Chromium Code Reviews
chromiumcodereview-hr@appspot.gserviceaccount.com (chromiumcodereview-hr) | Please choose your nickname with Settings | Help | Chromium Project | Gerrit Changes | Sign out
(81)

Unified Diff: media/audio/audio_output_device.cc

Issue 1830933002: Making AudioThread::Thread restartable *** DRAFT *** (Closed) Base URL: https://chromium.googlesource.com/chromium/src.git@master
Patch Set: Created 4 years, 9 months ago
Use n/p to move between diff chunks; N/P to move between comments. Draft comments are only viewable by you.
Jump to:
View side-by-side diff with in-line comments
Download patch
« media/audio/audio_output_device.h ('K') | « media/audio/audio_output_device.h ('k') | no next file » | no next file with comments »
Expand Comments ('e') | Collapse Comments ('c') | Show Comments Hide Comments ('s')
Index: media/audio/audio_output_device.cc
diff --git a/media/audio/audio_output_device.cc b/media/audio/audio_output_device.cc
index 7044e9ae770649c0477b9b37af56eca6fff01abb..58a67a748b9efa72015f8d69ba59acf3c7a97c5b 100644
--- a/media/audio/audio_output_device.cc
+++ b/media/audio/audio_output_device.cc
@@ -81,14 +81,21 @@ void AudioOutputDevice::Initialize(const AudioParameters& params,
RenderCallback* callback) {
DCHECK(!callback_) << "Calling Initialize() twice?";
Guido Urdaneta 2016/04/01 10:17:02 Why should Initialize by async? A common idiom is
Henrik Grunell 2016/04/04 08:08:59 Initialize() and Start() are serialized, so we're
DCHECK(params.IsValid());
- audio_parameters_ = params;
- callback_ = callback;
+ task_runner()->PostTask(
+ FROM_HERE, base::Bind(&AudioOutputDevice::InitializeOnIOThread, this,
+ params, callback));
}
AudioOutputDevice::~AudioOutputDevice() {
+ // I'm not sure if Stop() should remain mandatory. The purpose of that was
+ // to fix object lifetime issues. Probably we can do it in a cleaner way.
Guido Urdaneta 2016/04/01 10:17:01 The purpose of mandatory Stop() is to make sure yo
Henrik Grunell 2016/04/04 08:08:59 I wonder if we can post a task on the IO thread in
// The current design requires that the user calls Stop() before deleting
- // this class.
- DCHECK(audio_thread_.IsStopped());
+ // this class. Since this object is reference counted, we're the only one
+ // accessing data when we destruct.
+ DCHECK(state_ == IDLE || state_ == IPC_CLOSED);
+ // We need to ensure all the object lifetime stuff still works.
+ // Or even better, fix it to be more clear.
+ audio_thread_.Stop(base::MessageLoop::current());
}
void AudioOutputDevice::RequestDeviceAuthorization() {
@@ -100,20 +107,26 @@ void AudioOutputDevice::RequestDeviceAuthorization() {
void AudioOutputDevice::Start() {
DCHECK(callback_) << "Initialize hasn't been called";
- task_runner()->PostTask(FROM_HERE,
- base::Bind(&AudioOutputDevice::CreateStreamOnIOThread, this,
- audio_parameters_));
+ task_runner()->PostTask(
+ FROM_HERE, base::Bind(&AudioOutputDevice::CreateStreamOnIOThread, this));
}
void AudioOutputDevice::Stop() {
+ // TODO(olka): We need to clean this up; audio_thread_.Pause() must be called
+ // on IO thread only.
Henrik Grunell 2016/04/04 08:08:59 The problem here is that after Stop() returns, the
Henrik Grunell 2016/05/09 12:13:37 This has been addressed in the other CL.
{
base::AutoLock auto_lock(audio_thread_lock_);
- audio_thread_.Stop(base::MessageLoop::current());
+ // Play() is always called on IO thread. It means it can arrive after this
+ // Pause was executed, even if it was issued first. But we are safe, since
+ // Pause() will be called again in StopOnIOThread, so it's guaranteed to be
+ // called again after that sneaky Play(). What a hack.
+ // I'm not sure if we haven't broken this or about to break.
+ audio_thread_.Pause();
stopping_hack_ = true;
}
task_runner()->PostTask(FROM_HERE,
- base::Bind(&AudioOutputDevice::ShutDownOnIOThread, this));
+ base::Bind(&AudioOutputDevice::StopOnIOThread, this));
}
void AudioOutputDevice::Play() {
@@ -165,12 +178,29 @@ void AudioOutputDevice::RequestDeviceAuthorizationOnIOThread() {
DCHECK(task_runner()->BelongsToCurrentThread());
DCHECK_EQ(state_, IDLE);
state_ = AUTHORIZING;
+
+ // TODO(grunell): Store authorization so that we don't have to authorize
+ // again when re-starting. This involves mostly changes on the browser side:
+ // associate authorization by a new id (currently the stream id), don't forget
+ // authorization away after start or stop. Here we would have to explicitly
+ // make the browser side forget authorization in the dtor with a new
+ // AudioOutputIPC function.
ipc_->RequestDeviceAuthorization(this, session_id_, device_id_,
security_origin_);
}
-void AudioOutputDevice::CreateStreamOnIOThread(const AudioParameters& params) {
+void AudioOutputDevice::InitializeOnIOThread(const AudioParameters& params,
+ RenderCallback* callback) {
+ DCHECK(task_runner()->BelongsToCurrentThread());
+ DCHECK(!callback_);
+ audio_parameters_ = params;
+ callback_ = callback;
+}
+
