Bug 1083101 - Add a task scheduler to Moz2D. r=jrmuizel

gfx/2d/JobScheduler.cpp

@@ -0,0 +1,236 @@
+/* -*- Mode: C++; tab-width: 20; indent-tabs-mode: nil; c-basic-offset: 2 -*-
+ * This Source Code Form is subject to the terms of the Mozilla Public
+ * License, v. 2.0. If a copy of the MPL was not distributed with this
+ * file, You can obtain one at http://mozilla.org/MPL/2.0/. */
+
+#include "JobScheduler.h"
+
+namespace mozilla {
+namespace gfx {
+
+JobScheduler* JobScheduler::sSingleton = nullptr;
+
+bool JobScheduler::Init(uint32_t aNumThreads, uint32_t aNumQueues)
+{
+  MOZ_ASSERT(!sSingleton);
+  MOZ_ASSERT(aNumThreads >= aNumQueues);
+
+  sSingleton = new JobScheduler();
+  sSingleton->mNextQueue = 0;
+
+  for (uint32_t i = 0; i < aNumQueues; ++i) {
+    sSingleton->mDrawingQueues.push_back(new MultiThreadedJobQueue());
+  }
+
+  for (uint32_t i = 0; i < aNumThreads; ++i) {
+    sSingleton->mWorkerThreads.push_back(new WorkerThread(sSingleton->mDrawingQueues[i%aNumQueues]));
+  }
+  return true;
+}
+
+void JobScheduler::ShutDown()
+{
+  MOZ_ASSERT(IsEnabled());
+  if (!IsEnabled()) {
+    return;
+  }
+
+  for (auto queue : sSingleton->mDrawingQueues) {
+    queue->ShutDown();
+    delete queue;
+  }
+
+  for (WorkerThread* thread : sSingleton->mWorkerThreads) {
+    // this will block until the thread is joined.
+    delete thread;
+  }
+
+  sSingleton->mWorkerThreads.clear();
+  delete sSingleton;
+  sSingleton = nullptr;
+}
+
+JobStatus
+JobScheduler::ProcessJob(Job* aJob)
+{
+  MOZ_ASSERT(aJob);
+  auto status = aJob->Run();
+  if (status == JobStatus::Error || status == JobStatus::Complete) {
+    delete aJob;
+  }
+  return status;
+}
+
+void
+JobScheduler::SubmitJob(Job* aJob)
+{
+  MOZ_ASSERT(aJob);
+  RefPtr<SyncObject> start = aJob->GetStartSync();
+  if (start && start->Register(aJob)) {
+    // The Job buffer starts with a non-signaled sync object, it
+    // is now registered in the list of task buffers waiting on the
+    // sync object, so we should not place it in the queue.
+    return;
+  }
+
+  GetQueueForJob(aJob)->SubmitJob(aJob);
+}
+
+MultiThreadedJobQueue*
+JobScheduler::GetQueueForJob(Job* aJob)
+{
+  return aJob->IsPinnedToAThread() ? aJob->GetWorkerThread()->GetJobQueue()
+                                    : GetDrawingQueue();
+}
+
+Job::Job(SyncObject* aStart, SyncObject* aCompletion, WorkerThread* aThread)
+: mStartSync(aStart)
+, mCompletionSync(aCompletion)
+, mPinToThread(aThread)
+{
+  if (mStartSync) {
+    mStartSync->AddSubsequent(this);
+  }
+  if (mCompletionSync) {
+    mCompletionSync->AddPrerequisite(this);
+  }
+}
+
+Job::~Job()
+{
+  if (mCompletionSync) {
+    //printf(" -- Job %p dtor completion %p\n", this, mCompletionSync);
+    mCompletionSync->Signal();
+    mCompletionSync = nullptr;
+  }
+}
+
+JobStatus
+SetEventJob::Run()
+{
+  mEvent->Set();
+  return JobStatus::Complete;
+}
+
+SetEventJob::SetEventJob(EventObject* aEvent,
+                           SyncObject* aStart, SyncObject* aCompletion,
+                           WorkerThread* aWorker)
+: Job(aStart, aCompletion, aWorker)
+, mEvent(aEvent)
+{}
+
+SetEventJob::~SetEventJob()
+{}
+
+SyncObject::SyncObject(uint32_t aNumPrerequisites)
+: mSignals(aNumPrerequisites)
+#ifdef DEBUG
+, mNumPrerequisites(aNumPrerequisites)
+, mAddedPrerequisites(0)
+#endif
+{}
+
+SyncObject::~SyncObject()
+{
+  MOZ_ASSERT(mWaitingJobs.size() == 0);
+}
+
+bool
+SyncObject::Register(Job* aJob)
+{
+  MOZ_ASSERT(aJob);
+
+  // For now, ensure that when we schedule the first subsequent, we have already
+  // created all of the prerequisites. This is an arbitrary restriction because
+  // we specify the number of prerequisites in the constructor, but in the typical
+  // scenario, if the assertion FreezePrerequisite blows up here it probably means
+  // we got the initial nmber of prerequisites wrong. We can decide to remove
+  // this restriction if needed.
+  FreezePrerequisites();
+
+  int32_t signals = mSignals;
+
+  if (signals > 0) {
+    AddWaitingJob(aJob);
+    // Since Register and Signal can be called concurrently, it can happen that
+    // reading mSignals in Register happens before decrementing mSignals in Signal,
+    // but SubmitWaitingJobs happens before AddWaitingJob. This ordering means
+    // the SyncObject ends up in the signaled state with a task sitting in the
+    // waiting list. To prevent that we check mSignals a second time and submit
+    // again if signals reached zero in the mean time.
+    // We do this instead of holding a mutex around mSignals+mJobs to reduce
+    // lock contention.
+    int32_t signals2 = mSignals;
+    if (signals2 == 0) {
+      SubmitWaitingJobs();
+    }
+    return true;
+  }
+
+  return false;
+}
+
+void
+SyncObject::Signal()
+{
+  int32_t signals = --mSignals;
+  MOZ_ASSERT(signals >= 0);
+
+  if (signals == 0) {
+    SubmitWaitingJobs();
+  }
+}
+
+void
+SyncObject::AddWaitingJob(Job* aJob)
+{
+  MutexAutoLock lock(&mMutex);
+  mWaitingJobs.push_back(aJob);
+}
+
+void SyncObject::SubmitWaitingJobs()
+{
+  std::vector<Job*> tasksToSubmit;
+  {
+    // Scheduling the tasks can cause code that modifies <this>'s reference
+    // count to run concurrently, and cause the caller of this function to
+    // be owned by another thread. We need to make sure the reference count
+    // does not reach 0 on another thread before mWaitingJobs.clear(), so
+    // hold a strong ref to prevent that!
+    RefPtr<SyncObject> kungFuDeathGrip(this);
+
+    MutexAutoLock lock(&mMutex);
+    tasksToSubmit = Move(mWaitingJobs);
+    mWaitingJobs.clear();
+  }
+
+  for (Job* task : tasksToSubmit) {
+    JobScheduler::GetQueueForJob(task)->SubmitJob(task);
+  }
+}
+
+bool
+SyncObject::IsSignaled()
+{
+  return mSignals == 0;
+}
+
+void
+SyncObject::FreezePrerequisites()
+{
+  MOZ_ASSERT(mAddedPrerequisites == mNumPrerequisites);
+}
+
+void
+SyncObject::AddPrerequisite(Job* aJob)
+{
+  MOZ_ASSERT(++mAddedPrerequisites <= mNumPrerequisites);
+}
+
+void
+SyncObject::AddSubsequent(Job* aJob)
+{
+}
+
+} //namespace
+} //namespace

