10 #include "nodestack.h"
13 #include "clockvector.h"
14 #include "cyclegraph.h"
17 #include "threads-model.h"
18 #include "bugmessage.h"
20 #define INITIAL_THREAD_ID 0
23 * Structure for holding small ModelChecker members that should be snapshotted
25 struct model_snapshot_members {
26 model_snapshot_members() :
27 /* First thread created will have id INITIAL_THREAD_ID */
28 next_thread_id(INITIAL_THREAD_ID),
29 used_sequence_numbers(0),
32 failed_promise(false),
33 too_many_reads(false),
34 no_valid_reads(false),
35 bad_synchronization(false),
39 ~model_snapshot_members() {
40 for (unsigned int i = 0; i < bugs.size(); i++)
45 unsigned int next_thread_id;
46 modelclock_t used_sequence_numbers;
47 ModelAction *next_backtrack;
48 SnapVector<bug_message *> bugs;
52 /** @brief Incorrectly-ordered synchronization was made */
53 bool bad_synchronization;
59 /** @brief Constructor */
60 ModelExecution::ModelExecution(ModelChecker *m,
61 const struct model_params *params,
63 NodeStack *node_stack) :
68 thread_map(2), /* We'll always need at least 2 threads */
70 condvar_waiters_map(),
76 thrd_last_fence_release(),
77 node_stack(node_stack),
78 priv(new struct model_snapshot_members()),
79 mo_graph(new CycleGraph())
81 /* Initialize a model-checker thread, for special ModelActions */
82 model_thread = new Thread(get_next_id());
83 add_thread(model_thread);
84 scheduler->register_engine(this);
85 node_stack->register_engine(this);
88 /** @brief Destructor */
89 ModelExecution::~ModelExecution()
91 for (unsigned int i = 0; i < get_num_threads(); i++)
92 delete get_thread(int_to_id(i));
94 for (unsigned int i = 0; i < promises.size(); i++)
101 int ModelExecution::get_execution_number() const
103 return model->get_execution_number();
106 static action_list_t * get_safe_ptr_action(HashTable<const void *, action_list_t *, uintptr_t, 4> * hash, void * ptr)
108 action_list_t *tmp = hash->get(ptr);
110 tmp = new action_list_t();
116 static SnapVector<action_list_t> * get_safe_ptr_vect_action(HashTable<void *, SnapVector<action_list_t> *, uintptr_t, 4> * hash, void * ptr)
118 SnapVector<action_list_t> *tmp = hash->get(ptr);
120 tmp = new SnapVector<action_list_t>();
126 action_list_t * ModelExecution::get_actions_on_obj(void * obj, thread_id_t tid) const
128 SnapVector<action_list_t> *wrv = obj_thrd_map.get(obj);
131 unsigned int thread=id_to_int(tid);
132 if (thread < wrv->size())
133 return &(*wrv)[thread];
138 /** @return a thread ID for a new Thread */
139 thread_id_t ModelExecution::get_next_id()
141 return priv->next_thread_id++;
144 /** @return the number of user threads created during this execution */
145 unsigned int ModelExecution::get_num_threads() const
147 return priv->next_thread_id;
150 /** @return a sequence number for a new ModelAction */
151 modelclock_t ModelExecution::get_next_seq_num()
153 return ++priv->used_sequence_numbers;
157 * @brief Should the current action wake up a given thread?
159 * @param curr The current action
160 * @param thread The thread that we might wake up
161 * @return True, if we should wake up the sleeping thread; false otherwise
163 bool ModelExecution::should_wake_up(const ModelAction *curr, const Thread *thread) const
165 const ModelAction *asleep = thread->get_pending();
166 /* Don't allow partial RMW to wake anyone up */
169 /* Synchronizing actions may have been backtracked */
170 if (asleep->could_synchronize_with(curr))
172 /* All acquire/release fences and fence-acquire/store-release */
173 if (asleep->is_fence() && asleep->is_acquire() && curr->is_release())
175 /* Fence-release + store can awake load-acquire on the same location */
176 if (asleep->is_read() && asleep->is_acquire() && curr->same_var(asleep) && curr->is_write()) {
177 ModelAction *fence_release = get_last_fence_release(curr->get_tid());
178 if (fence_release && *(get_last_action(thread->get_id())) < *fence_release)
184 void ModelExecution::wake_up_sleeping_actions(ModelAction *curr)
186 for (unsigned int i = 0; i < get_num_threads(); i++) {
187 Thread *thr = get_thread(int_to_id(i));
188 if (scheduler->is_sleep_set(thr)) {
189 if (should_wake_up(curr, thr))
190 /* Remove this thread from sleep set */
191 scheduler->remove_sleep(thr);
196 /** @brief Alert the model-checker that an incorrectly-ordered
197 * synchronization was made */
198 void ModelExecution::set_bad_synchronization()
200 priv->bad_synchronization = true;
203 bool ModelExecution::assert_bug(const char *msg)
205 priv->bugs.push_back(new bug_message(msg));
207 if (isfeasibleprefix()) {
214 /** @return True, if any bugs have been reported for this execution */
215 bool ModelExecution::have_bug_reports() const
217 return priv->bugs.size() != 0;
220 SnapVector<bug_message *> * ModelExecution::get_bugs() const
226 * Check whether the current trace has triggered an assertion which should halt
229 * @return True, if the execution should be aborted; false otherwise
231 bool ModelExecution::has_asserted() const
233 return priv->asserted;
237 * Trigger a trace assertion which should cause this execution to be halted.
238 * This can be due to a detected bug or due to an infeasibility that should
241 void ModelExecution::set_assert()
243 priv->asserted = true;
247 * Check if we are in a deadlock. Should only be called at the end of an
248 * execution, although it should not give false positives in the middle of an
249 * execution (there should be some ENABLED thread).
251 * @return True if program is in a deadlock; false otherwise
253 bool ModelExecution::is_deadlocked() const
255 bool blocking_threads = false;
256 for (unsigned int i = 0; i < get_num_threads(); i++) {
257 thread_id_t tid = int_to_id(i);
260 Thread *t = get_thread(tid);
261 if (!t->is_model_thread() && t->get_pending())
262 blocking_threads = true;
264 return blocking_threads;
267 bool ModelExecution::is_yieldblocked() const
269 if (!params->yieldblock)
272 for (unsigned int i = 0; i < get_num_threads(); i++) {
273 thread_id_t tid = int_to_id(i);
274 Thread *t = get_thread(tid);
275 if (t->get_pending() && t->get_pending()->is_yield())
282 * Check if this is a complete execution. That is, have all thread completed
283 * execution (rather than exiting because sleep sets have forced a redundant
286 * @return True if the execution is complete.
