Use a single CGLContextObj to draw and update the IOSurface

Use a single CGLContextObj to draw and update the IOSurface

There is no reason that this can't be done, based on documentation,
but it may be that this will cause crashes during texture destruction
or context tear-down (see crbug.com/411782).

I have 3 theories as to why these crashes are occurring
1. Using an IOSurface in a CGLContextObj that is also used by a
   CAOpenGLLayer causes bugs
2. The crashes observed were because context destruction just crashes
   at a certain frequency, and previous patches made context destruction
   more frequent.
3. IOSurface-backed textures should be immutable -- no GL texture
   should be bound to one IOSurface then the other.

If theory 1 is correct, this will need to be reverted because the crashes
will re-occur. If theories 2 or 3 are correct, then this patch is safe.

BUG=420253
Committed: https://crrev.com/0ea6e165293b7d24deec0d29fb9d6333d3a71b6d
Cr-Commit-Position: refs/heads/master@{#298916}

[ccameron, 2014-10-03 18:54:03]

ccameron@chromium.org changed reviewers:
+ kbr@chromium.org

[ccameron, 2014-10-03 18:54:04]

This is the last cleanup patch for the IOSurface path (next step is to start the
switch over to remote layers).

This may just need to be reverted, if I see crashes resulting from it (I'd put
the odds at about 50%) ... remarks about white whales may be appropriate.

[Ken Russell (switch to Gerrit), 2014-10-03 21:53:37]

Could you please file a bug for this in case it needs to be reverted and
re-landed multiple times?

[ccameron, 2014-10-03 21:59:40]

On 2014/10/03 21:53:37, Ken Russell wrote:
> Could you please file a bug for this in case it needs to be reverted and
> re-landed multiple times?

Filed crbug.com/420253

[Ken Russell (switch to Gerrit), 2014-10-04 00:47:54]

Per my comments on https://codereview.chromium.org/625753002/ I don't understand
the current state of this code. Could we sit down and talk about these CLs
face-to-face next week?

[ccameron, 2014-10-04 00:50:13]

On 2014/10/04 00:47:54, Ken Russell wrote:
> Per my comments on https://codereview.chromium.org/625753002/ I don't
understand
> the current state of this code. Could we sit down and talk about these CLs
> face-to-face next week?

Sure -- that sounds like a good idea.

[ccameron, 2014-10-08 18:50:51]

This is the rebase of the second half of the patch. I listed the 3 theories of
why the GL driver is crashing in the description -- this will help narrow down
who is to blame.

I'll be waiting until after we have a Canary with the previous patch to land
this one (to be sure that this and not the other is the cause of crashes that
appear).

[Ken Russell (switch to Gerrit), 2014-10-08 21:43:16]

Thanks again for the offline conversation about this and the other CL. LGTM with
nits.

https://codereview.chromium.org/627003002/diff/40001/content/browser/composit...
File content/browser/compositor/io_surface_context_mac.mm (right):

https://codereview.chromium.org/627003002/diff/40001/content/browser/composit...
content/browser/compositor/io_surface_context_mac.mm:53: // need to be recreated
when transitioning contexts.
Comment should be updated at this point.

https://codereview.chromium.org/627003002/diff/40001/content/browser/composit...
content/browser/compositor/io_surface_context_mac.mm:55: pixel_format, NULL /*
share_context */, cgl_context.InitializeInto());
Delete commented-out "share_context".

[ccameron, 2014-10-09 07:32:51]

Thanks!

https://codereview.chromium.org/627003002/diff/40001/content/browser/composit...
File content/browser/compositor/io_surface_context_mac.mm (right):

https://codereview.chromium.org/627003002/diff/40001/content/browser/composit...
content/browser/compositor/io_surface_context_mac.mm:53: // need to be recreated
when transitioning contexts.
On 2014/10/08 21:43:16, Ken Russell wrote:
> Comment should be updated at this point.

Done.

https://codereview.chromium.org/627003002/diff/40001/content/browser/composit...
content/browser/compositor/io_surface_context_mac.mm:55: pixel_format, NULL /*
share_context */, cgl_context.InitializeInto());
On 2014/10/08 21:43:16, Ken Russell wrote:
> Delete commented-out "share_context".

Done.

[ccameron, 2014-10-09 07:32:56]

The CQ bit was checked by ccameron@chromium.org

[commit-bot: I haz the power, 2014-10-09 07:33:59]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/627003002/60001

[commit-bot: I haz the power, 2014-10-09 08:23:20]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2014-10-09 08:23:21]

Try jobs failed on following builders:
  linux_gpu_triggered_tests on tryserver.chromium.gpu
(http://build.chromium.org/p/tryserver.chromium.gpu/builders/linux_gpu_trigger...)

[ccameron, 2014-10-09 16:51:32]

The CQ bit was checked by ccameron@chromium.org

[commit-bot: I haz the power, 2014-10-09 16:52:56]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/627003002/60001

[commit-bot: I haz the power, 2014-10-09 17:07:23]

Committed patchset #4 (id:60001)

[commit-bot: I haz the power, 2014-10-09 17:09:10]

Patchset 4 (id:??) landed as
https://crrev.com/0ea6e165293b7d24deec0d29fb9d6333d3a71b6d
Cr-Commit-Position: refs/heads/master@{#298916}

[ccameron, 2014-10-09 20:35:45]

A revert of this CL (patchset #4 id:60001) has been created in
https://codereview.chromium.org/647553002/ by ccameron@chromium.org.

The reason for reverting is: I hear reports of instability on Mac try servers,
with crashes in the GL driver..