Fix hangs & crashes in teardown-during-Seek.

Fix hangs & crashes in teardown-during-Seek.

Hangs:
- AudioRendererImpl: Don't hold lock_ in the call to Join() b/c OnStop wants to
  acquire the lock before exiting.
- FFmpegDemuxerStream::Stop: signal EOS to waiting read callbacks (instead of
  silently dropping them on the floor) so decoders can exit instead of stalling
  (and causing renderers to stall).

Crashes:
- CompositeFilter::Stop: clear out the Seek callback when Stopping.
- FFmpegVideoDecoder::ProduceVideoFrame: early-return if already stopped (racing
  with webkit repaints makes this necessary).

Also fixed a bunch of mis-indentation where I saw it.

BUG=98975, 98948, 98955, 96861
TEST=trybots, manual


Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=104540

[Ami GONE FROM CHROMIUM, 2011-10-06 23:25:44]

The manual testing here was using scherkus' benchmark.html page (linked to from
the related BUGs).
I tried & failed to create a layouttest for this, and suspect writing a robust
regression test for this bug will take a while, so wanted to get this out first.

[acolwell GONE FROM CHROMIUM, 2011-10-07 15:50:46]

LGTM

http://codereview.chromium.org/8184003/diff/1/content/renderer/media/audio_re...
File content/renderer/media/audio_renderer_impl.cc (right):

http://codereview.chromium.org/8184003/diff/1/content/renderer/media/audio_re...
content/renderer/media/audio_renderer_impl.cc:121: CHECK(socket->Close());
Does this need to happen outside of the lock?

http://codereview.chromium.org/8184003/diff/1/media/filters/ffmpeg_demuxer.cc
File media/filters/ffmpeg_demuxer.cc (right):

http://codereview.chromium.org/8184003/diff/1/media/filters/ffmpeg_demuxer.cc...
media/filters/ffmpeg_demuxer.cc:57: class EndOfStreamBuffer : public Buffer {
It feels like this should go in media/base somewhere. Perhaps it could even be
hidden behind a Buffer.newEndOfStreamBuffer() static method or something like
that.

[Ami GONE FROM CHROMIUM, 2011-10-07 16:44:47]

http://codereview.chromium.org/8184003/diff/1/content/renderer/media/audio_re...
File content/renderer/media/audio_renderer_impl.cc (right):

http://codereview.chromium.org/8184003/diff/1/content/renderer/media/audio_re...
content/renderer/media/audio_renderer_impl.cc:121: CHECK(socket->Close());
On 2011/10/07 15:50:47, acolwell wrote:
> Does this need to happen outside of the lock?

Done.

http://codereview.chromium.org/8184003/diff/1/media/filters/ffmpeg_demuxer.cc
File media/filters/ffmpeg_demuxer.cc (right):

http://codereview.chromium.org/8184003/diff/1/media/filters/ffmpeg_demuxer.cc...
media/filters/ffmpeg_demuxer.cc:57: class EndOfStreamBuffer : public Buffer {
On 2011/10/07 15:50:47, acolwell wrote:
> It feels like this should go in media/base somewhere. Perhaps it could even be
> hidden behind a Buffer.newEndOfStreamBuffer() static method or something like
> that.

As soon as there's another customer, I'm happy for that to be the case.  Until
then I don't think it's a good idea.

[acolwell GONE FROM CHROMIUM, 2011-10-07 16:56:58]

http://codereview.chromium.org/8184003/diff/1/media/filters/ffmpeg_demuxer.cc
File media/filters/ffmpeg_demuxer.cc (right):

http://codereview.chromium.org/8184003/diff/1/media/filters/ffmpeg_demuxer.cc...
media/filters/ffmpeg_demuxer.cc:57: class EndOfStreamBuffer : public Buffer {
On 2011/10/07 16:44:47, Ami Fischman wrote:
> On 2011/10/07 15:50:47, acolwell wrote:
> > It feels like this should go in media/base somewhere. Perhaps it could even
be
> > hidden behind a Buffer.newEndOfStreamBuffer() static method or something
like
> > that.
> 
> As soon as there's another customer, I'm happy for that to be the case.  Until
> then I don't think it's a good idea.
ChunkDemuxer could be that customer. It currently uses a static function called
CreateEOSBuffer() to do this same thing in a slightly different way. In fact,
you might just mimic that solution instead of creating a class here. Up to you.

[Ami GONE FROM CHROMIUM, 2011-10-07 17:11:10]

http://codereview.chromium.org/8184003/diff/1/media/filters/ffmpeg_demuxer.cc
File media/filters/ffmpeg_demuxer.cc (right):

http://codereview.chromium.org/8184003/diff/1/media/filters/ffmpeg_demuxer.cc...
media/filters/ffmpeg_demuxer.cc:57: class EndOfStreamBuffer : public Buffer {
On 2011/10/07 16:56:59, acolwell wrote:
> ChunkDemuxer could be that customer. It currently uses a static function
called
> CreateEOSBuffer() to do this same thing in a slightly different way. In fact,
> you might just mimic that solution instead of creating a class here. Up to
you.

Doing the latter (mimicking).

[commit-bot: I haz the power, 2011-10-07 17:14:20]

CQ is trying the patch. Follow status at
https://chromium-status.appspot.com/cq/fischman@chromium.org/8184003/3003

[commit-bot: I haz the power, 2011-10-07 17:14:22]

Can't apply patch for file content/renderer/media/audio_renderer_impl.cc.
While running patch -p1 --forward --force;
patching file content/renderer/media/audio_renderer_impl.cc
Hunk #1 FAILED at 98.
Hunk #2 succeeded at 456 (offset 1 line).
1 out of 2 hunks FAILED -- saving rejects to file
content/renderer/media/audio_renderer_impl.cc.rej

[commit-bot: I haz the power, 2011-10-07 17:33:10]

CQ is trying the patch. Follow status at
https://chromium-status.appspot.com/cq/fischman@chromium.org/8184003/7002

[commit-bot: I haz the power, 2011-10-07 19:12:23]

Change committed as 104540