Bug#775756: -ao pulse stopped producing audio; alsa and oss still work

[josh at joshtriplett.org]

On Tue, Jan 20, 2015 at 11:53:54AM -0300, Felipe Sateler wrote:
> On Tue, Jan 20, 2015 at 11:49 AM, Josh Triplett <josh at joshtriplett.org> wrote:
> > On Tue, Jan 20, 2015 at 11:14:16AM -0300, Felipe Sateler wrote:
> >> On Tue, Jan 20, 2015 at 10:51 AM, Fabian Greffrath <fabian at greffrath.com> wrote:
> >> > Hi,
> >> >
> >> > Am Dienstag, den 20.01.2015, 10:11 -0300 schrieb Felipe Sateler:
> >> >> Pulseaudio remembers where a stream was playing to last time, and
> >> >> plugs the stream there.
> >> >
> >> > But then PA should be smart enough to realize when the device isn't even
> >> > connected anymore.
> >>
> >> Define enough ;). As far as I can tell, pulseaudio does detect if the
> >> hdmi device is plugged or not. However, behavior is a bit strange
> >> because different logic applies at different times.
> >>
> >> I'm guessing the problem here is that module-stream-restore is
> >> detecting it already knows about mplayer2, so it will move that stream
> >> to the hdmi device. However, being unplugged does not mean the HDMI
> >> device disappears (in cards where the hdmi device is a separate
> >> device). Therefore, silence.
> >
> > If it helps: snd_hda_codec_hdmi.
> >
> > At least in the PulseAudio control panel, the HDMI audio device did not
> > show up while I was observing the problem with mplayer2 not producing
> > audio output.
> 
> Please attach a verbose log of pulseaudio[1] while reproducing the
> problem, plus the output of `pactl list`

I don't yet have a recipe to reproduce the problem.  This time around,
connecting an HDMI audio device (DisplayPort, actually) and switching
audio back to speakers allowed me to *fix* the problem, such that the
default pulse audio output works again.

> If the device is not showing up in pavucontrol, then this would be
> indeed a bug in pulseaudio.

I can definitely confirm that the HDMI audio device did *not* show up in
the GNOME pulseaudio control panel when this problem had occurred; it
only showed the speaker device and the headphone device.  At that time,
switching the output back and forth between those did not cause mplayer2
to produce audio.

I definitely agree that the bug sounds more likely to fall in PulseAudio
than in mplayer2.

- Josh Triplett