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
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
base / synchronization / waitable_event_posix.cc [blame]
// Copyright 2012 The Chromium Authors
// Use of this source code is governed by a BSD-style license that can be
// found in the LICENSE file.
#ifdef UNSAFE_BUFFERS_BUILD
// TODO(crbug.com/40284755): Remove this and spanify to fix the errors.
#pragma allow_unsafe_buffers
#endif
#include "base/synchronization/waitable_event.h"
#include <stddef.h>
#include <limits>
#include <optional>
#include <vector>
#include "base/check_op.h"
#include "base/memory/stack_allocated.h"
#include "base/ranges/algorithm.h"
#include "base/synchronization/condition_variable.h"
#include "base/synchronization/lock.h"
#include "base/threading/scoped_blocking_call.h"
#include "base/threading/thread_restrictions.h"
#include "base/time/time.h"
#include "base/time/time_override.h"
// -----------------------------------------------------------------------------
// A WaitableEvent on POSIX is implemented as a wait-list. Currently we don't
// support cross-process events (where one process can signal an event which
// others are waiting on). Because of this, we can avoid having one thread per
// listener in several cases.
//
// The WaitableEvent maintains a list of waiters, protected by a lock. Each
// waiter is either an async wait, in which case we have a Task and the
// MessageLoop to run it on, or a blocking wait, in which case we have the
// condition variable to signal.
//
// Waiting involves grabbing the lock and adding oneself to the wait list. Async
// waits can be canceled, which means grabbing the lock and removing oneself
// from the list.
//
// Waiting on multiple events is handled by adding a single, synchronous wait to
// the wait-list of many events. An event passes a pointer to itself when
// firing a waiter and so we can store that pointer to find out which event
// triggered.
// -----------------------------------------------------------------------------
namespace base {
// -----------------------------------------------------------------------------
// This is just an abstract base class for waking the two types of waiters
// -----------------------------------------------------------------------------
WaitableEvent::WaitableEvent(ResetPolicy reset_policy,
InitialState initial_state)
: kernel_(new WaitableEventKernel(reset_policy, initial_state)) {}
void WaitableEvent::Reset() {
base::AutoLock locked(kernel_->lock_);
kernel_->signaled_ = false;
}
void WaitableEvent::SignalImpl() {
base::AutoLock locked(kernel_->lock_);
if (kernel_->signaled_)
return;
if (kernel_->manual_reset_) {
SignalAll();
kernel_->signaled_ = true;
} else {
// In the case of auto reset, if no waiters were woken, we remain
// signaled.
if (!SignalOne())
kernel_->signaled_ = true;
}
}
bool WaitableEvent::IsSignaled() const {
base::AutoLock locked(kernel_->lock_);
const bool result = kernel_->signaled_;
if (result && !kernel_->manual_reset_)
kernel_->signaled_ = false;
return result;
}
// -----------------------------------------------------------------------------
// Synchronous waits
// -----------------------------------------------------------------------------
// This is a synchronous waiter. The thread is waiting on the given condition
// variable and the fired flag in this object.
// -----------------------------------------------------------------------------
class SyncWaiter : public WaitableEvent::Waiter {
STACK_ALLOCATED();
public:
SyncWaiter()
: fired_(false), signaling_event_(nullptr), lock_(), cv_(&lock_) {}
bool Fire(WaitableEvent* signaling_event) override {
base::AutoLock locked(lock_);
if (fired_)
return false;
fired_ = true;
signaling_event_ = signaling_event;
cv_.Broadcast();
// Unlike AsyncWaiter objects, SyncWaiter objects are stack-allocated on
// the blocking thread's stack. There is no |delete this;| in Fire. The
// SyncWaiter object is destroyed when it goes out of scope.
return true;
}
WaitableEvent* signaling_event() const {
return signaling_event_;
}
// ---------------------------------------------------------------------------
// These waiters are always stack allocated and don't delete themselves. Thus
// there's no problem and the ABA tag is the same as the object pointer.
// ---------------------------------------------------------------------------
bool Compare(void* tag) override { return this == tag; }
// ---------------------------------------------------------------------------
// Called with lock held.
// ---------------------------------------------------------------------------
bool fired() const {
return fired_;
}
// ---------------------------------------------------------------------------
// During a TimedWait, we need a way to make sure that an auto-reset
// WaitableEvent doesn't think that this event has been signaled between
// unlocking it and removing it from the wait-list. Called with lock held.
