Android: Add strong binding for GPU process

Android: Put GPU process in foreground

That makes browser, foreground renderer, and GPU
have a similar priority. Otherwise the GPU process
might drop behind other (even unrelated bg) processes
and get killed too easily.

BUG=448549, 453671
TBR=ppi@chromium.org

Committed: https://crrev.com/e77a53958cc41a1df157929f492a52c564d37b1e
Cr-Commit-Position: refs/heads/master@{#314173}

[no sievers, 2015-01-30 22:57:32]

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

[no sievers, 2015-01-30 23:01:37]

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

[no sievers, 2015-01-30 23:03:01]

Ted, mind giving this a quick sanity check?

ppi@ definitely needs to look at this too, but I'd like to land this as a
bandaid to see how it affects some issues (GPU process getting potentially
killed very easily on some devices).

There is probably a way to make all this a bit nicer though (maybe through
inheritance or so, right now some stuff is a bit hardcoded to renderers).

[Ted C, 2015-01-30 23:38:14]

On 2015/01/30 23:03:01, sievers wrote:
> Ted, mind giving this a quick sanity check?
> 
> ppi@ definitely needs to look at this too, but I'd like to land this as a
> bandaid to see how it affects some issues (GPU process getting potentially
> killed very easily on some devices).
> 
> There is probably a way to make all this a bit nicer though (maybe through
> inheritance or so, right now some stuff is a bit hardcoded to renderers).

lgtm

[no sievers, 2015-01-31 02:23:14]

The CQ bit was checked by sievers@chromium.org

[commit-bot: I haz the power, 2015-01-31 02:23:39]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/889013002/1

[commit-bot: I haz the power, 2015-01-31 09:21:41]

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

[commit-bot: I haz the power, 2015-01-31 09:21:41]

Try jobs failed on following builders:
  win8_chromium_gn_rel on tryserver.chromium.win
(http://build.chromium.org/p/tryserver.chromium.win/builders/win8_chromium_gn_...)

[no sievers, 2015-02-02 18:45:44]

The CQ bit was checked by sievers@chromium.org

[commit-bot: I haz the power, 2015-02-02 18:47:13]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/889013002/1

[commit-bot: I haz the power, 2015-02-02 18:48:24]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2015-02-02 18:49:22]

Patchset 1 (id:??) landed as
https://crrev.com/e77a53958cc41a1df157929f492a52c564d37b1e
Cr-Commit-Position: refs/heads/master@{#314173}

[ppi, 2015-02-03 14:29:11]

Refactoring it at some point sounds good, I could imagine this becoming
ChildProcessConnectionBase, ChildRendererProcessConnection and
ChildGpuProcessConnection.

One comment below.

https://codereview.chromium.org/889013002/diff/1/content/public/android/java/...
File
content/public/android/java/src/org/chromium/content/browser/ChildProcessConnectionImpl.java
(right):

https://codereview.chromium.org/889013002/diff/1/content/public/android/java/...
content/public/android/java/src/org/chromium/content/browser/ChildProcessConnectionImpl.java:417:
synchronized (mLock) {
Should we put "assert !mAlwaysInForeground;" here too?

BTW, are we going to use separate gpu process when running on Svelte?

[no sievers, 2015-02-04 21:37:24]

On 2015/02/03 14:29:11, ppi wrote:
> Refactoring it at some point sounds good, I could imagine this becoming
> ChildProcessConnectionBase, ChildRendererProcessConnection and
> ChildGpuProcessConnection.
> 
> One comment below.
> 
>
https://codereview.chromium.org/889013002/diff/1/content/public/android/java/...
> File
>
content/public/android/java/src/org/chromium/content/browser/ChildProcessConnectionImpl.java
> (right):
> 
>
https://codereview.chromium.org/889013002/diff/1/content/public/android/java/...
>
content/public/android/java/src/org/chromium/content/browser/ChildProcessConnectionImpl.java:417:
> synchronized (mLock) {
> Should we put "assert !mAlwaysInForeground;" here too?
> 

Done in https://codereview.chromium.org/902593003

> BTW, are we going to use separate gpu process when running on Svelte?

No, I kept kInProcessGpu (at least for the time being) for those devices (see
content_startup_flags.cc) since we were worried about unexpected behavior wrt
Android's process management when under such memory constraints.