diff options
Diffstat (limited to 'drivers/misc')
| -rw-r--r-- | drivers/misc/Kconfig | 14 | ||||
| -rw-r--r-- | drivers/misc/Makefile | 2 | ||||
| -rw-r--r-- | drivers/misc/lkdtm.c | 29 | ||||
| -rw-r--r-- | drivers/misc/memory_state_time.c | 454 | ||||
| -rw-r--r-- | drivers/misc/uid_sys_stats.c | 505 |
5 files changed, 1001 insertions, 3 deletions
diff --git a/drivers/misc/Kconfig b/drivers/misc/Kconfig index 4bf7d50b1bc7..5847d3be0835 100644 --- a/drivers/misc/Kconfig +++ b/drivers/misc/Kconfig @@ -525,6 +525,20 @@ config VEXPRESS_SYSCFG bus. System Configuration interface is one of the possible means of generating transactions on this bus. +config UID_SYS_STATS + bool "Per-UID statistics" + depends on PROFILING + help + Per UID based cpu time statistics exported to /proc/uid_cputime + Per UID based io statistics exported to /proc/uid_io + Per UID based procstat control in /proc/uid_procstat + +config MEMORY_STATE_TIME + tristate "Memory freq/bandwidth time statistics" + depends on PROFILING + help + Memory time statistics exported to /sys/kernel/memory_state_time + source "drivers/misc/c2port/Kconfig" source "drivers/misc/eeprom/Kconfig" source "drivers/misc/cb710/Kconfig" diff --git a/drivers/misc/Makefile b/drivers/misc/Makefile index 537d7f3b78da..9a3b402921b2 100644 --- a/drivers/misc/Makefile +++ b/drivers/misc/Makefile @@ -56,3 +56,5 @@ obj-$(CONFIG_GENWQE) += genwqe/ obj-$(CONFIG_ECHO) += echo/ obj-$(CONFIG_VEXPRESS_SYSCFG) += vexpress-syscfg.o obj-$(CONFIG_CXL_BASE) += cxl/ +obj-$(CONFIG_UID_SYS_STATS) += uid_sys_stats.o +obj-$(CONFIG_MEMORY_STATE_TIME) += memory_state_time.o diff --git a/drivers/misc/lkdtm.c b/drivers/misc/lkdtm.c index 11fdadc68e53..2a6eaf1122b4 100644 --- a/drivers/misc/lkdtm.c +++ b/drivers/misc/lkdtm.c @@ -103,6 +103,7 @@ enum ctype { CT_EXEC_USERSPACE, CT_ACCESS_USERSPACE, CT_WRITE_RO, + CT_WRITE_RO_AFTER_INIT, CT_WRITE_KERN, }; @@ -140,6 +141,7 @@ static char* cp_type[] = { "EXEC_USERSPACE", "ACCESS_USERSPACE", "WRITE_RO", + "WRITE_RO_AFTER_INIT", "WRITE_KERN", }; @@ -162,6 +164,7 @@ static DEFINE_SPINLOCK(lock_me_up); static u8 data_area[EXEC_SIZE]; static const unsigned long rodata = 0xAA55AA55; +static unsigned long ro_after_init __ro_after_init = 0x55AA5500; module_param(recur_count, int, 0644); MODULE_PARM_DESC(recur_count, " Recursion level for the stack overflow test"); @@ -503,11 +506,28 @@ static void lkdtm_do_action(enum ctype which) break; } case CT_WRITE_RO: { - unsigned long *ptr; + /* Explicitly cast away "const" for the test. */ + unsigned long *ptr = (unsigned long *)&rodata; - ptr = (unsigned long *)&rodata; + pr_info("attempting bad rodata write at %p\n", ptr); + *ptr ^= 0xabcd1234; - pr_info("attempting bad write at %p\n", ptr); + break; + } + case CT_WRITE_RO_AFTER_INIT: { + unsigned long *ptr = &ro_after_init; + + /* + * Verify we were written to during init. Since an Oops + * is considered a "success", a failure is to just skip the + * real test. + */ + if ((*ptr & 0xAA) != 0xAA) { + pr_info("%p was NOT written during init!?