Linux內(nèi)核調(diào)試技術(shù)——進(jìn)程D狀態(tài)死鎖檢測
Linux的進(jìn)程存在多種狀態(tài),如TASK_RUNNING的運行態(tài)、EXIT_DEAD的停止態(tài)和TASK_INTERRUPTIBLE的接收信號的等待狀態(tài)等等(可在include/linux/sched.h中查看)。其中有一種狀態(tài)等待為TASK_UNINTERRUPTIBLE,稱為D狀態(tài),該種狀態(tài)下進(jìn)程不接收信號,只能通過wake_up喚醒。處于這種狀態(tài)的情況有很多,例如mutex鎖就可能會設(shè)置進(jìn)程于該狀態(tài),有時候進(jìn)程在等待某種IO資源就緒時(wait_event機制)會設(shè)置進(jìn)程進(jìn)入該狀態(tài)。一般情況下,進(jìn)程處于該狀態(tài)的時間不會太久,但若IO設(shè)備出現(xiàn)故障或者出現(xiàn)進(jìn)程死鎖等情況,進(jìn)程就可能長期處于該狀態(tài)而無法再返回到TASK_RUNNING態(tài)。因此,內(nèi)核為了便于發(fā)現(xiàn)這類情況設(shè)計出了hung task機制專門用于檢測長期處于D狀態(tài)的進(jìn)程并發(fā)出告警。本文分析內(nèi)核hung task機制的源碼并給出一個示例演示。
一、hung task機制分析
內(nèi)核在很早的版本中就已經(jīng)引入了hung task機制,本文以較新的Linux 4.1.15版本源碼為例進(jìn)行分析,代碼量并不多,源代碼文件為kernel/hung_task.c。
首先給出整體流程框圖和設(shè)計思想:
圖 D狀態(tài)死鎖流程圖
其核心思想為創(chuàng)建一個內(nèi)核監(jiān)測進(jìn)程循環(huán)監(jiān)測處于D狀態(tài)的每一個進(jìn)程(任務(wù)),統(tǒng)計它們在兩次檢測之間的調(diào)度次數(shù),如果發(fā)現(xiàn)有任務(wù)在兩次監(jiān)測之間沒有發(fā)生任何的調(diào)度則可判斷該進(jìn)程一直處于D狀態(tài),很有可能已經(jīng)死鎖,因此觸發(fā)報警日志打印,輸出進(jìn)程的基本信息,?;厮菀约凹拇嫫鞅4嫘畔⒁怨﹥?nèi)核開發(fā)人員定位。
下面詳細(xì)分析實現(xiàn)方式:
- [cpp] view plain copy 在CODE上查看代碼片派生到我的代碼片
- static int __init hung_task_init(void)
- {
- atomic_notifier_chain_register(&panic_notifier_list, &panic_block);
- watchdog_task = kthread_run(watchdog, NULL, "khungtaskd");
- return 0;
- }
- subsys_initcall(hung_task_init);
首先,若在內(nèi)核配置中啟用了該機制,在內(nèi)核的subsys初始化階段就會調(diào)用hung_task_init()函數(shù)啟用功能,首先向內(nèi)核的panic_notifier_list通知鏈注冊回調(diào):
- [cpp] view plain copy 在CODE上查看代碼片派生到我的代碼片
- static struct notifier_block panic_block = {
- .notifier_call = hung_task_panic,
- };
在內(nèi)核觸發(fā)panic時就會調(diào)用該hung_task_panic()函數(shù),這個函數(shù)的作用稍后再看。繼續(xù)往下初始化,調(diào)用kthread_run()函數(shù)創(chuàng)建了一個名為khungtaskd的線程,執(zhí)行watchdog()函數(shù),立即嘗試調(diào)度執(zhí)行。該線程就是專用于檢測D狀態(tài)死鎖進(jìn)程的后臺內(nèi)核線程。
- [cpp] view plain copy 在CODE上查看代碼片派生到我的代碼片
- /*
- * kthread which checks for tasks stuck in D state
- */
- static int watchdog(void *dummy)
- {
- set_user_nice(current, 0);
- for ( ; ; ) {
- unsigned long timeout = sysctl_hung_task_timeout_secs;
- while (schedule_timeout_interruptible(timeout_jiffies(timeout)))
- timeout = sysctl_hung_task_timeout_secs;
- if (atomic_xchg(&reset_hung_task, 0))
- continue;
- check_hung_uninterruptible_tasks(timeout);
- }
- return 0;
- }
本進(jìn)程首先設(shè)置優(yōu)先級為0,即一般優(yōu)先級,不影響其他進(jìn)程。然后進(jìn)入主循環(huán)(每隔timeout時間執(zhí)行一次),首先讓進(jìn)程睡眠,設(shè)置的睡眠時間為
CONFIG_DEFAULT_HUNG_TASK_TIMEOUT,可以通過內(nèi)核配置選項修改,默認(rèn)值為120s,睡眠結(jié)束被喚醒后判斷原子變量標(biāo)識reset_hung_task,若被置位則跳過本輪監(jiān)測,同時會清除該標(biāo)識。該標(biāo)識通過reset_hung_task_detector()函數(shù)設(shè)置(目前內(nèi)核中尚無其他程序使用該接口):