gfx/2d/JobScheduler.h

@@ -0,0 +1,231 @@
+/* -*- Mode: C++; tab-width: 20; indent-tabs-mode: nil; c-basic-offset: 2 -*-
+ * This Source Code Form is subject to the terms of the Mozilla Public
+ * License, v. 2.0. If a copy of the MPL was not distributed with this
+ * file, You can obtain one at http://mozilla.org/MPL/2.0/. */
+
+#ifndef MOZILLA_GFX_TASKSCHEDULER_H_
+#define MOZILLA_GFX_TASKSCHEDULER_H_
+
+#include "mozilla/RefPtr.h"
+#include "mozilla/gfx/Types.h"
+
+#ifdef WIN32
+#include "mozilla/gfx/JobScheduler_win32.h"
+#else
+#include "mozilla/gfx/JobScheduler_posix.h"
+#endif
+
+namespace mozilla {
+namespace gfx {
+
+class MultiThreadedJobQueue;
+class SyncObject;
+
+class JobScheduler {
+public:
+  /// Return one of the queues that the drawing worker threads pull from, chosen
+  /// pseudo-randomly.
+  static MultiThreadedJobQueue* GetDrawingQueue()
+  {
+    return sSingleton->mDrawingQueues[
+      sSingleton->mNextQueue++ % sSingleton->mDrawingQueues.size()
+    ];
+  }
+
+  /// Return one of the queues that the drawing worker threads pull from with a
+  /// hash to choose the queue.
+  ///
+  /// Calling this function several times with the same hash will yield the same queue.
+  static MultiThreadedJobQueue* GetDrawingQueue(uint32_t aHash)
+  {
+    return sSingleton->mDrawingQueues[
+      aHash % sSingleton->mDrawingQueues.size()
+    ];
+  }
+
+  /// Return the task queue associated to the worker the task is pinned to if
+  /// the task is pinned to a worker, or a random queue.
+  static MultiThreadedJobQueue* GetQueueForJob(Job* aJob);
+
+  /// Initialize the task scheduler with aNumThreads worker threads for drawing
+  /// and aNumQueues task queues.
+  ///
+  /// The number of threads must be superior or equal to the number of queues
+  /// (since for now a worker thread only pulls from one queue).
+  static bool Init(uint32_t aNumThreads, uint32_t aNumQueues);
+
+  /// Shut the scheduler down.
+  ///
+  /// This will block until worker threads are joined and deleted.
+  static void ShutDown();
+
+  /// Returns true if there is a successfully initialized JobScheduler singleton.
+  static bool IsEnabled() { return !!sSingleton; }
+
+  /// Submit a task buffer to its associated queue.
+  ///
+  /// The caller looses ownership of the task buffer.
+  static void SubmitJob(Job* aJobs);
+
+  /// Process commands until the command buffer needs to block on a sync object,
+  /// completes, yields, or encounters an error.
+  ///
+  /// Can be used on any thread. Worker threads basically loop over this, but the
+  /// main thread can also dequeue pending task buffers and process them alongside
+  /// the worker threads if it is about to block until completion anyway.
+  ///
+  /// The caller looses ownership of the task buffer.
+  static JobStatus ProcessJob(Job* aJobs);
+
+protected:
+  static JobScheduler* sSingleton;
+
+  // queues of Job that are ready to be processed
+  std::vector<MultiThreadedJobQueue*> mDrawingQueues;
+  std::vector<WorkerThread*> mWorkerThreads;
+  Atomic<uint32_t> mNextQueue;
+};
+
+/// Jobs are not reference-counted because they don't have shared ownership.
+/// The ownership of tasks can change when they are passed to certain methods
+/// of JobScheduler and SyncObject. See the docuumentaion of these classes.
+class Job {
+public:
+  Job(SyncObject* aStart, SyncObject* aCompletion, WorkerThread* aThread = nullptr);
+
+  virtual ~Job();
+
+  virtual JobStatus Run() = 0;
+
+  /// For use in JobScheduler::SubmitJob. Don't use it anywhere else.
+  //already_AddRefed<SyncObject> GetAndResetStartSync();
+  SyncObject* GetStartSync() { return mStartSync; }
+
+  bool IsPinnedToAThread() const { return !!mPinToThread; }
+
+  WorkerThread* GetWorkerThread() { return mPinToThread; }
+
+protected:
+  RefPtr<SyncObject> mStartSync;
+  RefPtr<SyncObject> mCompletionSync;
+  WorkerThread* mPinToThread;
+};
+
+class EventObject;
+
+/// This task will set an EventObject.
+///
