Re: [PATCH v2 0/2] Add snd_card_disconnect_sync() helper

[Takashi Iwai <tiwai@suse.de>]

On Fri, 13 Oct 2017 11:11:42 +0200,
Kuninori Morimoto wrote:
> 
> 
> Hi Takashi
> 
> > > I tested this patch-set.
> > > I noticed that it doesn't work if I used DPCM
> > > (Kernel has Oops).
> > > I will investigate it next week
> > 
> > Could you show the Oops message if you have?
> > My concern is whether it happens by stopping at disconnect, or it's
> > just another missing piece.
> 
> OK, but kernel log doesn't help you.
> see below
> 
> My environment now is I'm using DPCM.
> 	FE : rsnd
> 	FE : rsnd
> 	BE : ak4613
> 
> 1st issue is that kernel need below patch.
> I guess BE is using dummy driver, and it doesn't have ops(?).
> If this is needed, I can post it.

No, this can't be right.  Every PCM implementation mandates the
presence of a trigger callback.  It's a must.  If a fix needed, it has
to be fixed in the driver side.

> -----------------------
> diff --git a/sound/core/pcm_native.c b/sound/core/pcm_native.c
> index 2fec2fe..972408b 100644
> --- a/sound/core/pcm_native.c
> +++ b/sound/core/pcm_native.c
> @@ -1241,7 +1241,8 @@ static int snd_pcm_do_stop(struct snd_pcm_substream *substream, int state)
>  {
>  	if (substream->runtime->trigger_master == substream &&
>  	    snd_pcm_running(substream))
> -		substream->ops->trigger(substream, SNDRV_PCM_TRIGGER_STOP);
> +		if (substream->ops && substream->ops->trigger)
> +			substream->ops->trigger(substream, SNDRV_PCM_TRIGGER_STOP);
>  	return 0; /* unconditonally stop all substreams */
>  }
> -----------------------
> After this patch, my driver side clock counter mismatch Oops happen.
> It seems this is because FE side wasn't called snd_pcm_stop();
> (= BE side only called snd_pcm_stop()).

So the check might work around an Oops, but it's no "fix", per se.

> I could confirm this by printing who's stop was called by local quick hack.
> 
> Maybe timing reason, if kernel has Oops for some reasons,
> then, both BE/FE snd_pcm_stop() are called.
> If no Oops, BE snd_pcm_stop() only called.

Any pending delayed work (like rtd->delayed_work)?
This is flushed at soc_cleanup_card_resources(), but it's called at
card removal, thus it's too late for the hot-removal of the
component.


thanks,

Takashi