- [cpp] view plain copy 在CODE上查看代碼片派生到我的代碼片
- void reset_hung_task_detector(void)
- {
- atomic_set(&reset_hung_task, 1);
- }
- EXPORT_SYMBOL_GPL(reset_hung_task_detector);
接下來循環(huán)的***即為監(jiān)測函數(shù)check_hung_uninterruptible_tasks(),函數(shù)入?yún)楸O(jiān)測超時時間。
- [cpp] view plain copy 在CODE上查看代碼片派生到我的代碼片
- /*
- * Check whether a TASK_UNINTERRUPTIBLE does not get woken up for
- * a really long time (120 seconds). If that happens, print out
- * a warning.
- */
- static void check_hung_uninterruptible_tasks(unsigned long timeout)
- {
- int max_count = sysctl_hung_task_check_count;
- int batch_count = HUNG_TASK_BATCHING;
- struct task_struct *g, *t;
- /*
- * If the system crashed already then all bets are off,
- * do not report extra hung tasks:
- */
- if (test_taint(TAINT_DIE) || did_panic)
- return;
- rcu_read_lock();
- for_each_process_thread(g, t) {
- if (!max_count--)
- goto unlock;
- if (!--batch_count) {
- batch_count = HUNG_TASK_BATCHING;
- if (!rcu_lock_break(g, t))
- goto unlock;
- }
- /* use "==" to skip the TASK_KILLABLE tasks waiting on NFS */
- if (t->state == TASK_UNINTERRUPTIBLE)
- check_hung_task(t, timeout);
- }
- unlock:
- rcu_read_unlock();
- }
首先檢測內(nèi)核是否已經(jīng)DIE了或者已經(jīng)panic了,如果是則表明內(nèi)核已經(jīng)crash了,無需再進(jìn)行監(jiān)測了,直接返回即可。注意這里的did_panic標(biāo)識在前文中的panic通知鏈回調(diào)函數(shù)中hung_task_panic()置位:
- [cpp] view plain copy 在CODE上查看代碼片派生到我的代碼片
- static int
- hung_task_panic(struct notifier_block *this, unsigned long event, void *ptr)
- {
- did_panic = 1;
- return NOTIFY_DONE;
- }
接下去若尚無觸發(fā)內(nèi)核crash,則進(jìn)入監(jiān)測流程并逐一檢測內(nèi)核中的所有進(jìn)程(任務(wù)task),該過程在RCU加鎖的狀態(tài)下進(jìn)行,因此為了避免在進(jìn)程較多的情況下加鎖時間過長,這里設(shè)置了一個batch_count,一次最多檢測HUNG_TASK_BATCHING個進(jìn)程。于此同時用戶也可以設(shè)定***的檢測個數(shù)max_count=sysctl_hung_task_check_count,默認(rèn)值為***PID個數(shù)PID_MAX_LIMIT(通過sysctl命令設(shè)置)。
函數(shù)調(diào)用for_each_process_thread()函數(shù)輪詢內(nèi)核中的所有進(jìn)程(任務(wù)task),僅對狀態(tài)處于TASK_UNINTERRUPTIBLE狀態(tài)的進(jìn)程進(jìn)行超時判斷,調(diào)用check_hung_task()函數(shù),入?yún)閠ask_struct結(jié)構(gòu)和超時時間(120s):
- [cpp] view plain copy 在CODE上查看代碼片派生到我的代碼片
- static void check_hung_task(struct task_struct *t, unsigned long timeout)
- {
- unsigned long switch_count = t->nvcsw + t->nivcsw;
- /*
- * Ensure the task is not frozen.
- * Also, skip vfork and any other user process that freezer should skip.
- */
- if (unlikely(t->flags & (PF_FROZEN | PF_FREEZER_SKIP)))
- return;
- /*
- * When a freshly created task is scheduled once, changes its state to
- * TASK_UNINTERRUPTIBLE without having ever been switched out once, it
- * musn't be checked.
