Print this page
| Split |
Close |
| Expand all |
| Collapse all |
--- old/src/share/vm/utilities/workgroup.cpp
+++ new/src/share/vm/utilities/workgroup.cpp
1 1 /*
2 2 * Copyright (c) 2001, 2010, Oracle and/or its affiliates. All rights reserved.
3 3 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
4 4 *
5 5 * This code is free software; you can redistribute it and/or modify it
6 6 * under the terms of the GNU General Public License version 2 only, as
7 7 * published by the Free Software Foundation.
8 8 *
9 9 * This code is distributed in the hope that it will be useful, but WITHOUT
10 10 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
11 11 * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
12 12 * version 2 for more details (a copy is included in the LICENSE file that
13 13 * accompanied this code).
14 14 *
15 15 * You should have received a copy of the GNU General Public License version
16 16 * 2 along with this work; if not, write to the Free Software Foundation,
17 17 * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
18 18 *
19 19 * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
20 20 * or visit www.oracle.com if you need additional information or have any
21 21 * questions.
22 22 *
23 23 */
24 24
25 25 #include "precompiled.hpp"
26 26 #include "memory/allocation.hpp"
27 27 #include "memory/allocation.inline.hpp"
28 28 #include "runtime/os.hpp"
29 29 #include "utilities/workgroup.hpp"
30 30
31 31 // Definitions of WorkGang methods.
32 32
33 33 AbstractWorkGang::AbstractWorkGang(const char* name,
34 34 bool are_GC_task_threads,
35 35 bool are_ConcurrentGC_threads) :
36 36 _name(name),
37 37 _are_GC_task_threads(are_GC_task_threads),
38 38 _are_ConcurrentGC_threads(are_ConcurrentGC_threads) {
39 39
40 40 assert(!(are_GC_task_threads && are_ConcurrentGC_threads),
41 41 "They cannot both be STW GC and Concurrent threads" );
42 42
43 43 // Other initialization.
44 44 _monitor = new Monitor(/* priority */ Mutex::leaf,
45 45 /* name */ "WorkGroup monitor",
46 46 /* allow_vm_block */ are_GC_task_threads);
47 47 assert(monitor() != NULL, "Failed to allocate monitor");
48 48 _terminate = false;
49 49 _task = NULL;
50 50 _sequence_number = 0;
51 51 _started_workers = 0;
52 52 _finished_workers = 0;
53 53 }
54 54
55 55 WorkGang::WorkGang(const char* name,
56 56 int workers,
57 57 bool are_GC_task_threads,
58 58 bool are_ConcurrentGC_threads) :
59 59 AbstractWorkGang(name, are_GC_task_threads, are_ConcurrentGC_threads) {
60 60 // Save arguments.
61 61 _total_workers = workers;
62 62 }
63 63
64 64 GangWorker* WorkGang::allocate_worker(int which) {
65 65 GangWorker* new_worker = new GangWorker(this, which);
66 66 return new_worker;
67 67 }
68 68
69 69 // The current implementation will exit if the allocation
70 70 // of any worker fails. Still, return a boolean so that
71 71 // a future implementation can possibly do a partial
72 72 // initialization of the workers and report such to the
73 73 // caller.
74 74 bool WorkGang::initialize_workers() {
75 75
76 76 if (TraceWorkGang) {
77 77 tty->print_cr("Constructing work gang %s with %d threads",
78 78 name(),
79 79 total_workers());
80 80 }
81 81 _gang_workers = NEW_C_HEAP_ARRAY(GangWorker*, total_workers());
82 82 if (gang_workers() == NULL) {
83 83 vm_exit_out_of_memory(0, "Cannot create GangWorker array.");
84 84 return false;
85 85 }
86 86 os::ThreadType worker_type;
87 87 if (are_ConcurrentGC_threads()) {
88 88 worker_type = os::cgc_thread;
89 89 } else {
90 90 worker_type = os::pgc_thread;
91 91 }
92 92 for (int worker = 0; worker < total_workers(); worker += 1) {
93 93 GangWorker* new_worker = allocate_worker(worker);
94 94 assert(new_worker != NULL, "Failed to allocate GangWorker");
95 95 _gang_workers[worker] = new_worker;
96 96 if (new_worker == NULL || !os::create_thread(new_worker, worker_type)) {
97 97 vm_exit_out_of_memory(0, "Cannot create worker GC thread. Out of system resources.");
98 98 return false;
99 99 }
100 100 if (!DisableStartThread) {
101 101 os::start_thread(new_worker);
102 102 }
103 103 }
104 104 return true;
105 105 }
106 106
107 107 AbstractWorkGang::~AbstractWorkGang() {
108 108 if (TraceWorkGang) {
109 109 tty->print_cr("Destructing work gang %s", name());
110 110 }
111 111 stop(); // stop all the workers
112 112 for (int worker = 0; worker < total_workers(); worker += 1) {
113 113 delete gang_worker(worker);
114 114 }
115 115 delete gang_workers();
116 116 delete monitor();
117 117 }
118 118
119 119 GangWorker* AbstractWorkGang::gang_worker(int i) const {
120 120 // Array index bounds checking.
