OLD | NEW |
---|---|
1 // Copyright (c) 2012 The Chromium Authors. All rights reserved. | 1 // Copyright (c) 2012 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 #include "base/tracked_objects.h" | 5 #include "base/tracked_objects.h" |
6 | 6 |
7 #include <limits.h> | 7 #include <limits.h> |
8 #include <stdlib.h> | 8 #include <stdlib.h> |
9 | 9 |
10 #include "base/atomicops.h" | 10 #include "base/atomicops.h" |
11 #include "base/base_switches.h" | 11 #include "base/base_switches.h" |
12 #include "base/command_line.h" | 12 #include "base/command_line.h" |
13 #include "base/compiler_specific.h" | 13 #include "base/compiler_specific.h" |
14 #include "base/debug/leak_annotations.h" | 14 #include "base/debug/leak_annotations.h" |
15 #include "base/logging.h" | 15 #include "base/logging.h" |
16 #include "base/process/process_handle.h" | 16 #include "base/process/process_handle.h" |
17 #include "base/profiler/alternate_timer.h" | 17 #include "base/profiler/alternate_timer.h" |
18 #include "base/stl_util.h" | |
18 #include "base/strings/stringprintf.h" | 19 #include "base/strings/stringprintf.h" |
19 #include "base/third_party/valgrind/memcheck.h" | 20 #include "base/third_party/valgrind/memcheck.h" |
20 #include "base/tracking_info.h" | 21 #include "base/tracking_info.h" |
21 | 22 |
22 using base::TimeDelta; | 23 using base::TimeDelta; |
23 | 24 |
24 namespace base { | 25 namespace base { |
25 class TimeDelta; | 26 class TimeDelta; |
26 } | 27 } |
27 | 28 |
28 namespace tracked_objects { | 29 namespace tracked_objects { |
29 | 30 |
30 namespace { | 31 namespace { |
31 // Flag to compile out almost all of the task tracking code. | 32 // Flag to compile out almost all of the task tracking code. |
32 const bool kTrackAllTaskObjects = true; | 33 const bool kTrackAllTaskObjects = true; |
Ilya Sherman
2015/04/07 01:15:30
Is this flag still needed? It might be helpful to
vadimt
2015/04/07 21:44:14
Done; plase see (and LGTM) https://codereview.chro
| |
33 | 34 |
34 // TODO(jar): Evaluate the perf impact of enabling this. If the perf impact is | 35 // TODO(jar): Evaluate the perf impact of enabling this. If the perf impact is |
35 // negligible, enable by default. | 36 // negligible, enable by default. |
36 // Flag to compile out parent-child link recording. | 37 // Flag to compile out parent-child link recording. |
37 const bool kTrackParentChildLinks = false; | 38 const bool kTrackParentChildLinks = false; |
38 | 39 |
39 // When ThreadData is first initialized, should we start in an ACTIVE state to | 40 // When ThreadData is first initialized, should we start in an ACTIVE state to |
40 // record all of the startup-time tasks, or should we start up DEACTIVATED, so | 41 // record all of the startup-time tasks, or should we start up DEACTIVATED, so |
41 // that we only record after parsing the command line flag --enable-tracking. | 42 // that we only record after parsing the command line flag --enable-tracking. |
42 // Note that the flag may force either state, so this really controls only the | 43 // Note that the flag may force either state, so this really controls only the |
(...skipping 46 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... | |
89 } | 90 } |
90 return current_timing_enabled == ENABLED_TIMING; | 91 return current_timing_enabled == ENABLED_TIMING; |
91 } | 92 } |
92 | 93 |
93 } // namespace | 94 } // namespace |
94 | 95 |
95 //------------------------------------------------------------------------------ | 96 //------------------------------------------------------------------------------ |
96 // DeathData tallies durations when a death takes place. | 97 // DeathData tallies durations when a death takes place. |
97 | 98 |
98 DeathData::DeathData() { | 99 DeathData::DeathData() { |
99 Clear(); | 100 count_ = 0; |
101 sample_probability_count_ = 0; | |
102 run_duration_sum_ = 0; | |
103 run_duration_max_ = 0; | |
104 run_duration_sample_ = 0; | |
105 queue_duration_sum_ = 0; | |
106 queue_duration_max_ = 0; | |
107 queue_duration_sample_ = 0; | |
108 last_phase_snapshot_ = nullptr; | |
100 } | 109 } |
101 | 110 |
102 DeathData::DeathData(int count) { | 111 DeathData::DeathData(int count) : DeathData() { |
103 Clear(); | |
104 count_ = count; | 112 count_ = count; |
105 } | 113 } |
106 | 114 |
115 DeathData::~DeathData() { | |
116 while (last_phase_snapshot_) { | |
117 DeathDataPhaseSnapshot* snapshot = last_phase_snapshot_; | |
118 last_phase_snapshot_ = snapshot->prev; | |
119 delete snapshot; | |
120 } | |
121 } | |
122 | |
107 // TODO(jar): I need to see if this macro to optimize branching is worth using. | 123 // TODO(jar): I need to see if this macro to optimize branching is worth using. |
108 // | 124 // |
