Hi,
I tried to add dump_stack() in softdog like below:
100 static void watchdog_fire(unsigned long data)
101 {
102 if (test_and_clear_bit(0, &orphan_timer))
103 module_put(THIS_MODULE);
104
105 if (soft_noboot)
106 printk(KERN_CRIT PFX "Triggered - Reboot ignored.\n");
107 else if (soft_panic) {
108 printk(KERN_CRIT PFX "Initiating panic.\n");
109 panic("Software Watchdog Timer expired.");
110 } else {
111 printk(KERN_CRIT PFX "Initiating system reboot.\n");
dump_stack(); <--------add here to dump current
process call trace before restart
112 emergency_restart();
113 printk(KERN_CRIT PFX "Reboot didn't ?????\n");
114 }
115 }
I would like to see the stack trace of current process right before
emergency_restart to see what is going on in the user mode current
process to find clue what triggers the softdog timer out. I run a
stress program to trigger softdog timer out and see stack trace below:
#stress --cpu 1000 --io 1000 --vm 1000 --hdd 1000 --vm-bytes 128M --timeout 300s
stress: info: [12774] dispatching hogs: 1000 cpu, 1000 io, 1000 vm, 1000 hdd
SoftDog: Initiating system reboot.
Call Trace:
<IRQ> [<ffffffff882ce22e>] :softdog:watchdog_fire+0x8b/0xa0
[<ffffffff80090175>] run_timer_softirq+0x133/0x1af
[<ffffffff80010558>] __do_softirq+0x5e/0xd6
[<ffffffff8005a2fc>] call_softirq+0x1c/0x28
[<ffffffff80068863>] do_softirq+0x2c/0x85
[<ffffffff80059c8e>] apic_timer_interrupt+0x66/0x6c
<EOI>
I am hoping that dump_stack will dump call trace of stress program,
but instead, it dumps kernel function that kicks of watchdog_fire in
softdog which isn't really that helpful to find out what user land
process to cause the softdog timer out. do you have better idea how I
can do this.
Thanks
Vincent
--
To unsubscribe from this list: send the line "unsubscribe linux-watchdog" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