288 bool ModelExecution::is_complete_execution() const
290 if (is_yieldblocked())
292 for (unsigned int i = 0; i < get_num_threads(); i++)
293 if (is_enabled(int_to_id(i)))
299 * @brief Find the last fence-related backtracking conflict for a ModelAction
301 * This function performs the search for the most recent conflicting action
302 * against which we should perform backtracking, as affected by fence
303 * operations. This includes pairs of potentially-synchronizing actions which
304 * occur due to fence-acquire or fence-release, and hence should be explored in
305 * the opposite execution order.
307 * @param act The current action
308 * @return The most recent action which conflicts with act due to fences
310 ModelAction * ModelExecution::get_last_fence_conflict(ModelAction *act) const
312 /* Only perform release/acquire fence backtracking for stores */
313 if (!act->is_write())
316 /* Find a fence-release (or, act is a release) */
317 ModelAction *last_release;
318 if (act->is_release())
321 last_release = get_last_fence_release(act->get_tid());
325 /* Skip past the release */
326 const action_list_t *list = &action_trace;
327 action_list_t::const_reverse_iterator rit;
328 for (rit = list->rbegin(); rit != list->rend(); rit++)
329 if (*rit == last_release)
331 ASSERT(rit != list->rend());
336 * load --sb-> fence-acquire */
337 ModelVector<ModelAction *> acquire_fences(get_num_threads(), NULL);
338 ModelVector<ModelAction *> prior_loads(get_num_threads(), NULL);
339 bool found_acquire_fences = false;
340 for ( ; rit != list->rend(); rit++) {
341 ModelAction *prev = *rit;
342 if (act->same_thread(prev))
345 int tid = id_to_int(prev->get_tid());
347 if (prev->is_read() && act->same_var(prev)) {
348 if (prev->is_acquire()) {
349 /* Found most recent load-acquire, don't need
350 * to search for more fences */
351 if (!found_acquire_fences)
354 prior_loads[tid] = prev;
357 if (prev->is_acquire() && prev->is_fence() && !acquire_fences[tid]) {
358 found_acquire_fences = true;
359 acquire_fences[tid] = prev;
363 ModelAction *latest_backtrack = NULL;
364 for (unsigned int i = 0; i < acquire_fences.size(); i++)
365 if (acquire_fences[i] && prior_loads[i])
366 if (!latest_backtrack || *latest_backtrack < *acquire_fences[i])
367 latest_backtrack = acquire_fences[i];
368 return latest_backtrack;
372 * @brief Find the last backtracking conflict for a ModelAction
374 * This function performs the search for the most recent conflicting action
375 * against which we should perform backtracking. This primary includes pairs of
376 * synchronizing actions which should be explored in the opposite execution
379 * @param act The current action
380 * @return The most recent action which conflicts with act
382 ModelAction * ModelExecution::get_last_conflict(ModelAction *act) const
384 switch (act->get_type()) {
386 /* Only seq-cst fences can (directly) cause backtracking */
387 if (!act->is_seqcst())
392 ModelAction *ret = NULL;
394 /* linear search: from most recent to oldest */
395 action_list_t *list = obj_map.get(act->get_location());
396 action_list_t::reverse_iterator rit;
397 for (rit = list->rbegin(); rit != list->rend(); rit++) {
398 ModelAction *prev = *rit;
401 if (prev->could_synchronize_with(act)) {
407 ModelAction *ret2 = get_last_fence_conflict(act);
417 case ATOMIC_TRYLOCK: {
418 /* linear search: from most recent to oldest */
419 action_list_t *list = obj_map.get(act->get_location());
420 action_list_t::reverse_iterator rit;
421 for (rit = list->rbegin(); rit != list->rend(); rit++) {
422 ModelAction *prev = *rit;
423 if (act->is_conflicting_lock(prev))
428 case ATOMIC_UNLOCK: {
429 /* linear search: from most recent to oldest */
430 action_list_t *list = obj_map.get(act->get_location());
431 action_list_t::reverse_iterator rit;
432 for (rit = list->rbegin(); rit != list->rend(); rit++) {
433 ModelAction *prev = *rit;
434 if (!act->same_thread(prev) && prev->is_failed_trylock())
440 /* linear search: from most recent to oldest */
441 action_list_t *list = obj_map.get(act->get_location());
442 action_list_t::reverse_iterator rit;
443 for (rit = list->rbegin(); rit != list->rend(); rit++) {
444 ModelAction *prev = *rit;
445 if (!act->same_thread(prev) && prev->is_failed_trylock())
447 if (!act->same_thread(prev) && prev->is_notify())
453 case ATOMIC_NOTIFY_ALL:
454 case ATOMIC_NOTIFY_ONE: {
455 /* linear search: from most recent to oldest */
456 action_list_t *list = obj_map.get(act->get_location());
457 action_list_t::reverse_iterator rit;
458 for (rit = list->rbegin(); rit != list->rend(); rit++) {
459 ModelAction *prev = *rit;
460 if (!act->same_thread(prev) && prev->is_wait())
471 /** This method finds backtracking points where we should try to
472 * reorder the parameter ModelAction against.
474 * @param the ModelAction to find backtracking points for.
476 void ModelExecution::set_backtracking(ModelAction *act)
478 Thread *t = get_thread(act);
479 ModelAction *prev = get_last_conflict(act);
483 Node *node = prev->get_node()->get_parent();
485 /* See Dynamic Partial Order Reduction (addendum), POPL '05 */
486 int low_tid, high_tid;
487 if (node->enabled_status(t->get_id()) == THREAD_ENABLED) {
488 low_tid = id_to_int(act->get_tid());
489 high_tid = low_tid + 1;
492 high_tid = get_num_threads();
495 for (int i = low_tid; i < high_tid; i++) {
496 thread_id_t tid = int_to_id(i);
498 /* Make sure this thread can be enabled here. */
499 if (i >= node->get_num_threads())
502 /* See Dynamic Partial Order Reduction (addendum), POPL '05 */
503 /* Don't backtrack into a point where the thread is disabled or sleeping. */
504 if (node->enabled_status(tid) != THREAD_ENABLED)
507 /* Check if this has been explored already */
508 if (node->has_been_explored(tid))
511 /* See if fairness allows */
512 if (params->fairwindow != 0 && !node->has_priority(tid)) {
514 for (int t = 0; t < node->get_num_threads(); t++) {
515 thread_id_t tother = int_to_id(t);
516 if (node->is_enabled(tother) && node->has_priority(tother)) {
525 /* See if CHESS-like yield fairness allows */
526 if (params->yieldon) {
528 for (int t = 0; t < node->get_num_threads(); t++) {
529 thread_id_t tother = int_to_id(t);
530 if (node->is_enabled(tother) && node->has_priority_over(tid, tother)) {
539 /* Cache the latest backtracking point */
540 set_latest_backtrack(prev);
542 /* If this is a new backtracking point, mark the tree */
543 if (!node->set_backtrack(tid))
545 DEBUG("Setting backtrack: conflict = %d, instead tid = %d\n",
546 id_to_int(prev->get_tid()),
547 id_to_int(t->get_id()));
556 * @brief Cache the a backtracking point as the "most recent", if eligible
558 * Note that this does not prepare the NodeStack for this backtracking
559 * operation, it only caches the action on a per-execution basis
561 * @param act The operation at which we should explore a different next action
562 * (i.e., backtracking point)
563 * @return True, if this action is now the most recent backtracking point;
566 bool ModelExecution::set_latest_backtrack(ModelAction *act)
568 if (!priv->next_backtrack || *act > *priv->next_backtrack) {
569 priv->next_backtrack = act;
576 * Returns last backtracking point. The model checker will explore a different
577 * path for this point in the next execution.
