Remove reference counting from media::Pipeline.

Remove reference counting from media::Pipeline.

BUG=173313
R=acolwell@chromium.org

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=198760

[scherkus (not reviewing), 2013-04-26 04:37:31]

IIRC we settled on trying base::Unretained() but it still scares the heck out of
me! We should chat some more on Friday.

[acolwell GONE FROM CHROMIUM, 2013-04-26 18:30:21]

I agree that this change makes me nervous. We could add a bool that prevents
PostTasks() from happening after stop is called. We could also keep track of how
many pending posts we have and only call the stop callback once that count has
fallen to 0. Just some thoughts.

https://codereview.chromium.org/14371023/diff/2001/webkit/media/webmediaplaye...
File webkit/media/webmediaplayer_impl.cc (right):

https://codereview.chromium.org/14371023/diff/2001/webkit/media/webmediaplaye...
webkit/media/webmediaplayer_impl.cc:1215: 
pipeline_.reset() here? It would insure things go boom if WMPI gets unexpected
calls after Destroy().

[scherkus (not reviewing), 2013-04-30 03:26:38]

I tried using weak pointer but it made things a bit confusing since we can't use
it everywhere (e.g., callbacks that return a value such as GetMediaTime())

I did an audit of current PostTask() usage and after soaking with an asan build
for an hour I'm pretty confident that after Stop() completes there are no more
tasks referencing Pipeline since OnStopCompleted() is force-posted after all the
filter Stop()s complete.

Any other ideas?

https://codereview.chromium.org/14371023/diff/2001/webkit/media/webmediaplaye...
File webkit/media/webmediaplayer_impl.cc (right):

https://codereview.chromium.org/14371023/diff/2001/webkit/media/webmediaplaye...
webkit/media/webmediaplayer_impl.cc:1215: 
On 2013/04/26 18:30:21, acolwell wrote:
> pipeline_.reset() here? It would insure things go boom if WMPI gets unexpected
> calls after Destroy().

Done.

[acolwell GONE FROM CHROMIUM, 2013-05-03 17:13:22]

LGTM

[scherkus (not reviewing), 2013-05-03 17:23:39]

On 2013/05/03 17:13:22, acolwell wrote:
> LGTM

[scherkus (not reviewing), 2013-05-03 17:23:54]

On 2013/05/03 17:23:39, scherkus wrote:
> On 2013/05/03 17:13:22, acolwell wrote:
> > LGTM

crap hit send

anyway, thanks -- will target M29

[scherkus (not reviewing), 2013-05-07 17:43:37]

Committed patchset #3 manually as r198760 (presubmit successful).