[Cbe-oss-dev] PS3 sound driver / alsa problem with latest 2.6.28 kernel (2.6.28-09223-gc2fe70d)

[Stefan Wald]

Unfortunately, the software that produces this bug is too big to send
(hundreds of MB).
I think that the message 'DMA ch x is not stopped' is a follow-on problem
that shows up when the program terminates.
The main problem is that alsa all of a sudden stops returning data for
writei calls. (as said before this happens intermittendly.)
underruns are probabely involved in that situation.
Any data I can gather that would help isolate the problem?
Thanks.

> On Mon, 26 Jan 2009 21:54:01 +0100
> Stefan Wald <WALD at de.ibm.com> wrote:
>
> >
> > nobody have an idea?
> > can spe activity disturb alsa? what can i do to isolate the problem?
>
> The message 'DMA ch x is not stopped' meant that the driver found the
> DMA operation was still in progress when it was expected to be stopped.
> It was not expected to happen generally. The buffer underrun may cause
> the symptom, although I've never seen that case.
>
> Could you send me the program which can reproduce the symptom?
>
> regards
>
> > ----- Forwarded by Stefan Wald/Germany/IBM on 01/26/2009 09:51 PM -----
> >

> >   From:       Stefan Wald/Germany/IBM

> >

> >   To:         cbe-oss-dev at ozlabs.org

> >

> >   Date:       01/13/2009 09:29 PM

> >

> >   Subject:    Re: PS3 sound driver / alsa problem with latest 2.6.
> 28 kernel (2.6.28-09223-gc2fe70d)

> >

> >
> >
> >
> >
> > Hi, forgot to mention that I'm getting these dmesgs:
> > snd_ps3_verify_dma_stop: DMA ch 0 is not
> > stopped.<6>snd_ps3_verify_dma_stop: DMA ch 0 is not stopped.
> > hopefully this rings a bell...
> > Thanks,
> > Stefan,
> >
> > > cbe-oss-dev
> > >
> > > 01/12/2009 11:02 PM
> > >
> > > Hi,
> > > I experience an intermittend alsa sound lockup with the latest 2.6.
> > > 28 kernel on PS3. However, this problem has been around for a while
> > > ... probably since the very beginning ;-)
> > > I use ALSA in NONBLOCK mode and it generally works fine. But
> > > intermittendly it looks like alsa locks up (in most cases shortly
> > > after playback start).
> > > writei() then constantly returns -11 and snd_pcm_wait() never
> > > returns a 1 indicating that alsa is ready again to accept new frames.
> > > It's my impression that this lockup occurs more often in situations
> > > where there is high libspe activity with high memory utilization,
> > > but that's just a perception.
> > > My ALSA initialization is very similar to the one used in mplayer.
> > > I have played with the initialization: disabled underrun reporting,
> > > varied buffer sizes... but no change.
> > > Anything special I need to consider for PS3?
> > >
> > > Thanks,
> > > Stefan
> >
> > _______________________________________________
> > cbe-oss-dev mailing list
> > cbe-oss-dev at ozlabs.org
> > https://ozlabs.org/mailman/listinfo/cbe-oss-dev
> >
>
> --
> Masakazu Mokuno
>