Re: [PATCH 2/2] ALSA: usb-audio: Generic application of implicit fb to Roland/BOSS devices

From: Takashi Iwai <tiwai@suse.de>
To: Lucas <jaffa225man@gmail.com>
Cc: alsa-devel@alsa-project.org,
	"Keith A . Milner" <maillist@superlative.org>,
	Mike Oliphant <oliphant@nostatic.org>
Subject: Re: [PATCH 2/2] ALSA: usb-audio: Generic application of implicit fb to Roland/BOSS devices
Date: Sun, 25 Apr 2021 11:16:08 +0200	[thread overview]
Message-ID: <s5hlf964td3.wl-tiwai@suse.de> (raw)
In-Reply-To: <CAOsVg8rPfh-scSmmJSVASPTw7o3vw_zZ05YzVZ_JYFJHbcOcyg@mail.gmail.com>

On Fri, 23 Apr 2021 20:04:38 +0200,
Lucas wrote:
> 
> On Fri, Apr 23, 2021 at 2:47 AM Takashi Iwai <tiwai@suse.de> wrote:
> 
>     Is this really with snd-usb-audio driver?  Please check the entry in
>     /proc/asound/modules (or traverse via sysfs).
> 
> No, I see it's not.  It's using the correct snd_ua101 module

Then it's expected, and it's been so.  I haven't touched snd-ua101
driver code until now.

>     This rather sounds like a generic problem of the implicit feedback.
>     Please give the kernel log with snd_usb_audio.dyndbg=+p boot option
>     with some notion which timestamp corresponding to which action
>     (e.g. starting playback, stopping recording, etc).
> 
> Okay, that's good news!  Thanks, and here's that:

Thanks.

Could you try the patch below?


Takashi

--- a/sound/usb/endpoint.c
+++ b/sound/usb/endpoint.c
@@ -1443,11 +1443,11 @@ void snd_usb_endpoint_stop(struct snd_usb_endpoint *ep)
 	if (snd_BUG_ON(!atomic_read(&ep->running)))
 		return;
 
-	if (ep->sync_source)
-		WRITE_ONCE(ep->sync_source->sync_sink, NULL);
-
-	if (!atomic_dec_return(&ep->running))
+	if (!atomic_dec_return(&ep->running)) {
+		if (ep->sync_source)
+			WRITE_ONCE(ep->sync_source->sync_sink, NULL);
 		stop_urbs(ep, false);
+	}
 }
 
 /**
