[cpu].task->pid != kgdb_sstep_pid) && --sstep_tries) { atomic_set(&kgdb_active, -1); - touch_softlockup_watchdog_sync(); - clocksource_touch_watchdog(); + dbg_touch_watchdogs(); local_irq_restore(flags); goto acquirelock; @@ -659,8 +663,7 @@ kgdb_restore: tracing_on(); /* Free kgdb_active */ atomic_set(&kgdb_active, -1); - touch_softlockup_watchdog_sync(); - clocksource_touch_watchdog(); + dbg_touch_watchdogs(); local_irq_restore(flags); return kgdb_info[cpu].ret_state; -- cgit v1.2.3 From fb70b5888b70b0b50f738fbfc019445493112eb1 Mon Sep 17 00:00:00 2001 From: Jason Wessel Date: Fri, 13 Aug 2010 12:44:04 -0500 Subject: debug_core: stop rcu warnings on kernel resume When returning from the kernel debugger reset the rcu jiffies_stall value to prevent the rcu stall detector from sending NMI events which invoke a stack dump for each cpu in the system. Signed-off-by: Jason Wessel --- kernel/debug/debug_core.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'kernel') diff --git a/kernel/debug/debug_core.c b/kernel/debug/debug_core.c index c812857d0b80..5a3b04d20497 100644 --- a/kernel/debug/debug_core.c +++ b/kernel/debug/debug_core.c @@ -47,6 +47,7 @@ #include #include #include +#include #include #include @@ -474,6 +475,7 @@ static void dbg_touch_watchdogs(void) { touch_softlockup_watchdog_sync(); clocksource_touch_watchdog(); + rcu_cpu_stall_reset(); } static int kgdb_cpu_enter(struct kgdb_state *ks, struct pt_regs *regs) -- cgit v1.2.3 From f7030bbc446430ecd12c9ad02cf0ea94934e5f91 Mon Sep 17 00:00:00 2001 From: Jason Wessel Date: Mon, 11 Oct 2010 10:20:14 -0500 Subject: kdb: Allow kernel loadable modules to add kdb shell functions In order to allow kernel modules to dynamically add a command to the kdb shell the kdb_register, kdb_register_repeat, kdb_unregister, and kdb_printf need to be exported as GPL symbols. Any kernel module that adds a dynamic kdb shell function should only need to include linux/kdb.h. Signed-off-by: Jason Wessel --- kernel/debug/kdb/kdb_io.c | 2 +- kernel/debug/kdb/kdb_main.c | 4 ++++ kernel/debug/kdb/kdb_private.h | 39 --------------------------------------- 3 files changed, 5 insertions(+), 40 deletions(-) (limited to 'kernel') diff --git a/kernel/debug/kdb/kdb_io.c b/kernel/debug/kdb/kdb_io.c index c9b7f4f90bba..96fdaac46a80 100644 --- a/kernel/debug/kdb/kdb_io.c +++ b/kernel/debug/kdb/kdb_io.c @@ -823,4 +823,4 @@ int kdb_printf(const char *fmt, ...) return r; } - +EXPORT_SYMBOL_GPL(kdb_printf); diff --git a/kernel/debug/kdb/kdb_main.c b/kernel/debug/kdb/kdb_main.c index caf057a3de0e..5448990a299e 100644 --- a/kernel/debug/kdb/kdb_main.c +++ b/kernel/debug/kdb/kdb_main.c @@ -2783,6 +2783,8 @@ int kdb_register_repeat(char *cmd, return 0; } +EXPORT_SYMBOL_GPL(kdb_register_repeat); + /* * kdb_register - Compatibility register function for commands that do @@ -2805,6 +2807,7 @@ int kdb_register(char *cmd, return kdb_register_repeat(cmd, func, usage, help, minlen, KDB_REPEAT_NONE); } +EXPORT_SYMBOL_GPL(kdb_register); /* * kdb_unregister - This function is used to unregister a kernel @@ -2833,6 +2836,7 @@ int kdb_unregister(char *cmd) /* Couldn't find it. */ return 1; } +EXPORT_SYMBOL_GPL(kdb_unregister); /* Initialize the kdb command table. */ static void __init kdb_inittab(void) diff --git a/kernel/debug/kdb/kdb_private.h b/kernel/debug/kdb/kdb_private.h index be775f7e81e0..1921e6e4c0bc 100644 --- a/kernel/debug/kdb/kdb_private.h +++ b/kernel/debug/kdb/kdb_private.h @@ -15,29 +15,6 @@ #include #include "../debug_core.h" -/* Kernel Debugger Error codes. Must not overlap with command codes. */ -#define KDB_NOTFOUND (-1) -#define KDB_ARGCOUNT (-2) -#define KDB_BADWIDTH (-3) -#define KDB_BADRADIX (-4) -#define KDB_NOTENV (-5) -#define KDB_NOENVVALUE (-6) -#define KDB_NOTIMP (-7) -#define KDB_ENVFULL (-8) -#define KDB_ENVBUFFULL (-9) -#define KDB_TOOMANYBPT (-10) -#define KDB_TOOMANYDBREGS (-11) -#define KDB_DUPBPT (-12) -#define KDB_BPTNOTFOUND (-13) -#define KDB_BADMODE (-14) -#define KDB_BADINT (-15) -#define KDB_INVADDRFMT (-16) -#define KDB_BADREG (-17) -#define KDB_BADCPUNUM (-18) -#define KDB_BADLENGTH (-19) -#define KDB_NOBP (-20) -#define KDB_BADADDR (-21) - /* Kernel Debugger Command codes. Must not overlap with error codes. */ #define KDB_CMD_GO (-1001) #define KDB_CMD_CPU (-1002) @@ -93,17 +70,6 @@ */ #define KDB_MAXBPT 16 -/* Maximum number of arguments to a function */ -#define KDB_MAXARGS 16 - -typedef enum { - KDB_REPEAT_NONE = 0, /* Do not repeat this command */ - KDB_REPEAT_NO_ARGS, /* Repeat the command without arguments */ - KDB_REPEAT_WITH_ARGS, /* Repeat the command including its arguments */ -} kdb_repeat_t; - -typedef int (*kdb_func_t)(int, const char **); - /* Symbol table format returned by kallsyms. */ typedef struct __ksymtab { unsigned long value; /* Address of symbol */ @@ -123,11 +89,6 @@ extern int kallsyms_symbol_next(char *prefix_name, int flag); extern int kallsyms_symbol_complete(char *prefix_name, int max_len); /* Exported Symbols for kernel loadable modules to use. */ -extern int kdb_register(char *, kdb_func_t, char *, char *, short); -extern int kdb_register_repeat(char *, kdb_func_t, char *, char *, - short, kdb_repeat_t); -extern int kdb_unregister(char *); - extern int kdb_getarea_size(void *, unsigned long, size_t); extern int kdb_putarea_size(unsigned long, void *, size_t); -- cgit v1.2.3 From e3bda3ac33d3bf3e5a4049e2cabe82d3caaffc26 Mon Sep 17 00:00:00 2001 From: Jason Wessel Date: Mon, 11 Oct 2010 10:20:14 -0500 Subject: kdb,ftdump: Remove reference to internal kdb include Now that include/linux/kdb.h properly exports all the functions required to dynamically add a kdb shell command, the reference to the private kdb header can be removed. Signed-off-by: Jason Wessel --- kernel/trace/trace_kdb.c | 1 - 1 file changed, 1 deletion(-) (limited to 'kernel') diff --git a/kernel/trace/trace_kdb.c b/kernel/trace/trace_kdb.c index 7b8ecd751d93..3c5c5dfea0b3 100644 --- a/kernel/trace/trace_kdb.c +++ b/kernel/trace/trace_kdb.c @@ -13,7 +13,6 @@ #include #include -#include "../debug/kdb/kdb_private.h" #include "trace.h" #include "trace_output.h" -- cgit v1.2.3 From 75d14edee5689716b55afe467acfc13206a31f95 Mon Sep 17 00:00:00 2001 From: Jason Wessel Date: Mon, 11 Oct 2010 10:20:14 -0500 Subject: kdb: Fix oops in kdb_unregister Nothing should try to use kdb_commands directly as sometimes it is null. Instead, use the for_each_kdbcmd() iterator. This particular problem dates back to the initial kdb merge (2.6.35), but at that point nothing was dynamically unregistering commands from the kdb shell. Signed-off-by: Jason Wessel --- kernel/debug/kdb/kdb_main.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/debug/kdb/kdb_main.c b/kernel/debug/kdb/kdb_main.c index 5448990a299e..4226f32517d1 100644 --- a/kernel/debug/kdb/kdb_main.c +++ b/kernel/debug/kdb/kdb_main.c @@ -2826,7 +2826,7 @@ int kdb_unregister(char *cmd) /* * find the command. */ - for (i = 0, kp = kdb_commands; i < kdb_max_commands; i++, kp++) { + for_each_kdbcmd(kp, i) { if (kp->cmd_name && (strcmp(kp->cmd_name, cmd) == 0)) { kp->cmd_name = NULL; return 0; -- cgit v1.2.3 From 91b152aa85bbcf076e269565394c31964f940371 Mon Sep 17 00:00:00 2001 From: Jason Wessel Date: Mon, 23 Aug 2010 09:20:14 -0500 Subject: kdb,kgdb: fix sparse fixups Fix the following sparse warnings: kdb_main.c:328:5: warning: symbol 'kdbgetu64arg' was not declared. Should it be static? kgdboc.c:246:12: warning: symbol 'kgdboc_early_init' was not declared. Should it be static? kgdb.c:652:26: warning: incorrect type in argument 1 (different address spaces) kgdb.c:652:26: expected void const *ptr kgdb.c:652:26: got struct perf_event *[noderef] *pev The one in kgdb.c required the (void * __force) because of the return code from register_wide_hw_breakpoint looking like: return (void __percpu __force *)ERR_PTR(err); Signed-off-by: Jason Wessel --- kernel/debug/kdb/kdb_private.h | 9 +-------- 1 file changed, 1 insertion(+), 8 deletions(-) (limited to 'kernel') diff --git a/kernel/debug/kdb/kdb_private.h b/kernel/debug/kdb/kdb_private.h index 1921e6e4c0bc..35d69ed1dfb5 100644 --- a/kernel/debug/kdb/kdb_private.h +++ b/kernel/debug/kdb/kdb_private.h @@ -105,6 +105,7 @@ extern int kdb_getword(unsigned long *, unsigned long, size_t); extern int kdb_putword(unsigned long, unsigned long, size_t); extern int kdbgetularg(const char *, unsigned long *); +extern int kdbgetu64arg(const char *, u64 *); extern char *kdbgetenv(const char *); extern int kdbgetaddrarg(int, const char **, int*, unsigned long *, long *, char **); @@ -216,14 +217,6 @@ extern void kdb_ps1(const struct task_struct *p); extern void kdb_print_nameval(const char *name, unsigned long val); extern void kdb_send_sig_info(struct task_struct *p, struct siginfo *info); extern void kdb_meminfo_proc_show(void); -#ifdef CONFIG_KALLSYMS -extern const char *kdb_walk_kallsyms(loff_t *pos); -#else /* ! CONFIG_KALLSYMS */ -static inline const char *kdb_walk_kallsyms(loff_t *pos) -{ - return NULL; -} -#endif /* ! CONFIG_KALLSYMS */ extern char *kdb_getstr(char *, size_t, char *); /* Defines for kdb_symbol_print */ -- cgit v1.2.3 From c1bb9a9c1911036549c5cdfb23f32d7d20ffdc5a Mon Sep 17 00:00:00 2001 From: Dongdong Deng Date: Mon, 13 Sep 2010 06:58:00 -0500 Subject: debug_core: disable hw_breakpoints on all cores in kgdb_cpu_enter() The slave cpus do not have the hw breakpoints disabled upon entry to the debug_core and as a result could cause unrecoverable recursive faults on badly placed breakpoints, or get out of sync with the arch specific hw breakpoint operations. This patch addresses the problem by invoking kgdb_disable_hw_debug() earlier in kgdb_enter_cpu for each cpu that enters the debug core. The hw breakpoint dis/enable flow should be: master_debug_cpu slave_debug_cpu \ / kgdb_cpu_enter | kgdb_disable_hw_debug --> uninstall pre-enabled hw_breakpoint | do add/rm dis/enable operates to hw_breakpoints on master_debug_cpu.. | correct_hw_break --> correct/install the enabled hw_breakpoint | leave_kgdb Signed-off-by: Dongdong Deng Signed-off-by: Jason Wessel --- kernel/debug/debug_core.c | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/debug/debug_core.c b/kernel/debug/debug_core.c index 5a3b04d20497..bb9497724808 100644 --- a/kernel/debug/debug_core.c +++ b/kernel/debug/debug_core.c @@ -485,6 +485,9 @@ static int kgdb_cpu_enter(struct kgdb_state *ks, struct pt_regs *regs) int error; int i, cpu; int trace_on = 0; + + kgdb_disable_hw_debug(ks->linux_regs); + acquirelock: /* * Interrupts will be restored by the 'trap return' code, except when @@ -569,8 +572,6 @@ return_normal: if (dbg_io_ops->pre_exception) dbg_io_ops->pre_exception(); - kgdb_disable_hw_debug(ks->linux_regs); - /* * Get the passive CPU lock which will hold all the non-primary * CPU in a spin state while the debugger is active @@ -661,6 +662,8 @@ kgdb_restore: else kgdb_sstep_pid = 0; } + if (arch_kgdb_ops.correct_hw_break) + arch_kgdb_ops.correct_hw_break(); if (trace_on) tracing_on(); /* Free kgdb_active */ -- cgit v1.2.3 From dfee3a7b92208b30f77876068aece9ea571270c2 Mon Sep 17 00:00:00 2001 From: Jason Wessel Date: Fri, 21 May 2010 08:46:00 -0500 Subject: debug_core: refactor locking for master/slave cpus For quite some time there have been problems with memory barriers and various races with NMI on multi processor systems using the kernel debugger. The algorithm for entering the kernel debug core and resuming kernel execution was racy and had several known edge case problems with attempting to debug something on a heavily loaded system using breakpoints that are hit repeatedly and quickly. The prior "locking" design entry worked as follows: * The atomic counter kgdb_active was used with atomic exchange in order to elect a master cpu out of all the cpus that may have taken a debug exception. * The master cpu increments all elements of passive_cpu_wait[]. * The master cpu issues the round up cpus message. * Each "slave cpu" that enters the debug core increments its own element in cpu_in_kgdb[]. * Each "slave cpu" spins on passive_cpu_wait[] until it becomes 0. * The master cpu debugs the system. The new scheme removes the two arrays of atomic counters and replaces them with 2 single counters. One counter is used to count the number of cpus waiting to become a master cpu (because one or more hit an exception). The second counter is use to indicate how many cpus have entered as slave cpus. The new entry logic works as follows: * One or more cpus enters via kgdb_handle_exception() and increments the masters_in_kgdb. Each cpu attempts to get the spin lock called dbg_master_lock. * The master cpu sets kgdb_active to the current cpu. * The master cpu takes the spinlock dbg_slave_lock. * The master cpu asks to round up all the other cpus. * Each slave cpu that is not already in kgdb_handle_exception() will enter and increment slaves_in_kgdb. Each slave will now spin try_locking on dbg_slave_lock. * The master cpu waits for the sum of masters_in_kgdb and slaves_in_kgdb to be equal to the sum of the online cpus. * The master cpu debugs the system. In the new design the kgdb_active can only be changed while holding dbg_master_lock. Stress testing has not turned up any further entry/exit races that existed in the prior locking design. The prior locking design suffered from atomic variables not being truly atomic (in the capacity as used by kgdb) along with memory barrier races. Signed-off-by: Jason Wessel Acked-by: Dongdong Deng --- kernel/debug/debug_core.c | 105 ++++++++++++++++++++++++---------------------- kernel/debug/debug_core.h | 1 + 2 files changed, 56 insertions(+), 50 deletions(-) (limited to 'kernel') diff --git a/kernel/debug/debug_core.c b/kernel/debug/debug_core.c index bb9497724808..26dbdc37d219 100644 --- a/kernel/debug/debug_core.c +++ b/kernel/debug/debug_core.c @@ -110,13 +110,15 @@ static struct kgdb_bkpt kgdb_break[KGDB_MAX_BREAKPOINTS] = { */ atomic_t kgdb_active = ATOMIC_INIT(-1); EXPORT_SYMBOL_GPL(kgdb_active); +static DEFINE_RAW_SPINLOCK(dbg_master_lock); +static DEFINE_RAW_SPINLOCK(dbg_slave_lock); /* * We use NR_CPUs not PERCPU, in case kgdb is used to debug early * bootup code (which might not have percpu set up yet): */ -static atomic_t passive_cpu_wait[NR_CPUS]; -static atomic_t cpu_in_kgdb[NR_CPUS]; +static atomic_t masters_in_kgdb; +static atomic_t slaves_in_kgdb; static atomic_t kgdb_break_tasklet_var; atomic_t kgdb_setting_breakpoint; @@ -478,14 +480,23 @@ static void dbg_touch_watchdogs(void) rcu_cpu_stall_reset(); } -static int kgdb_cpu_enter(struct kgdb_state *ks, struct pt_regs *regs) +static int kgdb_cpu_enter(struct kgdb_state *ks, struct pt_regs *regs, + int exception_state) { unsigned long flags; int sstep_tries = 100; int error; - int i, cpu; + int cpu; int trace_on = 0; + int online_cpus = num_online_cpus(); + kgdb_info[ks->cpu].enter_kgdb++; + kgdb_info[ks->cpu].exception_state |= exception_state; + + if (exception_state == DCPU_WANT_MASTER) + atomic_inc(&masters_in_kgdb); + else + atomic_inc(&slaves_in_kgdb); kgdb_disable_hw_debug(ks->linux_regs); acquirelock: @@ -500,14 +511,15 @@ acquirelock: kgdb_info[cpu].task = current; kgdb_info[cpu].ret_state = 0; kgdb_info[cpu].irq_depth = hardirq_count() >> HARDIRQ_SHIFT; - /* - * Make sure the above info reaches the primary CPU before - * our cpu_in_kgdb[] flag setting does: - */ - atomic_inc(&cpu_in_kgdb[cpu]); - if (exception_level == 1) + /* Make sure the above info reaches the primary CPU */ + smp_mb(); + + if (exception_level == 1) { + if (raw_spin_trylock(&dbg_master_lock)) + atomic_xchg(&kgdb_active, cpu); goto cpu_master_loop; + } /* * CPU will loop if it is a slave or request to become a kgdb @@ -519,10 +531,12 @@ cpu_loop: kgdb_info[cpu].exception_state &= ~DCPU_NEXT_MASTER; goto cpu_master_loop; } else if (kgdb_info[cpu].exception_state & DCPU_WANT_MASTER) { - if (atomic_cmpxchg(&kgdb_active, -1, cpu) == cpu) + if (raw_spin_trylock(&dbg_master_lock)) { + atomic_xchg(&kgdb_active, cpu); break; + } } else if (kgdb_info[cpu].exception_state & DCPU_IS_SLAVE) { - if (!atomic_read(&passive_cpu_wait[cpu])) + if (!raw_spin_is_locked(&dbg_slave_lock)) goto return_normal; } else { return_normal: @@ -533,7 +547,11 @@ return_normal: arch_kgdb_ops.correct_hw_break(); if (trace_on) tracing_on(); - atomic_dec(&cpu_in_kgdb[cpu]); + kgdb_info[cpu].exception_state &= + ~(DCPU_WANT_MASTER | DCPU_IS_SLAVE); + kgdb_info[cpu].enter_kgdb--; + smp_mb__before_atomic_dec(); + atomic_dec(&slaves_in_kgdb); dbg_touch_watchdogs(); local_irq_restore(flags); return 0; @@ -551,6 +569,7 @@ return_normal: (kgdb_info[cpu].task && kgdb_info[cpu].task->pid != kgdb_sstep_pid) && --sstep_tries) { atomic_set(&kgdb_active, -1); + raw_spin_unlock(&dbg_master_lock); dbg_touch_watchdogs(); local_irq_restore(flags); @@ -576,10 +595,8 @@ return_normal: * Get the passive CPU lock which will hold all the non-primary * CPU in a spin state while the debugger is active */ - if (!kgdb_single_step) { - for (i = 0; i < NR_CPUS; i++) - atomic_inc(&passive_cpu_wait[i]); - } + if (!kgdb_single_step) + raw_spin_lock(&dbg_slave_lock); #ifdef CONFIG_SMP /* Signal the other CPUs to enter kgdb_wait() */ @@ -590,10 +607,9 @@ return_normal: /* * Wait for the other CPUs to be notified and be waiting for us: */ - for_each_online_cpu(i) { - while (kgdb_do_roundup && !atomic_read(&cpu_in_kgdb[i])) - cpu_relax(); - } + while (kgdb_do_roundup && (atomic_read(&masters_in_kgdb) + + atomic_read(&slaves_in_kgdb)) != online_cpus) + cpu_relax(); /* * At this point the primary processor is completely @@ -634,24 +650,11 @@ cpu_master_loop: if (dbg_io_ops->post_exception) dbg_io_ops->post_exception(); - atomic_dec(&cpu_in_kgdb[ks->cpu]); - if (!kgdb_single_step) { - for (i = NR_CPUS-1; i >= 0; i--) - atomic_dec(&passive_cpu_wait[i]); - /* - * Wait till all the CPUs have quit from the debugger, - * but allow a CPU that hit an exception and is - * waiting to become the master to remain in the debug - * core. - */ - for_each_online_cpu(i) { - while (kgdb_do_roundup && - atomic_read(&cpu_in_kgdb[i]) && - !(kgdb_info[i].exception_state & - DCPU_WANT_MASTER)) - cpu_relax(); - } + raw_spin_unlock(&dbg_slave_lock); + /* Wait till all the CPUs have quit from the debugger. */ + while (kgdb_do_roundup && atomic_read(&slaves_in_kgdb)) + cpu_relax(); } kgdb_restore: @@ -666,8 +669,15 @@ kgdb_restore: arch_kgdb_ops.correct_hw_break(); if (trace_on) tracing_on(); + + kgdb_info[cpu].exception_state &= + ~(DCPU_WANT_MASTER | DCPU_IS_SLAVE); + kgdb_info[cpu].enter_kgdb--; + smp_mb__before_atomic_dec(); + atomic_dec(&masters_in_kgdb); /* Free kgdb_active */ atomic_set(&kgdb_active, -1); + raw_spin_unlock(&dbg_master_lock); dbg_touch_watchdogs(); local_irq_restore(flags); @@ -686,7 +696,6 @@ kgdb_handle_exception(int evector, int signo, int ecode, struct pt_regs *regs) { struct kgdb_state kgdb_var; struct kgdb_state *ks = &kgdb_var; - int ret; ks->cpu = raw_smp_processor_id(); ks->ex_vector = evector; @@ -697,11 +706,10 @@ kgdb_handle_exception(int evector, int signo, int ecode, struct pt_regs *regs) if (kgdb_reenter_check(ks)) return 0; /* Ouch, double exception ! */ - kgdb_info[ks->cpu].exception_state |= DCPU_WANT_MASTER; - ret = kgdb_cpu_enter(ks, regs); - kgdb_info[ks->cpu].exception_state &= ~(DCPU_WANT_MASTER | - DCPU_IS_SLAVE); - return ret; + if (kgdb_info[ks->cpu].enter_kgdb != 0) + return 0; + + return kgdb_cpu_enter(ks, regs, DCPU_WANT_MASTER); } int kgdb_nmicallback(int cpu, void *regs) @@ -714,12 +722,9 @@ int kgdb_nmicallback(int cpu, void *regs) ks->cpu = cpu; ks->linux_regs = regs; - if (!atomic_read(&cpu_in_kgdb[cpu]) && - atomic_read(&kgdb_active) != -1 && - atomic_read(&kgdb_active) != cpu) { - kgdb_info[cpu].exception_state |= DCPU_IS_SLAVE; - kgdb_cpu_enter(ks, regs); - kgdb_info[cpu].exception_state &= ~DCPU_IS_SLAVE; + if (kgdb_info[ks->cpu].enter_kgdb == 0 && + raw_spin_is_locked(&dbg_master_lock)) { + kgdb_cpu_enter(ks, regs, DCPU_IS_SLAVE); return 0; } #endif diff --git a/kernel/debug/debug_core.h b/kernel/debug/debug_core.h index c5d753d80f67..3494c28a7e7a 100644 --- a/kernel/debug/debug_core.h +++ b/kernel/debug/debug_core.h @@ -40,6 +40,7 @@ struct debuggerinfo_struct { int exception_state; int ret_state; int irq_depth; + int enter_kgdb; }; extern struct debuggerinfo_struct kgdb_info[]; -- cgit v1.2.3 From 495363d380b4f4745bd8677912688654afc44020 Mon Sep 17 00:00:00 2001 From: Jason Wessel Date: Fri, 21 May 2010 08:46:00 -0500 Subject: kdb,debug_core: adjust master cpu switch logic against new debug_core locking The kdb shell needs to enforce switching back to the original CPU that took the exception before restoring normal kernel execution. Resuming from a different CPU than what took the original exception will cause problems with spin locks that are freed from the a different processor than had taken the lock. The special logic in dbg_cpu_switch() can go away entirely with because the state of what cpus want to be masters or slaves will remain unchanged between entry and exit of the debug_core exception context. Signed-off-by: Jason Wessel --- kernel/debug/debug_core.c | 16 ++-------------- kernel/debug/kdb/kdb_debugger.c | 3 +-- kernel/debug/kdb/kdb_main.c | 12 ++++++------ 3 files changed, 9 insertions(+), 22 deletions(-) (limited to 'kernel') diff --git a/kernel/debug/debug_core.c b/kernel/debug/debug_core.c index 26dbdc37d219..fec596da9bd0 100644 --- a/kernel/debug/debug_core.c +++ b/kernel/debug/debug_core.c @@ -460,19 +460,6 @@ static int kgdb_reenter_check(struct kgdb_state *ks) return 1; } -static void dbg_cpu_switch(int cpu, int next_cpu) -{ - /* Mark the cpu we are switching away from as a slave when it - * holds the kgdb_active token. This must be done so that the - * that all the cpus wait in for the debug core will not enter - * again as the master. */ - if (cpu == atomic_read(&kgdb_active)) { - kgdb_info[cpu].exception_state |= DCPU_IS_SLAVE; - kgdb_info[cpu].exception_state &= ~DCPU_WANT_MASTER; - } - kgdb_info[next_cpu].exception_state |= DCPU_NEXT_MASTER; -} - static void dbg_touch_watchdogs(void) { touch_softlockup_watchdog_sync(); @@ -638,7 +625,8 @@ cpu_master_loop: if (error == DBG_PASS_EVENT) { dbg_kdb_mode = !dbg_kdb_mode; } else if (error == DBG_SWITCH_CPU_EVENT) { - dbg_cpu_switch(cpu, dbg_switch_cpu); + kgdb_info[dbg_switch_cpu].exception_state |= + DCPU_NEXT_MASTER; goto cpu_loop; } else { kgdb_info[cpu].ret_state = error; diff --git a/kernel/debug/kdb/kdb_debugger.c b/kernel/debug/kdb/kdb_debugger.c index bf6e8270e957..dd0b1b7dd02c 100644 --- a/kernel/debug/kdb/kdb_debugger.c +++ b/kernel/debug/kdb/kdb_debugger.c @@ -86,7 +86,7 @@ int kdb_stub(struct kgdb_state *ks) } /* Set initial kdb state variables */ KDB_STATE_CLEAR(KGDB_TRANS); - kdb_initial_cpu = ks->cpu; + kdb_initial_cpu = atomic_read(&kgdb_active); kdb_current_task = kgdb_info[ks->cpu].task; kdb_current_regs = kgdb_info[ks->cpu].debuggerinfo; /* Remove any breakpoints as needed by kdb and clear single step */ @@ -105,7 +105,6 @@ int kdb_stub(struct kgdb_state *ks) ks->pass_exception = 1; KDB_FLAG_SET(CATASTROPHIC); } - kdb_initial_cpu = ks->cpu; if (KDB_STATE(SSBPT) && reason == KDB_REASON_SSTEP) { KDB_STATE_CLEAR(SSBPT); KDB_STATE_CLEAR(DOING_SS); diff --git a/kernel/debug/kdb/kdb_main.c b/kernel/debug/kdb/kdb_main.c index 4226f32517d1..d7bda21a106b 100644 --- a/kernel/debug/kdb/kdb_main.c +++ b/kernel/debug/kdb/kdb_main.c @@ -1749,13 +1749,13 @@ static int kdb_go(int argc, const char **argv) int nextarg; long offset; + if (raw_smp_processor_id() != kdb_initial_cpu) { + kdb_printf("go must execute on the entry cpu, " + "please use \"cpu %d\" and then execute go\n", + kdb_initial_cpu); + return KDB_BADCPUNUM; + } if (argc == 1) { - if (raw_smp_processor_id() != kdb_initial_cpu) { - kdb_printf("go
must be issued from the " - "initial cpu, do cpu %d first\n", - kdb_initial_cpu); - return KDB_ARGCOUNT; - } nextarg = 1; diag = kdbgetaddrarg(argc, argv, &nextarg, &addr, &offset, NULL); -- cgit v1.2.3 From aa7b250c252cc8e6b1daf0e1eada5eba42a1a68d Mon Sep 17 00:00:00 2001 From: Joe Perches Date: Thu, 21 Oct 2010 22:17:19 -0700 Subject: tracing: Fix 'faild' -> 'failed' typo Signed-off-by: Joe Perches Cc: Steven Rostedt Cc: Frederic Weisbecker Cc: Jiri Kosina LKML-Reference: Signed-off-by: Ingo Molnar --- kernel/trace/trace_kprobe.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/trace/trace_kprobe.c b/kernel/trace/trace_kprobe.c index 544301d29dee..b8d2852baa4a 100644 --- a/kernel/trace/trace_kprobe.c +++ b/kernel/trace/trace_kprobe.c @@ -648,7 +648,7 @@ static int register_trace_probe(struct trace_probe *tp) } ret = register_probe_event(tp); if (ret) { - pr_warning("Faild to register probe event(%d)\n", ret); + pr_warning("Failed to register probe event(%d)\n", ret); goto end; } -- cgit v1.2.3 From d4a6f3c32c39132318454e77d59ab14b06f6eb02 Mon Sep 17 00:00:00 2001 From: Rakib Mullick Date: Sun, 24 Oct 2010 16:28:47 +0600 Subject: sched_stat: Update sched_info_queue/dequeue() code comments Remove some sched_info_queue(), sched_info_dequeue() code comment. We no longer belongs to the era of O(1) and we don't use active or expired array anymore. Signed-off-by: Rakib Mullick Cc: Oleg Nesterov Cc: Peter Zijlstra LKML-Reference: Signed-off-by: Ingo Molnar --- kernel/sched_stats.h | 20 +------------------- 1 file changed, 1 insertion(+), 19 deletions(-) (limited to 'kernel') diff --git a/kernel/sched_stats.h b/kernel/sched_stats.h index 25c2f962f6fc..48ddf431db0e 100644 --- a/kernel/sched_stats.h +++ b/kernel/sched_stats.h @@ -157,15 +157,7 @@ static inline void sched_info_reset_dequeued(struct task_struct *t) } /* - * Called when a process is dequeued from the active array and given - * the cpu. We should note that with the exception of interactive - * tasks, the expired queue will become the active queue after the active - * queue is empty, without explicitly dequeuing and requeuing tasks in the - * expired queue. (Interactive tasks may be requeued directly to the - * active queue, thus delaying tasks in the expired queue from running; - * see scheduler_tick()). - * - * Though we are interested in knowing how long it was from the *first* time a + * We are interested in knowing how long it was from the *first* time a * task was queued to the time that it finally hit a cpu, we call this routine * from dequeue_task() to account for possible rq->clock skew across cpus. The * delta taken on each cpu would annul the skew. @@ -203,16 +195,6 @@ static void sched_info_arrive(struct task_struct *t) } /* - * Called when a process is queued into either the active or expired - * array. The time is noted and later used to determine how long we - * had to wait for us to reach the cpu. Since the expired queue will - * become the active queue after active queue is empty, without dequeuing - * and requeuing any tasks, we are interested in queuing to either. It - * is unusual but not impossible for tasks to be dequeued and immediately - * requeued in the same or another array: this can happen in sched_yield(), - * set_user_nice(), and even load_balance() as it moves tasks from runqueue - * to runqueue. - * * This function is only called from enqueue_task(), but also only updates * the timestamp if it is already not set. It's assumed that * sched_info_dequeued() will clear that stamp when appropriate. -- cgit v1.2.3 From 43948f50276eca010a22726860dfe9a4e8130136 Mon Sep 17 00:00:00 2001 From: Masami Hiramatsu Date: Mon, 25 Oct 2010 22:18:01 +0900 Subject: kprobes: Remove redundant text_mutex lock in optimize Remove text_mutex locking in optimize_all_kprobes, because this function doesn't modify text. It simply queues probes on optimization list for kprobe_optimizer worker thread. Signed-off-by: Masami Hiramatsu Cc: Ananth N Mavinakayanahalli Cc: Anil S Keshavamurthy Cc: David S. Miller Cc: Namhyung Kim Cc: Jason Baron Cc: Peter Zijlstra LKML-Reference: <20101025131801.19160.70939.stgit@ltc236.sdl.hitachi.co.jp> Signed-off-by: Ingo Molnar --- kernel/kprobes.c | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/kprobes.c b/kernel/kprobes.c index ec4210c6501e..7c44133f51ec 100644 --- a/kernel/kprobes.c +++ b/kernel/kprobes.c @@ -74,7 +74,8 @@ static struct hlist_head kretprobe_inst_table[KPROBE_TABLE_SIZE]; /* NOTE: change this value only with kprobe_mutex held */ static bool kprobes_all_disarmed; -static DEFINE_MUTEX(kprobe_mutex); /* Protects kprobe_table */ +/* This protects kprobe_table and optimizing_list */ +static DEFINE_MUTEX(kprobe_mutex); static DEFINE_PER_CPU(struct kprobe *, kprobe_instance) = NULL; static struct { spinlock_t lock ____cacheline_aligned_in_smp; @@ -595,6 +596,7 @@ static __kprobes void try_to_optimize_kprobe(struct kprobe *p) } #ifdef CONFIG_SYSCTL +/* This should be called with kprobe_mutex locked */ static void __kprobes optimize_all_kprobes(void) { struct hlist_head *head; @@ -607,17 +609,16 @@ static void __kprobes optimize_all_kprobes(void) return; kprobes_allow_optimization = true; - mutex_lock(&text_mutex); for (i = 0; i < KPROBE_TABLE_SIZE; i++) { head = &kprobe_table[i]; hlist_for_each_entry_rcu(p, node, head, hlist) if (!kprobe_disabled(p)) optimize_kprobe(p); } - mutex_unlock(&text_mutex); printk(KERN_INFO "Kprobes globally optimized\n"); } +/* This should be called with kprobe_mutex locked */ static void __kprobes unoptimize_all_kprobes(void) { struct hlist_head *head; -- cgit v1.2.3 From 5260562754c0aa4b95eebb1f851eaccce7286365 Mon Sep 17 00:00:00 2001 From: David Howells Date: Mon, 25 Oct 2010 23:41:11 +0100 Subject: MN10300: Fix the PERCPU() alignment to allow for workqueues In the MN10300 arch, we occasionally see an assertion being tripped in alloc_cwqs() at the following line: /* just in case, make sure it's actually aligned */ ---> BUG_ON(!IS_ALIGNED(wq->cpu_wq.v, align)); return wq->cpu_wq.v ? 0 : -ENOMEM; The values are: wa->cpu_wq.v => 0x902776e0 align => 0x100 and align is calculated by the following: const size_t align = max_t(size_t, 1 << WORK_STRUCT_FLAG_BITS, __alignof__(unsigned long long)); This is because the pointer in question (wq->cpu_wq.v) loses some of its lower bits to control flags, and so the object it points to must be sufficiently aligned to avoid the need to use those bits for pointing to things. Currently, 4 control bits and 4 colour bits are used in normal circumstances, plus a debugging bit if debugging is set. This requires the cpu_workqueue_struct struct to be at least 256 bytes aligned (or 512 bytes aligned with debugging). PERCPU() alignment on MN13000, however, is only 32 bytes as set in vmlinux.lds.S. So we set this to PAGE_SIZE (4096) to match most other arches and stick a comment in alloc_cwqs() for anyone else who triggers the assertion. Reported-by: Akira Takeuchi Signed-off-by: David Howells Acked-by: Mark Salter Cc: Tejun Heo Signed-off-by: Linus Torvalds --- kernel/workqueue.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/workqueue.c b/kernel/workqueue.c index 30acdb74cc23..e5ff2cbaadc2 100644 --- a/kernel/workqueue.c +++ b/kernel/workqueue.c @@ -2791,7 +2791,9 @@ static int alloc_cwqs(struct workqueue_struct *wq) } } - /* just in case, make sure it's actually aligned */ + /* just in case, make sure it's actually aligned + * - this is affected by PERCPU() alignment in vmlinux.lds.S + */ BUG_ON(!IS_ALIGNED(wq->cpu_wq.v, align)); return wq->cpu_wq.v ? 0 : -ENOMEM; } -- cgit v1.2.3 From 7e360c38abe2c70eae3ba5a8a17f17671d8b77c5 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Tue, 5 Oct 2010 09:32:55 +0200 Subject: fs: allow for more than 2^31 files Andrew, Could you please review this patch, you probably are the right guy to take it, because it crosses fs and net trees. Note : /proc/sys/fs/file-nr is a read-only file, so this patch doesnt depend on previous patch (sysctl: fix min/max handling in __do_proc_doulongvec_minmax()) Thanks ! [PATCH V4] fs: allow for more than 2^31 files Robin Holt tried to boot a 16TB system and found af_unix was overflowing a 32bit value : We were seeing a failure which prevented boot. The kernel was incapable of creating either a named pipe or unix domain socket. This comes down to a common kernel function called unix_create1() which does: atomic_inc(&unix_nr_socks); if (atomic_read(&unix_nr_socks) > 2 * get_max_files()) goto out; The function get_max_files() is a simple return of files_stat.max_files. files_stat.max_files is a signed integer and is computed in fs/file_table.c's files_init(). n = (mempages * (PAGE_SIZE / 1024)) / 10; files_stat.max_files = n; In our case, mempages (total_ram_pages) is approx 3,758,096,384 (0xe0000000). That leaves max_files at approximately 1,503,238,553. This causes 2 * get_max_files() to integer overflow. Fix is to let /proc/sys/fs/file-nr & /proc/sys/fs/file-max use long integers, and change af_unix to use an atomic_long_t instead of atomic_t. get_max_files() is changed to return an unsigned long. get_nr_files() is changed to return a long. unix_nr_socks is changed from atomic_t to atomic_long_t, while not strictly needed to address Robin problem. Before patch (on a 64bit kernel) : # echo 2147483648 >/proc/sys/fs/file-max # cat /proc/sys/fs/file-max -18446744071562067968 After patch: # echo 2147483648 >/proc/sys/fs/file-max # cat /proc/sys/fs/file-max 2147483648 # cat /proc/sys/fs/file-nr 704 0 2147483648 Reported-by: Robin Holt Signed-off-by: Eric Dumazet Acked-by: David Miller Reviewed-by: Robin Holt Tested-by: Robin Holt Signed-off-by: Al Viro --- kernel/sysctl.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/sysctl.c b/kernel/sysctl.c index 3a45c224770f..694b140852c2 100644 --- a/kernel/sysctl.c +++ b/kernel/sysctl.c @@ -1352,16 +1352,16 @@ static struct ctl_table fs_table[] = { { .procname = "file-nr", .data = &files_stat, - .maxlen = 3*sizeof(int), + .maxlen = sizeof(files_stat), .mode = 0444, .proc_handler = proc_nr_files, }, { .procname = "file-max", .data = &files_stat.max_files, - .maxlen = sizeof(int), + .maxlen = sizeof(files_stat.max_files), .mode = 0644, - .proc_handler = proc_dointvec, + .proc_handler = proc_doulongvec_minmax, }, { .procname = "nr_open", -- cgit v1.2.3 From cffbc8aa334f55c9ed42d25202eb3ebf3a97c195 Mon Sep 17 00:00:00 2001 From: Dave Chinner Date: Sat, 23 Oct 2010 05:03:02 -0400 Subject: fs: Convert nr_inodes and nr_unused to per-cpu counters The number of inodes allocated does not need to be tied to the addition or removal of an inode to/from a list. If we are not tied to a list lock, we could update the counters when inodes are initialised or destroyed, but to do that we need to convert the counters to be per-cpu (i.e. independent of a lock). This means that we have the freedom to change the list/locking implementation without needing to care about the counters. Based on a patch originally from Eric Dumazet. [AV: cleaned up a bit, fixed build breakage on weird configs Signed-off-by: Dave Chinner Reviewed-by: Christoph Hellwig Signed-off-by: Al Viro --- kernel/sysctl.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/sysctl.c b/kernel/sysctl.c index 694b140852c2..99a510cbfbb3 100644 --- a/kernel/sysctl.c +++ b/kernel/sysctl.c @@ -1340,14 +1340,14 @@ static struct ctl_table fs_table[] = { .data = &inodes_stat, .maxlen = 2*sizeof(int), .mode = 0444, - .proc_handler = proc_dointvec, + .proc_handler = proc_nr_inodes, }, { .procname = "inode-state", .data = &inodes_stat, .maxlen = 7*sizeof(int), .mode = 0444, - .proc_handler = proc_dointvec, + .proc_handler = proc_nr_inodes, }, { .procname = "file-nr", -- cgit v1.2.3 From 7de9c6ee3ecffd99e1628e81a5ea5468f7581a1f Mon Sep 17 00:00:00 2001 From: Al Viro Date: Sat, 23 Oct 2010 11:11:40 -0400 Subject: new helper: ihold() Clones an existing reference to inode; caller must already hold one. Signed-off-by: Al Viro --- kernel/futex.