12 #include "clockvector.h"
13 #include "cyclegraph.h"
16 #include "threads-model.h"
17 #include "bugmessage.h"
19 #define INITIAL_THREAD_ID 0
22 * Structure for holding small ModelChecker members that should be snapshotted
24 struct model_snapshot_members {
25 model_snapshot_members() :
26 /* First thread created will have id INITIAL_THREAD_ID */
27 next_thread_id(INITIAL_THREAD_ID),
28 used_sequence_numbers(0),
31 failed_promise(false),
32 hard_failed_promise(false),
33 too_many_reads(false),
34 no_valid_reads(false),
35 bad_synchronization(false),
40 ~model_snapshot_members() {
41 for (unsigned int i = 0; i < bugs.size(); i++)
46 unsigned int next_thread_id;
47 modelclock_t used_sequence_numbers;
48 ModelAction *next_backtrack;
49 SnapVector<bug_message *> bugs;
51 bool hard_failed_promise;
54 /** @brief Incorrectly-ordered synchronization was made */
55 bool bad_synchronization;
62 /** @brief Constructor */
63 ModelExecution::ModelExecution(ModelChecker *m,
64 const struct model_params *params,
66 NodeStack *node_stack) :
71 thread_map(2), /* We'll always need at least 2 threads */
75 condvar_waiters_map(),
82 thrd_last_fence_release(),
83 node_stack(node_stack),
84 priv(new struct model_snapshot_members()),
85 mo_graph(new CycleGraph())
87 /* Initialize a model-checker thread, for special ModelActions */
88 model_thread = new Thread(get_next_id()); // L: Create model thread
89 add_thread(model_thread); // L: Add model thread to scheduler
90 scheduler->register_engine(this);
91 node_stack->register_engine(this);
94 /** @brief Destructor */
95 ModelExecution::~ModelExecution()
97 for (unsigned int i = 0; i < get_num_threads(); i++)
98 delete get_thread(int_to_id(i));
100 for (unsigned int i = 0; i < promises.size(); i++)
107 int ModelExecution::get_execution_number() const
109 return model->get_execution_number();
112 static action_list_t * get_safe_ptr_action(HashTable<const void *, action_list_t *, uintptr_t, 4> * hash, void * ptr)
114 action_list_t *tmp = hash->get(ptr);
116 tmp = new action_list_t();
122 static SnapVector<action_list_t> * get_safe_ptr_vect_action(HashTable<void *, SnapVector<action_list_t> *, uintptr_t, 4> * hash, void * ptr)
124 SnapVector<action_list_t> *tmp = hash->get(ptr);
126 tmp = new SnapVector<action_list_t>();
132 action_list_t * ModelExecution::get_actions_on_obj(void * obj, thread_id_t tid) const
134 SnapVector<action_list_t> *wrv = obj_thrd_map.get(obj);
137 unsigned int thread=id_to_int(tid);
138 if (thread < wrv->size())
139 return &(*wrv)[thread];
144 /** @return a thread ID for a new Thread */
145 thread_id_t ModelExecution::get_next_id()
147 return priv->next_thread_id++;
150 /** @return the number of user threads created during this execution */
151 unsigned int ModelExecution::get_num_threads() const
153 return priv->next_thread_id;
156 /** @return a sequence number for a new ModelAction */
157 modelclock_t ModelExecution::get_next_seq_num()
159 return ++priv->used_sequence_numbers;
163 * @brief Should the current action wake up a given thread?
165 * @param curr The current action
166 * @param thread The thread that we might wake up
167 * @return True, if we should wake up the sleeping thread; false otherwise
169 bool ModelExecution::should_wake_up(const ModelAction *curr, const Thread *thread) const
171 const ModelAction *asleep = thread->get_pending();
172 /* Don't allow partial RMW to wake anyone up */
175 /* Synchronizing actions may have been backtracked */
176 if (asleep->could_synchronize_with(curr))
178 /* All acquire/release fences and fence-acquire/store-release */
179 if (asleep->is_fence() && asleep->is_acquire() && curr->is_release())
181 /* Fence-release + store can awake load-acquire on the same location */
182 if (asleep->is_read() && asleep->is_acquire() && curr->same_var(asleep) && curr->is_write()) {
183 ModelAction *fence_release = get_last_fence_release(curr->get_tid());
184 if (fence_release && *(get_last_action(thread->get_id())) < *fence_release)
190 void ModelExecution::wake_up_sleeping_actions(ModelAction *curr)
192 for (unsigned int i = 0; i < get_num_threads(); i++) {
193 Thread *thr = get_thread(int_to_id(i));
194 if (scheduler->is_sleep_set(thr)) {
195 if (should_wake_up(curr, thr))
196 /* Remove this thread from sleep set */
197 scheduler->remove_sleep(thr);
202 /** @brief Alert the model-checker that an incorrectly-ordered
203 * synchronization was made */
204 void ModelExecution::set_bad_synchronization()
206 priv->bad_synchronization = true;
209 /** @brief Alert the model-checker that an incorrectly-ordered
210 * synchronization was made */
211 void ModelExecution::set_bad_sc_read()
213 priv->bad_sc_read = true;
216 bool ModelExecution::assert_bug(const char *msg)
218 priv->bugs.push_back(new bug_message(msg));
220 if (isfeasibleprefix()) {
227 /** @return True, if any bugs have been reported for this execution */
228 bool ModelExecution::have_bug_reports() const
230 return priv->bugs.size() != 0;
233 SnapVector<bug_message *> * ModelExecution::get_bugs() const
239 * Check whether the current trace has triggered an assertion which should halt
242 * @return True, if the execution should be aborted; false otherwise
244 bool ModelExecution::has_asserted() const
246 return priv->asserted;
250 * Trigger a trace assertion which should cause this execution to be halted.
251 * This can be due to a detected bug or due to an infeasibility that should
254 void ModelExecution::set_assert()
256 priv->asserted = true;
260 * Check if we are in a deadlock. Should only be called at the end of an
261 * execution, although it should not give false positives in the middle of an
262 * execution (there should be some ENABLED thread).
264 * @return True if program is in a deadlock; false otherwise
266 bool ModelExecution::is_deadlocked() const
268 bool blocking_threads = false;
269 for (unsigned int i = 0; i < get_num_threads(); i++) {
270 thread_id_t tid = int_to_id(i);
273 Thread *t = get_thread(tid);
274 if (!t->is_model_thread() && t->get_pending())
275 blocking_threads = true;
277 return blocking_threads;
281 * @brief Check if we are yield-blocked
283 * A program can be "yield-blocked" if all threads are ready to execute a
286 * @return True if the program is yield-blocked; false otherwise
288 bool ModelExecution::is_yieldblocked() const
290 if (!params->yieldblock)
293 for (unsigned int i = 0; i < get_num_threads(); i++) {
294 thread_id_t tid = int_to_id(i);
295 Thread *t = get_thread(tid);
296 if (t->get_pending() && t->get_pending()->is_yield())
303 * Check if this is a complete execution. That is, have all thread completed
304 * execution (rather than exiting because sleep sets have forced a redundant
307 * @return True if the execution is complete.
309 bool ModelExecution::is_complete_execution() const
311 if (is_yieldblocked())
313 for (unsigned int i = 0; i < get_num_threads(); i++)
314 if (is_enabled(int_to_id(i)))
320 * @brief Find the last fence-related backtracking conflict for a ModelAction
322 * This function performs the search for the most recent conflicting action
323 * against which we should perform backtracking, as affected by fence
324 * operations. This includes pairs of potentially-synchronizing actions which
325 * occur due to fence-acquire or fence-release, and hence should be explored in
326 * the opposite execution order.
328 * @param act The current action
329 * @return The most recent action which conflicts with act due to fences
331 ModelAction * ModelExecution::get_last_fence_conflict(ModelAction *act) const
333 /* Only perform release/acquire fence backtracking for stores */
334 if (!act->is_write())
337 /* Find a fence-release (or, act is a release) */
338 ModelAction *last_release;
339 if (act->is_release())
342 last_release = get_last_fence_release(act->get_tid());
346 /* Skip past the release */
347 const action_list_t *list = &action_trace;
348 action_list_t::const_reverse_iterator rit;
349 for (rit = list->rbegin(); rit != list->rend(); rit++)
350 if (*rit == last_release)
352 ASSERT(rit != list->rend());
357 * load --sb-> fence-acquire */
358 ModelVector<ModelAction *> acquire_fences(get_num_threads(), NULL);
359 ModelVector<ModelAction *> prior_loads(get_num_threads(), NULL);
360 bool found_acquire_fences = false;
361 for ( ; rit != list->rend(); rit++) {
362 ModelAction *prev = *rit;
363 if (act->same_thread(prev))
366 int tid = id_to_int(prev->get_tid());
368 if (prev->is_read() && act->same_var(prev)) {
369 if (prev->is_acquire()) {
370 /* Found most recent load-acquire, don't need
371 * to search for more fences */
372 if (!found_acquire_fences)
375 prior_loads[tid] = prev;
378 if (prev->is_acquire() && prev->is_fence() && !acquire_fences[tid]) {
379 found_acquire_fences = true;
380 acquire_fences[tid] = prev;
384 ModelAction *latest_backtrack = NULL;
385 for (unsigned int i = 0; i < acquire_fences.size(); i++)
386 if (acquire_fences[i] && prior_loads[i])
387 if (!latest_backtrack || *latest_backtrack < *acquire_fences[i])
388 latest_backtrack = acquire_fences[i];
389 return latest_backtrack;
393 * @brief Find the last backtracking conflict for a ModelAction
395 * This function performs the search for the most recent conflicting action
396 * against which we should perform backtracking. This primary includes pairs of
397 * synchronizing actions which should be explored in the opposite execution
400 * @param act The current action
401 * @return The most recent action which conflicts with act
403 ModelAction * ModelExecution::get_last_conflict(ModelAction *act) const
405 switch (act->get_type()) {
407 /* Only seq-cst fences can (directly) cause backtracking */
408 if (!act->is_seqcst())
413 ModelAction *ret = NULL;
415 /* linear search: from most recent to oldest */
416 action_list_t *list = obj_map.get(act->get_location());
417 action_list_t::reverse_iterator rit;
418 for (rit = list->rbegin(); rit != list->rend(); rit++) {
419 ModelAction *prev = *rit;
422 if (prev->could_synchronize_with(act)) {
428 ModelAction *ret2 = get_last_fence_conflict(act);
438 case ATOMIC_TRYLOCK: {
439 /* linear search: from most recent to oldest */
440 action_list_t *list = obj_map.get(act->get_location());
441 action_list_t::reverse_iterator rit;
442 for (rit = list->rbegin(); rit != list->rend(); rit++) {
443 ModelAction *prev = *rit;
444 if (act->is_conflicting_lock(prev))
449 case ATOMIC_UNLOCK: {
450 /* linear search: from most recent to oldest */
451 action_list_t *list = obj_map.get(act->get_location());
452 action_list_t::reverse_iterator rit;
453 for (rit = list->rbegin(); rit != list->rend(); rit++) {
454 ModelAction *prev = *rit;
455 if (!act->same_thread(prev) && prev->is_failed_trylock())
461 /* linear search: from most recent to oldest */
462 action_list_t *list = obj_map.get(act->get_location());
463 action_list_t::reverse_iterator rit;
464 for (rit = list->rbegin(); rit != list->rend(); rit++) {
465 ModelAction *prev = *rit;
466 if (!act->same_thread(prev) && prev->is_failed_trylock())
468 if (!act->same_thread(prev) && prev->is_notify())
474 case ATOMIC_NOTIFY_ALL:
475 case ATOMIC_NOTIFY_ONE: {
476 /* linear search: from most recent to oldest */
477 action_list_t *list = obj_map.get(act->get_location());
478 action_list_t::reverse_iterator rit;
479 for (rit = list->rbegin(); rit != list->rend(); rit++) {
480 ModelAction *prev = *rit;
481 if (!act->same_thread(prev) && prev->is_wait())
492 /** This method finds backtracking points where we should try to
493 * reorder the parameter ModelAction against.
