On Tue, 20.01.09 02:21, Aleksander Kamenik (aleksander at krediidiinfo.ee) wrote:
>
> Lennart Poettering wrote:
> > On Tue, 20.01.09 00:45, Aleksander Kamenik (aleksander at krediidiinfo.ee) wrote:
> >
> >>> One thing you might try before you file that bug is commenting the
> >>> invocation of snd_pcm_hw_params_set_periods_integer() in that
> >>> function, recompiling and testing whether that changes the behaviour.
> >> I guess your idea was to include a goto finish; right before the
> >> problematic code with snd_pcm_hw_params_set_periods_near() by commenting
> >> the 'if ((ret = snd_pcm_hw_params_set_periods_integer(pcm_handle,
> >> hwparams)) < 0)' line?
> >
> > No, my idea was to comment the two lines that deal with
> > snd_pcm_hw_params_set_periods_integer(), as I said.
> >
> > i.e., comment these two lines:
> >
> > <snip>
> > if ((ret = snd_pcm_hw_params_set_periods_integer(pcm_handle, hwparams)) < 0)
> > goto finish;
> > </snip>
>
> Oh, that's how I understood it at the beginning and tried it. This
> didn't do anything at all, cause the comparison fails and goto finish;
> is not executed. I got exactly the same errors at the same place when
> running gdb. And the same behaviour. Of course I might have missed
> something, but it looked exactly like my previous run with the original
> code. That's why I thought you meant something else.
OK, if that minimal change doesn't really change anything, then please
file a bug against ALSA as mentioned.
>
> >
> >> However I also tried commenting out the
> >> snd_pcm_hw_params_set_periods_near() function calls. Thanks to this my
> >> USB sound card popped up in the pulse audio volume control. My
> >> integrated nvidia chip worked.
> >
> > Ah, that's interesting. Apparently your sound card does not support
> > integral number of periods. Weird thing. Could you please paste the
> > output of "pulseaudio -vvvv" with that minimal commenting change made?
>
> Right. Starting up pulseaudio with this patch:
>
> diff -Naur pulseaudio-0.9.14/src/modules/alsa-util.c
> pulseaudio-0.9.14.changed/src/modules/alsa-util.c
> --- pulseaudio-0.9.14/src/modules/alsa-util.c 2009-01-13
> 01:11:38.000000000 +0200
> +++ pulseaudio-0.9.14.changed/src/modules/alsa-util.c 2009-01-20
> 00:25:46.000000000 +0200
> @@ -369,11 +369,11 @@
>
> if ((ret = snd_pcm_hw_params_set_periods_integer(pcm_handle,
> hwparams)) < 0)
> goto finish;
> -
> +/*
> if (_periods > 0) {
>
> - /* First we pass 0 as direction to get exactly what we asked
> - * for. That this is necessary is presumably a bug in ALSA */
> + * First we pass 0 as direction to get exactly what we asked
> + * for. That this is necessary is presumably a bug in ALSA *
>
> dir = 0;
> if ((ret = snd_pcm_hw_params_set_periods_near(pcm_handle,
> hwparams, &_periods, &dir)) < 0) {
> @@ -385,7 +385,7 @@
> }
> }
> }
> -
> +*/
> if (_period_size > 0)
> if ((ret = snd_pcm_hw_params_set_buffer_size_near(pcm_handle,
> hwparams, &buffer_size)) < 0)
> goto finish;
>
> Result:
>
> http://pastebin.com/m5eb3e962
No, this patch is a bit too intrusive. We need to set the
buffer/period settings. On your setup ALSA refuses that in every way
so we this should be considered a bug in ALSA.
Lennart
--
Lennart Poettering Red Hat, Inc.
lennart [at] poettering [dot] net ICQ# 11060553
http://0pointer.net/lennart/ GnuPG 0x1A015CC4
