third_party/WebKit/Source/core/workers/WorkerBackingThread.h - Issue 2402983002: [TimeZoneMonitor] Decouple renderer side impl from content to blink.

Side by Side Diff: third_party/WebKit/Source/core/workers/WorkerBackingThread.h

Issue 2402983002: [TimeZoneMonitor] Decouple renderer side impl from content to blink. (Closed)

Patch Set: Rebase only Created 4 years, 2 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:

« no previous file with comments | « content/renderer/render_thread_impl.cc ('k') | third_party/WebKit/Source/core/workers/WorkerBackingThread.cpp » ('j') | third_party/WebKit/Source/core/workers/WorkerBackingThread.cpp » ('J')
Toggle Intra-line Diffs ('i') | Expand Comments ('e') | Collapse Comments ('c') | Hide Comments ('s')

OLD	NEW
1 // Copyright 2016 The Chromium Authors. All rights reserved.	1 // Copyright 2016 The Chromium Authors. All rights reserved.

2 // Use of this source code is governed by a BSD-style license that can be	2 // Use of this source code is governed by a BSD-style license that can be

3 // found in the LICENSE file.	3 // found in the LICENSE file.

4	4

5 #ifndef WorkerBackingThread_h	5 #ifndef WorkerBackingThread_h

6 #define WorkerBackingThread_h	6 #define WorkerBackingThread_h

7	7

8 #include "core/CoreExport.h"	8 #include "core/CoreExport.h"

9 #include "platform/heap/ThreadState.h"	9 #include "platform/heap/ThreadState.h"

10 #include "wtf/Forward.h"	10 #include "wtf/Forward.h"

(...skipping 51 matching lines...) Expand 10 before \| Expand all \| Expand 10 after Loading...
62 return *m_backingThread;	62 return *m_backingThread;

63 }	63 }

64	64

65 v8::Isolate* isolate() { return m_isolate; }	65 v8::Isolate* isolate() { return m_isolate; }

66	66

67 static void MemoryPressureNotificationToWorkerThreadIsolates(	67 static void MemoryPressureNotificationToWorkerThreadIsolates(

68 v8::MemoryPressureLevel);	68 v8::MemoryPressureLevel);

69	69

70 static void setRAILModeOnWorkerThreadIsolates(v8::RAILMode);	70 static void setRAILModeOnWorkerThreadIsolates(v8::RAILMode);

