gpu: Crash GPU process when context was lost through robustness

gpu: Crash GPU process when context was lost through robustness

This works around both recovery problems in drivers as well as
in our recovery flow (480545).

Note that this is already the default behavior on Windows.

BUG=458269, 475676, 503414, 480545
Committed: https://crrev.com/751a98d265bf854b6b6d216fb7d20ef6ec1313c8
Cr-Commit-Position: refs/heads/master@{#342928}

[no sievers, 2015-08-11 00:20:32]

sievers@chromium.org changed reviewers:
+ jbauman@chromium.org, kbr@chromium.org, piman@chromium.org

[no sievers, 2015-08-11 00:22:12]

ptal. Seems to do wonders on Qualcomm also as manually tested.

We can later look into scaling back the workaround, but for now I'd like to
leave it for drivers that don't support robustness (like Mali-400MP and
potentially other drivers on Windows).

[Ken Russell (switch to Gerrit), 2015-08-11 16:53:15]

This looks generally good but I wonder would it be better to put in a GPU driver
bug workaround to control this behavior?

I'd like to defer to jbauman@ for review as I'm OOO this week and can't devote
the time to thinking through the implications.

[jbauman, 2015-08-11 23:07:57]

lgtm. I think it makes sense to have this as the default until we're sure that
our ARB_robustness support is correct
(https://code.google.com/p/chromium/issues/detail?id=480545).

[no sievers, 2015-08-11 23:14:51]

The CQ bit was checked by sievers@chromium.org

[commit-bot: I haz the power, 2015-08-11 23:15:05]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1282973004/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1282973004/1

[commit-bot: I haz the power, 2015-08-11 23:23:18]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2015-08-11 23:24:03]

Patchset 1 (id:??) landed as
https://crrev.com/751a98d265bf854b6b6d216fb7d20ef6ec1313c8
Cr-Commit-Position: refs/heads/master@{#342928}