578 * @return The ModelAction at which the next execution should diverge.
580 ModelAction * ModelExecution::get_next_backtrack()
582 ModelAction *next = priv->next_backtrack;
583 priv->next_backtrack = NULL;
588 * Processes a read model action.
589 * @param curr is the read model action to process.
590 * @return True if processing this read updates the mo_graph.
592 bool ModelExecution::process_read(ModelAction *curr)
594 Node *node = curr->get_node();
596 bool updated = false;
597 switch (node->get_read_from_status()) {
598 case READ_FROM_PAST: {
599 const ModelAction *rf = node->get_read_from_past();
602 mo_graph->startChanges();
604 ASSERT(!is_infeasible());
605 if (!check_recency(curr, rf)) {
606 if (node->increment_read_from()) {
607 mo_graph->rollbackChanges();
610 priv->too_many_reads = true;
614 updated = r_modification_order(curr, rf);
616 mo_graph->commitChanges();
617 mo_check_promises(curr, true);
620 case READ_FROM_PROMISE: {
621 Promise *promise = curr->get_node()->get_read_from_promise();
622 if (promise->add_reader(curr))
623 priv->failed_promise = true;
624 curr->set_read_from_promise(promise);
625 mo_graph->startChanges();
626 if (!check_recency(curr, promise))
627 priv->too_many_reads = true;
628 updated = r_modification_order(curr, promise);
629 mo_graph->commitChanges();
632 case READ_FROM_FUTURE: {
633 /* Read from future value */
634 struct future_value fv = node->get_future_value();
635 Promise *promise = new Promise(this, curr, fv);
636 curr->set_read_from_promise(promise);
637 promises.push_back(promise);
638 mo_graph->startChanges();
639 updated = r_modification_order(curr, promise);
640 mo_graph->commitChanges();
646 get_thread(curr)->set_return_value(curr->get_return_value());
652 * Processes a lock, trylock, or unlock model action. @param curr is
653 * the read model action to process.
655 * The try lock operation checks whether the lock is taken. If not,
656 * it falls to the normal lock operation case. If so, it returns
659 * The lock operation has already been checked that it is enabled, so
660 * it just grabs the lock and synchronizes with the previous unlock.
662 * The unlock operation has to re-enable all of the threads that are
663 * waiting on the lock.
665 * @return True if synchronization was updated; false otherwise
667 bool ModelExecution::process_mutex(ModelAction *curr)
669 std::mutex *mutex = curr->get_mutex();
670 struct std::mutex_state *state = NULL;
673 state = mutex->get_state();
675 switch (curr->get_type()) {
676 case ATOMIC_TRYLOCK: {
677 bool success = !state->locked;
678 curr->set_try_lock(success);
680 get_thread(curr)->set_return_value(0);
683 get_thread(curr)->set_return_value(1);
685 //otherwise fall into the lock case
687 if (curr->get_cv()->getClock(state->alloc_tid) <= state->alloc_clock)
688 assert_bug("Lock access before initialization");
689 state->locked = get_thread(curr);
690 ModelAction *unlock = get_last_unlock(curr);
691 //synchronize with the previous unlock statement
692 if (unlock != NULL) {
693 synchronize(unlock, curr);
699 case ATOMIC_UNLOCK: {
700 /* wake up the other threads */
701 for (unsigned int i = 0; i < get_num_threads(); i++) {
702 Thread *t = get_thread(int_to_id(i));
703 Thread *curr_thrd = get_thread(curr);
704 if (t->waiting_on() == curr_thrd && t->get_pending()->is_lock())
708 /* unlock the lock - after checking who was waiting on it */
709 state->locked = NULL;
711 if (!curr->is_wait())
712 break; /* The rest is only for ATOMIC_WAIT */
714 /* Should we go to sleep? (simulate spurious failures) */
715 if (curr->get_node()->get_misc() == 0) {
716 get_safe_ptr_action(&condvar_waiters_map, curr->get_location())->push_back(curr);
718 scheduler->sleep(get_thread(curr));
722 case ATOMIC_NOTIFY_ALL: {
723 action_list_t *waiters = get_safe_ptr_action(&condvar_waiters_map, curr->get_location());
724 //activate all the waiting threads
725 for (action_list_t::iterator rit = waiters->begin(); rit != waiters->end(); rit++) {
726 scheduler->wake(get_thread(*rit));
731 case ATOMIC_NOTIFY_ONE: {
732 action_list_t *waiters = get_safe_ptr_action(&condvar_waiters_map, curr->get_location());
733 int wakeupthread = curr->get_node()->get_misc();
734 action_list_t::iterator it = waiters->begin();
735 advance(it, wakeupthread);
736 scheduler->wake(get_thread(*it));
748 * @brief Check if the current pending promises allow a future value to be sent
750 * If one of the following is true:
751 * (a) there are no pending promises
752 * (b) the reader and writer do not cross any promises
753 * Then, it is safe to pass a future value back now.
755 * Otherwise, we must save the pending future value until (a) or (b) is true
757 * @param writer The operation which sends the future value. Must be a write.
758 * @param reader The operation which will observe the value. Must be a read.
759 * @return True if the future value can be sent now; false if it must wait.
761 bool ModelExecution::promises_may_allow(const ModelAction *writer,
762 const ModelAction *reader) const
764 if (promises.empty())
766 for (int i = promises.size() - 1; i >= 0; i--) {
767 ModelAction *pr = promises[i]->get_reader(0);
768 //reader is after promise...doesn't cross any promise
771 //writer is after promise, reader before...bad...
779 * @brief Add a future value to a reader
781 * This function performs a few additional checks to ensure that the future
782 * value can be feasibly observed by the reader
784 * @param writer The operation whose value is sent. Must be a write.
785 * @param reader The read operation which may read the future value. Must be a read.
