Support stream textures with the synchronous compositor

Support stream textures with the synchronous compositor

This adds a thread-safe StreamTextureManager for GLInProcessContext.

It further adds some plumbing for the webkit thread to be able
 to create a StreamTextureFactory from the main thread (see render_view_impl.cc)
 with the ability to extract the thread-safe SurfaceTexture(Bridge) for a given stream.

This is nothing but a crazy amount of plumbing to achieve this:
- StreamTexture::Update() i.e. SurfaceTexture::updateTexImage needs to happen on the GL thread
with a current context
- The StreamTextureFactory and Proxy are referenced from the WebmediaPlayer instance which lives on
the main thread
- The callback for invalidating the video is posted to the
compositor thread directly (here: UI thread) bypassing the main thread

BUG=239760
NOTRY=True

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=219057

[no sievers, 2013-08-16 02:27:14]

Absolutely mindless and terrible.

[boliu, 2013-08-20 21:33:49]

Only looked at content

https://codereview.chromium.org/23234003/diff/4001/content/browser/android/in...
File content/browser/android/in_process/synchronous_compositor_impl.cc (right):

https://codereview.chromium.org/23234003/diff/4001/content/browser/android/in...
content/browser/android/in_process/synchronous_compositor_impl.cc:106: const
char* allowed_extensions = "*";
Leave a comment about deduping this code.

https://codereview.chromium.org/23234003/diff/4001/content/browser/android/in...
content/browser/android/in_process/synchronous_compositor_impl.cc:186:
gpu::GLInProcessContext* offscreen_context_as_gl_in_process_context_;
Indicate with var name or a comment that this is a pointer into
|offscreen_context_for_main_thread_| and that's why it's safe.

https://codereview.chromium.org/23234003/diff/4001/content/renderer/media/and...
File
content/renderer/media/android/stream_texture_factory_android_synchronous_impl.cc
(right):

https://codereview.chromium.org/23234003/diff/4001/content/renderer/media/and...
content/renderer/media/android/stream_texture_factory_android_synchronous_impl.cc:80:
LOG(ERROR) << "Already bound to thread.";
I think this can be called multiple times if the video is paused then played
again.

https://codereview.chromium.org/23234003/diff/4001/content/renderer/media/and...
content/renderer/media/android/stream_texture_factory_android_synchronous_impl.cc:83:
loop_ = base::MessageLoopProxy::current();
Hmm...should we have two thread checkers, one for main thread (constructor,
Release, SetClient), then create another thread checker here for
OnFrameAvailable.

Or maybe just DCHECK loop with current in OnFrameAvailable.

https://codereview.chromium.org/23234003/diff/4001/content/renderer/render_vi...
File content/renderer/render_view_impl.cc (right):

https://codereview.chromium.org/23234003/diff/4001/content/renderer/render_vi...
content/renderer/render_view_impl.cc:3031: if (SynchronousCompositorFactory*
factory =
This is a redundant check. UsingSynchronousRendererCompositor() checks exactly
that the factory is not null. But factory is android only code, so need
OS_ANDROID guards. :(

[no sievers, 2013-08-20 21:59:58]

https://codereview.chromium.org/23234003/diff/4001/content/browser/android/in...
File content/browser/android/in_process/synchronous_compositor_impl.cc (right):

https://codereview.chromium.org/23234003/diff/4001/content/browser/android/in...
content/browser/android/in_process/synchronous_compositor_impl.cc:106: const
char* allowed_extensions = "*";
On 2013/08/20 21:33:49, boliu wrote:
> Leave a comment about deduping this code.

Well, ideally we wouldn't have one thing (GLInProcessContext) wrapped in another
thing (WGC3D) wrapped in another thing (ContextProvider).
Ideally, we'd just not be using WGC3D in content/ anymore.

https://codereview.chromium.org/23234003/diff/4001/content/browser/android/in...
content/browser/android/in_process/synchronous_compositor_impl.cc:186:
gpu::GLInProcessContext* offscreen_context_as_gl_in_process_context_;
On 2013/08/20 21:33:49, boliu wrote:
> Indicate with var name or a comment that this is a pointer into
> |offscreen_context_for_main_thread_| and that's why it's safe.