495 * @param the ModelAction to find backtracking points for.
497 void ModelExecution::set_backtracking(ModelAction *act)
499 Thread *t = get_thread(act);
500 ModelAction *prev = get_last_conflict(act);
504 Node *node = prev->get_node()->get_parent();
506 /* See Dynamic Partial Order Reduction (addendum), POPL '05 */
507 int low_tid, high_tid;
508 if (node->enabled_status(t->get_id()) == THREAD_ENABLED) {
509 low_tid = id_to_int(act->get_tid());
510 high_tid = low_tid + 1;
513 high_tid = get_num_threads();
516 for (int i = low_tid; i < high_tid; i++) {
517 thread_id_t tid = int_to_id(i);
519 /* Make sure this thread can be enabled here. */
520 if (i >= node->get_num_threads())
523 /* See Dynamic Partial Order Reduction (addendum), POPL '05 */
524 /* Don't backtrack into a point where the thread is disabled or sleeping. */
525 if (node->enabled_status(tid) != THREAD_ENABLED)
528 /* Check if this has been explored already */
529 if (node->has_been_explored(tid))
532 /* See if fairness allows */
533 if (params->fairwindow != 0 && !node->has_priority(tid)) {
535 for (int t = 0; t < node->get_num_threads(); t++) {
536 thread_id_t tother = int_to_id(t);
537 if (node->is_enabled(tother) && node->has_priority(tother)) {
546 /* See if CHESS-like yield fairness allows */
547 if (params->yieldon) {
549 for (int t = 0; t < node->get_num_threads(); t++) {
550 thread_id_t tother = int_to_id(t);
551 if (node->is_enabled(tother) && node->has_priority_over(tid, tother)) {
560 /* Cache the latest backtracking point */
561 set_latest_backtrack(prev);
563 /* If this is a new backtracking point, mark the tree */
564 if (!node->set_backtrack(tid))
566 DEBUG("Setting backtrack: conflict = %d, instead tid = %d\n",
567 id_to_int(prev->get_tid()),
568 id_to_int(t->get_id()));
577 * @brief Cache the a backtracking point as the "most recent", if eligible
579 * Note that this does not prepare the NodeStack for this backtracking
580 * operation, it only caches the action on a per-execution basis
582 * @param act The operation at which we should explore a different next action
583 * (i.e., backtracking point)
584 * @return True, if this action is now the most recent backtracking point;
587 bool ModelExecution::set_latest_backtrack(ModelAction *act)
589 if (!priv->next_backtrack || *act > *priv->next_backtrack) {
590 priv->next_backtrack = act;
597 * Returns last backtracking point. The model checker will explore a different
598 * path for this point in the next execution.
599 * @return The ModelAction at which the next execution should diverge.
601 ModelAction * ModelExecution::get_next_backtrack()
603 ModelAction *next = priv->next_backtrack;
604 priv->next_backtrack = NULL;
609 * Processes a read model action.
610 * @param curr is the read model action to process.
611 * @return True if processing this read updates the mo_graph.
613 bool ModelExecution::process_read(ModelAction *curr)
615 Node *node = curr->get_node();
617 bool updated = false;
618 switch (node->get_read_from_status()) {
619 case READ_FROM_PAST: {
620 const ModelAction *rf = node->get_read_from_past();
623 mo_graph->startChanges();
625 ASSERT(!is_infeasible());
626 if (!check_recency(curr, rf)) {
627 if (node->increment_read_from()) {
628 mo_graph->rollbackChanges();
631 priv->too_many_reads = true;
635 updated = r_modification_order(curr, rf);
637 mo_graph->commitChanges();
638 mo_check_promises(curr, true);
641 case READ_FROM_PROMISE: {
642 Promise *promise = curr->get_node()->get_read_from_promise();
643 if (promise->add_reader(curr))
644 priv->failed_promise = true;
645 curr->set_read_from_promise(promise);
646 mo_graph->startChanges();
647 if (!check_recency(curr, promise))
648 priv->too_many_reads = true;
649 updated = r_modification_order(curr, promise);
650 mo_graph->commitChanges();
653 case READ_FROM_FUTURE: {
654 /* Read from future value */
655 struct future_value fv = node->get_future_value();
656 Promise *promise = new Promise(this, curr, fv);
657 curr->set_read_from_promise(promise);
658 promises.push_back(promise);
659 mo_graph->startChanges();
660 updated = r_modification_order(curr, promise);
661 mo_graph->commitChanges();
667 get_thread(curr)->set_return_value(curr->get_return_value());
673 * Processes a lock, trylock, or unlock model action. @param curr is
674 * the read model action to process.
676 * The try lock operation checks whether the lock is taken. If not,
677 * it falls to the normal lock operation case. If so, it returns
680 * The lock operation has already been checked that it is enabled, so
681 * it just grabs the lock and synchronizes with the previous unlock.
683 * The unlock operation has to re-enable all of the threads that are
684 * waiting on the lock.
686 * @return True if synchronization was updated; false otherwise
688 bool ModelExecution::process_mutex(ModelAction *curr)
690 std::mutex *mutex = curr->get_mutex();
691 struct std::mutex_state *state = NULL;
694 state = mutex->get_state();
696 switch (curr->get_type()) {
697 case ATOMIC_TRYLOCK: {
698 bool success = !state->locked;
699 curr->set_try_lock(success);
701 get_thread(curr)->set_return_value(0);
704 get_thread(curr)->set_return_value(1);
706 //otherwise fall into the lock case
708 if (curr->get_cv()->getClock(state->alloc_tid) <= state->alloc_clock)
709 assert_bug("Lock access before initialization");
710 state->locked = get_thread(curr);
711 ModelAction *unlock = get_last_unlock(curr);
712 //synchronize with the previous unlock statement
713 if (unlock != NULL) {
714 synchronize(unlock, curr);
720 case ATOMIC_UNLOCK: {
721 /* wake up the other threads */
722 for (unsigned int i = 0; i < get_num_threads(); i++) {
723 Thread *t = get_thread(int_to_id(i));
724 Thread *curr_thrd = get_thread(curr);
725 if (t->waiting_on() == curr_thrd && t->get_pending()->is_lock())
729 /* unlock the lock - after checking who was waiting on it */
730 state->locked = NULL;
732 if (!curr->is_wait())
733 break; /* The rest is only for ATOMIC_WAIT */
735 /* Should we go to sleep? (simulate spurious failures) */
736 if (curr->get_node()->get_misc() == 0) {
737 get_safe_ptr_action(&condvar_waiters_map, curr->get_location())->push_back(curr);
739 scheduler->sleep(get_thread(curr));
743 case ATOMIC_NOTIFY_ALL: {
744 action_list_t *waiters = get_safe_ptr_action(&condvar_waiters_map, curr->get_location());
745 //activate all the waiting threads
746 for (action_list_t::iterator rit = waiters->begin(); rit != waiters->end(); rit++) {
747 scheduler->wake(get_thread(*rit));
752 case ATOMIC_NOTIFY_ONE: {
753 action_list_t *waiters = get_safe_ptr_action(&condvar_waiters_map, curr->get_location());
754 int wakeupthread = curr->get_node()->get_misc();
755 action_list_t::iterator it = waiters->begin();
756 advance(it, wakeupthread);
757 scheduler->wake(get_thread(*it));
769 * @brief Check if the current pending promises allow a future value to be sent
771 * It is unsafe to pass a future value back if there exists a pending promise Pr
774 * reader --exec-> Pr --exec-> writer
776 * If such Pr exists, we must save the pending future value until Pr is
779 * @param writer The operation which sends the future value. Must be a write.
780 * @param reader The operation which will observe the value. Must be a read.
781 * @return True if the future value can be sent now; false if it must wait.
783 bool ModelExecution::promises_may_allow(const ModelAction *writer,
784 const ModelAction *reader) const
786 for (int i = promises.size() - 1; i >= 0; i--) {
787 ModelAction *pr = promises[i]->get_reader(0);
788 //reader is after promise...doesn't cross any promise
791 //writer is after promise, reader before...bad...
799 * @brief Add a future value to a reader
801 * This function performs a few additional checks to ensure that the future
802 * value can be feasibly observed by the reader
804 * @param writer The operation whose value is sent. Must be a write.
805 * @param reader The read operation which may read the future value. Must be a read.
807 void ModelExecution::add_future_value(const ModelAction *writer, ModelAction *reader)
809 /* Do more ambitious checks now that mo is more complete */
810 if (!mo_may_allow(writer, reader))
813 Node *node = reader->get_node();
815 /* Find an ancestor thread which exists at the time of the reader */
816 Thread *write_thread = get_thread(writer);
817 while (id_to_int(write_thread->get_id()) >= node->get_num_threads())
818 write_thread = write_thread->get_parent();
820 struct future_value fv = {
821 writer->get_write_value(),
822 writer->get_seq_number() + params->maxfuturedelay,
823 write_thread->get_id(),
825 if (node->add_future_value(fv))
826 set_latest_backtrack(reader);
830 * Process a write ModelAction
831 * @param curr The ModelAction to process
832 * @param work The work queue, for adding fixup work
833 * @return True if the mo_graph was updated or promises were resolved
835 bool ModelExecution::process_write(ModelAction *curr, work_queue_t *work)
837 /* Readers to which we may send our future value */
838 ModelVector<ModelAction *> send_fv;
840 const ModelAction *earliest_promise_reader;
841 bool updated_promises = false;
843 bool updated_mod_order = w_modification_order(curr, &send_fv);
844 Promise *promise = pop_promise_to_resolve(curr);
847 earliest_promise_reader = promise->get_reader(0);
848 updated_promises = resolve_promise(curr, promise, work);
850 earliest_promise_reader = NULL;
852 for (unsigned int i = 0; i < send_fv.size(); i++) {
853 ModelAction *read = send_fv[i];
855 /* Don't send future values to reads after the Promise we resolve */
856 if (!earliest_promise_reader || *read < *earliest_promise_reader) {
857 /* Check if future value can be sent immediately */
858 if (promises_may_allow(curr, read)) {
859 add_future_value(curr, read);
861 futurevalues.push_back(PendingFutureValue(curr, read));
866 /* Check the pending future values */
867 for (int i = (int)futurevalues.size() - 1; i >= 0; i--) {
868 struct PendingFutureValue pfv = futurevalues[i];
869 if (promises_may_allow(pfv.writer, pfv.reader)) {
870 add_future_value(pfv.writer, pfv.reader);
871 futurevalues.erase(futurevalues.begin() + i);
875 mo_graph->commitChanges();
876 mo_check_promises(curr, false);
878 get_thread(curr)->set_return_value(VALUE_NONE);
879 return updated_mod_order || updated_promises;
883 * Process a fence ModelAction
884 * @param curr The ModelAction to process
885 * @return True if synchronization was updated
887 bool ModelExecution::process_fence(ModelAction *curr)
890 * fence-relaxed: no-op
891 * fence-release: only log the occurence (not in this function), for
892 * use in later synchronization
893 * fence-acquire (this function): search for hypothetical release
895 * fence-seq-cst: MO constraints formed in {r,w}_modification_order
897 bool updated = false;
898 if (curr->is_acquire()) {
899 action_list_t *list = &action_trace;
900 action_list_t::reverse_iterator rit;
901 /* Find X : is_read(X) && X --sb-> curr */
902 for (rit = list->rbegin(); rit != list->rend(); rit++) {
903 ModelAction *act = *rit;
906 if (act->get_tid() != curr->get_tid())
908 /* Stop at the beginning of the thread */
909 if (act->is_thread_start())
911 /* Stop once we reach a prior fence-acquire */
912 if (act->is_fence() && act->is_acquire())
916 /* read-acquire will find its own release sequences */
917 if (act->is_acquire())
920 /* Establish hypothetical release sequences */
921 rel_heads_list_t release_heads;
922 get_release_seq_heads(curr, act, &release_heads);
923 for (unsigned int i = 0; i < release_heads.size(); i++)
924 synchronize(release_heads[i], curr);
925 if (release_heads.size() != 0)
933 * @brief Process the current action for thread-related activity
935 * Performs current-action processing for a THREAD_* ModelAction. Proccesses
936 * may include setting Thread status, completing THREAD_FINISH/THREAD_JOIN
937 * synchronization, etc. This function is a no-op for non-THREAD actions
938 * (e.g., ATOMIC_{READ,WRITE,RMW,LOCK}, etc.)