109 // This macro has no branching, so it is surely fast, and is equivalent to: | 125 // This macro has no branching, so it is surely fast, and is equivalent to: |
110 // if (assign_it) | 126 // if (assign_it) |
111 // target = source; | 127 // target = source; |
112 // We use a macro rather than a template to force this to inline. | 128 // We use a macro rather than a template to force this to inline. |
113 // Related code for calculating max is discussed on the web. | 129 // Related code for calculating max is discussed on the web. |
114 #define CONDITIONAL_ASSIGN(assign_it, target, source) \ | 130 #define CONDITIONAL_ASSIGN(assign_it, target, source) \ |
115 ((target) ^= ((target) ^ (source)) & -static_cast<int32>(assign_it)) | 131 ((target) ^= ((target) ^ (source)) & -static_cast<int32>(assign_it)) |
116 | 132 |
117 void DeathData::RecordDeath(const int32 queue_duration, | 133 void DeathData::RecordDeath(const int32 queue_duration, |
118 const int32 run_duration, | 134 const int32 run_duration, |
119 const uint32 random_number) { | 135 const uint32 random_number) { |
120 // We'll just clamp at INT_MAX, but we should note this in the UI as such. | 136 // We'll just clamp at INT_MAX, but we should note this in the UI as such. |
121 if (count_ < INT_MAX) | 137 if (count_ < INT_MAX) |
122 ++count_; | 138 ++count_; |
139 if (sample_probability_count_ < INT_MAX) | |
140 ++sample_probability_count_; | |
123 queue_duration_sum_ += queue_duration; | 141 queue_duration_sum_ += queue_duration; |
124 run_duration_sum_ += run_duration; | 142 run_duration_sum_ += run_duration; |
125 | 143 |
126 if (queue_duration_max_ < queue_duration) | 144 if (queue_duration_max_ < queue_duration) |
127 queue_duration_max_ = queue_duration; | 145 queue_duration_max_ = queue_duration; |
128 if (run_duration_max_ < run_duration) | 146 if (run_duration_max_ < run_duration) |
129 run_duration_max_ = run_duration; | 147 run_duration_max_ = run_duration; |
130 | 148 |
131 // Take a uniformly distributed sample over all durations ever supplied. | 149 // Take a uniformly distributed sample over all durations ever supplied during |
132 // The probability that we (instead) use this new sample is 1/count_. This | 150 // currrent profiling phase. |
133 // results in a completely uniform selection of the sample (at least when we | 151 // The probability that we (instead) use this new sample is |
134 // don't clamp count_... but that should be inconsequentially likely). | 152 // 1/sample_probability_count_. This results in a completely uniform selection |
135 // We ignore the fact that we correlated our selection of a sample to the run | 153 // of the sample (at least when we don't clamp sample_probability_count_... |
136 // and queue times (i.e., we used them to generate random_number). | 154 // but that should be inconsequentially likely). We ignore the fact that we |
137 CHECK_GT(count_, 0); | 155 // correlated our selection of a sample to the run and queue times (i.e., we |
138 if (0 == (random_number % count_)) { | 156 // used them to generate random_number). |
157 CHECK_GT(sample_probability_count_, 0); | |
158 if (0 == (random_number % sample_probability_count_)) { | |
139 queue_duration_sample_ = queue_duration; | 159 queue_duration_sample_ = queue_duration; |
140 run_duration_sample_ = run_duration; | 160 run_duration_sample_ = run_duration; |
141 } | 161 } |
142 } | 162 } |
143 | 163 |
144 int DeathData::count() const { return count_; } | 164 int DeathData::count() const { return count_; } |
145 | 165 |
146 int32 DeathData::run_duration_sum() const { return run_duration_sum_; } | 166 int32 DeathData::run_duration_sum() const { return run_duration_sum_; } |
147 | 167 |
148 int32 DeathData::run_duration_max() const { return run_duration_max_; } | 168 int32 DeathData::run_duration_max() const { return run_duration_max_; } |
149 | 169 |
150 int32 DeathData::run_duration_sample() const { | 170 int32 DeathData::run_duration_sample() const { |
151 return run_duration_sample_; | 171 return run_duration_sample_; |
152 } | 172 } |
153 | 173 |
154 int32 DeathData::queue_duration_sum() const { | 174 int32 DeathData::queue_duration_sum() const { |
155 return queue_duration_sum_; | 175 return queue_duration_sum_; |
156 } | 176 } |
157 | 177 |
158 int32 DeathData::queue_duration_max() const { | 178 int32 DeathData::queue_duration_max() const { |
159 return queue_duration_max_; | 179 return queue_duration_max_; |
160 } | 180 } |
161 | 181 |
162 int32 DeathData::queue_duration_sample() const { | 182 int32 DeathData::queue_duration_sample() const { |
163 return queue_duration_sample_; | 183 return queue_duration_sample_; |
164 } | 184 } |
165 | 185 |
166 void DeathData::Clear() { | 186 DeathDataPhaseSnapshot* DeathData::last_phase_snapshot() const { |