- */
- if (unlikely(!switch_count))
- return;
- if (switch_count != t->last_switch_count) {
- t->last_switch_count = switch_count;
- return;
- }
- trace_sched_process_hang(t);
- if (!sysctl_hung_task_warnings)
- return;
- if (sysctl_hung_task_warnings > 0)
- sysctl_hung_task_warnings--;
首先通過t->nvcsw和t->nivcsw的計數(shù)累加表示進(jìn)程從創(chuàng)建開始至今的調(diào)度次數(shù)總和,其中t->nvcsw表示進(jìn)程主動放棄CPU的次數(shù),t->nivcsw表示被強制搶占的次數(shù)。隨后函數(shù)判斷幾個標(biāo)識:(1)如果進(jìn)程被frozen了那就跳過檢測;(2)調(diào)度次數(shù)為0的不檢測。
接下來判斷從上一次檢測時保存的進(jìn)程調(diào)度次數(shù)和本次是否相同,若不相同則表明這輪timeout(120s)時間內(nèi)進(jìn)程發(fā)生了調(diào)度,則更新該調(diào)度值返回,否則則表明該進(jìn)程已經(jīng)有timeout(120s)時間沒有得到調(diào)度了,一直處于D狀態(tài)。接下來的trace_sched_process_hang()暫不清楚作用,然后判斷sysctl_hung_task_warnings標(biāo)識,它表示需要觸發(fā)報警的次數(shù),用戶也可以通過sysctl命令配置,默認(rèn)值為10,即若當(dāng)前檢測的進(jìn)程一直處于D狀態(tài),默認(rèn)情況下此處每2分鐘發(fā)出一次告警,一共發(fā)出10次,之后不再發(fā)出告警。下面來看告警代碼:
- [cpp] view plain copy 在CODE上查看代碼片派生到我的代碼片
- /*
- * Ok, the task did not get scheduled for more than 2 minutes,
- * complain:
- */
- pr_err("INFO: task %s:%d blocked for more than %ld seconds.\n",
- t->comm, t->pid, timeout);
- pr_err(" %s %s %.*s\n",
- print_tainted(), init_utsname()->release,
- (int)strcspn(init_utsname()->version, " "),
- init_utsname()->version);
- pr_err("\"echo 0 > /proc/sys/kernel/hung_task_timeout_secs\""
- " disables this message.\n");
- sched_show_task(t);
- debug_show_held_locks(t);
- touch_nmi_watchdog();
這里會在控制臺和日志中打印死鎖任務(wù)的名稱、PID號、超時時間、內(nèi)核tainted信息、sysinfo、內(nèi)核棧barktrace以及寄存器信息等。如果開啟了debug lock則打印鎖占用的情況,并touch nmi_watchdog以防止nmi_watchdog超時(對于我的ARM環(huán)境無需考慮nmi_watchdog)。
- [cpp] view plain copy 在CODE上查看代碼片派生到我的代碼片
- if (sysctl_hung_task_panic) {
- trigger_all_cpu_backtrace();
- panic("hung_task: blocked tasks");
- }
***如果設(shè)置了sysctl_hung_task_panic標(biāo)識則直接觸發(fā)panic(該值可通過內(nèi)核配置文件配置也可以通過sysctl設(shè)置)。
二、示例演示
演示環(huán)境:樹莓派b(Linux 4.1.15)
1、首先確認(rèn)內(nèi)核配置選項以確認(rèn)開啟hung stak機制
- [cpp] view plain copy 在CODE上查看代碼片派生到我的代碼片
- #include
- #include
- #include
- #include
- DEFINE_MUTEX(dlock);
- static int __init dlock_init(void)
- {
- mutex_lock(&dlock);
- mutex_lock(&dlock);
- return 0;
- }
- static void __exit dlock_exit(void)
- {
- return;
- }
- module_init(dlock_init);
- module_exit(dlock_exit);
- MODULE_LICENSE("GPL");
本示例程序定義了一個mutex鎖,然后在模塊的init函數(shù)中重復(fù)加鎖,人為造成死鎖現(xiàn)象(mutex_lock()函數(shù)會調(diào)用__mutex_lock_slowpath()將進(jìn)程設(shè)置為TASK_UNINTERRUPTIBLE狀態(tài)),進(jìn)程進(jìn)入D狀態(tài)后是無法退出的??梢酝ㄟ^ps命令來查看:
- root@apple:~# busybox ps
- PID USER TIME COMMAND
- ......