940 * @param curr The current action
941 * @return True if synchronization was updated or a thread completed
943 bool ModelExecution::process_thread_action(ModelAction *curr)
945 bool updated = false;
947 switch (curr->get_type()) {
948 case THREAD_CREATE: {
949 thrd_t *thrd = (thrd_t *)curr->get_location();
950 struct thread_params *params = (struct thread_params *)curr->get_value();
951 Thread *th = new Thread(get_next_id(), thrd, params->func, params->arg, get_thread(curr));
952 curr->set_thread_operand(th);
954 th->set_creation(curr);
955 /* Promises can be satisfied by children */
956 for (unsigned int i = 0; i < promises.size(); i++) {
957 Promise *promise = promises[i];
958 if (promise->thread_is_available(curr->get_tid()))
959 promise->add_thread(th->get_id());
963 case PTHREAD_CREATE: {
964 (*(pthread_t *)curr->get_location()) = pthread_counter++;
966 struct pthread_params *params = (struct pthread_params *)curr->get_value();
967 Thread *th = new Thread(get_next_id(), NULL, params->func, params->arg, get_thread(curr));
968 curr->set_thread_operand(th);
970 th->set_creation(curr);
972 if ( pthread_map.size() < pthread_counter )
973 pthread_map.resize( pthread_counter );
974 pthread_map[ pthread_counter-1 ] = th;
976 /* Promises can be satisfied by children */
977 for (unsigned int i = 0; i < promises.size(); i++) {
978 Promise *promise = promises[i];
979 if (promise->thread_is_available(curr->get_tid()))
980 promise->add_thread(th->get_id());
986 Thread *blocking = curr->get_thread_operand();
987 ModelAction *act = get_last_action(blocking->get_id());
988 synchronize(act, curr);
989 updated = true; /* trigger rel-seq checks */
993 Thread *blocking = curr->get_thread_operand();
994 ModelAction *act = get_last_action(blocking->get_id());
995 synchronize(act, curr);
996 updated = true; /* trigger rel-seq checks */
997 break; // WL: to be add (modified)
1000 case THREAD_FINISH: {
1001 Thread *th = get_thread(curr);
1002 /* Wake up any joining threads */
1003 for (unsigned int i = 0; i < get_num_threads(); i++) {
1004 Thread *waiting = get_thread(int_to_id(i));
1005 if (waiting->waiting_on() == th &&
1006 waiting->get_pending()->is_thread_join())
1007 scheduler->wake(waiting);
1010 /* Completed thread can't satisfy promises */
1011 for (unsigned int i = 0; i < promises.size(); i++) {
1012 Promise *promise = promises[i];
1013 if (promise->thread_is_available(th->get_id()))
1014 if (promise->eliminate_thread(th->get_id()))
1015 priv->failed_promise = true;
1017 updated = true; /* trigger rel-seq checks */
1020 case THREAD_START: {
1021 check_promises(curr->get_tid(), NULL, curr->get_cv());
1032 * @brief Process the current action for release sequence fixup activity
1034 * Performs model-checker release sequence fixups for the current action,
1035 * forcing a single pending release sequence to break (with a given, potential
1036 * "loose" write) or to complete (i.e., synchronize). If a pending release
1037 * sequence forms a complete release sequence, then we must perform the fixup
1038 * synchronization, mo_graph additions, etc.
1040 * @param curr The current action; must be a release sequence fixup action
1041 * @param work_queue The work queue to which to add work items as they are
1044 void ModelExecution::process_relseq_fixup(ModelAction *curr, work_queue_t *work_queue)
1046 const ModelAction *write = curr->get_node()->get_relseq_break();
1047 struct release_seq *sequence = pending_rel_seqs.back();
1048 pending_rel_seqs.pop_back();
1050 ModelAction *acquire = sequence->acquire;
1051 const ModelAction *rf = sequence->rf;
1052 const ModelAction *release = sequence->release;
1056 ASSERT(release->same_thread(rf));
1058 if (write == NULL) {
1060 * @todo Forcing a synchronization requires that we set
1061 * modification order constraints. For instance, we can't allow
1062 * a fixup sequence in which two separate read-acquire
1063 * operations read from the same sequence, where the first one
1064 * synchronizes and the other doesn't. Essentially, we can't
1065 * allow any writes to insert themselves between 'release' and
1069 /* Must synchronize */
1070 if (!synchronize(release, acquire))
1073 /* Propagate the changed clock vector */
1074 propagate_clockvector(acquire, work_queue);
1076 /* Break release sequence with new edges:
1077 * release --mo--> write --mo--> rf */
1078 mo_graph->addEdge(release, write);
1079 mo_graph->addEdge(write, rf);
1082 /* See if we have realized a data race */
1087 * Initialize the current action by performing one or more of the following
1088 * actions, as appropriate: merging RMWR and RMWC/RMW actions, stepping forward
1089 * in the NodeStack, manipulating backtracking sets, allocating and
1090 * initializing clock vectors, and computing the promises to fulfill.
1092 * @param curr The current action, as passed from the user context; may be
1093 * freed/invalidated after the execution of this function, with a different
1094 * action "returned" its place (pass-by-reference)
1095 * @return True if curr is a newly-explored action; false otherwise
1097 bool ModelExecution::initialize_curr_action(ModelAction **curr)
1099 ModelAction *newcurr;
1101 if ((*curr)->is_rmwc() || (*curr)->is_rmw()) {
1102 newcurr = process_rmw(*curr);
1105 if (newcurr->is_rmw())
1106 compute_promises(newcurr);
1112 (*curr)->set_seq_number(get_next_seq_num());
1114 newcurr = node_stack->explore_action(*curr, scheduler->get_enabled_array());
1116 /* First restore type and order in case of RMW operation */
1117 if ((*curr)->is_rmwr())
1118 newcurr->copy_typeandorder(*curr);
1120 ASSERT((*curr)->get_location() == newcurr->get_location());
1121 newcurr->copy_from_new(*curr);
1123 /* Discard duplicate ModelAction; use action from NodeStack */
1126 /* Always compute new clock vector */
1127 newcurr->create_cv(get_parent_action(newcurr->get_tid()));
1130 return false; /* Action was explored previously */
1134 /* Always compute new clock vector */
1135 newcurr->create_cv(get_parent_action(newcurr->get_tid()));
1137 /* Assign most recent release fence */
1138 newcurr->set_last_fence_release(get_last_fence_release(newcurr->get_tid()));
1141 * Perform one-time actions when pushing new ModelAction onto
1144 if (newcurr->is_write())
1145 compute_promises(newcurr);
1146 else if (newcurr->is_relseq_fixup())
1147 compute_relseq_breakwrites(newcurr);
1148 else if (newcurr->is_wait())
1149 newcurr->get_node()->set_misc_max(2);
1150 else if (newcurr->is_notify_one()) {
1151 newcurr->get_node()->set_misc_max(get_safe_ptr_action(&condvar_waiters_map, newcurr->get_location())->size());
1153 return true; /* This was a new ModelAction */
1158 * @brief Establish reads-from relation between two actions
1160 * Perform basic operations involved with establishing a concrete rf relation,
1161 * including setting the ModelAction data and checking for release sequences.
1163 * @param act The action that is reading (must be a read)
1164 * @param rf The action from which we are reading (must be a write)
1166 * @return True if this read established synchronization
1169 bool ModelExecution::read_from(ModelAction *act, const ModelAction *rf)
1172 ASSERT(rf->is_write());
1174 act->set_read_from(rf);
1175 if (act->is_acquire()) {
1176 rel_heads_list_t release_heads;
1177 get_release_seq_heads(act, act, &release_heads);
1178 int num_heads = release_heads.size();
1179 for (unsigned int i = 0; i < release_heads.size(); i++)
1180 if (!synchronize(release_heads[i], act))
1182 return num_heads > 0;
1188 * @brief Synchronizes two actions
1190 * When A synchronizes with B (or A --sw-> B), B inherits A's clock vector.
1191 * This function performs the synchronization as well as providing other hooks
1192 * for other checks along with synchronization.
1194 * @param first The left-hand side of the synchronizes-with relation
1195 * @param second The right-hand side of the synchronizes-with relation
1196 * @return True if the synchronization was successful (i.e., was consistent
1197 * with the execution order); false otherwise
1199 bool ModelExecution::synchronize(const ModelAction *first, ModelAction *second)
1201 if (*second < *first) {
1202 set_bad_synchronization();
1205 check_promises(first->get_tid(), second->get_cv(), first->get_cv());
1206 return second->synchronize_with(first);
1210 * Check promises and eliminate potentially-satisfying threads when a thread is
1211 * blocked (e.g., join, lock). A thread which is waiting on another thread can
1212 * no longer satisfy a promise generated from that thread.
1214 * @param blocker The thread on which a thread is waiting
1215 * @param waiting The waiting thread
1217 void ModelExecution::thread_blocking_check_promises(Thread *blocker, Thread *waiting)
1219 for (unsigned int i = 0; i < promises.size(); i++) {
1220 Promise *promise = promises[i];
1221 if (!promise->thread_is_available(waiting->get_id()))
1223 for (unsigned int j = 0; j < promise->get_num_readers(); j++) {
1224 ModelAction *reader = promise->get_reader(j);
1225 if (reader->get_tid() != blocker->get_id())
1227 if (promise->eliminate_thread(waiting->get_id())) {
1228 /* Promise has failed */
1229 priv->failed_promise = true;
1231 /* Only eliminate the 'waiting' thread once */
1239 * @brief Check whether a model action is enabled.