// ---------------------------------------------------------------------------
void Disable() {
fired_ = true;
}
base::Lock* lock() {
return &lock_;
}
base::ConditionVariable* cv() {
return &cv_;
}
private:
bool fired_;
WaitableEvent* signaling_event_ = nullptr; // The WaitableEvent which woke us
base::Lock lock_;
base::ConditionVariable cv_;
};
bool WaitableEvent::TimedWaitImpl(TimeDelta wait_delta) {
kernel_->lock_.Acquire();
if (kernel_->signaled_) {
if (!kernel_->manual_reset_) {
// In this case we were signaled when we had no waiters. Now that
// someone has waited upon us, we can automatically reset.
kernel_->signaled_ = false;
}
kernel_->lock_.Release();
return true;
}
SyncWaiter sw;
if (only_used_while_idle_) {
sw.cv()->declare_only_used_while_idle();
}
sw.lock()->Acquire();
Enqueue(&sw);
kernel_->lock_.Release();
// We are violating locking order here by holding the SyncWaiter lock but not
// the WaitableEvent lock. However, this is safe because we don't lock |lock_|
// again before unlocking it.
// TimeTicks takes care of overflow but we special case is_max() nonetheless
// to avoid invoking TimeTicksNowIgnoringOverride() unnecessarily (same for
// the increment step of the for loop if the condition variable returns
// early). Ref: https://crbug.com/910524#c7
const TimeTicks end_time =
wait_delta.is_max() ? TimeTicks::Max()
: subtle::TimeTicksNowIgnoringOverride() + wait_delta;
for (TimeDelta remaining = wait_delta; remaining.is_positive() && !sw.fired();
remaining = end_time.is_max()
? TimeDelta::Max()
: end_time - subtle::TimeTicksNowIgnoringOverride()) {
if (end_time.is_max())
sw.cv()->Wait();
else
sw.cv()->TimedWait(remaining);
}
// Get the SyncWaiter signaled state before releasing the lock.
const bool return_value = sw.fired();
// We can't acquire |lock_| before releasing the SyncWaiter lock (because of
// locking order), however, in between the two a signal could be fired and
// |sw| would accept it, however we will still return false, so the signal
// would be lost on an auto-reset WaitableEvent. Thus we call Disable which
// makes sw::Fire return false.
sw.Disable();
sw.lock()->Release();
// This is a bug that has been enshrined in the interface of WaitableEvent
// now: |Dequeue| is called even when |sw.fired()| is true, even though it'll
// always return false in that case. However, taking the lock ensures that
// |Signal| has completed before we return and means that a WaitableEvent can
// synchronise its own destruction.
kernel_->lock_.Acquire();
kernel_->Dequeue(&sw, &sw);
kernel_->lock_.Release();
return return_value;
}
// -----------------------------------------------------------------------------
// Synchronous waiting on multiple objects.
static bool // StrictWeakOrdering
cmp_fst_addr(const std::pair<WaitableEvent*, unsigned> &a,
const std::pair<WaitableEvent*, unsigned> &b) {
return a.first < b.first;
}
// static
// NO_THREAD_SAFETY_ANALYSIS: Complex control flow.
size_t WaitableEvent::WaitManyImpl(WaitableEvent** raw_waitables,
size_t count) NO_THREAD_SAFETY_ANALYSIS {
// We need to acquire the locks in a globally consistent order. Thus we sort
// the array of waitables by address. We actually sort a pairs so that we can
// map back to the original index values later.
std::vector<std::pair<WaitableEvent*, size_t> > waitables;
waitables.reserve(count);
for (size_t i = 0; i < count; ++i)
waitables.push_back(std::make_pair(raw_waitables[i], i));
DCHECK_EQ(count, waitables.size());
ranges::sort(waitables, cmp_fst_addr);
// The set of waitables must be distinct. Since we have just sorted by
// address, we can check this cheaply by comparing pairs of consecutive
// elements.
for (size_t i = 0; i < waitables.size() - 1; ++i) {
DCHECK(waitables[i].first != waitables[i+1].first);
}
SyncWaiter sw;
const size_t r = EnqueueMany(&waitables[0], count, &sw);
if (r < count) {
// One of the events is already signaled. The SyncWaiter has not been
// enqueued anywhere.
return waitables[r].second;
}
// At this point, we hold the locks on all the WaitableEvents and we have
// enqueued our waiter in them all.
sw.lock()->Acquire();
// Release the WaitableEvent locks in the reverse order
for (size_t i = 0; i < count; ++i) {
waitables[count - (1 + i)].first->kernel_->lock_.Release();
}
for (;;) {
if (sw.fired())
break;
sw.cv()->Wait();
}
sw.lock()->Release();
// The address of the WaitableEvent which fired is stored in the SyncWaiter.