787 void ModelExecution::add_future_value(const ModelAction *writer, ModelAction *reader)
789 /* Do more ambitious checks now that mo is more complete */
790 if (!mo_may_allow(writer, reader))
793 Node *node = reader->get_node();
795 /* Find an ancestor thread which exists at the time of the reader */
796 Thread *write_thread = get_thread(writer);
797 while (id_to_int(write_thread->get_id()) >= node->get_num_threads())
798 write_thread = write_thread->get_parent();
800 struct future_value fv = {
801 writer->get_write_value(),
802 writer->get_seq_number() + params->maxfuturedelay,
803 write_thread->get_id(),
805 if (node->add_future_value(fv))
806 set_latest_backtrack(reader);
810 * Process a write ModelAction
811 * @param curr The ModelAction to process
812 * @return True if the mo_graph was updated or promises were resolved
814 bool ModelExecution::process_write(ModelAction *curr)
816 /* Readers to which we may send our future value */
817 ModelVector<ModelAction *> send_fv;
819 const ModelAction *earliest_promise_reader;
820 bool updated_promises = false;
822 bool updated_mod_order = w_modification_order(curr, &send_fv);
823 Promise *promise = pop_promise_to_resolve(curr);
826 earliest_promise_reader = promise->get_reader(0);
827 updated_promises = resolve_promise(curr, promise);
829 earliest_promise_reader = NULL;
831 for (unsigned int i = 0; i < send_fv.size(); i++) {
832 ModelAction *read = send_fv[i];
834 /* Don't send future values to reads after the Promise we resolve */
835 if (!earliest_promise_reader || *read < *earliest_promise_reader) {
836 /* Check if future value can be sent immediately */
837 if (promises_may_allow(curr, read)) {
838 add_future_value(curr, read);
840 futurevalues.push_back(PendingFutureValue(curr, read));
845 /* Check the pending future values */
846 for (int i = (int)futurevalues.size() - 1; i >= 0; i--) {
847 struct PendingFutureValue pfv = futurevalues[i];
848 if (promises_may_allow(pfv.writer, pfv.reader)) {
849 add_future_value(pfv.writer, pfv.reader);
850 futurevalues.erase(futurevalues.begin() + i);
854 mo_graph->commitChanges();
855 mo_check_promises(curr, false);
857 get_thread(curr)->set_return_value(VALUE_NONE);
858 return updated_mod_order || updated_promises;
862 * Process a fence ModelAction
863 * @param curr The ModelAction to process
864 * @return True if synchronization was updated
866 bool ModelExecution::process_fence(ModelAction *curr)
869 * fence-relaxed: no-op
870 * fence-release: only log the occurence (not in this function), for
871 * use in later synchronization
872 * fence-acquire (this function): search for hypothetical release
874 * fence-seq-cst: MO constraints formed in {r,w}_modification_order
876 bool updated = false;
877 if (curr->is_acquire()) {
878 action_list_t *list = &action_trace;
879 action_list_t::reverse_iterator rit;
880 /* Find X : is_read(X) && X --sb-> curr */
881 for (rit = list->rbegin(); rit != list->rend(); rit++) {
882 ModelAction *act = *rit;
885 if (act->get_tid() != curr->get_tid())
887 /* Stop at the beginning of the thread */
888 if (act->is_thread_start())
890 /* Stop once we reach a prior fence-acquire */
891 if (act->is_fence() && act->is_acquire())
895 /* read-acquire will find its own release sequences */
896 if (act->is_acquire())
899 /* Establish hypothetical release sequences */
900 rel_heads_list_t release_heads;
901 get_release_seq_heads(curr, act, &release_heads);
902 for (unsigned int i = 0; i < release_heads.size(); i++)
903 synchronize(release_heads[i], curr);
904 if (release_heads.size() != 0)
912 * @brief Process the current action for thread-related activity
914 * Performs current-action processing for a THREAD_* ModelAction. Proccesses
915 * may include setting Thread status, completing THREAD_FINISH/THREAD_JOIN
916 * synchronization, etc. This function is a no-op for non-THREAD actions
917 * (e.g., ATOMIC_{READ,WRITE,RMW,LOCK}, etc.)
919 * @param curr The current action
920 * @return True if synchronization was updated or a thread completed
922 bool ModelExecution::process_thread_action(ModelAction *curr)
924 bool updated = false;
926 switch (curr->get_type()) {
927 case THREAD_CREATE: {
928 thrd_t *thrd = (thrd_t *)curr->get_location();
929 struct thread_params *params = (struct thread_params *)curr->get_value();
930 Thread *th = new Thread(get_next_id(), thrd, params->func, params->arg, get_thread(curr));
932 th->set_creation(curr);
933 /* Promises can be satisfied by children */
934 for (unsigned int i = 0; i < promises.size(); i++) {
935 Promise *promise = promises[i];
936 if (promise->thread_is_available(curr->get_tid()))
937 promise->add_thread(th->get_id());
942 Thread *blocking = curr->get_thread_operand();
943 ModelAction *act = get_last_action(blocking->get_id());
944 synchronize(act, curr);
945 updated = true; /* trigger rel-seq checks */
948 case THREAD_FINISH: {
949 Thread *th = get_thread(curr);
950 /* Wake up any joining threads */
951 for (unsigned int i = 0; i < get_num_threads(); i++) {
952 Thread *waiting = get_thread(int_to_id(i));
953 if (waiting->waiting_on() == th &&
954 waiting->get_pending()->is_thread_join())
955 scheduler->wake(waiting);
958 /* Completed thread can't satisfy promises */
959 for (unsigned int i = 0; i < promises.size(); i++) {
960 Promise *promise = promises[i];
961 if (promise->thread_is_available(th->get_id()))
962 if (promise->eliminate_thread(th->get_id()))
963 priv->failed_promise = true;
965 updated = true; /* trigger rel-seq checks */
969 check_promises(curr->get_tid(), NULL, curr->get_cv());
980 * @brief Process the current action for release sequence fixup activity
982 * Performs model-checker release sequence fixups for the current action,
983 * forcing a single pending release sequence to break (with a given, potential
984 * "loose" write) or to complete (i.e., synchronize). If a pending release
985 * sequence forms a complete release sequence, then we must perform the fixup
986 * synchronization, mo_graph additions, etc.