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/futex.c b/kernel/futex.c index a118bf160e0b..6c683b37f2ce 100644 --- a/kernel/futex.c +++ b/kernel/futex.c @@ -169,7 +169,7 @@ static void get_futex_key_refs(union futex_key *key) switch (key->both.offset & (FUT_OFF_INODE|FUT_OFF_MMSHARED)) { case FUT_OFF_INODE: - atomic_inc(&key->shared.inode->i_count); + ihold(key->shared.inode); break; case FUT_OFF_MMSHARED: atomic_inc(&key->private.mm->mm_count); -- cgit v1.2.3 From 85fe4025c616a7c0ed07bc2fc8c5371b07f3888c Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Sat, 23 Oct 2010 11:19:54 -0400 Subject: fs: do not assign default i_ino in new_inode Instead of always assigning an increasing inode number in new_inode move the call to assign it into those callers that actually need it. For now callers that need it is estimated conservatively, that is the call is added to all filesystems that do not assign an i_ino by themselves. For a few more filesystems we can avoid assigning any inode number given that they aren't user visible, and for others it could be done lazily when an inode number is actually needed, but that's left for later patches. Signed-off-by: Christoph Hellwig Signed-off-by: Dave Chinner Signed-off-by: Al Viro --- kernel/cgroup.c | 1 + 1 file changed, 1 insertion(+) (limited to 'kernel') diff --git a/kernel/cgroup.c b/kernel/cgroup.c index 7b69b8d0313d..9270d532ec3c 100644 --- a/kernel/cgroup.c +++ b/kernel/cgroup.c @@ -777,6 +777,7 @@ static struct inode *cgroup_new_inode(mode_t mode, struct super_block *sb) struct inode *inode = new_inode(sb); if (inode) { + inode->i_ino = get_next_ino(); inode->i_mode = mode; inode->i_uid = current_fsuid(); inode->i_gid = current_fsgid(); -- cgit v1.2.3 From 312d3ca856d369bb04d0443846b85b4cdde6fa8a Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Sun, 10 Oct 2010 05:36:23 -0400 Subject: fs: use percpu counter for nr_dentry and nr_dentry_unused The nr_dentry stat is a globally touched cacheline and atomic operation twice over the lifetime of a dentry. It is used for the benfit of userspace only. Turn it into a per-cpu counter and always decrement it in d_free instead of doing various batching operations to reduce lock hold times in the callers. Based on an earlier patch from Nick Piggin . Signed-off-by: Christoph Hellwig Signed-off-by: Al Viro --- kernel/sysctl.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/sysctl.c b/kernel/sysctl.c index 99a510cbfbb3..8b77ff5c502c 100644 --- a/kernel/sysctl.c +++ b/kernel/sysctl.c @@ -1377,7 +1377,7 @@ static struct ctl_table fs_table[] = { .data = &dentry_stat, .maxlen = 6*sizeof(int), .mode = 0444, - .proc_handler = proc_dointvec, + .proc_handler = proc_nr_dentry, }, { .procname = "overflowuid", -- cgit v1.2.3 From a9cea017411c95ec789092971f9baaef1f826883 Mon Sep 17 00:00:00 2001 From: Bjorn Helgaas Date: Tue, 26 Oct 2010 15:41:13 -0600 Subject: resources: add a default alignf to simplify find_resource() This removes a test from find_resource(), which is getting cluttered. No functional change. Signed-off-by: Bjorn Helgaas Signed-off-by: Jesse Barnes --- kernel/resource.c | 15 +++++++++++++-- 1 file changed, 13 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/resource.c b/kernel/resource.c index 7b36976e5dea..7dc8ad24f915 100644 --- a/kernel/resource.c +++ b/kernel/resource.c @@ -357,6 +357,14 @@ int __weak page_is_ram(unsigned long pfn) return walk_system_ram_range(pfn, 1, NULL, __is_ram) == 1; } +static resource_size_t simple_align_resource(void *data, + const struct resource *avail, + resource_size_t size, + resource_size_t align) +{ + return avail->start; +} + /* * Find empty slot in the resource tree given range and alignment. */ @@ -391,8 +399,8 @@ static int find_resource(struct resource *root, struct resource *new, if (tmp.end > max) tmp.end = max; tmp.start = ALIGN(tmp.start, align); - if (alignf) - tmp.start = alignf(alignf_data, &tmp, size, align); + + tmp.start = alignf(alignf_data, &tmp, size, align); if (tmp.start < tmp.end && tmp.end - tmp.start >= size - 1) { new->start = tmp.start; new->end = tmp.start + size - 1; @@ -428,6 +436,9 @@ int allocate_resource(struct resource *root, struct resource *new, { int err; + if (!alignf) + alignf = simple_align_resource; + write_lock(&resource_lock); err = find_resource(root, new, size, min, max, align, alignf, alignf_data); if (err >= 0 && __request_resource(root, new)) -- cgit v1.2.3 From 5d6b1fa301b13cc651ee717a9b518124dea2f814 Mon Sep 17 00:00:00 2001 From: Bjorn Helgaas Date: Tue, 26 Oct 2010 15:41:18 -0600 Subject: resources: factor out resource_clip() to simplify find_resource() This factors out the min/max clipping to simplify find_resource(). No functional change. Signed-off-by: Bjorn Helgaas Signed-off-by: Jesse Barnes --- kernel/resource.c | 15 +++++++++++---- 1 file changed, 11 insertions(+), 4 deletions(-) (limited to 'kernel') diff --git a/kernel/resource.c b/kernel/resource.c index 7dc8ad24f915..26e9f2546923 100644 --- a/kernel/resource.c +++ b/kernel/resource.c @@ -365,6 +365,15 @@ static resource_size_t simple_align_resource(void *data, return avail->start; } +static void resource_clip(struct resource *res, resource_size_t min, + resource_size_t max) +{ + if (res->start < min) + res->start = min; + if (res->end > max) + res->end = max; +} + /* * Find empty slot in the resource tree given range and alignment. */ @@ -394,10 +403,8 @@ static int find_resource(struct resource *root, struct resource *new, tmp.end = this->start - 1; else tmp.end = root->end; - if (tmp.start < min) - tmp.start = min; - if (tmp.end > max) - tmp.end = max; + + resource_clip(&tmp, min, max); tmp.start = ALIGN(tmp.start, align); tmp.start = alignf(alignf_data, &tmp, size, align); -- cgit v1.2.3 From 6909ba14c25b4db6be2ff89f4fa0fac2d70151a0 Mon Sep 17 00:00:00 2001 From: Bjorn Helgaas Date: Tue, 26 Oct 2010 15:41:23 -0600 Subject: resources: ensure callback doesn't allocate outside available space The alignment callback returns a proposed location, which may have been adjusted to avoid ISA aliases or for other architecture-specific reasons. We already had a check ("tmp.start < tmp.end") to make sure the callback doesn't return an area that extends past the available area. This patch reworks the check to make sure it doesn't return an area that extends either below or above the available area. Signed-off-by: Bjorn Helgaas Signed-off-by: Jesse Barnes --- kernel/resource.c | 16 +++++++++++----- 1 file changed, 11 insertions(+), 5 deletions(-) (limited to 'kernel') diff --git a/kernel/resource.c b/kernel/resource.c index 26e9f2546923..89d50412508c 100644 --- a/kernel/resource.c +++ b/kernel/resource.c @@ -374,6 +374,11 @@ static void resource_clip(struct resource *res, resource_size_t min, res->end = max; } +static bool resource_contains(struct resource *res1, struct resource *res2) +{ + return res1->start <= res2->start && res1->end >= res2->end; +} + /* * Find empty slot in the resource tree given range and alignment. */ @@ -387,7 +392,7 @@ static int find_resource(struct resource *root, struct resource *new, void *alignf_data) { struct resource *this = root->child; - struct resource tmp = *new; + struct resource tmp = *new, alloc; tmp.start = root->start; /* @@ -407,10 +412,11 @@ static int find_resource(struct resource *root, struct resource *new, resource_clip(&tmp, min, max); tmp.start = ALIGN(tmp.start, align); - tmp.start = alignf(alignf_data, &tmp, size, align); - if (tmp.start < tmp.end && tmp.end - tmp.start >= size - 1) { - new->start = tmp.start; - new->end = tmp.start + size - 1; + alloc.start = alignf(alignf_data, &tmp, size, align); + alloc.end = alloc.start + size - 1; + if (resource_contains(&tmp, &alloc)) { + new->start = alloc.start; + new->end = alloc.end; return 0; } if (!this) -- cgit v1.2.3 From a1862e31079149a52b6223776228c3aee493d4a7 Mon Sep 17 00:00:00 2001 From: Bjorn Helgaas Date: Tue, 26 Oct 2010 15:41:28 -0600 Subject: resources: handle overflow when aligning start of available area If tmp.start is near ~0, ALIGN(tmp.start) may overflow, which would make us think there's more available space than there really is. We would likely return something that conflicts with a previous resource, which would cause a failure when allocate_resource() requests the newly- allocated region. Reference: https://bugzilla.redhat.com/show_bug.cgi?id=646027 Reported-by: Fabrice Bellet Signed-off-by: Bjorn Helgaas Signed-off-by: Jesse Barnes --- kernel/resource.c | 21 +++++++++++++-------- 1 file changed, 13 insertions(+), 8 deletions(-) (limited to 'kernel') diff --git a/kernel/resource.c b/kernel/resource.c index 89d50412508c..e15b922d4ba4 100644 --- a/kernel/resource.c +++ b/kernel/resource.c @@ -392,7 +392,7 @@ static int find_resource(struct resource *root, struct resource *new, void *alignf_data) { struct resource *this = root->child; - struct resource tmp = *new, alloc; + struct resource tmp = *new, avail, alloc; tmp.start = root->start; /* @@ -410,14 +410,19 @@ static int find_resource(struct resource *root, struct resource *new, tmp.end = root->end; resource_clip(&tmp, min, max); - tmp.start = ALIGN(tmp.start, align); - alloc.start = alignf(alignf_data, &tmp, size, align); - alloc.end = alloc.start + size - 1; - if (resource_contains(&tmp, &alloc)) { - new->start = alloc.start; - new->end = alloc.end; - return 0; + /* Check for overflow after ALIGN() */ + avail = *new; + avail.start = ALIGN(tmp.start, align); + avail.end = tmp.end; + if (avail.start >= tmp.start) { + alloc.start = alignf(alignf_data, &avail, size, align); + alloc.end = alloc.start + size - 1; + if (resource_contains(&avail, &alloc)) { + new->start = alloc.start; + new->end = alloc.end; + return 0; + } } if (!this) break; -- cgit v1.2.3 From e7f8567db9a7f6b3151b0b275e245c1cef0d9c70 Mon Sep 17 00:00:00 2001 From: Bjorn Helgaas Date: Tue, 26 Oct 2010 15:41:33 -0600 Subject: resources: support allocating space within a region from the top down Allocate space from the top of a region first, then work downward, if an architecture desires this. When we allocate space from a resource, we look for gaps between children of the resource. Previously, we always looked at gaps from the bottom up. For example, given this: [mem 0xbff00000-0xf7ffffff] PCI Bus 0000:00 [mem 0xbff00000-0xbfffffff] gap -- available [mem 0xc0000000-0xdfffffff] PCI Bus 0000:02 [mem 0xe0000000-0xf7ffffff] gap -- available we attempted to allocate from the [mem 0xbff00000-0xbfffffff] gap first, then the [mem 0xe0000000-0xf7ffffff] gap. With this patch an architecture can choose to allocate from the top gap [mem 0xe0000000-0xf7ffffff] first. We can't do this across the board because iomem_resource.end is initialized to 0xffffffff_ffffffff on 64-bit architectures, and most machines can't address the entire 64-bit physical address space. Therefore, we only allocate top-down if the arch requests it by clearing "resource_alloc_from_bottom". Signed-off-by: Bjorn Helgaas Signed-off-by: Jesse Barnes --- kernel/resource.c | 98 ++++++++++++++++++++++++++++++++++++++++++++++++++++--- 1 file changed, 94 insertions(+), 4 deletions(-) (limited to 'kernel') diff --git a/kernel/resource.c b/kernel/resource.c index e15b922d4ba4..716b6804077e 100644 --- a/kernel/resource.c +++ b/kernel/resource.c @@ -40,6 +40,23 @@ EXPORT_SYMBOL(iomem_resource); static DEFINE_RWLOCK(resource_lock); +/* + * By default, we allocate free space bottom-up. The architecture can request + * top-down by clearing this flag. The user can override the architecture's + * choice with the "resource_alloc_from_bottom" kernel boot option, but that + * should only be a debugging tool. + */ +int resource_alloc_from_bottom = 1; + +static __init int setup_alloc_from_bottom(char *s) +{ + printk(KERN_INFO + "resource: allocating from bottom-up; please report a bug\n"); + resource_alloc_from_bottom = 1; + return 0; +} +early_param("resource_alloc_from_bottom", setup_alloc_from_bottom); + static void *r_next(struct seq_file *m, void *v, loff_t *pos) { struct resource *p = v; @@ -379,8 +396,75 @@ static bool resource_contains(struct resource *res1, struct resource *res2) return res1->start <= res2->start && res1->end >= res2->end; } +/* + * Find the resource before "child" in the sibling list of "root" children. + */ +static struct resource *find_sibling_prev(struct resource *root, struct resource *child) +{ + struct resource *this; + + for (this = root->child; this; this = this->sibling) + if (this->sibling == child) + return this; + + return NULL; +} + +/* + * Find empty slot in the resource tree given range and alignment. + * This version allocates from the end of the root resource first. + */ +static int find_resource_from_top(struct resource *root, struct resource *new, + resource_size_t size, resource_size_t min, + resource_size_t max, resource_size_t align, + resource_size_t (*alignf)(void *, + const struct resource *, + resource_size_t, + resource_size_t), + void *alignf_data) +{ + struct resource *this; + struct resource tmp, avail, alloc; + + tmp.start = root->end; + tmp.end = root->end; + + this = find_sibling_prev(root, NULL); + for (;;) { + if (this) { + if (this->end < root->end) + tmp.start = this->end + 1; + } else + tmp.start = root->start; + + resource_clip(&tmp, min, max); + + /* Check for overflow after ALIGN() */ + avail = *new; + avail.start = ALIGN(tmp.start, align); + avail.end = tmp.end; + if (avail.start >= tmp.start) { + alloc.start = alignf(alignf_data, &avail, size, align); + alloc.end = alloc.start + size - 1; + if (resource_contains(&avail, &alloc)) { + new->start = alloc.start; + new->end = alloc.end; + return 0; + } + } + + if (!this || this->start == root->start) + break; + + tmp.end = this->start - 1; + this = find_sibling_prev(root, this); + } + return -EBUSY; +} + /* * Find empty slot in the resource tree given range and alignment. + * This version allocates from the beginning of the root resource first. */ static int find_resource(struct resource *root, struct resource *new, resource_size_t size, resource_size_t min, @@ -396,14 +480,15 @@ static int find_resource(struct resource *root, struct resource *new, tmp.start = root->start; /* - * Skip past an allocated resource that starts at 0, since the assignment - * of this->start - 1 to tmp->end below would cause an underflow. + * Skip past an allocated resource that starts at 0, since the + * assignment of this->start - 1 to tmp->end below would cause an + * underflow. */ if (this && this->start == 0) { tmp.start = this->end + 1; this = this->sibling; } - for(;;) { + for (;;) { if (this) tmp.end = this->start - 1; else @@ -424,8 +509,10 @@ static int find_resource(struct resource *root, struct resource *new, return 0; } } + if (!this) break; + tmp.start = this->end + 1; this = this->sibling; } @@ -458,7 +545,10 @@ int allocate_resource(struct resource *root, struct resource *new, alignf = simple_align_resource; write_lock(&resource_lock); - err = find_resource(root, new, size, min, max, align, alignf, alignf_data); + if (resource_alloc_from_bottom) + err = find_resource(root, new, size, min, max, align, alignf, alignf_data); + else + err = find_resource_from_top(root, new, size, min, max, align, alignf, alignf_data); if (err >= 0 && __request_resource(root, new)) err = -EBUSY; write_unlock(&resource_lock); -- cgit v1.2.3 From 3d5992d2ac7dc09aed8ab537cba074589f0f0a52 Mon Sep 17 00:00:00 2001 From: Ying Han Date: Tue, 26 Oct 2010 14:21:23 -0700 Subject: oom: add per-mm oom disable count It's pointless to kill a task if another thread sharing its mm cannot be killed to allow future memory freeing. A subsequent patch will prevent kills in such cases, but first it's necessary to have a way to flag a task that shares memory with an OOM_DISABLE task that doesn't incur an additional tasklist scan, which would make select_bad_process() an O(n^2) function. This patch adds an atomic counter to struct mm_struct that follows how many threads attached to it have an oom_score_adj of OOM_SCORE_ADJ_MIN. They cannot be killed by the kernel, so their memory cannot be freed in oom conditions. This only requires task_lock() on the task that we're operating on, it does not require mm->mmap_sem since task_lock() pins the mm and the operation is atomic. [rientjes@google.com: changelog and sys_unshare() code] [rientjes@google.com: protect oom_disable_count with task_lock in fork] [rientjes@google.com: use old_mm for oom_disable_count in exec] Signed-off-by: Ying Han Signed-off-by: David Rientjes Cc: KAMEZAWA Hiroyuki Cc: KOSAKI Motohiro Cc: Rik van Riel Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/exit.c | 3 +++ kernel/fork.c | 15 ++++++++++++++- 2 files changed, 17 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/exit.c b/kernel/exit.c index e2bdf37f9fde..894179a32ec1 100644 --- a/kernel/exit.c +++ b/kernel/exit.c @@ -50,6 +50,7 @@ #include #include #include +#include #include #include @@ -687,6 +688,8 @@ static void exit_mm(struct task_struct * tsk) enter_lazy_tlb(mm, current); /* We don't want this task to be frozen prematurely */ clear_freeze_flag(tsk); + if (tsk->signal->oom_score_adj == OOM_SCORE_ADJ_MIN) + atomic_dec(&mm->oom_disable_count); task_unlock(tsk); mm_update_next_owner(mm); mmput(mm); diff --git a/kernel/fork.c b/kernel/fork.c index c445f8cc408d..e87aaaaf5131 100644 --- a/kernel/fork.c +++ b/kernel/fork.c @@ -65,6 +65,7 @@ #include #include #include +#include #include #include @@ -488,6 +489,7 @@ static struct mm_struct * mm_init(struct mm_struct * mm, struct task_struct *p) mm->cached_hole_size = ~0UL; mm_init_aio(mm); mm_init_owner(mm, p); + atomic_set(&mm->oom_disable_count, 0); if (likely(!mm_alloc_pgd(mm))) { mm->def_flags = 0; @@ -741,6 +743,8 @@ good_mm: /* Initializing for Swap token stuff */ mm->token_priority = 0; mm->last_interval = 0; + if (tsk->signal->oom_score_adj == OOM_SCORE_ADJ_MIN) + atomic_inc(&mm->oom_disable_count); tsk->mm = mm; tsk->active_mm = mm; @@ -1299,8 +1303,13 @@ bad_fork_cleanup_io: bad_fork_cleanup_namespaces: exit_task_namespaces(p); bad_fork_cleanup_mm: - if (p->mm) + if (p->mm) { + task_lock(p); + if (p->signal->oom_score_adj == OOM_SCORE_ADJ_MIN) + atomic_dec(&p->mm->oom_disable_count); + task_unlock(p); mmput(p->mm); + } bad_fork_cleanup_signal: if (!(clone_flags & CLONE_THREAD)) free_signal_struct(p->signal); @@ -1693,6 +1702,10 @@ SYSCALL_DEFINE1(unshare, unsigned long, unshare_flags) active_mm = current->active_mm; current->mm = new_mm; current->active_mm = new_mm; + if (current->signal->oom_score_adj == OOM_SCORE_ADJ_MIN) { + atomic_dec(&mm->oom_disable_count); + atomic_inc(&new_mm->oom_disable_count); + } activate_mm(active_mm, new_mm); new_mm = mm; } -- cgit v1.2.3 From 61ecdb801ef2cd28e32442383106d7837d76deac Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Tue, 26 Oct 2010 14:21:47 -0700 Subject: mm: strictly nested kmap_atomic() Ensure kmap_atomic() usage is strictly nested Signed-off-by: Peter Zijlstra Reviewed-by: Rik van Riel Acked-by: Chris Metcalf Cc: David Howells Cc: Hugh Dickins Cc: Ingo Molnar Cc: Thomas Gleixner Cc: "H. Peter Anvin" Cc: Steven Rostedt Cc: Russell King Cc: Ralf Baechle Cc: David Miller Cc: Paul Mackerras Cc: Benjamin Herrenschmidt Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/power/snapshot.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/power/snapshot.c b/kernel/power/snapshot.c index ac7eb109f196..9e3581f4619a 100644 --- a/kernel/power/snapshot.c +++ b/kernel/power/snapshot.c @@ -984,8 +984,8 @@ static void copy_data_page(unsigned long dst_pfn, unsigned long src_pfn) src = kmap_atomic(s_page, KM_USER0); dst = kmap_atomic(d_page, KM_USER1); do_copy_page(dst, src); - kunmap_atomic(src, KM_USER0); kunmap_atomic(dst, KM_USER1); + kunmap_atomic(src, KM_USER0); } else { if (PageHighMem(d_page)) { /* Page pointed to by src may contain some kernel @@ -2273,8 +2273,8 @@ swap_two_pages_data(struct page *p1, struct page *p2, void *buf) memcpy(buf, kaddr1, PAGE_SIZE); memcpy(kaddr1, kaddr2, PAGE_SIZE); memcpy(kaddr2, buf, PAGE_SIZE); - kunmap_atomic(kaddr1, KM_USER0); kunmap_atomic(kaddr2, KM_USER1); + kunmap_atomic(kaddr1, KM_USER0); } /** -- cgit v1.2.3 From 3ecb01df3261d3b1f02ccfcf8384e2a255d2a1d0 Mon Sep 17 00:00:00 2001 From: Jan Beulich Date: Tue, 26 Oct 2010 14:22:27 -0700 Subject: use clear_page()/copy_page() in favor of memset()/memcpy() on whole pages After all that's what they are intended for. Signed-off-by: Jan Beulich Cc: Miklos Szeredi Cc: "Eric W. Biederman" Cc: "Rafael J. Wysocki" Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/kexec.c | 2 +- kernel/power/snapshot.c | 14 +++++++------- kernel/power/swap.c | 6 +++--- 3 files changed, 11 insertions(+), 11 deletions(-) (limited to 'kernel') diff --git a/kernel/kexec.c b/kernel/kexec.c index c0613f7d6730..b55045bc7563 100644 --- a/kernel/kexec.c +++ b/kernel/kexec.c @@ -816,7 +816,7 @@ static int kimage_load_normal_segment(struct kimage *image, ptr = kmap(page); /* Start with a clear page */ - memset(ptr, 0, PAGE_SIZE); + clear_page(ptr); ptr += maddr & ~PAGE_MASK; mchunk = PAGE_SIZE - (maddr & ~PAGE_MASK); if (mchunk > mbytes) diff --git a/kernel/power/snapshot.c b/kernel/power/snapshot.c index 9e3581f4619a..0dac75ea4456 100644 --- a/kernel/power/snapshot.c +++ b/kernel/power/snapshot.c @@ -993,7 +993,7 @@ static void copy_data_page(unsigned long dst_pfn, unsigned long src_pfn) */ safe_copy_page(buffer, s_page); dst = kmap_atomic(d_page, KM_USER0); - memcpy(dst, buffer, PAGE_SIZE); + copy_page(dst, buffer); kunmap_atomic(dst, KM_USER0); } else { safe_copy_page(page_address(d_page), s_page); @@ -1687,7 +1687,7 @@ int snapshot_read_next(struct snapshot_handle *handle) memory_bm_position_reset(&orig_bm); memory_bm_position_reset(©_bm); } else if (handle->cur <= nr_meta_pages) { - memset(buffer, 0, PAGE_SIZE); + clear_page(buffer); pack_pfns(buffer, &orig_bm); } else { struct page *page; @@ -1701,7 +1701,7 @@ int snapshot_read_next(struct snapshot_handle *handle) void *kaddr; kaddr = kmap_atomic(page, KM_USER0); - memcpy(buffer, kaddr, PAGE_SIZE); + copy_page(buffer, kaddr); kunmap_atomic(kaddr, KM_USER0); handle->buffer = buffer; } else { @@ -1984,7 +1984,7 @@ static void copy_last_highmem_page(void) void *dst; dst = kmap_atomic(last_highmem_page, KM_USER0); - memcpy(dst, buffer, PAGE_SIZE); + copy_page(dst, buffer); kunmap_atomic(dst, KM_USER0); last_highmem_page = NULL; } @@ -2270,9 +2270,9 @@ swap_two_pages_data(struct page *p1, struct page *p2, void *buf) kaddr1 = kmap_atomic(p1, KM_USER0); kaddr2 = kmap_atomic(p2, KM_USER1); - memcpy(buf, kaddr1, PAGE_SIZE); - memcpy(kaddr1, kaddr2, PAGE_SIZE); - memcpy(kaddr2, buf, PAGE_SIZE); + copy_page(buf, kaddr1); + copy_page(kaddr1, kaddr2); + copy_page(kaddr2, buf); kunmap_atomic(kaddr2, KM_USER1); kunmap_atomic(kaddr1, KM_USER0); } diff --git a/kernel/power/swap.c b/kernel/power/swap.c index 916eaa790399..a0e4a86ccf94 100644 --- a/kernel/power/swap.c +++ b/kernel/power/swap.c @@ -251,7 +251,7 @@ static int write_page(void *buf, sector_t offset, struct bio **bio_chain) if (bio_chain) { src = (void *)__get_free_page(__GFP_WAIT | __GFP_HIGH); if (src) { - memcpy(src, buf, PAGE_SIZE); + copy_page(src, buf); } else { WARN_ON_ONCE(1); bio_chain = NULL; /* Go synchronous */ @@ -325,7 +325,7 @@ static int swap_write_page(struct swap_map_handle *handle, void *buf, error = write_page(handle->cur, handle->cur_swap, NULL); if (error) goto out; - memset(handle->cur, 0, PAGE_SIZE); + clear_page(handle->cur); handle->cur_swap = offset; handle->k = 0; } @@ -910,7 +910,7 @@ int swsusp_check(void) hib_resume_bdev = open_by_devnum(swsusp_resume_device, FMODE_READ); if (!IS_ERR(hib_resume_bdev)) { set_blocksize(hib_resume_bdev, PAGE_SIZE); - memset(swsusp_header, 0, PAGE_SIZE); + clear_page(swsusp_header); error = hib_bio_read_page(swsusp_resume_block, swsusp_header, NULL); if (error) -- cgit v1.2.3 From ca1cab37d91cbe8a8333732540d43cabb54cfa85 Mon Sep 17 00:00:00 2001 From: Andrew Morton Date: Tue, 26 Oct 2010 14:22:34 -0700 Subject: workqueues: s/ON_STACK/ONSTACK/ Silly though it is, completions and wait_queue_heads use foo_ONSTACK (COMPLETION_INITIALIZER_ONSTACK, DECLARE_COMPLETION_ONSTACK, __WAIT_QUEUE_HEAD_INIT_ONSTACK and DECLARE_WAIT_QUEUE_HEAD_ONSTACK) so I guess workqueues should do the same thing. s/INIT_WORK_ON_STACK/INIT_WORK_ONSTACK/ s/INIT_DELAYED_WORK_ON_STACK/INIT_DELAYED_WORK_ONSTACK/ Cc: Peter Zijlstra Acked-by: Tejun Heo Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/workqueue.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/workqueue.c b/kernel/workqueue.c index e5ff2cbaadc2..90db1bd1a978 100644 --- a/kernel/workqueue.c +++ b/kernel/workqueue.c @@ -2064,7 +2064,7 @@ static void insert_wq_barrier(struct cpu_workqueue_struct *cwq, * checks and call back into the fixup functions where we * might deadlock. */ - INIT_WORK_ON_STACK(&barr->work, wq_barrier_func); + INIT_WORK_ONSTACK(&barr->work, wq_barrier_func); __set_bit(WORK_STRUCT_PENDING_BIT, work_data_bits(&barr->work)); init_completion(&barr->done); -- cgit v1.2.3 From 571428be550fbe37160596995e96ad398873fcbd Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Tue, 26 Oct 2010 14:22:43 -0700 Subject: kernel/user.c: add lock release annotation on free_user() free_user() releases uidhash_lock but was missing annotation. Add it. This removes following sparse warnings: include/linux/spinlock.h:339:9: warning: context imbalance in 'free_user' - unexpected unlock kernel/user.c:120:6: warning: context imbalance in 'free_uid' - wrong count at exit Signed-off-by: Namhyung Kim Cc: Ingo Molnar Cc: Dhaval Giani Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/user.c | 1 + 1 file changed, 1 insertion(+) (limited to 'kernel') diff --git a/kernel/user.c b/kernel/user.c index 7e72614b736d..2c7d8d5914b1 100644 --- a/kernel/user.c +++ b/kernel/user.c @@ -91,6 +91,7 @@ static struct user_struct *uid_hash_find(uid_t uid, struct hlist_head *hashent) * upon function exit. */ static void free_user(struct user_struct *up, unsigned long flags) + __releases(&uidhash_lock) { uid_hash_remove(up); spin_unlock_irqrestore(&uidhash_lock, flags); -- cgit v1.2.3 From 518de9b39e854542de59bfb8b9f61c8f7ecf808b Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Tue, 26 Oct 2010 14:22:44 -0700 Subject: fs: allow for more than 2^31 files Robin Holt tried to boot a 16TB system and found af_unix was overflowing a 32bit value : We were seeing a failure which prevented boot. The kernel was incapable of creating either a named pipe or unix domain socket. This comes down to a common kernel function called unix_create1() which does: atomic_inc(&unix_nr_socks); if (atomic_read(&unix_nr_socks) > 2 * get_max_files()) goto out; The function get_max_files() is a simple return of files_stat.max_files. files_stat.max_files is a signed integer and is computed in fs/file_table.c's files_init(). n = (mempages * (PAGE_SIZE / 1024)) / 10; files_stat.max_files = n; In our case, mempages (total_ram_pages) is approx 3,758,096,384 (0xe0000000). That leaves max_files at approximately 1,503,238,553. This causes 2 * get_max_files() to integer overflow. Fix is to let /proc/sys/fs/file-nr & /proc/sys/fs/file-max use long integers, and change af_unix to use an atomic_long_t instead of atomic_t. get_max_files() is changed to return an unsigned long. get_nr_files() is changed to return a long. unix_nr_socks is changed from atomic_t to atomic_long_t, while not strictly needed to address Robin problem. Before patch (on a 64bit kernel) : # echo 2147483648 >/proc/sys/fs/file-max # cat /proc/sys/fs/file-max -18446744071562067968 After patch: # echo 2147483648 >/proc/sys/fs/file-max # cat /proc/sys/fs/file-max 2147483648 # cat /proc/sys/fs/file-nr 704 0 2147483648 Reported-by: Robin Holt Signed-off-by: Eric Dumazet Acked-by: David Miller Reviewed-by: Robin Holt Tested-by: Robin Holt Cc: Al Viro Cc: Christoph Hellwig Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/sysctl.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/sysctl.c b/kernel/sysctl.c index 3a45c224770f..694b140852c2 100644 --- a/kernel/sysctl.c +++ b/kernel/sysctl.c @@ -1352,16 +1352,16 @@ static struct ctl_table fs_table[] = { { .procname = "file-nr", .data = &files_stat, - .maxlen = 3*sizeof(int), + .maxlen = sizeof(files_stat), .mode = 0444, .proc_handler = proc_nr_files, }, { .procname = "file-max", .data = &files_stat.max_files, - .maxlen = sizeof(int), + .maxlen = sizeof(files_stat.max_files), .mode = 0644, - .proc_handler = proc_dointvec, + .proc_handler = proc_doulongvec_minmax, }, { .procname = "nr_open", -- cgit v1.2.3 From ca51c5a76345b28c6f1b742f9f5f0a6fc9afd9ca Mon Sep 17 00:00:00 2001 From: Rakib Mullick Date: Tue, 26 Oct 2010 14:22:44 -0700 Subject: kernel/stop_machine.c: fix unused variable warning kernel/stop_machine.c: In function `cpu_stopper_thread': kernel/stop_machine.c:265: warning: unused variable `ksym_buf' ksym_buf[] is unused if WARN_ON() is a no-op. Signed-off-by: Rakib Mullick Cc: Ingo Molnar Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/stop_machine.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/stop_machine.c b/kernel/stop_machine.c index 090c28812ce1..3cb4a9a8ae1c 100644 --- a/kernel/stop_machine.c +++ b/kernel/stop_machine.c @@ -262,7 +262,7 @@ repeat: cpu_stop_fn_t fn = work->fn; void *arg = work->arg; struct cpu_stop_done *done = work->done; - char ksym_buf[KSYM_NAME_LEN]; + char ksym_buf[KSYM_NAME_LEN] __maybe_unused; __set_current_state(TASK_RUNNING); -- cgit v1.2.3 From 4ce6494dbd8909718840bb88d5a699ef6ce5c212 Mon Sep 17 00:00:00 2001 From: Akinobu Mita Date: Tue, 26 Oct 2010 14:22:45 -0700 Subject: stop_machine: convert cpu notifier to return encapsulate errno value In commit e6bde73b07edeb703d4c89c1daabc09c303de11f ("cpu-hotplug: return better errno on cpu hotplug failure"), the cpu notifier can return an encapsulated errno value. This converts the cpu notifier to return an encapsulated errno value for stop_machine(). Signed-off-by: Akinobu Mita Cc: Rusty Russell Cc: Tejun Heo Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/stop_machine.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/stop_machine.c b/kernel/stop_machine.c index 3cb4a9a8ae1c..2df820b03beb 100644 --- a/kernel/stop_machine.c +++ b/kernel/stop_machine.c @@ -304,7 +304,7 @@ static int __cpuinit cpu_stop_cpu_callback(struct notifier_block *nfb, p = kthread_create(cpu_stopper_thread, stopper, "migration/%d", cpu); if (IS_ERR(p)) - return NOTIFY_BAD; + return notifier_from_errno(PTR_ERR(p)); get_task_struct(p); kthread_bind(p, cpu); sched_set_stop_task(cpu, p); @@ -372,7 +372,7 @@ static int __init cpu_stop_init(void) /* start one for the boot cpu */ err = cpu_stop_cpu_callback(&cpu_stop_cpu_notifier, CPU_UP_PREPARE, bcpu); - BUG_ON(err == NOTIFY_BAD); + BUG_ON(err != NOTIFY_OK); cpu_stop_cpu_callback(&cpu_stop_cpu_notifier, CPU_ONLINE, bcpu); register_cpu_notifier(&cpu_stop_cpu_notifier); -- cgit v1.2.3 From 6c095efd82e8f6a98515426a733110f91cf0a709 Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Tue, 26 Oct 2010 14:22:47 -0700 Subject: printk: fixup declaration of kmsg_reasons Move redundant 'const' after '*' to make pointer itself const Signed-off-by: Namhyung Kim Cc: Ingo Molnar Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/printk.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/printk.c b/kernel/printk.c index 2531017795f6..d18933a92819 100644 --- a/kernel/printk.c +++ b/kernel/printk.c @@ -1511,7 +1511,7 @@ int kmsg_dump_unregister(struct kmsg_dumper *dumper) } EXPORT_SYMBOL_GPL(kmsg_dump_unregister); -static const char const *kmsg_reasons[] = { +static const char * const kmsg_reasons[] = { [KMSG_DUMP_OOPS] = "oops", [KMSG_DUMP_PANIC] = "panic", [KMSG_DUMP_KEXEC] = "kexec", -- cgit v1.2.3 From 8155c02a44a95562e1ae0999360eb31288d7195a Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Tue, 26 Oct 2010 14:22:47 -0700 Subject: printk: add lock context annotation acquire_console_semaphore_for_printk() releases logbuf_lock but was missing proper annotation. Add it. Signed-off-by: Namhyung Kim Cc: Ingo Molnar Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/printk.c | 1 + 1 file changed, 1 insertion(+) (limited to 'kernel') diff --git a/kernel/printk.c b/kernel/printk.c index d18933a92819..6ff26151f4ff 100644 --- a/kernel/printk.c +++ b/kernel/printk.c @@ -647,6 +647,7 @@ static inline int can_use_console(unsigned int cpu) * released but interrupts still disabled. */ static int acquire_console_semaphore_for_printk(unsigned int cpu) + __releases(&logbuf_lock) { int retval = 0; -- cgit v1.2.3 From 674dff6507d3f9b110219ea125cf5e1213c9acef Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Tue, 26 Oct 2010 14:22:48 -0700 Subject: printk: change type of 'boot_delay' to int * get_option() takes its 2nd arg as int * so passing boot_delay to it caused following warnings from sparse: kernel/printk.c:223:27: warning: incorrect type in argument 2 (different signedness) kernel/printk.c:223:27: expected int *pint kernel/printk.c:223:27: got unsigned int static [toplevel] * Since boot_delay can't grow more than 10,000 changing it to 'int *' will not produce any problem. Signed-off-by: Namhyung Kim Cc: Ingo Molnar Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/printk.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/printk.c b/kernel/printk.c index 6ff26151f4ff..b2ebaee8c377 100644 --- a/kernel/printk.c +++ b/kernel/printk.c @@ -210,7 +210,7 @@ __setup("log_buf_len=", log_buf_len_setup); #ifdef CONFIG_BOOT_PRINTK_DELAY -static unsigned int boot_delay; /* msecs delay after each printk during bootup */ +static int boot_delay; /* msecs delay after each printk during bootup */ static unsigned long long loops_per_msec; /* based on boot_delay */ static int __init boot_delay_setup(char *str) -- cgit v1.2.3 From f5d87d851d76a390d0fab2f77bd1d563d69ee586 Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Tue, 26 Oct 2010 14:22:49 -0700 Subject: printk: declare printk_ratelimit_state in ratelimit.h Adding declaration of printk_ratelimit_state in ratelimit.h removes potential build breakage and following sparse warning: kernel/printk.c:1426:1: warning: symbol 'printk_ratelimit_state' was not declared. Should it be static? [akpm@linux-foundation.org: remove unneeded ifdef] Signed-off-by: Namhyung Kim Cc: Ingo Molnar Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/sysctl.c | 2 -- 1 file changed, 2 deletions(-) (limited to 'kernel') diff --git a/kernel/sysctl.c b/kernel/sysctl.c index 694b140852c2..48d9d689498f 100644 --- a/kernel/sysctl.c +++ b/kernel/sysctl.c @@ -161,8 +161,6 @@ extern int no_unaligned_warning; extern int unaligned_dump_stack; #endif -extern struct ratelimit_state printk_ratelimit_state; - #ifdef CONFIG_PROC_SYSCTL static int proc_do_cad_pid(struct ctl_table *table, int write, void __user *buffer, size_t *lenp, loff_t *ppos); -- cgit v1.2.3 From ee2f154a598e96df2ebb01648a7699373bc085c7 Mon Sep 17 00:00:00 2001 From: Randy Dunlap Date: Tue, 26 Oct 2010 14:17:25 -0700 Subject: docbook: add more wait/wake/completion to device-drivers docbook Add more wait, wake, and completion interfaces to the device-drivers docbook. Fix kernel-doc notation in the added files. Signed-off-by: Randy Dunlap Signed-off-by: Linus Torvalds --- kernel/wait.