Commit 6687a97d authored by Ingo Molnar's avatar Ingo Molnar Committed by Linus Torvalds

[PATCH] timer-irq-driven soft-watchdog, cleanups

Make the softlockup detector purely timer-interrupt driven, removing
softirq-context (timer) dependencies.  This means that if the softlockup
watchdog triggers, it has truly observed a longer than 10 seconds
scheduling delay of a SCHED_FIFO prio 99 task.

(the patch also turns off the softlockup detector during the initial bootup
phase and does small style fixes)
Signed-off-by: default avatarIngo Molnar <mingo@elte.hu>
Signed-off-by: default avatarIngo Molnar <mingo@elte.hu>
Signed-off-by: default avatarAndrew Morton <akpm@osdl.org>
Signed-off-by: default avatarLinus Torvalds <torvalds@osdl.org>
parent 6a4d11c2
...@@ -206,11 +206,11 @@ extern void update_process_times(int user); ...@@ -206,11 +206,11 @@ extern void update_process_times(int user);
extern void scheduler_tick(void); extern void scheduler_tick(void);
#ifdef CONFIG_DETECT_SOFTLOCKUP #ifdef CONFIG_DETECT_SOFTLOCKUP
extern void softlockup_tick(struct pt_regs *regs); extern void softlockup_tick(void);
extern void spawn_softlockup_task(void); extern void spawn_softlockup_task(void);
extern void touch_softlockup_watchdog(void); extern void touch_softlockup_watchdog(void);
#else #else
static inline void softlockup_tick(struct pt_regs *regs) static inline void softlockup_tick(void)
{ {
} }
static inline void spawn_softlockup_task(void) static inline void spawn_softlockup_task(void)
......
/* /*
* Detect Soft Lockups * Detect Soft Lockups
* *
* started by Ingo Molnar, (C) 2005, Red Hat * started by Ingo Molnar, Copyright (C) 2005, 2006 Red Hat, Inc.
* *
* this code detects soft lockups: incidents in where on a CPU * this code detects soft lockups: incidents in where on a CPU
* the kernel does not reschedule for 10 seconds or more. * the kernel does not reschedule for 10 seconds or more.
*/ */
#include <linux/mm.h> #include <linux/mm.h>
#include <linux/cpu.h> #include <linux/cpu.h>
#include <linux/init.h> #include <linux/init.h>
...@@ -17,13 +16,14 @@ ...@@ -17,13 +16,14 @@
static DEFINE_SPINLOCK(print_lock); static DEFINE_SPINLOCK(print_lock);
static DEFINE_PER_CPU(unsigned long, timestamp) = 0; static DEFINE_PER_CPU(unsigned long, touch_timestamp);
static DEFINE_PER_CPU(unsigned long, print_timestamp) = 0; static DEFINE_PER_CPU(unsigned long, print_timestamp);
static DEFINE_PER_CPU(struct task_struct *, watchdog_task); static DEFINE_PER_CPU(struct task_struct *, watchdog_task);
static int did_panic = 0; static int did_panic = 0;
static int softlock_panic(struct notifier_block *this, unsigned long event,
void *ptr) static int
softlock_panic(struct notifier_block *this, unsigned long event, void *ptr)
{ {
did_panic = 1; did_panic = 1;
...@@ -36,7 +36,7 @@ static struct notifier_block panic_block = { ...@@ -36,7 +36,7 @@ static struct notifier_block panic_block = {
void touch_softlockup_watchdog(void) void touch_softlockup_watchdog(void)
{ {
per_cpu(timestamp, raw_smp_processor_id()) = jiffies; per_cpu(touch_timestamp, raw_smp_processor_id()) = jiffies;
} }
EXPORT_SYMBOL(touch_softlockup_watchdog); EXPORT_SYMBOL(touch_softlockup_watchdog);
...@@ -44,25 +44,35 @@ EXPORT_SYMBOL(touch_softlockup_watchdog); ...@@ -44,25 +44,35 @@ EXPORT_SYMBOL(touch_softlockup_watchdog);
* This callback runs from the timer interrupt, and checks * This callback runs from the timer interrupt, and checks
* whether the watchdog thread has hung or not: * whether the watchdog thread has hung or not:
*/ */
void softlockup_tick(struct pt_regs *regs) void softlockup_tick(void)
{ {