121 121 GangWorker* result = NULL;
122 122 assert(gang_workers() != NULL, "No workers for indexing");
123 123 assert(((i >= 0) && (i < total_workers())), "Worker index out of bounds");
124 124 result = _gang_workers[i];
125 125 assert(result != NULL, "Indexing to null worker");
126 126 return result;
127 127 }
128 128
129 129 void WorkGang::run_task(AbstractGangTask* task) {
130 130 // This thread is executed by the VM thread which does not block
131 131 // on ordinary MutexLocker's.
132 132 MutexLockerEx ml(monitor(), Mutex::_no_safepoint_check_flag);
133 133 if (TraceWorkGang) {
134 134 tty->print_cr("Running work gang %s task %s", name(), task->name());
135 135 }
136 136 // Tell all the workers to run a task.
137 137 assert(task != NULL, "Running a null task");
138 138 // Initialize.
139 139 _task = task;
140 140 _sequence_number += 1;
141 141 _started_workers = 0;
142 142 _finished_workers = 0;
143 143 // Tell the workers to get to work.
144 144 monitor()->notify_all();
145 145 // Wait for them to be finished
146 146 while (finished_workers() < total_workers()) {
147 147 if (TraceWorkGang) {
148 148 tty->print_cr("Waiting in work gang %s: %d/%d finished sequence %d",
|
↓ open down ↓ |
148 lines elided |
↑ open up ↑ |
149 149 name(), finished_workers(), total_workers(),
150 150 _sequence_number);
151 151 }
152 152 monitor()->wait(/* no_safepoint_check */ true);
153 153 }
154 154 _task = NULL;
155 155 if (TraceWorkGang) {
156 156 tty->print_cr("/nFinished work gang %s: %d/%d sequence %d",
157 157 name(), finished_workers(), total_workers(),
158 158 _sequence_number);
159 - }
159 + }
160 160 }
161 161
162 162 void AbstractWorkGang::stop() {
163 163 // Tell all workers to terminate, then wait for them to become inactive.
164 164 MutexLockerEx ml(monitor(), Mutex::_no_safepoint_check_flag);
165 165 if (TraceWorkGang) {
166 166 tty->print_cr("Stopping work gang %s task %s", name(), task()->name());
167 167 }
168 168 _task = NULL;
169 169 _terminate = true;
170 170 monitor()->notify_all();
171 171 while (finished_workers() < total_workers()) {
172 172 if (TraceWorkGang) {
173 173 tty->print_cr("Waiting in work gang %s: %d/%d finished",
174 174 name(), finished_workers(), total_workers());
175 175 }
176 176 monitor()->wait(/* no_safepoint_check */ true);
177 177 }
178 178 }
179 179
180 180 void AbstractWorkGang::internal_worker_poll(WorkData* data) const {
181 181 assert(monitor()->owned_by_self(), "worker_poll is an internal method");
182 182 assert(data != NULL, "worker data is null");
183 183 data->set_terminate(terminate());
184 184 data->set_task(task());
185 185 data->set_sequence_number(sequence_number());
186 186 }
187 187
188 188 void AbstractWorkGang::internal_note_start() {
189 189 assert(monitor()->owned_by_self(), "note_finish is an internal method");
190 190 _started_workers += 1;
191 191 }
192 192
193 193 void AbstractWorkGang::internal_note_finish() {
194 194 assert(monitor()->owned_by_self(), "note_finish is an internal method");
195 195 _finished_workers += 1;
196 196 }
197 197
198 198 void AbstractWorkGang::print_worker_threads_on(outputStream* st) const {
199 199 uint num_thr = total_workers();
200 200 for (uint i = 0; i < num_thr; i++) {
201 201 gang_worker(i)->print_on(st);
202 202 st->cr();
203 203 }
204 204 }
205 205
206 206 void AbstractWorkGang::threads_do(ThreadClosure* tc) const {
207 207 assert(tc != NULL, "Null ThreadClosure");
208 208 uint num_thr = total_workers();
209 209 for (uint i = 0; i < num_thr; i++) {
210 210 tc->do_thread(gang_worker(i));
211 211 }
212 212 }
213 213
214 214 // GangWorker methods.