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/wait.c b/kernel/wait.c index c4bd3d825f35..b0310eb6cc1e 100644 --- a/kernel/wait.c +++ b/kernel/wait.c @@ -92,7 +92,7 @@ prepare_to_wait_exclusive(wait_queue_head_t *q, wait_queue_t *wait, int state) } EXPORT_SYMBOL(prepare_to_wait_exclusive); -/* +/** * finish_wait - clean up after waiting in a queue * @q: waitqueue waited on * @wait: wait descriptor @@ -127,11 +127,11 @@ void finish_wait(wait_queue_head_t *q, wait_queue_t *wait) } EXPORT_SYMBOL(finish_wait); -/* +/** * abort_exclusive_wait - abort exclusive waiting in a queue * @q: waitqueue waited on * @wait: wait descriptor - * @state: runstate of the waiter to be woken + * @mode: runstate of the waiter to be woken * @key: key to identify a wait bit queue or %NULL * * Sets current thread back to running state and removes -- cgit v1.2.3 From abbce906d05ec37289cd0c3b4e35b2db26eab19b Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Micha=C5=82=20Miros=C5=82aw?= Date: Mon, 20 Sep 2010 01:58:08 +0200 Subject: (trivial) Fix compiler warning in kernel/modules.c MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Building with CONFIG_KALLSYMS=n gives following warning: /mnt/src/linux-git/kernel/module.c: In function ‘post_relocation’: /mnt/src/linux-git/kernel/module.c:2534:2: warning: passing argument 2 of ‘add_kallsyms’ discards qualifiers from pointer target type /mnt/src/linux-git/kernel/module.c:2038:13: note: expected ‘struct load_info *’ but argument is of type ‘const struct load_info *’ Signed-off-by: MichaÅ‚ MirosÅ‚aw Signed-off-by: Rusty Russell --- kernel/module.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/module.c b/kernel/module.c index 2df46301a7a4..437a74a7524a 100644 --- a/kernel/module.c +++ b/kernel/module.c @@ -2037,7 +2037,7 @@ static inline void layout_symtab(struct module *mod, struct load_info *info) { } -static void add_kallsyms(struct module *mod, struct load_info *info) +static void add_kallsyms(struct module *mod, const struct load_info *info) { } #endif /* CONFIG_KALLSYMS */ -- cgit v1.2.3 From 3a5f65df5a0fcbaa35e5417c0420d691fee4ac56 Mon Sep 17 00:00:00 2001 From: David Howells Date: Wed, 27 Oct 2010 17:28:36 +0100 Subject: Typedef SMP call function pointer Typedef the pointer to the function to be called by smp_call_function() and friends: typedef void (*smp_call_func_t)(void *info); as it is used in a fair number of places. Signed-off-by: David Howells cc: linux-arch@vger.kernel.org --- kernel/smp.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) (limited to 'kernel') diff --git a/kernel/smp.c b/kernel/smp.c index ed6aacfcb7ef..12ed8b013e2d 100644 --- a/kernel/smp.c +++ b/kernel/smp.c @@ -267,7 +267,7 @@ static DEFINE_PER_CPU_SHARED_ALIGNED(struct call_single_data, csd_data); * * Returns 0 on success, else a negative status code. */ -int smp_call_function_single(int cpu, void (*func) (void *info), void *info, +int smp_call_function_single(int cpu, smp_call_func_t func, void *info, int wait) { struct call_single_data d = { @@ -336,7 +336,7 @@ EXPORT_SYMBOL(smp_call_function_single); * 3) any other online cpu in @mask */ int smp_call_function_any(const struct cpumask *mask, - void (*func)(void *info), void *info, int wait) + smp_call_func_t func, void *info, int wait) { unsigned int cpu; const struct cpumask *nodemask; @@ -416,7 +416,7 @@ void __smp_call_function_single(int cpu, struct call_single_data *data, * must be disabled when calling this function. */ void smp_call_function_many(const struct cpumask *mask, - void (*func)(void *), void *info, bool wait) + smp_call_func_t func, void *info, bool wait) { struct call_function_data *data; unsigned long flags; @@ -500,7 +500,7 @@ EXPORT_SYMBOL(smp_call_function_many); * You must not call this function with disabled interrupts or from a * hardware interrupt handler or from a bottom half handler. */ -int smp_call_function(void (*func)(void *), void *info, int wait) +int smp_call_function(smp_call_func_t func, void *info, int wait) { preempt_disable(); smp_call_function_many(cpu_online_mask, func, info, wait); -- cgit v1.2.3 From d5de4ddb1bc430289bede76c0d87cabee93f749a Mon Sep 17 00:00:00 2001 From: Tomasz Buchert Date: Wed, 27 Oct 2010 15:33:32 -0700 Subject: cgroup_freezer: unnecessary test in cgroup_freezing_or_frozen() The root freezer_state is always CGROUP_THAWED so we can remove the special case from the code. The test itself can be handy and is extracted to static function. Signed-off-by: Tomasz Buchert Cc: Matt Helsley Cc: Paul Menage Cc: Li Zefan Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/cgroup_freezer.c | 19 +++++++++---------- 1 file changed, 9 insertions(+), 10 deletions(-) (limited to 'kernel') diff --git a/kernel/cgroup_freezer.c b/kernel/cgroup_freezer.c index ce71ed53e88f..321e2a007d25 100644 --- a/kernel/cgroup_freezer.c +++ b/kernel/cgroup_freezer.c @@ -48,20 +48,19 @@ static inline struct freezer *task_freezer(struct task_struct *task) struct freezer, css); } -int cgroup_freezing_or_frozen(struct task_struct *task) +static inline int __cgroup_freezing_or_frozen(struct task_struct *task) { - struct freezer *freezer; - enum freezer_state state; + enum freezer_state state = task_freezer(task)->state; + return (state == CGROUP_FREEZING) || (state == CGROUP_FROZEN); +} +int cgroup_freezing_or_frozen(struct task_struct *task) +{ + int result; task_lock(task); - freezer = task_freezer(task); - if (!freezer->css.cgroup->parent) - state = CGROUP_THAWED; /* root cgroup can't be frozen */ - else - state = freezer->state; + result = __cgroup_freezing_or_frozen(task); task_unlock(task); - - return (state == CGROUP_FREEZING) || (state == CGROUP_FROZEN); + return result; } /* -- cgit v1.2.3 From 0bdba580ab052a21e3eda2764ed22d9ee962392b Mon Sep 17 00:00:00 2001 From: Tomasz Buchert Date: Wed, 27 Oct 2010 15:33:33 -0700 Subject: cgroup_freezer: fix can_attach() to prohibit moving from/to freezing/frozen cgroups It is possible to move a task from its cgroup even if this group is 'FREEZING'. This results in a nasty bug - the moved task will become frozen OUTSIDE its original cgroup and will remain in a permanent 'D' state. This patch allows to migrate the task only between THAWED cgroups. This behavior was observed and easily reproduced on a single core laptop. Notice that reproducibility depends highly on the machine used. Program and instructions how to reproduce the bug can be fetched from: http://pentium.hopto.org/~thinred/repos/linux-misc/freezer_bug.c Signed-off-by: Tomasz Buchert Cc: Matt Helsley Cc: Paul Menage Cc: Li Zefan Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/cgroup_freezer.c | 15 ++++++++------- 1 file changed, 8 insertions(+), 7 deletions(-) (limited to 'kernel') diff --git a/kernel/cgroup_freezer.c b/kernel/cgroup_freezer.c index 321e2a007d25..c287627bed3d 100644 --- a/kernel/cgroup_freezer.c +++ b/kernel/cgroup_freezer.c @@ -173,24 +173,25 @@ static int freezer_can_attach(struct cgroup_subsys *ss, /* * Anything frozen can't move or be moved to/from. - * - * Since orig_freezer->state == FROZEN means that @task has been - * frozen, so it's sufficient to check the latter condition. */ - if (is_task_frozen_enough(task)) + freezer = cgroup_freezer(new_cgroup); + if (freezer->state != CGROUP_THAWED) return -EBUSY; - freezer = cgroup_freezer(new_cgroup); - if (freezer->state == CGROUP_FROZEN) + rcu_read_lock(); + if (__cgroup_freezing_or_frozen(task)) { + rcu_read_unlock(); return -EBUSY; + } + rcu_read_unlock(); if (threadgroup) { struct task_struct *c; rcu_read_lock(); list_for_each_entry_rcu(c, &task->thread_group, thread_group) { - if (is_task_frozen_enough(c)) { + if (__cgroup_freezing_or_frozen(c)) { rcu_read_unlock(); return -EBUSY; } -- cgit v1.2.3 From 2d3cbf8bc852ac1bc3d098186143c5973f87b753 Mon Sep 17 00:00:00 2001 From: Tomasz Buchert Date: Wed, 27 Oct 2010 15:33:34 -0700 Subject: cgroup_freezer: update_freezer_state() does incorrect state transitions There are 4 state transitions possible for a freezer. Only FREEZING -> FROZEN transaction is done lazily. This patch allows update_freezer_state only to perform this transaction and renames the function to update_if_frozen. Moreover is_task_frozen_enough function is removed and its every occurence is replaced with frozen(). Therefore for a group to become FROZEN every task must be frozen. The previous version could trigger a following bug: When cgroup is in the process of freezing (but none of its tasks are frozen yet), update_freezer_state() (called from freezer_read or freezer_write) would incorrectly report that a group is 'THAWED' (because nfrozen = 0), allowing the transaction FREEZING -> THAWED without writing anything to 'freezer.state'. This is incorrect according to the documentation. This could result in a 'THAWED' cgroup with frozen tasks inside. A code to reproduce this bug is available here: http://pentium.hopto.org/~thinred/repos/linux-misc/freezer_bug2.c [akpm@linux-foundation.org: coding-style fixes] Signed-off-by: Tomasz Buchert Cc: Matt Helsley Cc: Paul Menage Cc: Li Zefan Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/cgroup_freezer.c | 38 +++++++++++++++----------------------- 1 file changed, 15 insertions(+), 23 deletions(-) (limited to 'kernel') diff --git a/kernel/cgroup_freezer.c b/kernel/cgroup_freezer.c index c287627bed3d..e7bebb7c6c38 100644 --- a/kernel/cgroup_freezer.c +++ b/kernel/cgroup_freezer.c @@ -153,13 +153,6 @@ static void freezer_destroy(struct cgroup_subsys *ss, kfree(cgroup_freezer(cgroup)); } -/* Task is frozen or will freeze immediately when next it gets woken */ -static bool is_task_frozen_enough(struct task_struct *task) -{ - return frozen(task) || - (task_is_stopped_or_traced(task) && freezing(task)); -} - /* * The call to cgroup_lock() in the freezer.state write method prevents * a write to that file racing against an attach, and hence the @@ -236,31 +229,30 @@ static void freezer_fork(struct cgroup_subsys *ss, struct task_struct *task) /* * caller must hold freezer->lock */ -static void update_freezer_state(struct cgroup *cgroup, +static void update_if_frozen(struct cgroup *cgroup, struct freezer *freezer) { struct cgroup_iter it; struct task_struct *task; unsigned int nfrozen = 0, ntotal = 0; + enum freezer_state old_state = freezer->state; cgroup_iter_start(cgroup, &it); while ((task = cgroup_iter_next(cgroup, &it))) { ntotal++; - if (is_task_frozen_enough(task)) + if (frozen(task)) nfrozen++; } - /* - * Transition to FROZEN when no new tasks can be added ensures - * that we never exist in the FROZEN state while there are unfrozen - * tasks. - */ - if (nfrozen == ntotal) - freezer->state = CGROUP_FROZEN; - else if (nfrozen > 0) - freezer->state = CGROUP_FREEZING; - else - freezer->state = CGROUP_THAWED; + if (old_state == CGROUP_THAWED) { + BUG_ON(nfrozen > 0); + } else if (old_state == CGROUP_FREEZING) { + if (nfrozen == ntotal) + freezer->state = CGROUP_FROZEN; + } else { /* old_state == CGROUP_FROZEN */ + BUG_ON(nfrozen != ntotal); + } + cgroup_iter_end(cgroup, &it); } @@ -279,7 +271,7 @@ static int freezer_read(struct cgroup *cgroup, struct cftype *cft, if (state == CGROUP_FREEZING) { /* We change from FREEZING to FROZEN lazily if the cgroup was * only partially frozen when we exitted write. */ - update_freezer_state(cgroup, freezer); + update_if_frozen(cgroup, freezer); state = freezer->state; } spin_unlock_irq(&freezer->lock); @@ -301,7 +293,7 @@ static int try_to_freeze_cgroup(struct cgroup *cgroup, struct freezer *freezer) while ((task = cgroup_iter_next(cgroup, &it))) { if (!freeze_task(task, true)) continue; - if (is_task_frozen_enough(task)) + if (frozen(task)) continue; if (!freezing(task) && !freezer_should_skip(task)) num_cant_freeze_now++; @@ -335,7 +327,7 @@ static int freezer_change_state(struct cgroup *cgroup, spin_lock_irq(&freezer->lock); - update_freezer_state(cgroup, freezer); + update_if_frozen(cgroup, freezer); if (goal_state == freezer->state) goto out; -- cgit v1.2.3 From 97978e6d1f2da0073416870410459694fbdbfd9b Mon Sep 17 00:00:00 2001 From: Daniel Lezcano Date: Wed, 27 Oct 2010 15:33:35 -0700 Subject: cgroup: add clone_children control file The ns_cgroup is a control group interacting with the namespaces. When a new namespace is created, a corresponding cgroup is automatically created too. The cgroup name is the pid of the process who did 'unshare' or the child of 'clone'. This cgroup is tied with the namespace because it prevents a process to escape the control group and use the post_clone callback, so the child cgroup inherits the values of the parent cgroup. Unfortunately, the more we use this cgroup and the more we are facing problems with it: (1) when a process unshares, the cgroup name may conflict with a previous cgroup with the same pid, so unshare or clone return -EEXIST (2) the cgroup creation is out of control because there may have an application creating several namespaces where the system will automatically create several cgroups in his back and let them on the cgroupfs (eg. a vrf based on the network namespace). (3) the mix of (1) and (2) force an administrator to regularly check and clean these cgroups. This patchset removes the ns_cgroup by adding a new flag to the cgroup and the cgroupfs mount option. It enables the copy of the parent cgroup when a child cgroup is created. We can then safely remove the ns_cgroup as this flag brings a compatibility. We have now to manually create and add the task to a cgroup, which is consistent with the cgroup framework. This patch: Sent as an answer to a previous thread around the ns_cgroup. https://lists.linux-foundation.org/pipermail/containers/2009-June/018627.html It adds a control file 'clone_children' for a cgroup. This control file is a boolean specifying if the child cgroup should be a clone of the parent cgroup or not. The default value is 'false'. This flag makes the child cgroup to call the post_clone callback of all the subsystem, if it is available. At present, the cpuset is the only one which had implemented the post_clone callback. The option can be set at mount time by specifying the 'clone_children' mount option. Signed-off-by: Daniel Lezcano Signed-off-by: Serge E. Hallyn Cc: Eric W. Biederman Acked-by: Paul Menage Reviewed-by: Li Zefan Cc: Jamal Hadi Salim Cc: Matt Helsley Acked-by: Balbir Singh Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/cgroup.c | 39 +++++++++++++++++++++++++++++++++++++++ 1 file changed, 39 insertions(+) (limited to 'kernel') diff --git a/kernel/cgroup.c b/kernel/cgroup.c index 9270d532ec3c..4b218a46ddd3 100644 --- a/kernel/cgroup.c +++ b/kernel/cgroup.c @@ -243,6 +243,11 @@ static int notify_on_release(const struct cgroup *cgrp) return test_bit(CGRP_NOTIFY_ON_RELEASE, &cgrp->flags); } +static int clone_children(const struct cgroup *cgrp) +{ + return test_bit(CGRP_CLONE_CHILDREN, &cgrp->flags); +} + /* * for_each_subsys() allows you to iterate on each subsystem attached to * an active hierarchy @@ -1040,6 +1045,8 @@ static int cgroup_show_options(struct seq_file *seq, struct vfsmount *vfs) seq_puts(seq, ",noprefix"); if (strlen(root->release_agent_path)) seq_printf(seq, ",release_agent=%s", root->release_agent_path); + if (clone_children(&root->top_cgroup)) + seq_puts(seq, ",clone_children"); if (strlen(root->name)) seq_printf(seq, ",name=%s", root->name); mutex_unlock(&cgroup_mutex); @@ -1050,6 +1057,7 @@ struct cgroup_sb_opts { unsigned long subsys_bits; unsigned long flags; char *release_agent; + bool clone_children; char *name; /* User explicitly requested empty subsystem */ bool none; @@ -1097,6 +1105,8 @@ static int parse_cgroupfs_options(char *data, struct cgroup_sb_opts *opts) opts->none = true; } else if (!strcmp(token, "noprefix")) { set_bit(ROOT_NOPREFIX, &opts->flags); + } else if (!strcmp(token, "clone_children")) { + opts->clone_children = true; } else if (!strncmp(token, "release_agent=", 14)) { /* Specifying two release agents is forbidden */ if (opts->release_agent) @@ -1355,6 +1365,8 @@ static struct cgroupfs_root *cgroup_root_from_opts(struct cgroup_sb_opts *opts) strcpy(root->release_agent_path, opts->release_agent); if (opts->name) strcpy(root->name, opts->name); + if (opts->clone_children) + set_bit(CGRP_CLONE_CHILDREN, &root->top_cgroup.flags); return root; } @@ -3173,6 +3185,23 @@ fail: return ret; } +static u64 cgroup_clone_children_read(struct cgroup *cgrp, + struct cftype *cft) +{ + return clone_children(cgrp); +} + +static int cgroup_clone_children_write(struct cgroup *cgrp, + struct cftype *cft, + u64 val) +{ + if (val) + set_bit(CGRP_CLONE_CHILDREN, &cgrp->flags); + else + clear_bit(CGRP_CLONE_CHILDREN, &cgrp->flags); + return 0; +} + /* * for the common functions, 'private' gives the type of file */ @@ -3203,6 +3232,11 @@ static struct cftype files[] = { .write_string = cgroup_write_event_control, .mode = S_IWUGO, }, + { + .name = "cgroup.clone_children", + .read_u64 = cgroup_clone_children_read, + .write_u64 = cgroup_clone_children_write, + }, }; static struct cftype cft_release_agent = { @@ -3332,6 +3366,9 @@ static long cgroup_create(struct cgroup *parent, struct dentry *dentry, if (notify_on_release(parent)) set_bit(CGRP_NOTIFY_ON_RELEASE, &cgrp->flags); + if (clone_children(parent)) + set_bit(CGRP_CLONE_CHILDREN, &cgrp->flags); + for_each_subsys(root, ss) { struct cgroup_subsys_state *css = ss->create(ss, cgrp); @@ -3346,6 +3383,8 @@ static long cgroup_create(struct cgroup *parent, struct dentry *dentry, goto err_destroy; } /* At error, ->destroy() callback has to free assigned ID. */ + if (clone_children(parent) && ss->post_clone) + ss->post_clone(ss, cgrp); } cgroup_lock_hierarchy(root); -- cgit v1.2.3 From 32a8cf235e2f192eb002755076994525cdbaa35a Mon Sep 17 00:00:00 2001 From: Daniel Lezcano Date: Wed, 27 Oct 2010 15:33:37 -0700 Subject: cgroup: make the mount options parsing more accurate Current behavior: ================= (1) When we mount a cgroup, we can specify the 'all' option which means to enable all the cgroup subsystems. This is the default option when no option is specified. (2) If we want to mount a cgroup with a subset of the supported cgroup subsystems, we have to specify a subsystems name list for the mount option. (3) If we specify another option like 'noprefix' or 'release_agent', the actual code wants the 'all' or a subsystem name option specified also. Not critical but a bit not friendly as we should assume (1) in this case. (4) Logically, the 'all' option is mutually exclusive with a subsystem name, but this is not detected. In other words: succeed : mount -t cgroup -o all,freezer cgroup /cgroup => is it 'all' or 'freezer' ? fails : mount -t cgroup -o noprefix cgroup /cgroup => succeed if we do '-o noprefix,all' The following patches consolidate a bit the mount options check. New behavior: ============= (1) untouched (2) untouched (3) the 'all' option will be by default when specifying other than a subsystem name option (4) raises an error In other words: fails : mount -t cgroup -o all,freezer cgroup /cgroup succeed : mount -t cgroup -o noprefix cgroup /cgroup For the sake of lisibility, the if ... then ... else ... if ... indentation when parsing the options has been changed to: if ... then ... continue fi Signed-off-by: Daniel Lezcano Signed-off-by: Serge E. Hallyn Reviewed-by: Li Zefan Reviewed-by: Paul Menage Cc: Eric W. Biederman Cc: Jamal Hadi Salim Cc: Matt Helsley Cc: Balbir Singh Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/cgroup.c | 90 ++++++++++++++++++++++++++++++++++++++------------------- 1 file changed, 60 insertions(+), 30 deletions(-) (limited to 'kernel') diff --git a/kernel/cgroup.c b/kernel/cgroup.c index 4b218a46ddd3..3e6517e51fd3 100644 --- a/kernel/cgroup.c +++ b/kernel/cgroup.c @@ -1074,7 +1074,8 @@ struct cgroup_sb_opts { */ static int parse_cgroupfs_options(char *data, struct cgroup_sb_opts *opts) { - char *token, *o = data ?: "all"; + char *token, *o = data; + bool all_ss = false, one_ss = false; unsigned long mask = (unsigned long)-1; int i; bool module_pin_failed = false; @@ -1090,24 +1091,27 @@ static int parse_cgroupfs_options(char *data, struct cgroup_sb_opts *opts) while ((token = strsep(&o, ",")) != NULL) { if (!*token) return -EINVAL; - if (!strcmp(token, "all")) { - /* Add all non-disabled subsystems */ - opts->subsys_bits = 0; - for (i = 0; i < CGROUP_SUBSYS_COUNT; i++) { - struct cgroup_subsys *ss = subsys[i]; - if (ss == NULL) - continue; - if (!ss->disabled) - opts->subsys_bits |= 1ul << i; - } - } else if (!strcmp(token, "none")) { + if (!strcmp(token, "none")) { /* Explicitly have no subsystems */ opts->none = true; - } else if (!strcmp(token, "noprefix")) { + continue; + } + if (!strcmp(token, "all")) { + /* Mutually exclusive option 'all' + subsystem name */ + if (one_ss) + return -EINVAL; + all_ss = true; + continue; + } + if (!strcmp(token, "noprefix")) { set_bit(ROOT_NOPREFIX, &opts->flags); - } else if (!strcmp(token, "clone_children")) { + continue; + } + if (!strcmp(token, "clone_children")) { opts->clone_children = true; - } else if (!strncmp(token, "release_agent=", 14)) { + continue; + } + if (!strncmp(token, "release_agent=", 14)) { /* Specifying two release agents is forbidden */ if (opts->release_agent) return -EINVAL; @@ -1115,7 +1119,9 @@ static int parse_cgroupfs_options(char *data, struct cgroup_sb_opts *opts) kstrndup(token + 14, PATH_MAX - 1, GFP_KERNEL); if (!opts->release_agent) return -ENOMEM; - } else if (!strncmp(token, "name=", 5)) { + continue; + } + if (!strncmp(token, "name=", 5)) { const char *name = token + 5; /* Can't specify an empty name */ if (!strlen(name)) @@ -1137,20 +1143,44 @@ static int parse_cgroupfs_options(char *data, struct cgroup_sb_opts *opts) GFP_KERNEL); if (!opts->name) return -ENOMEM; - } else { - struct cgroup_subsys *ss; - for (i = 0; i < CGROUP_SUBSYS_COUNT; i++) { - ss = subsys[i]; - if (ss == NULL) - continue; - if (!strcmp(token, ss->name)) { - if (!ss->disabled) - set_bit(i, &opts->subsys_bits); - break; - } - } - if (i == CGROUP_SUBSYS_COUNT) - return -ENOENT; + + continue; + } + + for (i = 0; i < CGROUP_SUBSYS_COUNT; i++) { + struct cgroup_subsys *ss = subsys[i]; + if (ss == NULL) + continue; + if (strcmp(token, ss->name)) + continue; + if (ss->disabled) + continue; + + /* Mutually exclusive option 'all' + subsystem name */ + if (all_ss) + return -EINVAL; + set_bit(i, &opts->subsys_bits); + one_ss = true; + + break; + } + if (i == CGROUP_SUBSYS_COUNT) + return -ENOENT; + } + + /* + * If the 'all' option was specified select all the subsystems, + * otherwise 'all, 'none' and a subsystem name options were not + * specified, let's default to 'all' + */ + if (all_ss || (!all_ss && !one_ss && !opts->none)) { + for (i = 0; i < CGROUP_SUBSYS_COUNT; i++) { + struct cgroup_subsys *ss = subsys[i]; + if (ss == NULL) + continue; + if (ss->disabled) + continue; + set_bit(i, &opts->subsys_bits); } } -- cgit v1.2.3 From f4a2589feaef0a9b737a3e582b37ee96695bb25f Mon Sep 17 00:00:00 2001 From: Evgeny Kuznetsov Date: Wed, 27 Oct 2010 15:33:37 -0700 Subject: cgroups: add check for strcpy destination string overflow Function "strcpy" is used without check for maximum allowed source string length and could cause destination string overflow. Check for string length is added before using "strcpy". Function now is return error if source string length is more than a maximum. akpm: presently considered NotABug, but add the check for general future-safeness and robustness. Signed-off-by: Evgeny Kuznetsov Acked-by: Paul Menage Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/cgroup.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'kernel') diff --git a/kernel/cgroup.c b/kernel/cgroup.c index 3e6517e51fd3..5cf366965d0c 100644 --- a/kernel/cgroup.c +++ b/kernel/cgroup.c @@ -1922,6 +1922,8 @@ static int cgroup_release_agent_write(struct cgroup *cgrp, struct cftype *cft, const char *buffer) { BUILD_BUG_ON(sizeof(cgrp->root->release_agent_path) < PATH_MAX); + if (strlen(buffer) >= PATH_MAX) + return -EINVAL; if (!cgroup_lock_live_group(cgrp)) return -ENODEV; strcpy(cgrp->root->release_agent_path, buffer); -- cgit v1.2.3 From 45531757b45cae0ce64c5aff08c2534d5a0fa3e7 Mon Sep 17 00:00:00 2001 From: Daniel Lezcano Date: Wed, 27 Oct 2010 15:33:38 -0700 Subject: cgroup: notify ns_cgroup deprecated The ns_cgroup will be removed very soon. Let's warn, for this version, ns_cgroup is deprecated. Make ns_cgroup and clone_children exclusive. If the clone_children is set and the ns_cgroup is mounted, let's fail with EINVAL when the ns_cgroup subsys is created (a printk will help the user to understand why the creation fails). Update the feature remove schedule file with the deprecated ns_cgroup. Signed-off-by: Daniel Lezcano Acked-by: Paul Menage Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/ns_cgroup.c | 8 ++++++++ 1 file changed, 8 insertions(+) (limited to 'kernel') diff --git a/kernel/ns_cgroup.c b/kernel/ns_cgroup.c index 2a5dfec8efe0..2c98ad94ba0e 100644 --- a/kernel/ns_cgroup.c +++ b/kernel/ns_cgroup.c @@ -85,6 +85,14 @@ static struct cgroup_subsys_state *ns_create(struct cgroup_subsys *ss, return ERR_PTR(-EPERM); if (!cgroup_is_descendant(cgroup, current)) return ERR_PTR(-EPERM); + if (test_bit(CGRP_CLONE_CHILDREN, &cgroup->flags)) { + printk("ns_cgroup can't be created with parent " + "'clone_children' set.\n"); + return ERR_PTR(-EINVAL); + } + + printk_once("ns_cgroup deprecated: consider using the " + "'clone_children' flag without the ns_cgroup.\n"); ns_cgroup = kzalloc(sizeof(*ns_cgroup), GFP_KERNEL); if (!ns_cgroup) -- cgit v1.2.3 From c4b5ed250eebf854d40f27b43362c80f115cb57a Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Wed, 27 Oct 2010 15:33:44 -0700 Subject: ptrace: annotate lock context change on exit_ptrace() exit_ptrace() releases and regrabs tasklist_lock but was missing proper annotation. Add it. Signed-off-by: Namhyung Kim Acked-by: Roland McGrath Cc: Ingo Molnar Cc: Oleg Nesterov Signed-off-by: Linus Torvalds --- kernel/ptrace.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'kernel') diff --git a/kernel/ptrace.c b/kernel/ptrace.c index f34d798ef4a2..4afd9b86cc0b 100644 --- a/kernel/ptrace.c +++ b/kernel/ptrace.c @@ -329,6 +329,8 @@ int ptrace_detach(struct task_struct *child, unsigned int data) * and reacquire the lock. */ void exit_ptrace(struct task_struct *tracer) + __releases(&tasklist_lock) + __acquires(&tasklist_lock) { struct task_struct *p, *n; LIST_HEAD(ptrace_dead); -- cgit v1.2.3 From 4abf986960ecda6a87fc2f795aacf888a2f0127e Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Wed, 27 Oct 2010 15:33:45 -0700 Subject: ptrace: change signature of sys_ptrace() and friends Since userspace API of ptrace syscall defines @addr and @data as void pointers, it would be more appropriate to define them as unsigned long in kernel. Therefore related functions are changed also. 'unsigned long' is typically used in other places in kernel as an opaque data type and that using this helps cleaning up a lot of warnings from sparse. Suggested-by: Arnd Bergmann Signed-off-by: Namhyung Kim Acked-by: Arnd Bergmann Acked-by: Roland McGrath Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/ptrace.c | 16 ++++++++++------ 1 file changed, 10 insertions(+), 6 deletions(-) (limited to 'kernel') diff --git a/kernel/ptrace.c b/kernel/ptrace.c index 4afd9b86cc0b..06981a8b271b 100644 --- a/kernel/ptrace.c +++ b/kernel/ptrace.c @@ -404,7 +404,7 @@ int ptrace_writedata(struct task_struct *tsk, char __user *src, unsigned long ds return copied; } -static int ptrace_setoptions(struct task_struct *child, long data) +static int ptrace_setoptions(struct task_struct *child, unsigned long data) { child->ptrace &= ~PT_TRACE_MASK; @@ -483,7 +483,8 @@ static int ptrace_setsiginfo(struct task_struct *child, const siginfo_t *info) #define is_sysemu_singlestep(request) 0 #endif -static int ptrace_resume(struct task_struct *child, long request, long data) +static int ptrace_resume(struct task_struct *child, long request, + unsigned long data) { if (!valid_signal(data)) return -EIO; @@ -560,7 +561,7 @@ static int ptrace_regset(struct task_struct *task, int req, unsigned int type, #endif int ptrace_request(struct task_struct *child, long request, - long addr, long data) + unsigned long addr, unsigned long data) { int ret = -EIO; siginfo_t siginfo; @@ -693,7 +694,8 @@ static struct task_struct *ptrace_get_task_struct(pid_t pid) #define arch_ptrace_attach(child) do { } while (0) #endif -SYSCALL_DEFINE4(ptrace, long, request, long, pid, long, addr, long, data) +SYSCALL_DEFINE4(ptrace, long, request, long, pid, unsigned long, addr, + unsigned long, data) { struct task_struct *child; long ret; @@ -734,7 +736,8 @@ SYSCALL_DEFINE4(ptrace, long, request, long, pid, long, addr, long, data) return ret; } -int generic_ptrace_peekdata(struct task_struct *tsk, long addr, long data) +int generic_ptrace_peekdata(struct task_struct *tsk, unsigned long addr, + unsigned long data) { unsigned long tmp; int copied; @@ -745,7 +748,8 @@ int generic_ptrace_peekdata(struct task_struct *tsk, long addr, long data) return put_user(tmp, (unsigned long __user *)data); } -int generic_ptrace_pokedata(struct task_struct *tsk, long addr, long data) +int generic_ptrace_pokedata(struct task_struct *tsk, unsigned long addr, + unsigned long data) { int copied; -- cgit v1.2.3 From 9fed81dc40f5a1ac2783bcc78d4029873be72894 Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Wed, 27 Oct 2010 15:33:46 -0700 Subject: ptrace: cleanup ptrace_request() Use new 'datavp' and 'datalp' variables to remove unnecesary castings. Signed-off-by: Namhyung Kim Acked-by: Roland McGrath Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/ptrace.c | 14 +++++++------- 1 file changed, 7 insertions(+), 7 deletions(-) (limited to 'kernel') diff --git a/kernel/ptrace.c b/kernel/ptrace.c index 06981a8b271b..ea7ce0215cd1 100644 --- a/kernel/ptrace.c +++ b/kernel/ptrace.c @@ -565,6 +565,8 @@ int ptrace_request(struct task_struct *child, long request, { int ret = -EIO; siginfo_t siginfo; + void __user *datavp = (void __user *) data; + unsigned long __user *datalp = datavp; switch (request) { case PTRACE_PEEKTEXT: @@ -581,19 +583,17 @@ int ptrace_request(struct task_struct *child, long request, ret = ptrace_setoptions(child, data); break; case PTRACE_GETEVENTMSG: - ret = put_user(child->ptrace_message, (unsigned long __user *) data); + ret = put_user(child->ptrace_message, datalp); break; case PTRACE_GETSIGINFO: ret = ptrace_getsiginfo(child, &siginfo); if (!ret) - ret = copy_siginfo_to_user((siginfo_t __user *) data, - &siginfo); + ret = copy_siginfo_to_user(datavp, &siginfo); break; case PTRACE_SETSIGINFO: - if (copy_from_user(&siginfo, (siginfo_t __user *) data, - sizeof siginfo)) + if (copy_from_user(&siginfo, datavp, sizeof siginfo)) ret = -EFAULT; else ret = ptrace_setsiginfo(child, &siginfo); @@ -624,7 +624,7 @@ int ptrace_request(struct task_struct *child, long request, } mmput(mm); - ret = put_user(tmp, (unsigned long __user *) data); + ret = put_user(tmp, datalp); break; } #endif @@ -653,7 +653,7 @@ int ptrace_request(struct task_struct *child, long request, case PTRACE_SETREGSET: { struct iovec kiov; - struct iovec __user *uiov = (struct iovec __user *) data; + struct iovec __user *uiov = datavp; if (!access_ok(VERIFY_WRITE, uiov, sizeof(*uiov))) return -EFAULT; -- cgit v1.2.3 From b8ed374e202e23caaf9bd77dcadc9de6447faaa8 Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Wed, 27 Oct 2010 15:34:06 -0700 Subject: signals: annotate lock_task_sighand() lock_task_sighand() grabs sighand->siglock in case of returning non-NULL but unlock_task_sighand() releases it unconditionally. This leads sparse to complain about the lock context imbalance. Rename and wrap lock_task_sighand() using __cond_lock() macro to make sparse happy. Suggested-by: Eric Dumazet Signed-off-by: Namhyung Kim Cc: Ingo Molnar Cc: Oleg Nesterov Cc: Roland McGrath Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/signal.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/signal.c b/kernel/signal.c index 919562c3d6b7..e921409b85a9 100644 --- a/kernel/signal.c +++ b/kernel/signal.c @@ -1105,7 +1105,8 @@ int zap_other_threads(struct task_struct *p) return count; } -struct sighand_struct *lock_task_sighand(struct task_struct *tsk, unsigned long *flags) +struct sighand_struct *__lock_task_sighand(struct task_struct *tsk, + unsigned long *flags) { struct sighand_struct *sighand; -- cgit v1.2.3 From b84011508360d6885a9d95a235ec77d56f133377 Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Wed, 27 Oct 2010 15:34:07 -0700 Subject: signals: annotate lock context change on ptrace_stop() ptrace_stop() releases and regrabs current->sighand->siglock but was missing proper annotation. Add it. Signed-off-by: Namhyung Kim Acked-by: Roland McGrath Cc: Ingo Molnar Cc: Oleg Nesterov Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/signal.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'kernel') diff --git a/kernel/signal.c b/kernel/signal.c index e921409b85a9..4e3cff10fdce 100644 --- a/kernel/signal.c +++ b/kernel/signal.c @@ -1618,6 +1618,8 @@ static int sigkill_pending(struct task_struct *tsk) * is gone, we keep current->exit_code unless clear_code. */ static void ptrace_stop(int exit_code, int clear_code, siginfo_t *info) + __releases(¤t->sighand->siglock) + __acquires(¤t->sighand->siglock) { if (arch_ptrace_stop_needed(exit_code, info)) { /* -- cgit v1.2.3 From 9b1bf12d5d51bca178dea21b04a0805e29d60cf1 Mon Sep 17 00:00:00 2001 From: KOSAKI Motohiro Date: Wed, 27 Oct 2010 15:34:08 -0700 Subject: signals: move cred_guard_mutex from task_struct to signal_struct Oleg Nesterov pointed out we have to prevent multiple-threads-inside-exec itself and we can reuse ->cred_guard_mutex for it. Yes, concurrent execve() has no worth. Let's move ->cred_guard_mutex from task_struct to signal_struct. It naturally prevent multiple-threads-inside-exec. Signed-off-by: KOSAKI Motohiro Reviewed-by: Oleg Nesterov Acked-by: Roland McGrath Acked-by: David Howells Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/cred.c | 4 +--- kernel/fork.c | 2 ++ kernel/ptrace.c | 4 ++-- 3 files changed, 5 insertions(+), 5 deletions(-) (limited to 'kernel') diff --git a/kernel/cred.c b/kernel/cred.c index 9a3e22641fe7..6a1aa004e376 100644 --- a/kernel/cred.c +++ b/kernel/cred.c @@ -325,7 +325,7 @@ EXPORT_SYMBOL(prepare_creds); /* * Prepare credentials for current to perform an execve() - * - The caller must hold current->cred_guard_mutex + * - The caller must hold ->cred_guard_mutex */ struct cred *prepare_exec_creds(void) { @@ -384,8 +384,6 @@ int copy_creds(struct task_struct *p, unsigned long clone_flags) struct cred *new; int ret; - mutex_init(&p->cred_guard_mutex); - if ( #ifdef CONFIG_KEYS !p->cred->thread_keyring && diff --git a/kernel/fork.c b/kernel/fork.c index e87aaaaf5131..3b159c5991b7 100644 --- a/kernel/fork.c +++ b/kernel/fork.c @@ -908,6 +908,8 @@ static int copy_signal(unsigned long clone_flags, struct task_struct *tsk) sig->oom_adj = current->signal->oom_adj; sig->oom_score_adj = current->signal->oom_score_adj; + mutex_init(&sig->cred_guard_mutex); + return 0; } diff --git a/kernel/ptrace.c b/kernel/ptrace.c index ea7ce0215cd1..99bbaa3e5b0d 100644 --- a/kernel/ptrace.c +++ b/kernel/ptrace.c @@ -181,7 +181,7 @@ int ptrace_attach(struct task_struct *task) * under ptrace. */ retval = -ERESTARTNOINTR; - if (mutex_lock_interruptible(&task->cred_guard_mutex)) + if (mutex_lock_interruptible(&task->signal->cred_guard_mutex)) goto out; task_lock(task); @@ -208,7 +208,7 @@ int ptrace_attach(struct task_struct *task) unlock_tasklist: write_unlock_irq(&tasklist_lock); unlock_creds: - mutex_unlock(&task->cred_guard_mutex); + mutex_unlock(&task->signal->cred_guard_mutex); out: return retval; } -- cgit v1.2.3 From d16e15f5b029fc7d03540ba0e5fb23b0abb0ebe0 Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Wed, 27 Oct 2010 15:34:10 -0700 Subject: exit: add lock context annotation on find_new_reaper() find_new_reaper() releases and regrabs tasklist_lock but was missing proper annotations. Add it. This remove following sparse warning: warning: context imbalance in 'find_new_reaper' - unexpected unlock Signed-off-by: Namhyung Kim Acked-by: Roland McGrath Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/exit.