988 * @param curr The current action; must be a release sequence fixup action
989 * @param work_queue The work queue to which to add work items as they are
992 void ModelExecution::process_relseq_fixup(ModelAction *curr, work_queue_t *work_queue)
994 const ModelAction *write = curr->get_node()->get_relseq_break();
995 struct release_seq *sequence = pending_rel_seqs.back();
996 pending_rel_seqs.pop_back();
998 ModelAction *acquire = sequence->acquire;
999 const ModelAction *rf = sequence->rf;
1000 const ModelAction *release = sequence->release;
1004 ASSERT(release->same_thread(rf));
1006 if (write == NULL) {
1008 * @todo Forcing a synchronization requires that we set
1009 * modification order constraints. For instance, we can't allow
1010 * a fixup sequence in which two separate read-acquire
1011 * operations read from the same sequence, where the first one
1012 * synchronizes and the other doesn't. Essentially, we can't
1013 * allow any writes to insert themselves between 'release' and
1017 /* Must synchronize */
1018 if (!synchronize(release, acquire))
1020 /* Re-check all pending release sequences */
1021 work_queue->push_back(CheckRelSeqWorkEntry(NULL));
1022 /* Re-check act for mo_graph edges */
1023 work_queue->push_back(MOEdgeWorkEntry(acquire));
1025 /* propagate synchronization to later actions */
1026 action_list_t::reverse_iterator rit = action_trace.rbegin();
1027 for (; (*rit) != acquire; rit++) {
1028 ModelAction *propagate = *rit;
1029 if (acquire->happens_before(propagate)) {
1030 synchronize(acquire, propagate);
1031 /* Re-check 'propagate' for mo_graph edges */
1032 work_queue->push_back(MOEdgeWorkEntry(propagate));
1036 /* Break release sequence with new edges:
1037 * release --mo--> write --mo--> rf */
1038 mo_graph->addEdge(release, write);
1039 mo_graph->addEdge(write, rf);
1042 /* See if we have realized a data race */
1047 * Initialize the current action by performing one or more of the following
1048 * actions, as appropriate: merging RMWR and RMWC/RMW actions, stepping forward
1049 * in the NodeStack, manipulating backtracking sets, allocating and
1050 * initializing clock vectors, and computing the promises to fulfill.
1052 * @param curr The current action, as passed from the user context; may be
1053 * freed/invalidated after the execution of this function, with a different
1054 * action "returned" its place (pass-by-reference)
1055 * @return True if curr is a newly-explored action; false otherwise
1057 bool ModelExecution::initialize_curr_action(ModelAction **curr)
1059 ModelAction *newcurr;
1061 if ((*curr)->is_rmwc() || (*curr)->is_rmw()) {
1062 newcurr = process_rmw(*curr);
1065 if (newcurr->is_rmw())
1066 compute_promises(newcurr);
1072 (*curr)->set_seq_number(get_next_seq_num());
1074 newcurr = node_stack->explore_action(*curr, scheduler->get_enabled_array());
1076 /* First restore type and order in case of RMW operation */
1077 if ((*curr)->is_rmwr())
1078 newcurr->copy_typeandorder(*curr);
1080 ASSERT((*curr)->get_location() == newcurr->get_location());
1081 newcurr->copy_from_new(*curr);
1083 /* Discard duplicate ModelAction; use action from NodeStack */
1086 /* Always compute new clock vector */
1087 newcurr->create_cv(get_parent_action(newcurr->get_tid()));
1090 return false; /* Action was explored previously */
1094 /* Always compute new clock vector */
1095 newcurr->create_cv(get_parent_action(newcurr->get_tid()));
1097 /* Assign most recent release fence */
1098 newcurr->set_last_fence_release(get_last_fence_release(newcurr->get_tid()));
1101 * Perform one-time actions when pushing new ModelAction onto
1104 if (newcurr->is_write())
1105 compute_promises(newcurr);
1106 else if (newcurr->is_relseq_fixup())
1107 compute_relseq_breakwrites(newcurr);
1108 else if (newcurr->is_wait())
1109 newcurr->get_node()->set_misc_max(2);
1110 else if (newcurr->is_notify_one()) {
1111 newcurr->get_node()->set_misc_max(get_safe_ptr_action(&condvar_waiters_map, newcurr->get_location())->size());
1113 return true; /* This was a new ModelAction */
1118 * @brief Establish reads-from relation between two actions
1120 * Perform basic operations involved with establishing a concrete rf relation,
1121 * including setting the ModelAction data and checking for release sequences.
1123 * @param act The action that is reading (must be a read)
1124 * @param rf The action from which we are reading (must be a write)
1126 * @return True if this read established synchronization
1128 bool ModelExecution::read_from(ModelAction *act, const ModelAction *rf)
1131 ASSERT(rf->is_write());
1133 act->set_read_from(rf);
1134 if (act->is_acquire()) {
1135 rel_heads_list_t release_heads;
1136 get_release_seq_heads(act, act, &release_heads);
1137 int num_heads = release_heads.size();
1138 for (unsigned int i = 0; i < release_heads.size(); i++)
1139 if (!synchronize(release_heads[i], act))
1141 return num_heads > 0;
1147 * @brief Synchronizes two actions
1149 * When A synchronizes with B (or A --sw-> B), B inherits A's clock vector.
1150 * This function performs the synchronization as well as providing other hooks
1151 * for other checks along with synchronization.
1153 * @param first The left-hand side of the synchronizes-with relation
1154 * @param second The right-hand side of the synchronizes-with relation
1155 * @return True if the synchronization was successful (i.e., was consistent
1156 * with the execution order); false otherwise
1158 bool ModelExecution::synchronize(const ModelAction *first, ModelAction *second)
1160 if (*second < *first) {
1161 set_bad_synchronization();
1164 check_promises(first->get_tid(), second->get_cv(), first->get_cv());
1165 return second->synchronize_with(first);
1169 * Check promises and eliminate potentially-satisfying threads when a thread is
1170 * blocked (e.g., join, lock). A thread which is waiting on another thread can
1171 * no longer satisfy a promise generated from that thread.
1173 * @param blocker The thread on which a thread is waiting
1174 * @param waiting The waiting thread
1176 void ModelExecution::thread_blocking_check_promises(Thread *blocker, Thread *waiting)
1178 for (unsigned int i = 0; i < promises.size(); i++) {
1179 Promise *promise = promises[i];
1180 if (!promise->thread_is_available(waiting->get_id()))
1182 for (unsigned int j = 0; j < promise->get_num_readers(); j++) {
1183 ModelAction *reader = promise->get_reader(j);
1184 if (reader->get_tid() != blocker->get_id())
1186 if (promise->eliminate_thread(waiting->get_id())) {
1187 /* Promise has failed */
1188 priv->failed_promise = true;
1190 /* Only eliminate the 'waiting' thread once */
1198 * @brief Check whether a model action is enabled.
1200 * Checks whether a lock or join operation would be successful (i.e., is the
1201 * lock already locked, or is the joined thread already complete). If not, put
1202 * the action in a waiter list.
1204 * @param curr is the ModelAction to check whether it is enabled.
1205 * @return a bool that indicates whether the action is enabled.
1207 bool ModelExecution::check_action_enabled(ModelAction *curr) {
1208 if (curr->is_lock()) {
1209 std::mutex *lock = curr->get_mutex();
1210 struct std::mutex_state *state = lock->get_state();
1213 } else if (curr->is_thread_join()) {
1214 Thread *blocking = curr->get_thread_operand();
1215 if (!blocking->is_complete()) {
1216 thread_blocking_check_promises(blocking, get_thread(curr));
1219 } else if (params->yieldblock && curr->is_yield()) {
1227 * This is the heart of the model checker routine. It performs model-checking
1228 * actions corresponding to a given "current action." Among other processes, it
1229 * calculates reads-from relationships, updates synchronization clock vectors,
1230 * forms a memory_order constraints graph, and handles replay/backtrack
1231 * execution when running permutations of previously-observed executions.