167 count_ = 0; | 187 return last_phase_snapshot_; |
168 run_duration_sum_ = 0; | 188 } |
189 | |
190 void DeathData::OnProfilingPhaseCompletion(int profiling_phase) { | |
191 // Snapshotting and storing current state. | |
192 last_phase_snapshot_ = new DeathDataPhaseSnapshot(profiling_phase, *this); | |
193 | |
194 // Not touching fields for which a delta can be computed by comparing with a | |
195 // snapshot from previos phase. Resetting other fields. Sample values will be | |
196 // reset upon next death recording because sample_probability_count_ is set to | |
197 // 0. | |
198 sample_probability_count_ = 0; | |
169 run_duration_max_ = 0; | 199 run_duration_max_ = 0; |
170 run_duration_sample_ = 0; | |
171 queue_duration_sum_ = 0; | |
172 queue_duration_max_ = 0; | 200 queue_duration_max_ = 0; |
173 queue_duration_sample_ = 0; | |
174 } | 201 } |
175 | 202 |
176 //------------------------------------------------------------------------------ | 203 //------------------------------------------------------------------------------ |
177 DeathDataSnapshot::DeathDataSnapshot() | 204 DeathDataSnapshot::DeathDataSnapshot() |
178 : count(-1), | 205 : count(-1), |
179 run_duration_sum(-1), | 206 run_duration_sum(-1), |
180 run_duration_max(-1), | 207 run_duration_max(-1), |
181 run_duration_sample(-1), | 208 run_duration_sample(-1), |
182 queue_duration_sum(-1), | 209 queue_duration_sum(-1), |
183 queue_duration_max(-1), | 210 queue_duration_max(-1), |
184 queue_duration_sample(-1) { | 211 queue_duration_sample(-1) { |
185 } | 212 } |
186 | 213 |
187 DeathDataSnapshot::DeathDataSnapshot( | 214 DeathDataSnapshot::DeathDataSnapshot( |
188 const tracked_objects::DeathData& death_data) | 215 const tracked_objects::DeathData& death_data) |
189 : count(death_data.count()), | 216 : count(death_data.count()), |
190 run_duration_sum(death_data.run_duration_sum()), | 217 run_duration_sum(death_data.run_duration_sum()), |
191 run_duration_max(death_data.run_duration_max()), | 218 run_duration_max(death_data.run_duration_max()), |
192 run_duration_sample(death_data.run_duration_sample()), | 219 run_duration_sample(death_data.run_duration_sample()), |
193 queue_duration_sum(death_data.queue_duration_sum()), | 220 queue_duration_sum(death_data.queue_duration_sum()), |
194 queue_duration_max(death_data.queue_duration_max()), | 221 queue_duration_max(death_data.queue_duration_max()), |
195 queue_duration_sample(death_data.queue_duration_sample()) { | 222 queue_duration_sample(death_data.queue_duration_sample()) { |
196 } | 223 } |
197 | 224 |
198 DeathDataSnapshot::~DeathDataSnapshot() { | 225 DeathDataSnapshot::~DeathDataSnapshot() { |
199 } | 226 } |
200 | 227 |
228 void DeathDataSnapshot::CalculateDelta(const DeathDataSnapshot& older) { | |
229 count -= older.count; | |
230 run_duration_sum -= older.run_duration_sum; | |
231 queue_duration_sum -= older.queue_duration_sum; | |
232 } | |
233 | |
201 //------------------------------------------------------------------------------ | 234 //------------------------------------------------------------------------------ |
202 BirthOnThread::BirthOnThread(const Location& location, | 235 BirthOnThread::BirthOnThread(const Location& location, |
203 const ThreadData& current) | 236 const ThreadData& current) |
204 : location_(location), | 237 : location_(location), |
205 birth_thread_(¤t) { | 238 birth_thread_(¤t) { |
206 } | 239 } |
207 | 240 |
208 //------------------------------------------------------------------------------ | 241 //------------------------------------------------------------------------------ |
209 BirthOnThreadSnapshot::BirthOnThreadSnapshot() { | 242 BirthOnThreadSnapshot::BirthOnThreadSnapshot() { |
210 } | 243 } |
(...skipping 172 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... | |
383 return; | 416 return; |
384 } | 417 } |
385 // We must NOT do any allocations during this callback. | 418 // We must NOT do any allocations during this callback. |
386 // Using the simple linked lists avoids all allocations. | 419 // Using the simple linked lists avoids all allocations. |
387 DCHECK_EQ(this->next_retired_worker_, reinterpret_cast<ThreadData*>(NULL)); | 420 DCHECK_EQ(this->next_retired_worker_, reinterpret_cast<ThreadData*>(NULL)); |
388 this->next_retired_worker_ = first_retired_worker_; | 421 this->next_retired_worker_ = first_retired_worker_; |
389 first_retired_worker_ = this; | 422 first_retired_worker_ = this; |
390 } | 423 } |
391 | 424 |
392 // static | 425 // static |
393 void ThreadData::Snapshot(ProcessDataSnapshot* process_data_snapshot) { | 426 void ThreadData::Snapshot(int current_profiling_phase, |
394 ThreadData::SnapshotCurrentPhase( | 427 ProcessDataSnapshot* process_data_snapshot) { |
