On Tue, 13 Jun 2023 11:55:16 +0200,
Takashi Iwai wrote:
>
> On Tue, 13 Jun 2023 11:34:53 +0200,
> Jon Hunter wrote:
> >
> > Commit 3ed2b549b39f ("ALSA: pcm: fix wait_time calculations") corrected
> > the PCM wait_time calculations and in doing so reduced the calculated
> > wait_time. This exposed an issue with the Tegra Master Volume Control
> > (MVC) device where the reduced wait_time caused the MVC to fail. For now
> > fix this by setting the default wait_time for Tegra to be 500ms.
> >
> > Fixes: 3ed2b549b39f ("ALSA: pcm: fix wait_time calculations")
> > Signed-off-by: Jon Hunter <jonathanh@xxxxxxxxxx>
>
> Hm, it's still not clear why it fails. The commit above changes the
> timeout of wait_for_avail() to the full-buffer + 10% margin. In
> thoery, the loop should abort after the full buffer read, and that
> must be enough. If there were a large FIFO behind, it might be
> overflow, but the fifo_size of Tegra driver seems 4, so it's
> negligible.
>
> If extending the timeout "fixes" the problem, it might indicate that
> the period update IRQ is triggered too late. Could you measure the
> timing of each snd_pcm_period_elapsed() and wait_for_avail() call?
OTOH, it's already at a pretty late stage for 6.4, and we need an
urgent regression fix. So it's better to paper over it now, while
hunting further for the real culprit.
Takashi
