Hi, I emailed alsa-devel yesterday reporting a possible deadlock in the alsa-pulse plugin where a call to snd_pcm_writei would block forever after pausing and then resuming with snd_pcm_pause. The original email is here: http://mailman.alsa-project.org/pipermail/alsa-devel/2010-November/033242.html I've since managed to write a testcase that reproduces the same problem using the PulseAudio client API directly. The testcase is attached. If you can reproduce the problem, it'll eventually stop making progess and continuously print 'wsize=3000, need >= 6000, corked=0'. The testcase may look unusual, but I think it's an accurate minimization of the problem I originally encountered with the alsa-pulse plugin. It seems that corking and uncorking the stream when the writable_size is less than minreq can result in playback not resuming after uncorking. The testcase tries to exacerbate the problem by writing additional data to the stream just before pausing so that writable_size is less than minreq would be. After uncorking, is_corked returns false and writable_size never grows. Thanks, -mjg -- Matthew Gregan |/ /| kinetik at flim.org -------------- next part -------------- A non-text attachment was scrubbed... Name: ptest.c Type: text/x-csrc Size: 5080 bytes Desc: not available URL: <http://lists.freedesktop.org/archives/pulseaudio-discuss/attachments/20101102/2e644f52/attachment.c>