Done.

https://codereview.chromium.org/23234003/diff/4001/content/renderer/media/and...
File
content/renderer/media/android/stream_texture_factory_android_synchronous_impl.cc
(right):

https://codereview.chromium.org/23234003/diff/4001/content/renderer/media/and...
content/renderer/media/android/stream_texture_factory_android_synchronous_impl.cc:80:
LOG(ERROR) << "Already bound to thread.";
On 2013/08/20 21:33:49, boliu wrote:
> I think this can be called multiple times if the video is paused then played
> again.

Argh, that was never intended, but ok.

https://codereview.chromium.org/23234003/diff/4001/content/renderer/media/and...
content/renderer/media/android/stream_texture_factory_android_synchronous_impl.cc:83:
loop_ = base::MessageLoopProxy::current();
On 2013/08/20 21:33:49, boliu wrote:
> Hmm...should we have two thread checkers, one for main thread (constructor,
> Release, SetClient), then create another thread checker here for
> OnFrameAvailable.

Constructor, Release and SetClient can be called on any thread. And it does not
have to be consistent.


> 
> Or maybe just DCHECK loop with current in OnFrameAvailable.

https://codereview.chromium.org/23234003/diff/4001/content/renderer/render_vi...
File content/renderer/render_view_impl.cc (right):

https://codereview.chromium.org/23234003/diff/4001/content/renderer/render_vi...
content/renderer/render_view_impl.cc:3031: if (SynchronousCompositorFactory*
factory =
On 2013/08/20 21:33:49, boliu wrote:
> This is a redundant check. UsingSynchronousRendererCompositor() checks exactly
> that the factory is not null. But factory is android only code, so need
> OS_ANDROID guards. :(

Done. The whole section here is already ifdef ANDROID.

[boliu, 2013-08-20 22:14:33]

content lgtm

https://codereview.chromium.org/23234003/diff/21001/content/renderer/render_v...
File content/renderer/render_view_impl.cc (right):

https://codereview.chromium.org/23234003/diff/21001/content/renderer/render_v...
content/renderer/render_view_impl.cc:3033: stream_texture_factory =
factory->CreateStreamTextureFactory(routing_id_);
nit: indent

[no sievers, 2013-08-20 22:24:58]

https://codereview.chromium.org/23234003/diff/21001/content/renderer/render_v...
File content/renderer/render_view_impl.cc (right):

https://codereview.chromium.org/23234003/diff/21001/content/renderer/render_v...
content/renderer/render_view_impl.cc:3033: stream_texture_factory =
factory->CreateStreamTextureFactory(routing_id_);
On 2013/08/20 22:14:33, boliu wrote:
> nit: indent

Done.

[no sievers, 2013-08-20 22:28:20]

+piman

This is all really in need of a complete overhaul, but this just fills in an
implementation that is compatible with the existing one (and without touching
the implementation used by Chrome).

(One motivation for having all this StreamTextureFactory/Proxy abstraction was
originally DEPS-related with MediaPlayer living in webkit/, which is not the
case anymore.)

[piman, 2013-08-22 01:23:56]

lgtm

[commit-bot: I haz the power, 2013-08-22 01:28:25]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/sievers@chromium.org/23234003/29001

[commit-bot: I haz the power, 2013-08-22 01:55:33]

Sorry for I got bad news for ya.
Compile failed with a clobber build on linux_rel.
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=linux_rel&...
Your code is likely broken or HEAD is junk. Please ensure your
code is not broken then alert the build sheriffs.
Look at the try server FAQ for more details.

[commit-bot: I haz the power, 2013-08-22 17:18:17]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/sievers@chromium.org/23234003/50001

[commit-bot: I haz the power, 2013-08-22 17:44:34]

Sorry for I got bad news for ya.
Compile failed with a clobber build on android_clang_dbg.
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=android_cl...
Your code is likely broken or HEAD is junk. Please ensure your
code is not broken then alert the build sheriffs.
Look at the try server FAQ for more details.

[commit-bot: I haz the power, 2013-08-22 18:43:49]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/sievers@chromium.org/23234003/65001

[commit-bot: I haz the power, 2013-08-22 18:54:13]

Change committed as 219057