Make RendererGpuVideoDecoderFactories live on arbitrary threads.

content/renderer/media/renderer_gpu_video_decoder_factories.cc

 // Copyright (c) 2012 The Chromium Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file.
 
 #include "content/renderer/media/renderer_gpu_video_decoder_factories.h"
 
 #include <GLES2/gl2.h>
 #include <GLES2/gl2ext.h>
 
 #include "base/bind.h"
 #include "content/child/child_thread.h"
 #include "content/common/gpu/client/gpu_channel_host.h"
 #include "content/common/gpu/client/webgraphicscontext3d_command_buffer_impl.h"
 #include "gpu/command_buffer/client/gles2_implementation.h"
 #include "gpu/ipc/command_buffer_proxy.h"
 #include "third_party/skia/include/core/SkPixelRef.h"
 
 namespace content {
 
 RendererGpuVideoDecoderFactories::~RendererGpuVideoDecoderFactories() {}
 RendererGpuVideoDecoderFactories::RendererGpuVideoDecoderFactories(
     GpuChannelHost* gpu_channel_host,
-    const scoped_refptr<base::MessageLoopProxy>& compositor_message_loop,
+    const scoped_refptr<base::MessageLoopProxy>& message_loop,
     WebGraphicsContext3DCommandBufferImpl* context)
-    : compositor_message_loop_(compositor_message_loop),
+    : message_loop_(message_loop),
       main_message_loop_(base::MessageLoopProxy::current()),
       gpu_channel_host_(gpu_channel_host),
       aborted_waiter_(true, false),
-      compositor_loop_async_waiter_(false, false),
+      message_loop_async_waiter_(false, false),
       render_thread_async_waiter_(false, false) {
-  if (compositor_message_loop_->BelongsToCurrentThread()) {
+  if (message_loop_->BelongsToCurrentThread()) {
     AsyncGetContext(context);
-    compositor_loop_async_waiter_.Reset();
+    message_loop_async_waiter_.Reset();
     return;
   }
-  // Threaded compositor requires us to wait for the context to be acquired.
-  compositor_message_loop_->PostTask(FROM_HERE, base::Bind(
+  // Wait for the context to be acquired.
+  message_loop_->PostTask(FROM_HERE, base::Bind(
       &RendererGpuVideoDecoderFactories::AsyncGetContext,
       // Unretained to avoid ref/deref'ing |*this|, which is not yet stored in a
       // scoped_refptr.  Safe because the Wait() below keeps us alive until this
       // task completes.
       base::Unretained(this),
       // OK to pass raw because the pointee is only deleted on the compositor
       // thread, and only as the result of a PostTask from the render thread
       // which can only happen after this function returns, so our PostTask will
       // run first.
       context));
-  compositor_loop_async_waiter_.Wait();
+  message_loop_async_waiter_.Wait();
 }
 
+RendererGpuVideoDecoderFactories::RendererGpuVideoDecoderFactories()
+    : aborted_waiter_(true, false),
+      message_loop_async_waiter_(false, false),
+      render_thread_async_waiter_(false, false) {}
+
 void RendererGpuVideoDecoderFactories::AsyncGetContext(
     WebGraphicsContext3DCommandBufferImpl* context) {
   context_ = context->AsWeakPtr();
   if (context_.get()) {
     if (context_->makeContextCurrent()) {
       // Called once per media player, but is a no-op after the first one in
       // each renderer.
       context_->insertEventMarkerEXT("GpuVDAContext3D");
     }
   }
-  compositor_loop_async_waiter_.Signal();
+  message_loop_async_waiter_.Signal();
 }
 
 media::VideoDecodeAccelerator*
 RendererGpuVideoDecoderFactories::CreateVideoDecodeAccelerator(
     media::VideoCodecProfile profile,
     media::VideoDecodeAccelerator::Client* client) {
-  if (compositor_message_loop_->BelongsToCurrentThread()) {
+  if (message_loop_->BelongsToCurrentThread()) {
     AsyncCreateVideoDecodeAccelerator(profile, client);
-    compositor_loop_async_waiter_.Reset();
+    message_loop_async_waiter_.Reset();
     return vda_.release();
   }
   // The VDA is returned in the vda_ member variable by the
   // AsyncCreateVideoDecodeAccelerator() function.
-  compositor_message_loop_->PostTask(FROM_HERE, base::Bind(
+  message_loop_->PostTask(FROM_HERE, base::Bind(
       &RendererGpuVideoDecoderFactories::AsyncCreateVideoDecodeAccelerator,
       this, profile, client));
 
