[alsa-devel] [PATCH v5] ASoC: omap-mcbsp: Add PM QoS support for McBSP to prevent glitches

[Matt Ranostay]

On Mon, Jan 9, 2017 at 3:56 PM, Tony Lindgren <tony at atomide.com> wrote:
> * Matt Ranostay <matt at ranostay.consulting> [170109 14:11]:
>> On Mon, Jan 9, 2017 at 1:56 PM, Tony Lindgren <tony at atomide.com> wrote:
>> > * Matt Ranostay <matt at ranostay.consulting> [170109 13:46]:
>> >> On Mon, Jan 9, 2017 at 11:21 AM, Tony Lindgren <tony at atomide.com> wrote:
>> >> > * Peter Ujfalusi <peter.ujfalusi at ti.com> [170109 05:13]:
>> >> >> On 01/05/2017 03:59 AM, Matt Ranostay wrote:
>> >> >> > We can get audio errors if hitting deeper idle states on omaps:
>> >> >> >
>> >> >> > [alsa.c:230] error: Fatal problem with alsa output, error -5.
>> >> >> > [audio.c:614] error: Error in writing audio (Input/output error?)!
>> >> >> >
>> >> >> > This seems to happen with off mode idle enabled as power for the
>> >> >> > whole SoC may get cut off between filling the McBSP fifo using DMA.
>> >> >> > While active DMA blocks deeper idle states in hardware, McBSP
>> >> >> > activity does not seem to do so.
>> >> >> >
>> >> >> > Basing the QoS latency calculation on the FIFO size, threshold,
>> >> >> > sample rate, and channels.
>> >> >>
>> >> >> Looks good to me, thank you!
>> >> >>
>> >> >> Acked-by: Peter Ujfalusi <peter.ujfalusi at ti.com>
>> >> >
>> >> > Noticed the following about 10 seconds into playing an mp3 file with
>> >> > mpg123 though:
>> >> >
>> >>
>> >> Didn't notice that happening for me. But haven't rebased for the last
>> >> couple days.
>> >> Does this happen 100% of the time?
>> >
>> > Yeah seems to based on three attempts. This with 4.10.0-rc2-next-20170109.
>> >
>>
>> Ok I'll test this myself this evening and report back.
>
> Oh and this is with omap2plus_defconfig with the following also enabled:
>
> CONFIG_DEBUG_LOCKDEP=y
> CONFIG_DEBUG_ATOMIC_SLEEP=y
>

Yeah I can reproduce the issue you are mentioning. Anyone have a
suggestion how on to resolve? Guessing the QoS function calls need to
be in an atomic context and the omap_mcbsp_* functions can sleep

> Tony