时间子系统16_soft lockup机制

//	1.debug选项LOCKUP_DETECTOR,开启/关闭kernel中的soft lockup和hard lockup探测
//	2.实现:kernel/watchdog.c

//	3.实现原理:
//		1.涉及到了3部分内容:kernel线程,时钟中断,NMI中断
//			优先级:kernel线程 < 时钟中断 < NMI中断。
//		2.利用它们之间优先级的区别,调试系统运行中的两种问题:
//			抢占被长时间关闭而导致进程无法调度(soft lockup)
//			中断被长时间关闭而导致更严重的问题(hard lockup)

//	参考:http://blog.csdn.net/panzhenjie/article/details/10074551
//	内核版本 3.8.6
//	smp per-cpu watchdog核心线程
1.1 static struct smp_hotplug_thread watchdog_threads = {
	.store			= &softlockup_watchdog,
	.thread_should_run	= watchdog_should_run,
	.thread_fn		= watchdog,
	.thread_comm		= "watchdog/%u",
	.setup			= watchdog_enable,
	.park			= watchdog_disable,
	.unpark			= watchdog_enable,
};

//	lockup detector模块初始化
//	函数任务:
//		1.计算hrtimer运行的频率
//		2.注册watchdog核心线程
//	注:
1.2 void __init lockup_detector_init(void)
{
	//计算hrtimer运行的频率
	set_sample_period();
	//注册watchdog核心线程
	if (smpboot_register_percpu_thread(&watchdog_threads)) {
		pr_err("Failed to create watchdog threads, disabled\n");
		watchdog_disabled = -ENODEV;
	}
}
//	设置watchdog timer运行频率
//	调用路径:	lockup_detector_init->get_softlockup_thresh
//	注:
//		1.sample_period,即watchdog timer运行的频率
//		2.watchdog timer在一次soft lockup时间阈值内运行5次
1.3 static void set_sample_period(void)
{
	sample_period = get_softlockup_thresh() * ((u64)NSEC_PER_SEC / 5);
}

//	认定发生了soft lockup的时间阈值
//	注:如果watchdog kthread在watchdog_thresh * 2 时间内未被调度,
//		则认为发生了soft lockup.
1.4 static int get_softlockup_thresh(void)
{
	return watchdog_thresh * 2;
}

//	启动指定cpu上lockup检测
//	函数任务:
//		1.初始化watchdog timer
//		2.初始化hard lockup的nmi中断事件
//		3.启动watchdog timer
//		4.设置watchdog kthread调度策略FIFO
//		5.更新watchdog时间戳

//	注:设置watchdog kthread为FIFO的调度策略保证了watchdog timer
//		唤醒kthread之后,它可以因高优先级切换到cpu上执行。
2.1 static void watchdog_enable(unsigned int cpu)
{
	struct hrtimer *hrtimer = &__raw_get_cpu_var(watchdog_hrtimer);
	//lockup检测使用的hrtimer
	hrtimer_init(hrtimer, CLOCK_MONOTONIC, HRTIMER_MODE_REL);
	hrtimer->function = watchdog_timer_fn;

	//第一次启动watchdog,暂停current
	if (!watchdog_enabled) {
		kthread_park(current);
		return;
	}	
	//hard lockup检测机制
	watchdog_nmi_enable(cpu);
	//hrtimer sample时间之后运行
	hrtimer_start(hrtimer, ns_to_ktime(sample_period),
		      HRTIMER_MODE_REL_PINNED);
	//watchdog进程FIFO策略
	watchdog_set_prio(SCHED_FIFO, MAX_RT_PRIO - 1);
	//执行一次更新
	__touch_watchdog();
}

//	关闭指定cpu上的lockup检测
//	函数任务:
//		1.恢复watchdog正常优先级
//		2.取消hrtimer
//		3.关闭hard lockup检测机制的nmi中断
2.2 static void watchdog_disable(unsigned int cpu)
{
	struct hrtimer *hrtimer = &__raw_get_cpu_var(watchdog_hrtimer);
	//恢复watchdog正常优先级
	watchdog_set_prio(SCHED_NORMAL, 0);
	//取消hrtimer
	hrtimer_cancel(hrtimer);
	//关闭hard lockup检测机制的nmi中断
	watchdog_nmi_disable(cpu);
}

