exo: Handle lost context situations properly.

exo: Handle lost context situations properly.

This implements proper handling of lost context by exosphere. It
also improves support for software compositing by not crashing.
True software compositing support will be added by follow up
patches.

BUG=549781
TEST=exo_unittests --gtest_filter=BufferTest.IsLost

Committed: https://crrev.com/ced21f867a76bc4858a9b6530820c7962286c270
Cr-Commit-Position: refs/heads/master@{#361228}

[reveman, 2015-11-23 03:43:40]

The CQ bit was checked by reveman@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2015-11-23 03:43:49]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1467943002/20001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1467943002/20001

[reveman, 2015-11-23 03:43:54]

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

[commit-bot: I haz the power, 2015-11-23 03:56:08]

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

[commit-bot: I haz the power, 2015-11-23 03:56:09]

Dry run: This issue passed the CQ dry run.

[piman, 2015-11-23 21:04:42]

lgtm

https://codereview.chromium.org/1467943002/diff/20001/components/exo/buffer.cc
File components/exo/buffer.cc (right):

https://codereview.chromium.org/1467943002/diff/20001/components/exo/buffer.c...
components/exo/buffer.cc:186: texture.reset();
I suggest doing this before getting the context_provider, because of potential
races.
texture->IsLost() may be the first thing that learns about the lost context, and
so SharedMainThreadContextProvider() may have returned an "old" channel instead
of recreating one, and when you try to recreate below, things fail.
OTOH, if you check texture first, if it's lost due to a lost channel, you are
guaranteed that SharedMainThreadContextProvider() will recreate a channel.

[reveman, 2015-11-23 23:25:28]

The CQ bit was checked by reveman@chromium.org

[reveman, 2015-11-23 23:25:29]

https://codereview.chromium.org/1467943002/diff/20001/components/exo/buffer.cc
File components/exo/buffer.cc (right):

https://codereview.chromium.org/1467943002/diff/20001/components/exo/buffer.c...
components/exo/buffer.cc:186: texture.reset();
On 2015/11/23 at 21:04:42, piman (slow to review) wrote:
> I suggest doing this before getting the context_provider, because of potential
races.
> texture->IsLost() may be the first thing that learns about the lost context,
and so SharedMainThreadContextProvider() may have returned an "old" channel
instead of recreating one, and when you try to recreate below, things fail.
> OTOH, if you check texture first, if it's lost due to a lost channel, you are
guaranteed that SharedMainThreadContextProvider() will recreate a channel.

Makes sense. Done.

[reveman, 2015-11-23 23:25:30]

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/1467943002/#ps40001
(title: "move SharedMainThreadContextProvider")

[commit-bot: I haz the power, 2015-11-23 23:27:46]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1467943002/40001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1467943002/40001

[commit-bot: I haz the power, 2015-11-24 00:42:59]

Committed patchset #3 (id:40001)

[commit-bot: I haz the power, 2015-11-24 00:43:40]

Patchset 3 (id:??) landed as
https://crrev.com/ced21f867a76bc4858a9b6530820c7962286c270
Cr-Commit-Position: refs/heads/master@{#361228}