[PATCH] ALSA: compress: allow pause and resume during draining

[Gyeongtaek Lee]

On Mon, 26 Oct 2020 18:01:33 +0100, Takashi Iwai wrote:
>On Mon, 26 Oct 2020 10:18:38 +0100,
>Gyeongtaek Lee wrote:
>> 
>> With a stream with low bitrate, user can't pause or resume the stream
>> near the end of the stream because current ALSA doesn't allow it.
>> If the stream has very low bitrate enough to store whole stream into
>> the buffer, user can't do anything except stop the stream and then
>> restart it from the first because most of applications call draining
>> after sending last frame to the kernel.
>> If pause, resume are allowed during draining, user experience can be
>> enhanced.
>> To prevent malfunction in HW drivers which don't support pause
>> during draining, pause during draining will only work if HW driver
>> enable this feature explicitly by calling
>> snd_compr_use_pause_in_draining().
>> 
>> Signed-off-by: Gyeongtaek Lee <gt82.lee at samsung.com>
>> Cc: stable at vger.kernel.org
>
>Could you restart the new thread?  It's been hanging too deeply and
>hard to read through.
>
>Since it's the revised patch, please give the revision number (v2 or
>such) and show what's different from the previous patches.
>
Ok. I'll send the patch again with [PATCH v2] prefix.
>About the changes:
>
>> +/**
>> + * snd_compr_use_pause_in_draining - Allow pause and resume in draining state
>> + * @substream: compress substream to set
>> + *
>> + * Allow pause and resume in draining state.
>> + * Only HW driver supports this transition can call this API.
>> + */
>> +static inline void snd_compr_use_pause_in_draining(
>> +					struct snd_compr_stream *substream)
>> +{
>> +	substream->device->use_pause_in_draining = true;
>> +}
>
>How to set the flag is an open question.  A natural way would be to
>set it somehow at creating the component object, but currently there
>seems no way to pass any flags.
Could you explain more about what is your concerning?
For me, calling snd_compr_use_pause_in_draining() in open() callback of 
snd_compr_ops was good enough.
I've tested it and it worked well on linux 5.4.
>
>
>thanks,
>
>Takashi
>