Re: [PATCH] media: Media Controller fix to not let stream_count go negative

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On 02/02/2016 03:53 PM, Sakari Ailus wrote:
> Hi Shuah,
> 
> On Sat, Jan 30, 2016 at 01:10:52PM -0700, Shuah Khan wrote:
>> Change media_entity_pipeline_stop() to not decrement
>> stream_count of an inactive media pipeline. Doing so,
>> results in preventing starting the pipeline.
>>
>> Signed-off-by: Shuah Khan <shuahkh@xxxxxxxxxxxxxxx>
>> ---
>>  drivers/media/media-entity.c | 18 ++++++++++++------
>>  1 file changed, 12 insertions(+), 6 deletions(-)
>>
>> diff --git a/drivers/media/media-entity.c b/drivers/media/media-entity.c
>> index e89d85a..f2e4360 100644
>> --- a/drivers/media/media-entity.c
>> +++ b/drivers/media/media-entity.c
>> @@ -452,9 +452,12 @@ error:
>>  	media_entity_graph_walk_start(graph, entity_err);
>>  
>>  	while ((entity_err = media_entity_graph_walk_next(graph))) {
>> -		entity_err->stream_count--;
>> -		if (entity_err->stream_count == 0)
>> -			entity_err->pipe = NULL;
>> +		/* don't let the stream_count go negative */
>> +		if (entity->stream_count > 0) {
>> +			entity_err->stream_count--;
>> +			if (entity_err->stream_count == 0)
>> +				entity_err->pipe = NULL;
>> +		}
>>  
>>  		/*
>>  		 * We haven't increased stream_count further than this
>> @@ -486,9 +489,12 @@ void media_entity_pipeline_stop(struct media_entity *entity)
>>  	media_entity_graph_walk_start(graph, entity);
>>  
>>  	while ((entity = media_entity_graph_walk_next(graph))) {
>> -		entity->stream_count--;
>> -		if (entity->stream_count == 0)
>> -			entity->pipe = NULL;
>> +		/* don't let the stream_count go negative */
>> +		if (entity->stream_count > 0) {
>> +			entity->stream_count--;
>> +			if (entity->stream_count == 0)
>> +				entity->pipe = NULL;
>> +		}
>>  	}
>>  
>>  	if (!--pipe->streaming_count)
> 
> Have you seen issues with a certain driver, for instance?
> 
> In the original design the streaming count is really a count --- streaming
> starts when count becomes non-zero, and stops when it reaches zero again.
> 
> The calls to media_entity_pipeline_start() and media_entity_pipeline_stop()
> should thus always be balanced. I'm fine with the patch, but the framework
> should shout loud when the count would be decremented when it's zero.
> 
> That was some four or more years ago. I have to say I really haven't been
> able to see good reasons for making this a count --- rather what's needed is
> to mark the entity as busy so that its link configuration isn't touched. The
> request API is a completely different matter then.
> 
> Such change would require more discussion IMHO.
> 

Yes. I found problems with au0828 and ALSA
media controller use-case. It got into a state
where pipeline was essentially locked in a state.
It took some work to debug and find that the
stream_count was negative.

Granted the start and stop should be balanced, however,
the media_entity_pipeline_stop() still needs to protect
the stream_count from going negative.

thanks,
-- Shuah

-- 
Shuah Khan
Sr. Linux Kernel Developer
Open Source Innovation Group
Samsung Research America (Silicon Valley)
shuahkh@xxxxxxxxxxxxxxx | (970) 217-8978
--
To unsubscribe from this list: send the line "unsubscribe linux-media" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html



[Index of Archives]     [Linux Input]     [Video for Linux]     [Gstreamer Embedded]     [Mplayer Users]     [Linux USB Devel]     [Linux Audio Users]     [Linux Kernel]     [Linux SCSI]     [Yosemite Backpacking]
  Powered by Linux