215 215
216 216 GangWorker::GangWorker(AbstractWorkGang* gang, uint id) {
217 217 _gang = gang;
218 218 set_id(id);
219 219 set_name("Gang worker#%d (%s)", id, gang->name());
220 220 }
221 221
222 222 void GangWorker::run() {
223 223 initialize();
224 224 loop();
225 225 }
226 226
227 227 void GangWorker::initialize() {
228 228 this->initialize_thread_local_storage();
229 229 assert(_gang != NULL, "No gang to run in");
230 230 os::set_priority(this, NearMaxPriority);
231 231 if (TraceWorkGang) {
232 232 tty->print_cr("Running gang worker for gang %s id %d",
233 233 gang()->name(), id());
234 234 }
235 235 // The VM thread should not execute here because MutexLocker's are used
236 236 // as (opposed to MutexLockerEx's).
237 237 assert(!Thread::current()->is_VM_thread(), "VM thread should not be part"
238 238 " of a work gang");
239 239 }
240 240
241 241 void GangWorker::loop() {
242 242 int previous_sequence_number = 0;
243 243 Monitor* gang_monitor = gang()->monitor();
244 244 for ( ; /* !terminate() */; ) {
245 245 WorkData data;
246 246 int part; // Initialized below.
247 247 {
248 248 // Grab the gang mutex.
249 249 MutexLocker ml(gang_monitor);
250 250 // Wait for something to do.
251 251 // Polling outside the while { wait } avoids missed notifies
252 252 // in the outer loop.
253 253 gang()->internal_worker_poll(&data);
254 254 if (TraceWorkGang) {
255 255 tty->print("Polled outside for work in gang %s worker %d",
256 256 gang()->name(), id());
257 257 tty->print(" terminate: %s",
258 258 data.terminate() ? "true" : "false");
259 259 tty->print(" sequence: %d (prev: %d)",
260 260 data.sequence_number(), previous_sequence_number);
261 261 if (data.task() != NULL) {
262 262 tty->print(" task: %s", data.task()->name());
263 263 } else {
264 264 tty->print(" task: NULL");
265 265 }
266 266 tty->cr();
267 267 }
268 268 for ( ; /* break or return */; ) {
269 269 // Terminate if requested.
270 270 if (data.terminate()) {
271 271 gang()->internal_note_finish();
272 272 gang_monitor->notify_all();
273 273 return;
274 274 }
275 275 // Check for new work.
276 276 if ((data.task() != NULL) &&
277 277 (data.sequence_number() != previous_sequence_number)) {
278 278 gang()->internal_note_start();
279 279 gang_monitor->notify_all();
280 280 part = gang()->started_workers() - 1;
281 281 break;
282 282 }
283 283 // Nothing to do.
284 284 gang_monitor->wait(/* no_safepoint_check */ true);
285 285 gang()->internal_worker_poll(&data);
286 286 if (TraceWorkGang) {
287 287 tty->print("Polled inside for work in gang %s worker %d",
288 288 gang()->name(), id());
289 289 tty->print(" terminate: %s",
290 290 data.terminate() ? "true" : "false");
291 291 tty->print(" sequence: %d (prev: %d)",
292 292 data.sequence_number(), previous_sequence_number);
293 293 if (data.task() != NULL) {
294 294 tty->print(" task: %s", data.task()->name());
295 295 } else {
296 296 tty->print(" task: NULL");
297 297 }
298 298 tty->cr();
299 299 }
300 300 }
301 301 // Drop gang mutex.