+/// Typically used as the final task, so that the main thread can block on the
+/// corresponfing EventObject until all of the tasks are processed.
+class SetEventJob : public Job
+{
+public:
+  explicit SetEventJob(EventObject* aEvent,
+                        SyncObject* aStart, SyncObject* aCompletion = nullptr,
+                        WorkerThread* aPinToWorker = nullptr);
+
+  ~SetEventJob();
+
+  JobStatus Run() override;
+
+  EventObject* GetEvent() { return mEvent; }
+
+protected:
+  RefPtr<EventObject> mEvent;
+};
+
+/// A synchronization object that can be used to express dependencies and ordering between
+/// tasks.
+///
+/// Jobs can register to SyncObjects in order to asynchronously wait for a signal.
+/// In practice, Job objects usually start with a sync object (startSyc) and end
+/// with another one (completionSync).
+/// a Job never gets processed before its startSync is in the signaled state, and
+/// signals its completionSync as soon as it finishes. This is how dependencies
+/// between tasks is expressed.
+class SyncObject final : public external::AtomicRefCounted<SyncObject> {
+public:
+  MOZ_DECLARE_REFCOUNTED_TYPENAME(SyncObject)
+
+  /// Create a synchronization object.
+  ///
+  /// aNumPrerequisites represents the number of times the object must be signaled
+  /// before actually entering the signaled state (in other words, it means the
+  /// number of dependencies of this sync object).
+  ///
+  /// Explicitly specifying the number of prerequisites when creating sync objects
+  /// makes it easy to start scheduling some of the prerequisite tasks while
+  /// creating the others, which is how we typically use the task scheduler.
+  /// Automatically determining the number of prerequisites using Job's constructor
+  /// brings the risk that the sync object enters the signaled state while we
+  /// are still adding prerequisites which is hard to fix without using muteces.
+  explicit SyncObject(uint32_t aNumPrerequisites = 1);
+
+  ~SyncObject();
+
+  /// Attempt to register a task.
+  ///
+  /// If the sync object is already in the signaled state, the buffer is *not*
+  /// registered and the sync object does not take ownership of the task.
+  /// If the object is not yet in the signaled state, it takes ownership of
+  /// the task and places it in a list of pending tasks.
+  /// Pending tasks will not be processed by the worker thread.
+  /// When the SyncObject reaches the signaled state, it places the pending
+  /// tasks back in the available buffer queue, so that they can be
+  /// scheduled again.
+  ///
+  /// Returns true if the SyncOject is not already in the signaled state.
+  /// This means that if this method returns true, the SyncObject has taken
+  /// ownership of the Job.
+  bool Register(Job* aJob);
+
+  /// Signal the SyncObject.
+  ///
+  /// This decrements an internal counter. The sync object reaches the signaled
+  /// state when the counter gets to zero.
+  void Signal();
+
+  /// Returns true if mSignals is equal to zero. In other words, returns true
+  /// if all prerequisite tasks have already signaled the sync object.
+  bool IsSignaled();
+
+  /// Asserts that the number of added prerequisites is equal to the number
+  /// specified in the constructor (does nothin in release builds).
+  void FreezePrerequisites();
+
+private:
+  // Called by Job's constructor
+  void AddSubsequent(Job* aJob);
+  void AddPrerequisite(Job* aJob);
+
+  void AddWaitingJob(Job* aJob);
+
+  void SubmitWaitingJobs();
+
+  std::vector<Job*> mWaitingJobs;
+  Mutex mMutex; // for concurrent access to mWaintingJobs
+  Atomic<int32_t> mSignals;
+
+#ifdef DEBUG
+  uint32_t mNumPrerequisites;
+  Atomic<uint32_t> mAddedPrerequisites;
+#endif
+
+  friend class Job;
+  friend class JobScheduler;
+};
+
+
+/// RAII helper.
+struct MutexAutoLock {
+    MutexAutoLock(Mutex* aMutex) : mMutex(aMutex) { mMutex->Lock(); }
+    ~MutexAutoLock() { mMutex->Unlock(); }
+protected:
+    Mutex* mMutex;
+};
+
+
+} // namespace
+} // namespace
+
+#endif