\n", ptr); + break; + } + + pr_info("attempting bad ro_after_init write at %p\n", ptr); *ptr ^= 0xabcd1234; break; @@ -817,6 +837,9 @@ static int __init lkdtm_module_init(void) int n_debugfs_entries = 1; /* Assume only the direct entry */ int i; + /* Make sure we can write to __ro_after_init values during __init */ + ro_after_init |= 0xAA; + /* Register debugfs interface */ lkdtm_debugfs_root = debugfs_create_dir("provoke-crash", NULL); if (!lkdtm_debugfs_root) { diff --git a/drivers/misc/memory_state_time.c b/drivers/misc/memory_state_time.c new file mode 100644 index 000000000000..34c797a06a31 --- /dev/null +++ b/drivers/misc/memory_state_time.c @@ -0,0 +1,454 @@ +/* drivers/misc/memory_state_time.c + * + * Copyright (C) 2016 Google, Inc. + * + * This software is licensed under the terms of the GNU General Public + * License version 2, as published by the Free Software Foundation, and + * may be copied, distributed, and modified under those terms. + * + * 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. + * + */ + +#include <linux/device.h> +#include <linux/err.h> +#include <linux/errno.h> +#include <linux/hashtable.h> +#include <linux/kconfig.h> +#include <linux/kernel.h> +#include <linux/kobject.h> +#include <linux/memory-state-time.h> +#include <linux/module.h> +#include <linux/mutex.h> +#include <linux/of_platform.h> +#include <linux/slab.h> +#include <linux/sysfs.h> +#include <linux/time.h> +#include <linux/timekeeping.h> +#include <linux/workqueue.h> + +#define KERNEL_ATTR_RO(_name) \ +static struct kobj_attribute _name##_attr = __ATTR_RO(_name) + +#define KERNEL_ATTR_RW(_name) \ +static struct kobj_attribute _name##_attr = \ + __ATTR(_name, 0644, _name##_show, _name##_store) + +#define FREQ_HASH_BITS 4 +DECLARE_HASHTABLE(freq_hash_table, FREQ_HASH_BITS); + +static DEFINE_MUTEX(mem_lock); + +#define TAG "memory_state_time" +#define BW_NODE "/soc/memory-state-time" +#define FREQ_TBL "freq-tbl" +#define BW_TBL "bw-buckets" +#define NUM_SOURCES "num-sources" + +#define LOWEST_FREQ 2 + +static int curr_bw; +static int curr_freq; +static u32 *bw_buckets; +static u32 *freq_buckets; +static int num_freqs; +static int num_buckets; +static int registered_bw_sources; +static u64 last_update; +static bool init_success; +static struct workqueue_struct *memory_wq; +static u32 num_sources = 10; +static int *bandwidths; + +struct freq_entry { + int freq; + u64 *buckets; /* Bandwidth buckets. */ + struct hlist_node hash; +}; + +struct queue_container { + struct work_struct update_state; + int value; + u64 time_now; + int id; + struct mutex *lock; +}; + +static int find_bucket(int bw) +{ + int i; + + if (bw_buckets != NULL) { + for (i = 0; i < num_buckets; i++) { + if (bw_buckets[i] > bw) { + pr_debug("Found bucket %d for bandwidth %d\n", + i, bw); + return i; + } + } + return num_buckets - 1; + } + return 0; +} + +static u64 get_time_diff(u64 time_now) +{ + u64 ms; + + ms = time_now - last_update; + last_update = time_now; + return ms; +} + +static ssize_t show_stat_show(struct kobject *kobj, + struct kobj_attribute *attr, char *buf) +{ + int i, j; + int len = 0; + struct freq_entry *freq_entry; + + for (i = 0; i < num_freqs; i++) { + hash_for_each_possible(freq_hash_table, freq_entry, hash, + freq_buckets[i]) { + if (freq_entry->freq == freq_buckets[i]) { + len += scnprintf(buf + len, PAGE_SIZE - len, + "%d ", freq_buckets[i]); + if (len >= PAGE_SIZE) + break; + for (j = 0; j < num_buckets; j++) { + len += scnprintf(buf + len, + PAGE_SIZE - len, + "%llu ", + freq_entry->buckets[j]); + } + len += scnprintf(buf + len, PAGE_SIZE - len, + "\n"); + } + } + } + pr_debug("Current Time: %llu\n", ktime_get_boot_ns()); + return len; +} +KERNEL_ATTR_RO(show_stat); + +static void update_table(u64 time_now) +{ + struct freq_entry *freq_entry; + + pr_debug("Last known bw %d freq %d\n", curr_bw, curr_freq); + hash_for_each_possible(freq_hash_table, freq_entry, hash, curr_freq) { + if (curr_freq == freq_entry->freq) { + freq_entry->buckets[find_bucket(curr_bw)] + += get_time_diff(time_now); + break; + } + } +} + +static bool freq_exists(int freq) +{ + int i; + + for (i = 0; i < num_freqs; i++) { + if (freq == freq_buckets[i]) + return true; + } + return false; +} + +static int calculate_total_bw(int bw, int index) +{ + int i; + int total_bw = 0; + + pr_debug("memory_state_time New bw %d for id %d\n", bw, index); + bandwidths[index] = bw; + for (i = 0; i < registered_bw_sources; i++) + total_bw += bandwidths[i]; + return total_bw; +} + +static void freq_update_do_work(struct work_struct *work) +{ + struct queue_container *freq_state_update + = container_of(work, struct queue_container, + update_state); + if (freq_state_update) { + mutex_lock(&mem_lock); + update_table(freq_state_update->time_now); + curr_freq = freq_state_update->value; + mutex_unlock(&mem_lock); + kfree(freq_state_update); + } +} + +static void bw_update_do_work(struct work_struct *work) +{ + struct queue_container *bw_state_update + = container_of(work, struct queue_container, + update_state); + if (bw_state_update) { + mutex_lock(&mem_lock); + update_table(bw_state_update->time_now); + curr_bw = calculate_total_bw(bw_state_update->value, + bw_state_update->id); + mutex_unlock(&mem_lock); + kfree(bw_state_update); + } +} + +static void memory_state_freq_update(struct memory_state_update_block *ub, + int value) +{ + if (IS_ENABLED(CONFIG_MEMORY_STATE_TIME)) { + if (freq_exists(value) && init_success) { + struct queue_container *freq_container + = kmalloc(sizeof(struct queue_container), + GFP_KERNEL); + if (!freq_container) + return; + INIT_WORK(&freq_container->update_state, + freq_update_do_work); + freq_container->time_now = ktime_get_boot_ns(); + freq_container->value = value; + pr_debug("Scheduling freq update in work queue\n"); + queue_work(memory_wq, &freq_container->update_state); + } else { + pr_debug("Freq does not exist.\n"); + } + } +} + +static void memory_state_bw_update(struct memory_state_update_block *ub, + int value) +{ + if (IS_ENABLED(CONFIG_MEMORY_STATE_TIME)) { + if (init_success) { + struct queue_container *bw_container + = kmalloc(sizeof(struct queue_container), + GFP_KERNEL); + if (!bw_container) + return; + INIT_WORK(&bw_container->update_state, + bw_update_do_work); + bw_container->time_now = ktime_get_boot_ns(); + bw_container->value = value; + bw_container->id = ub->id; + pr_debug("Scheduling bandwidth update in work queue\n"); + queue_work(memory_wq, &bw_container->update_state); + } + } +} + +struct memory_state_update_block *memory_state_register_frequency_source(void) +{ + struct memory_state_update_block *block; + + if (IS_ENABLED(CONFIG_MEMORY_STATE_TIME)) { + pr_debug("Allocating frequency source\n"); + block = kmalloc(sizeof(struct memory_state_update_block), + GFP_KERNEL); + if (!block) + return NULL; + block->update_call = memory_state_freq_update; + return block; + } + pr_err("Config option disabled.\n"); + return NULL; +} +EXPORT_SYMBOL_GPL(memory_state_register_frequency_source); + +struct memory_state_update_block *memory_state_register_bandwidth_source(void) +{ + struct memory_state_update_block *block; + + if (IS_ENABLED(CONFIG_MEMORY_STATE_TIME)) { + pr_debug("Allocating bandwidth source %d\n", + registered_bw_sources); + block = kmalloc(sizeof(struct memory_state_update_block), + GFP_KERNEL); + if (!block) + return NULL; + block->update_call = memory_state_bw_update; + if (registered_bw_sources < num_sources) { + block->id = registered_bw_sources++; + } else { + pr_err("Unable to allocate source; max number reached\n"); + kfree(block); + return NULL; + } + return block; + } + pr_err("Config option disabled.