[pulseaudio-discuss] Pulseaudio 0.16-test4 - my experience

Maxim Levitsky maximlevitsky at gmail.com
Sun Aug 16 20:19:15 PDT 2009


On Sun, 2009-08-16 at 23:12 +0200, Lennart Poettering wrote:
> On Tue, 11.08.09 00:48, Maxim Levitsky (maximlevitsky at gmail.com) wrote:
> 
> > > > Small progress, this only happens with pulse backend of canberra, which
> > > > isn't used in 9.04, which explains why I see that bug now.
> > > > It is probably the canberra bug, maybe it doesn't tell pulse that sound
> > > > is mono? (I think that notify sounds are mono)
> > > 
> > > Hmm, Ubuntu didn't use the pa backend for libcanberra? Oh my!
> > > 
> > > Hmm, we had some recent changes in the surround sound handling in
> > > libcanberra. Which version are you running?
> > 
> > I pulled it from git repository.
> > 
> > git log
> > 
> > commit 605aee559f5bd6d53c4c93a9c3bbeefdcf10cad2
> > Author: Lennart Poettering <lennart at poettering.net>
> > Date:   Wed Aug 5 03:19:44 2009 +0200
> > 
> >     prepare release 0.15
> > 
> > commit 593511bad132e0d262780918cce469ad69804b52
> > Author: Lennart Poettering <lennart at poettering.net>
> > Date:   Tue Aug 4 01:55:53 2009 +0200
> > 
> >     gtk: don't take gdk lock since it is not recursive and it might
> > already be taken
> 
> 
> That's pretty recent. ;-)
> 
> Hmm, could you please file a bug on fdo's bz, so that I don't forget
> about looking into this?
https://bugs.freedesktop.org/show_bug.cgi?id=23358
> 
> > > > Just one issue though. If a stream is played at low volume, and I click
> > > > a button, a noise sound is emmited and it almost covers the notify
> > > > sound.
> > > > maybe this is a HW fault, or I suspect that PA reconfigures the played
> > > > sound to play at lower volume, and this gives that noise sound. Anyway
> > > > this is a bug.
> > > 
> > > I don't follow. Could you explain this again?
> > Sure!
> > 
> > 
> > Let say I start a music player, and set its PA stream volume to low
> > (rhythbox for example does that)
> > 
> > Now I click on a button.
> > 
> > Now what happens:
> > 
> > 1 - PA sets hardware volume to high
> > 2 - PA lowers music stream volume to low, so it plays at same volume.
> > 3 - PA plays the notify sound.
> > 
> > What is wrong is that the above sequence triggers a short noise, which
> > origin isn't yet known to me (also it is possible to hear the music at
> > high volume for a moment).
> 
> Hmm, the short noise might be a problem with the volume setting
> funcitonality of your driver/hw? Can you trigger it by low-level alsa
> volume changes? Otherwise it might be triggered by a faulty
> snd_pcm_rewind() implementation in your driver?
Could be, but like I said, this sometimes is a hardware limitation of
the hardware.

I use typical snd-hda-intel with realtec ALC268 codec


> 
> There is a little race here indeed, since we cannot change the mixer
> volume and update our PCM data atomically at the same time. Usually
> the hw mixer change should come first, the PCM data adjustment second,
> which means this should annoy only if the overall volume is increased,
> not when it is decreased.
> 
> In my checkout i have now modified the flat volume logic so that sink
> input volume changes can increase the sink volume, but they never
> dcrease it again. As a side effect this will make your problem go away
> since a corollary of this is that a stream volume is always <= the
> sink volume, meaning that for event sounds the sink volume is never
> increased.
> 
> This new logic hence makes the sink volume something of a "barrier"
> for sink input volumes. By default, no stream can increase the sink
> volume beyond what was configured by the user.
This is very welcome change on its own, it was really annoying that
applications would hijack the master volume otherwise.

> 
> > Since the notify sound is itself very short, it is almost replaced by
> > this noise.
> > 
> > It is ether a hick-up in PA, while it changes the stream volume, or a
> > hardware problem.
> 
> Could be a driver problem, too. Which driver is this?
>  
> > I do like the idea of flat volumes though, this is quite right, and nice
> > addition. However in case this is a hardware fault, I think that final
> > volume of a card should also be software one, and hardware volume set to
> > high.
> > 
> > I don't know about this sound codec, but I do know that many sound cards
> > like to emit all kinds of noises, when their volumes are touched. 
> 
> According to Takashi noises like this and when playback is
> started/stopped should be considered driver problems and be fixed in
> software. So please make sure to file a bug against your driver!
I investigate a bit more.

Best regards,
	Maxim Levitky

> 
> Lennart
> 
> -- 
> Lennart Poettering                        Red Hat, Inc.
> lennart [at] poettering [dot] net
> http://0pointer.net/lennart/           GnuPG 0x1A015CC4
> _______________________________________________
> pulseaudio-discuss mailing list
> pulseaudio-discuss at mail.0pointer.de
> https://tango.0pointer.de/mailman/listinfo/pulseaudio-discuss




More information about the pulseaudio-discuss mailing list