Allow WebGL on Android if the GPU has it's own process (Take 2)

Allow WebGL on Android if the GPU has it's own process (Take 2)

First attempt (https://codereview.chromium.org/1247093006/) was reverted due to
failures on the Android bots. This is because the Android bots use content shell,
not Chromium, to run the tests and that uses in-process GPU. As such the test
were running without WebGL support.

This patch adds back in the EXT_robusness exceptions, which allowed the bots to
run WebGL content previously and won't appear to users as a regression.

BUG=512992
TBR=mkwst@chromium.org

Committed: https://crrev.com/e3677b649fff6a114d4c28de6afddef8cc5320db
Cr-Commit-Position: refs/heads/master@{#340385}

[bajones, 2015-07-24 21:20:37]

bajones@chromium.org changed reviewers:
+ aelias@chromium.org, kbr@chromium.org, mkwst@chromium.org, zmo@chromium.org

[bajones, 2015-07-24 21:20:37]

Please take another look. Only difference between this and the previous CL is
that this one does not remove the EXT_robustness exception in the software
rendering list. That was we don't regress anywhere (especially the bots)

[Ken Russell (switch to Gerrit), 2015-07-24 21:33:04]

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

[Ken Russell (switch to Gerrit), 2015-07-24 21:33:05]

+sievers

I think this is the most compatible change. The fact that the bots broke is
indicative that we might break WebGL support in WebView with the previous
strategy.

LGTM from my standpoint, but please wait for others' reviews.

[aelias_OOO_until_Jul13, 2015-07-24 22:47:42]

Good point about WebView.  lgtm

[no sievers, 2015-07-24 22:55:29]

lgtm. ContentShell uses in-process-gpu, whaat?

[Ken Russell (switch to Gerrit), 2015-07-24 23:01:57]

On 2015/07/24 22:55:29, sievers wrote:
> lgtm. ContentShell uses in-process-gpu, whaat?

This was a surprise to me too. The decision must depend on the platform; I'm
pretty sure content_shell uses a separate GPU process on desktop platforms.

The Android bots on the chromium.gpu waterfall therefore aren't testing what we
ship. We should convert them to build and run ChromePublic.apk, which I assume
is fully multi-process.

[bajones, 2015-07-24 23:21:41]

TBRing Mike, since he's already approved the previous CL and gpu_messages hasn't
changed since then.

[bajones, 2015-07-24 23:21:44]

The CQ bit was checked by bajones@chromium.org

[commit-bot: I haz the power, 2015-07-24 23:22:25]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1252943008/20001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1252943008/20001

[commit-bot: I haz the power, 2015-07-25 00:42:04]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2015-07-25 00:42:57]

Patchset 2 (id:??) landed as
https://crrev.com/e3677b649fff6a114d4c28de6afddef8cc5320db
Cr-Commit-Position: refs/heads/master@{#340385}

[Mike West, 2015-07-25 04:43:00]

LGTM.