1241 * Checks whether an operation would be successful (i.e., is a lock already
1242 * locked, or is the joined thread already complete).
1244 * For yield-blocking, yields are never enabled.
1246 * @param curr is the ModelAction to check whether it is enabled.
1247 * @return a bool that indicates whether the action is enabled.
1249 bool ModelExecution::check_action_enabled(ModelAction *curr) {
1250 if (curr->is_lock()) {
1251 std::mutex *lock = curr->get_mutex();
1252 struct std::mutex_state *state = lock->get_state();
1255 } else if (curr->is_thread_join()) {
1256 Thread *blocking = curr->get_thread_operand();
1257 if (!blocking->is_complete()) {
1258 thread_blocking_check_promises(blocking, get_thread(curr));
1261 } else if (params->yieldblock && curr->is_yield()) {
1269 * This is the heart of the model checker routine. It performs model-checking
1270 * actions corresponding to a given "current action." Among other processes, it
1271 * calculates reads-from relationships, updates synchronization clock vectors,
1272 * forms a memory_order constraints graph, and handles replay/backtrack
1273 * execution when running permutations of previously-observed executions.
1275 * @param curr The current action to process
1276 * @return The ModelAction that is actually executed; may be different than
1279 ModelAction * ModelExecution::check_current_action(ModelAction *curr)
1282 bool second_part_of_rmw = curr->is_rmwc() || curr->is_rmw();
1283 bool newly_explored = initialize_curr_action(&curr);
1287 wake_up_sleeping_actions(curr);
1289 /* Compute fairness information for CHESS yield algorithm */
1290 if (params->yieldon) {
1291 curr->get_node()->update_yield(scheduler);
1294 /* Add the action to lists before any other model-checking tasks */
1295 if (!second_part_of_rmw)
1296 add_action_to_lists(curr);
1298 /* Build may_read_from set for newly-created actions */
1299 if (newly_explored && curr->is_read())
1300 build_may_read_from(curr);
1302 /* Initialize work_queue with the "current action" work */
1303 work_queue_t work_queue(1, CheckCurrWorkEntry(curr));
1304 while (!work_queue.empty() && !has_asserted()) {
1305 WorkQueueEntry work = work_queue.front();
1306 work_queue.pop_front();
1308 switch (work.type) {
1309 case WORK_CHECK_CURR_ACTION: {
1310 ModelAction *act = work.action;
1311 bool update = false; /* update this location's release seq's */
1312 bool update_all = false; /* update all release seq's */
1314 if (process_thread_action(curr))
1317 if (act->is_read() && !second_part_of_rmw && process_read(act))
1320 if (act->is_write() && process_write(act, &work_queue))
1323 if (act->is_fence() && process_fence(act))
1326 if (act->is_mutex_op() && process_mutex(act))
1329 if (act->is_relseq_fixup())
1330 process_relseq_fixup(curr, &work_queue);
1333 work_queue.push_back(CheckRelSeqWorkEntry(NULL));
1335 work_queue.push_back(CheckRelSeqWorkEntry(act->get_location()));
1338 case WORK_CHECK_RELEASE_SEQ:
1339 resolve_release_sequences(work.location, &work_queue);
1341 case WORK_CHECK_MO_EDGES: {
1342 /** @todo Complete verification of work_queue */
1343 ModelAction *act = work.action;
1344 bool updated = false;
1346 if (act->is_read()) {
1347 const ModelAction *rf = act->get_reads_from();
1348 const Promise *promise = act->get_reads_from_promise();
1350 if (r_modification_order(act, rf))
1352 if (act->is_seqcst()) {
1353 ModelAction *last_sc_write = get_last_seq_cst_write(act);
1354 if (last_sc_write != NULL && rf->happens_before(last_sc_write)) {
1358 } else if (promise) {
1359 if (r_modification_order(act, promise))
1363 if (act->is_write()) {
1364 if (w_modification_order(act, NULL))
1367 mo_graph->commitChanges();
1370 work_queue.push_back(CheckRelSeqWorkEntry(act->get_location()));
1379 check_curr_backtracking(curr);
1380 set_backtracking(curr);
1384 void ModelExecution::check_curr_backtracking(ModelAction *curr)
1386 Node *currnode = curr->get_node();
1387 Node *parnode = currnode->get_parent();
1389 if ((parnode && !parnode->backtrack_empty()) ||
1390 !currnode->misc_empty() ||
1391 !currnode->read_from_empty() ||
1392 !currnode->promise_empty() ||
1393 !currnode->relseq_break_empty()) {
1394 set_latest_backtrack(curr);
1398 bool ModelExecution::promises_expired() const
1400 for (unsigned int i = 0; i < promises.size(); i++) {
1401 Promise *promise = promises[i];
1402 if (promise->get_expiration() < priv->used_sequence_numbers)
1409 * This is the strongest feasibility check available.
1410 * @return whether the current trace (partial or complete) must be a prefix of
1413 bool ModelExecution::isfeasibleprefix() const
1415 return pending_rel_seqs.size() == 0 && is_feasible_prefix_ignore_relseq();
1419 * Print disagnostic information about an infeasible execution
1420 * @param prefix A string to prefix the output with; if NULL, then a default
1421 * message prefix will be provided
1423 void ModelExecution::print_infeasibility(const char *prefix) const
1427 if (mo_graph->checkForCycles())
1428 ptr += sprintf(ptr, "[mo cycle]");
1429 if (priv->failed_promise || priv->hard_failed_promise)
1430 ptr += sprintf(ptr, "[failed promise]");
1431 if (priv->too_many_reads)
1432 ptr += sprintf(ptr, "[too many reads]");
1433 if (priv->no_valid_reads)
1434 ptr += sprintf(ptr, "[no valid reads-from]");
1435 if (priv->bad_synchronization)
1436 ptr += sprintf(ptr, "[bad sw ordering]");
1437 if (priv->bad_sc_read)
1438 ptr += sprintf(ptr, "[bad sc read]");
1439 if (promises_expired())
1440 ptr += sprintf(ptr, "[promise expired]");
1441 if (promises.size() != 0)
1442 ptr += sprintf(ptr, "[unresolved promise]");
1444 model_print("%s: %s", prefix ? prefix : "Infeasible", buf);
1448 * Returns whether the current completed trace is feasible, except for pending
1449 * release sequences.
1451 bool ModelExecution::is_feasible_prefix_ignore_relseq() const
1453 return !is_infeasible() && promises.size() == 0 && ! priv->failed_promise;
1458 * Check if the current partial trace is infeasible. Does not check any
1459 * end-of-execution flags, which might rule out the execution. Thus, this is
1460 * useful only for ruling an execution as infeasible.
1461 * @return whether the current partial trace is infeasible.
1463 bool ModelExecution::is_infeasible() const
1465 return mo_graph->checkForCycles() ||
1466 priv->no_valid_reads ||
1467 priv->too_many_reads ||
1468 priv->bad_synchronization ||
1469 priv->bad_sc_read ||
1470 priv->hard_failed_promise ||
1474 /** Close out a RMWR by converting previous RMWR into a RMW or READ. */
1475 ModelAction * ModelExecution::process_rmw(ModelAction *act) {
1476 ModelAction *lastread = get_last_action(act->get_tid());
1477 lastread->process_rmw(act);
1478 if (act->is_rmw()) {
1479 if (lastread->get_reads_from())
1480 mo_graph->addRMWEdge(lastread->get_reads_from(), lastread);
1482 mo_graph->addRMWEdge(lastread->get_reads_from_promise(), lastread);
1483 mo_graph->commitChanges();
1489 * A helper function for ModelExecution::check_recency, to check if the current
1490 * thread is able to read from a different write/promise for 'params.maxreads'
1491 * number of steps and if that write/promise should become visible (i.e., is
1492 * ordered later in the modification order). This helps model memory liveness.
1494 * @param curr The current action. Must be a read.
1495 * @param rf The write/promise from which we plan to read
1496 * @param other_rf The write/promise from which we may read
1497 * @return True if we were able to read from other_rf for params.maxreads steps
1499 template <typename T, typename U>
1500 bool ModelExecution::should_read_instead(const ModelAction *curr, const T *rf, const U *other_rf) const
1502 /* Need a different write/promise */
1503 if (other_rf->equals(rf))
1506 /* Only look for "newer" writes/promises */
1507 if (!mo_graph->checkReachable(rf, other_rf))
1510 SnapVector<action_list_t> *thrd_lists = obj_thrd_map.get(curr->get_location());
1511 action_list_t *list = &(*thrd_lists)[id_to_int(curr->get_tid())];
1512 action_list_t::reverse_iterator rit = list->rbegin();
1513 ASSERT((*rit) == curr);
1514 /* Skip past curr */
1517 /* Does this write/promise work for everyone? */
1518 for (int i = 0; i < params->maxreads; i++, rit++) {
1519 ModelAction *act = *rit;
1520 if (!act->may_read_from(other_rf))
1527 * Checks whether a thread has read from the same write or Promise for too many
1528 * times without seeing the effects of a later write/Promise.
1531 * 1) there must a different write/promise that we could read from,
1532 * 2) we must have read from the same write/promise in excess of maxreads times,
1533 * 3) that other write/promise must have been in the reads_from set for maxreads times, and
1534 * 4) that other write/promise must be mod-ordered after the write/promise we are reading.
1536 * If so, we decide that the execution is no longer feasible.
1538 * @param curr The current action. Must be a read.
1539 * @param rf The ModelAction/Promise from which we might read.
1540 * @return True if the read should succeed; false otherwise
1542 template <typename T>
1543 bool ModelExecution::check_recency(ModelAction *curr, const T *rf) const
1545 if (!params->maxreads)
1548 //NOTE: Next check is just optimization, not really necessary....
1549 if (curr->get_node()->get_read_from_past_size() +
1550 curr->get_node()->get_read_from_promise_size() <= 1)
1553 SnapVector<action_list_t> *thrd_lists = obj_thrd_map.get(curr->get_location());
1554 int tid = id_to_int(curr->get_tid());
1555 ASSERT(tid < (int)thrd_lists->size());
1556 action_list_t *list = &(*thrd_lists)[tid];
1557 action_list_t::reverse_iterator rit = list->rbegin();
1558 ASSERT((*rit) == curr);
1559 /* Skip past curr */
1562 action_list_t::reverse_iterator ritcopy = rit;
1563 /* See if we have enough reads from the same value */
1564 for (int count = 0; count < params->maxreads; ritcopy++, count++) {
1565 if (ritcopy == list->rend())
1567 ModelAction *act = *ritcopy;
1568 if (!act->is_read())
1570 if (act->get_reads_from_promise() && !act->get_reads_from_promise()->equals(rf))
1572 if (act->get_reads_from() && !act->get_reads_from()->equals(rf))
1574 if (act->get_node()->get_read_from_past_size() +
1575 act->get_node()->get_read_from_promise_size() <= 1)
1578 for (int i = 0; i < curr->get_node()->get_read_from_past_size(); i++) {
1579 const ModelAction *write = curr->get_node()->get_read_from_past(i);
1580 if (should_read_instead(curr, rf, write))
1581 return false; /* liveness failure */
1583 for (int i = 0; i < curr->get_node()->get_read_from_promise_size(); i++) {
1584 const Promise *promise = curr->get_node()->get_read_from_promise(i);
1585 if (should_read_instead(curr, rf, promise))
1586 return false; /* liveness failure */
1592 * @brief Updates the mo_graph with the constraints imposed from the current
1595 * Basic idea is the following: Go through each other thread and find
1596 * the last action that happened before our read. Two cases:
1598 * -# The action is a write: that write must either occur before
1599 * the write we read from or be the write we read from.