   base::WaitableEvent* objects[] = {&aborted_waiter_,
-                                    &compositor_loop_async_waiter_};
+                                    &message_loop_async_waiter_};
   if (base::WaitableEvent::WaitMany(objects, arraysize(objects)) == 0) {
     // If we are aborting and the VDA is created by the
     // AsyncCreateVideoDecodeAccelerator() function later we need to ensure
     // that it is destroyed on the same thread.
-    compositor_message_loop_->PostTask(FROM_HERE, base::Bind(
+    message_loop_->PostTask(FROM_HERE, base::Bind(
         &RendererGpuVideoDecoderFactories::AsyncDestroyVideoDecodeAccelerator,
         this));
     return NULL;
   }
   return vda_.release();
 }
 
 void RendererGpuVideoDecoderFactories::AsyncCreateVideoDecodeAccelerator(
       media::VideoCodecProfile profile,
       media::VideoDecodeAccelerator::Client* client) {
-  DCHECK(compositor_message_loop_->BelongsToCurrentThread());
+  DCHECK(message_loop_->BelongsToCurrentThread());
 
   if (context_.get() && context_->GetCommandBufferProxy()) {
     vda_ = gpu_channel_host_->CreateVideoDecoder(
         context_->GetCommandBufferProxy()->GetRouteID(), profile, client);
   }
-  compositor_loop_async_waiter_.Signal();
+  message_loop_async_waiter_.Signal();
 }
 
 uint32 RendererGpuVideoDecoderFactories::CreateTextures(
     int32 count, const gfx::Size& size,
     std::vector<uint32>* texture_ids,
     std::vector<gpu::Mailbox>* texture_mailboxes,
     uint32 texture_target) {
   uint32 sync_point = 0;
 
-  if (compositor_message_loop_->BelongsToCurrentThread()) {
+  if (message_loop_->BelongsToCurrentThread()) {
     AsyncCreateTextures(count, size, texture_target, &sync_point);
     texture_ids->swap(created_textures_);
     texture_mailboxes->swap(created_texture_mailboxes_);
-    compositor_loop_async_waiter_.Reset();
+    message_loop_async_waiter_.Reset();
     return sync_point;
   }
-  compositor_message_loop_->PostTask(FROM_HERE, base::Bind(
+  message_loop_->PostTask(FROM_HERE, base::Bind(
       &RendererGpuVideoDecoderFactories::AsyncCreateTextures, this,
       count, size, texture_target, &sync_point));
 
   base::WaitableEvent* objects[] = {&aborted_waiter_,
-                                    &compositor_loop_async_waiter_};
+                                    &message_loop_async_waiter_};
   if (base::WaitableEvent::WaitMany(objects, arraysize(objects)) == 0)
     return 0;
   texture_ids->swap(created_textures_);
   texture_mailboxes->swap(created_texture_mailboxes_);
   return sync_point;
 }
 
 void RendererGpuVideoDecoderFactories::AsyncCreateTextures(
     int32 count, const gfx::Size& size, uint32 texture_target,
     uint32* sync_point) {
-  DCHECK(compositor_message_loop_->BelongsToCurrentThread());
+  DCHECK(message_loop_->BelongsToCurrentThread());
   DCHECK(texture_target);
 
   if (!context_.get()) {
-    compositor_loop_async_waiter_.Signal();
+    message_loop_async_waiter_.Signal();
     return;
   }
   gpu::gles2::GLES2Implementation* gles2 = context_->GetImplementation();
   created_textures_.resize(count);
   created_texture_mailboxes_.resize(count);
   gles2->GenTextures(count, &created_textures_[0]);
   for (int i = 0; i < count; ++i) {
     gles2->ActiveTexture(GL_TEXTURE0);
     uint32 texture_id = created_textures_[i];
     gles2->BindTexture(texture_target, texture_id);
@@ skipping 10 matching lines @@
                                   created_texture_mailboxes_[i].name);
   }
 
   // We need a glFlush here to guarantee the decoder (in the GPU process) can
   // use the texture ids we return here.  Since textures are expected to be
   // reused, this should not be unacceptably expensive.
   gles2->Flush();
   DCHECK_EQ(gles2->GetError(), static_cast<GLenum>(GL_NO_ERROR));
 
