Blame view
kernel/softlockup.c
3.95 KB
8446f1d39 [PATCH] detect so... |
1 2 3 |
/* * Detect Soft Lockups * |
6687a97d4 [PATCH] timer-irq... |
4 |
* started by Ingo Molnar, Copyright (C) 2005, 2006 Red Hat, Inc. |
8446f1d39 [PATCH] detect so... |
5 6 7 8 |
* * this code detects soft lockups: incidents in where on a CPU * the kernel does not reschedule for 10 seconds or more. */ |
8446f1d39 [PATCH] detect so... |
9 10 11 12 13 14 15 16 17 |
#include <linux/mm.h> #include <linux/cpu.h> #include <linux/init.h> #include <linux/delay.h> #include <linux/kthread.h> #include <linux/notifier.h> #include <linux/module.h> static DEFINE_SPINLOCK(print_lock); |
6687a97d4 [PATCH] timer-irq... |
18 19 |
static DEFINE_PER_CPU(unsigned long, touch_timestamp); static DEFINE_PER_CPU(unsigned long, print_timestamp); |
8446f1d39 [PATCH] detect so... |
20 21 22 |
static DEFINE_PER_CPU(struct task_struct *, watchdog_task); static int did_panic = 0; |
6687a97d4 [PATCH] timer-irq... |
23 24 25 |
static int softlock_panic(struct notifier_block *this, unsigned long event, void *ptr) |
8446f1d39 [PATCH] detect so... |
26 27 28 29 30 31 32 33 34 35 36 37 |
{ did_panic = 1; return NOTIFY_DONE; } static struct notifier_block panic_block = { .notifier_call = softlock_panic, }; void touch_softlockup_watchdog(void) { |
bfe5d8341 [PATCH] Define __... |
38 |
__raw_get_cpu_var(touch_timestamp) = jiffies; |
8446f1d39 [PATCH] detect so... |
39 40 41 42 43 44 45 |
} EXPORT_SYMBOL(touch_softlockup_watchdog); /* * This callback runs from the timer interrupt, and checks * whether the watchdog thread has hung or not: */ |
6687a97d4 [PATCH] timer-irq... |
46 |
void softlockup_tick(void) |
8446f1d39 [PATCH] detect so... |
47 48 |
{ int this_cpu = smp_processor_id(); |
6687a97d4 [PATCH] timer-irq... |
49 |
unsigned long touch_timestamp = per_cpu(touch_timestamp, this_cpu); |
8446f1d39 [PATCH] detect so... |
50 |
|
6687a97d4 [PATCH] timer-irq... |
51 52 53 54 |
/* prevent double reports: */ if (per_cpu(print_timestamp, this_cpu) == touch_timestamp || did_panic || !per_cpu(watchdog_task, this_cpu)) |
8446f1d39 [PATCH] detect so... |
55 |
return; |
6687a97d4 [PATCH] timer-irq... |
56 57 58 |
/* do not print during early bootup: */ if (unlikely(system_state != SYSTEM_RUNNING)) { touch_softlockup_watchdog(); |
8446f1d39 [PATCH] detect so... |
59 |
return; |
6687a97d4 [PATCH] timer-irq... |
60 |
} |
8446f1d39 [PATCH] detect so... |
61 |
|
6687a97d4 [PATCH] timer-irq... |
62 63 64 65 66 67 68 |
/* Wake up the high-prio watchdog task every second: */ 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; |
8446f1d39 [PATCH] detect so... |
69 70 71 72 73 |
spin_lock(&print_lock); printk(KERN_ERR "BUG: soft lockup detected on CPU#%d! ", this_cpu); |
6687a97d4 [PATCH] timer-irq... |
74 |
dump_stack(); |
8446f1d39 [PATCH] detect so... |
75 76 77 78 79 80 81 82 83 84 |
spin_unlock(&print_lock); } } /* * The watchdog thread - runs every second and touches the timestamp. */ static int watchdog(void * __bind_cpu) { struct sched_param param = { .sched_priority = 99 }; |