1600 * -# The action is a read: the write that that action read from
1601 * must occur before the write we read from or be the same write.
1603 * @param curr The current action. Must be a read.
1604 * @param rf The ModelAction or Promise that curr reads from. Must be a write.
1605 * @return True if modification order edges were added; false otherwise
1607 template <typename rf_type>
1608 bool ModelExecution::r_modification_order(ModelAction *curr, const rf_type *rf)
1610 SnapVector<action_list_t> *thrd_lists = obj_thrd_map.get(curr->get_location());
1613 ASSERT(curr->is_read());
1615 /* Last SC fence in the current thread */
1616 ModelAction *last_sc_fence_local = get_last_seq_cst_fence(curr->get_tid(), NULL);
1617 ModelAction *last_sc_write = NULL;
1618 if (curr->is_seqcst())
1619 last_sc_write = get_last_seq_cst_write(curr);
1621 /* Iterate over all threads */
1622 for (i = 0; i < thrd_lists->size(); i++) {
1623 /* Last SC fence in thread i */
1624 ModelAction *last_sc_fence_thread_local = NULL;
1625 if (int_to_id((int)i) != curr->get_tid())
1626 last_sc_fence_thread_local = get_last_seq_cst_fence(int_to_id(i), NULL);
1628 /* Last SC fence in thread i, before last SC fence in current thread */
1629 ModelAction *last_sc_fence_thread_before = NULL;
1630 if (last_sc_fence_local)
1631 last_sc_fence_thread_before = get_last_seq_cst_fence(int_to_id(i), last_sc_fence_local);
1633 /* Iterate over actions in thread, starting from most recent */
1634 action_list_t *list = &(*thrd_lists)[i];
1635 action_list_t::reverse_iterator rit;
1636 for (rit = list->rbegin(); rit != list->rend(); rit++) {
1637 ModelAction *act = *rit;
1642 /* Don't want to add reflexive edges on 'rf' */
1643 if (act->equals(rf)) {
1644 if (act->happens_before(curr))
1650 if (act->is_write()) {
1651 /* C++, Section 29.3 statement 5 */
1652 if (curr->is_seqcst() && last_sc_fence_thread_local &&
1653 *act < *last_sc_fence_thread_local) {
1654 added = mo_graph->addEdge(act, rf) || added;
1657 /* C++, Section 29.3 statement 4 */
1658 else if (act->is_seqcst() && last_sc_fence_local &&
1659 *act < *last_sc_fence_local) {
1660 added = mo_graph->addEdge(act, rf) || added;
1663 /* C++, Section 29.3 statement 6 */
1664 else if (last_sc_fence_thread_before &&
1665 *act < *last_sc_fence_thread_before) {
1666 added = mo_graph->addEdge(act, rf) || added;
1672 * Include at most one act per-thread that "happens
1675 if (act->happens_before(curr)) {
1676 if (act->is_write()) {
1677 added = mo_graph->addEdge(act, rf) || added;
1679 const ModelAction *prevrf = act->get_reads_from();
1680 const Promise *prevrf_promise = act->get_reads_from_promise();
1682 if (!prevrf->equals(rf))
1683 added = mo_graph->addEdge(prevrf, rf) || added;
1684 } else if (!prevrf_promise->equals(rf)) {
1685 added = mo_graph->addEdge(prevrf_promise, rf) || added;
1694 * All compatible, thread-exclusive promises must be ordered after any
1695 * concrete loads from the same thread
1697 for (unsigned int i = 0; i < promises.size(); i++)
1698 if (promises[i]->is_compatible_exclusive(curr))
1699 added = mo_graph->addEdge(rf, promises[i]) || added;
1705 * Updates the mo_graph with the constraints imposed from the current write.
1707 * Basic idea is the following: Go through each other thread and find
1708 * the lastest action that happened before our write. Two cases:
1710 * (1) The action is a write => that write must occur before
1713 * (2) The action is a read => the write that that action read from
1714 * must occur before the current write.
1716 * This method also handles two other issues:
1718 * (I) Sequential Consistency: Making sure that if the current write is
1719 * seq_cst, that it occurs after the previous seq_cst write.
1721 * (II) Sending the write back to non-synchronizing reads.
1723 * @param curr The current action. Must be a write.
1724 * @param send_fv A vector for stashing reads to which we may pass our future
1725 * value. If NULL, then don't record any future values.
1726 * @return True if modification order edges were added; false otherwise
1728 bool ModelExecution::w_modification_order(ModelAction *curr, ModelVector<ModelAction *> *send_fv)
1730 SnapVector<action_list_t> *thrd_lists = obj_thrd_map.get(curr->get_location());
1733 ASSERT(curr->is_write());
1735 if (curr->is_seqcst()) {
1736 /* We have to at least see the last sequentially consistent write,
1737 so we are initialized. */
1738 ModelAction *last_seq_cst = get_last_seq_cst_write(curr);
1739 if (last_seq_cst != NULL) {
1740 added = mo_graph->addEdge(last_seq_cst, curr) || added;
1744 /* Last SC fence in the current thread */
1745 ModelAction *last_sc_fence_local = get_last_seq_cst_fence(curr->get_tid(), NULL);
1747 /* Iterate over all threads */
1748 for (i = 0; i < thrd_lists->size(); i++) {
1749 /* Last SC fence in thread i, before last SC fence in current thread */
1750 ModelAction *last_sc_fence_thread_before = NULL;
1751 if (last_sc_fence_local && int_to_id((int)i) != curr->get_tid())
1752 last_sc_fence_thread_before = get_last_seq_cst_fence(int_to_id(i), last_sc_fence_local);
1754 /* Iterate over actions in thread, starting from most recent */
1755 action_list_t *list = &(*thrd_lists)[i];
1756 action_list_t::reverse_iterator rit;
1757 for (rit = list->rbegin(); rit != list->rend(); rit++) {
1758 ModelAction *act = *rit;
1761 * 1) If RMW and it actually read from something, then we
1762 * already have all relevant edges, so just skip to next
1765 * 2) If RMW and it didn't read from anything, we should
1766 * whatever edge we can get to speed up convergence.
1768 * 3) If normal write, we need to look at earlier actions, so
1769 * continue processing list.
1771 if (curr->is_rmw()) {
1772 if (curr->get_reads_from() != NULL)
1780 /* C++, Section 29.3 statement 7 */
1781 if (last_sc_fence_thread_before && act->is_write() &&
1782 *act < *last_sc_fence_thread_before) {
1783 added = mo_graph->addEdge(act, curr) || added;
1788 * Include at most one act per-thread that "happens
1791 if (act->happens_before(curr)) {
1793 * Note: if act is RMW, just add edge:
1795 * The following edge should be handled elsewhere:
1796 * readfrom(act) --mo--> act
1798 if (act->is_write())
1799 added = mo_graph->addEdge(act, curr) || added;
1800 else if (act->is_read()) {
1801 //if previous read accessed a null, just keep going
1802 if (act->get_reads_from() == NULL) {
1803 added = mo_graph->addEdge(act->get_reads_from_promise(), curr) || added;
1805 added = mo_graph->addEdge(act->get_reads_from(), curr) || added;
1808 } else if (act->is_read() && !act->could_synchronize_with(curr) &&
1809 !act->same_thread(curr)) {
1810 /* We have an action that:
1811 (1) did not happen before us
1812 (2) is a read and we are a write
1813 (3) cannot synchronize with us
1814 (4) is in a different thread
1816 that read could potentially read from our write. Note that
1817 these checks are overly conservative at this point, we'll
1818 do more checks before actually removing the
1823 if (send_fv && thin_air_constraint_may_allow(curr, act) && check_coherence_promise(curr, act)) {
1824 if (!is_infeasible())
1825 send_fv->push_back(act);
1826 else if (curr->is_rmw() && act->is_rmw() && curr->get_reads_from() && curr->get_reads_from() == act->get_reads_from())
1827 add_future_value(curr, act);
1834 * All compatible, thread-exclusive promises must be ordered after any
1835 * concrete stores to the same thread, or else they can be merged with
1838 for (unsigned int i = 0; i < promises.size(); i++)
1839 if (promises[i]->is_compatible_exclusive(curr))
1840 added = mo_graph->addEdge(curr, promises[i]) || added;
1845 //This procedure uses cohere to prune future values that are
1846 //guaranteed to generate a coherence violation.
1848 //need to see if there is (1) a promise for thread write, (2)
1849 //the promise is sb before write, (3) the promise can only be
1850 //resolved by the thread read, and (4) the promise has same
1851 //location as read/write
1853 bool ModelExecution::check_coherence_promise(const ModelAction * write, const ModelAction *read) {
1854 thread_id_t write_tid=write->get_tid();
1855 for(unsigned int i = promises.size(); i>0; i--) {
1856 Promise *pr=promises[i-1];
1857 if (!pr->same_location(write))
1859 //the reading thread is the only thread that can resolve the promise
1860 if (pr->get_num_was_available_threads()==1 && pr->thread_was_available(read->get_tid())) {
1861 for(unsigned int j=0;j<pr->get_num_readers();j++) {
1862 ModelAction *prreader=pr->get_reader(j);
1863 //the writing thread reads from the promise before the write
1864 if (prreader->get_tid()==write_tid &&
1865 (*prreader)<(*write)) {
1866 if ((*read)>(*prreader)) {
1867 //check that we don't have a read between the read and promise
1868 //from the same thread as read
1870 for(const ModelAction *tmp=read;tmp!=prreader;) {
1871 tmp=tmp->get_node()->get_parent()->get_action();
1872 if (tmp->is_read() && tmp->same_thread(read)) {
1889 /** Arbitrary reads from the future are not allowed. Section 29.3
1890 * part 9 places some constraints. This method checks one result of constraint
1891 * constraint. Others require compiler support. */
1892 bool ModelExecution::thin_air_constraint_may_allow(const ModelAction *writer, const ModelAction *reader) const
1894 if (!writer->is_rmw())
1897 if (!reader->is_rmw())
1900 for (const ModelAction *search = writer->get_reads_from(); search != NULL; search = search->get_reads_from()) {
1901 if (search == reader)
1903 if (search->get_tid() == reader->get_tid() &&
1904 search->happens_before(reader))
1912 * Arbitrary reads from the future are not allowed. Section 29.3 part 9 places
1913 * some constraints. This method checks one the following constraint (others
1914 * require compiler support):
1916 * If X --hb-> Y --mo-> Z, then X should not read from Z.
1917 * If X --hb-> Y, A --rf-> Y, and A --mo-> Z, then X should not read from Z.