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'kernel') diff --git a/kernel/exit.c b/kernel/exit.c index 894179a32ec1..b194febf5799 100644 --- a/kernel/exit.c +++ b/kernel/exit.c @@ -703,6 +703,8 @@ static void exit_mm(struct task_struct * tsk) * space. */ static struct task_struct *find_new_reaper(struct task_struct *father) + __releases(&tasklist_lock) + __acquires(&tasklist_lock) { struct pid_namespace *pid_ns = task_active_pid_ns(father); struct task_struct *thread; -- cgit v1.2.3 From 478735e38887077ac77a9756121b6ce0cb956e2f Mon Sep 17 00:00:00 2001 From: KAMEZAWA Hiroyuki Date: Wed, 27 Oct 2010 15:34:15 -0700 Subject: /proc/stat: fix scalability of irq sum of all cpu In /proc/stat, the number of per-IRQ event is shown by making a sum each irq's events on all cpus. But we can make use of kstat_irqs(). kstat_irqs() do the same calculation, If !CONFIG_GENERIC_HARDIRQ, it's not a big cost. (Both of the number of cpus and irqs are small.) If a system is very big and CONFIG_GENERIC_HARDIRQ, it does for_each_irq() for_each_cpu() - look up a radix tree - read desc->irq_stat[cpu] This seems not efficient. This patch adds kstat_irqs() for CONFIG_GENRIC_HARDIRQ and change the calculation as for_each_irq() look up radix tree for_each_cpu() - read desc->irq_stat[cpu] This reduces cost. A test on (4096cpusp, 256 nodes, 4592 irqs) host (by Jack Steiner) %time cat /proc/stat > /dev/null Before Patch: 2.459 sec After Patch : .561 sec [akpm@linux-foundation.org: unexport kstat_irqs, coding-style tweaks] [akpm@linux-foundation.org: fix unused variable 'per_irq_sum'] Signed-off-by: KAMEZAWA Hiroyuki Tested-by: Jack Steiner Acked-by: Jack Steiner Cc: Yinghai Lu Cc: Ingo Molnar Cc: Thomas Gleixner Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/irq/irqdesc.c | 15 +++++++++++++++ 1 file changed, 15 insertions(+) (limited to 'kernel') diff --git a/kernel/irq/irqdesc.c b/kernel/irq/irqdesc.c index 9d917ff72675..9988d03797f5 100644 --- a/kernel/irq/irqdesc.c +++ b/kernel/irq/irqdesc.c @@ -393,3 +393,18 @@ unsigned int kstat_irqs_cpu(unsigned int irq, int cpu) struct irq_desc *desc = irq_to_desc(irq); return desc ? desc->kstat_irqs[cpu] : 0; } + +#ifdef CONFIG_GENERIC_HARDIRQS +unsigned int kstat_irqs(unsigned int irq) +{ + struct irq_desc *desc = irq_to_desc(irq); + int cpu; + int sum = 0; + + if (!desc) + return 0; + for_each_possible_cpu(cpu) + sum += desc->kstat_irqs[cpu]; + return sum; +} +#endif /* CONFIG_GENERIC_HARDIRQS */ -- cgit v1.2.3 From 85893120699f8bae8caa12a8ee18ab5fceac978e Mon Sep 17 00:00:00 2001 From: Jeff Mahoney Date: Wed, 27 Oct 2010 15:34:43 -0700 Subject: delayacct: align to 8 byte boundary on 64-bit systems prepare_reply() sets up an skb for the response. The payload contains: +--------------------------------+ | genlmsghdr - 4 bytes | +--------------------------------+ | NLA header - 4 bytes | /* Aggregate header */ +-+------------------------------+ | | NLA header - 4 bytes | /* PID header */ | +------------------------------+ | | pid/tgid - 4 bytes | | +------------------------------+ | | NLA header - 4 bytes | /* stats header */ | + -----------------------------+ <- oops. aligned on 4 byte boundary | | struct taskstats - 328 bytes | +-+------------------------------+ The start of the taskstats struct must be 8 byte aligned on IA64 (and other systems with 8 byte alignment rules for 64-bit types) or runtime alignment warnings will be issued. This patch pads the pid/tgid field out to sizeof(long), which forces the alignment of taskstats. The getdelays userspace code is ok with this since it assumes 32-bit pid/tgid and then honors that header's length field. An array is used to avoid exposing kernel memory contents to userspace in the response. Signed-off-by: Jeff Mahoney Cc: Balbir Singh Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/taskstats.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/taskstats.c b/kernel/taskstats.c index 11281d5792bd..5a651aa63d61 100644 --- a/kernel/taskstats.c +++ b/kernel/taskstats.c @@ -360,6 +360,12 @@ static struct taskstats *mk_reply(struct sk_buff *skb, int type, u32 pid) struct nlattr *na, *ret; int aggr; + /* If we don't pad, we end up with alignment on a 4 byte boundary. + * This causes lots of runtime warnings on systems requiring 8 byte + * alignment */ + u32 pids[2] = { pid, 0 }; + int pid_size = ALIGN(sizeof(pid), sizeof(long)); + aggr = (type == TASKSTATS_TYPE_PID) ? TASKSTATS_TYPE_AGGR_PID : TASKSTATS_TYPE_AGGR_TGID; @@ -367,7 +373,7 @@ static struct taskstats *mk_reply(struct sk_buff *skb, int type, u32 pid) na = nla_nest_start(skb, aggr); if (!na) goto err; - if (nla_put(skb, type, sizeof(pid), &pid) < 0) + if (nla_put(skb, type, pid_size, pids) < 0) goto err; ret = nla_reserve(skb, TASKSTATS_TYPE_STATS, sizeof(struct taskstats)); if (!ret) -- cgit v1.2.3 From 9323312592cca636d7c2580dc85fa4846efa86a2 Mon Sep 17 00:00:00 2001 From: Michael Holzheu Date: Wed, 27 Oct 2010 15:34:44 -0700 Subject: taskstats: separate taskstats commands Move each taskstats command into a single function. This makes the code more readable and makes it easier to add new commands. Signed-off-by: Michael Holzheu Acked-by: Balbir Singh Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/taskstats.c | 118 +++++++++++++++++++++++++++++++++++------------------ 1 file changed, 78 insertions(+), 40 deletions(-) (limited to 'kernel') diff --git a/kernel/taskstats.c b/kernel/taskstats.c index 5a651aa63d61..9970cae04f15 100644 --- a/kernel/taskstats.c +++ b/kernel/taskstats.c @@ -430,39 +430,46 @@ err: return rc; } -static int taskstats_user_cmd(struct sk_buff *skb, struct genl_info *info) +static int cmd_attr_register_cpumask(struct genl_info *info) { - int rc; - struct sk_buff *rep_skb; - struct taskstats *stats; - size_t size; cpumask_var_t mask; + int rc; if (!alloc_cpumask_var(&mask, GFP_KERNEL)) return -ENOMEM; - rc = parse(info->attrs[TASKSTATS_CMD_ATTR_REGISTER_CPUMASK], mask); if (rc < 0) - goto free_return_rc; - if (rc == 0) { - rc = add_del_listener(info->snd_pid, mask, REGISTER); - goto free_return_rc; - } + goto out; + rc = add_del_listener(info->snd_pid, mask, REGISTER); +out: + free_cpumask_var(mask); + return rc; +} +static int cmd_attr_deregister_cpumask(struct genl_info *info) +{ + cpumask_var_t mask; + int rc; + + if (!alloc_cpumask_var(&mask, GFP_KERNEL)) + return -ENOMEM; rc = parse(info->attrs[TASKSTATS_CMD_ATTR_DEREGISTER_CPUMASK], mask); if (rc < 0) - goto free_return_rc; - if (rc == 0) { - rc = add_del_listener(info->snd_pid, mask, DEREGISTER); -free_return_rc: - free_cpumask_var(mask); - return rc; - } + goto out; + rc = add_del_listener(info->snd_pid, mask, DEREGISTER); +out: free_cpumask_var(mask); + return rc; +} + +static int cmd_attr_pid(struct genl_info *info) +{ + struct taskstats *stats; + struct sk_buff *rep_skb; + size_t size; + u32 pid; + int rc; - /* - * Size includes space for nested attributes - */ size = nla_total_size(sizeof(u32)) + nla_total_size(sizeof(struct taskstats)) + nla_total_size(0); @@ -471,33 +478,64 @@ free_return_rc: return rc; rc = -EINVAL; - if (info->attrs[TASKSTATS_CMD_ATTR_PID]) { - u32 pid = nla_get_u32(info->attrs[TASKSTATS_CMD_ATTR_PID]); - stats = mk_reply(rep_skb, TASKSTATS_TYPE_PID, pid); - if (!stats) - goto err; - - rc = fill_pid(pid, NULL, stats); - if (rc < 0) - goto err; - } else if (info->attrs[TASKSTATS_CMD_ATTR_TGID]) { - u32 tgid = nla_get_u32(info->attrs[TASKSTATS_CMD_ATTR_TGID]); - stats = mk_reply(rep_skb, TASKSTATS_TYPE_TGID, tgid); - if (!stats) - goto err; - - rc = fill_tgid(tgid, NULL, stats); - if (rc < 0) - goto err; - } else + pid = nla_get_u32(info->attrs[TASKSTATS_CMD_ATTR_PID]); + stats = mk_reply(rep_skb, TASKSTATS_TYPE_PID, pid); + if (!stats) + goto err; + + rc = fill_pid(pid, NULL, stats); + if (rc < 0) + goto err; + return send_reply(rep_skb, info); +err: + nlmsg_free(rep_skb); + return rc; +} + +static int cmd_attr_tgid(struct genl_info *info) +{ + struct taskstats *stats; + struct sk_buff *rep_skb; + size_t size; + u32 tgid; + int rc; + + size = nla_total_size(sizeof(u32)) + + nla_total_size(sizeof(struct taskstats)) + nla_total_size(0); + + rc = prepare_reply(info, TASKSTATS_CMD_NEW, &rep_skb, size); + if (rc < 0) + return rc; + + rc = -EINVAL; + tgid = nla_get_u32(info->attrs[TASKSTATS_CMD_ATTR_TGID]); + stats = mk_reply(rep_skb, TASKSTATS_TYPE_TGID, tgid); + if (!stats) goto err; + rc = fill_tgid(tgid, NULL, stats); + if (rc < 0) + goto err; return send_reply(rep_skb, info); err: nlmsg_free(rep_skb); return rc; } +static int taskstats_user_cmd(struct sk_buff *skb, struct genl_info *info) +{ + if (info->attrs[TASKSTATS_CMD_ATTR_REGISTER_CPUMASK]) + return cmd_attr_register_cpumask(info); + else if (info->attrs[TASKSTATS_CMD_ATTR_DEREGISTER_CPUMASK]) + return cmd_attr_deregister_cpumask(info); + else if (info->attrs[TASKSTATS_CMD_ATTR_PID]) + return cmd_attr_pid(info); + else if (info->attrs[TASKSTATS_CMD_ATTR_TGID]) + return cmd_attr_tgid(info); + else + return -EINVAL; +} + static struct taskstats *taskstats_tgid_alloc(struct task_struct *tsk) { struct signal_struct *sig = tsk->signal; -- cgit v1.2.3 From 3d9e0cf1fe007b88db55d43dfdb6839e1a029ca5 Mon Sep 17 00:00:00 2001 From: Michael Holzheu Date: Wed, 27 Oct 2010 15:34:44 -0700 Subject: taskstats: split fill_pid function Separate the finding of a task_struct by pid or tgid from filling the taskstats data. This makes the code more readable. Signed-off-by: Michael Holzheu Acked-by: Balbir Singh Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/taskstats.c | 50 +++++++++++++++++++++----------------------------- 1 file changed, 21 insertions(+), 29 deletions(-) (limited to 'kernel') diff --git a/kernel/taskstats.c b/kernel/taskstats.c index 9970cae04f15..c8231fb15708 100644 --- a/kernel/taskstats.c +++ b/kernel/taskstats.c @@ -175,22 +175,8 @@ static void send_cpu_listeners(struct sk_buff *skb, up_write(&listeners->sem); } -static int fill_pid(pid_t pid, struct task_struct *tsk, - struct taskstats *stats) +static void fill_stats(struct task_struct *tsk, struct taskstats *stats) { - int rc = 0; - - if (!tsk) { - rcu_read_lock(); - tsk = find_task_by_vpid(pid); - if (tsk) - get_task_struct(tsk); - rcu_read_unlock(); - if (!tsk) - return -ESRCH; - } else - get_task_struct(tsk); - memset(stats, 0, sizeof(*stats)); /* * Each accounting subsystem adds calls to its functions to @@ -209,17 +195,27 @@ static int fill_pid(pid_t pid, struct task_struct *tsk, /* fill in extended acct fields */ xacct_add_tsk(stats, tsk); +} - /* Define err: label here if needed */ - put_task_struct(tsk); - return rc; +static int fill_stats_for_pid(pid_t pid, struct taskstats *stats) +{ + struct task_struct *tsk; + rcu_read_lock(); + tsk = find_task_by_vpid(pid); + if (tsk) + get_task_struct(tsk); + rcu_read_unlock(); + if (!tsk) + return -ESRCH; + fill_stats(tsk, stats); + put_task_struct(tsk); + return 0; } -static int fill_tgid(pid_t tgid, struct task_struct *first, - struct taskstats *stats) +static int fill_stats_for_tgid(pid_t tgid, struct taskstats *stats) { - struct task_struct *tsk; + struct task_struct *tsk, *first; unsigned long flags; int rc = -ESRCH; @@ -228,8 +224,7 @@ static int fill_tgid(pid_t tgid, struct task_struct *first, * leaders who are already counted with the dead tasks */ rcu_read_lock(); - if (!first) - first = find_task_by_vpid(tgid); + first = find_task_by_vpid(tgid); if (!first || !lock_task_sighand(first, &flags)) goto out; @@ -268,7 +263,6 @@ out: return rc; } - static void fill_tgid_exit(struct task_struct *tsk) { unsigned long flags; @@ -483,7 +477,7 @@ static int cmd_attr_pid(struct genl_info *info) if (!stats) goto err; - rc = fill_pid(pid, NULL, stats); + rc = fill_stats_for_pid(pid, stats); if (rc < 0) goto err; return send_reply(rep_skb, info); @@ -513,7 +507,7 @@ static int cmd_attr_tgid(struct genl_info *info) if (!stats) goto err; - rc = fill_tgid(tgid, NULL, stats); + rc = fill_stats_for_tgid(tgid, stats); if (rc < 0) goto err; return send_reply(rep_skb, info); @@ -599,9 +593,7 @@ void taskstats_exit(struct task_struct *tsk, int group_dead) if (!stats) goto err; - rc = fill_pid(-1, tsk, stats); - if (rc < 0) - goto err; + fill_stats(tsk, stats); /* * Doesn't matter if tsk is the leader or the last group member leaving -- cgit v1.2.3 From d57af9b2142f31a39dcfdeb30776baadfc802827 Mon Sep 17 00:00:00 2001 From: Michael Holzheu Date: Wed, 27 Oct 2010 15:34:45 -0700 Subject: taskstats: use real microsecond granularity for CPU times The taskstats interface uses microsecond granularity for the user and system time values. The conversion from cputime to the taskstats values uses the cputime_to_msecs primitive which effectively limits the granularity to milliseconds. Add the cputime_to_usecs primitive for architectures that have better, more precise CPU time values. Remove cputime_to_msecs primitive because there are no more users left. Signed-off-by: Michael Holzheu Acked-by: Balbir Singh Cc: Luck Tony Cc: Shailabh Nagar Cc: Martin Schwidefsky Cc: Oleg Nesterov Cc: Benjamin Herrenschmidt Cc: Heiko Carstens Cc: Thomas Gleixner Cc: Shailabh Nagar Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/tsacct.c | 10 ++++------ 1 file changed, 4 insertions(+), 6 deletions(-) (limited to 'kernel') diff --git a/kernel/tsacct.c b/kernel/tsacct.c index 0a67e041edf8..24dc60d9fa1f 100644 --- a/kernel/tsacct.c +++ b/kernel/tsacct.c @@ -63,12 +63,10 @@ void bacct_add_tsk(struct taskstats *stats, struct task_struct *tsk) stats->ac_ppid = pid_alive(tsk) ? rcu_dereference(tsk->real_parent)->tgid : 0; rcu_read_unlock(); - stats->ac_utime = cputime_to_msecs(tsk->utime) * USEC_PER_MSEC; - stats->ac_stime = cputime_to_msecs(tsk->stime) * USEC_PER_MSEC; - stats->ac_utimescaled = - cputime_to_msecs(tsk->utimescaled) * USEC_PER_MSEC; - stats->ac_stimescaled = - cputime_to_msecs(tsk->stimescaled) * USEC_PER_MSEC; + stats->ac_utime = cputime_to_usecs(tsk->utime); + stats->ac_stime = cputime_to_usecs(tsk->stime); + stats->ac_utimescaled = cputime_to_usecs(tsk->utimescaled); + stats->ac_stimescaled = cputime_to_usecs(tsk->stimescaled); stats->ac_minflt = tsk->min_flt; stats->ac_majflt = tsk->maj_flt; -- cgit v1.2.3 From 5de1cb2d0f1c1e5475d2bedf65b76828f8cdde22 Mon Sep 17 00:00:00 2001 From: Huang Shijie Date: Wed, 27 Oct 2010 15:34:52 -0700 Subject: kernel/resource.c: handle reinsertion of an already-inserted resource If the same resource is inserted to the resource tree (maybe not on purpose), a dead loop will be created. In this situation, The kernel does not report any warning or error :( The command below will show a endless print. #cat /proc/iomem [akpm@linux-foundation.org: add WARN_ON()] Signed-off-by: Huang Shijie Cc: Jesse Barnes Cc: Bjorn Helgaas Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/resource.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'kernel') diff --git a/kernel/resource.c b/kernel/resource.c index 7b36976e5dea..9c9841cb6902 100644 --- a/kernel/resource.c +++ b/kernel/resource.c @@ -453,6 +453,8 @@ static struct resource * __insert_resource(struct resource *parent, struct resou if (first == parent) return first; + if (WARN_ON(first == new)) /* duplicated insertion */ + return first; if ((first->start > new->start) || (first->end < new->end)) break; -- cgit v1.2.3 From 61d8e11e519ee7912ab59610fba1aaf08e3c1d84 Mon Sep 17 00:00:00 2001 From: Zimny Lech Date: Wed, 27 Oct 2010 15:34:53 -0700 Subject: Remove duplicate includes from many files Signed-off-by: Zimny Lech Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/trace/trace_kprobe.c | 1 - 1 file changed, 1 deletion(-) (limited to 'kernel') diff --git a/kernel/trace/trace_kprobe.c b/kernel/trace/trace_kprobe.c index b8d2852baa4a..2dec9bcde8b4 100644 --- a/kernel/trace/trace_kprobe.c +++ b/kernel/trace/trace_kprobe.c @@ -31,7 +31,6 @@ #include #include #include -#include #include #include "trace.h" -- cgit v1.2.3 From b842f8faf6c7dc2005c6a70631c1a91bac02f180 Mon Sep 17 00:00:00 2001 From: Jason Baron Date: Fri, 1 Oct 2010 17:23:41 -0400 Subject: jump label: Fix module __init section race Jump label uses is_module_text_address() to ensure that the module __init sections are valid before updating them. However, between the check for a valid module __init section and the subsequent jump label update, the module's __init section could be freed out from under us. We fix this potential race by adding a notifier callback to the MODULE_STATE_LIVE state. This notifier is called *after* the __init section has been run but before it is going to be freed. In the callback, the jump label code zeros the key value for any __init jump code within the module, and we add a check for a non-zero key value when we update jump labels. In this way we require no additional data structures. Thanks to Mathieu Desnoyers for pointing out this race condition. Reported-by: Mathieu Desnoyers Cc: Masami Hiramatsu Signed-off-by: Jason Baron LKML-Reference: [ Renamed remove_module_init() to remove_jump_label_module_init() as suggested by Masami Hiramatsu. ] Signed-off-by: Steven Rostedt --- kernel/jump_label.c | 41 ++++++++++++++++++++++++++++++++++++++++- 1 file changed, 40 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/jump_label.c b/kernel/jump_label.c index 7be868bf25c6..be9e105345eb 100644 --- a/kernel/jump_label.c +++ b/kernel/jump_label.c @@ -168,7 +168,8 @@ void jump_label_update(unsigned long key, enum jump_label_type type) count = e_module->nr_entries; iter = e_module->table; while (count--) { - if (kernel_text_address(iter->code)) + if (iter->key && + kernel_text_address(iter->code)) arch_jump_label_transform(iter, type); iter++; } @@ -366,6 +367,39 @@ static void remove_jump_label_module(struct module *mod) } } +static void remove_jump_label_module_init(struct module *mod) +{ + struct hlist_head *head; + struct hlist_node *node, *node_next, *module_node, *module_node_next; + struct jump_label_entry *e; + struct jump_label_module_entry *e_module; + struct jump_entry *iter; + int i, count; + + /* if the module doesn't have jump label entries, just return */ + if (!mod->num_jump_entries) + return; + + for (i = 0; i < JUMP_LABEL_TABLE_SIZE; i++) { + head = &jump_label_table[i]; + hlist_for_each_entry_safe(e, node, node_next, head, hlist) { + hlist_for_each_entry_safe(e_module, module_node, + module_node_next, + &(e->modules), hlist) { + if (e_module->mod != mod) + continue; + count = e_module->nr_entries; + iter = e_module->table; + while (count--) { + if (within_module_init(iter->code, mod)) + iter->key = 0; + iter++; + } + } + } + } +} + static int jump_label_module_notify(struct notifier_block *self, unsigned long val, void *data) @@ -386,6 +420,11 @@ jump_label_module_notify(struct notifier_block *self, unsigned long val, remove_jump_label_module(mod); mutex_unlock(&jump_label_mutex); break; + case MODULE_STATE_LIVE: + mutex_lock(&jump_label_mutex); + remove_jump_label_module_init(mod); + mutex_unlock(&jump_label_mutex); + break; } return ret; } -- cgit v1.2.3 From 91bad2f8d3057482b9afb599f14421b007136960 Mon Sep 17 00:00:00 2001 From: Jason Baron Date: Fri, 1 Oct 2010 17:23:48 -0400 Subject: jump label: Fix deadlock b/w jump_label_mutex vs. text_mutex register_kprobe() downs the 'text_mutex' and then calls jump_label_text_reserved(), which downs the 'jump_label_mutex'. However, the jump label code takes those mutexes in the reverse order. Fix by requiring the caller of jump_label_text_reserved() to do the jump label locking via the newly added: jump_label_lock(), jump_label_unlock(). Currently, kprobes is the only user of jump_label_text_reserved(). Reported-by: Ingo Molnar Acked-by: Masami Hiramatsu Signed-off-by: Jason Baron LKML-Reference: <759032c48d5e30c27f0bba003d09bffa8e9f28bb.1285965957.git.jbaron@redhat.com> Signed-off-by: Steven Rostedt --- kernel/jump_label.c | 33 +++++++++++++++++++++------------ kernel/kprobes.c | 6 ++++++ 2 files changed, 27 insertions(+), 12 deletions(-) (limited to 'kernel') diff --git a/kernel/jump_label.c b/kernel/jump_label.c index be9e105345eb..12cce78e9568 100644 --- a/kernel/jump_label.c +++ b/kernel/jump_label.c @@ -39,6 +39,16 @@ struct jump_label_module_entry { struct module *mod; }; +void jump_label_lock(void) +{ + mutex_lock(&jump_label_mutex); +} + +void jump_label_unlock(void) +{ + mutex_unlock(&jump_label_mutex); +} + static int jump_label_cmp(const void *a, const void *b) { const struct jump_entry *jea = a; @@ -152,7 +162,7 @@ void jump_label_update(unsigned long key, enum jump_label_type type) struct jump_label_module_entry *e_module; int count; - mutex_lock(&jump_label_mutex); + jump_label_lock(); entry = get_jump_label_entry((jump_label_t)key); if (entry) { count = entry->nr_entries; @@ -175,7 +185,7 @@ void jump_label_update(unsigned long key, enum jump_label_type type) } } } - mutex_unlock(&jump_label_mutex); + jump_label_unlock(); } static int addr_conflict(struct jump_entry *entry, void *start, void *end) @@ -232,6 +242,7 @@ out: * overlaps with any of the jump label patch addresses. Code * that wants to modify kernel text should first verify that * it does not overlap with any of the jump label addresses. + * Caller must hold jump_label_mutex. * * returns 1 if there is an overlap, 0 otherwise */ @@ -242,7 +253,6 @@ int jump_label_text_reserved(void *start, void *end) struct jump_entry *iter_stop = __start___jump_table; int conflict = 0; - mutex_lock(&jump_label_mutex); iter = iter_start; while (iter < iter_stop) { if (addr_conflict(iter, start, end)) { @@ -257,7 +267,6 @@ int jump_label_text_reserved(void *start, void *end) conflict = module_conflict(start, end); #endif out: - mutex_unlock(&jump_label_mutex); return conflict; } @@ -268,7 +277,7 @@ static __init int init_jump_label(void) struct jump_entry *iter_stop = __stop___jump_table; struct jump_entry *iter; - mutex_lock(&jump_label_mutex); + jump_label_lock(); ret = build_jump_label_hashtable(__start___jump_table, __stop___jump_table); iter = iter_start; @@ -276,7 +285,7 @@ static __init int init_jump_label(void) arch_jump_label_text_poke_early(iter->code); iter++; } - mutex_unlock(&jump_label_mutex); + jump_label_unlock(); return ret; } early_initcall(init_jump_label); @@ -409,21 +418,21 @@ jump_label_module_notify(struct notifier_block *self, unsigned long val, switch (val) { case MODULE_STATE_COMING: - mutex_lock(&jump_label_mutex); + jump_label_lock(); ret = add_jump_label_module(mod); if (ret) remove_jump_label_module(mod); - mutex_unlock(&jump_label_mutex); + jump_label_unlock(); break; case MODULE_STATE_GOING: - mutex_lock(&jump_label_mutex); + jump_label_lock(); remove_jump_label_module(mod); - mutex_unlock(&jump_label_mutex); + jump_label_unlock(); break; case MODULE_STATE_LIVE: - mutex_lock(&jump_label_mutex); + jump_label_lock(); remove_jump_label_module_init(mod); - mutex_unlock(&jump_label_mutex); + jump_label_unlock(); break; } return ret; diff --git a/kernel/kprobes.c b/kernel/kprobes.c index 99865c33a60d..9437e14f36bd 100644 --- a/kernel/kprobes.c +++ b/kernel/kprobes.c @@ -1146,13 +1146,16 @@ int __kprobes register_kprobe(struct kprobe *p) return ret; preempt_disable(); + jump_label_lock(); if (!kernel_text_address((unsigned long) p->addr) || in_kprobes_functions((unsigned long) p->addr) || ftrace_text_reserved(p->addr, p->addr) || jump_label_text_reserved(p->addr, p->addr)) { preempt_enable(); + jump_label_unlock(); return -EINVAL; } + jump_label_unlock(); /* User can pass only KPROBE_FLAG_DISABLED to register_kprobe */ p->flags &= KPROBE_FLAG_DISABLED; @@ -1187,6 +1190,8 @@ int __kprobes register_kprobe(struct kprobe *p) INIT_LIST_HEAD(&p->list); mutex_lock(&kprobe_mutex); + jump_label_lock(); /* needed to call jump_label_text_reserved() */ + get_online_cpus(); /* For avoiding text_mutex deadlock. */ mutex_lock(&text_mutex); @@ -1214,6 +1219,7 @@ int __kprobes register_kprobe(struct kprobe *p) out: mutex_unlock(&text_mutex); put_online_cpus(); + jump_label_unlock(); mutex_unlock(&kprobe_mutex); if (probed_mod) -- cgit v1.2.3 From f7e835710ab5f6e43933c983f38f2d2e262b718c Mon Sep 17 00:00:00 2001 From: Al Viro Date: Mon, 26 Jul 2010 13:23:11 +0400 Subject: convert cgroup and cpuset Signed-off-by: Al Viro --- kernel/cgroup.c | 11 +++++------ kernel/cpuset.c | 13 ++++++------- 2 files changed, 11 insertions(+), 13 deletions(-) (limited to 'kernel') diff --git a/kernel/cgroup.c b/kernel/cgroup.c index 5cf366965d0c..66a416b42c18 100644 --- a/kernel/cgroup.c +++ b/kernel/cgroup.c @@ -1460,9 +1460,9 @@ static int cgroup_get_rootdir(struct super_block *sb) return 0; } -static int cgroup_get_sb(struct file_system_type *fs_type, +static struct dentry *cgroup_mount(struct file_system_type *fs_type, int flags, const char *unused_dev_name, - void *data, struct vfsmount *mnt) + void *data) { struct cgroup_sb_opts opts; struct cgroupfs_root *root; @@ -1596,10 +1596,9 @@ static int cgroup_get_sb(struct file_system_type *fs_type, drop_parsed_module_refcounts(opts.subsys_bits); } - simple_set_mnt(mnt, sb); kfree(opts.release_agent); kfree(opts.name); - return 0; + return dget(sb->s_root); drop_new_super: deactivate_locked_super(sb); @@ -1608,7 +1607,7 @@ static int cgroup_get_sb(struct file_system_type *fs_type, out_err: kfree(opts.release_agent); kfree(opts.name); - return ret; + return ERR_PTR(ret); } static void cgroup_kill_sb(struct super_block *sb) { @@ -1658,7 +1657,7 @@ static void cgroup_kill_sb(struct super_block *sb) { static struct file_system_type cgroup_fs_type = { .name = "cgroup", - .get_sb = cgroup_get_sb, + .mount = cgroup_mount, .kill_sb = cgroup_kill_sb, }; diff --git a/kernel/cpuset.c b/kernel/cpuset.c index 51b143e2a07a..4349935c2ad8 100644 --- a/kernel/cpuset.c +++ b/kernel/cpuset.c @@ -231,18 +231,17 @@ static DEFINE_SPINLOCK(cpuset_buffer_lock); * users. If someone tries to mount the "cpuset" filesystem, we * silently switch it to mount "cgroup" instead */ -static int cpuset_get_sb(struct file_system_type *fs_type, - int flags, const char *unused_dev_name, - void *data, struct vfsmount *mnt) +static struct dentry *cpuset_mount(struct file_system_type *fs_type, + int flags, const char *unused_dev_name, void *data) { struct file_system_type *cgroup_fs = get_fs_type("cgroup"); - int ret = -ENODEV; + struct dentry *ret = ERR_PTR(-ENODEV); if (cgroup_fs) { char mountopts[] = "cpuset,noprefix," "release_agent=/sbin/cpuset_release_agent"; - ret = cgroup_fs->get_sb(cgroup_fs, flags, - unused_dev_name, mountopts, mnt); + ret = cgroup_fs->mount(cgroup_fs, flags, + unused_dev_name, mountopts); put_filesystem(cgroup_fs); } return ret; @@ -250,7 +249,7 @@ static int cpuset_get_sb(struct file_system_type *fs_type, static struct file_system_type cpuset_fs_type = { .name = "cpuset", - .get_sb = cpuset_get_sb, + .mount = cpuset_mount, }; /* -- cgit v1.2.3 From de31c3ca8179d7c21def7ecb56e4fec0c8659d36 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 18 Oct 2010 10:38:58 -0400 Subject: jump label: Fix error with preempt disable holding mutex Kprobes and jump label were having a race between mutexes that was fixed by reordering the jump label. But this reordering moved the jump label mutex into a preempt disable location. This patch does a little fiddling to move the grabbing of the jump label mutex from inside the preempt disable section and still keep the order correct between the mutex and the kprobes lock. Reported-by: Ingo Molnar Acked-by: Masami Hiramatsu Cc: Jason Baron Signed-off-by: Steven Rostedt --- kernel/kprobes.c | 26 +++++++++++++------------- 1 file changed, 13 insertions(+), 13 deletions(-) (limited to 'kernel') diff --git a/kernel/kprobes.c b/kernel/kprobes.c index 9437e14f36bd..9737a76e106f 100644 --- a/kernel/kprobes.c +++ b/kernel/kprobes.c @@ -1145,17 +1145,13 @@ int __kprobes register_kprobe(struct kprobe *p) if (ret) return ret; - preempt_disable(); jump_label_lock(); + preempt_disable(); if (!kernel_text_address((unsigned long) p->addr) || in_kprobes_functions((unsigned long) p->addr) || ftrace_text_reserved(p->addr, p->addr) || - jump_label_text_reserved(p->addr, p->addr)) { - preempt_enable(); - jump_label_unlock(); - return -EINVAL; - } - jump_label_unlock(); + jump_label_text_reserved(p->addr, p->addr)) + goto fail_with_jump_label; /* User can pass only KPROBE_FLAG_DISABLED to register_kprobe */ p->flags &= KPROBE_FLAG_DISABLED; @@ -1169,10 +1165,9 @@ int __kprobes register_kprobe(struct kprobe *p) * We must hold a refcount of the probed module while updating * its code to prohibit unexpected unloading. */ - if (unlikely(!try_module_get(probed_mod))) { - preempt_enable(); - return -EINVAL; - } + if (unlikely(!try_module_get(probed_mod))) + goto fail_with_jump_label; + /* * If the module freed .init.text, we couldn't insert * kprobes in there. @@ -1180,11 +1175,11 @@ int __kprobes register_kprobe(struct kprobe *p) if (within_module_init((unsigned long)p->addr, probed_mod) && probed_mod->state != MODULE_STATE_COMING) { module_put(probed_mod); - preempt_enable(); - return -EINVAL; + goto fail_with_jump_label; } } preempt_enable(); + jump_label_unlock(); p->nmissed = 0; INIT_LIST_HEAD(&p->list); @@ -1226,6 +1221,11 @@ out: module_put(probed_mod); return ret; + +fail_with_jump_label: + preempt_enable(); + jump_label_unlock(); + return -EINVAL; } EXPORT_SYMBOL_GPL(register_kprobe); -- cgit v1.2.3 From 95bcd683fb694a3e2d0538bf486430a0dfbb4111 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Fri, 29 Oct 2010 11:02:43 -0400 Subject: jump label: Make arch_jump_label_text_poke_early() optional Some archs do not need to do anything special for jump labels on startup (like MIPS). This patch adds a weak function stub for arch_jump_label_text_poke_early(); Cc: Jason Baron Cc: David Miller Cc: David Daney Suggested-by: Thomas Gleixner LKML-Reference: <1286218615-24011-2-git-send-email-ddaney@caviumnetworks.com> LKML-Reference: <20101015201037.703989993@goodmis.org> Signed-off-by: Steven Rostedt --- kernel/jump_label.c | 7 +++++++ 1 file changed, 7 insertions(+) (limited to 'kernel') diff --git a/kernel/jump_label.c b/kernel/jump_label.c index 12cce78e9568..3b79bd938330 100644 --- a/kernel/jump_label.c +++ b/kernel/jump_label.c @@ -270,6 +270,13 @@ out: return conflict; } +/* + * Not all archs need this. + */ +void __weak arch_jump_label_text_poke_early(jump_label_t addr) +{ +} + static __init int init_jump_label(void) { int ret; -- cgit v1.2.3 From 931ea24819f2bd40cca2dc214558bfcc3c91549e Mon Sep 17 00:00:00 2001 From: Jason Wessel Date: Fri, 29 Oct 2010 08:04:16 -0500 Subject: kdb: fix per_cpu command to remove supress mask Rusty pointed out that the per_cpu command uses up lots of space on the stack and the cpu supress mask is probably not needed. This patch removes the need for the supress mask as well as fixing up the following problems with the kdb per_cpu command: * The per_cpu command should allow an address as an argument * When you have more data than can be displayed on one screen allow the user to break out of the print loop. Reported-by: Rusty Russell Signed-off-by: Jason Wessel --- kernel/debug/kdb/kdb_main.c | 46 +++++++++++---------------------------------- 1 file changed, 11 insertions(+), 35 deletions(-) (limited to 'kernel') diff --git a/kernel/debug/kdb/kdb_main.c b/kernel/debug/kdb/kdb_main.c index d7bda21a106b..9755ac05e446 100644 --- a/kernel/debug/kdb/kdb_main.c +++ b/kernel/debug/kdb/kdb_main.c @@ -2603,20 +2603,17 @@ static int kdb_summary(int argc, const char **argv) */ static int kdb_per_cpu(int argc, const char **argv) { - char buf[256], fmtstr[64]; - kdb_symtab_t symtab; - cpumask_t suppress = CPU_MASK_NONE; - int cpu, diag; - unsigned long addr, val, bytesperword = 0, whichcpu = ~0UL; + char fmtstr[64]; + int cpu, diag, nextarg = 1; + unsigned long addr, symaddr, val, bytesperword = 0, whichcpu = ~0UL; if (argc < 1 || argc > 3) return KDB_ARGCOUNT; - snprintf(buf, sizeof(buf), "per_cpu__%s", argv[1]); - if (!kdbgetsymval(buf, &symtab)) { - kdb_printf("%s is not a per_cpu variable\n", argv[1]); - return KDB_BADADDR; - } + diag = kdbgetaddrarg(argc, argv, &nextarg, &symaddr, NULL, NULL); + if (diag) + return diag; + if (argc >= 2) { diag = kdbgetularg(argv[2], &bytesperword); if (diag) @@ -2649,46 +2646,25 @@ static int kdb_per_cpu(int argc, const char **argv) #define KDB_PCU(cpu) 0 #endif #endif - for_each_online_cpu(cpu) { + if (KDB_FLAG(CMD_INTERRUPT)) + return 0; + if (whichcpu != ~0UL && whichcpu != cpu) continue; - addr = symtab.sym_start + KDB_PCU(cpu); + addr = symaddr + KDB_PCU(cpu); diag = kdb_getword(&val, addr, bytesperword); if (diag) { kdb_printf("%5d " kdb_bfd_vma_fmt0 " - unable to " "read, diag=%d\n", cpu, addr, diag); continue; } -#ifdef CONFIG_SMP - if (!val) { - cpu_set(cpu, suppress); - continue; - } -#endif /* CONFIG_SMP */ kdb_printf("%5d ", cpu); kdb_md_line(fmtstr, addr, bytesperword == KDB_WORD_SIZE, 1, bytesperword, 1, 1, 0); } - if (cpus_weight(suppress) == 0) - return 0; - kdb_printf("Zero suppressed cpu(s):"); - for (cpu = first_cpu(suppress); cpu < num_possible_cpus(); - cpu = next_cpu(cpu, suppress)) { - kdb_printf(" %d", cpu); - if (cpu == num_possible_cpus() - 1 || - next_cpu(cpu, suppress) != cpu + 1) - continue; - while (cpu < num_possible_cpus() && - next_cpu(cpu, suppress) == cpu + 1) - ++cpu; - kdb_printf("-%d", cpu); - } - kdb_printf("\n"); - #undef KDB_PCU - return 0; } -- cgit v1.2.3 From 578bd4dfcda63d2ef15f025f1d5d55c0e56b9660 Mon Sep 17 00:00:00 2001 From: Jason Wessel Date: Fri, 29 Oct 2010 13:14:41 -0500 Subject: kdb: Fix early debugging crash regression The kdb_current legally be equal to NULL in the early boot of the x86 arch. The problem pcan be observed by booting with the kernel arguments: earlyprintk=vga ekgdboc=kbd kgdbwait The kdb shell will oops on entry and recursively fault because it cannot get past the final stage of shell initialization. Signed-off-by: Jason Wessel --- kernel/debug/kdb/kdb_main.