Skip to contentSkip to conversation

alsaloop falls into loop after SIGCONT

alsa-utils Dimitrios Semitsoglou-Tsiapos 1 participant 1 message Aug 19, 2017

lore
Share & tools
lore
#0 lore raw

Hello everyone,

I've ran into some issues while writing a wrapper around alsaloop. It's possible to make it fall into infinite loops of silence. Here is how to trigger one such loop:

alsaloop -C cloop -P Audio -t 100000 -v... & # Read attached asoundrc sleep 2
mpc play
pkill -STOP -x alsaloop; sleep 0.1; pkill -CONT -x alsaloop

On the great majority of times, this happens:

...
pool took 68171us
playback hw:Audio/capture cloop: pollfds handle
playback hw:Audio: delay 2160 / 38400 / 0
capture cloop: delay 15120 / 38400 / 0
playback hw:Audio/capture cloop: prevents = 0x4, crevents = 0x0 playback hw:Audio/capture cloop: queued 5040/18000 samples
playback hw:Audio: end delay 5040 / 38400 / 0
capture cloop: end delay 18000 / 38400 / 0
playback hw:Audio/capture cloop: processing time 69us
pool took 192670us
playback hw:Audio/capture cloop: pollfds handle
playback hw:Audio: delay error: Broken pipe / 38400 / 0
capture cloop: delay 8400 / 38400 / 0
playback hw:Audio/capture cloop: prevents = 0xc, crevents = 0x1 underrun for playback hw:Audio
playback hw:Audio/capture cloop: xrun sync 0 1
sync: cdelay=18000(18000), pdelay=9600(9600), fill=4800 (delay=27600), src_out=0
sync: cbufcount=9600, pbufcount=9600
sync: capt stop removed 9600 samples
playback hw:Audio/capture cloop: xrun sync 0 1
sync: cdelay=18000(18000), pdelay=0(0), fill=4800 (delay=18000), src_out=0
sync: cbufcount=0, pbufcount=0
sync: capt stop removed 0 samples
playback hw:Audio/capture cloop: xrun sync 0 1
sync: cdelay=18000(18000), pdelay=0(0), fill=4800 (delay=18000), src_out=0
sync: cbufcount=0, pbufcount=0
sync: capt stop removed 0 samples
playback hw:Audio/capture cloop: xrun sync 0 1
...

Moving outwards, some observations,

* remove_samples() falls into if (loop->play->buf == loop->capt->buf), which is what decides to always remove 0 samples after some point (as seen in output).

* cdelay (from snd_pcm_delay(capt->handle, &cdelay)) is consistently 18000, meaning that delay1 > fill in xrun_sync() will always be true as fill is fixed during the run. If one forces this check to fail (by setting cdelay to 0), alsaloop recovers, with no audible delay compared to what's being played back.

* It is this code that leads to xrun() in the first place:

	if (avail == -EPIPE) {
		if ((err = xrun(lhandle)) < 0)

* The issue is hinted even earlier, as after

	err = snd_pcm_poll_descriptors_revents(play->handle, fds,
					       play->pollfd_count,
					       &prevents);

(prevents & POLLERR) is true.

* Depending on when STOP is sent, the same sometimes holds true after thread_job1's

	err = poll(pfds, j, wake);

on pfds[i].revents.

I am not sure what the correct way of fixing this is, especially because I couldn't figure out why snd_pcm_delay returns this specific value. I can only guess it is related to aloop.

Other ways of throwing alsaloop into infinite loops is unplugging a soundcard that it's using (I believe it should exit in this case) and suspending/resuming the system. I haven't gone through the codepaths in those cases carefully yet, but I'm interested in fixing them eventually.

Please let me know if I can provide more useful output, Dimitri

[alsa-info.sh]http://www.alsa-project.org/db/?f=89eb0056b0876a3b4fb377c39d5bc81405a308a7

# playback PCM device: using loopback subdevice 0,0
pcm.amix {
  type dmix
  ipc_key 219345
  slave.pcm "hw:Loopback,0,0"
}
# capture PCM device: using loopback subdevice 0,1
pcm.asnoop {
  type dsnoop
  ipc_key 219346
  slave.pcm "hw:Loopback,0,1"
}
# duplex device combining our PCM devices defined above
pcm.aduplex {
  type asym
  playback.pcm "amix"
  capture.pcm "asnoop"
}

# ------------------------------------------------------

# for jack alsa_in and alsa_out: looped-back signal at other ends
pcm.ploop {
  type plug
  slave.pcm "hw:Loopback,1,1"
}

pcm.cloop {
type dsnoop
ipc_key 219348
slave.pcm "hw:Loopback,1,0"
}

# ------------------------------------------------------

# default device

pcm.!default {
type plug
slave.pcm "aduplex"
}