On Thu, 2014-06-05 at 16:28 +0300, Nikos Chantziaras wrote: > On 05/06/14 16:17, Tanu Kaskinen wrote: > > On Wed, 2014-06-04 at 10:47 +0300, Nikos Chantziaras wrote: > >> When I boot the system (Gentoo), "pacmd list-sinks" says: > >> > >> left: 65520 / 100% / -0.01 dB, right: 65520 / 100% / -0.01 dB > >> > >> I have to first apply a manual volume change (with pavucontrol or kmix) > >> for it to get to 65536 (0dB.) > >> > >> I'm using PulseAudio 5.0 with software volume (hardware volume is not > >> supported by PulseAudio for my sound card, which is an Asus Xonar D1.) > >> > >> There is no "restore volume at startup" business going on. This is how > >> PA initializes the volume. Is this a known bug? > > > > No, it's not known (or at least I don't remember a similar report). What > > do you mean by there being no volume restoring going on? Have you > > removed module-device-restore from the configuration, or do you mean > > that there's no external program restoring the volume on boot? > > I meant that I have no external tools that would restore volumes on > startup (like kmix does by default, and I have disabled that option.) > Except for the ALSA hardware volume, which PA is unable to use and it > don't affect this specific issue. > > > > Can you reproduce the bug with these steps: > > > > 1. Disable autospawning: > > echo "autospawn = no" >> ~/.config/pulse/client.conf > > > > 2. Stop pulseaudio: > > killall pulseaudio > > > > 3. Start pulseaudio in a terminal with verbose logging: > > pulseaudio -vvv > > > > 4. Check the volume: > > pactl list sinks > > > > If you can reproduce the bug (i.e. the volume reported by pactl is not > > 0.00 dB), could you attach the log output? > > Yes, I can reproduce this. If I raise the volume to 0 db manually, then > exit PA with Ctrl+C and then run it again, the volume is again -0.01 dB. > > I am attaching the output of "pulseaudio -vvv". The wrong volume is coming from the module-device-restore database. Setting the volume to 0 db manually should update the volume in the database, but apparently it doesn't. This patch might fix the problem (so this probably was a known bug after all): http://cgit.freedesktop.org/pulseaudio/pulseaudio/commit/src/pulsecore/sink.c?id=e0e6bf687573ea8ee3b6b53f8e44bcb30f4914ec -- Tanu