1919 bool ModelExecution::mo_may_allow(const ModelAction *writer, const ModelAction *reader)
1921 SnapVector<action_list_t> *thrd_lists = obj_thrd_map.get(reader->get_location());
1923 /* Iterate over all threads */
1924 for (i = 0; i < thrd_lists->size(); i++) {
1925 const ModelAction *write_after_read = NULL;
1927 /* Iterate over actions in thread, starting from most recent */
1928 action_list_t *list = &(*thrd_lists)[i];
1929 action_list_t::reverse_iterator rit;
1930 for (rit = list->rbegin(); rit != list->rend(); rit++) {
1931 ModelAction *act = *rit;
1933 /* Don't disallow due to act == reader */
1934 if (!reader->happens_before(act) || reader == act)
1936 else if (act->is_write())
1937 write_after_read = act;
1938 else if (act->is_read() && act->get_reads_from() != NULL)
1939 write_after_read = act->get_reads_from();
1942 if (write_after_read && write_after_read != writer && mo_graph->checkReachable(write_after_read, writer))
1949 * Finds the head(s) of the release sequence(s) containing a given ModelAction.
1950 * The ModelAction under consideration is expected to be taking part in
1951 * release/acquire synchronization as an object of the "reads from" relation.
1952 * Note that this can only provide release sequence support for RMW chains
1953 * which do not read from the future, as those actions cannot be traced until
1954 * their "promise" is fulfilled. Similarly, we may not even establish the
1955 * presence of a release sequence with certainty, as some modification order
1956 * constraints may be decided further in the future. Thus, this function
1957 * "returns" two pieces of data: a pass-by-reference vector of @a release_heads
1958 * and a boolean representing certainty.
1960 * @param rf The action that might be part of a release sequence. Must be a
1962 * @param release_heads A pass-by-reference style return parameter. After
1963 * execution of this function, release_heads will contain the heads of all the
1964 * relevant release sequences, if any exists with certainty
1965 * @param pending A pass-by-reference style return parameter which is only used
1966 * when returning false (i.e., uncertain). Returns most information regarding
1967 * an uncertain release sequence, including any write operations that might
1968 * break the sequence.
1969 * @return true, if the ModelExecution is certain that release_heads is complete;
1972 bool ModelExecution::release_seq_heads(const ModelAction *rf,
1973 rel_heads_list_t *release_heads,
1974 struct release_seq *pending) const
1976 /* Only check for release sequences if there are no cycles */
1977 if (mo_graph->checkForCycles())
1980 for ( ; rf != NULL; rf = rf->get_reads_from()) {
1981 ASSERT(rf->is_write());
1983 if (rf->is_release())
1984 release_heads->push_back(rf);
1985 else if (rf->get_last_fence_release())
1986 release_heads->push_back(rf->get_last_fence_release());
1988 break; /* End of RMW chain */
1990 /** @todo Need to be smarter here... In the linux lock
1991 * example, this will run to the beginning of the program for
1993 /** @todo The way to be smarter here is to keep going until 1
1994 * thread has a release preceded by an acquire and you've seen
1997 /* acq_rel RMW is a sufficient stopping condition */
1998 if (rf->is_acquire() && rf->is_release())
1999 return true; /* complete */
2002 /* read from future: need to settle this later */
2004 return false; /* incomplete */
2007 if (rf->is_release())
2008 return true; /* complete */
2010 /* else relaxed write
2011 * - check for fence-release in the same thread (29.8, stmt. 3)
2012 * - check modification order for contiguous subsequence
2013 * -> rf must be same thread as release */
2015 const ModelAction *fence_release = rf->get_last_fence_release();
2016 /* Synchronize with a fence-release unconditionally; we don't need to
2017 * find any more "contiguous subsequence..." for it */
2019 release_heads->push_back(fence_release);
2021 int tid = id_to_int(rf->get_tid());
2022 SnapVector<action_list_t> *thrd_lists = obj_thrd_map.get(rf->get_location());
2023 action_list_t *list = &(*thrd_lists)[tid];
2024 action_list_t::const_reverse_iterator rit;
2026 /* Find rf in the thread list */
2027 rit = std::find(list->rbegin(), list->rend(), rf);
2028 ASSERT(rit != list->rend());
2030 /* Find the last {write,fence}-release */
2031 for (; rit != list->rend(); rit++) {
2032 if (fence_release && *(*rit) < *fence_release)
2034 if ((*rit)->is_release())
2037 if (rit == list->rend()) {
2038 /* No write-release in this thread */
2039 return true; /* complete */
2040 } else if (fence_release && *(*rit) < *fence_release) {
2041 /* The fence-release is more recent (and so, "stronger") than
2042 * the most recent write-release */
2043 return true; /* complete */
2044 } /* else, need to establish contiguous release sequence */
2045 ModelAction *release = *rit;
2047 ASSERT(rf->same_thread(release));
2049 pending->writes.clear();
2051 bool certain = true;
2052 for (unsigned int i = 0; i < thrd_lists->size(); i++) {
2053 if (id_to_int(rf->get_tid()) == (int)i)
2055 list = &(*thrd_lists)[i];
2057 /* Can we ensure no future writes from this thread may break
2058 * the release seq? */
2059 bool future_ordered = false;
2061 ModelAction *last = get_last_action(int_to_id(i));
2062 Thread *th = get_thread(int_to_id(i));
2063 if ((last && rf->happens_before(last)) ||
2066 future_ordered = true;
2068 ASSERT(!th->is_model_thread() || future_ordered);
2070 for (rit = list->rbegin(); rit != list->rend(); rit++) {
2071 const ModelAction *act = *rit;
2072 /* Reach synchronization -> this thread is complete */
2073 if (act->happens_before(release))
2075 if (rf->happens_before(act)) {
2076 future_ordered = true;
2080 /* Only non-RMW writes can break release sequences */
2081 if (!act->is_write() || act->is_rmw())
2084 /* Check modification order */
2085 if (mo_graph->checkReachable(rf, act)) {
2086 /* rf --mo--> act */
2087 future_ordered = true;
2090 if (mo_graph->checkReachable(act, release))
2091 /* act --mo--> release */
2093 if (mo_graph->checkReachable(release, act) &&
2094 mo_graph->checkReachable(act, rf)) {
2095 /* release --mo-> act --mo--> rf */
2096 return true; /* complete */
2098 /* act may break release sequence */
2099 pending->writes.push_back(act);
2102 if (!future_ordered)
2103 certain = false; /* This thread is uncertain */
2107 release_heads->push_back(release);
2108 pending->writes.clear();
2110 pending->release = release;
2117 * An interface for getting the release sequence head(s) with which a
2118 * given ModelAction must synchronize. This function only returns a non-empty
2119 * result when it can locate a release sequence head with certainty. Otherwise,
2120 * it may mark the internal state of the ModelExecution so that it will handle
2121 * the release sequence at a later time, causing @a acquire to update its
2122 * synchronization at some later point in execution.
2124 * @param acquire The 'acquire' action that may synchronize with a release
2126 * @param read The read action that may read from a release sequence; this may
2127 * be the same as acquire, or else an earlier action in the same thread (i.e.,
2128 * when 'acquire' is a fence-acquire)
2129 * @param release_heads A pass-by-reference return parameter. Will be filled
2130 * with the head(s) of the release sequence(s), if they exists with certainty.
2131 * @see ModelExecution::release_seq_heads
2133 void ModelExecution::get_release_seq_heads(ModelAction *acquire,
2134 ModelAction *read, rel_heads_list_t *release_heads)
2136 const ModelAction *rf = read->get_reads_from();
2137 struct release_seq *sequence = (struct release_seq *)snapshot_calloc(1, sizeof(struct release_seq));
2138 sequence->acquire = acquire;
2139 sequence->read = read;
2141 if (!release_seq_heads(rf, release_heads, sequence)) {
2142 /* add act to 'lazy checking' list */
2143 pending_rel_seqs.push_back(sequence);
2145 snapshot_free(sequence);
2150 * @brief Propagate a modified clock vector to actions later in the execution
2153 * After an acquire operation lazily completes a release-sequence
2154 * synchronization, we must update all clock vectors for operations later than
2155 * the acquire in the execution order.
2157 * @param acquire The ModelAction whose clock vector must be propagated
2158 * @param work The work queue to which we can add work items, if this
2159 * propagation triggers more updates (e.g., to the modification order)
2161 void ModelExecution::propagate_clockvector(ModelAction *acquire, work_queue_t *work)
2163 /* Re-check all pending release sequences */
2164 work->push_back(CheckRelSeqWorkEntry(NULL));
2165 /* Re-check read-acquire for mo_graph edges */
2166 work->push_back(MOEdgeWorkEntry(acquire));
2168 /* propagate synchronization to later actions */
2169 action_list_t::reverse_iterator rit = action_trace.rbegin();
2170 for (; (*rit) != acquire; rit++) {
2171 ModelAction *propagate = *rit;
2172 if (acquire->happens_before(propagate)) {
2173 synchronize(acquire, propagate);
2174 /* Re-check 'propagate' for mo_graph edges */
2175 work->push_back(MOEdgeWorkEntry(propagate));
2181 * Attempt to resolve all stashed operations that might synchronize with a
2182 * release sequence for a given location. This implements the "lazy" portion of
2183 * determining whether or not a release sequence was contiguous, since not all
2184 * modification order information is present at the time an action occurs.
2186 * @param location The location/object that should be checked for release
2187 * sequence resolutions. A NULL value means to check all locations.
2188 * @param work_queue The work queue to which to add work items as they are
2190 * @return True if any updates occurred (new synchronization, new mo_graph
2193 bool ModelExecution::resolve_release_sequences(void *location, work_queue_t *work_queue)
2195 bool updated = false;
2196 SnapVector<struct release_seq *>::iterator it = pending_rel_seqs.begin();
2197 while (it != pending_rel_seqs.end()) {
2198 struct release_seq *pending = *it;
2199 ModelAction *acquire = pending->acquire;
2200 const ModelAction *read = pending->read;
2202 /* Only resolve sequences on the given location, if provided */
2203 if (location && read->get_location() != location) {
2208 const ModelAction *rf = read->get_reads_from();
2209 rel_heads_list_t release_heads;
2211 complete = release_seq_heads(rf, &release_heads, pending);
2212 for (unsigned int i = 0; i < release_heads.size(); i++)
2213 if (!acquire->has_synchronized_with(release_heads[i]))
2214 if (synchronize(release_heads[i], acquire))
2218 /* Propagate the changed clock vector */
2219 propagate_clockvector(acquire, work_queue);
2222 it = pending_rel_seqs.erase(it);
2223 snapshot_free(pending);
2229 // If we resolved promises or data races, see if we have realized a data race.
2236 * Performs various bookkeeping operations for the current ModelAction. For
2237 * instance, adds action to the per-object, per-thread action vector and to the
2238 * action trace list of all thread actions.
2240 * @param act is the ModelAction to add.
