From 16bf134840da3920ded1290973c56ec214636f12 Mon Sep 17 00:00:00 2001 From: Hugh Dickins Date: Tue, 14 Feb 2006 13:52:59 -0800 Subject: [PATCH] compound page: no access_process_vm check The PageCompound check before access_process_vm's set_page_dirty_lock is no longer necessary, so remove it. But leave the PageCompound checks in bio_set_pages_dirty, dio_bio_complete and nfs_free_user_pages: at least some of those were introduced as a little optimization on hugetlb pages. Signed-off-by: Hugh Dickins Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/ptrace.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/ptrace.c b/kernel/ptrace.c index 5f33cdb6fff5..d2cf144d0af5 100644 --- a/kernel/ptrace.c +++ b/kernel/ptrace.c @@ -242,8 +242,7 @@ int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, in if (write) { copy_to_user_page(vma, page, addr, maddr + offset, buf, bytes); - if (!PageCompound(page)) - set_page_dirty_lock(page); + set_page_dirty_lock(page); } else { copy_from_user_page(vma, page, addr, buf, maddr + offset, bytes); -- cgit v1.2.3 From d6077cb80cde4506720f9165eba99ee07438513f Mon Sep 17 00:00:00 2001 From: "Chen, Kenneth W" Date: Tue, 14 Feb 2006 13:53:10 -0800 Subject: [PATCH] sched: revert "filter affine wakeups" Revert commit d7102e95b7b9c00277562c29aad421d2d521c5f6: [PATCH] sched: filter affine wakeups Apparently caused more than 10% performance regression for aim7 benchmark. The setup in use is 16-cpu HP rx8620, 64Gb of memory and 12 MSA1000s with 144 disks. Each disk is 72Gb with a single ext3 filesystem (courtesy of HP, who supplied benchmark results). The problem is, for aim7, the wake-up pattern is random, but it still needs load balancing action in the wake-up path to achieve best performance. With the above commit, lack of load balancing hurts that workload. However, for workloads like database transaction processing, the requirement is exactly opposite. In the wake up path, best performance is achieved with absolutely zero load balancing. We simply wake up the process on the CPU that it was previously run. Worst performance is obtained when we do load balancing at wake up. There isn't an easy way to auto detect the workload characteristics. Ingo's earlier patch that detects idle CPU and decide whether to load balance or not doesn't perform with aim7 either since all CPUs are busy (it causes even bigger perf. regression). Revert commit d7102e95b7b9c00277562c29aad421d2d521c5f6, which causes more than 10% performance regression with aim7. Signed-off-by: Ken Chen Acked-by: Ingo Molnar Cc: Nick Piggin Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/sched.c | 10 +--------- 1 file changed, 1 insertion(+), 9 deletions(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index 87d93be336a1..66d957227de9 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -1204,9 +1204,6 @@ static int try_to_wake_up(task_t *p, unsigned int state, int sync) } } - if (p->last_waker_cpu != this_cpu) - goto out_set_cpu; - if (unlikely(!cpu_isset(this_cpu, p->cpus_allowed))) goto out_set_cpu; @@ -1277,8 +1274,6 @@ out_set_cpu: cpu = task_cpu(p); } - p->last_waker_cpu = this_cpu; - out_activate: #endif /* CONFIG_SMP */ if (old_state == TASK_UNINTERRUPTIBLE) { @@ -1360,12 +1355,9 @@ void fastcall sched_fork(task_t *p, int clone_flags) #ifdef CONFIG_SCHEDSTATS memset(&p->sched_info, 0, sizeof(p->sched_info)); #endif -#if defined(CONFIG_SMP) - p->last_waker_cpu = cpu; -#if defined(__ARCH_WANT_UNLOCKED_CTXSW) +#if defined(CONFIG_SMP) && defined(__ARCH_WANT_UNLOCKED_CTXSW) p->oncpu = 0; #endif -#endif #ifdef CONFIG_PREEMPT /* Want to start with kernel preemption disabled. */ task_thread_info(p)->preempt_count = 1; -- cgit v1.2.3 From 06027bdd278a32a84b273e41db68a5db8ffd2bb6 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Tue, 14 Feb 2006 13:53:15 -0800 Subject: [PATCH] hrtimer: round up relative start time on low-res arches CONFIG_TIME_LOW_RES is a temporary way for architectures to signal that they simply return xtime in do_gettimeoffset(). In this corner-case we want to round up by resolution when starting a relative timer, to avoid short timeouts. This will go away with the GTOD framework. Signed-off-by: Ingo Molnar Cc: Roman Zippel Cc: Thomas Gleixner Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/hrtimer.c | 13 ++++++++++++- 1 file changed, 12 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/hrtimer.c b/kernel/hrtimer.c index 2b6e1757aedd..5ae51f1bc7c8 100644 --- a/kernel/hrtimer.c +++ b/kernel/hrtimer.c @@ -418,8 +418,19 @@ hrtimer_start(struct hrtimer *timer, ktime_t tim, const enum hrtimer_mode mode) /* Switch the timer base, if necessary: */ new_base = switch_hrtimer_base(timer, base); - if (mode == HRTIMER_REL) + if (mode == HRTIMER_REL) { tim = ktime_add(tim, new_base->get_time()); + /* + * CONFIG_TIME_LOW_RES is a temporary way for architectures + * to signal that they simply return xtime in + * do_gettimeoffset(). In this case we want to round up by + * resolution when starting a relative timer, to avoid short + * timeouts. This will go away with the GTOD framework. + */ +#ifdef CONFIG_TIME_LOW_RES + tim = ktime_add(tim, base->resolution); +#endif + } timer->expires = tim; enqueue_hrtimer(timer, new_base); -- cgit v1.2.3