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/debug/kdb/kdb_main.c b/kernel/debug/kdb/kdb_main.c index 9755ac05e446..37755d621924 100644 --- a/kernel/debug/kdb/kdb_main.c +++ b/kernel/debug/kdb/kdb_main.c @@ -1127,7 +1127,7 @@ static int kdb_local(kdb_reason_t reason, int error, struct pt_regs *regs, /* special case below */ } else { kdb_printf("\nEntering kdb (current=0x%p, pid %d) ", - kdb_current, kdb_current->pid); + kdb_current, kdb_current ? kdb_current->pid : 0); #if defined(CONFIG_SMP) kdb_printf("on processor %d ", raw_smp_processor_id()); #endif -- cgit v1.2.3 From d7ba979d45272385ce0fdf141d922e61ff48e07b Mon Sep 17 00:00:00 2001 From: Dongdong Deng Date: Wed, 18 Aug 2010 06:02:00 -0500 Subject: debug_core,x86,blackfin: Clean up hw debug disable API The kgdb_disable_hw_debug() was an architecture specific function for disabling all hardware breakpoints on a per cpu basis when entering the debug core. This patch will remove the weak function kdbg_disable_hw_debug() and change it into a call back which lives with the rest of hw breakpoint call backs in struct kgdb_arch. Signed-off-by: Dongdong Deng Signed-off-by: Jason Wessel --- kernel/debug/debug_core.c | 16 +++------------- 1 file changed, 3 insertions(+), 13 deletions(-) (limited to 'kernel') diff --git a/kernel/debug/debug_core.c b/kernel/debug/debug_core.c index fec596da9bd0..cefd4a11f6d9 100644 --- a/kernel/debug/debug_core.c +++ b/kernel/debug/debug_core.c @@ -209,18 +209,6 @@ int __weak kgdb_skipexception(int exception, struct pt_regs *regs) return 0; } -/** - * kgdb_disable_hw_debug - Disable hardware debugging while we in kgdb. - * @regs: Current &struct pt_regs. - * - * This function will be called if the particular architecture must - * disable hardware debugging while it is processing gdb packets or - * handling exception. - */ -void __weak kgdb_disable_hw_debug(struct pt_regs *regs) -{ -} - /* * Some architectures need cache flushes when we set/clear a * breakpoint: @@ -484,7 +472,9 @@ static int kgdb_cpu_enter(struct kgdb_state *ks, struct pt_regs *regs, atomic_inc(&masters_in_kgdb); else atomic_inc(&slaves_in_kgdb); - kgdb_disable_hw_debug(ks->linux_regs); + + if (arch_kgdb_ops.disable_hw_break) + arch_kgdb_ops.disable_hw_break(regs); acquirelock: /* -- cgit v1.2.3 From d29be158a68254f58cf1fbf60ce1e89557a321aa Mon Sep 17 00:00:00 2001 From: Miloslav Trmac Date: Thu, 16 Sep 2010 18:14:11 -0400 Subject: Audit: add support to match lsm labels on user audit messages Add support for matching by security label (e.g. SELinux context) of the sender of an user-space audit record. The audit filter code already allows user space to configure such filters, but they were ignored during evaluation. This patch implements evaluation of these filters. For example, after application of this patch, PAM authentication logs caused by cron can be disabled using auditctl -a user,never -F subj_type=crond_t Signed-off-by: Miloslav Trmac Acked-by: Eric Paris Signed-off-by: Al Viro --- kernel/auditfilter.c | 12 ++++++++++++ 1 file changed, 12 insertions(+) (limited to 'kernel') diff --git a/kernel/auditfilter.c b/kernel/auditfilter.c index eb7675499fb5..add2819af71b 100644 --- a/kernel/auditfilter.c +++ b/kernel/auditfilter.c @@ -1252,6 +1252,18 @@ static int audit_filter_user_rules(struct netlink_skb_parms *cb, case AUDIT_LOGINUID: result = audit_comparator(cb->loginuid, f->op, f->val); break; + case AUDIT_SUBJ_USER: + case AUDIT_SUBJ_ROLE: + case AUDIT_SUBJ_TYPE: + case AUDIT_SUBJ_SEN: + case AUDIT_SUBJ_CLR: + if (f->lsm_rule) + result = security_audit_rule_match(cb->sid, + f->type, + f->op, + f->lsm_rule, + NULL); + break; } if (!result) -- cgit v1.2.3 From b8800aa5d9c7e4e2869321c77b80f322a0d9663a Mon Sep 17 00:00:00 2001 From: Stephen Hemminger Date: Wed, 20 Oct 2010 17:23:50 -0700 Subject: audit: make functions static I was doing some namespace checks and found some simple stuff in audit that could be cleaned up. Make some functions static, and put const on make_reply payload arg. Signed-off-by: Stephen Hemminger Signed-off-by: Al Viro --- kernel/audit.c | 6 +++--- kernel/audit.h | 5 +---- kernel/audit_watch.c | 4 ++-- 3 files changed, 6 insertions(+), 9 deletions(-) (limited to 'kernel') diff --git a/kernel/audit.c b/kernel/audit.c index d96045789b54..a300931fc45f 100644 --- a/kernel/audit.c +++ b/kernel/audit.c @@ -506,7 +506,7 @@ int audit_send_list(void *_dest) } struct sk_buff *audit_make_reply(int pid, int seq, int type, int done, - int multi, void *payload, int size) + int multi, const void *payload, int size) { struct sk_buff *skb; struct nlmsghdr *nlh; @@ -555,8 +555,8 @@ static int audit_send_reply_thread(void *arg) * Allocates an skb, builds the netlink message, and sends it to the pid. * No failure notifications. */ -void audit_send_reply(int pid, int seq, int type, int done, int multi, - void *payload, int size) +static void audit_send_reply(int pid, int seq, int type, int done, int multi, + const void *payload, int size) { struct sk_buff *skb; struct task_struct *tsk; diff --git a/kernel/audit.h b/kernel/audit.h index f7206db4e13d..91e7071c4d2c 100644 --- a/kernel/audit.h +++ b/kernel/audit.h @@ -84,10 +84,7 @@ extern int audit_compare_dname_path(const char *dname, const char *path, int *dirlen); extern struct sk_buff * audit_make_reply(int pid, int seq, int type, int done, int multi, - void *payload, int size); -extern void audit_send_reply(int pid, int seq, int type, - int done, int multi, - void *payload, int size); + const void *payload, int size); extern void audit_panic(const char *message); struct audit_netlink_list { diff --git a/kernel/audit_watch.c b/kernel/audit_watch.c index f0c9b2e7542d..d2e3c7866460 100644 --- a/kernel/audit_watch.c +++ b/kernel/audit_watch.c @@ -60,7 +60,7 @@ struct audit_parent { }; /* fsnotify handle. */ -struct fsnotify_group *audit_watch_group; +static struct fsnotify_group *audit_watch_group; /* fsnotify events we care about. */ #define AUDIT_FS_WATCH (FS_MOVE | FS_CREATE | FS_DELETE | FS_DELETE_SELF |\ @@ -123,7 +123,7 @@ void audit_put_watch(struct audit_watch *watch) } } -void audit_remove_watch(struct audit_watch *watch) +static void audit_remove_watch(struct audit_watch *watch) { list_del(&watch->wlist); audit_put_parent(watch->parent); -- cgit v1.2.3 From f7a998a9491f2da1d3e44d150aa611d10093da4f Mon Sep 17 00:00:00 2001 From: Al Viro Date: Sat, 30 Oct 2010 02:18:32 -0400 Subject: in untag_chunk() we need to do alloc_chunk() a bit earlier ... while we are not holding spinlocks. Signed-off-by: Al Viro --- kernel/audit_tree.c | 9 +++++++-- 1 file changed, 7 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/audit_tree.c b/kernel/audit_tree.c index 7f18d3a4527e..37b2bea170c8 100644 --- a/kernel/audit_tree.c +++ b/kernel/audit_tree.c @@ -223,7 +223,7 @@ static void untag_chunk(struct node *p) { struct audit_chunk *chunk = find_chunk(p); struct fsnotify_mark *entry = &chunk->mark; - struct audit_chunk *new; + struct audit_chunk *new = NULL; struct audit_tree *owner; int size = chunk->count - 1; int i, j; @@ -232,9 +232,14 @@ static void untag_chunk(struct node *p) spin_unlock(&hash_lock); + if (size) + new = alloc_chunk(size); + spin_lock(&entry->lock); if (chunk->dead || !entry->i.inode) { spin_unlock(&entry->lock); + if (new) + free_chunk(new); goto out; } @@ -255,9 +260,9 @@ static void untag_chunk(struct node *p) goto out; } - new = alloc_chunk(size); if (!new) goto Fallback; + fsnotify_duplicate_mark(&new->mark, entry); if (fsnotify_add_mark(&new->mark, new->mark.group, new->mark.i.inode, NULL, 1)) { free_chunk(new); -- cgit v1.2.3 From 3c80fe4ac9cfb13b1bfa4edf1544e8b656716694 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Wed, 9 Dec 2009 14:19:31 +0000 Subject: audit: Call tty_audit_push_task() outside preempt disabled While auditing all tasklist_lock read_lock sites I stumbled over the following call chain: audit_prepare_user_tty() read_lock(&tasklist_lock); tty_audit_push_task(); mutex_lock(&buf->mutex); --> buf->mutex is locked with preemption disabled. Solve this by acquiring a reference to the task struct under rcu_read_lock and call tty_audit_push_task outside of the preempt disabled region. Move all code which needs to be protected by sighand lock into tty_audit_push_task() and use lock/unlock_sighand as we do not hold tasklist_lock. Signed-off-by: Thomas Gleixner Cc: Al Viro Cc: Eric Paris Cc: Oleg Nesterov Signed-off-by: Al Viro --- kernel/audit.c | 25 +++++++++---------------- 1 file changed, 9 insertions(+), 16 deletions(-) (limited to 'kernel') diff --git a/kernel/audit.c b/kernel/audit.c index a300931fc45f..8429afea37bf 100644 --- a/kernel/audit.c +++ b/kernel/audit.c @@ -467,23 +467,16 @@ static int audit_prepare_user_tty(pid_t pid, uid_t loginuid, u32 sessionid) struct task_struct *tsk; int err; - read_lock(&tasklist_lock); + rcu_read_lock(); tsk = find_task_by_vpid(pid); - err = -ESRCH; - if (!tsk) - goto out; - err = 0; - - spin_lock_irq(&tsk->sighand->siglock); - if (!tsk->signal->audit_tty) - err = -EPERM; - spin_unlock_irq(&tsk->sighand->siglock); - if (err) - goto out; - - tty_audit_push_task(tsk, loginuid, sessionid); -out: - read_unlock(&tasklist_lock); + if (!tsk) { + rcu_read_unlock(); + return -ESRCH; + } + get_task_struct(tsk); + rcu_read_unlock(); + err = tty_audit_push_task(tsk, loginuid, sessionid); + put_task_struct(tsk); return err; } -- cgit v1.2.3 From 207032051a5ed38df332729ba42e98e9a1e60434 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Wed, 9 Dec 2009 14:19:35 +0000 Subject: audit: Do not send uninitialized data for AUDIT_TTY_GET audit_receive_msg() sends uninitialized data for AUDIT_TTY_GET when the task was not found. Send reply only when task was found. Signed-off-by: Thomas Gleixner Cc: Al Viro Cc: Eric Paris Signed-off-by: Al Viro --- kernel/audit.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/audit.c b/kernel/audit.c index 8429afea37bf..57f4038694d1 100644 --- a/kernel/audit.c +++ b/kernel/audit.c @@ -884,8 +884,10 @@ static int audit_receive_msg(struct sk_buff *skb, struct nlmsghdr *nlh) spin_unlock_irq(&tsk->sighand->siglock); } read_unlock(&tasklist_lock); - audit_send_reply(NETLINK_CB(skb).pid, seq, AUDIT_TTY_GET, 0, 0, - &s, sizeof(s)); + + if (!err) + audit_send_reply(NETLINK_CB(skb).pid, seq, + AUDIT_TTY_GET, 0, 0, &s, sizeof(s)); break; } case AUDIT_TTY_SET: { -- cgit v1.2.3 From ab263f47c9781a644de8b28013434b645082922e Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Wed, 9 Dec 2009 14:19:41 +0000 Subject: audit: Use rcu for task lookup protection Protect the task lookups in audit_receive_msg() with rcu_read_lock() instead of tasklist_lock and use lock/unlock_sighand to protect against the exit race. Signed-off-by: Thomas Gleixner Cc: Al Viro Cc: Eric Paris Cc: Oleg Nesterov Signed-off-by: Al Viro --- kernel/audit.c | 30 ++++++++++++++---------------- 1 file changed, 14 insertions(+), 16 deletions(-) (limited to 'kernel') diff --git a/kernel/audit.c b/kernel/audit.c index 57f4038694d1..77770a034d59 100644 --- a/kernel/audit.c +++ b/kernel/audit.c @@ -873,17 +873,16 @@ static int audit_receive_msg(struct sk_buff *skb, struct nlmsghdr *nlh) case AUDIT_TTY_GET: { struct audit_tty_status s; struct task_struct *tsk; + unsigned long flags; - read_lock(&tasklist_lock); + rcu_read_lock(); tsk = find_task_by_vpid(pid); - if (!tsk) - err = -ESRCH; - else { - spin_lock_irq(&tsk->sighand->siglock); + if (tsk && lock_task_sighand(tsk, &flags)) { s.enabled = tsk->signal->audit_tty != 0; - spin_unlock_irq(&tsk->sighand->siglock); - } - read_unlock(&tasklist_lock); + unlock_task_sighand(tsk, &flags); + } else + err = -ESRCH; + rcu_read_unlock(); if (!err) audit_send_reply(NETLINK_CB(skb).pid, seq, @@ -893,22 +892,21 @@ static int audit_receive_msg(struct sk_buff *skb, struct nlmsghdr *nlh) case AUDIT_TTY_SET: { struct audit_tty_status *s; struct task_struct *tsk; + unsigned long flags; if (nlh->nlmsg_len < sizeof(struct audit_tty_status)) return -EINVAL; s = data; if (s->enabled != 0 && s->enabled != 1) return -EINVAL; - read_lock(&tasklist_lock); + rcu_read_lock(); tsk = find_task_by_vpid(pid); - if (!tsk) - err = -ESRCH; - else { - spin_lock_irq(&tsk->sighand->siglock); + if (tsk && lock_task_sighand(tsk, &flags)) { tsk->signal->audit_tty = s->enabled != 0; - spin_unlock_irq(&tsk->sighand->siglock); - } - read_unlock(&tasklist_lock); + unlock_task_sighand(tsk, &flags); + } else + err = -ESRCH; + rcu_read_unlock(); break; } default: -- cgit v1.2.3 From 120a795da07c9a02221ca23464c28a7c6ad7de1d Mon Sep 17 00:00:00 2001 From: Al Viro Date: Sat, 30 Oct 2010 02:54:44 -0400 Subject: audit mmap Normal syscall audit doesn't catch 5th argument of syscall. It also doesn't catch the contents of userland structures pointed to be syscall argument, so for both old and new mmap(2) ABI it doesn't record the descriptor we are mapping. For old one it also misses flags. Signed-off-by: Al Viro --- kernel/auditsc.c | 16 ++++++++++++++++ 1 file changed, 16 insertions(+) (limited to 'kernel') diff --git a/kernel/auditsc.c b/kernel/auditsc.c index 1b31c130d034..f49a0318c2ed 100644 --- a/kernel/auditsc.c +++ b/kernel/auditsc.c @@ -241,6 +241,10 @@ struct audit_context { pid_t pid; struct audit_cap_data cap; } capset; + struct { + int fd; + int flags; + } mmap; }; int fds[2]; @@ -1305,6 +1309,10 @@ static void show_special(struct audit_context *context, int *call_panic) audit_log_cap(ab, "cap_pp", &context->capset.cap.permitted); audit_log_cap(ab, "cap_pe", &context->capset.cap.effective); break; } + case AUDIT_MMAP: { + audit_log_format(ab, "fd=%d flags=0x%x", context->mmap.fd, + context->mmap.flags); + break; } } audit_log_end(ab); } @@ -2476,6 +2484,14 @@ void __audit_log_capset(pid_t pid, context->type = AUDIT_CAPSET; } +void __audit_mmap_fd(int fd, int flags) +{ + struct audit_context *context = current->audit_context; + context->mmap.fd = fd; + context->mmap.flags = flags; + context->type = AUDIT_MMAP; +} + /** * audit_core_dumps - record information about processes that end abnormally * @signr: signal value -- cgit v1.2.3 From 408af87a397a8ddef56ad39a79481f592aa1ac1a Mon Sep 17 00:00:00 2001 From: Jesper Juhl Date: Thu, 4 Nov 2010 21:44:41 +0100 Subject: Clean up relay_alloc_page_array() slightly by using vzalloc rather than vmalloc and memset We can optimize kernel/relay.c::relay_alloc_page_array() slightly by using vzalloc. The patch makes these changes: - use vzalloc instead of vmalloc+memset. - remove redundant local variable 'array'. - declare local 'pa_size' as const. Cuts down nicely on both source and object-code size. Signed-off-by: Jesper Juhl Acked-by: Pekka Enberg Acked-by: Mathieu Desnoyers Signed-off-by: Linus Torvalds --- kernel/relay.c | 15 ++++----------- 1 file changed, 4 insertions(+), 11 deletions(-) (limited to 'kernel') diff --git a/kernel/relay.c b/kernel/relay.c index c7cf397fb929..859ea5a9605f 100644 --- a/kernel/relay.c +++ b/kernel/relay.c @@ -70,17 +70,10 @@ static const struct vm_operations_struct relay_file_mmap_ops = { */ static struct page **relay_alloc_page_array(unsigned int n_pages) { - struct page **array; - size_t pa_size = n_pages * sizeof(struct page *); - - if (pa_size > PAGE_SIZE) { - array = vmalloc(pa_size); - if (array) - memset(array, 0, pa_size); - } else { - array = kzalloc(pa_size, GFP_KERNEL); - } - return array; + const size_t pa_size = n_pages * sizeof(struct page *); + if (pa_size > PAGE_SIZE) + return vzalloc(pa_size); + return kzalloc(pa_size, GFP_KERNEL); } /* -- cgit v1.2.3 From e0a70217107e6f9844628120412cb27bb4cea194 Mon Sep 17 00:00:00 2001 From: Oleg Nesterov Date: Fri, 5 Nov 2010 16:53:42 +0100 Subject: posix-cpu-timers: workaround to suppress the problems with mt exec posix-cpu-timers.c correctly assumes that the dying process does posix_cpu_timers_exit_group() and removes all !CPUCLOCK_PERTHREAD timers from signal->cpu_timers list. But, it also assumes that timer->it.cpu.task is always the group leader, and thus the dead ->task means the dead thread group. This is obviously not true after de_thread() changes the leader. After that almost every posix_cpu_timer_ method has problems. It is not simple to fix this bug correctly. First of all, I think that timer->it.cpu should use struct pid instead of task_struct. Also, the locking should be reworked completely. In particular, tasklist_lock should not be used at all. This all needs a lot of nontrivial and hard-to-test changes. Change __exit_signal() to do posix_cpu_timers_exit_group() when the old leader dies during exec. This is not the fix, just the temporary hack to hide the problem for 2.6.37 and stable. IOW, this is obviously wrong but this is what we currently have anyway: cpu timers do not work after mt exec. In theory this change adds another race. The exiting leader can detach the timers which were attached to the new leader. However, the window between de_thread() and release_task() is small, we can pretend that sys_timer_create() was called before de_thread(). Signed-off-by: Oleg Nesterov Signed-off-by: Linus Torvalds --- kernel/exit.c | 8 ++++++++ 1 file changed, 8 insertions(+) (limited to 'kernel') diff --git a/kernel/exit.c b/kernel/exit.c index b194febf5799..21aa7b3001fb 100644 --- a/kernel/exit.c +++ b/kernel/exit.c @@ -95,6 +95,14 @@ static void __exit_signal(struct task_struct *tsk) tty = sig->tty; sig->tty = NULL; } else { + /* + * This can only happen if the caller is de_thread(). + * FIXME: this is the temporary hack, we should teach + * posix-cpu-timers to handle this case correctly. + */ + if (unlikely(has_group_leader_pid(tsk))) + posix_cpu_timers_exit_group(tsk); + /* * If there is any task waiting for the group exit * then notify it: -- cgit v1.2.3 From 433039e97f672b81e6c8f6daef385dcf035c6e29 Mon Sep 17 00:00:00 2001 From: David Daney Date: Fri, 5 Nov 2010 16:17:39 -0700 Subject: watchdog: Fix section mismatch and potential undefined behavior. Commit d9ca07a05ce1 ("watchdog: Avoid kernel crash when disabling watchdog") introduces a section mismatch. Now that we reference no_watchdog from non-__init code it can no longer be __initdata. Signed-off-by: David Daney Cc: Stephane Eranian Cc: Peter Zijlstra Cc: Ingo Molnar Signed-off-by: Linus Torvalds --- kernel/watchdog.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/watchdog.c b/kernel/watchdog.c index bafba687a6d8..6e3c41a4024c 100644 --- a/kernel/watchdog.c +++ b/kernel/watchdog.c @@ -43,7 +43,7 @@ static DEFINE_PER_CPU(unsigned long, hrtimer_interrupts_saved); static DEFINE_PER_CPU(struct perf_event *, watchdog_ev); #endif -static int __initdata no_watchdog; +static int no_watchdog; /* boot commands */ -- cgit v1.2.3 From becf91f18750cf1c60828aa2ee63a36b05c2e4d0 Mon Sep 17 00:00:00 2001 From: Heiko Carstens Date: Wed, 10 Nov 2010 10:05:56 +0100 Subject: [S390] ftrace: build without frame pointers on s390 s390 doesn't need FRAME_POINTERS in order to have a working function tracer. We don't need frame pointers in order to get strack traces since we always have valid backchains by using the -mkernel-backchain gcc option. Signed-off-by: Heiko Carstens Signed-off-by: Martin Schwidefsky --- kernel/trace/Kconfig | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/trace/Kconfig b/kernel/trace/Kconfig index e04b8bcdef88..ea37e2ff4164 100644 --- a/kernel/trace/Kconfig +++ b/kernel/trace/Kconfig @@ -126,7 +126,7 @@ if FTRACE config FUNCTION_TRACER bool "Kernel Function Tracer" depends on HAVE_FUNCTION_TRACER - select FRAME_POINTER if (!ARM_UNWIND) + select FRAME_POINTER if !ARM_UNWIND && !S390 select KALLSYMS select GENERIC_TRACER select CONTEXT_SWITCH_TRACER -- cgit v1.2.3 From c0deae8c9587419ab13874b74425ce2eb2e18508 Mon Sep 17 00:00:00 2001 From: Sergey Senozhatsky Date: Wed, 3 Nov 2010 18:52:56 +0200 Subject: posix-cpu-timers: Rcu_read_lock/unlock protect find_task_by_vpid call Commit 4221a9918e38b7494cee341dda7b7b4bb8c04bde "Add RCU check for find_task_by_vpid()" introduced rcu_lockdep_assert to find_task_by_pid_ns. Add rcu_read_lock/rcu_read_unlock to call find_task_by_vpid. Tetsuo Handa wrote: | Quoting from one of posts in that thead | http://kerneltrap.org/mailarchive/linux-kernel/2010/2/8/4536388 | || Usually tasklist gives enough protection, but if copy_process() fails || it calls free_pid() lockless and does call_rcu(delayed_put_pid(). || This means, without rcu lock find_pid_ns() can't scan the hash table || safely. Thomas Gleixner wrote: | We can remove the tasklist_lock while at it. rcu_read_lock is enough. Patch also replaces thread_group_leader with has_group_leader_pid in accordance to comment by Oleg Nesterov: | ... thread_group_leader() check is not relaible without | tasklist. If we race with de_thread() find_task_by_vpid() can find | the new leader before it updates its ->group_leader. | | perhaps it makes sense to change posix_cpu_timer_create() to use | has_group_leader_pid() instead, just to make this code not look racy | and avoid adding new problems. Signed-off-by: Sergey Senozhatsky Cc: Peter Zijlstra Cc: Stanislaw Gruszka Reviewed-by: Oleg Nesterov LKML-Reference: <20101103165256.GD30053@swordfish.minsk.epam.com> Signed-off-by: Thomas Gleixner --- kernel/posix-cpu-timers.c | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-) (limited to 'kernel') diff --git a/kernel/posix-cpu-timers.c b/kernel/posix-cpu-timers.c index 6842eeba5879..05bb7173850e 100644 --- a/kernel/posix-cpu-timers.c +++ b/kernel/posix-cpu-timers.c @@ -37,13 +37,13 @@ static int check_clock(const clockid_t which_clock) if (pid == 0) return 0; - read_lock(&tasklist_lock); + rcu_read_lock(); p = find_task_by_vpid(pid); if (!p || !(CPUCLOCK_PERTHREAD(which_clock) ? - same_thread_group(p, current) : thread_group_leader(p))) { + same_thread_group(p, current) : has_group_leader_pid(p))) { error = -EINVAL; } - read_unlock(&tasklist_lock); + rcu_read_unlock(); return error; } @@ -390,7 +390,7 @@ int posix_cpu_timer_create(struct k_itimer *new_timer) INIT_LIST_HEAD(&new_timer->it.cpu.entry); - read_lock(&tasklist_lock); + rcu_read_lock(); if (CPUCLOCK_PERTHREAD(new_timer->it_clock)) { if (pid == 0) { p = current; @@ -404,7 +404,7 @@ int posix_cpu_timer_create(struct k_itimer *new_timer) p = current->group_leader; } else { p = find_task_by_vpid(pid); - if (p && !thread_group_leader(p)) + if (p && !has_group_leader_pid(p)) p = NULL; } } @@ -414,7 +414,7 @@ int posix_cpu_timer_create(struct k_itimer *new_timer) } else { ret = -EINVAL; } - read_unlock(&tasklist_lock); + rcu_read_unlock(); return ret; } -- cgit v1.2.3 From 4c115e951d80aff126468adaec7a6c7854f61ab8 Mon Sep 17 00:00:00 2001 From: Darren Hart Date: Thu, 4 Nov 2010 15:00:00 -0400 Subject: futex: Address compiler warnings in exit_robust_list MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Since commit 1dcc41bb (futex: Change 3rd arg of fetch_robust_entry() to unsigned int*) some gcc versions decided to emit the following warning: kernel/futex.c: In function ‘exit_robust_list’: kernel/futex.c:2492: warning: ‘next_pi’ may be used uninitialized in this function The commit did not introduce the warning as gcc should have warned before that commit as well. It's just gcc being silly. The code path really can't result in next_pi being unitialized (or should not), but let's keep the build clean. Annotate next_pi as an uninitialized_var. [ tglx: Addressed the same issue in futex_compat.c and massaged the changelog ] Signed-off-by: Darren Hart Tested-by: Matt Fleming Tested-by: Uwe Kleine-König Cc: Peter Zijlstra Cc: Eric Dumazet Cc: John Kacur Cc: Ingo Molnar LKML-Reference: <1288897200-13008-1-git-send-email-dvhart@linux.intel.com> Signed-off-by: Thomas Gleixner --- kernel/futex.c | 3 ++- kernel/futex_compat.c | 3 ++- 2 files changed, 4 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/futex.c b/kernel/futex.c index 6c683b37f2ce..40a8777a27d0 100644 --- a/kernel/futex.c +++ b/kernel/futex.c @@ -2489,7 +2489,8 @@ void exit_robust_list(struct task_struct *curr) { struct robust_list_head __user *head = curr->robust_list; struct robust_list __user *entry, *next_entry, *pending; - unsigned int limit = ROBUST_LIST_LIMIT, pi, next_pi, pip; + unsigned int limit = ROBUST_LIST_LIMIT, pi, pip; + unsigned int uninitialized_var(next_pi); unsigned long futex_offset; int rc; diff --git a/kernel/futex_compat.c b/kernel/futex_compat.c index 06da4dfc339b..a7934ac75e5b 100644 --- a/kernel/futex_compat.c +++ b/kernel/futex_compat.c @@ -49,7 +49,8 @@ void compat_exit_robust_list(struct task_struct *curr) { struct compat_robust_list_head __user *head = curr->compat_robust_list; struct robust_list __user *entry, *next_entry, *pending; - unsigned int limit = ROBUST_LIST_LIMIT, pi, next_pi, pip; + unsigned int limit = ROBUST_LIST_LIMIT, pi, pip; + unsigned int uninitialized_var(next_pi); compat_uptr_t uentry, next_uentry, upending; compat_long_t futex_offset; int rc; -- cgit v1.2.3 From 02e031cbc843b010e72fcc05c76113c688b2860f Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Wed, 10 Nov 2010 14:54:09 +0100 Subject: block: remove REQ_HARDBARRIER REQ_HARDBARRIER is dead now, so remove the leftovers. What's left at this point is: - various checks inside the block layer. - sanity checks in bio based drivers. - now unused bio_empty_barrier helper. - Xen blockfront use of BLKIF_OP_WRITE_BARRIER - it's dead for a while, but Xen really needs to sort out it's barrier situaton. - setting of ordered tags in uas - dead code copied from old scsi drivers. - scsi different retry for barriers - it's dead and should have been removed when flushes were converted to FS requests. - blktrace handling of barriers - removed. Someone who knows blktrace better should add support for REQ_FLUSH and REQ_FUA, though. Signed-off-by: Christoph Hellwig Signed-off-by: Jens Axboe --- kernel/trace/blktrace.c | 4 ---- 1 file changed, 4 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/blktrace.c b/kernel/trace/blktrace.c index bc251ed66724..7b8ec0281548 100644 --- a/kernel/trace/blktrace.c +++ b/kernel/trace/blktrace.c @@ -168,7 +168,6 @@ static int act_log_check(struct blk_trace *bt, u32 what, sector_t sector, static const u32 ddir_act[2] = { BLK_TC_ACT(BLK_TC_READ), BLK_TC_ACT(BLK_TC_WRITE) }; -#define BLK_TC_HARDBARRIER BLK_TC_BARRIER #define BLK_TC_RAHEAD BLK_TC_AHEAD /* The ilog2() calls fall out because they're constant */ @@ -196,7 +195,6 @@ static void __blk_add_trace(struct blk_trace *bt, sector_t sector, int bytes, return; what |= ddir_act[rw & WRITE]; - what |= MASK_TC_BIT(rw, HARDBARRIER); what |= MASK_TC_BIT(rw, SYNC); what |= MASK_TC_BIT(rw, RAHEAD); what |= MASK_TC_BIT(rw, META); @@ -1807,8 +1805,6 @@ void blk_fill_rwbs(char *rwbs, u32 rw, int bytes) if (rw & REQ_RAHEAD) rwbs[i++] = 'A'; - if (rw & REQ_HARDBARRIER) - rwbs[i++] = 'B'; if (rw & REQ_SYNC) rwbs[i++] = 'S'; if (rw & REQ_META) -- cgit v1.2.3 From eed01528a45dc4138e9a08064b4b6cc1a9426899 Mon Sep 17 00:00:00 2001 From: Stephane Eranian Date: Tue, 26 Oct 2010 16:08:01 +0200 Subject: perf_events: Fix time tracking in samples This patch corrects time tracking in samples. Without this patch both time_enabled and time_running are bogus when user asks for PERF_SAMPLE_READ. One uses PERF_SAMPLE_READ to sample the values of other counters in each sample. Because of multiplexing, it is necessary to know both time_enabled, time_running to be able to scale counts correctly. In this second version of the patch, we maintain a shadow copy of ctx->time which allows us to compute ctx->time without calling update_context_time() from NMI context. We avoid the issue that update_context_time() must always be called with ctx->lock held. We do not keep shadow copies of the other event timings because if the lead event is overflowing then it is active and thus it's been scheduled in via event_sched_in() in which case neither tstamp_stopped, tstamp_running can be modified. This timing logic only applies to samples when PERF_SAMPLE_READ is used. Note that this patch does not address timing issues related to sampling inheritance between tasks. This will be addressed in a future patch. With this patch, the libpfm4 example task_smpl now reports correct counts (shown on 2.4GHz Core 2): $ task_smpl -p 2400000000 -e unhalted_core_cycles:u,instructions_retired:u,baclears noploop 5 noploop for 5 seconds IIP:0x000000004006d6 PID:5596 TID:5596 TIME:466,210,211,430 STREAM_ID:33 PERIOD:2,400,000,000 ENA=1,010,157,814 RUN=1,010,157,814 NR=3 2,400,000,254 unhalted_core_cycles:u (33) 2,399,273,744 instructions_retired:u (34) 53,340 baclears (35) Signed-off-by: Stephane Eranian Signed-off-by: Peter Zijlstra LKML-Reference: <4cc6e14b.1e07e30a.256e.5190@mx.google.com> Signed-off-by: Ingo Molnar --- kernel/perf_event.c | 42 ++++++++++++++++++++++++++++++++++-------- 1 file changed, 34 insertions(+), 8 deletions(-) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index 517d827f4982..cb6c0d2af68f 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -674,6 +674,8 @@ event_sched_in(struct perf_event *event, event->tstamp_running += ctx->time - event->tstamp_stopped; + event->shadow_ctx_time = ctx->time - ctx->timestamp; + if (!is_software_event(event)) cpuctx->active_oncpu++; ctx->nr_active++; @@ -3396,7 +3398,8 @@ static u32 perf_event_tid(struct perf_event *event, struct task_struct *p) } static void perf_output_read_one(struct perf_output_handle *handle, - struct perf_event *event) + struct perf_event *event, + u64 enabled, u64 running) { u64 read_format = event->attr.read_format; u64 values[4]; @@ -3404,11 +3407,11 @@ static void perf_output_read_one(struct perf_output_handle *handle, values[n++] = perf_event_count(event); if (read_format & PERF_FORMAT_TOTAL_TIME_ENABLED) { - values[n++] = event->total_time_enabled + + values[n++] = enabled + atomic64_read(&event->child_total_time_enabled); } if (read_format & PERF_FORMAT_TOTAL_TIME_RUNNING) { - values[n++] = event->total_time_running + + values[n++] = running + atomic64_read(&event->child_total_time_running); } if (read_format & PERF_FORMAT_ID) @@ -3421,7 +3424,8 @@ static void perf_output_read_one(struct perf_output_handle *handle, * XXX PERF_FORMAT_GROUP vs inherited events seems difficult. */ static void perf_output_read_group(struct perf_output_handle *handle, - struct perf_event *event) + struct perf_event *event, + u64 enabled, u64 running) { struct perf_event *leader = event->group_leader, *sub; u64 read_format = event->attr.read_format; @@ -3431,10 +3435,10 @@ static void perf_output_read_group(struct perf_output_handle *handle, values[n++] = 1 + leader->nr_siblings; if (read_format & PERF_FORMAT_TOTAL_TIME_ENABLED) - values[n++] = leader->total_time_enabled; + values[n++] = enabled; if (read_format & PERF_FORMAT_TOTAL_TIME_RUNNING) - values[n++] = leader->total_time_running; + values[n++] = running; if (leader != event) leader->pmu->read(leader); @@ -3459,13 +3463,35 @@ static void perf_output_read_group(struct perf_output_handle *handle, } } +#define PERF_FORMAT_TOTAL_TIMES (PERF_FORMAT_TOTAL_TIME_ENABLED|\ + PERF_FORMAT_TOTAL_TIME_RUNNING) + static void perf_output_read(struct perf_output_handle *handle, struct perf_event *event) { + u64 enabled = 0, running = 0, now, ctx_time; + u64 read_format = event->attr.read_format; + + /* + * compute total_time_enabled, total_time_running + * based on snapshot values taken when the event + * was last scheduled in. + * + * we cannot simply called update_context_time() + * because of locking issue as we are called in + * NMI context + */ + if (read_format & PERF_FORMAT_TOTAL_TIMES) { + now = perf_clock(); + ctx_time = event->shadow_ctx_time + now; + enabled = ctx_time - event->tstamp_enabled; + running = ctx_time - event->tstamp_running; + } + if (event->attr.read_format & PERF_FORMAT_GROUP) - perf_output_read_group(handle, event); + perf_output_read_group(handle, event, enabled, running); else - perf_output_read_one(handle, event); + perf_output_read_one(handle, event, enabled, running); } void perf_output_sample(struct perf_output_handle *handle, -- cgit v1.2.3 From aae6d3ddd8b90f5b2c8d79a2b914d1706d124193 Mon Sep 17 00:00:00 2001 From: Suresh Siddha Date: Fri, 17 Sep 2010 15:02:32 -0700 Subject: sched: Use group weight, idle cpu metrics to fix imbalances during idle Currently we consider a sched domain to be well balanced when the imbalance is less than the domain's imablance_pct. As the number of cores and threads are increasing, current values of imbalance_pct (for example 25% for a NUMA domain) are not enough to detect imbalances like: a) On a WSM-EP system (two sockets, each having 6 cores and 12 logical threads), 24 cpu-hogging tasks get scheduled as 13 on one socket and 11 on another socket. Leading to an idle HT cpu. b) On a hypothetial 2 socket NHM-EX system (each socket having 8 cores and 16 logical threads), 16 cpu-hogging tasks can get scheduled as 9 on one socket and 7 on another socket. Leaving one core in a socket idle whereas in another socket we have a core having both its HT siblings busy. While this issue can be fixed by decreasing the domain's imbalance_pct (by making it a function of number of logical cpus in the domain), it can potentially cause more task migrations across sched groups in an overloaded case. Fix this by using imbalance_pct only during newly_idle and busy load balancing. And during idle load balancing, check if there is an imbalance in number of idle cpu's across the busiest and this sched_group or if the busiest group has more tasks than its weight that the idle cpu in this_group can pull. Reported-by: Nikhil Rao Signed-off-by: Suresh Siddha Signed-off-by: Peter Zijlstra LKML-Reference: <1284760952.2676.11.camel@sbsiddha-MOBL3.sc.intel.com> Signed-off-by: Ingo Molnar --- kernel/sched.c | 2 ++ kernel/sched_fair.c | 34 +++++++++++++++++++++++++++++++--- 2 files changed, 33 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index aa14a56f9d03..36a088018fe0 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -6960,6 +6960,8 @@ static void init_sched_groups_power(int cpu, struct sched_domain *sd) if (cpu != group_first_cpu(sd->groups)) return; + sd->groups->group_weight = cpumask_weight(sched_group_cpus(sd->groups)); + child = sd->child; sd->groups->cpu_power = 0; diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c index f4f6a8326dd0..034c4f410b36 100644 --- a/kernel/sched_fair.c +++ b/kernel/sched_fair.c @@ -2035,13 +2035,16 @@ struct sd_lb_stats { unsigned long this_load_per_task; unsigned long this_nr_running; unsigned long this_has_capacity; + unsigned int this_idle_cpus; /* Statistics of the busiest group */ + unsigned int busiest_idle_cpus; unsigned long max_load; unsigned long busiest_load_per_task; unsigned long busiest_nr_running; unsigned long busiest_group_capacity; unsigned long busiest_has_capacity; + unsigned int busiest_group_weight; int group_imb; /* Is there imbalance in this sd */ #if defined(CONFIG_SCHED_MC) || defined(CONFIG_SCHED_SMT) @@ -2063,6 +2066,8 @@ struct sg_lb_stats { unsigned long sum_nr_running; /* Nr tasks running in the group */ unsigned long sum_weighted_load; /* Weighted load of group's tasks */ unsigned long group_capacity; + unsigned long idle_cpus; + unsigned long group_weight; int group_imb; /* Is there an imbalance in the group ? */ int group_has_capacity; /* Is there extra capacity in the group? */ }; @@ -2431,7 +2436,8 @@ static inline void update_sg_lb_stats(struct sched_domain *sd, sgs->group_load += load; sgs->sum_nr_running += rq->nr_running; sgs->sum_weighted_load += weighted_cpuload(i); - + if (idle_cpu(i)) + sgs->idle_cpus++; } /* @@ -2469,6 +2475,7 @@ static inline void update_sg_lb_stats(struct sched_domain *sd, sgs->group_capacity = DIV_ROUND_CLOSEST(group->cpu_power, SCHED_LOAD_SCALE); if (!sgs->group_capacity) sgs->group_capacity = fix_small_capacity(sd, group); + sgs->group_weight = group->group_weight; if (sgs->group_capacity > sgs->sum_nr_running) sgs->group_has_capacity = 1; @@ -2576,13 +2583,16 @@ static inline void update_sd_lb_stats(struct sched_domain *sd, int this_cpu, sds->this_nr_running = sgs.sum_nr_running; sds->this_load_per_task = sgs.sum_weighted_load; sds->this_has_capacity = sgs.group_has_capacity; + sds->this_idle_cpus = sgs.idle_cpus; } else if (update_sd_pick_busiest(sd, sds, sg, &sgs, this_cpu)) { sds->max_load = sgs.avg_load; sds->busiest = sg; sds->busiest_nr_running = sgs.sum_nr_running; + sds->busiest_idle_cpus = sgs.idle_cpus; sds->busiest_group_capacity = sgs.group_capacity; sds->busiest_load_per_task = sgs.sum_weighted_load; sds->busiest_has_capacity = sgs.group_has_capacity; + sds->busiest_group_weight = sgs.group_weight; sds->group_imb = sgs.group_imb; } @@ -2860,8 +2870,26 @@ find_busiest_group(struct sched_domain *sd, int this_cpu, if (sds.this_load >= sds.avg_load) goto out_balanced; - if (100 * sds.max_load <= sd->imbalance_pct * sds.this_load) - goto out_balanced; + /* + * In the CPU_NEWLY_IDLE, use imbalance_pct to be conservative. + * And to check for busy balance use !idle_cpu instead of + * CPU_NOT_IDLE. This is because HT siblings will use CPU_NOT_IDLE + * even when they are idle. + */ + if (idle == CPU_NEWLY_IDLE || !idle_cpu(this_cpu)) { + if (100 * sds.max_load <= sd->imbalance_pct * sds.this_load) + goto out_balanced; + } else { + /* + * This cpu is idle. If the busiest group load doesn't + * have more tasks than the number of available cpu's and + * there is no imbalance between this and busiest group + * wrt to idle cpu's, it is balanced. + */ + if ((sds.this_idle_cpus <= sds.busiest_idle_cpus + 1) && + sds.busiest_nr_running <= sds.busiest_group_weight) + goto out_balanced; + } force_balance: /* Looks like there is an imbalance. Compute it */ -- cgit v1.2.3 From 2d46709082c062cae7cce1a15f8cd4cd81b92d88 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Tue, 9 Nov 2010 14:36:52 +0100 Subject: sched: Fix runnable condition for stoptask Heiko reported that the TASK_RUNNING check is not sufficient for CONFIG_PREEMPT=y since we can get preempted with !TASK_RUNNING. He suggested adding a ->se.on_rq test to the existing TASK_RUNNING one, however TASK_RUNNING will always have ->se.on_rq, so we might as well reduce that to a single test. [ stop tasks should never get preempted, but its good to handle this case correctly should this ever happen ] Reported-by: Heiko Carstens Signed-off-by: Peter Zijlstra LKML-Reference: Signed-off-by: Ingo Molnar --- kernel/sched_stoptask.