   *sync_point = gles2->InsertSyncPointCHROMIUM();
-  compositor_loop_async_waiter_.Signal();
+  message_loop_async_waiter_.Signal();
 }
 
 void RendererGpuVideoDecoderFactories::DeleteTexture(uint32 texture_id) {
-  if (compositor_message_loop_->BelongsToCurrentThread()) {
+  if (message_loop_->BelongsToCurrentThread()) {
     AsyncDeleteTexture(texture_id);
     return;
   }
-  compositor_message_loop_->PostTask(FROM_HERE, base::Bind(
+  message_loop_->PostTask(FROM_HERE, base::Bind(
       &RendererGpuVideoDecoderFactories::AsyncDeleteTexture, this, texture_id));
 }
 
 void RendererGpuVideoDecoderFactories::AsyncDeleteTexture(uint32 texture_id) {
-  DCHECK(compositor_message_loop_->BelongsToCurrentThread());
+  DCHECK(message_loop_->BelongsToCurrentThread());
   if (!context_.get())
     return;
 
   gpu::gles2::GLES2Implementation* gles2 = context_->GetImplementation();
   gles2->DeleteTextures(1, &texture_id);
   DCHECK_EQ(gles2->GetError(), static_cast<GLenum>(GL_NO_ERROR));
 }
 
 void RendererGpuVideoDecoderFactories::WaitSyncPoint(uint32 sync_point) {
-  if (compositor_message_loop_->BelongsToCurrentThread()) {
+  if (message_loop_->BelongsToCurrentThread()) {
     AsyncWaitSyncPoint(sync_point);
     return;

[Ami GONE FROM CHROMIUM, 2013/07/22 19:46:21]

message_loop_async_waiter_.Reset();

[wuchengli, 2013/07/23 16:29:28]

Done.

   }
 
-  compositor_message_loop_->PostTask(FROM_HERE, base::Bind(
+  message_loop_->PostTask(FROM_HERE, base::Bind(
       &RendererGpuVideoDecoderFactories::AsyncWaitSyncPoint,
       this,
       sync_point));
   base::WaitableEvent* objects[] = {&aborted_waiter_,
-                                    &compositor_loop_async_waiter_};
+                                    &message_loop_async_waiter_};
   base::WaitableEvent::WaitMany(objects, arraysize(objects));
 }
 
 void RendererGpuVideoDecoderFactories::AsyncWaitSyncPoint(uint32 sync_point) {
-  DCHECK(compositor_message_loop_->BelongsToCurrentThread());
+  DCHECK(message_loop_->BelongsToCurrentThread());
   if (!context_) {
-    compositor_loop_async_waiter_.Signal();
+    message_loop_async_waiter_.Signal();
     return;
   }
 
   gpu::gles2::GLES2Implementation* gles2 = context_->GetImplementation();
   gles2->WaitSyncPointCHROMIUM(sync_point);
-  compositor_loop_async_waiter_.Signal();
+  message_loop_async_waiter_.Signal();
 }
 
 void RendererGpuVideoDecoderFactories::ReadPixels(
     uint32 texture_id, uint32 texture_target, const gfx::Size& size,
     const SkBitmap& pixels) {
   // SkBitmaps use the SkPixelRef object to refcount the underlying pixels.
   // Multiple SkBitmaps can share a SkPixelRef instance. We use this to
   // ensure that the underlying pixels in the SkBitmap passed in remain valid
   // until the AsyncReadPixels() call completes.
   read_pixels_bitmap_.setPixelRef(pixels.pixelRef());
 
-  if (!compositor_message_loop_->BelongsToCurrentThread()) {
-    compositor_message_loop_->PostTask(FROM_HERE, base::Bind(
+  if (!message_loop_->BelongsToCurrentThread()) {
+    message_loop_->PostTask(FROM_HERE, base::Bind(
         &RendererGpuVideoDecoderFactories::AsyncReadPixels, this,
         texture_id, texture_target, size));
     base::WaitableEvent* objects[] = {&aborted_waiter_,
-                                      &compositor_loop_async_waiter_};
+                                      &message_loop_async_waiter_};
     if (base::WaitableEvent::WaitMany(objects, arraysize(objects)) == 0)
       return;
   } else {
     AsyncReadPixels(texture_id, texture_target, size);

[Ami GONE FROM CHROMIUM, 2013/07/22 19:46:21]

message_loop_async_waiter_.Reset();

[wuchengli, 2013/07/23 16:29:28]

Done.

