Re: frequent lockups in 3.18rc4

From: Will Deacon
Date: Thu Dec 04 2014 - 05:51:05 EST

On Mon, Dec 01, 2014 at 06:36:04PM +0000, Linus Torvalds wrote:
> On Mon, Dec 1, 2014 at 10:25 AM, Kirill A. Shutemov
> <kirill@xxxxxxxxxxxxx> wrote:
> >
> > No idea about oom_score, but kernel happily accepts chmod on any file
> > under /proc/PID/net/.
> /proc used to accept that fairly widely, but no, we tightened things
> down, and core /proc files end up not accepting chmod. See
> 'proc_setattr()':
> if (attr->ia_valid & ATTR_MODE)
> return -EPERM;
> although particular /proc files could choose to not use 'proc_setattr'
> if they want to.
> The '/proc/pid/net' subtree is obviously not doing that. No idea why,
> and probably for no good reason.

I just hit another one of these, but it's slightly different this time:

[child1:811] [2219] execve(name="/proc/610/attr/keycreate", argv=0x3a044bf0, envp=0x3a04c810)

this guy disappears off into the execve and never returns. A little later,
another guy gets stuck on a completion after a sync:

[child0:856] [155] sync()

trinity-c0 D ffffffc000087570 0 856 612 0x00000000
Call trace:
[<ffffffc000087570>] __switch_to+0x74/0x8c
[<ffffffc0005350b4>] __schedule+0x204/0x670
[<ffffffc000535544>] schedule+0x24/0x74
[<ffffffc0005380a4>] schedule_timeout+0x134/0x18c
[<ffffffc000536204>] wait_for_common+0x9c/0x144
[<ffffffc0005362bc>] wait_for_completion+0x10/0x1c
[<ffffffc0001bbc14>] sync_inodes_sb+0x98/0x194
[<ffffffc0001c0244>] sync_inodes_one_sb+0x10/0x1c
[<ffffffc0001984c8>] iterate_supers+0x10c/0x114
[<ffffffc0001c04c0>] sys_sync+0x38/0xa4

The backtrace for 811 looks bogus to me (or we're missing some entries):

trinity-c1 R running task 0 811 612 0x00000000
Call trace:
[<ffffffc000087570>] __switch_to+0x74/0x8c
[<ffffffc0000ecb48>] __handle_domain_irq+0x9c/0xf4
[<ffffffc000301da4>] __this_cpu_preempt_check+0x14/0x20
[<ffffffc000538a2c>] _raw_spin_lock_irq+0x18/0x58
[<ffffffc000538cb4>] _raw_spin_unlock_irq+0x1c/0x48
[<ffffffc0000fa6bc>] run_timer_softirq+0x68/0x240
[<ffffffc0000b5b2c>] __do_softirq+0x110/0x244
[<ffffffc000301d84>] debug_smp_processor_id+0x18/0x24

and, as before, it has a weird child process that I can't backtrace:

trinity-c1 R running task 0 861 811 0x00000000
Call trace:

The RCU stall detector gets cross too, but the stall ends before it has
a chance to dump anything.

To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at
Please read the FAQ at