softlock---Kernel panic - not syncing: softlockup: hung tasks

第一次解一个softlockup的bug，FAE回复的信息都不能很好的分析问题的根源，而且之前对softlockup的认识比较肤浅，

所以自己捣鼓这怎么让kernel自己产生个softlockup的warning出来。

 

基本原理：

每个CPU上起一个kernel thread，这个thread就用当前时间去更新一个percpu的全局时间锉；同时在tick中断中检查当前的时间

和这个全局的时间锉之间的差值是不是大于一个设定的阈值，对于linux-2.6.27,这个阈值设的是60s，如果60s这个内核线程还没

有得到一次调度的机会，那就说明这个CPU上发生死锁了。

 

在do_settimefoday中的clock_was_set 之前加上

preempt_disable();

while(1);   这种状态是irq 使能了，但是抢占关闭了，没有任何任务能够抢占CPU17上的settimeofday这个task，也正因为如此

                  该CPU上的softlockup内核线程得不到运行，全局时间锉与当前时间差值超过阈值后，在tick中断中检测这种情况。

 

然后遍一个调用settimefoday的小程序，运行，过60s后出现如下warning

 

 

 

 

用同样的方法在ARM上试

BUG: soft lockup - CPU#2 stuck for 22s! [settimeofday:485]
Modules linked in:

Pid: 485, comm:         settimeofday
CPU: 2    Not tainted  (3.3.0-rc3+ #48)
PC is at do_settimeofday+0xd8/0xe4
LR is at do_settimeofday+0xd7/0xe4
pc : [<800491e8>]    lr : [<800491e7>]    psr: 40000033  arm-non-linux-addr2line -e vmlinux800491e8 可以得到对应是哪一行代码
sp : 86d97f30  ip : 00000000  fp : 00000000
r10: 00000000  r9 : 86d96000  r8 : 806dc144
r7 : 00000001  r6 : 00000421  r5 : 806dc140  r4 : 80711180
r3 : 806e2d9c  r2 : 00000000  r1 : 00000422  r0 : 00000000
Flags: nZcv  IRQs on  FIQs on  Mode SVC_32  ISA Thumb  Segment user
Control: 50c53c7d  Table: 66dd806a  DAC: 00000015
[<80012639>] (unwind_backtrace+0x1/0xa4) from [<8005ddf9>] (watchdog_timer_fn+0x145/0x190)
[<8005ddf9>] (watchdog_timer_fn+0x145/0x190) from [<80036f91>] (hrtimer_run_queues+0xf9/0x240)
[<80036f91>] (hrtimer_run_queues+0xf9/0x240) from [<80027a01>] (run_local_timers+0xd/0x18)
[<80027a01>] (run_local_timers+0xd/0x18) from [<80027a33>] (update_process_times+0x27/0x54)
[<80027a33>] (update_process_times+0x27/0x54) from [<8004d4a3>] (tick_periodic+0x4b/0xb0)
[<8004d4a3>] (tick_periodic+0x4b/0xb0) from [<8004d593>] (tick_handle_periodic+0x23/0x78)
[<8004d593>] (tick_handle_periodic+0x23/0x78) from [<80012153>] (twd_handler+0x23/0x2c)
[<80012153>] (twd_handler+0x23/0x2c) from [<80060a5b>] (handle_percpu_devid_irq+0x73/0xc8)
[<80060a5b>] (handle_percpu_devid_irq+0x73/0xc8) from [<8005e163>] (generic_handle_irq+0x23/0x34)
[<8005e163>] (generic_handle_irq+0x23/0x34) from [<8000df8b>] (handle_IRQ+0x4f/0x94)
[<8000df8b>] (handle_IRQ+0x4f/0x94) from [<80008431>] (gic_handle_irq+0x29/0x84)
[<80008431>] (gic_handle_irq+0x29/0x84) from [<8000d1bf>] (__irq_svc+0x3f/0x80)
Exception stack(0x86d97ee8 to 0x86d97f30)
7ee0:                   00000000 00000422 00000000 806e2d9c 80711180 806dc140
7f00: 00000421 00000001 806dc144 86d96000 00000000 00000000 00000000 86d97f30
7f20: 800491e7 800491e8 40000033 ffffffff
[<8000d1bf>] (__irq_svc+0x3f/0x80) from [<800491e8>] (do_settimeofday+0xd8/0xe4)  死锁的位置在do_settimeofday+0xd8
[<800491e8>] (do_settimeofday+0xd8/0xe4) from [<80020ced>] (do_sys_settimeofday+0x49/0xa8)
[<80020ced>] (do_sys_settimeofday+0x49/0xa8) from [<80020d71>] (sys_settimeofday+0x25/0xb8)
[<80020d71>] (sys_settimeofday+0x25/0xb8) from [<8000d681>] (ret_fast_syscall+0x1/0x5c)

 

小试了下牛刀，等待FAE回复。

 