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/sched_stoptask.c b/kernel/sched_stoptask.c index 45bddc0c1048..755483b2a2ad 100644 --- a/kernel/sched_stoptask.c +++ b/kernel/sched_stoptask.c @@ -26,7 +26,7 @@ static struct task_struct *pick_next_task_stop(struct rq *rq) { struct task_struct *stop = rq->stop; - if (stop && stop->state == TASK_RUNNING) + if (stop && stop->se.on_rq) return stop; return NULL; -- cgit v1.2.3 From 43e60861fe9d39740cf5b355f58fecedf0d8e9ba Mon Sep 17 00:00:00 2001 From: Mark Brown Date: Thu, 11 Nov 2010 01:51:26 +0100 Subject: PM / OPP: Hide OPP configuration when SoCs do not provide an implementation Since the OPP API is only useful with an appropraite SoC-specific implementation there is no point in offering the ability to enable the API on general systems. Provide an ARCH_HAS OPP Kconfig symbol which masks out the option unless selected by an implementation. Signed-off-by: Mark Brown Acked-by: Nishanth Menon Acked-by: Kevin Hilman Signed-off-by: Rafael J. Wysocki --- kernel/power/Kconfig | 4 ++++ 1 file changed, 4 insertions(+) (limited to 'kernel') diff --git a/kernel/power/Kconfig b/kernel/power/Kconfig index 29bff6117abc..a5aff3ebad38 100644 --- a/kernel/power/Kconfig +++ b/kernel/power/Kconfig @@ -246,9 +246,13 @@ config PM_OPS depends on PM_SLEEP || PM_RUNTIME default y +config ARCH_HAS_OPP + bool + config PM_OPP bool "Operating Performance Point (OPP) Layer library" depends on PM + depends on ARCH_HAS_OPP ---help--- SOCs have a standard set of tuples consisting of frequency and voltage pairs that the device will support per voltage domain. This -- cgit v1.2.3 From 13b9b6e746d753d43270a78dd39694912646b5d9 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Wed, 10 Nov 2010 22:19:24 -0500 Subject: tracing: Fix module use of trace_bprintk() On use of trace_printk() there's a macro that determines if the format is static or a variable. If it is static, it defaults to __trace_bprintk() otherwise it uses __trace_printk(). A while ago, Lai Jiangshan added __trace_bprintk(). In that patch, we discussed a way to allow modules to use it. The difference between __trace_bprintk() and __trace_printk() is that for faster processing, just the format and args are stored in the trace instead of running it through a sprintf function. In order to do this, the format used by the __trace_bprintk() had to be persistent. See commit 1ba28e02a18cbdbea123836f6c98efb09cbf59ec The problem comes with trace_bprintk() where the module is unloaded. The pointer left in the buffer is still pointing to the format. To solve this issue, the formats in the module were copied into kernel core. If the same format was used, they would use the same copy (to prevent memory leak). This all worked well until we tried to merge everything. At the time this was written, Lai Jiangshan, Frederic Weisbecker, Ingo Molnar and myself were all touching the same code. When this was merged, we lost the part of it that was in module.c. This kept out the copying of the formats and unloading the module could cause bad pointers left in the ring buffer. This patch adds back (with updates required for current kernel) the module code that sets up the necessary pointers. Cc: Lai Jiangshan Cc: Rusty Russell Signed-off-by: Steven Rostedt --- kernel/module.c | 12 ++++++++++++ 1 file changed, 12 insertions(+) (limited to 'kernel') diff --git a/kernel/module.c b/kernel/module.c index 437a74a7524a..d190664f25ff 100644 --- a/kernel/module.c +++ b/kernel/module.c @@ -2326,6 +2326,18 @@ static void find_module_sections(struct module *mod, struct load_info *info) kmemleak_scan_area(mod->trace_events, sizeof(*mod->trace_events) * mod->num_trace_events, GFP_KERNEL); #endif +#ifdef CONFIG_TRACING + mod->trace_bprintk_fmt_start = section_objs(info, "__trace_printk_fmt", + sizeof(*mod->trace_bprintk_fmt_start), + &mod->num_trace_bprintk_fmt); + /* + * This section contains pointers to allocated objects in the trace + * code and not scanning it leads to false positives. + */ + kmemleak_scan_area(mod->trace_bprintk_fmt_start, + sizeof(*mod->trace_bprintk_fmt_start) * + mod->num_trace_bprintk_fmt, GFP_KERNEL); +#endif #ifdef CONFIG_FTRACE_MCOUNT_RECORD /* sechdrs[0].sh_size is always zero */ mod->ftrace_callsites = section_objs(info, "__mcount_loc", -- cgit v1.2.3 From 1e5a74059f9059d330744eac84873b1b99657008 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Sun, 31 Oct 2010 12:37:04 +0100 Subject: sched: Fix cross-sched-class wakeup preemption Instead of dealing with sched classes inside each check_preempt_curr() implementation, pull out this logic into the generic wakeup preemption path. This fixes a hang in KVM (and others) where we are waiting for the stop machine thread to run ... Reported-by: Markus Trippelsdorf Tested-by: Marcelo Tosatti Tested-by: Sergey Senozhatsky Signed-off-by: Peter Zijlstra LKML-Reference: <1288891946.2039.31.camel@laptop> Signed-off-by: Ingo Molnar --- kernel/sched.c | 37 ++++++++++++++++++++++++++----------- kernel/sched_fair.c | 6 ------ kernel/sched_stoptask.c | 2 +- 3 files changed, 27 insertions(+), 18 deletions(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index 36a088018fe0..dc91a4d09ac3 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -560,18 +560,8 @@ struct rq { static DEFINE_PER_CPU_SHARED_ALIGNED(struct rq, runqueues); -static inline -void check_preempt_curr(struct rq *rq, struct task_struct *p, int flags) -{ - rq->curr->sched_class->check_preempt_curr(rq, p, flags); - /* - * A queue event has occurred, and we're going to schedule. In - * this case, we can save a useless back to back clock update. - */ - if (test_tsk_need_resched(p)) - rq->skip_clock_update = 1; -} +static void check_preempt_curr(struct rq *rq, struct task_struct *p, int flags); static inline int cpu_of(struct rq *rq) { @@ -2118,6 +2108,31 @@ static inline void check_class_changed(struct rq *rq, struct task_struct *p, p->sched_class->prio_changed(rq, p, oldprio, running); } +static void check_preempt_curr(struct rq *rq, struct task_struct *p, int flags) +{ + const struct sched_class *class; + + if (p->sched_class == rq->curr->sched_class) { + rq->curr->sched_class->check_preempt_curr(rq, p, flags); + } else { + for_each_class(class) { + if (class == rq->curr->sched_class) + break; + if (class == p->sched_class) { + resched_task(rq->curr); + break; + } + } + } + + /* + * A queue event has occurred, and we're going to schedule. In + * this case, we can save a useless back to back clock update. + */ + if (test_tsk_need_resched(rq->curr)) + rq->skip_clock_update = 1; +} + #ifdef CONFIG_SMP /* * Is this task likely cache-hot: diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c index 034c4f410b36..52ab113d8bb9 100644 --- a/kernel/sched_fair.c +++ b/kernel/sched_fair.c @@ -1654,12 +1654,6 @@ static void check_preempt_wakeup(struct rq *rq, struct task_struct *p, int wake_ struct cfs_rq *cfs_rq = task_cfs_rq(curr); int scale = cfs_rq->nr_running >= sched_nr_latency; - if (unlikely(rt_prio(p->prio))) - goto preempt; - - if (unlikely(p->sched_class != &fair_sched_class)) - return; - if (unlikely(se == pse)) return; diff --git a/kernel/sched_stoptask.c b/kernel/sched_stoptask.c index 755483b2a2ad..2bf6b47058c1 100644 --- a/kernel/sched_stoptask.c +++ b/kernel/sched_stoptask.c @@ -19,7 +19,7 @@ select_task_rq_stop(struct rq *rq, struct task_struct *p, static void check_preempt_curr_stop(struct rq *rq, struct task_struct *p, int flags) { - resched_task(rq->curr); /* we preempt everything */ + /* we're never preempted */ } static struct task_struct *pick_next_task_stop(struct rq *rq) -- cgit v1.2.3 From 3c502e7a0255d82621ff25d60cc816624830497e Mon Sep 17 00:00:00 2001 From: Jason Wessel Date: Thu, 4 Nov 2010 17:33:01 -0500 Subject: perf,hw_breakpoint: Initialize hardware api earlier When using early debugging, the kernel does not initialize the hw_breakpoint API early enough and causes the late initialization of the kernel debugger to fail. The boot arguments are: earlyprintk=vga ekgdboc=kbd kgdbwait Then simply type "go" at the kdb prompt and boot. The kernel will later emit the message: kgdb: Could not allocate hwbreakpoints And at that point the kernel debugger will cease to work correctly. The solution is to initialize the hw_breakpoint at the same time that all the other perf call backs are initialized instead of using a core_initcall() initialization which happens well after the kernel debugger can make use of hardware breakpoints. Signed-off-by: Jason Wessel CC: Frederic Weisbecker CC: Ingo Molnar CC: Peter Zijlstra LKML-Reference: <4CD3396D.1090308@windriver.com> Signed-off-by: Frederic Weisbecker --- kernel/hw_breakpoint.c | 3 +-- kernel/perf_event.c | 6 ++++++ 2 files changed, 7 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/hw_breakpoint.c b/kernel/hw_breakpoint.c index 2c9120f0afca..e5325825aeb6 100644 --- a/kernel/hw_breakpoint.c +++ b/kernel/hw_breakpoint.c @@ -620,7 +620,7 @@ static struct pmu perf_breakpoint = { .read = hw_breakpoint_pmu_read, }; -static int __init init_hw_breakpoint(void) +int __init init_hw_breakpoint(void) { unsigned int **task_bp_pinned; int cpu, err_cpu; @@ -655,6 +655,5 @@ static int __init init_hw_breakpoint(void) return -ENOMEM; } -core_initcall(init_hw_breakpoint); diff --git a/kernel/perf_event.c b/kernel/perf_event.c index 517d827f4982..05b7d8c72c6c 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -31,6 +31,7 @@ #include #include #include +#include #include @@ -6295,6 +6296,8 @@ perf_cpu_notify(struct notifier_block *self, unsigned long action, void *hcpu) void __init perf_event_init(void) { + int ret; + perf_event_init_all_cpus(); init_srcu_struct(&pmus_srcu); perf_pmu_register(&perf_swevent); @@ -6302,4 +6305,7 @@ void __init perf_event_init(void) perf_pmu_register(&perf_task_clock); perf_tp_register(); perf_cpu_notifier(perf_cpu_notify); + + ret = init_hw_breakpoint(); + WARN(ret, "hw_breakpoint initialization failed with: %d", ret); } -- cgit v1.2.3 From 834b40380e93e36f1c9b48ec1d280cebe3d7bd8c Mon Sep 17 00:00:00 2001 From: Alexey Khoroshilov Date: Thu, 11 Nov 2010 14:05:14 -0800 Subject: kernel/range.c: fix clean_sort_range() for the case of full array clean_sort_range() should return a number of nonempty elements of range array, but if the array is full clean_sort_range() returns 0. The problem is that the number of nonempty elements is evaluated by finding the first empty element of the array. If there is no such element it returns an initial value of local variable nr_range that is zero. The fix is trivial: it changes initial value of nr_range to size of the array. The bug can lead to loss of information regarding all ranges, since typically returned value of clean_sort_range() is considered as an actual number of ranges in the array after a series of add/subtract operations. Found by Analytical Verification project of Linux Verification Center (linuxtesting.org), thanks to Alexander Kolosov. Signed-off-by: Alexey Khoroshilov Cc: Yinghai Lu Cc: "H. Peter Anvin" Cc: Geert Uytterhoeven Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/range.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/range.c b/kernel/range.c index 471b66acabb5..37fa9b99ad58 100644 --- a/kernel/range.c +++ b/kernel/range.c @@ -119,7 +119,7 @@ static int cmp_range(const void *x1, const void *x2) int clean_sort_range(struct range *range, int az) { - int i, j, k = az - 1, nr_range = 0; + int i, j, k = az - 1, nr_range = az; for (i = 0; i < k; i++) { if (range[i].end) -- cgit v1.2.3 From 38715258aa2e8cd94bd4aafadc544e5104efd551 Mon Sep 17 00:00:00 2001 From: Ken Chen Date: Thu, 11 Nov 2010 14:05:16 -0800 Subject: latencytop: fix per task accumulator Per task latencytop accumulator prematurely terminates due to erroneous placement of latency_record_count. It should be incremented whenever a new record is allocated instead of increment on every latencytop event. Also fix search iterator to only search known record events instead of blindly searching all pre-allocated space. Signed-off-by: Ken Chen Reviewed-by: Arjan van de Ven Cc: Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/latencytop.c | 17 ++++++++--------- 1 file changed, 8 insertions(+), 9 deletions(-) (limited to 'kernel') diff --git a/kernel/latencytop.c b/kernel/latencytop.c index 877fb306d415..17110a4a4fc2 100644 --- a/kernel/latencytop.c +++ b/kernel/latencytop.c @@ -194,14 +194,7 @@ __account_scheduler_latency(struct task_struct *tsk, int usecs, int inter) account_global_scheduler_latency(tsk, &lat); - /* - * short term hack; if we're > 32 we stop; future we recycle: - */ - tsk->latency_record_count++; - if (tsk->latency_record_count >= LT_SAVECOUNT) - goto out_unlock; - - for (i = 0; i < LT_SAVECOUNT; i++) { + for (i = 0; i < tsk->latency_record_count; i++) { struct latency_record *mylat; int same = 1; @@ -227,8 +220,14 @@ __account_scheduler_latency(struct task_struct *tsk, int usecs, int inter) } } + /* + * short term hack; if we're > 32 we stop; future we recycle: + */ + if (tsk->latency_record_count >= LT_SAVECOUNT) + goto out_unlock; + /* Allocated a new one: */ - i = tsk->latency_record_count; + i = tsk->latency_record_count++; memcpy(&tsk->latency_record[i], &lat, sizeof(struct latency_record)); out_unlock: -- cgit v1.2.3 From eaf06b241b091357e72b76863ba16e89610d31bd Mon Sep 17 00:00:00 2001 From: Dan Rosenberg Date: Thu, 11 Nov 2010 14:05:18 -0800 Subject: Restrict unprivileged access to kernel syslog The kernel syslog contains debugging information that is often useful during exploitation of other vulnerabilities, such as kernel heap addresses. Rather than futilely attempt to sanitize hundreds (or thousands) of printk statements and simultaneously cripple useful debugging functionality, it is far simpler to create an option that prevents unprivileged users from reading the syslog. This patch, loosely based on grsecurity's GRKERNSEC_DMESG, creates the dmesg_restrict sysctl. When set to "0", the default, no restrictions are enforced. When set to "1", only users with CAP_SYS_ADMIN can read the kernel syslog via dmesg(8) or other mechanisms. [akpm@linux-foundation.org: explain the config option in kernel.txt] Signed-off-by: Dan Rosenberg Acked-by: Ingo Molnar Acked-by: Eugene Teo Acked-by: Kees Cook Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/printk.c | 6 ++++++ kernel/sysctl.c | 9 +++++++++ 2 files changed, 15 insertions(+) (limited to 'kernel') diff --git a/kernel/printk.c b/kernel/printk.c index b2ebaee8c377..38e7d5868d60 100644 --- a/kernel/printk.c +++ b/kernel/printk.c @@ -261,6 +261,12 @@ static inline void boot_delay_msec(void) } #endif +#ifdef CONFIG_SECURITY_DMESG_RESTRICT +int dmesg_restrict = 1; +#else +int dmesg_restrict; +#endif + int do_syslog(int type, char __user *buf, int len, bool from_file) { unsigned i, j, limit, count; diff --git a/kernel/sysctl.c b/kernel/sysctl.c index c33a1edb799f..b65bf634035e 100644 --- a/kernel/sysctl.c +++ b/kernel/sysctl.c @@ -703,6 +703,15 @@ static struct ctl_table kern_table[] = { .extra2 = &ten_thousand, }, #endif + { + .procname = "dmesg_restrict", + .data = &dmesg_restrict, + .maxlen = sizeof(int), + .mode = 0644, + .proc_handler = proc_dointvec_minmax, + .extra1 = &zero, + .extra2 = &one, + }, { .procname = "ngroups_max", .data = &ngroups_max, -- cgit v1.2.3 From 91e86e560d0b3ce4c5fc64fd2bbb99f856a30a4e Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Wed, 10 Nov 2010 12:56:12 +0100 Subject: tracing: Fix recursive user stack trace The user stack trace can fault when examining the trace. Which would call the do_page_fault handler, which would trace again, which would do the user stack trace, which would fault and call do_page_fault again ... Thus this is causing a recursive bug. We need to have a recursion detector here. [ Resubmitted by Jiri Olsa ] [ Eric Dumazet recommended using __this_cpu_* instead of __get_cpu_* ] Cc: Eric Dumazet Signed-off-by: Jiri Olsa LKML-Reference: <1289390172-9730-3-git-send-email-jolsa@redhat.com> Signed-off-by: Steven Rostedt --- kernel/trace/trace.c | 19 +++++++++++++++++++ 1 file changed, 19 insertions(+) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 82d9b8106cd0..ee6a7339cf0e 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -1284,6 +1284,8 @@ void trace_dump_stack(void) __ftrace_trace_stack(global_trace.buffer, flags, 3, preempt_count()); } +static DEFINE_PER_CPU(int, user_stack_count); + void ftrace_trace_userstack(struct ring_buffer *buffer, unsigned long flags, int pc) { @@ -1302,6 +1304,18 @@ ftrace_trace_userstack(struct ring_buffer *buffer, unsigned long flags, int pc) if (unlikely(in_nmi())) return; + /* + * prevent recursion, since the user stack tracing may + * trigger other kernel events. + */ + preempt_disable(); + if (__this_cpu_read(user_stack_count)) + goto out; + + __this_cpu_inc(user_stack_count); + + + event = trace_buffer_lock_reserve(buffer, TRACE_USER_STACK, sizeof(*entry), flags, pc); if (!event) @@ -1319,6 +1333,11 @@ ftrace_trace_userstack(struct ring_buffer *buffer, unsigned long flags, int pc) save_stack_trace_user(&trace); if (!filter_check_discard(call, entry, buffer, event)) ring_buffer_unlock_commit(buffer, event); + + __this_cpu_dec(user_stack_count); + + out: + preempt_enable(); } #ifdef UNUSED -- cgit v1.2.3 From 00fafcda1773245a5292f953321ec3f0668c8c28 Mon Sep 17 00:00:00 2001 From: Colin Cross Date: Mon, 15 Nov 2010 22:45:22 +0100 Subject: PM / PM QoS: Fix reversed min and max pm_qos_get_value had min and max reversed, causing all pm_qos requests to have no effect. Signed-off-by: Colin Cross Acked-by: mark Signed-off-by: Rafael J. Wysocki Cc: stable@kernel.org --- kernel/pm_qos_params.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/pm_qos_params.c b/kernel/pm_qos_params.c index c7a8f453919e..aeaa7f846821 100644 --- a/kernel/pm_qos_params.c +++ b/kernel/pm_qos_params.c @@ -121,10 +121,10 @@ static inline int pm_qos_get_value(struct pm_qos_object *o) switch (o->type) { case PM_QOS_MIN: - return plist_last(&o->requests)->prio; + return plist_first(&o->requests)->prio; case PM_QOS_MAX: - return plist_first(&o->requests)->prio; + return plist_last(&o->requests)->prio; default: /* runtime check for not using enum */ -- cgit v1.2.3 From 12b3052c3ee8f508b2c7ee4ddd63ed03423409d8 Mon Sep 17 00:00:00 2001 From: Eric Paris Date: Mon, 15 Nov 2010 18:36:29 -0500 Subject: capabilities/syslog: open code cap_syslog logic to fix build failure The addition of CONFIG_SECURITY_DMESG_RESTRICT resulted in a build failure when CONFIG_PRINTK=n. This is because the capabilities code which used the new option was built even though the variable in question didn't exist. The patch here fixes this by moving the capabilities checks out of the LSM and into the caller. All (known) LSMs should have been calling the capabilities hook already so it actually makes the code organization better to eliminate the hook altogether. Signed-off-by: Eric Paris Acked-by: James Morris Signed-off-by: Linus Torvalds --- kernel/printk.c | 15 ++++++++++++++- 1 file changed, 14 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/printk.c b/kernel/printk.c index 38e7d5868d60..9a2264fc42ca 100644 --- a/kernel/printk.c +++ b/kernel/printk.c @@ -274,7 +274,20 @@ int do_syslog(int type, char __user *buf, int len, bool from_file) char c; int error = 0; - error = security_syslog(type, from_file); + /* + * If this is from /proc/kmsg we only do the capabilities checks + * at open time. + */ + if (type == SYSLOG_ACTION_OPEN || !from_file) { + if (dmesg_restrict && !capable(CAP_SYS_ADMIN)) + return -EPERM; + if ((type != SYSLOG_ACTION_READ_ALL && + type != SYSLOG_ACTION_SIZE_BUFFER) && + !capable(CAP_SYS_ADMIN)) + return -EPERM; + } + + error = security_syslog(type); if (error) return error; -- cgit v1.2.3 From df6e61d4ca268dc8706db38222fde9f04701566c Mon Sep 17 00:00:00 2001 From: Joe Perches Date: Mon, 15 Nov 2010 21:17:27 -0800 Subject: kernel/sysctl.c: Fix build failure with !CONFIG_PRINTK Sigh... Signed-off-by: Joe Perches Acked-by: Eric Paris Signed-off-by: Linus Torvalds --- kernel/sysctl.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/sysctl.c b/kernel/sysctl.c index b65bf634035e..5abfa1518554 100644 --- a/kernel/sysctl.c +++ b/kernel/sysctl.c @@ -702,7 +702,6 @@ static struct ctl_table kern_table[] = { .extra1 = &zero, .extra2 = &ten_thousand, }, -#endif { .procname = "dmesg_restrict", .data = &dmesg_restrict, @@ -712,6 +711,7 @@ static struct ctl_table kern_table[] = { .extra1 = &zero, .extra2 = &one, }, +#endif { .procname = "ngroups_max", .data = &ngroups_max, -- cgit v1.2.3 From 59365d136d205cc20fe666ca7f89b1c5001b0d5a Mon Sep 17 00:00:00 2001 From: Marcus Meissner Date: Tue, 16 Nov 2010 11:46:03 +0100 Subject: kernel: make /proc/kallsyms mode 400 to reduce ease of attacking Making /proc/kallsyms readable only for root by default makes it slightly harder for attackers to write generic kernel exploits by removing one source of knowledge where things are in the kernel. This is the second submit, discussion happened on this on first submit and mostly concerned that this is just one hole of the sieve ... but one of the bigger ones. Changing the permissions of at least System.map and vmlinux is also required to fix the same set, but a packaging issue. Target of this starter patch and follow ups is removing any kind of kernel space address information leak from the kernel. [ Side note: the default of root-only reading is the "safe" value, and it's easy enough to then override at any time after boot. The /proc filesystem allows root to change the permissions with a regular chmod, so you can "revert" this at run-time by simply doing chmod og+r /proc/kallsyms as root if you really want regular users to see the kernel symbols. It does help some tools like "perf" figure them out without any setup, so it may well make sense in some situations. - Linus ] Signed-off-by: Marcus Meissner Acked-by: Tejun Heo Acked-by: Eugene Teo Reviewed-by: Jesper Juhl Signed-off-by: Linus Torvalds --- kernel/kallsyms.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/kallsyms.c b/kernel/kallsyms.c index 6f6d091b5757..a8db2570f99a 100644 --- a/kernel/kallsyms.c +++ b/kernel/kallsyms.c @@ -546,7 +546,7 @@ static const struct file_operations kallsyms_operations = { static int __init kallsyms_init(void) { - proc_create("kallsyms", 0444, NULL, &kallsyms_operations); + proc_create("kallsyms", 0400, NULL, &kallsyms_operations); return 0; } device_initcall(kallsyms_init); -- cgit v1.2.3 From 451a3c24b0135bce54542009b5fde43846c7cf67 Mon Sep 17 00:00:00 2001 From: Arnd Bergmann Date: Wed, 17 Nov 2010 16:26:55 +0100 Subject: BKL: remove extraneous #include The big kernel lock has been removed from all these files at some point, leaving only the #include. Remove this too as a cleanup. Signed-off-by: Arnd Bergmann Signed-off-by: Linus Torvalds --- kernel/trace/trace.c | 1 - 1 file changed, 1 deletion(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 82d9b8106cd0..042084157980 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -17,7 +17,6 @@ #include #include #include -#include #include #include #include -- cgit v1.2.3 From 85e76ab50aecbdc9011806f2f8943450ccb0d93c Mon Sep 17 00:00:00 2001 From: Jovi Zhang Date: Wed, 10 Nov 2010 07:22:18 -0600 Subject: kdb: fix memory leak in kdb_main.c Call kfree in the error path as well as the success path in kdb_ll(). Signed-off-by: Jovi Zhang Signed-off-by: Jason Wessel --- kernel/debug/kdb/kdb_main.c | 13 +++++++------ 1 file changed, 7 insertions(+), 6 deletions(-) (limited to 'kernel') diff --git a/kernel/debug/kdb/kdb_main.c b/kernel/debug/kdb/kdb_main.c index 37755d621924..3ab3feee7840 100644 --- a/kernel/debug/kdb/kdb_main.c +++ b/kernel/debug/kdb/kdb_main.c @@ -2361,7 +2361,7 @@ static int kdb_pid(int argc, const char **argv) */ static int kdb_ll(int argc, const char **argv) { - int diag; + int diag = 0; unsigned long addr; long offset = 0; unsigned long va; @@ -2400,20 +2400,21 @@ static int kdb_ll(int argc, const char **argv) char buf[80]; if (KDB_FLAG(CMD_INTERRUPT)) - return 0; + goto out; sprintf(buf, "%s " kdb_machreg_fmt "\n", command, va); diag = kdb_parse(buf); if (diag) - return diag; + goto out; addr = va + linkoffset; if (kdb_getword(&va, addr, sizeof(va))) - return 0; + goto out; } - kfree(command); - return 0; +out: + kfree(command); + return diag; } static int kdb_kgdb(int argc, const char **argv) -- cgit v1.2.3 From 5450d904054b4ed582793ad6ecb5469f03cc4c46 Mon Sep 17 00:00:00 2001 From: Jovi Zhang Date: Wed, 10 Nov 2010 07:22:18 -0600 Subject: kdb: fix crash when KDB_BASE_CMD_MAX is exceeded When the number of dyanmic kdb commands exceeds KDB_BASE_CMD_MAX, the kernel will fault. Signed-off-by: Jovi Zhang Signed-off-by: Jason Wessel --- kernel/debug/kdb/kdb_main.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) (limited to 'kernel') diff --git a/kernel/debug/kdb/kdb_main.c b/kernel/debug/kdb/kdb_main.c index 3ab3feee7840..a6e729766821 100644 --- a/kernel/debug/kdb/kdb_main.c +++ b/kernel/debug/kdb/kdb_main.c @@ -82,7 +82,7 @@ static kdbtab_t kdb_base_commands[50]; #define for_each_kdbcmd(cmd, num) \ for ((cmd) = kdb_base_commands, (num) = 0; \ num < kdb_max_commands; \ - num == KDB_BASE_CMD_MAX ? cmd = kdb_commands : cmd++, num++) + num++, num == KDB_BASE_CMD_MAX ? cmd = kdb_commands : cmd++) typedef struct _kdbmsg { int km_diag; /* kdb diagnostic */ @@ -646,7 +646,7 @@ static int kdb_defcmd2(const char *cmdstr, const char *argv0) } if (!s->usable) return KDB_NOTIMP; - s->command = kmalloc((s->count + 1) * sizeof(*(s->command)), GFP_KDB); + s->command = kzalloc((s->count + 1) * sizeof(*(s->command)), GFP_KDB); if (!s->command) { kdb_printf("Could not allocate new kdb_defcmd table for %s\n", cmdstr); @@ -2740,13 +2740,13 @@ int kdb_register_repeat(char *cmd, } if (kdb_commands) { memcpy(new, kdb_commands, - kdb_max_commands * sizeof(*new)); + (kdb_max_commands - KDB_BASE_CMD_MAX) * sizeof(*new)); kfree(kdb_commands); } memset(new + kdb_max_commands, 0, kdb_command_extend * sizeof(*new)); kdb_commands = new; - kp = kdb_commands + kdb_max_commands; + kp = kdb_commands + kdb_max_commands - KDB_BASE_CMD_MAX; kdb_max_commands += kdb_command_extend; } -- cgit v1.2.3 From b5482cfa1c95a188b3054fa33274806add91bbe5 Mon Sep 17 00:00:00 2001 From: Alex Shi Date: Tue, 16 Nov 2010 17:34:02 +0800 Subject: sched: Fix volanomark performance regression Commit fab4762 triggers excessive idle balancing, causing a ~30% loss in volanomark throughput. Remove idle balancing throttle reset. Originally-by: Alex Shi Signed-off-by: Mike Galbraith Acked-by: Nikhil Rao Signed-off-by: Peter Zijlstra LKML-Reference: <1289928732.5169.211.camel@maggy.simson.net> Signed-off-by: Ingo Molnar --- kernel/sched_fair.c | 4 ---- 1 file changed, 4 deletions(-) (limited to 'kernel') diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c index 52ab113d8bb9..ba0556dc7c06 100644 --- a/kernel/sched_fair.c +++ b/kernel/sched_fair.c @@ -1758,10 +1758,6 @@ static void pull_task(struct rq *src_rq, struct task_struct *p, set_task_cpu(p, this_cpu); activate_task(this_rq, p, 0); check_preempt_curr(this_rq, p, 0); - - /* re-arm NEWIDLE balancing when moving tasks */ - src_rq->avg_idle = this_rq->avg_idle = 2*sysctl_sched_migration_cost; - this_rq->idle_stamp = 0; } /* -- cgit v1.2.3 From d5ad140bc1505a98c0f040937125bfcbb508078f Mon Sep 17 00:00:00 2001 From: Nikhil Rao Date: Wed, 17 Nov 2010 11:42:04 -0800 Subject: sched: Fix idle balancing An earlier commit reverts idle balancing throttling reset to fix a 30% regression in volanomark throughput. We still need to reset idle_stamp when we pull a task in newidle balance. Reported-by: Alex Shi Signed-off-by: Nikhil Rao Signed-off-by: Peter Zijlstra LKML-Reference: <1290022924-3548-1-git-send-email-ncrao@google.com> Signed-off-by: Ingo Molnar --- kernel/sched_fair.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c index ba0556dc7c06..00ebd7686676 100644 --- a/kernel/sched_fair.c +++ b/kernel/sched_fair.c @@ -3215,8 +3215,10 @@ static void idle_balance(int this_cpu, struct rq *this_rq) interval = msecs_to_jiffies(sd->balance_interval); if (time_after(next_balance, sd->last_balance + interval)) next_balance = sd->last_balance + interval; - if (pulled_task) + if (pulled_task) { + this_rq->idle_stamp = 0; break; + } } raw_spin_lock(&this_rq->lock); -- cgit v1.2.3 From 8882135bcd332f294df5455747ea43ba9e6f77ad Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Tue, 9 Nov 2010 19:01:43 +0100 Subject: perf: Fix owner-list vs exit Oleg noticed that a perf-fd keeping a reference on the creating task leads to a few funny side effects. There's two different aspects to this: - kernel based perf-events, these should not take out a reference on the creating task and appear on the task's event list since they're not bound to fds nor visible to userspace. - fork() and pthread_create(), these can lead to the creating task dying (and thus the task's event-list becomming useless) but keeping the list and ref alive until the event is closed. Combined they lead to malfunction of the ptrace hw_tracepoints. Cure this by not considering kernel based perf_events for the owner-list and destroying the owner-list when the owner dies. Reported-by: Oleg Nesterov Signed-off-by: Peter Zijlstra Acked-by: Oleg Nesterov LKML-Reference: <1289576883.2084.286.camel@laptop> Signed-off-by: Ingo Molnar --- kernel/perf_event.c | 63 +++++++++++++++++++++++++++++++++++++++++++---------- 1 file changed, 51 insertions(+), 12 deletions(-) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index f818d9d2dc93..671f6c8c8a32 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -2235,11 +2235,6 @@ int perf_event_release_kernel(struct perf_event *event) raw_spin_unlock_irq(&ctx->lock); mutex_unlock(&ctx->mutex); - mutex_lock(&event->owner->perf_event_mutex); - list_del_init(&event->owner_entry); - mutex_unlock(&event->owner->perf_event_mutex); - put_task_struct(event->owner); - free_event(event); return 0; @@ -2252,9 +2247,43 @@ EXPORT_SYMBOL_GPL(perf_event_release_kernel); static int perf_release(struct inode *inode, struct file *file) { struct perf_event *event = file->private_data; + struct task_struct *owner; file->private_data = NULL; + rcu_read_lock(); + owner = ACCESS_ONCE(event->owner); + /* + * Matches the smp_wmb() in perf_event_exit_task(). If we observe + * !owner it means the list deletion is complete and we can indeed + * free this event, otherwise we need to serialize on + * owner->perf_event_mutex. + */ + smp_read_barrier_depends(); + if (owner) { + /* + * Since delayed_put_task_struct() also drops the last + * task reference we can safely take a new reference + * while holding the rcu_read_lock(). + */ + get_task_struct(owner); + } + rcu_read_unlock(); + + if (owner) { + mutex_lock(&owner->perf_event_mutex); + /* + * We have to re-check the event->owner field, if it is cleared + * we raced with perf_event_exit_task(), acquiring the mutex + * ensured they're done, and we can proceed with freeing the + * event. + */ + if (event->owner) + list_del_init(&event->owner_entry); + mutex_unlock(&owner->perf_event_mutex); + put_task_struct(owner); + } + return perf_event_release_kernel(event); } @@ -5678,7 +5707,7 @@ SYSCALL_DEFINE5(perf_event_open, mutex_unlock(&ctx->mutex); event->owner = current; - get_task_struct(current); + mutex_lock(¤t->perf_event_mutex); list_add_tail(&event->owner_entry, ¤t->perf_event_list); mutex_unlock(¤t->perf_event_mutex); @@ -5746,12 +5775,6 @@ perf_event_create_kernel_counter(struct perf_event_attr *attr, int cpu, ++ctx->generation; mutex_unlock(&ctx->mutex); - event->owner = current; - get_task_struct(current); - mutex_lock(¤t->perf_event_mutex); - list_add_tail(&event->owner_entry, ¤t->perf_event_list); - mutex_unlock(¤t->perf_event_mutex); - return event; err_free: @@ -5902,8 +5925,24 @@ again: */ void perf_event_exit_task(struct task_struct *child) { + struct perf_event *event, *tmp; int ctxn; + mutex_lock(&child->perf_event_mutex); + list_for_each_entry_safe(event, tmp, &child->perf_event_list, + owner_entry) { + list_del_init(&event->owner_entry); + + /* + * Ensure the list deletion is visible before we clear + * the owner, closes a race against perf_release() where + * we need to serialize on the owner->perf_event_mutex. + */ + smp_wmb(); + event->owner = NULL; + } + mutex_unlock(&child->perf_event_mutex); + for_each_task_context_nr(ctxn) perf_event_exit_task_context(child, ctxn); } -- cgit v1.2.3 From 94e8ba728640dc01375a14e337f3b892bfacbeeb Mon Sep 17 00:00:00 2001 From: Sergio Aguirre Date: Tue, 16 Nov 2010 12:02:47 -0600 Subject: irq_work: Drop cmpxchg() result The compiler warned us about: kernel/irq_work.c: In function 'irq_work_run': kernel/irq_work.c:148: warning: value computed is not used Dropping the cmpxchg() result is indeed weird, but correct - so annotate away the warning. Signed-off-by: Sergio Aguirre Cc: Huang Ying Cc: Martin Schwidefsky Cc: Kyle McMartin Signed-off-by: Peter Zijlstra LKML-Reference: <1289930567-17828-1-git-send-email-saaguirre@ti.com> Signed-off-by: Ingo Molnar --- kernel/irq_work.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/irq_work.c b/kernel/irq_work.c index f16763ff8481..90f881904bb1 100644 --- a/kernel/irq_work.c +++ b/kernel/irq_work.c @@ -145,7 +145,9 @@ void irq_work_run(void) * Clear the BUSY bit and return to the free state if * no-one else claimed it meanwhile. */ - cmpxchg(&entry->next, next_flags(NULL, IRQ_WORK_BUSY), NULL); + (void)cmpxchg(&entry->next, + next_flags(NULL, IRQ_WORK_BUSY), + NULL); } } EXPORT_SYMBOL_GPL(irq_work_run); -- cgit v1.2.3 From 33e0d57f5d2f079104611be9f3fccc27ef2c6b24 Mon Sep 17 00:00:00 2001 From: Linus Torvalds Date: Fri, 19 Nov 2010 11:54:40 -0800 Subject: Revert "kernel: make /proc/kallsyms mode 400 to reduce ease of attacking" This reverts commit 59365d136d205cc20fe666ca7f89b1c5001b0d5a. It turns out that this can break certain existing user land setups. Quoth Sarah Sharp: "On Wednesday, I updated my branch to commit 460781b from linus' tree, and my box would not boot. klogd segfaulted, which stalled the whole system. At first I thought it actually hung the box, but it continued booting after 5 minutes, and I was able to log in. It dropped back to the text console instead of the graphical bootup display for that period of time. dmesg surprisingly still works. I've bisected the problem down to this commit (commit 59365d136d205cc20fe666ca7f89b1c5001b0d5a) The box is running klogd 1.5.5ubuntu3 (from Jaunty). Yes, I know that's old. I read the bit in the commit about changing the permissions of kallsyms after boot, but if I can't boot that doesn't help." So let's just keep the old default, and encourage distributions to do the "chmod -r /proc/kallsyms" in their bootup scripts. This is not worth a kernel option to change default behavior, since it's so easily done in user space. Reported-and-bisected-by: Sarah Sharp Cc: Marcus Meissner Cc: Tejun Heo Cc: Eugene Teo Cc: Jesper Juhl Signed-off-by: Linus Torvalds --- kernel/kallsyms.