gfx/2d/JobScheduler_posix.cpp

@@ -0,0 +1,202 @@
+/* -*- Mode: C++; tab-width: 20; indent-tabs-mode: nil; c-basic-offset: 2 -*-
+ * This Source Code Form is subject to the terms of the Mozilla Public
+ * License, v. 2.0. If a copy of the MPL was not distributed with this
+ * file, You can obtain one at http://mozilla.org/MPL/2.0/. */
+
+#include "JobScheduler.h"
+#include "mozilla/gfx/Logging.h"
+
+using namespace std;
+
+namespace mozilla {
+namespace gfx {
+
+MultiThreadedJobQueue::MultiThreadedJobQueue()
+: mThreadsCount(0)
+, mShuttingDown(false)
+{}
+
+MultiThreadedJobQueue::~MultiThreadedJobQueue()
+{
+  MOZ_ASSERT(mJobs.empty());
+}
+
+bool
+MultiThreadedJobQueue::WaitForJob(Job*& aOutJob)
+{
+  return PopJob(aOutJob, BLOCKING);
+}
+
+bool
+MultiThreadedJobQueue::PopJob(Job*& aOutJobs, AccessType aAccess)
+{
+  for (;;) {
+    MutexAutoLock lock(&mMutex);
+
+    while (aAccess == BLOCKING && !mShuttingDown && mJobs.empty()) {
+      mAvailableCondvar.Wait(&mMutex);
+    }
+
+    if (mShuttingDown) {
+      return false;
+    }
+
+    if (mJobs.empty()) {
+      if (aAccess == NON_BLOCKING) {
+        return false;
+      }
+      continue;
+    }
+
+    Job* task = mJobs.front();
+    MOZ_ASSERT(task);
+
+    mJobs.pop_front();
+
+    aOutJobs = task;
+    return true;
+  }
+}
+
+void
+MultiThreadedJobQueue::SubmitJob(Job* aJobs)
+{
+  MOZ_ASSERT(aJobs);
+  MutexAutoLock lock(&mMutex);
+  mJobs.push_back(aJobs);
+  mAvailableCondvar.Broadcast();
+}
+
+size_t
+MultiThreadedJobQueue::NumJobs()
+{
+  MutexAutoLock lock(&mMutex);
+  return mJobs.size();
+}
+
+bool
+MultiThreadedJobQueue::IsEmpty()
+{
+  MutexAutoLock lock(&mMutex);
+  return mJobs.empty();
+}
+
+void
+MultiThreadedJobQueue::ShutDown()
+{
+  MutexAutoLock lock(&mMutex);
+  mShuttingDown = true;
+  while (mThreadsCount) {
+    mAvailableCondvar.Broadcast();
+    mShutdownCondvar.Wait(&mMutex);
+  }
+}
+
+void
+MultiThreadedJobQueue::RegisterThread()
+{
+  mThreadsCount += 1;
+}
+
+void
+MultiThreadedJobQueue::UnregisterThread()
+{
+  MutexAutoLock lock(&mMutex);
+  mThreadsCount -= 1;
+  if (mThreadsCount == 0) {
+    mShutdownCondvar.Broadcast();
+  }
+}
+
+void* ThreadCallback(void* threadData)
+{
+  WorkerThread* thread = (WorkerThread*)threadData;
+  thread->Run();
+  return nullptr;
+}
+
+WorkerThread::WorkerThread(MultiThreadedJobQueue* aJobQueue)
+: mQueue(aJobQueue)
+{
+  aJobQueue->RegisterThread();
+  pthread_create(&mThread, nullptr, ThreadCallback, this);
+}
+
+WorkerThread::~WorkerThread()
+{
+  pthread_join(mThread, nullptr);
+}
+
+void
+WorkerThread::SetName(const char* aName)
+{
+  // Call this from the thread itself because of Mac.
+#ifdef XP_MACOSX
+  pthread_setname_np(aName);
+#elif defined(__DragonFly__) || defined(__FreeBSD__) || defined(__OpenBSD__)
+  pthread_set_name_np(mThread, aName);
+#elif defined(__NetBSD__)
+  pthread_setname_np(mThread, "%s", (void*)aName);
+#else
+  pthread_setname_np(mThread, aName);
+#endif
+}
+
+void
+WorkerThread::Run()
+{
+  SetName("gfx worker");
+
+  for (;;) {
+    Job* commands = nullptr;
+    if (!mQueue->WaitForJob(commands)) {
+      mQueue->UnregisterThread();
+      return;
+    }
+
+    JobStatus status = JobScheduler::ProcessJob(commands);
+
+    if (status == JobStatus::Error) {
+      // Don't try to handle errors for now, but that's open to discussions.
+      // I expect errors to be mostly OOM issues.
+      MOZ_CRASH();
+    }
+  }
+}
+
+EventObject::EventObject()
+: mIsSet(false)
+{}
+
+EventObject::~EventObject()
+{}
+
+bool
+EventObject::Peak()
+{
+  MutexAutoLock lock(&mMutex);
+  return mIsSet;
+}
+
+void
+EventObject::Set()
+{
+  MutexAutoLock lock(&mMutex);
+  if (!mIsSet) {
+    mIsSet = true;
+    mCond.Broadcast();
+  }
+}
+
+void
+EventObject::Wait()
+{
+  MutexAutoLock lock(&mMutex);
+  if (mIsSet) {
+    return;
+  }
+  mCond.Wait(&mMutex);
+}
+
+} // namespce
+} // namespce