 =============================================================================

333473

BUG: soft lockup - CPU#0 stuck for 61s! [events/0:51] 
Modules linked in: OSLXcontext ippathmgt kmd kprn x_tables ip6_tables ip_tables ipv6 binfmt_misc 8021q 
Cpu 0 
$ 0   : 0000000000000000 ffffffff81410000 0000000000000003 0000000000000000 
$ 4   : 0000000000000010 0000000000000010 0000000000000010 a80000041fa0bdd0 
$ 8   : 0000000000000000 a80000041fbed080 0000000000000000 0000000000000001 
$12   : a80000041fa0bfe0 0000000000008c00 ffffffff813b4000 000000000000009a 
$16   : ffffffff813d0000 a8000004132faf00 000000000000000f 0000000000000001 
$20   : 0000000000000000 ffffffff8119a920 0000000000000000 fffffffffffffffe 
$24   : 0000000000000002 ffffffff81103e40 
$28   : a80000041fa08000 a80000041fa0be00 ffffffff8136e3f0 ffffffff811898cc 
Hi    : 0000000000000000 
Lo    : 0000000000000000 
epc   : ffffffff81189784 smp_call_function_mask+0x144/0x258 
    Not tainted 
ra    : ffffffff811898cc smp_call_function+0x34/0x78 
Status: 10008ce3    KX SX UX KERNEL EXL IE 
Cause : 40808000 
PrId  : 000d030b (Cavium Octeon) 
BUG: soft lockup - CPU#10 stuck for 61s! [ntpdate:3032] 
Kernel panic - not syncing: softlockup: hung tasks 
------------[ cut here ]------------ 
WARNING: at kernel/smp.c:332 smp_call_function_mask+0x250/0x258() 
Modules linked in: OSLXcontext ippathmgt kmd kprn x_tables ip6_tables ip_tables ipv6 binfmt_misc 8021q 
Call Trace: 
[] dump_stack+0x8/0x34 
[] warn_on_slowpath+0x58/0x90 
[] smp_call_function_mask+0x250/0x258 
[] smp_call_function+0x34/0x78 
[] panic+0x98/0x220 
[] softlockup_tick+0x290/0x310 
[] update_process_times+0x34/0x78 
[] tick_sched_timer+0x58/0xd8 
[] __run_hrtimer+0xb0/0xc8 
[] hrtimer_interrupt+0x158/0x210 
[] c0_compare_interrupt+0x11c/0x148 
[] _handle_IRQ_event+0x5c/0xb8 
[] handle_IRQ_event+0x88/0x140 
[] handle_percpu_irq+0x78/0xd8 
[] plat_irq_dispatch+0xc8/0x1d0 
[] ret_from_irq+0x0/0x4 
[] smp_call_function_mask+0x144/0x258 
[] smp_call_function+0x34/0x78 
[] on_each_cpu+0x2c/0x88 
[] run_workqueue+0xc8/0x1a0 
[] worker_thread+0x60/0xd0 
[] kthread+0x5c/0xb0 
[] kernel_thread_helper+0x10/0x18 
 
---[ end trace 1770b5013095d915 ]--- 
Rebooting in 10 seconds..<3> 
Modules linked in: OSLXcontext ippathmgt kmd kprn x_tables ip6_tables ip_tables ipv6 binfmt_misc 8021q 
Cpu 10 
$ 0   : 0000000000000000 a80000041cf3be70 0000000000000003 0000000000000000 
$ 4   : 0000000000000010 0000000000000010 0000000000000010 a80000041cf3bca0 
$ 8   : 0000000000000000 000000003b9aca00 ffffffff81410000 ffffffff81412520 
$12   : 0000000000000000 ffffffffc0000008 ffffffff81191258 fffffffff0000000 
$16   : ffffffff813d0000 a80000041feaca80 000000000000000f 0000000000000001 
$20   : 0000000000000000 ffffffff8117e610 0000000000000000 0000000000000001 
$24   : 0000000000000042 ffffffff81103e40 
$28   : a80000041cf38000 a80000041cf3bcd0 0000000010013fe0 ffffffff811898cc 
Hi    : 03d0900000000000 
Lo    : 0487ab0000000000 
epc   : ffffffff81189778 smp_call_function_mask+0x138/0x258 
    Tainted: G        W 
ra    : ffffffff811898cc smp_call_function+0x34/0x78 
Status: 10008ce3    KX SX UX KERNEL EXL IE  
Cause : 40808000 
PrId  : 000d030b (Cavium Octeon) 
Kernel panic - not syncing: softlockup: hung tasks 
------------[ cut here ]------------ 
WARNING: at kernel/smp.c:332 smp_call_function_mask+0x250/0x258() 
Modules linked in: OSLXcontext ippathmgt kmd kprn x_tables ip6_tables ip_tables ipv6 binfmt_misc 8021q 
Call Trace: 
[] dump_stack+0x8/0x34 
[] warn_on_slowpath+0x58/0x90 
[] smp_call_function_mask+0x250/0x258 
[] smp_call_function+0x34/0x78 
[] panic+0x98/0x220 
[] softlockup_tick+0x290/0x310 
[] update_process_times+0x34/0x78 
[] tick_sched_timer+0x58/0xd8 
[] __run_hrtimer+0xb0/0xc8 
[] hrtimer_interrupt+0x158/0x210 
[] c0_compare_interrupt+0x11c/0x148 
[] _handle_IRQ_event+0x5c/0xb8 
[] handle_IRQ_event+0x88/0x140 
[] handle_percpu_irq+0x78/0xd8 
[] plat_irq_dispatch+0xc8/0x1d0 
[] ret_from_irq+0x0/0x4 
[] smp_call_function_mask+0x138/0x258 
[] smp_call_function+0x34/0x78 
[] on_each_cpu+0x2c/0x88 
[] do_settimeofday+0x14c/0x180 
[] SyS_clock_settime+0xec/0x108 
[] compat_sys_clock_settime+0x78/0x90 
[] handle_sysn32+0x44/0x84