1233 * @param curr The current action to process
1234 * @return The ModelAction that is actually executed; may be different than
1237 ModelAction * ModelExecution::check_current_action(ModelAction *curr)
1240 bool second_part_of_rmw = curr->is_rmwc() || curr->is_rmw();
1241 bool newly_explored = initialize_curr_action(&curr);
1245 wake_up_sleeping_actions(curr);
1247 /* Compute fairness information for CHESS yield algorithm */
1248 if (params->yieldon) {
1249 curr->get_node()->update_yield(scheduler);
1252 /* Add the action to lists before any other model-checking tasks */
1253 if (!second_part_of_rmw)
1254 add_action_to_lists(curr);
1256 /* Build may_read_from set for newly-created actions */
1257 if (newly_explored && curr->is_read())
1258 build_may_read_from(curr);
1260 /* Initialize work_queue with the "current action" work */
1261 work_queue_t work_queue(1, CheckCurrWorkEntry(curr));
1262 while (!work_queue.empty() && !has_asserted()) {
1263 WorkQueueEntry work = work_queue.front();
1264 work_queue.pop_front();
1266 switch (work.type) {
1267 case WORK_CHECK_CURR_ACTION: {
1268 ModelAction *act = work.action;
1269 bool update = false; /* update this location's release seq's */
1270 bool update_all = false; /* update all release seq's */
1272 if (process_thread_action(curr))
1275 if (act->is_read() && !second_part_of_rmw && process_read(act))
1278 if (act->is_write() && process_write(act))
1281 if (act->is_fence() && process_fence(act))
1284 if (act->is_mutex_op() && process_mutex(act))
1287 if (act->is_relseq_fixup())
1288 process_relseq_fixup(curr, &work_queue);
1291 work_queue.push_back(CheckRelSeqWorkEntry(NULL));
1293 work_queue.push_back(CheckRelSeqWorkEntry(act->get_location()));
1296 case WORK_CHECK_RELEASE_SEQ:
1297 resolve_release_sequences(work.location, &work_queue);
1299 case WORK_CHECK_MO_EDGES: {
1300 /** @todo Complete verification of work_queue */
1301 ModelAction *act = work.action;
1302 bool updated = false;
1304 if (act->is_read()) {
1305 const ModelAction *rf = act->get_reads_from();
1306 const Promise *promise = act->get_reads_from_promise();
1308 if (r_modification_order(act, rf))
1310 } else if (promise) {
1311 if (r_modification_order(act, promise))
1315 if (act->is_write()) {
1316 if (w_modification_order(act, NULL))
1319 mo_graph->commitChanges();
1322 work_queue.push_back(CheckRelSeqWorkEntry(act->get_location()));
1331 check_curr_backtracking(curr);
1332 set_backtracking(curr);
1336 void ModelExecution::check_curr_backtracking(ModelAction *curr)
1338 Node *currnode = curr->get_node();
1339 Node *parnode = currnode->get_parent();
1341 if ((parnode && !parnode->backtrack_empty()) ||
1342 !currnode->misc_empty() ||
1343 !currnode->read_from_empty() ||
1344 !currnode->promise_empty() ||
1345 !currnode->relseq_break_empty()) {
1346 set_latest_backtrack(curr);
1350 bool ModelExecution::promises_expired() const
1352 for (unsigned int i = 0; i < promises.size(); i++) {
1353 Promise *promise = promises[i];
1354 if (promise->get_expiration() < priv->used_sequence_numbers)
1361 * This is the strongest feasibility check available.
1362 * @return whether the current trace (partial or complete) must be a prefix of
1365 bool ModelExecution::isfeasibleprefix() const
1367 return pending_rel_seqs.size() == 0 && is_feasible_prefix_ignore_relseq();
1371 * Print disagnostic information about an infeasible execution
1372 * @param prefix A string to prefix the output with; if NULL, then a default
1373 * message prefix will be provided
1375 void ModelExecution::print_infeasibility(const char *prefix) const
1379 if (mo_graph->checkForCycles())
1380 ptr += sprintf(ptr, "[mo cycle]");
1381 if (priv->failed_promise)
1382 ptr += sprintf(ptr, "[failed promise]");
1383 if (priv->too_many_reads)
1384 ptr += sprintf(ptr, "[too many reads]");
1385 if (priv->no_valid_reads)
1386 ptr += sprintf(ptr, "[no valid reads-from]");
1387 if (priv->bad_synchronization)
1388 ptr += sprintf(ptr, "[bad sw ordering]");
1389 if (promises_expired())
1390 ptr += sprintf(ptr, "[promise expired]");
1391 if (promises.size() != 0)
1392 ptr += sprintf(ptr, "[unresolved promise]");
1394 model_print("%s: %s\n", prefix ? prefix : "Infeasible", buf);
1398 * Returns whether the current completed trace is feasible, except for pending
1399 * release sequences.
1401 bool ModelExecution::is_feasible_prefix_ignore_relseq() const
1403 return !is_infeasible() && promises.size() == 0;
1407 * Check if the current partial trace is infeasible. Does not check any
1408 * end-of-execution flags, which might rule out the execution. Thus, this is
1409 * useful only for ruling an execution as infeasible.
1410 * @return whether the current partial trace is infeasible.
1412 bool ModelExecution::is_infeasible() const
1414 return mo_graph->checkForCycles() ||
1415 priv->no_valid_reads ||
1416 priv->failed_promise ||
1417 priv->too_many_reads ||
1418 priv->bad_synchronization ||
1422 /** Close out a RMWR by converting previous RMWR into a RMW or READ. */
1423 ModelAction * ModelExecution::process_rmw(ModelAction *act) {
1424 ModelAction *lastread = get_last_action(act->get_tid());
1425 lastread->process_rmw(act);
1426 if (act->is_rmw()) {
1427 if (lastread->get_reads_from())
1428 mo_graph->addRMWEdge(lastread->get_reads_from(), lastread);
1430 mo_graph->addRMWEdge(lastread->get_reads_from_promise(), lastread);
1431 mo_graph->commitChanges();
1437 * A helper function for ModelExecution::check_recency, to check if the current
1438 * thread is able to read from a different write/promise for 'params.maxreads'
1439 * number of steps and if that write/promise should become visible (i.e., is
1440 * ordered later in the modification order). This helps model memory liveness.
1442 * @param curr The current action. Must be a read.