int this_cpu = smp_processor_id(); int this_cpu = smp_processor_id();
unsigned long timestamp = per_cpu(timestamp, this_cpu); unsigned long touch_timestamp = per_cpu(touch_timestamp, this_cpu);
if (per_cpu(print_timestamp, this_cpu) == timestamp) /* prevent double reports: */
if (per_cpu(print_timestamp, this_cpu) == touch_timestamp ||
did_panic ||
!per_cpu(watchdog_task, this_cpu))
return; return;
/* Do not cause a second panic when there already was one */ /* do not print during early bootup: */
if (did_panic) if (unlikely(system_state != SYSTEM_RUNNING)) {
touch_softlockup_watchdog();
return; return;
}
if (time_after(jiffies, timestamp + 10*HZ)) { /* Wake up the high-prio watchdog task every second: */
per_cpu(print_timestamp, this_cpu) = timestamp; if (time_after(jiffies, touch_timestamp + HZ))
wake_up_process(per_cpu(watchdog_task, this_cpu));
/* Warn about unreasonable 10+ seconds delays: */
if (time_after(jiffies, touch_timestamp + 10*HZ)) {
per_cpu(print_timestamp, this_cpu) = touch_timestamp;
spin_lock(&print_lock); spin_lock(&print_lock);
printk(KERN_ERR "BUG: soft lockup detected on CPU#%d!\n", printk(KERN_ERR "BUG: soft lockup detected on CPU#%d!\n",
this_cpu); this_cpu);
show_regs(regs); dump_stack();
spin_unlock(&print_lock); spin_unlock(&print_lock);
} }
} }
...@@ -77,18 +87,16 @@ static int watchdog(void * __bind_cpu) ...@@ -77,18 +87,16 @@ static int watchdog(void * __bind_cpu)
sched_setscheduler(current, SCHED_FIFO, &param); sched_setscheduler(current, SCHED_FIFO, &param);
current->flags |= PF_NOFREEZE; current->flags |= PF_NOFREEZE;
set_current_state(TASK_INTERRUPTIBLE);
/* /*
* Run briefly once per second - if this gets delayed for * Run briefly once per second to reset the softlockup timestamp.
* more than 10 seconds then the debug-printout triggers * If this gets delayed for more than 10 seconds then the
* in softlockup_tick(): * debug-printout triggers in softlockup_tick().
*/ */
while (!kthread_should_stop()) { while (!kthread_should_stop()) {
msleep_interruptible(1000); set_current_state(TASK_INTERRUPTIBLE);
touch_softlockup_watchdog(); touch_softlockup_watchdog();
schedule();
} }
__set_current_state(TASK_RUNNING);
return 0; return 0;
} }
...@@ -114,7 +122,6 @@ cpu_callback(struct notifier_block *nfb, unsigned long action, void *hcpu) ...@@ -114,7 +122,6 @@ cpu_callback(struct notifier_block *nfb, unsigned long action, void *hcpu)
kthread_bind(p, hotcpu); kthread_bind(p, hotcpu);
break; break;
case CPU_ONLINE: case CPU_ONLINE:
wake_up_process(per_cpu(watchdog_task, hotcpu)); wake_up_process(per_cpu(watchdog_task, hotcpu));
break; break;
#ifdef CONFIG_HOTPLUG_CPU #ifdef CONFIG_HOTPLUG_CPU
...@@ -146,4 +153,3 @@ __init void spawn_softlockup_task(void) ...@@ -146,4 +153,3 @@ __init void spawn_softlockup_task(void)
notifier_chain_register(&panic_notifier_list, &panic_block); notifier_chain_register(&panic_notifier_list, &panic_block);
} }
...@@ -915,6 +915,7 @@ static void run_timer_softirq(struct softirq_action *h) ...@@ -915,6 +915,7 @@ static void run_timer_softirq(struct softirq_action *h)
void run_local_timers(void) void run_local_timers(void)
{ {
raise_softirq(TIMER_SOFTIRQ); raise_softirq(TIMER_SOFTIRQ);
softlockup_tick();
} }
/* /*
...@@ -945,7 +946,6 @@ void do_timer(struct pt_regs *regs) ...@@ -945,7 +946,6 @@ void do_timer(struct pt_regs *regs)
/* prevent loading jiffies before storing new jiffies_64 value. */ /* prevent loading jiffies before storing new jiffies_64 value. */
barrier(); barrier();
update_times(); update_times();
softlockup_tick(regs);
} }
#ifdef __ARCH_WANT_SYS_ALARM #ifdef __ARCH_WANT_SYS_ALARM
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment