Re: [PATCH v13 32/53] ALSA: usb-audio: Check for support for requested audio format

[Takashi Iwai]

On Tue, 06 Feb 2024 15:50:21 +0100,
Greg KH wrote:
> 
> On Tue, Feb 06, 2024 at 02:12:44PM +0100, Takashi Iwai wrote:
> > On Sat, 03 Feb 2024 03:36:24 +0100,
> > Wesley Cheng wrote:
> > > 
> > > Allow for checks on a specific USB audio device to see if a requested PCM
> > > format is supported.  This is needed for support when playback is
> > > initiated by the ASoC USB backend path.
> > > 
> > > Signed-off-by: Wesley Cheng <quic_wcheng@xxxxxxxxxxx>
> > 
> > Just cosmetic:
> > 
> > > +struct snd_usb_stream *snd_usb_find_suppported_substream(int card_idx,
> > > +			struct snd_pcm_hw_params *params, int direction)
> > > +{
> > > +	struct snd_usb_audio *chip;
> > > +	struct snd_usb_substream *subs;
> > > +	struct snd_usb_stream *as;
> > > +	const struct audioformat *fmt;
> > > +
> > > +	/*
> > > +	 * Register mutex is held when populating and clearing usb_chip
> > > +	 * array.
> > > +	 */
> > > +	mutex_lock(&register_mutex);
> > > +	chip = usb_chip[card_idx];
> > > +	if (!chip) {
> > > +		mutex_unlock(&register_mutex);
> > > +		return NULL;
> > > +	}
> > > +
> > > +	if (enable[card_idx]) {
> > > +		list_for_each_entry(as, &chip->pcm_list, list) {
> > > +			subs = &as->substream[direction];
> > > +			fmt = snd_usb_find_substream_format(subs, params);
> > > +			if (fmt) {
> > > +				mutex_unlock(&register_mutex);
> > > +				return as;
> > > +			}
> > > +		}
> > > +	}
> > > +	mutex_unlock(&register_mutex);
> > 
> > I prefer having the single lock/unlock call pair, e.g.
> > 
> > 	struct snd_usb_stream *as, *ret;
> > 
> > 	ret = NULL;
> > 	mutex_lock(&register_mutex);
> > 	chip = usb_chip[card_idx];
> > 	if (chip && enable[card_idx]) {
> > 		list_for_each_entry(as, &chip->pcm_list, list) {
> > 			subs = &as->substream[direction];
> > 			if (snd_usb_find_substream_format(subs, params)) {
> > 				ret = as;
> > 				break;
> > 			}
> > 		}
> > 	}
> > 	mutex_unlock(&register_mutex);
> > 	return ret;
> > }
> > 
> > In this case, we shouldn't reuse "as" for the return value since it
> > can be non-NULL after the loop end.
> 
> Why not just use guard(mutex) for this, making it all not an issue?

Heh, it's too new ;)
That should work gracefully, yes.


Takashi