1443 * @param rf The write/promise from which we plan to read
1444 * @param other_rf The write/promise from which we may read
1445 * @return True if we were able to read from other_rf for params.maxreads steps
1447 template <typename T, typename U>
1448 bool ModelExecution::should_read_instead(const ModelAction *curr, const T *rf, const U *other_rf) const
1450 /* Need a different write/promise */
1451 if (other_rf->equals(rf))
1454 /* Only look for "newer" writes/promises */
1455 if (!mo_graph->checkReachable(rf, other_rf))
1458 SnapVector<action_list_t> *thrd_lists = obj_thrd_map.get(curr->get_location());
1459 action_list_t *list = &(*thrd_lists)[id_to_int(curr->get_tid())];
1460 action_list_t::reverse_iterator rit = list->rbegin();
1461 ASSERT((*rit) == curr);
1462 /* Skip past curr */
1465 /* Does this write/promise work for everyone? */
1466 for (int i = 0; i < params->maxreads; i++, rit++) {
1467 ModelAction *act = *rit;
1468 if (!act->may_read_from(other_rf))
1475 * Checks whether a thread has read from the same write or Promise for too many
1476 * times without seeing the effects of a later write/Promise.
1479 * 1) there must a different write/promise that we could read from,
1480 * 2) we must have read from the same write/promise in excess of maxreads times,
1481 * 3) that other write/promise must have been in the reads_from set for maxreads times, and
1482 * 4) that other write/promise must be mod-ordered after the write/promise we are reading.
1484 * If so, we decide that the execution is no longer feasible.
1486 * @param curr The current action. Must be a read.
1487 * @param rf The ModelAction/Promise from which we might read.
1488 * @return True if the read should succeed; false otherwise
1490 template <typename T>
1491 bool ModelExecution::check_recency(ModelAction *curr, const T *rf) const
1493 if (!params->maxreads)
1496 //NOTE: Next check is just optimization, not really necessary....
1497 if (curr->get_node()->get_read_from_past_size() +
1498 curr->get_node()->get_read_from_promise_size() <= 1)
1501 SnapVector<action_list_t> *thrd_lists = obj_thrd_map.get(curr->get_location());
1502 int tid = id_to_int(curr->get_tid());
1503 ASSERT(tid < (int)thrd_lists->size());
1504 action_list_t *list = &(*thrd_lists)[tid];
1505 action_list_t::reverse_iterator rit = list->rbegin();
1506 ASSERT((*rit) == curr);
1507 /* Skip past curr */
1510 action_list_t::reverse_iterator ritcopy = rit;
1511 /* See if we have enough reads from the same value */
1512 for (int count = 0; count < params->maxreads; ritcopy++, count++) {
1513 if (ritcopy == list->rend())
1515 ModelAction *act = *ritcopy;
1516 if (!act->is_read())
1518 if (act->get_reads_from_promise() && !act->get_reads_from_promise()->equals(rf))
1520 if (act->get_reads_from() && !act->get_reads_from()->equals(rf))
1522 if (act->get_node()->get_read_from_past_size() +
1523 act->get_node()->get_read_from_promise_size() <= 1)
1526 for (int i = 0; i < curr->get_node()->get_read_from_past_size(); i++) {
1527 const ModelAction *write = curr->get_node()->get_read_from_past(i);
1528 if (should_read_instead(curr, rf, write))
1529 return false; /* liveness failure */
1531 for (int i = 0; i < curr->get_node()->get_read_from_promise_size(); i++) {
1532 const Promise *promise = curr->get_node()->get_read_from_promise(i);
1533 if (should_read_instead(curr, rf, promise))
1534 return false; /* liveness failure */
1540 * @brief Updates the mo_graph with the constraints imposed from the current
1543 * Basic idea is the following: Go through each other thread and find
1544 * the last action that happened before our read. Two cases:
1546 * -# The action is a write: that write must either occur before
1547 * the write we read from or be the write we read from.
1548 * -# The action is a read: the write that that action read from
1549 * must occur before the write we read from or be the same write.
1551 * @param curr The current action. Must be a read.
1552 * @param rf The ModelAction or Promise that curr reads from. Must be a write.
1553 * @return True if modification order edges were added; false otherwise
1555 template <typename rf_type>
1556 bool ModelExecution::r_modification_order(ModelAction *curr, const rf_type *rf)
1558 SnapVector<action_list_t> *thrd_lists = obj_thrd_map.get(curr->get_location());
1561 ASSERT(curr->is_read());
1563 /* Last SC fence in the current thread */
1564 ModelAction *last_sc_fence_local = get_last_seq_cst_fence(curr->get_tid(), NULL);
1565 ModelAction *last_sc_write = NULL;
1566 if (curr->is_seqcst())
1567 last_sc_write = get_last_seq_cst_write(curr);
1569 /* Iterate over all threads */
1570 for (i = 0; i < thrd_lists->size(); i++) {
1571 /* Last SC fence in thread i */
1572 ModelAction *last_sc_fence_thread_local = NULL;
1573 if (int_to_id((int)i) != curr->get_tid())
1574 last_sc_fence_thread_local = get_last_seq_cst_fence(int_to_id(i), NULL);
1576 /* Last SC fence in thread i, before last SC fence in current thread */
1577 ModelAction *last_sc_fence_thread_before = NULL;
1578 if (last_sc_fence_local)
1579 last_sc_fence_thread_before = get_last_seq_cst_fence(int_to_id(i), last_sc_fence_local);
1581 /* Iterate over actions in thread, starting from most recent */
1582 action_list_t *list = &(*thrd_lists)[i];
1583 action_list_t::reverse_iterator rit;
1584 for (rit = list->rbegin(); rit != list->rend(); rit++) {
1585 ModelAction *act = *rit;
1590 /* Don't want to add reflexive edges on 'rf' */
1591 if (act->equals(rf)) {
1592 if (act->happens_before(curr))
1598 if (act->is_write()) {
1599 /* C++, Section 29.3 statement 5 */
1600 if (curr->is_seqcst() && last_sc_fence_thread_local &&
1601 *act < *last_sc_fence_thread_local) {
1602 added = mo_graph->addEdge(act, rf) || added;
1605 /* C++, Section 29.3 statement 4 */
1606 else if (act->is_seqcst() && last_sc_fence_local &&
1607 *act < *last_sc_fence_local) {
1608 added = mo_graph->addEdge(act, rf) || added;
1611 /* C++, Section 29.3 statement 6 */
1612 else if (last_sc_fence_thread_before &&
1613 *act < *last_sc_fence_thread_before) {
1614 added = mo_graph->addEdge(act, rf) || added;
1619 /* C++, Section 29.3 statement 3 (second subpoint) */
1620 if (curr->is_seqcst() && last_sc_write && act == last_sc_write) {
1621 added = mo_graph->addEdge(act, rf) || added;
1626 * Include at most one act per-thread that "happens
1629 if (act->happens_before(curr)) {
1630 if (act->is_write()) {
1631 added = mo_graph->addEdge(act, rf) || added;
1633 const ModelAction *prevrf = act->get_reads_from();
1634 const Promise *prevrf_promise = act->get_reads_from_promise();
1636 if (!prevrf->equals(rf))
1637 added = mo_graph->addEdge(prevrf, rf) || added;
1638 } else if (!prevrf_promise->equals(rf)) {
1639 added = mo_graph->addEdge(prevrf_promise, rf) || added;
1648 * All compatible, thread-exclusive promises must be ordered after any
1649 * concrete loads from the same thread
1651 for (unsigned int i = 0; i < promises.size(); i++)
1652 if (promises[i]->is_compatible_exclusive(curr))
1653 added = mo_graph->addEdge(rf, promises[i]) || added;
1659 * Updates the mo_graph with the constraints imposed from the current write.