395 &process_data_snapshot->phased_process_data_snapshots[0]); | 428 if (!kTrackAllTaskObjects) |
429 return; // Not compiled in. | |
430 | |
431 BirthCountMap birth_counts; | |
432 | |
433 // Get an unchanging copy of a ThreadData list. | |
434 ThreadData* my_list = ThreadData::first(); | |
435 | |
436 // Gather data serially. | |
437 // This hackish approach *can* get some slighly corrupt tallies, as we are | |
438 // grabbing values without the protection of a lock, but it has the advantage | |
439 // of working even with threads that don't have message loops. If a user | |
440 // sees any strangeness, they can always just run their stats gathering a | |
441 // second time. | |
442 for (ThreadData* thread_data = my_list; thread_data; | |
443 thread_data = thread_data->next()) { | |
444 thread_data->SnapshotExecutedTasks( | |
445 current_profiling_phase, | |
446 &process_data_snapshot->phased_process_data_snapshots, &birth_counts); | |
447 } | |
448 | |
449 // Add births that are still active -- i.e. objects that have tallied a birth, | |
450 // but have not yet tallied a matching death, and hence must be either | |
451 // running, queued up, or being held in limbo for future posting. | |
452 for (const auto& birth_count : birth_counts) { | |
453 if (birth_count.second > 0) { | |
454 process_data_snapshot | |
455 ->phased_process_data_snapshots[current_profiling_phase] | |
456 .tasks.push_back(TaskSnapshot( | |
457 BirthOnThreadSnapshot(*birth_count.first), | |
458 DeathDataSnapshot(DeathData(birth_count.second)), "Still_Alive")); | |
459 } | |
460 } | |
461 } | |
462 | |
463 // static | |
464 void ThreadData::OnProfilingPhaseCompletion(int profiling_phase) { | |
465 if (!kTrackAllTaskObjects) | |
466 return; // Not compiled in. | |
467 | |
468 // Get an unchanging copy of a ThreadData list. | |
469 ThreadData* my_list = ThreadData::first(); | |
470 | |
471 // Add snapshots for all death datas in all threads serially. | |
472 // This hackish approach *can* get some slighly corrupt tallies, as we are | |
473 // grabbing values without the protection of a lock, but it has the advantage | |
474 // of working even with threads that don't have message loops. Any corruption | |
475 // shouldn't cause "cascading damage" to anything else (in later phases). | |
476 for (ThreadData* thread_data = my_list; thread_data; | |
477 thread_data = thread_data->next()) { | |
478 thread_data->OnProfilingPhaseCompletionOnThread(profiling_phase); | |
479 } | |
396 } | 480 } |
397 | 481 |
398 Births* ThreadData::TallyABirth(const Location& location) { | 482 Births* ThreadData::TallyABirth(const Location& location) { |
399 BirthMap::iterator it = birth_map_.find(location); | 483 BirthMap::iterator it = birth_map_.find(location); |
400 Births* child; | 484 Births* child; |
401 if (it != birth_map_.end()) { | 485 if (it != birth_map_.end()) { |
402 child = it->second; | 486 child = it->second; |
403 child->RecordBirth(); | 487 child->RecordBirth(); |
404 } else { | 488 } else { |
405 child = new Births(location, *this); // Leak this. | 489 child = new Births(location, *this); // Leak this. |
(...skipping 11 matching lines...) Expand all Loading... | |
417 // Lock since the map may get relocated now, and other threads sometimes | 501 // Lock since the map may get relocated now, and other threads sometimes |
418 // snapshot it (but they lock before copying it). | 502 // snapshot it (but they lock before copying it). |
419 base::AutoLock lock(map_lock_); | 503 base::AutoLock lock(map_lock_); |
420 parent_child_set_.insert(pair); | 504 parent_child_set_.insert(pair); |
421 } | 505 } |
422 } | 506 } |
423 | 507 |
424 return child; | 508 return child; |
425 } | 509 } |
426 | 510 |
427 void ThreadData::TallyADeath(const Births& birth, | 511 void ThreadData::TallyADeath(const Births& births, |
428 int32 queue_duration, | 512 int32 queue_duration, |
429 const TaskStopwatch& stopwatch) { | 513 const TaskStopwatch& stopwatch) { |
430 int32 run_duration = stopwatch.RunDurationMs(); | 514 int32 run_duration = stopwatch.RunDurationMs(); |
431 | 515 |
432 // Stir in some randomness, plus add constant in case durations are zero. | 516 // Stir in some randomness, plus add constant in case durations are zero. |
433 const uint32 kSomePrimeNumber = 2147483647; | 517 const uint32 kSomePrimeNumber = 2147483647; |
434 random_number_ += queue_duration + run_duration + kSomePrimeNumber; | 518 random_number_ += queue_duration + run_duration + kSomePrimeNumber; |
435 // An address is going to have some randomness to it as well ;-). | 519 // An address is going to have some randomness to it as well ;-). |
436 random_number_ ^= static_cast<uint32>(&birth - reinterpret_cast<Births*>(0)); | 520 random_number_ ^= static_cast<uint32>(&births - reinterpret_cast<Births*>(0)); |