71	71

	72 static void PostTaskToAllThreads(const base::Closure& closure);
	blundell 2016/10/11 11:29:06 Do the WorkerBackingThreads correspond exactly to Do the WorkerBackingThreads correspond exactly to workers in content::PostTaskToAllWebWorkers()? blundell 2016/10/11 11:29:06 Could we reuse WorkerThread::workerThread() instea Could we reuse WorkerThread::workerThread() instead of creating parallel functionality? leonhsl(Using Gerrit) 2016/10/12 03:55:55 Seems multiple blink::WorkerThread correspond with Show quoted text On 2016/10/11 11:29:06, blundell wrote: > Could we reuse WorkerThread::workerThread() instead of creating parallel > functionality? Seems multiple blink::WorkerThread correspond with 1 blink::WorkerBackingThread, and content::WorkerThreadRegistry is tracking blink::WorkerBackingThread, so I'm focusing on the latter one. leonhsl(Using Gerrit) 2016/10/12 03:55:55 Yeah this has exactly the same effect with content Show quoted text On 2016/10/11 11:29:06, blundell wrote: > Do the WorkerBackingThreads correspond exactly to workers in > content::PostTaskToAllWebWorkers()? Yeah this has exactly the same effect with content::WorkerThreadRegistry::PostTaskToAllThreads(), which is posting task to all threads registered via blink::Platform API at https://cs.chromium.org/chromium/src/third_party/WebKit/Source/core/workers/W... . As we want to do the same work inside blink, I maintains the worker threads via a static blink::threadRunners(), maybe this is not proper way but I did not find out how to enumerate all worker threads inside blink. #I noticed haraken@ has helped asking nhiroki@ about this. nhiroki 2016/10/12 06:20:20 (This is not related to this CL and this looks lik Show quoted text On 2016/10/11 11:29:06, blundell wrote: > Do the WorkerBackingThreads correspond exactly to workers in > content::PostTaskToAllWebWorkers()? (This is not related to this CL and this looks like an issue of worker infra, but let me leave a comment for the future) Apparently, PostTaskToAllWebWorkers() doesn't post a task to WorkerBackingThreads when they don't own a platform thread. AnimationWorklet/AudioWorklet/CompositorWorker are such cases and these haven't been shipped yet, so this is not a real problem. We may need to fix it before shipping them. === Details === WorkerBackingThread creates and owns a dedicated platform thread for typical workers, but it doesn't create a thread and instead borrows an existing platform thread for some special workers (e.g. CompositorWorker is hosted on the existing compositor thread[1]). Note that these special workers have NOT been shipped yet. WorkerThreadRegistry::PostTaskToAllThreads() posts a task to \|task_runner_map_\|. This map is updated on WorkerThreadRegistry::DidStartCurrentWorkerThread()[2]. This is called from WorkerBackingThread::initialize() only when WorkerBackingThread owns a platform thread[3]. [1] https://cs.chromium.org/chromium/src/third_party/WebKit/Source/modules/compos... [2] https://cs.chromium.org/chromium/src/content/child/worker_thread_registry.cc?... [3] https://cs.chromium.org/chromium/src/third_party/WebKit/Source/core/workers/W... nhiroki 2016/10/12 06:32:46 Filed an issue: https://bugs.chromium.org/p/chromi Show quoted text On 2016/10/12 06:20:20, nhiroki wrote: > On 2016/10/11 11:29:06, blundell wrote: > > Do the WorkerBackingThreads correspond exactly to workers in > > content::PostTaskToAllWebWorkers()? > > (This is not related to this CL and this looks like an issue of worker infra, > but let me leave a comment for the future) > > Apparently, PostTaskToAllWebWorkers() doesn't post a task to > WorkerBackingThreads when they don't own a platform thread. > AnimationWorklet/AudioWorklet/CompositorWorker are such cases and these haven't > been shipped yet, so this is not a real problem. We may need to fix it before > shipping them. > > === Details === > > WorkerBackingThread creates and owns a dedicated platform thread for typical > workers, but it doesn't create a thread and instead borrows an existing platform > thread for some special workers (e.g. CompositorWorker is hosted on the existing > compositor thread[1]). Note that these special workers have NOT been shipped > yet. > > WorkerThreadRegistry::PostTaskToAllThreads() posts a task to \|task_runner_map_\|. > This map is updated on WorkerThreadRegistry::DidStartCurrentWorkerThread()[2]. > This is called from WorkerBackingThread::initialize() only when > WorkerBackingThread owns a platform thread[3]. > > [1] > https://cs.chromium.org/chromium/src/third_party/WebKit/Source/modules/compos... > [2] > https://cs.chromium.org/chromium/src/content/child/worker_thread_registry.cc?... > [3] > https://cs.chromium.org/chromium/src/third_party/WebKit/Source/core/workers/W... Filed an issue: https://bugs.chromium.org/p/chromium/issues/detail?id=655028
	73

72 private:	74 private:

73 WorkerBackingThread(const char* name,	75 WorkerBackingThread(const char* name,

74 bool shouldCallGCOnShutdown,	76 bool shouldCallGCOnShutdown,

75 BlinkGC::ThreadHeapMode);	77 BlinkGC::ThreadHeapMode);

76 WorkerBackingThread(WebThread*, bool shouldCallGCOnSHutdown);	78 WorkerBackingThread(WebThread*, bool shouldCallGCOnSHutdown);

77	79

78 std::unique_ptr<WebThreadSupportingGC> m_backingThread;	80 std::unique_ptr<WebThreadSupportingGC> m_backingThread;

79 v8::Isolate* m_isolate = nullptr;	81 v8::Isolate* m_isolate = nullptr;

80 bool m_isOwningThread;	82 bool m_isOwningThread;

81 bool m_shouldCallGCOnShutdown;	83 bool m_shouldCallGCOnShutdown;

82 };	84 };

83	85

84 } // namespace blink	86 } // namespace blink

85	87

86 #endif // WorkerBackingThread_h	88 #endif // WorkerBackingThread_h

OLD	NEW