\n"); + return NULL; +} +EXPORT_SYMBOL_GPL(memory_state_register_bandwidth_source); + +/* Buckets are designated by their maximum. + * Returns the buckets decided by the capability of the device. + */ +static int get_bw_buckets(struct device *dev) +{ + int ret, lenb; + struct device_node *node = dev->of_node; + + of_property_read_u32(node, NUM_SOURCES, &num_sources); + if (of_find_property(node, BW_TBL, &lenb)) { + bandwidths = devm_kzalloc(dev, + sizeof(*bandwidths) * num_sources, GFP_KERNEL); + if (!bandwidths) + return -ENOMEM; + lenb /= sizeof(*bw_buckets); + bw_buckets = devm_kzalloc(dev, lenb * sizeof(*bw_buckets), + GFP_KERNEL); + if (!bw_buckets) { + devm_kfree(dev, bandwidths); + return -ENOMEM; + } + ret = of_property_read_u32_array(node, BW_TBL, bw_buckets, + lenb); + if (ret < 0) { + devm_kfree(dev, bandwidths); + devm_kfree(dev, bw_buckets); + pr_err("Unable to read bandwidth table from device tree.\n"); + return ret; + } + } + curr_bw = 0; + num_buckets = lenb; + return 0; +} + +/* Adds struct freq_entry nodes to the hashtable for each compatible frequency. + * Returns the supported number of frequencies. + */ +static int freq_buckets_init(struct device *dev) +{ + struct freq_entry *freq_entry; + int i; + int ret, lenf; + struct device_node *node = dev->of_node; + + if (of_find_property(node, FREQ_TBL, &lenf)) { + lenf /= sizeof(*freq_buckets); + freq_buckets = devm_kzalloc(dev, lenf * sizeof(*freq_buckets), + GFP_KERNEL); + if (!freq_buckets) + return -ENOMEM; + pr_debug("freqs found len %d\n", lenf); + ret = of_property_read_u32_array(node, FREQ_TBL, freq_buckets, + lenf); + if (ret < 0) { + devm_kfree(dev, freq_buckets); + pr_err("Unable to read frequency table from device tree.\n"); + return ret; + } + pr_debug("ret freq %d\n", ret); + } + num_freqs = lenf; + curr_freq = freq_buckets[LOWEST_FREQ]; + + for (i = 0; i < num_freqs; i++) { + freq_entry = devm_kzalloc(dev, sizeof(struct freq_entry), + GFP_KERNEL); + if (!freq_entry) + return -ENOMEM; + freq_entry->buckets = devm_kzalloc(dev, sizeof(u64)*num_buckets, + GFP_KERNEL); + if (!freq_entry->buckets) { + devm_kfree(dev, freq_entry); + return -ENOMEM; + } + pr_debug("memory_state_time Adding freq to ht %d\n", + freq_buckets[i]); + freq_entry->freq = freq_buckets[i]; + hash_add(freq_hash_table, &freq_entry->hash, freq_buckets[i]); + } + return 0; +} + +struct kobject *memory_kobj; +EXPORT_SYMBOL_GPL(memory_kobj); + +static struct attribute *memory_attrs[] = { + &show_stat_attr.attr, + NULL +}; + +static struct attribute_group memory_attr_group = { + .attrs = memory_attrs, +}; + +static int memory_state_time_probe(struct platform_device *pdev) +{ + int error; + + error = get_bw_buckets(&pdev->dev); + if (error) + return error; + error = freq_buckets_init(&pdev->dev); + if (error) + return error; + last_update = ktime_get_boot_ns(); + init_success = true; + + pr_debug("memory_state_time initialized with num_freqs %d\n", + num_freqs); + return 0; +} + +static const struct of_device_id match_table[] = { + { .compatible = "memory-state-time" }, + {} +}; + +static struct platform_driver memory_state_time_driver = { + .probe = memory_state_time_probe, + .driver = { + .name = "memory-state-time", + .of_match_table = match_table, + .owner = THIS_MODULE, + }, +}; + +static int __init memory_state_time_init(void) +{ + int error; + + hash_init(freq_hash_table); + memory_wq = create_singlethread_workqueue("memory_wq"); + if (!memory_wq) { + pr_err("Unable to create workqueue.