- 521 root 0:00 insmod dlock.ko
- ......
然后查看該進(jìn)程的狀態(tài),可見已經(jīng)進(jìn)入了D狀態(tài)。
- root@apple:~# cat /proc/521/status
- Name: insmod
- State: D (disk sleep)
- Tgid: 521
- Ngid: 0
- Pid: 521
至此在等待兩分鐘后調(diào)試串口就會輸出以下信息,可見每兩分鐘就會輸出一次:
- [ 360.625466] INFO: task insmod:521 blocked for more than 120 seconds.
- [ 360.631878] Tainted: G O 4.1.15 #5
- [ 360.637042] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
- [ 360.644986] [] (__schedule) from [] (schedule+0x40/0xa4)
- [ 360.652129] [] (schedule) from [] (schedule_preempt_disabled+0x18/0x1c)
- [ 360.660570] [] (schedule_preempt_disabled) from [] (__mutex_lock_slowpath+0x6c/0xe4)
- [ 360.670142] [] (__mutex_lock_slowpath) from [] (mutex_lock+0x44/0x48)
- [ 360.678432] [] (mutex_lock) from [] (dlock_init+0x20/0x2c [dlock])
- [ 360.686480] [] (dlock_init [dlock]) from [] (do_one_initcall+0x90/0x1e8)
- [ 360.694976] [] (do_one_initcall) from [] (do_init_module+0x6c/0x1c0)
- [ 360.703170] [] (do_init_module) from [] (load_module+0x1690/0x1d34)
- [ 360.711284] [] (load_module) from [] (SyS_init_module+0xdc/0x130)
- [ 360.719239] [] (SyS_init_module) from [] (ret_fast_syscall+0x0/0x54)
- [ 480.725351] INFO: task insmod:521 blocked for more than 120 seconds.
- [ 480.731759] Tainted: G O 4.1.15 #5
- [ 480.736917] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
- [ 480.744842] [] (__schedule) from [] (schedule+0x40/0xa4)
- [ 480.752029] [] (schedule) from [] (schedule_preempt_disabled+0x18/0x1c)
- [ 480.760479] [] (schedule_preempt_disabled) from [] (__mutex_lock_slowpath+0x6c/0xe4)
- [ 480.770066] [] (__mutex_lock_slowpath) from [] (mutex_lock+0x44/0x48)
- [ 480.778363] [] (mutex_lock) from [] (dlock_init+0x20/0x2c [dlock])
- [ 480.786402] [] (dlock_init [dlock]) from [] (do_one_initcall+0x90/0x1e8)
- [ 480.794897] [] (do_one_initcall) from [] (do_init_module+0x6c/0x1c0)
- [ 480.803085] [] (do_init_module) from [] (load_module+0x1690/0x1d34)
- [ 480.811188] [] (load_module) from [] (SyS_init_module+0xdc/0x130)
- [ 480.819113] [] (SyS_init_module) from [] (ret_fast_syscall+0x0/0x54)
- [ 600.825353] INFO: task insmod:521 blocked for more than 120 seconds.
- [ 600.831759] Tainted: G O 4.1.15 #5
- [ 600.836916] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
- [ 600.844865] [] (__schedule) from [] (schedule+0x40/0xa4)
- [ 600.852005] [] (schedule) from [] (schedule_preempt_disabled+0x18/0x1c)
- [ 600.860445] [] (schedule_preempt_disabled) from [] (__mutex_lock_slowpath+0x6c/0xe4)
- [ 600.870014] [] (__mutex_lock_slowpath) from [] (mutex_lock+0x44/0x48)
- [ 600.878303] [] (mutex_lock) from [] (dlock_init+0x20/0x2c [dlock])
- [ 600.886339] [] (dlock_init [dlock]) from [] (do_one_initcall+0x90/0x1e8)
- [ 600.894835] [] (do_one_initcall) from [] (do_init_module+0x6c/0x1c0)
- [ 600.903023] [] (do_init_module) from [] (load_module+0x1690/0x1d34)
- [ 600.911133] [] (load_module) from [] (SyS_init_module+0xdc/0x130)
- [ 600.919059] [] (SyS_init_module) from [] (ret_fast_syscall+0x0/0x54)
三、總結(jié)
D狀態(tài)死鎖一般在驅(qū)動開發(fā)的過程中比較常見,且不太容易定位,內(nèi)核提供這種hung task機制,開發(fā)人員只需要將這些輸出的定位信息抓取并保留下來就可以快速的進(jìn)行定位。