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/kallsyms.c b/kernel/kallsyms.c index a8db2570f99a..6f6d091b5757 100644 --- a/kernel/kallsyms.c +++ b/kernel/kallsyms.c @@ -546,7 +546,7 @@ static const struct file_operations kallsyms_operations = { static int __init kallsyms_init(void) { - proc_create("kallsyms", 0400, NULL, &kallsyms_operations); + proc_create("kallsyms", 0444, NULL, &kallsyms_operations); return 0; } device_initcall(kallsyms_init); -- cgit v1.2.3 From e5cba24e3f018d4beb6acd101a82483c98f91ce7 Mon Sep 17 00:00:00 2001 From: Hitoshi Mitake Date: Fri, 26 Nov 2010 12:06:44 +0100 Subject: workqueue: check the allocation of system_unbound_wq I found a trivial bug on initialization of workqueue. Current init_workqueues doesn't check the result of allocation of system_unbound_wq, this should be checked like other queues. Signed-off-by: Hitoshi Mitake Cc: Arjan van de Ven Cc: David Howells Signed-off-by: Tejun Heo --- kernel/workqueue.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/workqueue.c b/kernel/workqueue.c index 90db1bd1a978..ca017ce8bc6b 100644 --- a/kernel/workqueue.c +++ b/kernel/workqueue.c @@ -3692,7 +3692,8 @@ static int __init init_workqueues(void) system_nrt_wq = alloc_workqueue("events_nrt", WQ_NON_REENTRANT, 0); system_unbound_wq = alloc_workqueue("events_unbound", WQ_UNBOUND, WQ_UNBOUND_MAX_ACTIVE); - BUG_ON(!system_wq || !system_long_wq || !system_nrt_wq); + BUG_ON(!system_wq || !system_long_wq || !system_nrt_wq || + !system_unbound_wq); return 0; } early_initcall(init_workqueues); -- cgit v1.2.3 From dddd3379a619a4cb8247bfd3c94ca9ae3797aa2e Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Wed, 24 Nov 2010 10:05:55 +0100 Subject: perf: Fix inherit vs. context rotation bug It was found that sometimes children of tasks with inherited events had one extra event. Eventually it turned out to be due to the list rotation no being exclusive with the list iteration in the inheritance code. Cure this by temporarily disabling the rotation while we inherit the events. Signed-off-by: Thomas Gleixner Signed-off-by: Peter Zijlstra LKML-Reference: Cc: Signed-off-by: Ingo Molnar --- kernel/perf_event.c | 22 ++++++++++++++++++++-- 1 file changed, 20 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index 671f6c8c8a32..f365dd8ef8b0 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -1622,8 +1622,12 @@ static void rotate_ctx(struct perf_event_context *ctx) { raw_spin_lock(&ctx->lock); - /* Rotate the first entry last of non-pinned groups */ - list_rotate_left(&ctx->flexible_groups); + /* + * Rotate the first entry last of non-pinned groups. Rotation might be + * disabled by the inheritance code. + */ + if (!ctx->rotate_disable) + list_rotate_left(&ctx->flexible_groups); raw_spin_unlock(&ctx->lock); } @@ -6162,6 +6166,7 @@ int perf_event_init_context(struct task_struct *child, int ctxn) struct perf_event *event; struct task_struct *parent = current; int inherited_all = 1; + unsigned long flags; int ret = 0; child->perf_event_ctxp[ctxn] = NULL; @@ -6202,6 +6207,15 @@ int perf_event_init_context(struct task_struct *child, int ctxn) break; } + /* + * We can't hold ctx->lock when iterating the ->flexible_group list due + * to allocations, but we need to prevent rotation because + * rotate_ctx() will change the list from interrupt context. + */ + raw_spin_lock_irqsave(&parent_ctx->lock, flags); + parent_ctx->rotate_disable = 1; + raw_spin_unlock_irqrestore(&parent_ctx->lock, flags); + list_for_each_entry(event, &parent_ctx->flexible_groups, group_entry) { ret = inherit_task_group(event, parent, parent_ctx, child, ctxn, &inherited_all); @@ -6209,6 +6223,10 @@ int perf_event_init_context(struct task_struct *child, int ctxn) break; } + raw_spin_lock_irqsave(&parent_ctx->lock, flags); + parent_ctx->rotate_disable = 0; + raw_spin_unlock_irqrestore(&parent_ctx->lock, flags); + child_ctx = child->perf_event_ctxp[ctxn]; if (child_ctx && inherited_all) { -- cgit v1.2.3 From ee6dcfa40a50fe12a3ae0fb4d2653c66c3ed6556 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Fri, 26 Nov 2010 13:49:04 +0100 Subject: perf: Fix the software context switch counter Stephane noticed that because the perf_sw_event() call is inside the perf_event_task_sched_out() call it won't get called unless we have a per-task counter. Reported-by: Stephane Eranian Signed-off-by: Peter Zijlstra LKML-Reference: Signed-off-by: Ingo Molnar --- kernel/perf_event.c | 2 -- 1 file changed, 2 deletions(-) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index f365dd8ef8b0..eac7e3364335 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -1287,8 +1287,6 @@ void __perf_event_task_sched_out(struct task_struct *task, { int ctxn; - perf_sw_event(PERF_COUNT_SW_CONTEXT_SWITCHES, 1, 1, NULL, 0); - for_each_task_context_nr(ctxn) perf_event_context_sched_out(task, ctxn, next); } -- cgit v1.2.3 From 49f4138346b3cec2706adff02658fe27ceb1e46f Mon Sep 17 00:00:00 2001 From: Heiko Carstens Date: Fri, 26 Nov 2010 13:42:47 +0100 Subject: printk: Fix wake_up_klogd() vs cpu hotplug wake_up_klogd() may get called from preemptible context but uses __raw_get_cpu_var() to write to a per cpu variable. If it gets preempted between getting the address and writing to it, the cpu in question could be offline if the process gets scheduled back and hence writes to the per cpu data of an offline cpu. This buggy behaviour was introduced with fa33507a "printk: robustify printk, fix #2" which was supposed to fix a "using smp_processor_id() in preemptible" warning. Let's use this_cpu_write() instead which disables preemption and makes sure that the outlined scenario cannot happen. Signed-off-by: Heiko Carstens Acked-by: Eric Dumazet Signed-off-by: Peter Zijlstra LKML-Reference: <20101126124247.GC7023@osiris.boeblingen.de.ibm.com> Signed-off-by: Ingo Molnar --- kernel/printk.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/printk.c b/kernel/printk.c index 9a2264fc42ca..cf7588e93f6f 100644 --- a/kernel/printk.c +++ b/kernel/printk.c @@ -1088,7 +1088,7 @@ int printk_needs_cpu(int cpu) void wake_up_klogd(void) { if (waitqueue_active(&log_wait)) - __raw_get_cpu_var(printk_pending) = 1; + this_cpu_write(printk_pending, 1); } /** -- cgit v1.2.3 From 61ab25447ad6334a74e32f60efb135a3467223f8 Mon Sep 17 00:00:00 2001 From: Heiko Carstens Date: Fri, 26 Nov 2010 13:00:59 +0100 Subject: nohz: Fix printk_needs_cpu() return value on offline cpus This patch fixes a hang observed with 2.6.32 kernels where timers got enqueued on offline cpus. printk_needs_cpu() may return 1 if called on offline cpus. When a cpu gets offlined it schedules the idle process which, before killing its own cpu, will call tick_nohz_stop_sched_tick(). That function in turn will call printk_needs_cpu() in order to check if the local tick can be disabled. On offline cpus this function should naturally return 0 since regardless if the tick gets disabled or not the cpu will be dead short after. That is besides the fact that __cpu_disable() should already have made sure that no interrupts on the offlined cpu will be delivered anyway. In this case it prevents tick_nohz_stop_sched_tick() to call select_nohz_load_balancer(). No idea if that really is a problem. However what made me debug this is that on 2.6.32 the function get_nohz_load_balancer() is used within __mod_timer() to select a cpu on which a timer gets enqueued. If printk_needs_cpu() returns 1 then the nohz_load_balancer cpu doesn't get updated when a cpu gets offlined. It may contain the cpu number of an offline cpu. In turn timers get enqueued on an offline cpu and not very surprisingly they never expire and cause system hangs. This has been observed 2.6.32 kernels. On current kernels __mod_timer() uses get_nohz_timer_target() which doesn't have that problem. However there might be other problems because of the too early exit tick_nohz_stop_sched_tick() in case a cpu goes offline. Easiest way to fix this is just to test if the current cpu is offline and call printk_tick() directly which clears the condition. Alternatively I tried a cpu hotplug notifier which would clear the condition, however between calling the notifier function and printk_needs_cpu() something could have called printk() again and the problem is back again. This seems to be the safest fix. Signed-off-by: Heiko Carstens Signed-off-by: Peter Zijlstra Cc: stable@kernel.org LKML-Reference: <20101126120235.406766476@de.ibm.com> Signed-off-by: Ingo Molnar --- kernel/printk.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'kernel') diff --git a/kernel/printk.c b/kernel/printk.c index cf7588e93f6f..a23315dc4498 100644 --- a/kernel/printk.c +++ b/kernel/printk.c @@ -1082,6 +1082,8 @@ void printk_tick(void) int printk_needs_cpu(int cpu) { + if (unlikely(cpu_is_offline(cpu))) + printk_tick(); return per_cpu(printk_pending, cpu); } -- cgit v1.2.3 From 364829b1263b44aa60383824e4c1289d83d78ca7 Mon Sep 17 00:00:00 2001 From: Slava Pestov Date: Wed, 24 Nov 2010 15:13:16 -0800 Subject: tracing: Fix panic when lseek() called on "trace" opened for writing The file_ops struct for the "trace" special file defined llseek as seq_lseek(). However, if the file was opened for writing only, seq_open() was not called, and the seek would dereference a null pointer, file->private_data. This patch introduces a new wrapper for seq_lseek() which checks if the file descriptor is opened for reading first. If not, it does nothing. Cc: Signed-off-by: Slava Pestov LKML-Reference: <1290640396-24179-1-git-send-email-slavapestov@google.com> Signed-off-by: Steven Rostedt --- kernel/trace/trace.c | 10 +++++++++- 1 file changed, 9 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index ee6a7339cf0e..21db0deb5c7d 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -2339,11 +2339,19 @@ tracing_write_stub(struct file *filp, const char __user *ubuf, return count; } +static loff_t tracing_seek(struct file *file, loff_t offset, int origin) +{ + if (file->f_mode & FMODE_READ) + return seq_lseek(file, offset, origin); + else + return 0; +} + static const struct file_operations tracing_fops = { .open = tracing_open, .read = seq_read, .write = tracing_write_stub, - .llseek = seq_lseek, + .llseek = tracing_seek, .release = tracing_release, }; -- cgit v1.2.3 From 25c9170ed64a6551beefe9315882f754e14486f4 Mon Sep 17 00:00:00 2001 From: Kenji Kaneshige Date: Tue, 30 Nov 2010 17:36:08 +0900 Subject: genirq: Fix incorrect proc spurious output Since commit a1afb637(switch /proc/irq/*/spurious to seq_file) all /proc/irq/XX/spurious files show the information of irq 0. Current irq_spurious_proc_open() passes on NULL as the 3rd argument, which is used as an IRQ number in irq_spurious_proc_show(), to the single_open(). Because of this, all the /proc/irq/XX/spurious file shows IRQ 0 information regardless of the IRQ number. To fix the problem, irq_spurious_proc_open() must pass on the appropreate data (IRQ number) to single_open(). Signed-off-by: Kenji Kaneshige Reviewed-by: Yong Zhang LKML-Reference: <4CF4B778.90604@jp.fujitsu.com> Cc: stable@kernel.org [2.6.33+] Signed-off-by: Thomas Gleixner --- kernel/irq/proc.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/irq/proc.c b/kernel/irq/proc.c index 01b1d3a88983..6c8a2a9f8a7b 100644 --- a/kernel/irq/proc.c +++ b/kernel/irq/proc.c @@ -214,7 +214,7 @@ static int irq_spurious_proc_show(struct seq_file *m, void *v) static int irq_spurious_proc_open(struct inode *inode, struct file *file) { - return single_open(file, irq_spurious_proc_show, NULL); + return single_open(file, irq_spurious_proc_show, PDE(inode)->data); } static const struct file_operations irq_spurious_proc_fops = { -- cgit v1.2.3 From 33dd94ae1ccbfb7bf0fb6c692bc3d1c4269e6177 Mon Sep 17 00:00:00 2001 From: Nelson Elhage Date: Thu, 2 Dec 2010 14:31:21 -0800 Subject: do_exit(): make sure that we run with get_fs() == USER_DS If a user manages to trigger an oops with fs set to KERNEL_DS, fs is not otherwise reset before do_exit(). do_exit may later (via mm_release in fork.c) do a put_user to a user-controlled address, potentially allowing a user to leverage an oops into a controlled write into kernel memory. This is only triggerable in the presence of another bug, but this potentially turns a lot of DoS bugs into privilege escalations, so it's worth fixing. I have proof-of-concept code which uses this bug along with CVE-2010-3849 to write a zero to an arbitrary kernel address, so I've tested that this is not theoretical. A more logical place to put this fix might be when we know an oops has occurred, before we call do_exit(), but that would involve changing every architecture, in multiple places. Let's just stick it in do_exit instead. [akpm@linux-foundation.org: update code comment] Signed-off-by: Nelson Elhage Cc: KOSAKI Motohiro Cc: Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/exit.c | 9 +++++++++ 1 file changed, 9 insertions(+) (limited to 'kernel') diff --git a/kernel/exit.c b/kernel/exit.c index 21aa7b3001fb..676149a4ac5f 100644 --- a/kernel/exit.c +++ b/kernel/exit.c @@ -914,6 +914,15 @@ NORET_TYPE void do_exit(long code) if (unlikely(!tsk->pid)) panic("Attempted to kill the idle task!"); + /* + * If do_exit is called because this processes oopsed, it's possible + * that get_fs() was left as KERNEL_DS, so reset it to USER_DS before + * continuing. Amongst other possible reasons, this is to prevent + * mm_release()->clear_child_tid() from writing to a user-controlled + * kernel address. + */ + set_fs(USER_DS); + tracehook_report_exit(&code); validate_creds_for_do_exit(tsk); -- cgit v1.2.3 From 9f339caf8454f0c21983111350ede93983db4340 Mon Sep 17 00:00:00 2001 From: Bojan Smojver Date: Thu, 25 Nov 2010 23:41:39 +0100 Subject: PM / Hibernate: Use async I/O when reading compressed hibernation image This is a fix for reading LZO compressed image using async I/O. Essentially, instead of having just one page into which we keep reading blocks from swap, we allocate enough of them to cover the largest compressed size and then let block I/O pick them all up. Once we have them all (and here we wait), we decompress them, as usual. Obviously, the very first block we still pick up synchronously, because we need to know the size of the lot before we pick up the rest. Also fixed the copyright line, which I've forgotten before. Signed-off-by: Bojan Smojver Signed-off-by: Rafael J. Wysocki --- kernel/power/swap.c | 53 ++++++++++++++++++++++++++++++++++++++--------------- 1 file changed, 38 insertions(+), 15 deletions(-) (limited to 'kernel') diff --git a/kernel/power/swap.c b/kernel/power/swap.c index a0e4a86ccf94..baf667bb2794 100644 --- a/kernel/power/swap.c +++ b/kernel/power/swap.c @@ -6,6 +6,7 @@ * * Copyright (C) 1998,2001-2005 Pavel Machek * Copyright (C) 2006 Rafael J. Wysocki + * Copyright (C) 2010 Bojan Smojver * * This file is released under the GPLv2. * @@ -753,30 +754,43 @@ static int load_image_lzo(struct swap_map_handle *handle, { unsigned int m; int error = 0; + struct bio *bio; struct timeval start; struct timeval stop; unsigned nr_pages; - size_t off, unc_len, cmp_len; - unsigned char *unc, *cmp, *page; + size_t i, off, unc_len, cmp_len; + unsigned char *unc, *cmp, *page[LZO_CMP_PAGES]; - page = (void *)__get_free_page(__GFP_WAIT | __GFP_HIGH); - if (!page) { - printk(KERN_ERR "PM: Failed to allocate LZO page\n"); - return -ENOMEM; + for (i = 0; i < LZO_CMP_PAGES; i++) { + page[i] = (void *)__get_free_page(__GFP_WAIT | __GFP_HIGH); + if (!page[i]) { + printk(KERN_ERR "PM: Failed to allocate LZO page\n"); + + while (i) + free_page((unsigned long)page[--i]); + + return -ENOMEM; + } } unc = vmalloc(LZO_UNC_SIZE); if (!unc) { printk(KERN_ERR "PM: Failed to allocate LZO uncompressed\n"); - free_page((unsigned long)page); + + for (i = 0; i < LZO_CMP_PAGES; i++) + free_page((unsigned long)page[i]); + return -ENOMEM; } cmp = vmalloc(LZO_CMP_SIZE); if (!cmp) { printk(KERN_ERR "PM: Failed to allocate LZO compressed\n"); + vfree(unc); - free_page((unsigned long)page); + for (i = 0; i < LZO_CMP_PAGES; i++) + free_page((unsigned long)page[i]); + return -ENOMEM; } @@ -787,6 +801,7 @@ static int load_image_lzo(struct swap_map_handle *handle, if (!m) m = 1; nr_pages = 0; + bio = NULL; do_gettimeofday(&start); error = snapshot_write_next(snapshot); @@ -794,11 +809,11 @@ static int load_image_lzo(struct swap_map_handle *handle, goto out_finish; for (;;) { - error = swap_read_page(handle, page, NULL); /* sync */ + error = swap_read_page(handle, page[0], NULL); /* sync */ if (error) break; - cmp_len = *(size_t *)page; + cmp_len = *(size_t *)page[0]; if (unlikely(!cmp_len || cmp_len > lzo1x_worst_compress(LZO_UNC_SIZE))) { printk(KERN_ERR "PM: Invalid LZO compressed length\n"); @@ -806,13 +821,20 @@ static int load_image_lzo(struct swap_map_handle *handle, break; } - memcpy(cmp, page, PAGE_SIZE); - for (off = PAGE_SIZE; off < LZO_HEADER + cmp_len; off += PAGE_SIZE) { - error = swap_read_page(handle, page, NULL); /* sync */ + for (off = PAGE_SIZE, i = 1; + off < LZO_HEADER + cmp_len; off += PAGE_SIZE, i++) { + error = swap_read_page(handle, page[i], &bio); if (error) goto out_finish; + } - memcpy(cmp + off, page, PAGE_SIZE); + error = hib_wait_on_bio_chain(&bio); /* need all data now */ + if (error) + goto out_finish; + + for (off = 0, i = 0; + off < LZO_HEADER + cmp_len; off += PAGE_SIZE, i++) { + memcpy(cmp + off, page[i], PAGE_SIZE); } unc_len = LZO_UNC_SIZE; @@ -857,7 +879,8 @@ out_finish: vfree(cmp); vfree(unc); - free_page((unsigned long)page); + for (i = 0; i < LZO_CMP_PAGES; i++) + free_page((unsigned long)page[i]); return error; } -- cgit v1.2.3 From c9e664f1fdf34aa8cede047b206deaa8f1945af0 Mon Sep 17 00:00:00 2001 From: "Rafael J. Wysocki" Date: Fri, 3 Dec 2010 22:57:45 +0100 Subject: PM / Hibernate: Fix memory corruption related to swap There is a problem that swap pages allocated before the creation of a hibernation image can be released and used for storing the contents of different memory pages while the image is being saved. Since the kernel stored in the image doesn't know of that, it causes memory corruption to occur after resume from hibernation, especially on systems with relatively small RAM that need to swap often. This issue can be addressed by keeping the GFP_IOFS bits clear in gfp_allowed_mask during the entire hibernation, including the saving of the image, until the system is finally turned off or the hibernation is aborted. Unfortunately, for this purpose it's necessary to rework the way in which the hibernate and suspend code manipulates gfp_allowed_mask. This change is based on an earlier patch from Hugh Dickins. Signed-off-by: Rafael J. Wysocki Reported-by: Ondrej Zary Acked-by: Hugh Dickins Reviewed-by: KAMEZAWA Hiroyuki Cc: stable@kernel.org --- kernel/power/hibernate.c | 22 ++++++++++++---------- kernel/power/suspend.c | 5 ++--- kernel/power/user.c | 2 ++ 3 files changed, 16 insertions(+), 13 deletions(-) (limited to 'kernel') diff --git a/kernel/power/hibernate.c b/kernel/power/hibernate.c index 657272e91d0a..048d0b514831 100644 --- a/kernel/power/hibernate.c +++ b/kernel/power/hibernate.c @@ -327,7 +327,6 @@ static int create_image(int platform_mode) int hibernation_snapshot(int platform_mode) { int error; - gfp_t saved_mask; error = platform_begin(platform_mode); if (error) @@ -339,7 +338,7 @@ int hibernation_snapshot(int platform_mode) goto Close; suspend_console(); - saved_mask = clear_gfp_allowed_mask(GFP_IOFS); + pm_restrict_gfp_mask(); error = dpm_suspend_start(PMSG_FREEZE); if (error) goto Recover_platform; @@ -348,7 +347,10 @@ int hibernation_snapshot(int platform_mode) goto Recover_platform; error = create_image(platform_mode); - /* Control returns here after successful restore */ + /* + * Control returns here (1) after the image has been created or the + * image creation has failed and (2) after a successful restore. + */ Resume_devices: /* We may need to release the preallocated image pages here. */ @@ -357,7 +359,10 @@ int hibernation_snapshot(int platform_mode) dpm_resume_end(in_suspend ? (error ? PMSG_RECOVER : PMSG_THAW) : PMSG_RESTORE); - set_gfp_allowed_mask(saved_mask); + + if (error || !in_suspend) + pm_restore_gfp_mask(); + resume_console(); Close: platform_end(platform_mode); @@ -452,17 +457,16 @@ static int resume_target_kernel(bool platform_mode) int hibernation_restore(int platform_mode) { int error; - gfp_t saved_mask; pm_prepare_console(); suspend_console(); - saved_mask = clear_gfp_allowed_mask(GFP_IOFS); + pm_restrict_gfp_mask(); error = dpm_suspend_start(PMSG_QUIESCE); if (!error) { error = resume_target_kernel(platform_mode); dpm_resume_end(PMSG_RECOVER); } - set_gfp_allowed_mask(saved_mask); + pm_restore_gfp_mask(); resume_console(); pm_restore_console(); return error; @@ -476,7 +480,6 @@ int hibernation_restore(int platform_mode) int hibernation_platform_enter(void) { int error; - gfp_t saved_mask; if (!hibernation_ops) return -ENOSYS; @@ -492,7 +495,6 @@ int hibernation_platform_enter(void) entering_platform_hibernation = true; suspend_console(); - saved_mask = clear_gfp_allowed_mask(GFP_IOFS); error = dpm_suspend_start(PMSG_HIBERNATE); if (error) { if (hibernation_ops->recover) @@ -536,7 +538,6 @@ int hibernation_platform_enter(void) Resume_devices: entering_platform_hibernation = false; dpm_resume_end(PMSG_RESTORE); - set_gfp_allowed_mask(saved_mask); resume_console(); Close: @@ -646,6 +647,7 @@ int hibernate(void) swsusp_free(); if (!error) power_down(); + pm_restore_gfp_mask(); } else { pr_debug("PM: Image restored successfully.\n"); } diff --git a/kernel/power/suspend.c b/kernel/power/suspend.c index 7335952ee473..ecf770509d0d 100644 --- a/kernel/power/suspend.c +++ b/kernel/power/suspend.c @@ -197,7 +197,6 @@ static int suspend_enter(suspend_state_t state) int suspend_devices_and_enter(suspend_state_t state) { int error; - gfp_t saved_mask; if (!suspend_ops) return -ENOSYS; @@ -208,7 +207,7 @@ int suspend_devices_and_enter(suspend_state_t state) goto Close; } suspend_console(); - saved_mask = clear_gfp_allowed_mask(GFP_IOFS); + pm_restrict_gfp_mask(); suspend_test_start(); error = dpm_suspend_start(PMSG_SUSPEND); if (error) { @@ -225,7 +224,7 @@ int suspend_devices_and_enter(suspend_state_t state) suspend_test_start(); dpm_resume_end(PMSG_RESUME); suspend_test_finish("resume devices"); - set_gfp_allowed_mask(saved_mask); + pm_restore_gfp_mask(); resume_console(); Close: if (suspend_ops->end) diff --git a/kernel/power/user.c b/kernel/power/user.c index e819e17877ca..1b2ea31e6bd8 100644 --- a/kernel/power/user.c +++ b/kernel/power/user.c @@ -263,6 +263,7 @@ static long snapshot_ioctl(struct file *filp, unsigned int cmd, case SNAPSHOT_UNFREEZE: if (!data->frozen || data->ready) break; + pm_restore_gfp_mask(); thaw_processes(); usermodehelper_enable(); data->frozen = 0; @@ -275,6 +276,7 @@ static long snapshot_ioctl(struct file *filp, unsigned int cmd, error = -EPERM; break; } + pm_restore_gfp_mask(); error = hibernation_snapshot(data->platform_support); if (!error) error = put_user(in_suspend, (int __user *)arg); -- cgit v1.2.3 From 5167695753c63444a9e6cbbef136200a16c7a225 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Tue, 7 Dec 2010 14:18:20 +0100 Subject: perf: Fix duplicate events with multiple-pmu vs software events Because the multi-pmu bits can share contexts between struct pmu instances we could get duplicate events by iterating the pmu list. Signed-off-by: Peter Zijlstra Signed-off-by: Thomas Gleixner LKML-Reference: Signed-off-by: Ingo Molnar --- kernel/perf_event.c | 35 +++++++++++++++++++++++++++++------ 1 file changed, 29 insertions(+), 6 deletions(-) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index eac7e3364335..7b870174c56d 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -3824,6 +3824,8 @@ static void perf_event_task_event(struct perf_task_event *task_event) rcu_read_lock(); list_for_each_entry_rcu(pmu, &pmus, entry) { cpuctx = get_cpu_ptr(pmu->pmu_cpu_context); + if (cpuctx->active_pmu != pmu) + goto next; perf_event_task_ctx(&cpuctx->ctx, task_event); ctx = task_event->task_ctx; @@ -3959,6 +3961,8 @@ static void perf_event_comm_event(struct perf_comm_event *comm_event) rcu_read_lock(); list_for_each_entry_rcu(pmu, &pmus, entry) { cpuctx = get_cpu_ptr(pmu->pmu_cpu_context); + if (cpuctx->active_pmu != pmu) + goto next; perf_event_comm_ctx(&cpuctx->ctx, comm_event); ctxn = pmu->task_ctx_nr; @@ -4144,6 +4148,8 @@ got_name: rcu_read_lock(); list_for_each_entry_rcu(pmu, &pmus, entry) { cpuctx = get_cpu_ptr(pmu->pmu_cpu_context); + if (cpuctx->active_pmu != pmu) + goto next; perf_event_mmap_ctx(&cpuctx->ctx, mmap_event, vma->vm_flags & VM_EXEC); @@ -5145,20 +5151,36 @@ static void *find_pmu_context(int ctxn) return NULL; } -static void free_pmu_context(void * __percpu cpu_context) +static void update_pmu_context(struct pmu *pmu, struct pmu *old_pmu) { - struct pmu *pmu; + int cpu; + + for_each_possible_cpu(cpu) { + struct perf_cpu_context *cpuctx; + + cpuctx = per_cpu_ptr(pmu->pmu_cpu_context, cpu); + + if (cpuctx->active_pmu == old_pmu) + cpuctx->active_pmu = pmu; + } +} + +static void free_pmu_context(struct pmu *pmu) +{ + struct pmu *i; mutex_lock(&pmus_lock); /* * Like a real lame refcount. */ - list_for_each_entry(pmu, &pmus, entry) { - if (pmu->pmu_cpu_context == cpu_context) + list_for_each_entry(i, &pmus, entry) { + if (i->pmu_cpu_context == pmu->pmu_cpu_context) { + update_pmu_context(i, pmu); goto out; + } } - free_percpu(cpu_context); + free_percpu(pmu->pmu_cpu_context); out: mutex_unlock(&pmus_lock); } @@ -5190,6 +5212,7 @@ int perf_pmu_register(struct pmu *pmu) cpuctx->ctx.pmu = pmu; cpuctx->jiffies_interval = 1; INIT_LIST_HEAD(&cpuctx->rotation_list); + cpuctx->active_pmu = pmu; } got_cpu_context: @@ -5241,7 +5264,7 @@ void perf_pmu_unregister(struct pmu *pmu) synchronize_rcu(); free_percpu(pmu->pmu_disable_count); - free_pmu_context(pmu->pmu_cpu_context); + free_pmu_context(pmu); } struct pmu *perf_init_event(struct perf_event *event) -- cgit v1.2.3 From 0f004f5a696a9434b7214d0d3cbd0525ee77d428 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Tue, 30 Nov 2010 19:48:45 +0100 Subject: sched: Cure more NO_HZ load average woes There's a long-running regression that proved difficult to fix and which is hitting certain people and is rather annoying in its effects. Damien reported that after 74f5187ac8 (sched: Cure load average vs NO_HZ woes) his load average is unnaturally high, he also noted that even with that patch reverted the load avgerage numbers are not correct. The problem is that the previous patch only solved half the NO_HZ problem, it addressed the part of going into NO_HZ mode, not of comming out of NO_HZ mode. This patch implements that missing half. When comming out of NO_HZ mode there are two important things to take care of: - Folding the pending idle delta into the global active count. - Correctly aging the averages for the idle-duration. So with this patch the NO_HZ interaction should be complete and behaviour between CONFIG_NO_HZ=[yn] should be equivalent. Furthermore, this patch slightly changes the load average computation by adding a rounding term to the fixed point multiplication. Reported-by: Damien Wyart Reported-by: Tim McGrath Tested-by: Damien Wyart Tested-by: Orion Poplawski Tested-by: Kyle McMartin Signed-off-by: Peter Zijlstra Cc: stable@kernel.org Cc: Chase Douglas LKML-Reference: <1291129145.32004.874.camel@laptop> Signed-off-by: Ingo Molnar --- kernel/sched.c | 150 ++++++++++++++++++++++++++++++++++++++++++++++++++++----- kernel/timer.c | 2 +- 2 files changed, 140 insertions(+), 12 deletions(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index dc91a4d09ac3..6b7c26a1a097 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -3119,6 +3119,15 @@ static long calc_load_fold_active(struct rq *this_rq) return delta; } +static unsigned long +calc_load(unsigned long load, unsigned long exp, unsigned long active) +{ + load *= exp; + load += active * (FIXED_1 - exp); + load += 1UL << (FSHIFT - 1); + return load >> FSHIFT; +} + #ifdef CONFIG_NO_HZ /* * For NO_HZ we delay the active fold to the next LOAD_FREQ update. @@ -3148,6 +3157,128 @@ static long calc_load_fold_idle(void) return delta; } + +/** + * fixed_power_int - compute: x^n, in O(log n) time + * + * @x: base of the power + * @frac_bits: fractional bits of @x + * @n: power to raise @x to. + * + * By exploiting the relation between the definition of the natural power + * function: x^n := x*x*...*x (x multiplied by itself for n times), and + * the binary encoding of numbers used by computers: n := \Sum n_i * 2^i, + * (where: n_i \elem {0, 1}, the binary vector representing n), + * we find: x^n := x^(\Sum n_i * 2^i) := \Prod x^(n_i * 2^i), which is + * of course trivially computable in O(log_2 n), the length of our binary + * vector. + */ +static unsigned long +fixed_power_int(unsigned long x, unsigned int frac_bits, unsigned int n) +{ + unsigned long result = 1UL << frac_bits; + + if (n) for (;;) { + if (n & 1) { + result *= x; + result += 1UL << (frac_bits - 1); + result >>= frac_bits; + } + n >>= 1; + if (!n) + break; + x *= x; + x += 1UL << (frac_bits - 1); + x >>= frac_bits; + } + + return result; +} + +/* + * a1 = a0 * e + a * (1 - e) + * + * a2 = a1 * e + a * (1 - e) + * = (a0 * e + a * (1 - e)) * e + a * (1 - e) + * = a0 * e^2 + a * (1 - e) * (1 + e) + * + * a3 = a2 * e + a * (1 - e) + * = (a0 * e^2 + a * (1 - e) * (1 + e)) * e + a * (1 - e) + * = a0 * e^3 + a * (1 - e) * (1 + e + e^2) + * + * ... + * + * an = a0 * e^n + a * (1 - e) * (1 + e + ... + e^n-1) [1] + * = a0 * e^n + a * (1 - e) * (1 - e^n)/(1 - e) + * = a0 * e^n + a * (1 - e^n) + * + * [1] application of the geometric series: + * + * n 1 - x^(n+1) + * S_n := \Sum x^i = ------------- + * i=0 1 - x + */ +static unsigned long +calc_load_n(unsigned long load, unsigned long exp, + unsigned long active, unsigned int n) +{ + + return calc_load(load, fixed_power_int(exp, FSHIFT, n), active); +} + +/* + * NO_HZ can leave us missing all per-cpu ticks calling + * calc_load_account_active(), but since an idle CPU folds its delta into + * calc_load_tasks_idle per calc_load_account_idle(), all we need to do is fold + * in the pending idle delta if our idle period crossed a load cycle boundary. + * + * Once we've updated the global active value, we need to apply the exponential + * weights adjusted to the number of cycles missed. + */ +static void calc_global_nohz(unsigned long ticks) +{ + long delta, active, n; + + if (time_before(jiffies, calc_load_update)) + return; + + /* + * If we crossed a calc_load_update boundary, make sure to fold + * any pending idle changes, the respective CPUs might have + * missed the tick driven calc_load_account_active() update + * due to NO_HZ. + */ + delta = calc_load_fold_idle(); + if (delta) + atomic_long_add(delta, &calc_load_tasks); + + /* + * If we were idle for multiple load cycles, apply them. + */ + if (ticks >= LOAD_FREQ) { + n = ticks / LOAD_FREQ; + + active = atomic_long_read(&calc_load_tasks); + active = active > 0 ? active * FIXED_1 : 0; + + avenrun[0] = calc_load_n(avenrun[0], EXP_1, active, n); + avenrun[1] = calc_load_n(avenrun[1], EXP_5, active, n); + avenrun[2] = calc_load_n(avenrun[2], EXP_15, active, n); + + calc_load_update += n * LOAD_FREQ; + } + + /* + * Its possible the remainder of the above division also crosses + * a LOAD_FREQ period, the regular check in calc_global_load() + * which comes after this will take care of that. + * + * Consider us being 11 ticks before a cycle completion, and us + * sleeping for 4*LOAD_FREQ + 22 ticks, then the above code will + * age us 4 cycles, and the test in calc_global_load() will + * pick up the final one. + */ +} #else static void calc_load_account_idle(struct rq *this_rq) { @@ -3157,6 +3288,10 @@ static inline long calc_load_fold_idle(void) { return 0; } + +static void calc_global_nohz(unsigned long ticks) +{ +} #endif /** @@ -3174,24 +3309,17 @@ void get_avenrun(unsigned long *loads, unsigned long offset, int shift) loads[2] = (avenrun[2] + offset) << shift; } -static unsigned long -calc_load(unsigned long load, unsigned long exp, unsigned long active) -{ - load *= exp; - load += active * (FIXED_1 - exp); - return load >> FSHIFT; -} - /* * calc_load - update the avenrun load estimates 10 ticks after the * CPUs have updated calc_load_tasks. */ -void calc_global_load(void) +void calc_global_load(unsigned long ticks) { - unsigned long upd = calc_load_update + 10; long active; - if (time_before(jiffies, upd)) + calc_global_nohz(ticks); + + if (time_before(jiffies, calc_load_update + 10)) return; active = atomic_long_read(&calc_load_tasks); diff --git a/kernel/timer.c b/kernel/timer.c index 68a9ae7679b7..7bd715fda974 100644 --- a/kernel/timer.c +++ b/kernel/timer.c @@ -1319,7 +1319,7 @@ void do_timer(unsigned long ticks) { jiffies_64 += ticks; update_wall_time(); - calc_global_load(); + calc_global_load(ticks); } #ifdef __ARCH_WANT_SYS_ALARM -- cgit v1.2.3 From f26f9aff6aaf67e9a430d16c266f91b13a5bff64 Mon Sep 17 00:00:00 2001 From: Mike Galbraith Date: Wed, 8 Dec 2010 11:05:42 +0100 Subject: Sched: fix skip_clock_update optimization idle_balance() drops/retakes rq->lock, leaving the previous task vulnerable to set_tsk_need_resched(). Clear it after we return from balancing instead, and in setup_thread_stack() as well, so no successfully descheduled or never scheduled task has it set. Need resched confused the skip_clock_update logic, which assumes that the next call to update_rq_clock() will come nearly immediately after being set. Make the optimization robust against the waking a sleeper before it sucessfully deschedules case by checking that the current task has not been dequeued before setting the flag, since it is that useless clock update we're trying to save, and clear unconditionally in schedule() proper instead of conditionally in put_prev_task(). Signed-off-by: Mike Galbraith Reported-by: Bjoern B. Brandenburg Tested-by: Yong Zhang Signed-off-by: Peter Zijlstra Cc: stable@kernel.org LKML-Reference: <1291802742.1417.9.camel@marge.simson.net> Signed-off-by: Ingo Molnar --- kernel/fork.c | 1 + kernel/sched.c | 26 ++++++++++++++------------ 2 files changed, 15 insertions(+), 12 deletions(-) (limited to 'kernel') diff --git a/kernel/fork.c b/kernel/fork.c index 3b159c5991b7..5447dc7defa9 100644 --- a/kernel/fork.c +++ b/kernel/fork.c @@ -273,6 +273,7 @@ static struct task_struct *dup_task_struct(struct task_struct *orig) setup_thread_stack(tsk, orig); clear_user_return_notifier(tsk); + clear_tsk_need_resched(tsk); stackend = end_of_stack(tsk); *stackend = STACK_END_MAGIC; /* for overflow detection */ diff --git a/kernel/sched.c b/kernel/sched.c index 6b7c26a1a097..da14302a9857 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -641,17 +641,18 @@ static void sched_irq_time_avg_update(struct rq *rq, u64 irq_time); inline void update_rq_clock(struct rq *rq) { - if (!rq->skip_clock_update) { - int cpu = cpu_of(rq); - u64 irq_time; + int cpu = cpu_of(rq); + u64 irq_time; - rq->clock = sched_clock_cpu(cpu); - irq_time = irq_time_cpu(cpu); - if (rq->clock - irq_time > rq->clock_task) - rq->clock_task = rq->clock - irq_time; + if (rq->skip_clock_update) + return; - sched_irq_time_avg_update(rq, irq_time); - } + rq->clock = sched_clock_cpu(cpu); + irq_time = irq_time_cpu(cpu); + if (rq->clock - irq_time > rq->clock_task) + rq->clock_task = rq->clock - irq_time; + + sched_irq_time_avg_update(rq, irq_time); } /* @@ -2129,7 +2130,7 @@ static void check_preempt_curr(struct rq *rq, struct task_struct *p, int flags) * A queue event has occurred, and we're going to schedule. In * this case, we can save a useless back to back clock update. */ - if (test_tsk_need_resched(rq->curr)) + if (rq->curr->se.on_rq && test_tsk_need_resched(rq->curr)) rq->skip_clock_update = 1; } @@ -3973,7 +3974,6 @@ static void put_prev_task(struct rq *rq, struct task_struct *prev) { if (prev->se.on_rq) update_rq_clock(rq); - rq->skip_clock_update = 0; prev->sched_class->put_prev_task(rq, prev); } @@ -4031,7 +4031,6 @@ need_resched_nonpreemptible: hrtick_clear(rq); raw_spin_lock_irq(&rq->lock); - clear_tsk_need_resched(prev); switch_count = &prev->nivcsw; if (prev->state && !