[alsa-devel] [PATCH 2/5] ASoC: dwc: Don't allow negative use counts
Signed-off-by: Andrew Jackson Andrew.Jackson@arm.com --- sound/soc/dwc/designware_i2s.c | 3 ++- 1 files changed, 2 insertions(+), 1 deletions(-)
diff --git a/sound/soc/dwc/designware_i2s.c b/sound/soc/dwc/designware_i2s.c index 08f0229..f8946bd 100644 --- a/sound/soc/dwc/designware_i2s.c +++ b/sound/soc/dwc/designware_i2s.c @@ -280,7 +280,8 @@ static int dw_i2s_trigger(struct snd_pcm_substream *substream, case SNDRV_PCM_TRIGGER_STOP: case SNDRV_PCM_TRIGGER_SUSPEND: case SNDRV_PCM_TRIGGER_PAUSE_PUSH: - dev->active--; + if (dev->active > 0) + dev->active--; i2s_stop(dev, substream); break; default:
On Wed, Dec 03, 2014 at 04:38:55PM +0000, Andrew Jackson wrote:
case SNDRV_PCM_TRIGGER_STOP: case SNDRV_PCM_TRIGGER_SUSPEND: case SNDRV_PCM_TRIGGER_PAUSE_PUSH:
dev->active--;
if (dev->active > 0)dev->active--;
How is this triggering - this sounds like you're papering over some other bug somewhere?
On Wed, Dec 3, 2014 at 10:55 PM, Mark Brown broonie@kernel.org wrote:
On Wed, Dec 03, 2014 at 04:38:55PM +0000, Andrew Jackson wrote:
case SNDRV_PCM_TRIGGER_STOP: case SNDRV_PCM_TRIGGER_SUSPEND: case SNDRV_PCM_TRIGGER_PAUSE_PUSH:
dev->active--;
if (dev->active > 0)dev->active--;How is this triggering - this sounds like you're papering over some other bug somewhere?
On Thu, Dec 4, 2014 at 12:10 PM, rajeev kumar rajeevkumar.linux@gmail.com wrote:
On Wed, Dec 3, 2014 at 10:55 PM, Mark Brown broonie@kernel.org wrote:
On Wed, Dec 03, 2014 at 04:38:55PM +0000, Andrew Jackson wrote:
case SNDRV_PCM_TRIGGER_STOP: case SNDRV_PCM_TRIGGER_SUSPEND: case SNDRV_PCM_TRIGGER_PAUSE_PUSH:
dev->active--;
if (dev->active > 0)dev->active--;How is this triggering - this sounds like you're papering over some other bug somewhere?
This check can be removed as it is not going to triggered.
B'rgds ~Rajeev
On 12/04/14 06:43, rajeev kumar wrote:
On Thu, Dec 4, 2014 at 12:10 PM, rajeev kumar rajeevkumar.linux@gmail.com wrote:
On Wed, Dec 3, 2014 at 10:55 PM, Mark Brown broonie@kernel.org wrote:
On Wed, Dec 03, 2014 at 04:38:55PM +0000, Andrew Jackson wrote:
case SNDRV_PCM_TRIGGER_STOP: case SNDRV_PCM_TRIGGER_SUSPEND: case SNDRV_PCM_TRIGGER_PAUSE_PUSH:
dev->active--;
if (dev->active > 0)dev->active--;How is this triggering - this sounds like you're papering over some other bug somewhere?
This check can be removed as it is not going to triggered.
As I said in my email to Mark, I couldn't convince myself that STOP/SUSPEND would only be called once so it seemed like a loophole which /might/ result in the device not functioning correctly.
If this can never happen, I'll drop the patch.
Andrew
B'rgds ~Rajeev
On 12/03/14 17:25, Mark Brown wrote:
On Wed, Dec 03, 2014 at 04:38:55PM +0000, Andrew Jackson wrote:
case SNDRV_PCM_TRIGGER_STOP: case SNDRV_PCM_TRIGGER_SUSPEND: case SNDRV_PCM_TRIGGER_PAUSE_PUSH:
dev->active--;
if (dev->active > 0)dev->active--;How is this triggering - this sounds like you're papering over some other bug somewhere?
When I looked at the code paths I couldn't convince myself that STOP wouldn't be called more than once. Then actuve would be negative and the device might not be restartable. I didn't have a problem per se, it was just that it seemed to be something of a loophole.
Andrew
On Thu, Dec 04, 2014 at 09:00:35AM +0000, Andrew Jackson wrote:
On 12/03/14 17:25, Mark Brown wrote:
On Wed, Dec 03, 2014 at 04:38:55PM +0000, Andrew Jackson wrote:
case SNDRV_PCM_TRIGGER_STOP: case SNDRV_PCM_TRIGGER_SUSPEND: case SNDRV_PCM_TRIGGER_PAUSE_PUSH:
dev->active--;
if (dev->active > 0)dev->active--;
How is this triggering - this sounds like you're papering over some other bug somewhere?
When I looked at the code paths I couldn't convince myself that STOP wouldn't be called more than once. Then actuve would be negative and the device might not be restartable. I didn't have a problem per se, it was just that it seemed to be something of a loophole.
If you're just adding the check on the off chance that it might fire you need to add a warning message as well - what your change does is make the code look like it's supposed to have broken reference counting since it has a check to silently fix up and ignore problems.
participants (3)
-
Andrew Jackson -
Mark Brown -
rajeev kumar