\n"); + return -EINVAL; + } + /* + * Create sys/kernel directory for memory_state_time. + */ + memory_kobj = kobject_create_and_add(TAG, kernel_kobj); + if (!memory_kobj) { + pr_err("Unable to allocate memory_kobj for sysfs directory.\n"); + error = -ENOMEM; + goto wq; + } + error = sysfs_create_group(memory_kobj, &memory_attr_group); + if (error) { + pr_err("Unable to create sysfs folder.\n"); + goto kobj; + } + + error = platform_driver_register(&memory_state_time_driver); + if (error) { + pr_err("Unable to register memory_state_time platform driver.\n"); + goto group; + } + return 0; + +group: sysfs_remove_group(memory_kobj, &memory_attr_group); +kobj: kobject_put(memory_kobj); +wq: destroy_workqueue(memory_wq); + return error; +} +module_init(memory_state_time_init); diff --git a/drivers/misc/uid_sys_stats.c b/drivers/misc/uid_sys_stats.c new file mode 100644 index 000000000000..ad21276c8d9e --- /dev/null +++ b/drivers/misc/uid_sys_stats.c @@ -0,0 +1,505 @@ +/* drivers/misc/uid_cputime.c + * + * Copyright (C) 2014 - 2015 Google, Inc. + * + * This software is licensed under the terms of the GNU General Public + * License version 2, as published by the Free Software Foundation, and + * may be copied, distributed, and modified under those terms. + * + * 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. + * + */ + +#include <linux/atomic.h> +#include <linux/err.h> +#include <linux/hashtable.h> +#include <linux/init.h> +#include <linux/kernel.h> +#include <linux/list.h> +#include <linux/proc_fs.h> +#include <linux/profile.h> +#include <linux/rtmutex.h> +#include <linux/sched.h> +#include <linux/seq_file.h> +#include <linux/slab.h> +#include <linux/uaccess.h> + +#define UID_HASH_BITS 10 +DECLARE_HASHTABLE(hash_table, UID_HASH_BITS); + +static DEFINE_RT_MUTEX(uid_lock); +static struct proc_dir_entry *cpu_parent; +static struct proc_dir_entry *io_parent; +static struct proc_dir_entry *proc_parent; + +struct io_stats { + u64 read_bytes; + u64 write_bytes; + u64 rchar; + u64 wchar; + u64 fsync; +}; + +#define UID_STATE_FOREGROUND 0 +#define UID_STATE_BACKGROUND 1 +#define UID_STATE_BUCKET_SIZE 2 + +#define UID_STATE_TOTAL_CURR 2 +#define UID_STATE_TOTAL_LAST 3 +#define UID_STATE_SIZE 4 + +struct uid_entry { + uid_t uid; + cputime_t utime; + cputime_t stime; + cputime_t active_utime; + cputime_t active_stime; + int state; + struct io_stats io[UID_STATE_SIZE]; + struct hlist_node hash; +}; + +static struct uid_entry *find_uid_entry(uid_t uid) +{ + struct uid_entry *uid_entry; + hash_for_each_possible(hash_table, uid_entry, hash, uid) { + if (uid_entry->uid == uid) + return uid_entry; + } + return NULL; +} + +static struct uid_entry *find_or_register_uid(uid_t uid) +{ + struct uid_entry *uid_entry; + + uid_entry = find_uid_entry(uid); + if (uid_entry) + return uid_entry; + + uid_entry = kzalloc(sizeof(struct uid_entry), GFP_ATOMIC); + if (!uid_entry) + return NULL; + + uid_entry->uid = uid; + + hash_add(hash_table, &uid_entry->hash, uid); + + return uid_entry; +} + +static int uid_cputime_show(struct seq_file *m, void *v) +{ + struct uid_entry *uid_entry; + struct task_struct *task, *temp; + struct user_namespace *user_ns = current_user_ns(); + cputime_t utime; + cputime_t stime; + unsigned long bkt; + uid_t