(preempt_count() & PREEMPT_ACTIVE)) { @@ -4063,6 +4062,8 @@ need_resched_nonpreemptible: put_prev_task(rq, prev); next = pick_next_task(rq); + clear_tsk_need_resched(prev); + rq->skip_clock_update = 0; if (likely(prev != next)) { sched_info_switch(prev, next); @@ -4071,6 +4072,7 @@ need_resched_nonpreemptible: rq->nr_switches++; rq->curr = next; ++*switch_count; + WARN_ON_ONCE(test_tsk_need_resched(next)); context_switch(rq, prev, next); /* unlocks the rq */ /* -- cgit v1.2.3 From dbd87b5af055a0cc9bba17795c9a2b0d17795389 Mon Sep 17 00:00:00 2001 From: Heiko Carstens Date: Wed, 1 Dec 2010 10:11:09 +0100 Subject: nohz: Fix get_next_timer_interrupt() vs cpu hotplug This fixes a bug as seen on 2.6.32 based kernels where timers got enqueued on offline cpus. If a cpu goes offline it might still have pending timers. These will be migrated during CPU_DEAD handling after the cpu is offline. However while the cpu is going offline it will schedule the idle task which will then call tick_nohz_stop_sched_tick(). That function in turn will call get_next_timer_intterupt() to figure out if the tick of the cpu can be stopped or not. If it turns out that the next tick is just one jiffy off (delta_jiffies == 1) tick_nohz_stop_sched_tick() incorrectly assumes that the tick should not stop and takes an early exit and thus it won't update the load balancer cpu. Just afterwards the cpu will be killed and the load balancer cpu could be the offline cpu. On 2.6.32 based kernel get_nohz_load_balancer() gets called to decide on which cpu a timer should be enqueued (see __mod_timer()). Which leads to the possibility that timers get enqueued on an offline cpu. These will never expire and can cause a system hang. This has been observed 2.6.32 kernels. On current kernels __mod_timer() uses get_nohz_timer_target() which doesn't have that problem. However there might be other problems because of the too early exit tick_nohz_stop_sched_tick() in case a cpu goes offline. The easiest and probably safest fix seems to be to let get_next_timer_interrupt() just lie and let it say there isn't any pending timer if the current cpu is offline. I also thought of moving migrate_[hr]timers() from CPU_DEAD to CPU_DYING, but seeing that there already have been fixes at least in the hrtimer code in this area I'm afraid that this could add new subtle bugs. Signed-off-by: Heiko Carstens Signed-off-by: Peter Zijlstra LKML-Reference: <20101201091109.GA8984@osiris.boeblingen.de.ibm.com> Cc: stable@kernel.org Signed-off-by: Ingo Molnar --- kernel/timer.c | 6 ++++++ 1 file changed, 6 insertions(+) (limited to 'kernel') diff --git a/kernel/timer.c b/kernel/timer.c index 7bd715fda974..353b9227c2ec 100644 --- a/kernel/timer.c +++ b/kernel/timer.c @@ -1252,6 +1252,12 @@ unsigned long get_next_timer_interrupt(unsigned long now) struct tvec_base *base = __get_cpu_var(tvec_bases); unsigned long expires; + /* + * Pretend that there is no timer pending if the cpu is offline. + * Possible pending timers will be migrated later to an active cpu. + */ + if (cpu_is_offline(smp_processor_id())) + return now + NEXT_TIMER_MAX_DELTA; spin_lock(&base->lock); if (time_before_eq(base->next_timer, base->timer_jiffies)) base->next_timer = __next_timer_interrupt(base); -- cgit v1.2.3 From 2d64672ed38721b7a3815009d79bfb90a1f34a17 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Fri, 3 Dec 2010 23:12:33 -0500 Subject: workqueue: It is likely that WORKER_NOT_RUNNING is true Running the annotate branch profiler on three boxes, including my main box that runs firefox, evolution, xchat, and is part of the distcc farm, showed this with the likelys in the workqueue code: correct incorrect % Function File Line ------- --------- - -------- ---- ---- 96 996253 99 wq_worker_sleeping workqueue.c 703 96 996247 99 wq_worker_waking_up workqueue.c 677 The likely()s in this case were assuming that WORKER_NOT_RUNNING will most likely be false. But this is not the case. The reason is (and shown by adding trace_printks and testing it) that most of the time WORKER_PREP is set. In worker_thread() we have: worker_clr_flags(worker, WORKER_PREP); [ do work stuff ] worker_set_flags(worker, WORKER_PREP, false); (that 'false' means not to wake up an idle worker) The wq_worker_sleeping() is called from schedule when a worker thread is putting itself to sleep. Which happens most of the time outside of that [ do work stuff ]. The wq_worker_waking_up is called by the wakeup worker code, which is also callod outside that [ do work stuff ]. Thus, the likely and unlikely used by those two functions are actually backwards. Remove the annotation and let gcc figure it out. Acked-by: Tejun Heo Signed-off-by: Steven Rostedt Signed-off-by: Tejun Heo --- kernel/workqueue.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/workqueue.c b/kernel/workqueue.c index ca017ce8bc6b..e785b0f2aea5 100644 --- a/kernel/workqueue.c +++ b/kernel/workqueue.c @@ -661,7 +661,7 @@ void wq_worker_waking_up(struct task_struct *task, unsigned int cpu) { struct worker *worker = kthread_data(task); - if (likely(!(worker->flags & WORKER_NOT_RUNNING))) + if (!(worker->flags & WORKER_NOT_RUNNING)) atomic_inc(get_gcwq_nr_running(cpu)); } @@ -687,7 +687,7 @@ struct task_struct *wq_worker_sleeping(struct task_struct *task, struct global_cwq *gcwq = get_gcwq(cpu); atomic_t *nr_running = get_gcwq_nr_running(cpu); - if (unlikely(worker->flags & WORKER_NOT_RUNNING)) + if (worker->flags & WORKER_NOT_RUNNING) return NULL; /* this can only happen on the local cpu */ -- cgit v1.2.3 From ce677831a4abd0f9f957c90ac6f6a0d0472bafb4 Mon Sep 17 00:00:00 2001 From: Dan Carpenter Date: Sun, 24 Oct 2010 21:50:42 +0200 Subject: perf: Fix off by one in perf_swevent_init() The perf_swevent_enabled[] array has PERF_COUNT_SW_MAX elements. Signed-off-by: Dan Carpenter Signed-off-by: Peter Zijlstra LKML-Reference: <20101024195041.GT5985@bicker> Signed-off-by: Ingo Molnar --- kernel/perf_event.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/perf_event.c b/kernel/perf_event.c index 7b870174c56d..2870feee81dd 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -4719,7 +4719,7 @@ static int perf_swevent_init(struct perf_event *event) break; } - if (event_id > PERF_COUNT_SW_MAX) + if (event_id >= PERF_COUNT_SW_MAX) return -ENOENT; if (!event->parent) { -- cgit v1.2.3 From fe44d62122829959e960bc699318d58966922a69 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Thu, 9 Dec 2010 14:15:34 +0100 Subject: sched: Fix the irqtime code to deal with u64 wraps Some ARM systems have a short sched_clock() [ which needs to be fixed too ], but this exposed a bug in the irq_time code as well, it doesn't deal with wraps at all. Fix the irq_time code to deal with u64 wraps by re-writing the code to only use delta increments, which avoids the whole issue. Reviewed-by: Venkatesh Pallipadi Reported-by: Mikael Pettersson Tested-by: Mikael Pettersson Signed-off-by: Peter Zijlstra LKML-Reference: <1292242433.6803.199.camel@twins> Signed-off-by: Ingo Molnar --- kernel/sched.c | 83 +++++++++++++++++++++++++++++++++++----------------------- 1 file changed, 50 insertions(+), 33 deletions(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index da14302a9857..79b557c63381 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -636,23 +636,18 @@ static inline struct task_group *task_group(struct task_struct *p) #endif /* CONFIG_CGROUP_SCHED */ -static u64 irq_time_cpu(int cpu); -static void sched_irq_time_avg_update(struct rq *rq, u64 irq_time); +static void update_rq_clock_task(struct rq *rq, s64 delta); -inline void update_rq_clock(struct rq *rq) +static void update_rq_clock(struct rq *rq) { - int cpu = cpu_of(rq); - u64 irq_time; + s64 delta; if (rq->skip_clock_update) return; - rq->clock = sched_clock_cpu(cpu); - irq_time = irq_time_cpu(cpu); - if (rq->clock - irq_time > rq->clock_task) - rq->clock_task = rq->clock - irq_time; - - sched_irq_time_avg_update(rq, irq_time); + delta = sched_clock_cpu(cpu_of(rq)) - rq->clock; + rq->clock += delta; + update_rq_clock_task(rq, delta); } /* @@ -1946,19 +1941,20 @@ void disable_sched_clock_irqtime(void) sched_clock_irqtime = 0; } -static u64 irq_time_cpu(int cpu) +static inline u64 irq_time_cpu(int cpu) { - if (!sched_clock_irqtime) - return 0; - return per_cpu(cpu_softirq_time, cpu) + per_cpu(cpu_hardirq_time, cpu); } +/* + * Called before incrementing preempt_count on {soft,}irq_enter + * and before decrementing preempt_count on {soft,}irq_exit. + */ void account_system_vtime(struct task_struct *curr) { unsigned long flags; + s64 delta; int cpu; - u64 now, delta; if (!sched_clock_irqtime) return; @@ -1966,9 +1962,9 @@ void account_system_vtime(struct task_struct *curr) local_irq_save(flags); cpu = smp_processor_id(); - now = sched_clock_cpu(cpu); - delta = now - per_cpu(irq_start_time, cpu); - per_cpu(irq_start_time, cpu) = now; + delta = sched_clock_cpu(cpu) - __this_cpu_read(irq_start_time); + __this_cpu_add(irq_start_time, delta); + /* * We do not account for softirq time from ksoftirqd here. * We want to continue accounting softirq time to ksoftirqd thread @@ -1976,33 +1972,54 @@ void account_system_vtime(struct task_struct *curr) * that do not consume any time, but still wants to run. */ if (hardirq_count()) - per_cpu(cpu_hardirq_time, cpu) += delta; + __this_cpu_add(cpu_hardirq_time, delta); else if (in_serving_softirq() && !(curr->flags & PF_KSOFTIRQD)) - per_cpu(cpu_softirq_time, cpu) += delta; + __this_cpu_add(cpu_softirq_time, delta); local_irq_restore(flags); } EXPORT_SYMBOL_GPL(account_system_vtime); -static void sched_irq_time_avg_update(struct rq *rq, u64 curr_irq_time) +static void update_rq_clock_task(struct rq *rq, s64 delta) { - if (sched_clock_irqtime && sched_feat(NONIRQ_POWER)) { - u64 delta_irq = curr_irq_time - rq->prev_irq_time; - rq->prev_irq_time = curr_irq_time; - sched_rt_avg_update(rq, delta_irq); - } + s64 irq_delta; + + irq_delta = irq_time_cpu(cpu_of(rq)) - rq->prev_irq_time; + + /* + * Since irq_time is only updated on {soft,}irq_exit, we might run into + * this case when a previous update_rq_clock() happened inside a + * {soft,}irq region. + * + * When this happens, we stop ->clock_task and only update the + * prev_irq_time stamp to account for the part that fit, so that a next + * update will consume the rest. This ensures ->clock_task is + * monotonic. + * + * It does however cause some slight miss-attribution of {soft,}irq + * time, a more accurate solution would be to update the irq_time using + * the current rq->clock timestamp, except that would require using + * atomic ops. + */ + if (irq_delta > delta) + irq_delta = delta; + + rq->prev_irq_time += irq_delta; + delta -= irq_delta; + rq->clock_task += delta; + + if (irq_delta && sched_feat(NONIRQ_POWER)) + sched_rt_avg_update(rq, irq_delta); } -#else +#else /* CONFIG_IRQ_TIME_ACCOUNTING */ -static u64 irq_time_cpu(int cpu) +static void update_rq_clock_task(struct rq *rq, s64 delta) { - return 0; + rq->clock_task += delta; } -static void sched_irq_time_avg_update(struct rq *rq, u64 curr_irq_time) { } - -#endif +#endif /* CONFIG_IRQ_TIME_ACCOUNTING */ #include "sched_idletask.c" #include "sched_fair.c" -- cgit v1.2.3 From 8e92c20183ed0579d94501311b81c42b65cb2129 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Thu, 9 Dec 2010 14:15:34 +0100 Subject: sched: Fix the irqtime code for 32bit Since the irqtime accounting is using non-atomic u64 and can be read from remote cpus (writes are strictly cpu local, reads are not) we have to deal with observing partial updates. When we do observe partial updates the clock movement (in particular, ->clock_task movement) will go funny (in either direction), a subsequent clock update (observing the full update) will make it go funny in the oposite direction. Since we rely on these clocks to be strictly monotonic we cannot suffer backwards motion. One possible solution would be to simply ignore all backwards deltas, but that will lead to accounting artefacts, most notable: clock_task + irq_time != clock, this inaccuracy would end up in user visible stats. Therefore serialize the reads using a seqcount. Reviewed-by: Venkatesh Pallipadi Reported-by: Mikael Pettersson Tested-by: Mikael Pettersson Signed-off-by: Peter Zijlstra LKML-Reference: <1292242434.6803.200.camel@twins> Signed-off-by: Ingo Molnar --- kernel/sched.c | 51 +++++++++++++++++++++++++++++++++++++++++++++------ 1 file changed, 45 insertions(+), 6 deletions(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index 79b557c63381..456c99054160 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -1920,10 +1920,9 @@ static void deactivate_task(struct rq *rq, struct task_struct *p, int flags) * They are read and saved off onto struct rq in update_rq_clock(). * This may result in other CPU reading this CPU's irq time and can * race with irq/account_system_vtime on this CPU. We would either get old - * or new value (or semi updated value on 32 bit) with a side effect of - * accounting a slice of irq time to wrong task when irq is in progress - * while we read rq->clock. That is a worthy compromise in place of having - * locks on each irq in account_system_time. + * or new value with a side effect of accounting a slice of irq time to wrong + * task when irq is in progress while we read rq->clock. That is a worthy + * compromise in place of having locks on each irq in account_system_time. */ static DEFINE_PER_CPU(u64, cpu_hardirq_time); static DEFINE_PER_CPU(u64, cpu_softirq_time); @@ -1941,10 +1940,48 @@ void disable_sched_clock_irqtime(void) sched_clock_irqtime = 0; } -static inline u64 irq_time_cpu(int cpu) +#ifndef CONFIG_64BIT +static DEFINE_PER_CPU(seqcount_t, irq_time_seq); + +static inline void irq_time_write_begin(void) +{ + __this_cpu_inc(irq_time_seq.sequence); + smp_wmb(); +} + +static inline void irq_time_write_end(void) +{ + smp_wmb(); + __this_cpu_inc(irq_time_seq.sequence); +} + +static inline u64 irq_time_read(int cpu) +{ + u64 irq_time; + unsigned seq; + + do { + seq = read_seqcount_begin(&per_cpu(irq_time_seq, cpu)); + irq_time = per_cpu(cpu_softirq_time, cpu) + + per_cpu(cpu_hardirq_time, cpu); + } while (read_seqcount_retry(&per_cpu(irq_time_seq, cpu), seq)); + + return irq_time; +} +#else /* CONFIG_64BIT */ +static inline void irq_time_write_begin(void) +{ +} + +static inline void irq_time_write_end(void) +{ +} + +static inline u64 irq_time_read(int cpu) { return per_cpu(cpu_softirq_time, cpu) + per_cpu(cpu_hardirq_time, cpu); } +#endif /* CONFIG_64BIT */ /* * Called before incrementing preempt_count on {soft,}irq_enter @@ -1965,6 +2002,7 @@ void account_system_vtime(struct task_struct *curr) delta = sched_clock_cpu(cpu) - __this_cpu_read(irq_start_time); __this_cpu_add(irq_start_time, delta); + irq_time_write_begin(); /* * We do not account for softirq time from ksoftirqd here. * We want to continue accounting softirq time to ksoftirqd thread @@ -1976,6 +2014,7 @@ void account_system_vtime(struct task_struct *curr) else if (in_serving_softirq() && !(curr->flags & PF_KSOFTIRQD)) __this_cpu_add(cpu_softirq_time, delta); + irq_time_write_end(); local_irq_restore(flags); } EXPORT_SYMBOL_GPL(account_system_vtime); @@ -1984,7 +2023,7 @@ static void update_rq_clock_task(struct rq *rq, s64 delta) { s64 irq_delta; - irq_delta = irq_time_cpu(cpu_of(rq)) - rq->prev_irq_time; + irq_delta = irq_time_read(cpu_of(rq)) - rq->prev_irq_time; /* * Since irq_time is only updated on {soft,}irq_exit, we might run into -- cgit v1.2.3 From 1497dd1d29c6a53fcd3c80f7ac8d0e0239e7389e Mon Sep 17 00:00:00 2001 From: Takashi Iwai Date: Fri, 10 Dec 2010 00:16:39 +0100 Subject: PM / Hibernate: Fix PM_POST_* notification with user-space suspend The user-space hibernation sends a wrong notification after the image restoration because of thinko for the file flag check. RDONLY corresponds to hibernation and WRONLY to restoration, confusingly. Signed-off-by: Takashi Iwai Signed-off-by: Rafael J. Wysocki Cc: stable@kernel.org --- kernel/power/user.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/power/user.c b/kernel/power/user.c index 1b2ea31e6bd8..c36c3b9e8a84 100644 --- a/kernel/power/user.c +++ b/kernel/power/user.c @@ -137,7 +137,7 @@ static int snapshot_release(struct inode *inode, struct file *filp) free_all_swap_pages(data->swap); if (data->frozen) thaw_processes(); - pm_notifier_call_chain(data->mode == O_WRONLY ? + pm_notifier_call_chain(data->mode == O_RDONLY ? PM_POST_HIBERNATION : PM_POST_RESTORE); atomic_inc(&snapshot_device_available); -- cgit v1.2.3 From be8cd644c49dca4212e975455c8e7119b848ebe8 Mon Sep 17 00:00:00 2001 From: "Rafael J. Wysocki" Date: Sat, 11 Dec 2010 21:46:44 +0100 Subject: PM / Hibernate: Restore old swap signature to avoid user space breakage Commit 3624eb0 (PM / Hibernate: Modify signature used to mark swap) attempted to modify hibernate signature used to mark swap partitions containing hibernation images, so that old kernels don't try to handle compressed images. However, this change broke resume from hibernation on Fedora 14 that apparently doesn't pass the resume= argument to the kernel and tries to trigger resume from early user space. This doesn't work, because the signature is now different, so the old signature has to be restored to avoid the problem. Addresses https://bugzilla.kernel.org/show_bug.cgi?id=22732 . Reported-by: Dr. David Alan Gilbert Reported-by: Zhang Rui Reported-by: Pascal Chapperon Signed-off-by: Rafael J. Wysocki --- kernel/power/swap.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/power/swap.c b/kernel/power/swap.c index baf667bb2794..8c7e4832b9be 100644 --- a/kernel/power/swap.c +++ b/kernel/power/swap.c @@ -30,7 +30,7 @@ #include "power.h" -#define HIBERNATE_SIG "LINHIB0001" +#define HIBERNATE_SIG "S1SUSPEND" /* * The swap map is a data structure used for keeping track of each page -- cgit v1.2.3 From c0f5ac5426f7fd82b23dd5c6a1e633b290294a08 Mon Sep 17 00:00:00 2001 From: Bjorn Helgaas Date: Thu, 16 Dec 2010 10:38:41 -0700 Subject: Revert "resources: support allocating space within a region from the top down" This reverts commit e7f8567db9a7f6b3151b0b275e245c1cef0d9c70. Acked-by: H. Peter Anvin Signed-off-by: Bjorn Helgaas Signed-off-by: Jesse Barnes --- kernel/resource.c | 98 +++---------------------------------------------------- 1 file changed, 4 insertions(+), 94 deletions(-) (limited to 'kernel') diff --git a/kernel/resource.c b/kernel/resource.c index 9fad33efd0db..560659f7baef 100644 --- a/kernel/resource.c +++ b/kernel/resource.c @@ -40,23 +40,6 @@ EXPORT_SYMBOL(iomem_resource); static DEFINE_RWLOCK(resource_lock); -/* - * By default, we allocate free space bottom-up. The architecture can request - * top-down by clearing this flag. The user can override the architecture's - * choice with the "resource_alloc_from_bottom" kernel boot option, but that - * should only be a debugging tool. - */ -int resource_alloc_from_bottom = 1; - -static __init int setup_alloc_from_bottom(char *s) -{ - printk(KERN_INFO - "resource: allocating from bottom-up; please report a bug\n"); - resource_alloc_from_bottom = 1; - return 0; -} -early_param("resource_alloc_from_bottom", setup_alloc_from_bottom); - static void *r_next(struct seq_file *m, void *v, loff_t *pos) { struct resource *p = v; @@ -396,75 +379,8 @@ static bool resource_contains(struct resource *res1, struct resource *res2) return res1->start <= res2->start && res1->end >= res2->end; } -/* - * Find the resource before "child" in the sibling list of "root" children. - */ -static struct resource *find_sibling_prev(struct resource *root, struct resource *child) -{ - struct resource *this; - - for (this = root->child; this; this = this->sibling) - if (this->sibling == child) - return this; - - return NULL; -} - -/* - * Find empty slot in the resource tree given range and alignment. - * This version allocates from the end of the root resource first. - */ -static int find_resource_from_top(struct resource *root, struct resource *new, - resource_size_t size, resource_size_t min, - resource_size_t max, resource_size_t align, - resource_size_t (*alignf)(void *, - const struct resource *, - resource_size_t, - resource_size_t), - void *alignf_data) -{ - struct resource *this; - struct resource tmp, avail, alloc; - - tmp.start = root->end; - tmp.end = root->end; - - this = find_sibling_prev(root, NULL); - for (;;) { - if (this) { - if (this->end < root->end) - tmp.start = this->end + 1; - } else - tmp.start = root->start; - - resource_clip(&tmp, min, max); - - /* Check for overflow after ALIGN() */ - avail = *new; - avail.start = ALIGN(tmp.start, align); - avail.end = tmp.end; - if (avail.start >= tmp.start) { - alloc.start = alignf(alignf_data, &avail, size, align); - alloc.end = alloc.start + size - 1; - if (resource_contains(&avail, &alloc)) { - new->start = alloc.start; - new->end = alloc.end; - return 0; - } - } - - if (!this || this->start == root->start) - break; - - tmp.end = this->start - 1; - this = find_sibling_prev(root, this); - } - return -EBUSY; -} - /* * Find empty slot in the resource tree given range and alignment. - * This version allocates from the beginning of the root resource first. */ static int find_resource(struct resource *root, struct resource *new, resource_size_t size, resource_size_t min, @@ -480,15 +396,14 @@ static int find_resource(struct resource *root, struct resource *new, tmp.start = root->start; /* - * Skip past an allocated resource that starts at 0, since the - * assignment of this->start - 1 to tmp->end below would cause an - * underflow. + * Skip past an allocated resource that starts at 0, since the assignment + * of this->start - 1 to tmp->end below would cause an underflow. */ if (this && this->start == 0) { tmp.start = this->end + 1; this = this->sibling; } - for (;;) { + for(;;) { if (this) tmp.end = this->start - 1; else @@ -509,10 +424,8 @@ static int find_resource(struct resource *root, struct resource *new, return 0; } } - if (!this) break; - tmp.start = this->end + 1; this = this->sibling; } @@ -545,10 +458,7 @@ int allocate_resource(struct resource *root, struct resource *new, alignf = simple_align_resource; write_lock(&resource_lock); - if (resource_alloc_from_bottom) - err = find_resource(root, new, size, min, max, align, alignf, alignf_data); - else - err = find_resource_from_top(root, new, size, min, max, align, alignf, alignf_data); + err = find_resource(root, new, size, min, max, align, alignf, alignf_data); if (err >= 0 && __request_resource(root, new)) err = -EBUSY; write_unlock(&resource_lock); -- cgit v1.2.3 From fcb119183c73bf0781009713f303e28b1fb13d3e Mon Sep 17 00:00:00 2001 From: Bjorn Helgaas Date: Thu, 16 Dec 2010 10:38:46 -0700 Subject: resources: add arch hook for preventing allocation in reserved areas This adds arch_remove_reservations(), which an arch can implement if it needs to protect part of the address space from allocation. Sometimes that can be done by just putting a region in the resource tree, but there are cases where that doesn't work well. For example, x86 BIOS E820 reservations are not related to devices, so they may overlap part of, all of, or more than a device resource, so they may not end up at the correct spot in the resource tree. Acked-by: H. Peter Anvin Signed-off-by: Bjorn Helgaas Signed-off-by: Jesse Barnes --- kernel/resource.c | 6 ++++++ 1 file changed, 6 insertions(+) (limited to 'kernel') diff --git a/kernel/resource.c b/kernel/resource.c index 560659f7baef..798e2fae2a06 100644 --- a/kernel/resource.c +++ b/kernel/resource.c @@ -357,6 +357,10 @@ int __weak page_is_ram(unsigned long pfn) return walk_system_ram_range(pfn, 1, NULL, __is_ram) == 1; } +void __weak arch_remove_reservations(struct resource *avail) +{ +} + static resource_size_t simple_align_resource(void *data, const struct resource *avail, resource_size_t size, @@ -394,6 +398,7 @@ static int find_resource(struct resource *root, struct resource *new, struct resource *this = root->child; struct resource tmp = *new, avail, alloc; + tmp.flags = new->flags; tmp.start = root->start; /* * Skip past an allocated resource that starts at 0, since the assignment @@ -410,6 +415,7 @@ static int find_resource(struct resource *root, struct resource *new, tmp.end = root->end; resource_clip(&tmp, min, max); + arch_remove_reservations(&tmp); /* Check for overflow after ALIGN() */ avail = *new; -- cgit v1.2.3 From 050c6c9b896625d9fa498265be17b82c5fc65257 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Sun, 19 Dec 2010 23:24:27 +0100 Subject: sched: Remove debugging check Linus reported that the new warning introduced by commit f26f9aff6aaf "Sched: fix skip_clock_update optimization" triggers. The need_resched flag can be set by other CPUs asynchronously so this debug check is bogus - remove it. Reported-by: Linus Torvalds Cc: Peter Zijlstra Cc: Mike Galbraith LKML-Reference: Signed-off-by: Ingo Molnar --- kernel/sched.c | 1 - 1 file changed, 1 deletion(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index 456c99054160..297d1a0eedb0 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -4128,7 +4128,6 @@ need_resched_nonpreemptible: rq->nr_switches++; rq->curr = next; ++*switch_count; - WARN_ON_ONCE(test_tsk_need_resched(next)); context_switch(rq, prev, next); /* unlocks the rq */ /* -- cgit v1.2.3 From 4f32e9b1f812fd6c00cc85a127583fefbdedaedc Mon Sep 17 00:00:00 2001 From: Yong Zhang Date: Wed, 22 Dec 2010 10:27:53 +0100 Subject: kthread_work: make lockdep happy spinlock in kthread_worker and wait_queue_head in kthread_work both should be lockdep sensible, so change the interface to make it suiltable for CONFIG_LOCKDEP. tj: comment update Reported-by: Nicolas Signed-off-by: Yong Zhang Signed-off-by: Andy Walls Tested-by: Andy Walls Cc: Tejun Heo Cc: Andrew Morton Signed-off-by: Tejun Heo --- kernel/kthread.c | 11 +++++++++++ 1 file changed, 11 insertions(+) (limited to 'kernel') diff --git a/kernel/kthread.c b/kernel/kthread.c index 2dc3786349d1..ca61bbdd44b2 100644 --- a/kernel/kthread.c +++ b/kernel/kthread.c @@ -265,6 +265,17 @@ int kthreadd(void *unused) return 0; } +void __init_kthread_worker(struct kthread_worker *worker, + const char *name, + struct lock_class_key *key) +{ + spin_lock_init(&worker->lock); + lockdep_set_class_and_name(&worker->lock, key, name); + INIT_LIST_HEAD(&worker->work_list); + worker->task = NULL; +} +EXPORT_SYMBOL_GPL(__init_kthread_worker); + /** * kthread_worker_fn - kthread function to process kthread_worker * @worker_ptr: pointer to initialized kthread_worker -- cgit v1.2.3 From 4be2c95d1f7706ca0e74499f2bd118e1cee19669 Mon Sep 17 00:00:00 2001 From: Jeff Mahoney Date: Tue, 21 Dec 2010 17:24:30 -0800 Subject: taskstats: pad taskstats netlink response for aligment issues on ia64 The taskstats structure is internally aligned on 8 byte boundaries but the layout of the aggregrate reply, with two NLA headers and the pid (each 4 bytes), actually force the entire structure to be unaligned. This causes the kernel to issue unaligned access warnings on some architectures like ia64. Unfortunately, some software out there doesn't properly unroll the NLA packet and assumes that the start of the taskstats structure will always be 20 bytes from the start of the netlink payload. Aligning the start of the taskstats structure breaks this software, which we don't want. So, for now the alignment only happens on architectures that require it and those users will have to update to fixed versions of those packages. Space is reserved in the packet only when needed. This ifdef should be removed in several years e.g. 2012 once we can be confident that fixed versions are installed on most systems. We add the padding before the aggregate since the aggregate is already a defined type. Commit 85893120 ("delayacct: align to 8 byte boundary on 64-bit systems") previously addressed the alignment issues by padding out the pid field. This was supposed to be a compatible change but the circumstances described above mean that it wasn't. This patch backs out that change, since it was a hack, and introduces a new NULL attribute type to provide the padding. Padding the response with 4 bytes avoids allocating an aligned taskstats structure and copying it back. Since the structure weighs in at 328 bytes, it's too big to do it on the stack. Signed-off-by: Jeff Mahoney Reported-by: Brian Rogers Cc: Jeff Mahoney Cc: Guillaume Chazarain Cc: Balbir Singh Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/taskstats.c | 57 +++++++++++++++++++++++++++++++++++++++++------------- 1 file changed, 44 insertions(+), 13 deletions(-) (limited to 'kernel') diff --git a/kernel/taskstats.c b/kernel/taskstats.c index c8231fb15708..3308fd7f1b52 100644 --- a/kernel/taskstats.c +++ b/kernel/taskstats.c @@ -349,25 +349,47 @@ static int parse(struct nlattr *na, struct cpumask *mask) return ret; } +#ifdef CONFIG_IA64 +#define TASKSTATS_NEEDS_PADDING 1 +#endif + static struct taskstats *mk_reply(struct sk_buff *skb, int type, u32 pid) { struct nlattr *na, *ret; int aggr; - /* If we don't pad, we end up with alignment on a 4 byte boundary. - * This causes lots of runtime warnings on systems requiring 8 byte - * alignment */ - u32 pids[2] = { pid, 0 }; - int pid_size = ALIGN(sizeof(pid), sizeof(long)); - aggr = (type == TASKSTATS_TYPE_PID) ? TASKSTATS_TYPE_AGGR_PID : TASKSTATS_TYPE_AGGR_TGID; + /* + * The taskstats structure is internally aligned on 8 byte + * boundaries but the layout of the aggregrate reply, with + * two NLA headers and the pid (each 4 bytes), actually + * force the entire structure to be unaligned. This causes + * the kernel to issue unaligned access warnings on some + * architectures like ia64. Unfortunately, some software out there + * doesn't properly unroll the NLA packet and assumes that the start + * of the taskstats structure will always be 20 bytes from the start + * of the netlink payload. Aligning the start of the taskstats + * structure breaks this software, which we don't want. So, for now + * the alignment only happens on architectures that require it + * and those users will have to update to fixed versions of those + * packages. Space is reserved in the packet only when needed. + * This ifdef should be removed in several years e.g. 2012 once + * we can be confident that fixed versions are installed on most + * systems. We add the padding before the aggregate since the + * aggregate is already a defined type. + */ +#ifdef TASKSTATS_NEEDS_PADDING + if (nla_put(skb, TASKSTATS_TYPE_NULL, 0, NULL) < 0) + goto err; +#endif na = nla_nest_start(skb, aggr); if (!na) goto err; - if (nla_put(skb, type, pid_size, pids) < 0) + + if (nla_put(skb, type, sizeof(pid), &pid) < 0) goto err; ret = nla_reserve(skb, TASKSTATS_TYPE_STATS, sizeof(struct taskstats)); if (!ret) @@ -456,6 +478,18 @@ out: return rc; } +static size_t taskstats_packet_size(void) +{ + size_t size; + + size = nla_total_size(sizeof(u32)) + + nla_total_size(sizeof(struct taskstats)) + nla_total_size(0); +#ifdef TASKSTATS_NEEDS_PADDING + size += nla_total_size(0); /* Padding for alignment */ +#endif + return size; +} + static int cmd_attr_pid(struct genl_info *info) { struct taskstats *stats; @@ -464,8 +498,7 @@ static int cmd_attr_pid(struct genl_info *info) u32 pid; int rc; - size = nla_total_size(sizeof(u32)) + - nla_total_size(sizeof(struct taskstats)) + nla_total_size(0); + size = taskstats_packet_size(); rc = prepare_reply(info, TASKSTATS_CMD_NEW, &rep_skb, size); if (rc < 0) @@ -494,8 +527,7 @@ static int cmd_attr_tgid(struct genl_info *info) u32 tgid; int rc; - size = nla_total_size(sizeof(u32)) + - nla_total_size(sizeof(struct taskstats)) + nla_total_size(0); + size = taskstats_packet_size(); rc = prepare_reply(info, TASKSTATS_CMD_NEW, &rep_skb, size); if (rc < 0) @@ -570,8 +602,7 @@ void taskstats_exit(struct task_struct *tsk, int group_dead) /* * Size includes space for nested attributes */ - size = nla_total_size(sizeof(u32)) + - nla_total_size(sizeof(struct taskstats)) + nla_total_size(0); + size = taskstats_packet_size(); is_thread_group = !!taskstats_tgid_alloc(tsk); if (is_thread_group) { -- cgit v1.2.3 From e1e359273576ee8fe27021356b064c772ed29af3 Mon Sep 17 00:00:00 2001 From: David Sharp Date: Wed, 22 Dec 2010 16:38:24 -0800 Subject: ring_buffer: Off-by-one and duplicate events in ring_buffer_read_page Fix two related problems in the event-copying loop of ring_buffer_read_page. The loop condition for copying events is off-by-one. "len" is the remaining space in the caller-supplied page. "size" is the size of the next event (or two events). If len == size, then there is just enough space for the next event. size was set to rb_event_ts_length, which may include the size of two events if the first event is a time-extend, in order to assure time- extends are kept together with the event after it. However, rb_advance_reader always advances by one event. This would result in the event after any time-extend being duplicated. Instead, get the size of a single event for the memcpy, but use rb_event_ts_length for the loop condition. Signed-off-by: David Sharp LKML-Reference: <1293064704-8101-1-git-send-email-dhsharp@google.com> LKML-Reference: Signed-off-by: Steven Rostedt --- kernel/trace/ring_buffer.c | 9 ++++++++- 1 file changed, 8 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c index 9ed509a015d8..bd1c35a4fbcc 100644 --- a/kernel/trace/ring_buffer.c +++ b/kernel/trace/ring_buffer.c @@ -3853,6 +3853,13 @@ int ring_buffer_read_page(struct ring_buffer *buffer, /* Need to copy one event at a time */ do { + /* We need the size of one event, because + * rb_advance_reader only advances by one event, + * whereas rb_event_ts_length may include the size of + * one or two events. + * We have already ensured there's enough space if this + * is a time extend. */ + size = rb_event_length(event); memcpy(bpage->data + pos, rpage->data + rpos, size); len -= size; @@ -3867,7 +3874,7 @@ int ring_buffer_read_page(struct ring_buffer *buffer, event = rb_reader_event(cpu_buffer); /* Always keep the time extend and data together */ size = rb_event_ts_length(event); - } while (len > size); + } while (len >= size); /* update bpage */ local_set(&bpage->commit, pos); -- cgit v1.2.3