8446f1d39 [PATCH] detect so... |
85 86 87 |
sched_setscheduler(current, SCHED_FIFO, ¶m); current->flags |= PF_NOFREEZE; |
8446f1d39 [PATCH] detect so... |
88 |
/* |
6687a97d4 [PATCH] timer-irq... |
89 90 91 |
* Run briefly once per second to reset the softlockup timestamp. * If this gets delayed for more than 10 seconds then the * debug-printout triggers in softlockup_tick(). |
8446f1d39 [PATCH] detect so... |
92 93 |
*/ while (!kthread_should_stop()) { |
6687a97d4 [PATCH] timer-irq... |
94 |
set_current_state(TASK_INTERRUPTIBLE); |
8446f1d39 [PATCH] detect so... |
95 |
touch_softlockup_watchdog(); |
6687a97d4 [PATCH] timer-irq... |
96 |
schedule(); |
8446f1d39 [PATCH] detect so... |
97 |
} |
8446f1d39 [PATCH] detect so... |
98 99 100 101 102 103 104 |
return 0; } /* * Create/destroy watchdog threads as CPUs come and go: */ |
8c78f3075 [PATCH] cpu hotpl... |
105 |
static int __cpuinit |
8446f1d39 [PATCH] detect so... |
106 107 108 109 110 111 112 113 114 115 116 117 118 119 |
cpu_callback(struct notifier_block *nfb, unsigned long action, void *hcpu) { int hotcpu = (unsigned long)hcpu; struct task_struct *p; switch (action) { case CPU_UP_PREPARE: BUG_ON(per_cpu(watchdog_task, hotcpu)); p = kthread_create(watchdog, hcpu, "watchdog/%d", hotcpu); if (IS_ERR(p)) { printk("watchdog for %i failed ", hotcpu); return NOTIFY_BAD; } |
185ae6d7a [PATCH] timer irq... |
120 |
per_cpu(touch_timestamp, hotcpu) = jiffies; |
8446f1d39 [PATCH] detect so... |
121 122 123 124 |
per_cpu(watchdog_task, hotcpu) = p; kthread_bind(p, hotcpu); break; case CPU_ONLINE: |
8446f1d39 [PATCH] detect so... |
125 126 127 128 |
wake_up_process(per_cpu(watchdog_task, hotcpu)); break; #ifdef CONFIG_HOTPLUG_CPU case CPU_UP_CANCELED: |
fc75cdfa5 [PATCH] cpu hotpl... |
129 130 |
if (!per_cpu(watchdog_task, hotcpu)) break; |
8446f1d39 [PATCH] detect so... |
131 |
/* Unbind so it can run. Fall thru. */ |
a4c4af7c8 [PATCH] cpu hoptl... |
132 133 |
kthread_bind(per_cpu(watchdog_task, hotcpu), any_online_cpu(cpu_online_map)); |
8446f1d39 [PATCH] detect so... |
134 135 136 137 138 139 140 141 142 |
case CPU_DEAD: p = per_cpu(watchdog_task, hotcpu); per_cpu(watchdog_task, hotcpu) = NULL; kthread_stop(p); break; #endif /* CONFIG_HOTPLUG_CPU */ } return NOTIFY_OK; } |
8c78f3075 [PATCH] cpu hotpl... |
143 |
static struct notifier_block __cpuinitdata cpu_nfb = { |
8446f1d39 [PATCH] detect so... |
144 145 146 147 148 149 |
.notifier_call = cpu_callback }; __init void spawn_softlockup_task(void) { void *cpu = (void *)(long)smp_processor_id(); |
07dccf334 [PATCH] check ret... |
150 |
int err = cpu_callback(&cpu_nfb, CPU_UP_PREPARE, cpu); |
8446f1d39 [PATCH] detect so... |
151 |
|
07dccf334 [PATCH] check ret... |
152 |
BUG_ON(err == NOTIFY_BAD); |
8446f1d39 [PATCH] detect so... |
153 154 |
cpu_callback(&cpu_nfb, CPU_ONLINE, cpu); register_cpu_notifier(&cpu_nfb); |
e041c6834 [PATCH] Notifier ... |
155 |
atomic_notifier_chain_register(&panic_notifier_list, &panic_block); |
8446f1d39 [PATCH] detect so... |
156 |
} |