2242 void ModelExecution::add_action_to_lists(ModelAction *act)
2244 int tid = id_to_int(act->get_tid());
2245 ModelAction *uninit = NULL;
2247 action_list_t *list = get_safe_ptr_action(&obj_map, act->get_location());
2248 if (list->empty() && act->is_atomic_var()) {
2249 uninit = get_uninitialized_action(act);
2250 uninit_id = id_to_int(uninit->get_tid());
2251 list->push_front(uninit);
2253 list->push_back(act);
2255 action_trace.push_back(act);
2257 action_trace.push_front(uninit);
2259 SnapVector<action_list_t> *vec = get_safe_ptr_vect_action(&obj_thrd_map, act->get_location());
2260 if (tid >= (int)vec->size())
2261 vec->resize(priv->next_thread_id);
2262 (*vec)[tid].push_back(act);
2264 (*vec)[uninit_id].push_front(uninit);
2266 if ((int)thrd_last_action.size() <= tid)
2267 thrd_last_action.resize(get_num_threads());
2268 thrd_last_action[tid] = act;
2270 thrd_last_action[uninit_id] = uninit;
2272 if (act->is_fence() && act->is_release()) {
2273 if ((int)thrd_last_fence_release.size() <= tid)
2274 thrd_last_fence_release.resize(get_num_threads());
2275 thrd_last_fence_release[tid] = act;
2278 if (act->is_wait()) {
2279 void *mutex_loc = (void *) act->get_value();
2280 get_safe_ptr_action(&obj_map, mutex_loc)->push_back(act);
2282 SnapVector<action_list_t> *vec = get_safe_ptr_vect_action(&obj_thrd_map, mutex_loc);
2283 if (tid >= (int)vec->size())
2284 vec->resize(priv->next_thread_id);
2285 (*vec)[tid].push_back(act);
2290 * @brief Get the last action performed by a particular Thread
2291 * @param tid The thread ID of the Thread in question
2292 * @return The last action in the thread
2294 ModelAction * ModelExecution::get_last_action(thread_id_t tid) const
2296 int threadid = id_to_int(tid);
2297 if (threadid < (int)thrd_last_action.size())
2298 return thrd_last_action[id_to_int(tid)];
2304 * @brief Get the last fence release performed by a particular Thread
2305 * @param tid The thread ID of the Thread in question
2306 * @return The last fence release in the thread, if one exists; NULL otherwise
2308 ModelAction * ModelExecution::get_last_fence_release(thread_id_t tid) const
2310 int threadid = id_to_int(tid);
2311 if (threadid < (int)thrd_last_fence_release.size())
2312 return thrd_last_fence_release[id_to_int(tid)];
2318 * Gets the last memory_order_seq_cst write (in the total global sequence)
2319 * performed on a particular object (i.e., memory location), not including the
2321 * @param curr The current ModelAction; also denotes the object location to
2323 * @return The last seq_cst write
2325 ModelAction * ModelExecution::get_last_seq_cst_write(ModelAction *curr) const
2327 void *location = curr->get_location();
2328 action_list_t *list = obj_map.get(location);
2329 /* Find: max({i in dom(S) | seq_cst(t_i) && isWrite(t_i) && samevar(t_i, t)}) */
2330 action_list_t::reverse_iterator rit;
2331 for (rit = list->rbegin(); (*rit) != curr; rit++)
2333 rit++; /* Skip past curr */
2334 for ( ; rit != list->rend(); rit++)
2335 if ((*rit)->is_write() && (*rit)->is_seqcst())
2341 * Gets the last memory_order_seq_cst fence (in the total global sequence)
2342 * performed in a particular thread, prior to a particular fence.
2343 * @param tid The ID of the thread to check
2344 * @param before_fence The fence from which to begin the search; if NULL, then
2345 * search for the most recent fence in the thread.
2346 * @return The last prior seq_cst fence in the thread, if exists; otherwise, NULL
2348 ModelAction * ModelExecution::get_last_seq_cst_fence(thread_id_t tid, const ModelAction *before_fence) const
2350 /* All fences should have location FENCE_LOCATION */
2351 action_list_t *list = obj_map.get(FENCE_LOCATION);
2356 action_list_t::reverse_iterator rit = list->rbegin();
2359 for (; rit != list->rend(); rit++)
2360 if (*rit == before_fence)
2363 ASSERT(*rit == before_fence);
2367 for (; rit != list->rend(); rit++)
2368 if ((*rit)->is_fence() && (tid == (*rit)->get_tid()) && (*rit)->is_seqcst())
2374 * Gets the last unlock operation performed on a particular mutex (i.e., memory
2375 * location). This function identifies the mutex according to the current
2376 * action, which is presumed to perform on the same mutex.
2377 * @param curr The current ModelAction; also denotes the object location to
2379 * @return The last unlock operation
2381 ModelAction * ModelExecution::get_last_unlock(ModelAction *curr) const
2383 void *location = curr->get_location();
2385 action_list_t *list = obj_map.get(location);
2386 /* Find: max({i in dom(S) | isUnlock(t_i) && samevar(t_i, t)}) */
2387 action_list_t::reverse_iterator rit;
2388 for (rit = list->rbegin(); rit != list->rend(); rit++)
2389 if ((*rit)->is_unlock() || (*rit)->is_wait())
2394 ModelAction * ModelExecution::get_parent_action(thread_id_t tid) const
2396 ModelAction *parent = get_last_action(tid);
2398 parent = get_thread(tid)->get_creation();
2403 * Returns the clock vector for a given thread.
2404 * @param tid The thread whose clock vector we want
2405 * @return Desired clock vector
2407 ClockVector * ModelExecution::get_cv(thread_id_t tid) const
2409 return get_parent_action(tid)->get_cv();
2413 * @brief Find the promise (if any) to resolve for the current action and
2414 * remove it from the pending promise vector
2415 * @param curr The current ModelAction. Should be a write.
2416 * @return The Promise to resolve, if any; otherwise NULL
2418 Promise * ModelExecution::pop_promise_to_resolve(const ModelAction *curr)
2420 for (unsigned int i = 0; i < promises.size(); i++)
2421 if (curr->get_node()->get_promise(i)) {
2422 Promise *ret = promises[i];
2423 promises.erase(promises.begin() + i);
2430 * Resolve a Promise with a current write.
2431 * @param write The ModelAction that is fulfilling Promises
2432 * @param promise The Promise to resolve
2433 * @param work The work queue, for adding new fixup work
2434 * @return True if the Promise was successfully resolved; false otherwise
2436 bool ModelExecution::resolve_promise(ModelAction *write, Promise *promise,
2439 ModelVector<ModelAction *> actions_to_check;
2441 for (unsigned int i = 0; i < promise->get_num_readers(); i++) {
2442 ModelAction *read = promise->get_reader(i);
2443 if (read_from(read, write)) {
2444 /* Propagate the changed clock vector */
2445 propagate_clockvector(read, work);
2447 actions_to_check.push_back(read);
2449 /* Make sure the promise's value matches the write's value */
2450 ASSERT(promise->is_compatible(write) && promise->same_value(write));
2451 if (!mo_graph->resolvePromise(promise, write))
2452 priv->hard_failed_promise = true;
2455 * @todo It is possible to end up in an inconsistent state, where a
2456 * "resolved" promise may still be referenced if
2457 * CycleGraph::resolvePromise() failed, so don't delete 'promise'.
2459 * Note that the inconsistency only matters when dumping mo_graph to
2465 //Check whether reading these writes has made threads unable to
2467 for (unsigned int i = 0; i < actions_to_check.size(); i++) {
2468 ModelAction *read = actions_to_check[i];
2469 mo_check_promises(read, true);
2476 * Compute the set of promises that could potentially be satisfied by this
2477 * action. Note that the set computation actually appears in the Node, not in
2479 * @param curr The ModelAction that may satisfy promises
2481 void ModelExecution::compute_promises(ModelAction *curr)
2483 for (unsigned int i = 0; i < promises.size(); i++) {
2484 Promise *promise = promises[i];
2485 if (!promise->is_compatible(curr) || !promise->same_value(curr))
2488 bool satisfy = true;
2489 for (unsigned int j = 0; j < promise->get_num_readers(); j++) {
2490 const ModelAction *act = promise->get_reader(j);
2491 if (act->happens_before(curr) ||
2492 act->could_synchronize_with(curr)) {
2498 curr->get_node()->set_promise(i);
2502 /** Checks promises in response to change in ClockVector Threads. */
2503 void ModelExecution::check_promises(thread_id_t tid, ClockVector *old_cv, ClockVector *merge_cv)
2505 for (unsigned int i = 0; i < promises.size(); i++) {
2506 Promise *promise = promises[i];
2507 if (!promise->thread_is_available(tid))
2509 for (unsigned int j = 0; j < promise->get_num_readers(); j++) {
2510 const ModelAction *act = promise->get_reader(j);
2511 if ((!old_cv || !old_cv->synchronized_since(act)) &&
2512 merge_cv->synchronized_since(act)) {
2513 if (promise->eliminate_thread(tid)) {
2514 /* Promise has failed */
2515 priv->failed_promise = true;
2523 void ModelExecution::check_promises_thread_disabled()
2525 for (unsigned int i = 0; i < promises.size(); i++) {
2526 Promise *promise = promises[i];
2527 if (promise->has_failed()) {
2528 priv->failed_promise = true;
2535 * @brief Checks promises in response to addition to modification order for
2538 * We test whether threads are still available for satisfying promises after an
2539 * addition to our modification order constraints. Those that are unavailable
2540 * are "eliminated". Once all threads are eliminated from satisfying a promise,
2541 * that promise has failed.
2543 * @param act The ModelAction which updated the modification order
2544 * @param is_read_check Should be true if act is a read and we must check for
2545 * updates to the store from which it read (there is a distinction here for
2546 * RMW's, which are both a load and a store)
2548 void ModelExecution::mo_check_promises(const ModelAction *act, bool is_read_check)
2550 const ModelAction *write = is_read_check ? act->get_reads_from() : act;
2552 for (unsigned int i = 0; i < promises.size(); i++) {
2553 Promise *promise = promises[i];
2555 // Is this promise on the same location?
2556 if (!promise->same_location(write))
2559 for (unsigned int j = 0; j < promise->get_num_readers(); j++) {
2560 const ModelAction *pread = promise->get_reader(j);
2561 if (!pread->happens_before(act))
2563 if (mo_graph->checkPromise(write, promise)) {
2564 priv->hard_failed_promise = true;
2570 // Don't do any lookups twice for the same thread
2571 if (!promise->thread_is_available(act->get_tid()))
2574 if (mo_graph->checkReachable(promise, write)) {
2575 if (mo_graph->checkPromise(write, promise)) {
2576 priv->hard_failed_promise = true;
2584 * Compute the set of writes that may break the current pending release
2585 * sequence. This information is extracted from previou release sequence
2588 * @param curr The current ModelAction. Must be a release sequence fixup
2591 void ModelExecution::compute_relseq_breakwrites(ModelAction *curr)
2593 if (pending_rel_seqs.empty())
2596 struct release_seq *pending = pending_rel_seqs.back();
2597 for (unsigned int i = 0; i < pending->writes.size(); i++) {
2598 const ModelAction *write = pending->writes[i];
2599 curr->get_node()->add_relseq_break(write);
2602 /* NULL means don't break the sequence; just synchronize */
2603 curr->get_node()->add_relseq_break(NULL);
2607 * Build up an initial set of all past writes that this 'read' action may read
2608 * from, as well as any previously-observed future values that must still be valid.
2610 * @param curr is the current ModelAction that we are exploring; it must be a
2613 void ModelExecution::build_may_read_from(ModelAction *curr)
2615 SnapVector<action_list_t> *thrd_lists = obj_thrd_map.get(curr->get_location());
2617 ASSERT(curr->is_read());
2619 ModelAction *last_sc_write = NULL;
2621 if (curr->is_seqcst())
2622 last_sc_write = get_last_seq_cst_write(curr);
2624 /* Iterate over all threads */
2625 for (i = 0; i < thrd_lists->size(); i++) {
2626 /* Iterate over actions in thread, starting from most recent */
2627 action_list_t *list = &(*thrd_lists)[i];
2628 action_list_t::reverse_iterator rit;
2629 for (rit = list->rbegin(); rit != list->rend(); rit++) {
2630 ModelAction *act = *rit;
2632 /* Only consider 'write' actions */
2633 if (!act->is_write() || act == curr)
2636 /* Don't consider more than one seq_cst write if we are a seq_cst read. */
2637 bool allow_read = true;
2639 if (curr->is_seqcst() && (act->is_seqcst() || (last_sc_write != NULL && act->happens_before(last_sc_write))) && act != last_sc_write)
2641 else if (curr->get_sleep_flag() && !curr->is_seqcst() && !sleep_can_read_from(curr, act))
2645 /* Only add feasible reads */
2646 mo_graph->startChanges();
2647 r_modification_order(curr, act);
2648 if (!is_infeasible())
2649 curr->get_node()->add_read_from_past(act);
2650 mo_graph->rollbackChanges();
2653 /* Include at most one act per-thread that "happens before" curr */
2654 if (act->happens_before(curr))
2659 /* Inherit existing, promised future values */
2660 for (i = 0; i < promises.size(); i++) {
2661 const Promise *promise = promises[i];
2662 const ModelAction *promise_read = promise->get_reader(0);
2663 if (promise_read->same_var(curr)) {
2664 /* Only add feasible future-values */
2665 mo_graph->startChanges();
2666 r_modification_order(curr, promise);
2667 if (!is_infeasible())
2668 curr->get_node()->add_read_from_promise(promise_read);
2669 mo_graph->rollbackChanges();
2673 /* We may find no valid may-read-from only if the execution is doomed */
2674 if (!curr->get_node()->read_from_size()) {
2675 priv->no_valid_reads = true;
2679 if (DBG_ENABLED()) {
2680 model_print("Reached read action:\n");
2682 model_print("Printing read_from_past\n");
2683 curr->get_node()->print_read_from_past();
2684 model_print("End printing read_from_past\n");
2688 bool ModelExecution::sleep_can_read_from(ModelAction *curr, const ModelAction *write)
2690 for ( ; write != NULL; write = write->get_reads_from()) {
2691 /* UNINIT actions don't have a Node, and they never sleep */
2692 if (write->is_uninitialized())
2694 Node *prevnode = write->get_node()->get_parent();
2696 bool thread_sleep = prevnode->enabled_status(curr->get_tid()) == THREAD_SLEEP_SET;
2697 if (write->is_release() && thread_sleep)
2699 if (!write->is_rmw())
2706 * @brief Get an action representing an uninitialized atomic
2708 * This function may create a new one or try to retrieve one from the NodeStack
2710 * @param curr The current action, which prompts the creation of an UNINIT action
2711 * @return A pointer to the UNINIT ModelAction
2713 ModelAction * ModelExecution::get_uninitialized_action(const ModelAction *curr) const
2715 Node *node = curr->get_node();
2716 ModelAction *act = node->get_uninit_action();
2718 act = new ModelAction(ATOMIC_UNINIT, std::memory_order_relaxed, curr->get_location(), params->uninitvalue, model_thread);
2719 node->set_uninit_action(act);
2721 act->create_cv(NULL);
2725 static void print_list(const action_list_t *list)
2727 action_list_t::const_iterator it;
2729 model_print("------------------------------------------------------------------------------------\n");
2730 model_print("# t Action type MO Location Value Rf CV\n");
2731 model_print("------------------------------------------------------------------------------------\n");
2733 unsigned int hash = 0;
2735 for (it = list->begin(); it != list->end(); it++) {
2736 const ModelAction *act = *it;
2737 if (act->get_seq_number() > 0)
2739 hash = hash^(hash<<3)^((*it)->hash());
2741 model_print("HASH %u\n", hash);
2742 model_print("------------------------------------------------------------------------------------\n");
2745 #if SUPPORT_MOD_ORDER_DUMP
2746 void ModelExecution::dumpGraph(char *filename) const
2749 sprintf(buffer, "%s.dot", filename);
2750 FILE *file = fopen(buffer, "w");
2751 fprintf(file, "digraph %s {\n", filename);
2752 mo_graph->dumpNodes(file);
2753 ModelAction **thread_array = (ModelAction **)model_calloc(1, sizeof(ModelAction *) * get_num_threads());
2755 for (action_list_t::const_iterator it = action_trace.begin(); it != action_trace.end(); it++) {
2756 ModelAction *act = *it;
2757 if (act->is_read()) {
2758 mo_graph->dot_print_node(file, act);
2759 if (act->get_reads_from())
2760 mo_graph->dot_print_edge(file,
2761 act->get_reads_from(),
2763 "label=\"rf\", color=red, weight=2");
2765 mo_graph->dot_print_edge(file,
2766 act->get_reads_from_promise(),
2768 "label=\"rf\", color=red");
2770 if (thread_array[act->get_tid()]) {
2771 mo_graph->dot_print_edge(file,
2772 thread_array[id_to_int(act->get_tid())],
2774 "label=\"sb\", color=blue, weight=400");
2777 thread_array[act->get_tid()] = act;
2779 fprintf(file, "}\n");
2780 model_free(thread_array);
2785 /** @brief Prints an execution trace summary. */
2786 void ModelExecution::print_summary() const
2788 #if SUPPORT_MOD_ORDER_DUMP
2789 char buffername[100];
2790 sprintf(buffername, "exec%04u", get_execution_number());
2791 mo_graph->dumpGraphToFile(buffername);
2792 sprintf(buffername, "graph%04u", get_execution_number());
2793 dumpGraph(buffername);
2796 model_print("Execution trace %d:", get_execution_number());
2797 if (isfeasibleprefix()) {
2798 if (is_yieldblocked())
2799 model_print(" YIELD BLOCKED");
2800 if (scheduler->all_threads_sleeping())
2801 model_print(" SLEEP-SET REDUNDANT");
2802 if (have_bug_reports())
2803 model_print(" DETECTED BUG(S)");
2805 print_infeasibility(" INFEASIBLE");
2808 print_list(&action_trace);
2811 if (!promises.empty()) {
2812 model_print("Pending promises:\n");
2813 for (unsigned int i = 0; i < promises.size(); i++) {
2814 model_print(" [P%u] ", i);
2815 promises[i]->print();
2822 * Add a Thread to the system for the first time. Should only be called once
2824 * @param t The Thread to add
2826 void ModelExecution::add_thread(Thread *t)
2828 unsigned int i = id_to_int(t->get_id());
2829 if (i >= thread_map.size())
2830 thread_map.resize(i + 1);
2832 if (!t->is_model_thread())
2833 scheduler->add_thread(t);
2837 * @brief Get a Thread reference by its ID
2838 * @param tid The Thread's ID
2839 * @return A Thread reference
2841 Thread * ModelExecution::get_thread(thread_id_t tid) const
2843 unsigned int i = id_to_int(tid);
2844 if (i < thread_map.size())
2845 return thread_map[i];
2850 * @brief Get a reference to the Thread in which a ModelAction was executed
2851 * @param act The ModelAction
2852 * @return A Thread reference
2854 Thread * ModelExecution::get_thread(const ModelAction *act) const
2856 return get_thread(act->get_tid());
2860 * @brief Get a Thread reference by its pthread ID
2861 * @param index The pthread's ID
2862 * @return A Thread reference
2864 Thread * ModelExecution::get_pthread(pthread_t pid) {
2865 if (pid < pthread_counter + 1) return pthread_map[pid];
2870 * @brief Get a Promise's "promise number"
2872 * A "promise number" is an index number that is unique to a promise, valid
2873 * only for a specific snapshot of an execution trace. Promises may come and go
2874 * as they are generated an resolved, so an index only retains meaning for the
2877 * @param promise The Promise to check
2878 * @return The promise index, if the promise still is valid; otherwise -1
2880 int ModelExecution::get_promise_number(const Promise *promise) const
2882 for (unsigned int i = 0; i < promises.size(); i++)
2883 if (promises[i] == promise)
2890 * @brief Check if a Thread is currently enabled
2891 * @param t The Thread to check
2892 * @return True if the Thread is currently enabled
2894 bool ModelExecution::is_enabled(Thread *t) const
2896 return scheduler->is_enabled(t);
2900 * @brief Check if a Thread is currently enabled
2901 * @param tid The ID of the Thread to check
2902 * @return True if the Thread is currently enabled
2904 bool ModelExecution::is_enabled(thread_id_t tid) const
2906 return scheduler->is_enabled(tid);
2910 * @brief Select the next thread to execute based on the curren action
2912 * RMW actions occur in two parts, and we cannot split them. And THREAD_CREATE
2913 * actions should be followed by the execution of their child thread. In either
2914 * case, the current action should determine the next thread schedule.
2916 * @param curr The current action
2917 * @return The next thread to run, if the current action will determine this
2918 * selection; otherwise NULL
2920 Thread * ModelExecution::action_select_next_thread(const ModelAction *curr) const
2922 /* Do not split atomic RMW */
2923 if (curr->is_rmwr())
2924 return get_thread(curr);
2925 if (curr->is_write()) {
2926 // std::memory_order order = curr->get_mo();
2928 // case std::memory_order_relaxed:
2929 // return get_thread(curr);
2930 // case std::memory_order_release:
2931 // return get_thread(curr);
2938 /* Follow CREATE with the created thread */
2939 /* which is not needed, because model.cc takes care of this */
2940 if (curr->get_type() == THREAD_CREATE)
2941 return curr->get_thread_operand();
2942 if (curr->get_type() == PTHREAD_CREATE) {
2943 return curr->get_thread_operand();
2948 /** @return True if the execution has taken too many steps */
2949 bool ModelExecution::too_many_steps() const
2951 return params->bound != 0 && priv->used_sequence_numbers > params->bound;
2955 * Takes the next step in the execution, if possible.
2956 * @param curr The current step to take
2957 * @return Returns the next Thread to run, if any; NULL if this execution
2960 Thread * ModelExecution::take_step(ModelAction *curr)
2962 Thread *curr_thrd = get_thread(curr);
2963 ASSERT(curr_thrd->get_state() == THREAD_READY);
2965 ASSERT(check_action_enabled(curr)); /* May have side effects? */
2966 curr = check_current_action(curr);
2969 if (curr_thrd->is_blocked() || curr_thrd->is_complete())
2970 scheduler->remove_thread(curr_thrd);
2972 return action_select_next_thread(curr);
2976 * Launch end-of-execution release sequence fixups only when
2977 * the execution is otherwise feasible AND there are:
2979 * (1) pending release sequences
2980 * (2) pending assertions that could be invalidated by a change
2981 * in clock vectors (i.e., data races)
2982 * (3) no pending promises
2984 void ModelExecution::fixup_release_sequences()
2986 while (!pending_rel_seqs.empty() &&
2987 is_feasible_prefix_ignore_relseq() &&
2988 haveUnrealizedRaces()) {
2989 model_print("*** WARNING: release sequence fixup action "
2990 "(%zu pending release seuqence(s)) ***\n",
2991 pending_rel_seqs.size());
2992 ModelAction *fixup = new ModelAction(MODEL_FIXUP_RELSEQ,
2993 std::memory_order_seq_cst, NULL, VALUE_NONE,