302 302 }
303 303 if (TraceWorkGang) {
304 304 tty->print("Work for work gang %s id %d task %s part %d",
305 305 gang()->name(), id(), data.task()->name(), part);
306 306 }
307 307 assert(data.task() != NULL, "Got null task");
308 308 data.task()->work(part);
309 309 {
310 310 if (TraceWorkGang) {
311 311 tty->print("Finish for work gang %s id %d task %s part %d",
312 312 gang()->name(), id(), data.task()->name(), part);
313 313 }
314 314 // Grab the gang mutex.
315 315 MutexLocker ml(gang_monitor);
316 316 gang()->internal_note_finish();
317 317 // Tell the gang you are done.
318 318 gang_monitor->notify_all();
319 319 // Drop the gang mutex.
320 320 }
321 321 previous_sequence_number = data.sequence_number();
322 322 }
323 323 }
324 324
325 325 bool GangWorker::is_GC_task_thread() const {
326 326 return gang()->are_GC_task_threads();
327 327 }
328 328
329 329 bool GangWorker::is_ConcurrentGC_thread() const {
330 330 return gang()->are_ConcurrentGC_threads();
331 331 }
332 332
333 333 void GangWorker::print_on(outputStream* st) const {
334 334 st->print("\"%s\" ", name());
335 335 Thread::print_on(st);
336 336 st->cr();
337 337 }
338 338
339 339 // Printing methods
340 340
341 341 const char* AbstractWorkGang::name() const {
342 342 return _name;
343 343 }
344 344
345 345 #ifndef PRODUCT
346 346
347 347 const char* AbstractGangTask::name() const {
348 348 return _name;
349 349 }
350 350
351 351 #endif /* PRODUCT */
352 352
353 353 // *** WorkGangBarrierSync
354 354
355 355 WorkGangBarrierSync::WorkGangBarrierSync()
356 356 : _monitor(Mutex::safepoint, "work gang barrier sync", true),
357 357 _n_workers(0), _n_completed(0), _should_reset(false) {
358 358 }
359 359
360 360 WorkGangBarrierSync::WorkGangBarrierSync(int n_workers, const char* name)
361 361 : _monitor(Mutex::safepoint, name, true),
362 362 _n_workers(n_workers), _n_completed(0), _should_reset(false) {
363 363 }
364 364
365 365 void WorkGangBarrierSync::set_n_workers(int n_workers) {
366 366 _n_workers = n_workers;
367 367 _n_completed = 0;
368 368 _should_reset = false;
369 369 }
370 370
371 371 void WorkGangBarrierSync::enter() {
372 372 MutexLockerEx x(monitor(), Mutex::_no_safepoint_check_flag);
373 373 if (should_reset()) {
374 374 // The should_reset() was set and we are the first worker to enter
375 375 // the sync barrier. We will zero the n_completed() count which
376 376 // effectively resets the barrier.
377 377 zero_completed();
378 378 set_should_reset(false);
379 379 }
380 380 inc_completed();
381 381 if (n_completed() == n_workers()) {
382 382 // At this point we would like to reset the barrier to be ready in
383 383 // case it is used again. However, we cannot set n_completed() to
384 384 // 0, even after the notify_all(), given that some other workers
385 385 // might still be waiting for n_completed() to become ==
386 386 // n_workers(). So, if we set n_completed() to 0, those workers
387 387 // will get stuck (as they will wake up, see that n_completed() !=
388 388 // n_workers() and go back to sleep). Instead, we raise the
389 389 // should_reset() flag and the barrier will be reset the first
390 390 // time a worker enters it again.
391 391 set_should_reset(true);
392 392 monitor()->notify_all();
393 393 } else {
394 394 while (n_completed() != n_workers()) {
395 395 monitor()->wait(/* no_safepoint_check */ true);
396 396 }
397 397 }
398 398 }
399 399
400 400 // SubTasksDone functions.