gfx/2d/JobScheduler_posix.h

@@ -0,0 +1,187 @@
+/* -*- Mode: C++; tab-width: 20; indent-tabs-mode: nil; c-basic-offset: 2 -*-
+ * This Source Code Form is subject to the terms of the Mozilla Public
+ * License, v. 2.0. If a copy of the MPL was not distributed with this
+ * file, You can obtain one at http://mozilla.org/MPL/2.0/. */
+
+#ifndef WIN32
+#ifndef MOZILLA_GFX_TASKSCHEDULER_POSIX_H_
+#define MOZILLA_GFX_TASKSCHEDULER_POSIX_H_
+
+#include <string>
+#include <vector>
+#include <list>
+#include <pthread.h>
+#include <stdint.h>
+#include <stdio.h>
+
+#include "mozilla/RefPtr.h"
+#include "mozilla/DebugOnly.h"
+
+namespace mozilla {
+namespace gfx {
+
+class Job;
+class PosixCondVar;
+
+class Mutex {
+public:
+  Mutex() {
+    DebugOnly<int> err = pthread_mutex_init(&mMutex, nullptr);
+    MOZ_ASSERT(!err);
+  }
+
+  ~Mutex() {
+    DebugOnly<int> err = pthread_mutex_destroy(&mMutex);
+    MOZ_ASSERT(!err);
+  }
+
+  void Lock() {
+    DebugOnly<int> err = pthread_mutex_lock(&mMutex);
+    MOZ_ASSERT(!err);
+  }
+
+  void Unlock() {
+    DebugOnly<int> err = pthread_mutex_unlock(&mMutex);
+    MOZ_ASSERT(!err);
+  }
+
+protected:
+  pthread_mutex_t mMutex;
+  friend class PosixCondVar;
+};
+
+// posix platforms only!
+class PosixCondVar {
+public:
+  PosixCondVar() {
+    DebugOnly<int> err = pthread_cond_init(&mCond, nullptr);
+    MOZ_ASSERT(!err);
+  }
+
+  ~PosixCondVar() {
+    DebugOnly<int> err = pthread_cond_destroy(&mCond);
+    MOZ_ASSERT(!err);
+  }
+
+  void Wait(Mutex* aMutex) {
+    DebugOnly<int> err = pthread_cond_wait(&mCond, &aMutex->mMutex);
+    MOZ_ASSERT(!err);
+  }
+
+  void Broadcast() {
+    DebugOnly<int> err = pthread_cond_broadcast(&mCond);
+    MOZ_ASSERT(!err);
+  }
+
+protected:
+  pthread_cond_t mCond;
+};
+
+
+/// A simple and naive multithreaded task queue
+///
+/// The public interface of this class must remain identical to its equivalent
+/// in JobScheduler_win32.h
+class MultiThreadedJobQueue {
+public:
+  enum AccessType {
+    BLOCKING,
+    NON_BLOCKING
+  };
+
+  // Producer thread
+  MultiThreadedJobQueue();
+
+  // Producer thread
+  ~MultiThreadedJobQueue();
+
+  // Worker threads
+  bool WaitForJob(Job*& aOutJob);
+
+  // Any thread
+  bool PopJob(Job*& aOutJob, AccessType aAccess);
+
+  // Any threads
+  void SubmitJob(Job* aJob);
+
+  // Producer thread
+  void ShutDown();
+
+  // Any thread
+  size_t NumJobs();
+
+  // Any thread
+  bool IsEmpty();
+
+  // Producer thread
+  void RegisterThread();
+
+  // Worker threads
+  void UnregisterThread();
+
+protected:
+
+  std::list<Job*> mJobs;
+  Mutex mMutex;
+  PosixCondVar mAvailableCondvar;
+  PosixCondVar mShutdownCondvar;
+  int32_t mThreadsCount;
+  bool mShuttingDown;
+
+  friend class WorkerThread;
+};
+
+/// Worker thread that continuously dequeues Jobs from a MultiThreadedJobQueue
+/// and process them.
+///
+/// The public interface of this class must remain identical to its equivalent
+/// in JobScheduler_win32.h
+class WorkerThread {
+public:
+  explicit WorkerThread(MultiThreadedJobQueue* aJobQueue);
+
+  ~WorkerThread();
+
+  void Run();
+
+  MultiThreadedJobQueue* GetJobQueue() { return mQueue; }
+protected:
+  void SetName(const char* name);
+
+  MultiThreadedJobQueue* mQueue;
+  pthread_t mThread;
+};
+
+/// An object that a thread can synchronously wait on.
+/// Usually set by a SetEventJob.
+class EventObject : public external::AtomicRefCounted<EventObject>
+{
+public:
+  MOZ_DECLARE_REFCOUNTED_TYPENAME(EventObject)
+
+  EventObject();
+
+  ~EventObject();
+
+  /// Synchronously wait until the event is set.
+  void Wait();
+
+  /// Return true if the event is set, without blocking.
+  bool Peak();
+
+  /// Set the event.
+  void Set();
+
+protected:
+  Mutex mMutex;
+  PosixCondVar mCond;
+  bool mIsSet;
+};
+
+} // namespace
+} // namespace
+
+#include "JobScheduler.h"
+
+#endif
+#endif