uid; + + rt_mutex_lock(&uid_lock); + + hash_for_each(hash_table, bkt, uid_entry, hash) { + uid_entry->active_stime = 0; + uid_entry->active_utime = 0; + } + + read_lock(&tasklist_lock); + do_each_thread(temp, task) { + uid = from_kuid_munged(user_ns, task_uid(task)); + uid_entry = find_or_register_uid(uid); + if (!uid_entry) { + read_unlock(&tasklist_lock); + rt_mutex_unlock(&uid_lock); + pr_err("%s: failed to find the uid_entry for uid %d\n", + __func__, uid); + return -ENOMEM; + } + task_cputime_adjusted(task, &utime, &stime); + uid_entry->active_utime += utime; + uid_entry->active_stime += stime; + } while_each_thread(temp, task); + read_unlock(&tasklist_lock); + + hash_for_each(hash_table, bkt, uid_entry, hash) { + cputime_t total_utime = uid_entry->utime + + uid_entry->active_utime; + cputime_t total_stime = uid_entry->stime + + uid_entry->active_stime; + seq_printf(m, "%d: %llu %llu\n", uid_entry->uid, + (unsigned long long)jiffies_to_msecs( + cputime_to_jiffies(total_utime)) * USEC_PER_MSEC, + (unsigned long long)jiffies_to_msecs( + cputime_to_jiffies(total_stime)) * USEC_PER_MSEC); + } + + rt_mutex_unlock(&uid_lock); + return 0; +} + +static int uid_cputime_open(struct inode *inode, struct file *file) +{ + return single_open(file, uid_cputime_show, PDE_DATA(inode)); +} + +static const struct file_operations uid_cputime_fops = { + .open = uid_cputime_open, + .read = seq_read, + .llseek = seq_lseek, + .release = single_release, +}; + +static int uid_remove_open(struct inode *inode, struct file *file) +{ + return single_open(file, NULL, NULL); +} + +static ssize_t uid_remove_write(struct file *file, + const char __user *buffer, size_t count, loff_t *ppos) +{ + struct uid_entry *uid_entry; + struct hlist_node *tmp; + char uids[128]; + char *start_uid, *end_uid = NULL; + long int uid_start = 0, uid_end = 0; + + if (count >= sizeof(uids)) + count = sizeof(uids) - 1; + + if (copy_from_user(uids, buffer, count)) + return -EFAULT; + + uids[count] = '\0'; + end_uid = uids; + start_uid = strsep(&end_uid, "-"); + + if (!start_uid || !end_uid) + return -EINVAL; + + if (kstrtol(start_uid, 10, &uid_start) != 0 || + kstrtol(end_uid, 10, &uid_end) != 0) { + return -EINVAL; + } + rt_mutex_lock(&uid_lock); + + for (; uid_start <= uid_end; uid_start++) { + hash_for_each_possible_safe(hash_table, uid_entry, tmp, + hash, (uid_t)uid_start) { + if (uid_start == uid_entry->uid) { + hash_del(&uid_entry->hash); + kfree(uid_entry); + } + } + } + + rt_mutex_unlock(&uid_lock); + return count; +} + +static const struct file_operations uid_remove_fops = { + .open = uid_remove_open, + .release = single_release, + .write = uid_remove_write, +}; + +static u64 compute_write_bytes(struct task_struct *task) +{ + if (task->ioac.write_bytes <= task->ioac.cancelled_write_bytes) + return 0; + + return task->ioac.write_bytes - task->ioac.cancelled_write_bytes; +} + +static void add_uid_io_curr_stats(struct uid_entry *uid_entry, + struct task_struct *task) +{ + struct io_stats *io_curr = &uid_entry->io[UID_STATE_TOTAL_CURR]; + + io_curr->read_bytes += task->ioac.read_bytes; + io_curr->write_bytes += compute_write_bytes(task); + io_curr->rchar += task->ioac.rchar; + io_curr->wchar += task->ioac.wchar; + io_curr->fsync += task->ioac.syscfs; +} + +static void clean_uid_io_last_stats(struct uid_entry *uid_entry, + struct task_struct *task) +{ + struct io_stats *io_last = &uid_entry->io[UID_STATE_TOTAL_LAST]; + + io_last->read_bytes -= task->ioac.read_bytes; + io_last->write_bytes -= compute_write_bytes(task); + io_last->rchar -= task->ioac.rchar; + io_last->wchar -= task->ioac.wchar; + io_last->fsync -= task->ioac.syscfs; +} + +static void update_io_stats_all_locked(void) +{ + struct uid_entry *uid_entry; + struct task_struct *task, *temp; + struct io_stats *io_bucket, *io_curr, *io_last; + struct user_namespace *user_ns = current_user_ns(); + unsigned long bkt; + uid_t uid; + + hash_for_each(hash_table, bkt, uid_entry, hash) + memset(&uid_entry->io[UID_STATE_TOTAL_CURR], 0, + sizeof(struct io_stats)); + + rcu_read_lock(); + do_each_thread(temp, task) { + uid = from_kuid_munged(user_ns, task_uid(task)); + uid_entry = find_or_register_uid(uid); + if (!uid_entry) + continue; + add_uid_io_curr_stats(uid_entry, task); + } while_each_thread(temp, task); + rcu_read_unlock(); + + hash_for_each(hash_table, bkt, uid_entry, hash) { + io_bucket = &uid_entry->io[uid_entry->state]; + io_curr = &uid_entry->io[UID_STATE_TOTAL_CURR]; + io_last = &uid_entry->io[UID_STATE_TOTAL_LAST]; + + io_bucket->read_bytes += + io_curr->read_bytes - io_last->read_bytes; + io_bucket->write_bytes += + io_curr->write_bytes - io_last->write_bytes; + io_bucket->rchar += io_curr->rchar - io_last->rchar; + io_bucket->wchar += io_curr->wchar - io_last->wchar; + io_bucket->fsync += io_curr->fsync - io_last->fsync; + + io_last->read_bytes = io_curr->read_bytes; + io_last->write_bytes = io_curr->write_bytes; + io_last->rchar = io_curr->rchar; + io_last->wchar = io_curr->wchar; + io_last->fsync = io_curr->fsync; + } +} + +static void update_io_stats_uid_locked(uid_t target_uid) +{ + struct uid_entry *uid_entry; + struct task_struct *task, *temp; + struct io_stats *io_bucket, *io_curr, *io_last; + struct user_namespace *user_ns = current_user_ns(); + + uid_entry = find_or_register_uid(target_uid); + if (!uid_entry) + return; + + memset(&uid_entry->io[UID_STATE_TOTAL_CURR], 0, + sizeof(struct io_stats)); + + rcu_read_lock(); + do_each_thread(temp, task) { + if (from_kuid_munged(user_ns, task_uid(task)) != target_uid) + continue; + add_uid_io_curr_stats(uid_entry, task); + } while_each_thread(temp, task); + rcu_read_unlock(); + + io_bucket = &uid_entry->io[uid_entry->state]; + io_curr = &uid_entry->io[UID_STATE_TOTAL_CURR]; + io_last = &uid_entry->io[UID_STATE_TOTAL_LAST]; + + io_bucket->read_bytes += + io_curr->read_bytes - io_last->read_bytes; + io_bucket->write_bytes += + io_curr->write_bytes - io_last->write_bytes; + io_bucket->rchar += io_curr->rchar - io_last->rchar; + io_bucket->wchar += io_curr->wchar - io_last->wchar; + io_bucket->fsync += io_curr->fsync - io_last->fsync; + + io_last->read_bytes = io_curr->read_bytes; + io_last->write_bytes = io_curr->write_bytes; + io_last->rchar = io_curr->rchar; + io_last->wchar = io_curr->wchar; + io_last->fsync = io_curr->fsync; +} + +static int uid_io_show(struct seq_file *m, void *v) +{ + struct uid_entry *uid_entry; + unsigned long bkt; + + rt_mutex_lock(&uid_lock); + + update_io_stats_all_locked(); + + hash_for_each(hash_table, bkt, uid_entry, hash) { + seq_printf(m, "%d %llu %llu %llu %llu %llu %llu %llu %llu %llu %llu\n", + uid_entry->uid, + uid_entry->io[UID_STATE_FOREGROUND].rchar, + uid_entry->io[UID_STATE_FOREGROUND].wchar, + uid_entry->io[UID_STATE_FOREGROUND].read_bytes, + uid_entry->io[UID_STATE_FOREGROUND].write_bytes, + uid_entry->io[UID_STATE_BACKGROUND].rchar, + uid_entry->io[UID_STATE_BACKGROUND].wchar, + uid_entry->io[UID_STATE_BACKGROUND].read_bytes, + uid_entry->io[UID_STATE_BACKGROUND].write_bytes, + uid_entry->io[UID_STATE_FOREGROUND].fsync, + uid_entry->io[UID_STATE_BACKGROUND].fsync); + } + + rt_mutex_unlock(&uid_lock); + + return 