Pepper: Add a nullptr check.

Pepper: Associate TextureMailboxes with the Graphics3D they were obtained from.

This allows the TextureMailbox to be returned to the appropriate Graphics3D,
which implicitly deals with trying to return a TextureMailbox to a null
Graphics3D.

BUG=602484
Committed: https://crrev.com/5fcc55335e3b5ba36a0de1a2e7a0976c859f6fff
Cr-Commit-Position: refs/heads/master@{#391934}

[erikchen, 2016-05-05 20:15:29]

erikchen@chromium.org changed reviewers:
+ piman@chromium.org

[erikchen, 2016-05-05 20:15:29]

piman: Please review.

https://codereview.chromium.org/1951263003/diff/1/content/renderer/pepper/pep...
File content/renderer/pepper/pepper_plugin_instance_impl.cc (right):

https://codereview.chromium.org/1951263003/diff/1/content/renderer/pepper/pep...
content/renderer/pepper/pepper_plugin_instance_impl.cc:781:
bound_graphics_3d_->ReturnFrontBuffer(
I don't think this one needs a nullcheck, as there are no crashes with it in the
signature, and this isn't coming from a callback from the compositor.

[piman, 2016-05-05 20:25:41]

https://codereview.chromium.org/1951263003/diff/1/content/renderer/pepper/pep...
File content/renderer/pepper/pepper_plugin_instance_impl.cc (right):

https://codereview.chromium.org/1951263003/diff/1/content/renderer/pepper/pep...
content/renderer/pepper/pepper_plugin_instance_impl.cc:826: if (removed &&
!is_committed_texture && bound_graphics_3d_) {
Actually, this makes me think a little more.
If the plugin binds a Graphics3D, swaps a couple of frames, then binds another
Graphics3D, we will be returning the mailbox to the wrong context (i.e. we won't
find the texture in GLES2DecoderImpl::ReturnFrontBuffer).

Should we bind the release callback to the Graphics3D that gave us the texture?

[erikchen, 2016-05-05 21:08:20]

Description was changed from

==========
Pepper: Add a nullptr check.

It is possible for the bound_graphics_3d_ to have been destroyed before the
texture is finished being consumed.

BUG=602484
==========

to

==========
Pepper: Associate TextureMailboxes with the Graphics3D they were obtained from.

This allows the TextureMailbox to be returned to the appropriate Graphics3D,
which implicitly deals with trying to return a TextureMailbox to a null
Graphics3D.

BUG=602484
==========

[erikchen, 2016-05-05 21:13:08]

piman: PTAL

https://codereview.chromium.org/1951263003/diff/1/content/renderer/pepper/pep...
File content/renderer/pepper/pepper_plugin_instance_impl.cc (right):

https://codereview.chromium.org/1951263003/diff/1/content/renderer/pepper/pep...
content/renderer/pepper/pepper_plugin_instance_impl.cc:826: if (removed &&
!is_committed_texture && bound_graphics_3d_) {
On 2016/05/05 20:25:41, piman wrote:
> Actually, this makes me think a little more.
> If the plugin binds a Graphics3D, swaps a couple of frames, then binds another
> Graphics3D, we will be returning the mailbox to the wrong context (i.e. we
won't
> find the texture in GLES2DecoderImpl::ReturnFrontBuffer).
> 
> Should we bind the release callback to the Graphics3D that gave us the
texture?

Yes, we should.

[piman, 2016-05-05 21:14:37]

lgtm

[erikchen, 2016-05-05 21:15:44]

The CQ bit was checked by erikchen@chromium.org

[erikchen, 2016-05-05 21:15:45]

The patchset sent to the CQ was uploaded after l-g-t-m from piman@chromium.org
Link to the patchset: https://codereview.chromium.org/1951263003/#ps40001
(title: " ")

[commit-bot: I haz the power, 2016-05-05 21:16:17]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1951263003/40001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1951263003/40001

[commit-bot: I haz the power, 2016-05-05 22:49:11]

Committed patchset #3 (id:40001)

[commit-bot: I haz the power, 2016-05-05 22:50:45]

Description was changed from

==========
Pepper: Associate TextureMailboxes with the Graphics3D they were obtained from.

This allows the TextureMailbox to be returned to the appropriate Graphics3D,
which implicitly deals with trying to return a TextureMailbox to a null
Graphics3D.

BUG=602484
==========

to

==========
Pepper: Associate TextureMailboxes with the Graphics3D they were obtained from.

This allows the TextureMailbox to be returned to the appropriate Graphics3D,
which implicitly deals with trying to return a TextureMailbox to a null
Graphics3D.

BUG=602484

Committed: https://crrev.com/5fcc55335e3b5ba36a0de1a2e7a0976c859f6fff
Cr-Commit-Position: refs/heads/master@{#391934}
==========

[commit-bot: I haz the power, 2016-05-05 22:50:46]

Patchset 3 (id:??) landed as
https://crrev.com/5fcc55335e3b5ba36a0de1a2e7a0976c859f6fff
Cr-Commit-Position: refs/heads/master@{#391934}

[ilja, 2016-05-10 02:16:43]

A revert of this CL (patchset #3 id:40001) has been created in
https://codereview.chromium.org/1963633003/ by ihf@chromium.org.

The reason for reverting is: Breaks WebGL on CrOS, crbug.com/610411..