401 401
402 402 SubTasksDone::SubTasksDone(int n) :
403 403 _n_tasks(n), _n_threads(1), _tasks(NULL) {
404 404 _tasks = NEW_C_HEAP_ARRAY(jint, n);
405 405 guarantee(_tasks != NULL, "alloc failure");
406 406 clear();
407 407 }
408 408
409 409 bool SubTasksDone::valid() {
410 410 return _tasks != NULL;
411 411 }
412 412
413 413 void SubTasksDone::set_n_threads(int t) {
414 414 #ifdef ASSERT
415 415 assert(_claimed == 0 || _threads_completed == _n_threads,
416 416 "should not be called while tasks are being processed!");
417 417 #endif
418 418 _n_threads = (t == 0 ? 1 : t);
419 419 }
420 420
421 421 void SubTasksDone::clear() {
422 422 for (int i = 0; i < _n_tasks; i++) {
423 423 _tasks[i] = 0;
424 424 }
425 425 _threads_completed = 0;
426 426 #ifdef ASSERT
427 427 _claimed = 0;
428 428 #endif
429 429 }
430 430
431 431 bool SubTasksDone::is_task_claimed(int t) {
432 432 assert(0 <= t && t < _n_tasks, "bad task id.");
433 433 jint old = _tasks[t];
434 434 if (old == 0) {
435 435 old = Atomic::cmpxchg(1, &_tasks[t], 0);
436 436 }
437 437 assert(_tasks[t] == 1, "What else?");
438 438 bool res = old != 0;
439 439 #ifdef ASSERT
440 440 if (!res) {
441 441 assert(_claimed < _n_tasks, "Too many tasks claimed; missing clear?");
442 442 Atomic::inc(&_claimed);
443 443 }
444 444 #endif
445 445 return res;
446 446 }
447 447
448 448 void SubTasksDone::all_tasks_completed() {
449 449 jint observed = _threads_completed;
450 450 jint old;
451 451 do {
452 452 old = observed;
453 453 observed = Atomic::cmpxchg(old+1, &_threads_completed, old);
454 454 } while (observed != old);
455 455 // If this was the last thread checking in, clear the tasks.
456 456 if (observed+1 == _n_threads) clear();
457 457 }
458 458
459 459
460 460 SubTasksDone::~SubTasksDone() {
461 461 if (_tasks != NULL) FREE_C_HEAP_ARRAY(jint, _tasks);
462 462 }
463 463
464 464 // *** SequentialSubTasksDone
465 465
466 466 void SequentialSubTasksDone::clear() {
467 467 _n_tasks = _n_claimed = 0;
468 468 _n_threads = _n_completed = 0;
469 469 }
470 470
471 471 bool SequentialSubTasksDone::valid() {
472 472 return _n_threads > 0;
473 473 }
474 474
475 475 bool SequentialSubTasksDone::is_task_claimed(int& t) {
476 476 jint* n_claimed_ptr = &_n_claimed;
477 477 t = *n_claimed_ptr;
478 478 while (t < _n_tasks) {
479 479 jint res = Atomic::cmpxchg(t+1, n_claimed_ptr, t);
480 480 if (res == t) {
481 481 return false;
482 482 }
483 483 t = *n_claimed_ptr;
484 484 }
485 485 return true;
486 486 }
487 487
488 488 bool SequentialSubTasksDone::all_tasks_completed() {
489 489 jint* n_completed_ptr = &_n_completed;
490 490 jint complete = *n_completed_ptr;
491 491 while (true) {
492 492 jint res = Atomic::cmpxchg(complete+1, n_completed_ptr, complete);
493 493 if (res == complete) {
494 494 break;
495 495 }
496 496 complete = res;
497 497 }
498 498 if (complete+1 == _n_threads) {
499 499 clear();
500 500 return true;
501 501 }
502 502 return false;
503 503 }
504 504
505 505 bool FreeIdSet::_stat_init = false;
506 506 FreeIdSet* FreeIdSet::_sets[NSets];
507 507 bool FreeIdSet::_safepoint;
508 508
509 509 FreeIdSet::FreeIdSet(int sz, Monitor* mon) :
510 510 _sz(sz), _mon(mon), _hd(0), _waiters(0), _index(-1), _claimed(0)
511 511 {
512 512 _ids = new int[sz];
513 513 for (int i = 0; i < sz; i++) _ids[i] = i+1;
514 514 _ids[sz-1] = end_of_list; // end of list.
