cc: Tell scheduler the output surface is lost before releasing it.

cc: Tell scheduler the output surface is lost before releasing it.

This matches the behaviour of ReleaseOutputSurface, and will prevent
the scheduler from trying to activate when the surface is released. At
the time the surface is lost in the scheduler, it will drop the active
tree and activate the pending tree immediately. This avoids the state
where we have a pending tree when the output surface is released, after
doing ReleaseTreeResources.

R=enne
BUG=624075
CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel

Committed: https://crrev.com/e702a318e280000fc7f7d81c08c8ed41caa14735
Cr-Commit-Position: refs/heads/master@{#403990}

[danakj, 2016-07-06 20:38:52]

Description was changed from

==========
cc: Tell scheduler the output surface is lost before releasing it.

This matches the behaviour of ReleaseOutputSurface, and will prevent
the scheduler from trying to activate when the surface is released. At
the time the surface is lost in the scheduler, it will drop the active
tree and activate the pending tree immediately. This avoids the state
where we have a pending tree when the output surface is released, after
doing ReleaseTreeResources.

R=enne
BUG=624075
==========

to

==========
cc: Tell scheduler the output surface is lost before releasing it.

This matches the behaviour of ReleaseOutputSurface, and will prevent
the scheduler from trying to activate when the surface is released. At
the time the surface is lost in the scheduler, it will drop the active
tree and activate the pending tree immediately. This avoids the state
where we have a pending tree when the output surface is released, after
doing ReleaseTreeResources.

R=enne
BUG=624075
CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel
==========

[danakj, 2016-07-06 20:39:11]

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

[commit-bot: I haz the power, 2016-07-06 20:40:02]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[danakj, 2016-07-06 20:42:27]

I agree this isn't super ideal. The original goal was to avoid OutputSurface
calling back to its client (LTHI) during shutdown. The amount of work we're
doing here is non-trivial tho, when we're going to delete it all immediately
anyway. I'm not sure what a LTHI::ReleaseOutputSurfaceForShutdown might look
like. Deleting the pending tree will make it disagree with the scheduler state
machine. It could do *less* maybe, and only detach the output surface or
something.

[enne (OOO), 2016-07-06 20:44:54]

I think this is fine as a crash fix.  I can't think of a cleaner way to shut
things down without adding a lot of complication or extra scheduler states, even
though I don't like having to do this extra activation work when shutting down.

lgtm

[danakj, 2016-07-06 20:48:53]

The CQ bit was unchecked by danakj@chromium.org

[danakj, 2016-07-06 21:12:21]

The CQ bit was checked by danakj@chromium.org

[commit-bot: I haz the power, 2016-07-06 21:13:06]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2016-07-06 23:57:09]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2016-07-06 23:59:06]

Description was changed from

==========
cc: Tell scheduler the output surface is lost before releasing it.

This matches the behaviour of ReleaseOutputSurface, and will prevent
the scheduler from trying to activate when the surface is released. At
the time the surface is lost in the scheduler, it will drop the active
tree and activate the pending tree immediately. This avoids the state
where we have a pending tree when the output surface is released, after
doing ReleaseTreeResources.

R=enne
BUG=624075
CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel
==========

to

==========
cc: Tell scheduler the output surface is lost before releasing it.

This matches the behaviour of ReleaseOutputSurface, and will prevent
the scheduler from trying to activate when the surface is released. At
the time the surface is lost in the scheduler, it will drop the active
tree and activate the pending tree immediately. This avoids the state
where we have a pending tree when the output surface is released, after
doing ReleaseTreeResources.

R=enne
BUG=624075
CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel

Committed: https://crrev.com/e702a318e280000fc7f7d81c08c8ed41caa14735
Cr-Commit-Position: refs/heads/master@{#403990}
==========

[commit-bot: I haz the power, 2016-07-06 23:59:07]

Patchset 1 (id:??) landed as
https://crrev.com/e702a318e280000fc7f7d81c08c8ed41caa14735
Cr-Commit-Position: refs/heads/master@{#403990}