437 | 521 |
438 // We don't have queue durations without OS timer. OS timer is automatically | 522 // We don't have queue durations without OS timer. OS timer is automatically |
439 // used for task-post-timing, so the use of an alternate timer implies all | 523 // used for task-post-timing, so the use of an alternate timer implies all |
440 // queue times are invalid, unless it was explicitly said that we can trust | 524 // queue times are invalid, unless it was explicitly said that we can trust |
441 // the alternate timer. | 525 // the alternate timer. |
442 if (kAllowAlternateTimeSourceHandling && | 526 if (kAllowAlternateTimeSourceHandling && |
443 now_function_ && | 527 now_function_ && |
444 !now_function_is_time_) { | 528 !now_function_is_time_) { |
445 queue_duration = 0; | 529 queue_duration = 0; |
446 } | 530 } |
447 | 531 |
448 DeathMap::iterator it = death_map_.find(&birth); | 532 DeathMap::iterator it = death_map_.find(&births); |
449 DeathData* death_data; | 533 DeathData* death_data; |
450 if (it != death_map_.end()) { | 534 if (it != death_map_.end()) { |
451 death_data = &it->second; | 535 death_data = &it->second; |
452 } else { | 536 } else { |
453 base::AutoLock lock(map_lock_); // Lock as the map may get relocated now. | 537 base::AutoLock lock(map_lock_); // Lock as the map may get relocated now. |
454 death_data = &death_map_[&birth]; | 538 death_data = &death_map_[&births]; |
455 } // Release lock ASAP. | 539 } // Release lock ASAP. |
456 death_data->RecordDeath(queue_duration, run_duration, random_number_); | 540 death_data->RecordDeath(queue_duration, run_duration, random_number_); |
457 | 541 |
458 if (!kTrackParentChildLinks) | 542 if (!kTrackParentChildLinks) |
459 return; | 543 return; |
460 if (!parent_stack_.empty()) { // We might get turned off. | 544 if (!parent_stack_.empty()) { // We might get turned off. |
461 DCHECK_EQ(parent_stack_.top(), &birth); | 545 DCHECK_EQ(parent_stack_.top(), &births); |
462 parent_stack_.pop(); | 546 parent_stack_.pop(); |
463 } | 547 } |
464 } | 548 } |
465 | 549 |
466 // static | 550 // static |
467 Births* ThreadData::TallyABirthIfActive(const Location& location) { | 551 Births* ThreadData::TallyABirthIfActive(const Location& location) { |
468 if (!kTrackAllTaskObjects) | 552 if (!kTrackAllTaskObjects) |
469 return NULL; // Not compiled in. | 553 return NULL; // Not compiled in. |
470 | 554 |
471 if (!TrackingStatus()) | 555 if (!TrackingStatus()) |
472 return NULL; | 556 return NULL; |
473 ThreadData* current_thread_data = Get(); | 557 ThreadData* current_thread_data = Get(); |
474 if (!current_thread_data) | 558 if (!current_thread_data) |
475 return NULL; | 559 return NULL; |
476 return current_thread_data->TallyABirth(location); | 560 return current_thread_data->TallyABirth(location); |
477 } | 561 } |
478 | 562 |
479 // static | 563 // static |
480 void ThreadData::TallyRunOnNamedThreadIfTracking( | 564 void ThreadData::TallyRunOnNamedThreadIfTracking( |
481 const base::TrackingInfo& completed_task, | 565 const base::TrackingInfo& completed_task, |
482 const TaskStopwatch& stopwatch) { | 566 const TaskStopwatch& stopwatch) { |
483 if (!kTrackAllTaskObjects) | 567 if (!kTrackAllTaskObjects) |
484 return; // Not compiled in. | 568 return; // Not compiled in. |
485 | 569 |
486 // Even if we have been DEACTIVATED, we will process any pending births so | 570 // Even if we have been DEACTIVATED, we will process any pending births so |
487 // that our data structures (which counted the outstanding births) remain | 571 // that our data structures (which counted the outstanding births) remain |
488 // consistent. | 572 // consistent. |
489 const Births* birth = completed_task.birth_tally; | 573 const Births* births = completed_task.birth_tally; |
490 if (!birth) | 574 if (!births) |
491 return; | 575 return; |
492 ThreadData* current_thread_data = stopwatch.GetThreadData(); | 576 ThreadData* current_thread_data = stopwatch.GetThreadData(); |
493 if (!current_thread_data) | 577 if (!current_thread_data) |
494 return; | 578 return; |
495 | 579 |
496 // Watch out for a race where status_ is changing, and hence one or both | 580 // Watch out for a race where status_ is changing, and hence one or both |
497 // of start_of_run or end_of_run is zero. In that case, we didn't bother to | 581 // of start_of_run or end_of_run is zero. In that case, we didn't bother to |
498 // get a time value since we "weren't tracking" and we were trying to be | 582 // get a time value since we "weren't tracking" and we were trying to be |
499 // efficient by not calling for a genuine time value. For simplicity, we'll | 583 // efficient by not calling for a genuine time value. For simplicity, we'll |
