[lttng-dev] ***UNCHECKED*** Re: Re: urcu workqueue thread uses 99% of cpu while workqueue is empty

[Mathieu Desnoyers]

----- On Jun 22, 2022, at 9:19 AM, Mathieu Desnoyers mathieu.desnoyers at efficios.com wrote:

> ----- On Jun 22, 2022, at 3:45 AM, Minlan Wang wangminlan at szsandstone.com wrote:
> 
> [...]
> 
> Hi Minlan,
> 
>> --
>> 1.8.3.1
>> 
>> 
>> And the lttng session is configured to trace these events:
>> kernel: syscall futex
> 
> On the kernel side, in addition to the syscall futex, I would really like to see
> what
> happens in the scheduler, mainly the wait/wakeup tracepoints. This can be added
> by using:
> 
> lttng enable-event -k 'sched_*'
> 
> This should help us confirm whether we indeed have a situation where queued wake
> ups
> happen to wake up a wait happening only later, which is unexpected by the
> current liburcu
> userspace code.
> 
> [...]
> 
>> ---
>> 
>> The babletrace output of this session is pretty big, 6 MB in size, i put it in
>> the attachment trace_0622.tar.bz2.
>> Let my know if your mailbox can't handle such big attachment.
> 
> It would be even better if you can share the binary trace, because then it's
> easy to
> load it in trace compass, cut away time ranges that don't matter, and lots of
> other
> useful stuff.

I just found the relevant snippet of documentation in futex(5):

       FUTEX_WAIT
              Returns 0 if the caller was woken up.  Note that a  wake-up  can
              also  be caused by common futex usage patterns in unrelated code
              that happened to have previously used the  futex  word's  memory
              location  (e.g., typical futex-based implementations of Pthreads
              mutexes can cause this under some conditions).  Therefore, call‐
              ers should always conservatively assume that a return value of 0
              can mean a spurious wake-up, and  use  the  futex  word's  value
              (i.e.,  the user-space synchronization scheme) to decide whether
              to continue to block or not.

I'm pretty sure this is what is happening here.

Thanks,

Mathieu


> 
> Thanks,
> 
> Mathieu
> 
> --
> Mathieu Desnoyers
> EfficiOS Inc.
> http://www.efficios.com

-- 
Mathieu Desnoyers
EfficiOS Inc.
http://www.efficios.com