gfx/2d/JobScheduler_win32.h

@@ -0,0 +1,76 @@
+/* -*- Mode: C++; tab-width: 20; indent-tabs-mode: nil; c-basic-offset: 2 -*-
+ * This Source Code Form is subject to the terms of the Mozilla Public
+ * License, v. 2.0. If a copy of the MPL was not distributed with this
+ * file, You can obtain one at http://mozilla.org/MPL/2.0/. */
+
+#ifdef WIN32
+#ifndef MOZILLA_GFX_TASKSCHEDULER_WIN32_H_
+#define MOZILLA_GFX_TASKSCHEDULER_WIN32_H_
+
+#define NOT_IMPLEMENTED MOZ_CRASH("Not implemented")
+
+#include "mozilla/RefPtr.h"
+
+namespace mozilla {
+namespace gfx {
+
+class WorkerThread;
+class Job;
+
+class Mutex {
+public:
+  Mutex() { NOT_IMPLEMENTED; }
+  ~Mutex() { NOT_IMPLEMENTED; }
+  void Lock() { NOT_IMPLEMENTED; }
+  void Unlock() { NOT_IMPLEMENTED; }
+};
+
+// The public interface of this class must remain identical to its equivalent
+// in JobScheduler_posix.h
+class MultiThreadedJobQueue {
+public:
+  enum AccessType {
+    BLOCKING,
+    NON_BLOCKING
+  };
+
+  bool WaitForJob(Job*& aOutCommands) { NOT_IMPLEMENTED; }
+  bool PopJob(Job*& aOutCommands, AccessType aAccess) { NOT_IMPLEMENTED; }
+  void SubmitJob(Job* aCommands) { NOT_IMPLEMENTED; }
+  void ShutDown() { NOT_IMPLEMENTED; }
+  size_t NumJobs() { NOT_IMPLEMENTED;  }
+  bool IsEmpty() { NOT_IMPLEMENTED; }
+  void RegisterThread() { NOT_IMPLEMENTED; }
+  void UnregisterThread() { NOT_IMPLEMENTED; }
+
+  friend class WorkerThread;
+};
+
+
+// The public interface of this class must remain identical to its equivalent
+// in JobScheduler_posix.h
+class EventObject : public external::AtomicRefCounted<EventObject>
+{
+public:
+  MOZ_DECLARE_REFCOUNTED_TYPENAME(EventObject)
+
+  EventObject() { NOT_IMPLEMENTED; }
+  ~EventObject() { NOT_IMPLEMENTED; }
+  void Wait() { NOT_IMPLEMENTED; }
+  bool Peak() { NOT_IMPLEMENTED; }
+  void Set() { NOT_IMPLEMENTED; }
+};
+
+// The public interface of this class must remain identical to its equivalent
+// in JobScheduler_posix.h
+class WorkerThread {
+public:
+  explicit WorkerThread(MultiThreadedJobQueue* aJobQueue) { NOT_IMPLEMENTED; }
+  void Run();
+};
+
+} // namespace
+} // namespace
+
+#endif
+#endif

gfx/2d/Types.h

@@ -289,6 +289,13 @@ struct GradientStop
   Color color;
 };
 
+enum class JobStatus {
+    Complete,
+    Wait,
+    Yield,
+    Error
+};
+
 } // namespace gfx
 } // namespace mozilla
 

gfx/2d/moz.build

@@ -26,6 +26,9 @@ EXPORTS.mozilla.gfx += [
     'Helpers.h',
     'HelpersCairo.h',
     'IterableArena.h',
+    'JobScheduler.h',
+    'JobScheduler_posix.h',
+    'JobScheduler_win32.h',
     'Logging.h',
     'Matrix.h',
     'NumericTools.h',
@@ -71,6 +74,11 @@ elif CONFIG['MOZ_WIDGET_TOOLKIT'] == 'windows':
     ]
     DEFINES['WIN32'] = True
 