515 515 if (_stat_init) {
516 516 for (int j = 0; j < NSets; j++) _sets[j] = NULL;
517 517 _stat_init = true;
518 518 }
519 519 // Add to sets. (This should happen while the system is still single-threaded.)
520 520 for (int j = 0; j < NSets; j++) {
521 521 if (_sets[j] == NULL) {
522 522 _sets[j] = this;
523 523 _index = j;
524 524 break;
525 525 }
526 526 }
527 527 guarantee(_index != -1, "Too many FreeIdSets in use!");
528 528 }
529 529
530 530 FreeIdSet::~FreeIdSet() {
531 531 _sets[_index] = NULL;
532 532 }
533 533
534 534 void FreeIdSet::set_safepoint(bool b) {
535 535 _safepoint = b;
536 536 if (b) {
537 537 for (int j = 0; j < NSets; j++) {
538 538 if (_sets[j] != NULL && _sets[j]->_waiters > 0) {
539 539 Monitor* mon = _sets[j]->_mon;
540 540 mon->lock_without_safepoint_check();
541 541 mon->notify_all();
542 542 mon->unlock();
543 543 }
544 544 }
545 545 }
546 546 }
547 547
548 548 #define FID_STATS 0
549 549
550 550 int FreeIdSet::claim_par_id() {
551 551 #if FID_STATS
552 552 thread_t tslf = thr_self();
553 553 tty->print("claim_par_id[%d]: sz = %d, claimed = %d\n", tslf, _sz, _claimed);
554 554 #endif
555 555 MutexLockerEx x(_mon, Mutex::_no_safepoint_check_flag);
556 556 while (!_safepoint && _hd == end_of_list) {
557 557 _waiters++;
558 558 #if FID_STATS
559 559 if (_waiters > 5) {
560 560 tty->print("claim_par_id waiting[%d]: %d waiters, %d claimed.\n",
561 561 tslf, _waiters, _claimed);
562 562 }
563 563 #endif
564 564 _mon->wait(Mutex::_no_safepoint_check_flag);
565 565 _waiters--;
566 566 }
567 567 if (_hd == end_of_list) {
568 568 #if FID_STATS
569 569 tty->print("claim_par_id[%d]: returning EOL.\n", tslf);
570 570 #endif
571 571 return -1;
572 572 } else {
573 573 int res = _hd;
574 574 _hd = _ids[res];
575 575 _ids[res] = claimed; // For debugging.
576 576 _claimed++;
577 577 #if FID_STATS
578 578 tty->print("claim_par_id[%d]: returning %d, claimed = %d.\n",
579 579 tslf, res, _claimed);
580 580 #endif
581 581 return res;
582 582 }
583 583 }
584 584
585 585 bool FreeIdSet::claim_perm_id(int i) {
586 586 assert(0 <= i && i < _sz, "Out of range.");
587 587 MutexLockerEx x(_mon, Mutex::_no_safepoint_check_flag);
588 588 int prev = end_of_list;
589 589 int cur = _hd;
590 590 while (cur != end_of_list) {
591 591 if (cur == i) {
592 592 if (prev == end_of_list) {
593 593 _hd = _ids[cur];
594 594 } else {
595 595 _ids[prev] = _ids[cur];
596 596 }
597 597 _ids[cur] = claimed;
598 598 _claimed++;
599 599 return true;
600 600 } else {
601 601 prev = cur;
602 602 cur = _ids[cur];
603 603 }
604 604 }
605 605 return false;
606 606
607 607 }
608 608
609 609 void FreeIdSet::release_par_id(int id) {
610 610 MutexLockerEx x(_mon, Mutex::_no_safepoint_check_flag);
611 611 assert(_ids[id] == claimed, "Precondition.");
612 612 _ids[id] = _hd;
613 613 _hd = id;
614 614 _claimed--;
615 615 #if FID_STATS
616 616 tty->print("[%d] release_par_id(%d), waiters =%d, claimed = %d.\n",
617 617 thr_self(), id, _waiters, _claimed);
618 618 #endif
619 619 if (_waiters > 0)
620 620 // Notify all would be safer, but this is OK, right?
621 621 _mon->notify_all();
622 622 }
|
↓ open down ↓ |
453 lines elided |
↑ open up ↑ |
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX