Re: Soft lockup during suspend since ~2.6.36

[Thilo-Alexander Ginkel]

On Sun, Apr 3, 2011 at 19:32, Thilo-Alexander Ginkel <thilo@xxxxxxxxxx> wrote:
> A good night of sleep and some Googling later I am under the
> impression that this issue was already supposedly fixed by the
> following commit a long time ago:
> Âhttp://git.kernel.org/?p=linux/kernel/git/jikos/hid.git;a=commitdiff;h=9c9e54a8df0be48aa359744f412377cc55c3b7d2
> Âhttp://permalink.gmane.org/gmane.linux.kernel/1024386
>
> This makes some sense as versions closer to 2.6.36 only lock up
> sporadically whereas the original bug locks up reliably. I'll try to
> get a backtrace for the sporadic bug, but may only be able to do so
> using a tainted kernel as my text console does not survive
> suspend/resume without proprietary drivers (and the serial console
> also breaks during the relevant suspend phase). I hope, this is
> acceptable.

All right. I have actually been able to obtain a backtrace for a
2.6.36 kernel, which only shows this problem sporadically (so this
probably has a different root cause).

It all starts with the suspend sequence hanging while shutting down
non-boot CPUs (the CPU# varies from case to case):
  https://secure.tgbyte.de/dropbox/ueph3Ohm-1.jpg

Then, after some time, the following backtrace is written:
  https://secure.tgbyte.de/dropbox/ueph3Ohm-2.jpg

Transcript:
| RIP: _raw_spin_unlock_irqrestore
| Call Stack:
|   _set_cpus_allowed_ptr
|   worker_maybe_bind_and_lock
|   rescuer_thread
|   rescuer_thread
|   kernel_thread_helper
|   kthread
|   kernel_thread_helper
|   kthread

After some more time, the following backtrace is printed:
  https://secure.tgbyte.de/dropbox/ueph3Ohm-3.jpg

Transcript:
| RIP: worker_maybe_bind_and_lock
| Call Stack:
|   rescuer_thread
|   rescuer_thread
|   kernel_thread_helper
|   kthread
|   kernel_thread_helper
|   kthread

>From then on these two backtraces are printed in an alternating fashion.

Unfortunately, the top of the output is missing as it does not fit on
screen (does someone know an even smaller console font?), but I assume
it's the deadlock detection.

Thanks,
Thilo
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/