+if CONFIG['MOZ_WIDGET_TOOLKIT'] != 'windows':
+    SOURCES += [
+        'JobScheduler_posix.cpp',
+    ]
+
 if CONFIG['MOZ_ENABLE_SKIA']:
     UNIFIED_SOURCES += [
         'convolver.cpp',
@@ -130,6 +138,7 @@ UNIFIED_SOURCES += [
     'FilterProcessing.cpp',
     'FilterProcessingScalar.cpp',
     'ImageScaling.cpp',
+    'JobScheduler.cpp',
     'Matrix.cpp',
     'Path.cpp',
     'PathCairo.cpp',

gfx/tests/gtest/TestJobScheduler.cpp

@@ -0,0 +1,246 @@
+/* vim:set ts=2 sw=2 sts=2 et: */
+/* Any copyright is dedicated to the Public Domain.
+ * http://creativecommons.org/publicdomain/zero/1.0/
+ */
+
+#ifndef WIN32
+
+#include "gtest/gtest.h"
+#include "gmock/gmock.h"
+
+#include "mozilla/gfx/JobScheduler.h"
+
+#include <pthread.h>
+#include <sched.h>
+#include <stdlib.h>
+#include <time.h>
+
+namespace test_scheduler {
+
+using namespace mozilla::gfx;
+using namespace mozilla;
+
+// Artificially cause threads to yield randomly in an attempt to make racy
+// things more apparent (if any).
+void MaybeYieldThread()
+{
+  if (rand() % 5 == 0) {
+    sched_yield();
+  }
+}
+
+/// Used by the TestCommand to check that tasks are processed in the right order.
+struct SanityChecker {
+  std::vector<uint64_t> mAdvancements;
+  mozilla::gfx::Mutex mMutex;
+
+  explicit SanityChecker(uint64_t aNumCmdBuffers)
+  {
+    for (uint32_t i = 0; i < aNumCmdBuffers; ++i) {
+      mAdvancements.push_back(0);
+    }
+  }
+
+  virtual void Check(uint64_t aJobId, uint64_t aCmdId)
+  {
+    MaybeYieldThread();
+    MutexAutoLock lock(&mMutex);
+    ASSERT_EQ(mAdvancements[aJobId], aCmdId-1);
+    mAdvancements[aJobId] = aCmdId;
+  }
+};
+
+/// Run checks that are specific to TestSchulerJoin.
+struct JoinTestSanityCheck : public SanityChecker {
+  bool mSpecialJobHasRun;
+
+  explicit JoinTestSanityCheck(uint64_t aNumCmdBuffers)
+  : SanityChecker(aNumCmdBuffers)
+  , mSpecialJobHasRun(false)
+  {}
+
+  virtual void Check(uint64_t aJobId, uint64_t aCmdId) override
+  {
+    // Job 0 is the special task executed when everything is joined after task 1
+    if (aCmdId == 0) {
+      ASSERT_FALSE(mSpecialJobHasRun);
+      mSpecialJobHasRun = true;
+      for (auto advancement : mAdvancements) {
+        // Because of the synchronization point (beforeFilter), all
+        // task buffers should have run task 1 when task 0 is run.
+        ASSERT_EQ(advancement, (uint32_t)1);
+      }
+    } else {
+      // This check does not apply to task 0.
+      SanityChecker::Check(aJobId, aCmdId);
+    }
+
+    if (aCmdId == 2) {
+      ASSERT_TRUE(mSpecialJobHasRun);
+    }
+  }
+};
+
+class TestJob : public Job
+{
+public:
+  TestJob(uint64_t aCmdId, uint64_t aJobId, SanityChecker* aChecker,
+           SyncObject* aStart, SyncObject* aCompletion)
+  : Job(aStart, aCompletion, nullptr)
+  , mCmdId(aCmdId)
+  , mCmdBufferId(aJobId)
+  , mSanityChecker(aChecker)
+  {}
+
+  JobStatus Run()
+  {
+    MaybeYieldThread();
+    mSanityChecker->Check(mCmdBufferId, mCmdId);
+    MaybeYieldThread();
+    return JobStatus::Complete;
+  }
+
+  uint64_t mCmdId;
+  uint64_t mCmdBufferId;
+  SanityChecker* mSanityChecker;
+};
+
+/// This test creates aNumCmdBuffers task buffers with sync objects set up
+/// so that all tasks will join after command 5 before a task buffer runs
+/// a special task (task 0) after which all task buffers fork again.
+/// This simulates the kind of scenario where all tiles must join at
+/// a certain point to execute, say, a filter, and fork again after the filter
+/// has been processed.
+/// The main thread is only blocked when waiting for the completion of the entire
+/// task stream (it doesn't have to wait at the filter's sync points to orchestrate it).
+void TestSchedulerJoin(uint32_t aNumThreads, uint32_t aNumCmdBuffers)
+{
+  JoinTestSanityCheck check(aNumCmdBuffers);
+
+  RefPtr<SyncObject> beforeFilter = new SyncObject(aNumCmdBuffers);
+  RefPtr<SyncObject> afterFilter = new SyncObject();
+  RefPtr<SyncObject> completion = new SyncObject(aNumCmdBuffers);
+
+
