1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
|
/* 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_dom_RemoteWorkerOp_h
#define mozilla_dom_RemoteWorkerOp_h
#include "mozilla/RefPtr.h"
#include "mozilla/Variant.h"
#include "mozilla/dom/WorkerPrivate.h"
namespace mozilla::dom {
class RemoteWorkerChild;
class RemtoeWorkerNonfLifeCycleOpControllerChild;
class RemoteWorkerOp;
// class RemoteWorkerNonLifeCycleOpControllerChild;
namespace remoteworker {
struct WorkerPrivateAccessibleState {
~WorkerPrivateAccessibleState();
RefPtr<WorkerPrivate> mWorkerPrivate;
};
// Initial state, mWorkerPrivate is initially null but will be initialized on
// the main thread by ExecWorkerOnMainThread when the WorkerPrivate is
// created. The state will transition to Running or Canceled, also from the
// main thread.
struct Pending : WorkerPrivateAccessibleState {
nsTArray<RefPtr<RemoteWorkerOp>> mPendingOps;
};
// Running, with the state transition happening on the main thread as a result
// of the worker successfully processing our initialization runnable,
// indicating that top-level script execution successfully completed. Because
// all of our state transitions happen on the main thread and are posed in
// terms of the main thread's perspective of the worker's state, it's very
// possible for us to skip directly from Pending to Canceled because we decide
// to cancel/terminate the worker prior to it finishing script loading or
// reporting back to us.
struct Running : WorkerPrivateAccessibleState {};
// Cancel() has been called on the WorkerPrivate on the main thread by a
// TerminationOp, top-level script evaluation has failed and canceled the
// worker, or in the case of a SharedWorker, close() has been called on
// the global scope by content code and the worker has advanced to the
// Canceling state. (Dedicated Workers can also self close, but they will
// never be RemoteWorkers. Although a SharedWorker can own DedicatedWorkers.)
// Browser shutdown will result in a TerminationOp thanks to use of a shutdown
// blocker in the parent, so the RuntimeService shouldn't get involved, but we
// would also handle that case acceptably too.
//
// Because worker self-closing is still handled by dispatching a runnable to
// the main thread to effectively call WorkerPrivate::Cancel(), there isn't
// a race between a worker deciding to self-close and our termination ops.
//
// In this state, we have dropped the reference to the WorkerPrivate and will
// no longer be dispatching runnables to the worker. We wait in this state
// until the termination lambda is invoked letting us know that the worker has
// entirely shutdown and we can advanced to the Killed state.
struct Canceled {};
// The worker termination lambda has been invoked and we know the Worker is
// entirely shutdown. (Inherently it is possible for us to advance to this
// state while the nsThread for the worker is still in the process of
// shutting down, but no more worker code will run on it.)
//
// This name is chosen to match the Worker's own state model.
struct Killed {};
using RemoteWorkerState = Variant<Pending, Running, Canceled, Killed>;
} // namespace remoteworker
class RemoteWorkerOp {
public:
NS_INLINE_DECL_PURE_VIRTUAL_REFCOUNTING
virtual ~RemoteWorkerOp() = default;
virtual bool MaybeStart(RemoteWorkerChild* aOwner,
remoteworker::RemoteWorkerState& aState) = 0;
virtual void StartOnMainThread(RefPtr<RemoteWorkerChild>& aOwner) = 0;
virtual void Start(RemoteWorkerNonLifeCycleOpControllerChild* aOwner,
remoteworker::RemoteWorkerState& aState) = 0;
virtual void Cancel() = 0;
};
} // namespace mozilla::dom
#endif
|