+void AudioOutputDevice::CreateStreamOnIOThread() {
DCHECK(task_runner()->BelongsToCurrentThread());
+ DCHECK(callback_ || state_ == IPC_CLOSED);
+
switch (state_) {
case IPC_CLOSED:
if (callback_)
@@ -181,7 +211,7 @@ void AudioOutputDevice::CreateStreamOnIOThread(const AudioParameters& params) {
if (did_receive_auth_.IsSignaled() && device_id_.empty() &&
security_origin_.unique()) {
state_ = CREATING_STREAM;
- ipc_->CreateStream(this, params);
+ ipc_->CreateStream(this, audio_parameters_);
} else {
RequestDeviceAuthorizationOnIOThread();
start_on_authorized_ = true;
@@ -194,7 +224,7 @@ void AudioOutputDevice::CreateStreamOnIOThread(const AudioParameters& params) {
case AUTHORIZED:
state_ = CREATING_STREAM;
- ipc_->CreateStream(this, params);
+ ipc_->CreateStream(this, audio_parameters_);
start_on_authorized_ = false;
break;
@@ -230,7 +260,7 @@ void AudioOutputDevice::PauseOnIOThread() {
play_on_start_ = false;
}
-void AudioOutputDevice::ShutDownOnIOThread() {
+void AudioOutputDevice::StopOnIOThread() {
DCHECK(task_runner()->BelongsToCurrentThread());
// Close the stream, if we haven't already.
@@ -240,19 +270,16 @@ void AudioOutputDevice::ShutDownOnIOThread() {
}
start_on_authorized_ = false;
- // We can run into an issue where ShutDownOnIOThread is called right after
+ // We can run into an issue where StopOnIOThread is called right after
// OnStreamCreated is called in cases where Start/Stop are called before we
// get the OnStreamCreated callback. To handle that corner case, we call
// Stop(). In most cases, the thread will already be stopped.
Guido Urdaneta 2016/04/01 10:17:01 now you call Pause() instead of Stop()
Henrik Grunell 2016/05/09 12:13:37 Done in other CL.
- //
- // Another situation is when the IO thread goes away before Stop() is called
- // in which case, we cannot use the message loop to close the thread handle
- // and can't rely on the main thread existing either.
+
base::AutoLock auto_lock_(audio_thread_lock_);
- base::ThreadRestrictions::ScopedAllowIO allow_io;
- audio_thread_.Stop(NULL);
+ audio_thread_.Pause();
audio_callback_.reset();
stopping_hack_ = false;
+ callback_ = nullptr;
}
void AudioOutputDevice::SetVolumeOnIOThread(double volume) {
@@ -276,13 +303,8 @@ void AudioOutputDevice::OnStateChanged(AudioOutputIPCDelegateState state) {
break;
case AUDIO_OUTPUT_IPC_DELEGATE_STATE_ERROR:
DLOG(WARNING) << "AudioOutputDevice::OnStateChanged(ERROR)";
- // Don't dereference the callback object if the audio thread
- // is stopped or stopping. That could mean that the callback
- // object has been deleted.
- // TODO(tommi): Add an explicit contract for clearing the callback
- // object. Possibly require calling Initialize again or provide
- // a callback object via Start() and clear it in Stop().
- if (!audio_thread_.IsStopped())
+ // The callback object is cleared in Stop().
+ if (callback_)
callback_->OnRenderError();
break;
default:
@@ -315,7 +337,7 @@ void AudioOutputDevice::OnDeviceAuthorized(
did_receive_auth_.Signal();
}
if (start_on_authorized_)
- CreateStreamOnIOThread(audio_parameters_);
+ CreateStreamOnIOThread();
} else {
// Closing IPC forces a Signal(), so no clients are locked waiting
// indefinitely after this method returns.
@@ -343,9 +365,9 @@ void AudioOutputDevice::OnStreamCreated(
return;
// We can receive OnStreamCreated() on the IO thread after the client has
- // called Stop() but before ShutDownOnIOThread() is processed. In such a
+ // called Stop() but before StopOnIOThread() is processed. In such a
// situation |callback_| might point to freed memory. Instead of starting
- // |audio_thread_| do nothing and wait for ShutDownOnIOThread() to get called.
+ // |audio_thread_| do nothing and wait for StopOnIOThread() to get called.
//
// TODO(scherkus): The real fix is to have sane ownership semantics. The fact
// that |callback_| (which should own and outlive this object!) can point to
@@ -362,8 +384,13 @@ void AudioOutputDevice::OnStreamCreated(
DCHECK(audio_thread_.IsStopped());
audio_callback_.reset(new AudioOutputDevice::AudioThreadCallback(
audio_parameters_, handle, length, callback_));
- audio_thread_.Start(audio_callback_.get(), socket_handle,
- "AudioOutputDevice", true);
+
+ // If the audio thread hasn't been started before, start it.
+ if (audio_thread_.IsStopped()) {
+ audio_thread_.Start("AudioOutputDevice", true);
+ }
+ audio_thread_.Play(audio_callback_.get(), socket_handle);
+
state_ = PAUSED;
// We handle the case where Play() and/or Pause() may have been called
@@ -384,7 +411,7 @@ void AudioOutputDevice::OnIPCClosed() {
void AudioOutputDevice::WillDestroyCurrentMessageLoop() {
LOG(ERROR) << "IO loop going away before the audio device has been stopped";
- ShutDownOnIOThread();
+ StopOnIOThread();
}
// AudioOutputDevice::AudioThreadCallback
« media/audio/audio_output_device.h ('K') | « media/audio/audio_output_device.h ('k') | no next file » | no next file with comments »

Powered by Google App Engine
This is Rietveld 408576698