500 // use a default zero duration when we can't calculate a true value. | 584 // use a default zero duration when we can't calculate a true value. |
501 TrackedTime start_of_run = stopwatch.StartTime(); | 585 TrackedTime start_of_run = stopwatch.StartTime(); |
502 int32 queue_duration = 0; | 586 int32 queue_duration = 0; |
503 if (!start_of_run.is_null()) { | 587 if (!start_of_run.is_null()) { |
504 queue_duration = (start_of_run - completed_task.EffectiveTimePosted()) | 588 queue_duration = (start_of_run - completed_task.EffectiveTimePosted()) |
505 .InMilliseconds(); | 589 .InMilliseconds(); |
506 } | 590 } |
507 current_thread_data->TallyADeath(*birth, queue_duration, stopwatch); | 591 current_thread_data->TallyADeath(*births, queue_duration, stopwatch); |
508 } | 592 } |
509 | 593 |
510 // static | 594 // static |
511 void ThreadData::TallyRunOnWorkerThreadIfTracking( | 595 void ThreadData::TallyRunOnWorkerThreadIfTracking( |
512 const Births* birth, | 596 const Births* births, |
513 const TrackedTime& time_posted, | 597 const TrackedTime& time_posted, |
514 const TaskStopwatch& stopwatch) { | 598 const TaskStopwatch& stopwatch) { |
515 if (!kTrackAllTaskObjects) | 599 if (!kTrackAllTaskObjects) |
516 return; // Not compiled in. | 600 return; // Not compiled in. |
517 | 601 |
518 // Even if we have been DEACTIVATED, we will process any pending births so | 602 // Even if we have been DEACTIVATED, we will process any pending births so |
519 // that our data structures (which counted the outstanding births) remain | 603 // that our data structures (which counted the outstanding births) remain |
520 // consistent. | 604 // consistent. |
521 if (!birth) | 605 if (!births) |
522 return; | 606 return; |
523 | 607 |
524 // TODO(jar): Support the option to coalesce all worker-thread activity under | 608 // TODO(jar): Support the option to coalesce all worker-thread activity under |
525 // one ThreadData instance that uses locks to protect *all* access. This will | 609 // one ThreadData instance that uses locks to protect *all* access. This will |
526 // reduce memory (making it provably bounded), but run incrementally slower | 610 // reduce memory (making it provably bounded), but run incrementally slower |
527 // (since we'll use locks on TallyABirth and TallyADeath). The good news is | 611 // (since we'll use locks on TallyABirth and TallyADeath). The good news is |
528 // that the locks on TallyADeath will be *after* the worker thread has run, | 612 // that the locks on TallyADeath will be *after* the worker thread has run, |
529 // and hence nothing will be waiting for the completion (... besides some | 613 // and hence nothing will be waiting for the completion (... besides some |
530 // other thread that might like to run). Also, the worker threads tasks are | 614 // other thread that might like to run). Also, the worker threads tasks are |
531 // generally longer, and hence the cost of the lock may perchance be amortized | 615 // generally longer, and hence the cost of the lock may perchance be amortized |
532 // over the long task's lifetime. | 616 // over the long task's lifetime. |
533 ThreadData* current_thread_data = stopwatch.GetThreadData(); | 617 ThreadData* current_thread_data = stopwatch.GetThreadData(); |
534 if (!current_thread_data) | 618 if (!current_thread_data) |
535 return; | 619 return; |
536 | 620 |
537 TrackedTime start_of_run = stopwatch.StartTime(); | 621 TrackedTime start_of_run = stopwatch.StartTime(); |
538 int32 queue_duration = 0; | 622 int32 queue_duration = 0; |
539 if (!start_of_run.is_null()) { | 623 if (!start_of_run.is_null()) { |
540 queue_duration = (start_of_run - time_posted).InMilliseconds(); | 624 queue_duration = (start_of_run - time_posted).InMilliseconds(); |
541 } | 625 } |
542 current_thread_data->TallyADeath(*birth, queue_duration, stopwatch); | 626 current_thread_data->TallyADeath(*births, queue_duration, stopwatch); |
543 } | 627 } |
544 | 628 |
545 // static | 629 // static |
546 void ThreadData::TallyRunInAScopedRegionIfTracking( | 630 void ThreadData::TallyRunInAScopedRegionIfTracking( |
547 const Births* birth, | 631 const Births* births, |
548 const TaskStopwatch& stopwatch) { | 632 const TaskStopwatch& stopwatch) { |
549 if (!kTrackAllTaskObjects) | 633 if (!kTrackAllTaskObjects) |
550 return; // Not compiled in. | 634 return; // Not compiled in. |
551 | 635 |
552 // Even if we have been DEACTIVATED, we will process any pending births so | 636 // Even if we have been DEACTIVATED, we will process any pending births so |
553 // that our data structures (which counted the outstanding births) remain | 637 // that our data structures (which counted the outstanding births) remain |
554 // consistent. | 638 // consistent. |
555 if (!birth) | 639 if (!births) |
