capture a top java exception on certain Samsung devices

capture a top java exception on certain Samsung devices

This is one of the top crasher on some samsung devices
https://crash.corp.google.com/search?query=android.media.MediaCodec.flush
Capture some exceptions while we are trying to look into the problem

BUG=278464
Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=221858

[qinmin, 2013-08-23 20:35:59]

PTAL

[Yaron, 2013-08-23 20:54:15]

On 2013/08/23 20:35:59, qinmin wrote:
> PTAL

What side effects can happen if we think we've flushed but haven't? I see the
exception is from the first call so fFlushed won't get set, but have you tried
repro'ing and seeing the underlying cause. It'd be good to see what the
IllegalStateException is.

[qinmin, 2013-08-23 21:24:40]

The IllegalStateException is a very general exception thrown by media framework.
I cannot repro the issue on a galaxy S3 and 2 galaxy S4s (although the crash
report shows some are from galaxy S4). 
Will try to get a device that can reproduce this.
My pure guess is that if we caught the exception, the next queueInputBuffer()
call will fail. And we will abort the media playback 

On 2013/08/23 20:54:15, Yaron wrote:
> On 2013/08/23 20:35:59, qinmin wrote:
> > PTAL
> 
> What side effects can happen if we think we've flushed but haven't? I see the
> exception is from the first call so fFlushed won't get set, but have you tried
> repro'ing and seeing the underlying cause. It'd be good to see what the
> IllegalStateException is.

[Yaron, 2013-08-23 22:42:12]

On 2013/08/23 21:24:40, qinmin wrote:
> The IllegalStateException is a very general exception thrown by media
framework.
> I cannot repro the issue on a galaxy S3 and 2 galaxy S4s (although the crash
> report shows some are from galaxy S4). 
> Will try to get a device that can reproduce this.
> My pure guess is that if we caught the exception, the next queueInputBuffer()
> call will fail. And we will abort the media playback 
> 
> On 2013/08/23 20:54:15, Yaron wrote:
> > On 2013/08/23 20:35:59, qinmin wrote:
> > > PTAL
> > 
> > What side effects can happen if we think we've flushed but haven't? I see
the
> > exception is from the first call so fFlushed won't get set, but have you
tried
> > repro'ing and seeing the underlying cause. It'd be good to see what the
> > IllegalStateException is.

That's probably better than crashing but worth seeing if we can repro first.

[qinmin, 2013-09-06 21:25:22]

PTAL again, the crash is now almost 30% on beta, and we still haven't got a
repro.
Sending an error to the webkit seems a safer way to handle this rather than
crash the browser process

[acolwell GONE FROM CHROMIUM, 2013-09-06 21:37:44]

lgtm

https://codereview.chromium.org/23321004/diff/7001/media/base/android/java/sr...
File media/base/android/java/src/org/chromium/media/MediaCodecBridge.java
(right):

https://codereview.chromium.org/23321004/diff/7001/media/base/android/java/sr...
media/base/android/java/src/org/chromium/media/MediaCodecBridge.java:202:
mMediaCodec.flush();
Why was the call order changed?

https://codereview.chromium.org/23321004/diff/7001/media/base/android/media_c...
File media/base/android/media_codec_bridge.h (right):

https://codereview.chromium.org/23321004/diff/7001/media/base/android/media_c...
media/base/android/media_codec_bridge.h:57: // Returns MEDIA_CODEC_ERROR if
unexpected error happens, or Media_CODEC_OK
nit: s/if/if an/

[xhwang, 2013-09-06 23:18:47]

lgtm % acolwell's question

[qinmin, 2013-09-06 23:22:50]

https://codereview.chromium.org/23321004/diff/7001/media/base/android/java/sr...
File media/base/android/java/src/org/chromium/media/MediaCodecBridge.java
(right):

https://codereview.chromium.org/23321004/diff/7001/media/base/android/java/sr...
media/base/android/java/src/org/chromium/media/MediaCodecBridge.java:202:
mMediaCodec.flush();
If MediaCodec throws an exception, mAudioTrack.flush() will not be called. As a
result, we might still play some audio after we report the error to webkit. 
On the otherhand, AudioTrack is more robust compared to MediaCodec. 

On 2013/09/06 21:37:44, acolwell wrote:
> Why was the call order changed?

https://codereview.chromium.org/23321004/diff/7001/media/base/android/media_c...
File media/base/android/media_codec_bridge.h (right):

https://codereview.chromium.org/23321004/diff/7001/media/base/android/media_c...
media/base/android/media_codec_bridge.h:57: // Returns MEDIA_CODEC_ERROR if
unexpected error happens, or Media_CODEC_OK
On 2013/09/06 21:37:44, acolwell wrote:
> nit: s/if/if an/

Done.

[commit-bot: I haz the power, 2013-09-06 23:25:32]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/qinmin@chromium.org/23321004/15001

[commit-bot: I haz the power, 2013-09-07 02:04:34]

Change committed as 221858