Re: A reliable kernel panic (3.6.2) and system crash when visiting a particular website

[Takashi Iwai]

At Thu, 08 Nov 2012 08:31:35 +0100,
Daniel Mack wrote:
(snip)
> >> We can't simply stop both endpoints in the prepare callback.
> > 
> > The new function doesn't stop the stream by itself but it just syncs
> > if the stream is being stopped beforehand.  So, it's safe to call it
> > there.
> > 
> > Maybe the name was confusing.  It should have been like
> > snd_usb_endpoint_sync_pending_stop() or such.
> 
> Ah, right. I was errornously looking closer to Alan's patch but then
> replied to yours. Alright then - thanks for explaining :)

OK, thanks for checking.

FWIW, below is the patch I applied now to for-linus branch.
Renamed the function, added the comment and put NULL check to the
function to simplify.


Takashi

---
From: Takashi Iwai <tiwai@xxxxxxx>
Subject: [PATCH] ALSA: usb-audio: Fix crash at re-preparing the PCM stream

There are bug reports of a crash with USB-audio devices when PCM
prepare is performed immediately after the stream is stopped via
trigger callback.  It turned out that the problem is that we don't
wait until all URBs are killed.

This patch adds a new function to synchronize the pending stop
operation on an endpoint, and calls in the prepare callback for
avoiding the crash above.

Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=49181

Reported-and-tested-by: Artem S. Tashkinov <t.artem@xxxxxxxxx>
Cc: <stable@xxxxxxxxxxxxxxx> [v3.6]
Signed-off-by: Takashi Iwai <tiwai@xxxxxxx>
---
 sound/usb/endpoint.c | 13 +++++++++++++
 sound/usb/endpoint.h |  1 +
 sound/usb/pcm.c      |  3 +++
 3 files changed, 17 insertions(+)

diff --git a/sound/usb/endpoint.c b/sound/usb/endpoint.c
index 7f78c6d..34de6f2 100644
--- a/sound/usb/endpoint.c
+++ b/sound/usb/endpoint.c
@@ -35,6 +35,7 @@
 
 #define EP_FLAG_ACTIVATED	0
 #define EP_FLAG_RUNNING		1
+#define EP_FLAG_STOPPING	2
 
 /*
  * snd_usb_endpoint is a model that abstracts everything related to an
@@ -502,10 +503,20 @@ static int wait_clear_urbs(struct snd_usb_endpoint *ep)
 	if (alive)
 		snd_printk(KERN_ERR "timeout: still %d active urbs on EP #%x\n",
 					alive, ep->ep_num);
+	clear_bit(EP_FLAG_STOPPING, &ep->flags);
 
 	return 0;
 }
 
+/* sync the pending stop operation;
+ * this function itself doesn't trigger the stop operation
+ */
+void snd_usb_endpoint_sync_pending_stop(struct snd_usb_endpoint *ep)
+{
+	if (ep && test_bit(EP_FLAG_STOPPING, &ep->flags))
+		wait_clear_urbs(ep);
+}
+
 /*
  * unlink active urbs.
  */
@@ -918,6 +929,8 @@ void snd_usb_endpoint_stop(struct snd_usb_endpoint *ep,
 
 		if (wait)
 			wait_clear_urbs(ep);
+		else
+			set_bit(EP_FLAG_STOPPING, &ep->flags);
 	}
 }
 
diff --git a/sound/usb/endpoint.h b/sound/usb/endpoint.h
index 6376ccf..3d4c970 100644
--- a/sound/usb/endpoint.h
+++ b/sound/usb/endpoint.h
@@ -19,6 +19,7 @@ int snd_usb_endpoint_set_params(struct snd_usb_endpoint *ep,
 int  snd_usb_endpoint_start(struct snd_usb_endpoint *ep, int can_sleep);
 void snd_usb_endpoint_stop(struct snd_usb_endpoint *ep,
 			   int force, int can_sleep, int wait);
+void snd_usb_endpoint_sync_pending_stop(struct snd_usb_endpoint *ep);
 int  snd_usb_endpoint_activate(struct snd_usb_endpoint *ep);
 int  snd_usb_endpoint_deactivate(struct snd_usb_endpoint *ep);
 void snd_usb_endpoint_free(struct list_head *head);
diff --git a/sound/usb/pcm.c b/sound/usb/pcm.c
index 37428f7..5c12a3f 100644
--- a/sound/usb/pcm.c
+++ b/sound/usb/pcm.c
@@ -568,6 +568,9 @@ static int snd_usb_pcm_prepare(struct snd_pcm_substream *substream)
 		goto unlock;
 	}
 
+	snd_usb_endpoint_sync_pending_stop(subs->sync_endpoint);
+	snd_usb_endpoint_sync_pending_stop(subs->data_endpoint);
+
 	ret = set_format(subs, subs->cur_audiofmt);
 	if (ret < 0)
 		goto unlock;
-- 
1.8.0


--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/