556 return; | 640 return; |
557 | 641 |
558 ThreadData* current_thread_data = stopwatch.GetThreadData(); | 642 ThreadData* current_thread_data = stopwatch.GetThreadData(); |
559 if (!current_thread_data) | 643 if (!current_thread_data) |
560 return; | 644 return; |
561 | 645 |
562 int32 queue_duration = 0; | 646 int32 queue_duration = 0; |
563 current_thread_data->TallyADeath(*birth, queue_duration, stopwatch); | 647 current_thread_data->TallyADeath(*births, queue_duration, stopwatch); |
564 } | |
565 | |
566 // static | |
567 void ThreadData::SnapshotAllExecutedTasks( | |
568 ProcessDataPhaseSnapshot* process_data_phase, | |
569 BirthCountMap* birth_counts) { | |
570 if (!kTrackAllTaskObjects) | |
571 return; // Not compiled in. | |
572 | |
573 // Get an unchanging copy of a ThreadData list. | |
574 ThreadData* my_list = ThreadData::first(); | |
575 | |
576 // Gather data serially. | |
577 // This hackish approach *can* get some slighly corrupt tallies, as we are | |
578 // grabbing values without the protection of a lock, but it has the advantage | |
579 // of working even with threads that don't have message loops. If a user | |
580 // sees any strangeness, they can always just run their stats gathering a | |
581 // second time. | |
582 for (ThreadData* thread_data = my_list; | |
583 thread_data; | |
584 thread_data = thread_data->next()) { | |
585 thread_data->SnapshotExecutedTasks(process_data_phase, birth_counts); | |
586 } | |
587 } | |
588 | |
589 // static | |
590 void ThreadData::SnapshotCurrentPhase( | |
591 ProcessDataPhaseSnapshot* process_data_phase) { | |
592 // Add births that have run to completion to |collected_data|. | |
593 // |birth_counts| tracks the total number of births recorded at each location | |
594 // for which we have not seen a death count. | |
595 BirthCountMap birth_counts; | |
596 ThreadData::SnapshotAllExecutedTasks(process_data_phase, &birth_counts); | |
597 | |
598 // Add births that are still active -- i.e. objects that have tallied a birth, | |
599 // but have not yet tallied a matching death, and hence must be either | |
600 // running, queued up, or being held in limbo for future posting. | |
601 for (const auto& birth_count : birth_counts) { | |
602 if (birth_count.second > 0) { | |
603 process_data_phase->tasks.push_back(TaskSnapshot( | |
604 *birth_count.first, DeathData(birth_count.second), "Still_Alive")); | |
605 } | |
606 } | |
607 } | 648 } |
608 | 649 |
609 void ThreadData::SnapshotExecutedTasks( | 650 void ThreadData::SnapshotExecutedTasks( |
610 ProcessDataPhaseSnapshot* process_data_phase, | 651 int current_profiling_phase, |
652 PhasedProcessDataSnapshotMap* phased_process_data_snapshots, | |
611 BirthCountMap* birth_counts) { | 653 BirthCountMap* birth_counts) { |
612 // Get copy of data, so that the data will not change during the iterations | 654 // Get copy of data, so that the data will not change during the iterations |
613 // and processing. | 655 // and processing. |
614 ThreadData::BirthMap birth_map; | 656 BirthMap birth_map; |
615 ThreadData::DeathMap death_map; | 657 DeathsSnapshot deaths; |
616 ThreadData::ParentChildSet parent_child_set; | 658 ParentChildSet parent_child_set; |
617 SnapshotMaps(&birth_map, &death_map, &parent_child_set); | 659 SnapshotMaps(current_profiling_phase, &birth_map, &deaths, &parent_child_set); |
618 | |
619 for (const auto& death : death_map) { | |
620 process_data_phase->tasks.push_back( | |
621 TaskSnapshot(*death.first, death.second, thread_name())); | |
622 (*birth_counts)[death.first] -= death.first->birth_count(); | |
623 } | |
624 | 660 |
625 for (const auto& birth : birth_map) { | 661 for (const auto& birth : birth_map) { |
626 (*birth_counts)[birth.second] += birth.second->birth_count(); | 662 (*birth_counts)[birth.second] += birth.second->birth_count(); |
627 } | 663 } |
628 | 664 |
629 if (!kTrackParentChildLinks) | 665 for (const auto& death : deaths) { |
630 return; | 666 (*birth_counts)[death.first] -= death.first->birth_count(); |
631 | 667 |
632 for (const auto& parent_child : parent_child_set) { | 668 for (const DeathDataPhaseSnapshot* phase = &death.second; phase; |
633 process_data_phase->descendants.push_back( | 669 phase = phase->prev) { |
634 ParentChildPairSnapshot(parent_child)); | 670 DeathDataSnapshot death_data = phase->death_data; |
671 | |
672 if (phase->prev) | |
673 death_data.CalculateDelta(phase->prev->death_data); | |
674 | |
675 if (death_data.count > 0) | |
676 (*phased_process_data_snapshots)[phase->profiling_phase] | |
677 .tasks.push_back(TaskSnapshot(BirthOnThreadSnapshot(*death.first), | |
678 death_data, thread_name())); | |
679 } | |
635 } | 680 } |
636 } | 681 } |
637 | 682 |
638 // This may be called from another thread. | 683 // This may be called from another thread. |
