[pulseaudio-discuss] realtime kernel RLIMIT_RTTIME crashes, rewinds

[David Henningsson]

On 2011-05-04 08:19, Tanu Kaskinen wrote:
> Sorry for the delayed reply.
>
> On Wed, 2011-04-27 at 13:23 +0300, Dan Muresan wrote:
>> I've patched (as in attachment) src/daemon/main.c to set the soft
>> rttime limit lower than the hard limit. This way I was able to get a
>> backtrace in gdb when pulseaudio received SIGXCPU due to excessive
>> real-time CPU usage:
>>
>> #0  0x0012d422 in ?? ()
>> #1  0x00475edb in write () at ../sysdeps/unix/syscall-template.S:82
>> #2  0x00153a55 in pa_fdsem_post (f=0x806e178) at pulsecore/fdsem.c:205
>> #3  0x0013cb0f in push (l=0x8071bc8, p=0x8, wait_op=false) at
>> pulsecore/asyncq.c:161
>> #4  0x0013d291 in pa_asyncq_post (l=0x8071bc8, p=0x80736d8) at
>> pulsecore/asyncq.c:203
>> #5  0x0013c1ee in pa_asyncmsgq_post (a=0x8071a90, object=0xb6a16a08,
>> code=7, userdata=0x0, offset=0, chunk=0xbffff590, free_cb=0) at
>> pulsecore/asyncmsgq.c:139
>> #6  0x009811d7 in pstream_memblock_callback (p=0x808cf78, channel=0,
>> offset=0, seek=PA_SEEK_RELATIVE, chunk=0xbffff590, userdata=0x80b6480)
>> at pulsecore/protocol-native.c:4445
>> #7  0x001dd20b in ?? () from /usr/lib/libpulsecommon-0.9.22.so
>> #8  0x001c823e in ?? () from /usr/lib/libpulsecommon-0.9.22.so
>> #9  0x0021b5fb in pa_mainloop_dispatch () from /usr/lib/libpulse.so.0
>> #10 0x0021bb11 in pa_mainloop_iterate () from /usr/lib/libpulse.so.0
>> #11 0x0021bbd4 in pa_mainloop_run () from /usr/lib/libpulse.so.0
>> #12 0x08052e85 in main (argc=1, argv=0xbffff854) at daemon/main.c:974
>>
>> Can someone point out the likely cause, or is this not enough
>> information? This is pulseaudio 0.9.22 from Ubuntu Natty.
>
> I would guess that the excessive cpu use is caused by doing rewinds in
> rapid succession. That's just a guess, though - the backtrace doesn't
> really give any clue. There have been issues with rewinds happening in
> some kind of a loop. David Henningson has been working on this problem
> and he has made at least a few patches that help with the problem. The
> patches are in Pulseaudio's git repository at least. Since David works
> for Canonical, I would expect the fixes to be also in Ubuntu, but I
> don't know for sure.
>
> David, what's the status of the "fighting rewinds" patches in Ubuntu
> Natty? Is the problem supposed to be fixed already?
>

My rewind patches are both in Natty and stable-queue, but this bug is 
something different.
IIRC I had a quick look at it, didn't have time to look really deep, so 
I just responded in the launchpad bug Dan also posted in (see 
https://bugs.launchpad.net/ubuntu/+source/linux-rt/+bug/367671 )

-- 
David Henningsson
http://launchpad.net/~diwic