0; +} + +static int uid_io_open(struct inode *inode, struct file *file) +{ + return single_open(file, uid_io_show, PDE_DATA(inode)); +} + +static const struct file_operations uid_io_fops = { + .open = uid_io_open, + .read = seq_read, + .llseek = seq_lseek, + .release = single_release, +}; + +static int uid_procstat_open(struct inode *inode, struct file *file) +{ + return single_open(file, NULL, NULL); +} + +static ssize_t uid_procstat_write(struct file *file, + const char __user *buffer, size_t count, loff_t *ppos) +{ + struct uid_entry *uid_entry; + uid_t uid; + int argc, state; + char input[128]; + + if (count >= sizeof(input)) + return -EINVAL; + + if (copy_from_user(input, buffer, count)) + return -EFAULT; + + input[count] = '\0'; + + argc = sscanf(input, "%u %d", &uid, &state); + if (argc != 2) + return -EINVAL; + + if (state != UID_STATE_BACKGROUND && state != UID_STATE_FOREGROUND) + return -EINVAL; + + rt_mutex_lock(&uid_lock); + + uid_entry = find_or_register_uid(uid); + if (!uid_entry) { + rt_mutex_unlock(&uid_lock); + return -EINVAL; + } + + if (uid_entry->state == state) { + rt_mutex_unlock(&uid_lock); + return count; + } + + update_io_stats_uid_locked(uid); + + uid_entry->state = state; + + rt_mutex_unlock(&uid_lock); + + return count; +} + +static const struct file_operations uid_procstat_fops = { + .open = uid_procstat_open, + .release = single_release, + .write = uid_procstat_write, +}; + +static int process_notifier(struct notifier_block *self, + unsigned long cmd, void *v) +{ + struct task_struct *task = v; + struct uid_entry *uid_entry; + cputime_t utime, stime; + uid_t uid; + + if (!task) + return NOTIFY_OK; + + rt_mutex_lock(&uid_lock); + uid = from_kuid_munged(current_user_ns(), task_uid(task)); + uid_entry = find_or_register_uid(uid); + if (!uid_entry) { + pr_err("%s: failed to find uid %d\n", __func__, uid); + goto exit; + } + + task_cputime_adjusted(task, &utime, &stime); + uid_entry->utime += utime; + uid_entry->stime += stime; + + update_io_stats_uid_locked(uid); + clean_uid_io_last_stats(uid_entry, task); + +exit: + rt_mutex_unlock(&uid_lock); + return NOTIFY_OK; +} + +static struct notifier_block process_notifier_block = { + .notifier_call = process_notifier, +}; + +static int __init proc_uid_sys_stats_init(void) +{ + hash_init(hash_table); + + cpu_parent = proc_mkdir("uid_cputime", NULL); + if (!cpu_parent) { + pr_err("%s: failed to create uid_cputime proc entry\n", + __func__); + goto err; + } + + proc_create_data("remove_uid_range", 0222, cpu_parent, + &uid_remove_fops, NULL); + proc_create_data("show_uid_stat", 0444, cpu_parent, + &uid_cputime_fops, NULL); + + io_parent = proc_mkdir("uid_io", NULL); + if (!io_parent) { + pr_err("%s: failed to create uid_io proc entry\n", + __func__); + goto err; + } + + proc_create_data("stats", 0444, io_parent, + &uid_io_fops, NULL); + + proc_parent = proc_mkdir("uid_procstat", NULL); + if (!proc_parent) { + pr_err("%s: failed to create uid_procstat proc entry\n", + __func__); + goto err; + } + + proc_create_data("set", 0222, proc_parent, + &uid_procstat_fops, NULL); + + profile_event_register(PROFILE_TASK_EXIT, &process_notifier_block); + + return 0; + +err: + remove_proc_subtree("uid_cputime", NULL); + remove_proc_subtree("uid_io", NULL); + remove_proc_subtree("uid_procstat", NULL); + return -ENOMEM; +} + +early_initcall(proc_uid_sys_stats_init); |