+  for (uint32_t i = 0; i < aNumCmdBuffers; ++i) {
+    Job* t1 = new TestJob(1, i, &check, nullptr, beforeFilter);
+    JobScheduler::SubmitJob(t1);
+    MaybeYieldThread();
+  }
+  beforeFilter->FreezePrerequisites();
+
+  // This task buffer is executed when all other tasks have joined after task 1
+  JobScheduler::SubmitJob(
+    new TestJob(0, 0, &check, beforeFilter, afterFilter)
+  );
+  afterFilter->FreezePrerequisites();
+
+  for (uint32_t i = 0; i < aNumCmdBuffers; ++i) {
+    Job* t2 = new TestJob(2, i, &check, afterFilter, completion);
+    JobScheduler::SubmitJob(t2);
+    MaybeYieldThread();
+  }
+  completion->FreezePrerequisites();
+
+  RefPtr<EventObject> waitForCompletion = new EventObject();
+  auto evtJob = new SetEventJob(waitForCompletion, completion);
+  JobScheduler::SubmitJob(evtJob);
+
+  MaybeYieldThread();
+
+  waitForCompletion->Wait();
+
+  MaybeYieldThread();
+
+  for (auto advancement : check.mAdvancements) {
+    ASSERT_TRUE(advancement == 2);
+  }
+}
+
+/// This test creates several chains of 10 task, tasks of a given chain are executed
+/// sequentially, and chains are exectuted in parallel.
+/// This simulates the typical scenario where we want to process sequences of drawing
+/// commands for several tiles in parallel.
+void TestSchedulerChain(uint32_t aNumThreads, uint32_t aNumCmdBuffers)
+{
+  SanityChecker check(aNumCmdBuffers);
+
+  RefPtr<SyncObject> completion = new SyncObject(aNumCmdBuffers);
+
+  uint32_t numJobs = 10;
+
+  for (uint32_t i = 0; i < aNumCmdBuffers; ++i) {
+
+    std::vector<RefPtr<SyncObject>> syncs;
+    std::vector<Job*> tasks;
+    syncs.reserve(numJobs);
+    tasks.reserve(numJobs);
+
+    for (uint32_t t = 0; t < numJobs-1; ++t) {
+      syncs.push_back(new SyncObject());
+      tasks.push_back(new TestJob(t+1, i, &check, t == 0 ? nullptr
+                                                          : syncs[t-1].get(),
+                                   syncs[t]));
+      syncs.back()->FreezePrerequisites();
+    }
+
+    tasks.push_back(new TestJob(numJobs, i, &check, syncs.back(), completion));
+
+    if (i % 2 == 0) {
+      // submit half of the tasks in order
+      for (Job* task : tasks) {
+        JobScheduler::SubmitJob(task);
+        MaybeYieldThread();
+      }
+    } else {
+      // ... and submit the other half in reverse order
+      for (int32_t reverse = numJobs-1; reverse >= 0; --reverse) {
+        JobScheduler::SubmitJob(tasks[reverse]);
+        MaybeYieldThread();
+      }
+    }
+  }
+  completion->FreezePrerequisites();
+
+  RefPtr<EventObject> waitForCompletion = new EventObject();
+  auto evtJob = new SetEventJob(waitForCompletion, completion);
+  JobScheduler::SubmitJob(evtJob);
+
+  MaybeYieldThread();
+
+  waitForCompletion->Wait();
+
+  for (auto advancement : check.mAdvancements) {
+    ASSERT_TRUE(advancement == numJobs);
+  }
+}
+
+} // namespace test_scheduler
+
+TEST(Moz2D, JobScheduler_Join) {
+  srand(time(nullptr));
+  for (uint32_t threads = 1; threads < 8; ++threads) {
+    for (uint32_t queues = 1; queues < threads; ++queues) {
+      for (uint32_t buffers = 1; buffers < 100; buffers += 3) {
+        mozilla::gfx::JobScheduler::Init(threads, queues);
+        test_scheduler::TestSchedulerJoin(threads, buffers);
+        mozilla::gfx::JobScheduler::ShutDown();
+      }
+    }
+  }
+}
+
+TEST(Moz2D, JobScheduler_Chain) {
+  srand(time(nullptr));
+  for (uint32_t threads = 1; threads < 8; ++threads) {
+    for (uint32_t queues = 1; queues < threads; ++queues) {
+      for (uint32_t buffers = 1; buffers < 100; buffers += 3) {
+        mozilla::gfx::JobScheduler::Init(threads, queues);
+        test_scheduler::TestSchedulerChain(threads, buffers);
+        mozilla::gfx::JobScheduler::ShutDown();
+      }
+    }
+  }
+}
+
+#endif

gfx/tests/gtest/moz.build

@@ -14,6 +14,7 @@ UNIFIED_SOURCES += [
     'TestCompositor.cpp',
     'TestGfxPrefs.cpp',
     'TestGfxWidgets.cpp',
+    'TestJobScheduler.cpp',
     'TestLayers.cpp',
     'TestMoz2D.cpp',
     'TestQcms.cpp',