[vlc-devel] vlc: svn commit r23022 (funman)

Sigmund Augdal sigmund.augdal at gmail.com
Mon Jan 14 23:49:48 CET 2008


On 1/14/08, Rafaël Carré <funman at videolan.org> wrote:
>
> Le Sun, 13 Jan 2008 21:18:43 +0100,
> "Sigmund Augdal" <sigmund.augdal at gmail.com> a écrit :
>
> > On 1/13/08, Rafaël Carré <funman at videolan.org> wrote:
> > >
> > > Le Sat, 12 Jan 2008 21:01:31 +0100,
> > > "Sigmund Augdal" <dnumgis at videolan.org> a écrit :
> > >
> > > > On 11/11/07, Subversion daemon <svn at videolan.org> wrote:
> > > > >
> > > > > r23022 | funman | 2007-11-11 19:48:07 +0100 (Sun, 11 Nov 2007)
> > > > > | 3 lines Changed paths:
> > > > >    M /trunk/modules/audio_output/alsa.c
> > > > >
> > > > > alsa: use the pcm status to compute the delay. fix #894 (spdif
> > > > > passthrough)
> > > > > Please report loudly any regression, using spdif or not.
> > > >
> > > >
> > > > THIS CHANGESET CAUSES MY VLC TO SPAM THE LOG WITH MESSAGES LIKE
> > > > THIS: [00000375] main audio output debug: audio output is
> > > > starving (30777), playing silence
> > >
> > > Thanks for reporting loudly ;)
> > >
> > > > (I'm not sure if the audio output has actually degraded at all,
> > > > but you asked for loud replies for any regressions)
> > >
> > > I'm experiencing the same problem with OSS, so I doubt it is an ALSA
> > > specific regression.
> > > Did you try reverting alsa.c ?
> > >
> > > Note that to get rid of these messages, i use pause twice.
> >
> >
> > I get these messages perhaps 90% of the time with trunk, but not ever
> > if I revert this specific change. I also don't get these with OSS.
>
> thresh also reported a "clicking sound" problem with this changeset.
>
> I will study again alsa api, and put some more verbose debug to find
> what's wrong.
>
> I suppose I didn't catch some corner case.


I did investigate a bit here and found that for me
snd_pcm_status_get_delay() allways return 0, while snd_pcm_delay()
return the value 15535, but sometimes slightly lower. Some times below
15400, but never above 15535 as far as I can see.
The crittical point here is anyway that the replacement function used
in the abovementioned changeset allways returns zero.

> Sigmund
> >
> > > Sigmund
> > > >
> > > > > http://trac.videolan.org/vlc/changeset/23022
> > > > > _______________________________________________
> > > > > vlc-devel mailing list
> > > > > To unsubscribe or modify your subscription options:
> > > > > http://mailman.videolan.org/listinfo/vlc-devel
> > > > >
> > >
> > >
> > > --
> > > Rafaël Carré
> > >
> > > _______________________________________________
> > > vlc-devel mailing list
> > > To unsubscribe or modify your subscription options:
> > > http://mailman.videolan.org/listinfo/vlc-devel
> > >
> > >
> > >
>
>
> --
> Rafaël Carré
>
> _______________________________________________
> vlc-devel mailing list
> To unsubscribe or modify your subscription options:
> http://mailman.videolan.org/listinfo/vlc-devel
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.videolan.org/pipermail/vlc-devel/attachments/20080114/41300bf8/attachment.html>


More information about the vlc-devel mailing list