//	watchdog核心线程可运行的条件
//	函数任务:
//		1.保证watchdog kthread 运行频率 <= watchdog timer 运行频率
//	注:
//		soft_lockup_hrtimer_cnt代表watchdog核心线程运行的次数
//		hrtimer_interrupts代表watchdog timer运行的次数
2.3 static int watchdog_should_run(unsigned int cpu)
{
	return __this_cpu_read(hrtimer_interrupts) !=
		__this_cpu_read(soft_lockup_hrtimer_cnt);
}

//	watchdog核心线程函数
//	函数任务:
//		1.更新soft_lockup_hrtimer_cnt=hrtimer_interrupts
//		2.更新watchdog运行时间戳
2.4 static void watchdog(unsigned int cpu)
{
	__this_cpu_write(soft_lockup_hrtimer_cnt,
			 __this_cpu_read(hrtimer_interrupts));
	__touch_watchdog();
}

//	更新watchdog运行时间戳
2.5 static void __touch_watchdog(void)
{
	int this_cpu = smp_processor_id();

	__this_cpu_write(watchdog_touch_ts, get_timestamp(this_cpu));
}
//	定时器函数
//	函数主要任务:
//		1.获取watchdog上次运行的时间戳
//		2.递增watchdog timer运行次数
//		3.检查watchdog时间戳,是否发生了soft lockup
//			3.1 如果发生了,dump堆栈,打印信息
//		4.重调度timer
//	注:
//		在watchdog timer运行时唤醒watchdog kthread,保证kthread与timer相同的运行频率
3.1 static enum hrtimer_restart watchdog_timer_fn(struct hrtimer *hrtimer)
{
	//watchdog上次运行的时间戳
	unsigned long touch_ts = __this_cpu_read(watchdog_touch_ts);
	struct pt_regs *regs = get_irq_regs();
	int duration;
	//在唤醒watchdog kthread之前递增hrtimer_interrupts,保证kthread更新其时间戳
	watchdog_interrupt_count();
	//唤醒watchdog kthread,保证kthread与timer相同的运行频率
	wake_up_process(__this_cpu_read(softlockup_watchdog));
	//再次调度hrtimer下一个周期运行
	hrtimer_forward_now(hrtimer, ns_to_ktime(sample_period));

	...

	//检测是否发生soft lockup
	duration = is_softlockup(touch_ts);
	if (unlikely(duration)) {
		printk(KERN_EMERG "BUG: soft lockup - CPU#%d stuck for %us! [%s:%d]\n",
			smp_processor_id(), duration,
			current->comm, task_pid_nr(current));
		print_modules();
		print_irqtrace_events(current);
		//dump 寄存器和堆栈
		if (regs)
			show_regs(regs);
		else
			dump_stack();

		if (softlockup_panic)
			panic("softlockup: hung tasks");
	} 
	return HRTIMER_RESTART;
}
//	检查抢占被关闭的时间间隔
//		watchdog kthread在watchdog timer的中断上下文中被唤醒,
//		当中断退出时,kthread会抢占cpu上的当前进程。如果
//		抢占被关闭的话,则不会发生抢占,watchdog便无法更新时
//		间戳,当抢占关闭的时间超过阈值时,核心认为发生了soft
//		lock up。
//	注:soft lockup阈值 watchdog_thresh * 2 (20s)
3.2 static int is_softlockup(unsigned long touch_ts)
{
	//当前时间戳
	unsigned long now = get_timestamp(smp_processor_id());
	//watchdog在 watchdog_thresh * 2 时间内未被调度过
	if (time_after(now, touch_ts + get_softlockup_thresh()))
		return now - touch_ts;

	return 0;
}

上一篇:python自动化测试开发利器sublime环境搭建实战


下一篇:自动化设计-自动化测试环境搭建<一>