aw: Fix HardwareView for instrumentation tests

aw: Fix HardwareView for instrumentation tests

* Implement detachGLFunctor. The contract here is after detachGLFunctor
  is called, DrawGL is no longer called until the next requestDrawGL.
  Implement just with a mFunctorAttached variable.
* Only create one HardwareView. Each GLSurfaceView creates its own GL
  thread, but we only assume a single render thread. Using SurfaceView
  or hosting multiple AwContents in the same GLSurfaceView is too hard,
  so just do the simplest thing, and create the first view has hardware
  accelerated.

BUG=416981
Committed: https://crrev.com/cc5ed8647904a5bab9a1062a2a3a786e0152cfd7
Cr-Commit-Position: refs/heads/master@{#296261}

[boliu, 2014-09-23 20:16:25]

boliu@chromium.org changed reviewers:
+ hush@chromium.org, sievers@chromium.org

[no sievers, 2014-09-23 20:24:38]

lgtm!

https://codereview.chromium.org/594333002/diff/1/android_webview/test/shell/s...
File
android_webview/test/shell/src/org/chromium/android_webview/test/AwTestContainerView.java
(right):

https://codereview.chromium.org/594333002/diff/1/android_webview/test/shell/s...
android_webview/test/shell/src/org/chromium/android_webview/test/AwTestContainerView.java:189:
return;
we even don't process GL here if mNeedsProcessGL?

Does the caller guarantee that all GL was flushed (with ScopedAllowGL) when it
detaches the functor (and that no GL is issue afterwards)?

https://codereview.chromium.org/594333002/diff/1/android_webview/test/shell/s...
android_webview/test/shell/src/org/chromium/android_webview/test/AwTestContainerView.java:426:
if (mHardwareView == null) return;
nit: if (mHardwareView) mHardwareView.detachGLFunctor();

[boliu, 2014-09-23 20:29:32]

https://codereview.chromium.org/594333002/diff/1/android_webview/test/shell/s...
File
android_webview/test/shell/src/org/chromium/android_webview/test/AwTestContainerView.java
(right):

https://codereview.chromium.org/594333002/diff/1/android_webview/test/shell/s...
android_webview/test/shell/src/org/chromium/android_webview/test/AwTestContainerView.java:189:
return;
On 2014/09/23 20:24:38, sievers wrote:
> we even don't process GL here if mNeedsProcessGL?
> 
> Does the caller guarantee that all GL was flushed (with ScopedAllowGL) when it
> detaches the functor (and that no GL is issue afterwards)?

Hmm good question. Yes it's fine for because webview always flushes first with
kModeProcess and waitForCompletion = true before detaching functor.

[boliu, 2014-09-23 20:35:21]

https://codereview.chromium.org/594333002/diff/1/android_webview/test/shell/s...
File
android_webview/test/shell/src/org/chromium/android_webview/test/AwTestContainerView.java
(right):

https://codereview.chromium.org/594333002/diff/1/android_webview/test/shell/s...
android_webview/test/shell/src/org/chromium/android_webview/test/AwTestContainerView.java:426:
if (mHardwareView == null) return;
On 2014/09/23 20:24:38, sievers wrote:
> nit: if (mHardwareView) mHardwareView.detachGLFunctor();

Done. Except it's (mHardwareView != null) since this is java, not c++.

[boliu, 2014-09-23 20:35:41]

The CQ bit was checked by boliu@chromium.org

[boliu, 2014-09-23 20:36:32]

Down to 6 failing tests with this :D

http://build.chromium.org/p/tryserver.chromium.linux/builders/android_dbg_tes...

[commit-bot: I haz the power, 2014-09-23 20:39:43]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/594333002/20001

[hush (inactive), 2014-09-23 20:41:02]

https://codereview.chromium.org/594333002/diff/20001/android_webview/test/she...
File
android_webview/test/shell/src/org/chromium/android_webview/test/AwTestContainerView.java
(right):

https://codereview.chromium.org/594333002/diff/20001/android_webview/test/she...
android_webview/test/shell/src/org/chromium/android_webview/test/AwTestContainerView.java:230:
public AwTestContainerView(Context context, boolean hardwareAccelerated) {
so even you pass hardwareAccelerated to be true here, you may not get a hardware
enabled AwTestContainerView returned to you.
I'm not sure if there are any current instrumentation tests that assume hardware
acceleration to be on that will be broken by this.

[boliu, 2014-09-23 20:42:33]

https://codereview.chromium.org/594333002/diff/20001/android_webview/test/she...
File
android_webview/test/shell/src/org/chromium/android_webview/test/AwTestContainerView.java
(right):

https://codereview.chromium.org/594333002/diff/20001/android_webview/test/she...
android_webview/test/shell/src/org/chromium/android_webview/test/AwTestContainerView.java:230:
public AwTestContainerView(Context context, boolean hardwareAccelerated) {
On 2014/09/23 20:41:02, hush wrote:
> so even you pass hardwareAccelerated to be true here, you may not get a
hardware
> enabled AwTestContainerView returned to you.
> I'm not sure if there are any current instrumentation tests that assume
hardware
> acceleration to be on that will be broken by this.

All instrumentation tests are software only right now:
https://code.google.com/p/chromium/codesearch#chromium/src/android_webview/ja...

[commit-bot: I haz the power, 2014-09-23 21:45:41]

Committed patchset #2 (id:20001) as f626dfa3d91f13af0c0245e847160ceafd70b9be

[commit-bot: I haz the power, 2014-09-23 21:46:23]

Patchset 2 (id:??) landed as
https://crrev.com/cc5ed8647904a5bab9a1062a2a3a786e0152cfd7
Cr-Commit-Position: refs/heads/master@{#296261}