   }
   read_pixels_bitmap_.setPixelRef(NULL);
 }
 
 void RendererGpuVideoDecoderFactories::AsyncReadPixels(
     uint32 texture_id, uint32 texture_target, const gfx::Size& size) {
-  DCHECK(compositor_message_loop_->BelongsToCurrentThread());
+  DCHECK(message_loop_->BelongsToCurrentThread());
   if (!context_.get()) {
-    compositor_loop_async_waiter_.Signal();
+    message_loop_async_waiter_.Signal();
     return;
   }
 
   gpu::gles2::GLES2Implementation* gles2 = context_->GetImplementation();
 
   GLuint tmp_texture;
   gles2->GenTextures(1, &tmp_texture);
   gles2->BindTexture(texture_target, tmp_texture);
   gles2->TexParameteri(texture_target, GL_TEXTURE_MIN_FILTER, GL_LINEAR);
   gles2->TexParameteri(texture_target, GL_TEXTURE_MAG_FILTER, GL_LINEAR);
   gles2->TexParameteri(texture_target, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE);
   gles2->TexParameteri(texture_target, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE);
   context_->copyTextureCHROMIUM(
       texture_target, texture_id, tmp_texture, 0, GL_RGBA, GL_UNSIGNED_BYTE);
 
   GLuint fb;
   gles2->GenFramebuffers(1, &fb);
   gles2->BindFramebuffer(GL_FRAMEBUFFER, fb);
   gles2->FramebufferTexture2D(GL_FRAMEBUFFER, GL_COLOR_ATTACHMENT0,
                               texture_target, tmp_texture, 0);
   gles2->PixelStorei(GL_PACK_ALIGNMENT, 4);
   gles2->ReadPixels(0, 0, size.width(), size.height(), GL_BGRA_EXT,
       GL_UNSIGNED_BYTE, read_pixels_bitmap_.pixelRef()->pixels());
   gles2->DeleteFramebuffers(1, &fb);
   gles2->DeleteTextures(1, &tmp_texture);
   DCHECK_EQ(gles2->GetError(), static_cast<GLenum>(GL_NO_ERROR));
-  compositor_loop_async_waiter_.Signal();
+  message_loop_async_waiter_.Signal();
 }
 
 base::SharedMemory* RendererGpuVideoDecoderFactories::CreateSharedMemory(
     size_t size) {
   if (main_message_loop_->BelongsToCurrentThread()) {
     return ChildThread::current()->AllocateSharedMemory(size);
   }
   main_message_loop_->PostTask(FROM_HERE, base::Bind(
       &RendererGpuVideoDecoderFactories::AsyncCreateSharedMemory, this,
       size));
 
   base::WaitableEvent* objects[] = {&aborted_waiter_,
                                     &render_thread_async_waiter_};
   if (base::WaitableEvent::WaitMany(objects, arraysize(objects)) == 0)
     return NULL;
   return shared_memory_segment_.release();
 }
 
 void RendererGpuVideoDecoderFactories::AsyncCreateSharedMemory(size_t size) {
   DCHECK_EQ(base::MessageLoop::current(),
             ChildThread::current()->message_loop());
 
   shared_memory_segment_.reset(
       ChildThread::current()->AllocateSharedMemory(size));
   render_thread_async_waiter_.Signal();
 }
 
 scoped_refptr<base::MessageLoopProxy>
 RendererGpuVideoDecoderFactories::GetMessageLoop() {
-  return compositor_message_loop_;
+  return message_loop_;
 }
 
 void RendererGpuVideoDecoderFactories::Abort() {
   aborted_waiter_.Signal();
 }
 
 bool RendererGpuVideoDecoderFactories::IsAborted() {
   return aborted_waiter_.IsSignaled();
 }
 
+scoped_refptr<media::GpuVideoDecoder::Factories>
+RendererGpuVideoDecoderFactories::Clone() {
+  scoped_refptr<RendererGpuVideoDecoderFactories> factories =
+      new RendererGpuVideoDecoderFactories();

[Ami GONE FROM CHROMIUM, 2013/07/22 19:46:21]

AFAICT you can reuse the non-trivial ctor and then just set main_message_loop_
explicitly.
(if you do this then you can revert the IMPLICIT->COPY_ASSIGN diff in the
header)

[wuchengli, 2013/07/23 16:29:28]

Non-trial ctor takes raw pointer of WebGraphicsContext3DCommandBufferImpl. But
Clone() is called on libjingle thread, not |message_loop_| (media thread). So we
cannot dereference the weak pointer |context_| here.

In my understanding, make a copy of WeakPtr on a different thread is OK.

+  factories->message_loop_ = message_loop_;
+  factories->main_message_loop_ = main_message_loop_;
+  factories->gpu_channel_host_ = gpu_channel_host_;
+  factories->context_ = context_;
+  return factories;
+}
+
 void RendererGpuVideoDecoderFactories::AsyncDestroyVideoDecodeAccelerator() {
   // OK to release because Destroy() will delete the VDA instance.
   if (vda_)
     vda_.release()->Destroy();
 }
 
 }  // namespace content