WaitableEvent *const signaled_event = sw.signaling_event();
// This will store the index of the raw_waitables which fired.
size_t signaled_index = 0;
// Take the locks of each WaitableEvent in turn (except the signaled one) and
// remove our SyncWaiter from the wait-list
for (size_t i = 0; i < count; ++i) {
if (raw_waitables[i] != signaled_event) {
raw_waitables[i]->kernel_->lock_.Acquire();
// There's no possible ABA issue with the address of the SyncWaiter here
// because it lives on the stack. Thus the tag value is just the pointer
// value again.
raw_waitables[i]->kernel_->Dequeue(&sw, &sw);
raw_waitables[i]->kernel_->lock_.Release();
} else {
// By taking this lock here we ensure that |Signal| has completed by the
// time we return, because |Signal| holds this lock. This matches the
// behaviour of |Wait| and |TimedWait|.
raw_waitables[i]->kernel_->lock_.Acquire();
raw_waitables[i]->kernel_->lock_.Release();
signaled_index = i;
}
}
return signaled_index;
}
// -----------------------------------------------------------------------------
// If return value == count:
// The locks of the WaitableEvents have been taken in order and the Waiter has
// been enqueued in the wait-list of each. None of the WaitableEvents are
// currently signaled
// else:
// None of the WaitableEvent locks are held. The Waiter has not been enqueued
// in any of them and the return value is the index of the WaitableEvent which
// was signaled with the lowest input index from the original WaitMany call.
// -----------------------------------------------------------------------------
// static
// NO_THREAD_SAFETY_ANALYSIS: Complex control flow.
size_t WaitableEvent::EnqueueMany(std::pair<WaitableEvent*, size_t>* waitables,
size_t count,
Waiter* waiter) NO_THREAD_SAFETY_ANALYSIS {
size_t winner = count;
size_t winner_index = count;
for (size_t i = 0; i < count; ++i) {
auto& kernel = waitables[i].first->kernel_;
kernel->lock_.Acquire();
if (kernel->signaled_ && waitables[i].second < winner) {
winner = waitables[i].second;
winner_index = i;
}
}
// No events signaled. All locks acquired. Enqueue the Waiter on all of them
// and return.
if (winner == count) {
for (size_t i = 0; i < count; ++i)
waitables[i].first->Enqueue(waiter);
return count;
}
// Unlock in reverse order and possibly clear the chosen winner's signal
// before returning its index.
for (auto* w = waitables + count - 1; w >= waitables; --w) {
auto& kernel = w->first->kernel_;
if (w->second == winner) {
if (!kernel->manual_reset_)
kernel->signaled_ = false;
}
kernel->lock_.Release();
}
return winner_index;
}
// -----------------------------------------------------------------------------
// -----------------------------------------------------------------------------
// Private functions...
WaitableEvent::WaitableEventKernel::WaitableEventKernel(
ResetPolicy reset_policy,
InitialState initial_state)
: manual_reset_(reset_policy == ResetPolicy::MANUAL),
signaled_(initial_state == InitialState::SIGNALED) {}
WaitableEvent::WaitableEventKernel::~WaitableEventKernel() = default;
// -----------------------------------------------------------------------------
// Wake all waiting waiters. Called with lock held.
// -----------------------------------------------------------------------------
bool WaitableEvent::SignalAll() {
bool signaled_at_least_one = false;
for (Waiter* i : kernel_->waiters_) {
if (i->Fire(this))
signaled_at_least_one = true;
}
kernel_->waiters_.clear();
return signaled_at_least_one;
}
// ---------------------------------------------------------------------------
// Try to wake a single waiter. Return true if one was woken. Called with lock
// held.
// ---------------------------------------------------------------------------
bool WaitableEvent::SignalOne() {
for (;;) {
if (kernel_->waiters_.empty())
return false;
const bool r = (*kernel_->waiters_.begin())->Fire(this);
kernel_->waiters_.pop_front();
if (r)
return true;
}
}
// -----------------------------------------------------------------------------
// Add a waiter to the list of those waiting. Called with lock held.
// -----------------------------------------------------------------------------
void WaitableEvent::Enqueue(Waiter* waiter) {
kernel_->waiters_.push_back(waiter);
}
// -----------------------------------------------------------------------------
// Remove a waiter from the list of those waiting. Return true if the waiter was
// actually removed. Called with lock held.
// -----------------------------------------------------------------------------
bool WaitableEvent::WaitableEventKernel::Dequeue(Waiter* waiter, void* tag) {
for (auto i = waiters_.begin(); i != waiters_.end(); ++i) {
if (*i == waiter && (*i)->Compare(tag)) {
waiters_.erase(i);
return true;
}
}
return false;
}
// -----------------------------------------------------------------------------
} // namespace base