639 void ThreadData::SnapshotMaps(BirthMap* birth_map, | 684 void ThreadData::SnapshotMaps(int profiling_phase, |
640 DeathMap* death_map, | 685 BirthMap* birth_map, |
686 DeathsSnapshot* deaths, | |
641 ParentChildSet* parent_child_set) { | 687 ParentChildSet* parent_child_set) { |
642 base::AutoLock lock(map_lock_); | 688 base::AutoLock lock(map_lock_); |
689 | |
643 for (const auto& birth : birth_map_) | 690 for (const auto& birth : birth_map_) |
644 (*birth_map)[birth.first] = birth.second; | 691 (*birth_map)[birth.first] = birth.second; |
645 for (const auto& death : death_map_) | 692 |
646 (*death_map)[death.first] = death.second; | 693 for (const auto& death : death_map_) { |
694 deaths->push_back(DeathsSnapshot::value_type( | |
695 death.first, DeathDataPhaseSnapshot(profiling_phase, death.second))); | |
696 } | |
647 | 697 |
648 if (!kTrackParentChildLinks) | 698 if (!kTrackParentChildLinks) |
649 return; | 699 return; |
650 | 700 |
651 for (const auto& parent_child : parent_child_set_) | 701 for (const auto& parent_child : parent_child_set_) |
652 parent_child_set->insert(parent_child); | 702 parent_child_set->insert(parent_child); |
653 } | 703 } |
654 | 704 |
705 void ThreadData::OnProfilingPhaseCompletionOnThread(int profiling_phase) { | |
706 base::AutoLock lock(map_lock_); | |
707 | |
708 for (auto& death : death_map_) { | |
709 death.second.OnProfilingPhaseCompletion(profiling_phase); | |
710 } | |
711 } | |
712 | |
655 static void OptionallyInitializeAlternateTimer() { | 713 static void OptionallyInitializeAlternateTimer() { |
656 NowFunction* alternate_time_source = GetAlternateTimeSource(); | 714 NowFunction* alternate_time_source = GetAlternateTimeSource(); |
657 if (alternate_time_source) | 715 if (alternate_time_source) |
658 ThreadData::SetAlternateTimeSource(alternate_time_source); | 716 ThreadData::SetAlternateTimeSource(alternate_time_source); |
659 } | 717 } |
660 | 718 |
661 bool ThreadData::Initialize() { | 719 bool ThreadData::Initialize() { |
662 if (!kTrackAllTaskObjects) | 720 if (!kTrackAllTaskObjects) |
663 return false; // Not compiled in. | 721 return false; // Not compiled in. |
664 if (status_ >= DEACTIVATED) | 722 if (status_ >= DEACTIVATED) |
(...skipping 259 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... | |
924 | 982 |
925 ThreadData* TaskStopwatch::GetThreadData() const { | 983 ThreadData* TaskStopwatch::GetThreadData() const { |
926 #if DCHECK_IS_ON() | 984 #if DCHECK_IS_ON() |
927 DCHECK(state_ != CREATED); | 985 DCHECK(state_ != CREATED); |
928 #endif | 986 #endif |
929 | 987 |
930 return current_thread_data_; | 988 return current_thread_data_; |
931 } | 989 } |
932 | 990 |
933 //------------------------------------------------------------------------------ | 991 //------------------------------------------------------------------------------ |
992 // DeathDataPhaseSnapshot | |
993 | |
994 DeathDataPhaseSnapshot::DeathDataPhaseSnapshot(int profiling_phase, | |
995 const DeathData& death_data) | |
996 : profiling_phase(profiling_phase), | |
997 death_data(death_data), | |
998 prev(death_data.last_phase_snapshot()) { | |
999 } | |
1000 | |
1001 //------------------------------------------------------------------------------ | |
1002 // TaskSnapshot | |
1003 | |
934 TaskSnapshot::TaskSnapshot() { | 1004 TaskSnapshot::TaskSnapshot() { |
935 } | 1005 } |
936 | 1006 |
937 TaskSnapshot::TaskSnapshot(const BirthOnThread& birth, | 1007 TaskSnapshot::TaskSnapshot(const BirthOnThreadSnapshot& birth, |
938 const DeathData& death_data, | 1008 const DeathDataSnapshot& death_data, |
939 const std::string& death_thread_name) | 1009 const std::string& death_thread_name) |
940 : birth(birth), | 1010 : birth(birth), |
941 death_data(death_data), | 1011 death_data(death_data), |
942 death_thread_name(death_thread_name) { | 1012 death_thread_name(death_thread_name) { |
943 } | 1013 } |
944 | 1014 |
945 TaskSnapshot::~TaskSnapshot() { | 1015 TaskSnapshot::~TaskSnapshot() { |
946 } | 1016 } |
947 | 1017 |
948 //------------------------------------------------------------------------------ | 1018 //------------------------------------------------------------------------------ |
(...skipping 28 matching lines...) Expand all Loading... | |
977 : process_id(base::GetCurrentProcId()) { | 1047 : process_id(base::GetCurrentProcId()) { |
978 #else | 1048 #else |
979 : process_id(base::kNullProcessId) { | 1049 : process_id(base::kNullProcessId) { |
980 #endif | 1050 #endif |
981 } | 1051 } |
982 | 1052 |
983 ProcessDataSnapshot::~ProcessDataSnapshot() { | 1053 ProcessDataSnapshot::~ProcessDataSnapshot() { |
984 } | 1054 } |
985 | 1055 |
986 } // namespace tracked_objects | 1056 } // namespace tracked_objects |
OLD | NEW |