1661 * Basic idea is the following: Go through each other thread and find
1662 * the lastest action that happened before our write. Two cases:
1664 * (1) The action is a write => that write must occur before
1667 * (2) The action is a read => the write that that action read from
1668 * must occur before the current write.
1670 * This method also handles two other issues:
1672 * (I) Sequential Consistency: Making sure that if the current write is
1673 * seq_cst, that it occurs after the previous seq_cst write.
1675 * (II) Sending the write back to non-synchronizing reads.
1677 * @param curr The current action. Must be a write.
1678 * @param send_fv A vector for stashing reads to which we may pass our future
1679 * value. If NULL, then don't record any future values.
1680 * @return True if modification order edges were added; false otherwise
1682 bool ModelExecution::w_modification_order(ModelAction *curr, ModelVector<ModelAction *> *send_fv)
1684 SnapVector<action_list_t> *thrd_lists = obj_thrd_map.get(curr->get_location());
1687 ASSERT(curr->is_write());
1689 if (curr->is_seqcst()) {
1690 /* We have to at least see the last sequentially consistent write,
1691 so we are initialized. */
1692 ModelAction *last_seq_cst = get_last_seq_cst_write(curr);
1693 if (last_seq_cst != NULL) {
1694 added = mo_graph->addEdge(last_seq_cst, curr) || added;
1698 /* Last SC fence in the current thread */
1699 ModelAction *last_sc_fence_local = get_last_seq_cst_fence(curr->get_tid(), NULL);
1701 /* Iterate over all threads */
1702 for (i = 0; i < thrd_lists->size(); i++) {
1703 /* Last SC fence in thread i, before last SC fence in current thread */
1704 ModelAction *last_sc_fence_thread_before = NULL;
1705 if (last_sc_fence_local && int_to_id((int)i) != curr->get_tid())
1706 last_sc_fence_thread_before = get_last_seq_cst_fence(int_to_id(i), last_sc_fence_local);
1708 /* Iterate over actions in thread, starting from most recent */
1709 action_list_t *list = &(*thrd_lists)[i];
1710 action_list_t::reverse_iterator rit;
1711 for (rit = list->rbegin(); rit != list->rend(); rit++) {
1712 ModelAction *act = *rit;
1715 * 1) If RMW and it actually read from something, then we
1716 * already have all relevant edges, so just skip to next
1719 * 2) If RMW and it didn't read from anything, we should
1720 * whatever edge we can get to speed up convergence.
1722 * 3) If normal write, we need to look at earlier actions, so
1723 * continue processing list.
1725 if (curr->is_rmw()) {
1726 if (curr->get_reads_from() != NULL)
1734 /* C++, Section 29.3 statement 7 */
1735 if (last_sc_fence_thread_before && act->is_write() &&
1736 *act < *last_sc_fence_thread_before) {
1737 added = mo_graph->addEdge(act, curr) || added;
1742 * Include at most one act per-thread that "happens
1745 if (act->happens_before(curr)) {
1747 * Note: if act is RMW, just add edge:
1749 * The following edge should be handled elsewhere:
1750 * readfrom(act) --mo--> act
1752 if (act->is_write())
1753 added = mo_graph->addEdge(act, curr) || added;
1754 else if (act->is_read()) {
1755 //if previous read accessed a null, just keep going
1756 if (act->get_reads_from() == NULL)
1758 added = mo_graph->addEdge(act->get_reads_from(), curr) || added;
1761 } else if (act->is_read() && !act->could_synchronize_with(curr) &&
1762 !act->same_thread(curr)) {
1763 /* We have an action that:
1764 (1) did not happen before us
1765 (2) is a read and we are a write
1766 (3) cannot synchronize with us
1767 (4) is in a different thread
1769 that read could potentially read from our write. Note that
1770 these checks are overly conservative at this point, we'll
1771 do more checks before actually removing the
1775 if (send_fv && thin_air_constraint_may_allow(curr, act)) {
1776 if (!is_infeasible())
1777 send_fv->push_back(act);
1778 else if (curr->is_rmw() && act->is_rmw() && curr->get_reads_from() && curr->get_reads_from() == act->get_reads_from())
1779 add_future_value(curr, act);
1786 * All compatible, thread-exclusive promises must be ordered after any
1787 * concrete stores to the same thread, or else they can be merged with
1790 for (unsigned int i = 0; i < promises.size(); i++)
1791 if (promises[i]->is_compatible_exclusive(curr))
1792 added = mo_graph->addEdge(curr, promises[i]) || added;
1797 /** Arbitrary reads from the future are not allowed. Section 29.3
1798 * part 9 places some constraints. This method checks one result of constraint
1799 * constraint. Others require compiler support. */
1800 bool ModelExecution::thin_air_constraint_may_allow(const ModelAction *writer, const ModelAction *reader) const
1802 if (!writer->is_rmw())
1805 if (!reader->is_rmw())
1808 for (const ModelAction *search = writer->get_reads_from(); search != NULL; search = search->get_reads_from()) {
1809 if (search == reader)
1811 if (search->get_tid() == reader->get_tid() &&
1812 search->happens_before(reader))
1820 * Arbitrary reads from the future are not allowed. Section 29.3 part 9 places
1821 * some constraints. This method checks one the following constraint (others
1822 * require compiler support):
1824 * If X --hb-> Y --mo-> Z, then X should not read from Z.
1826 bool ModelExecution::mo_may_allow(const ModelAction *writer, const ModelAction *reader)
1828 SnapVector<action_list_t> *thrd_lists = obj_thrd_map.get(reader->get_location());
1830 /* Iterate over all threads */
1831 for (i = 0; i < thrd_lists->size(); i++) {
1832 const ModelAction *write_after_read = NULL;
1834 /* Iterate over actions in thread, starting from most recent */
1835 action_list_t *list = &(*thrd_lists)[i];
1836 action_list_t::reverse_iterator rit;
1837 for (rit = list->rbegin(); rit != list->rend(); rit++) {
1838 ModelAction *act = *rit;
1840 /* Don't disallow due to act == reader */
1841 if (!reader->happens_before(act) || reader == act)
1843 else if (act->is_write())
1844 write_after_read = act;
1845 else if (act->is_read() && act->get_reads_from() != NULL)
1846 write_after_read = act->get_reads_from();
1849 if (write_after_read && write_after_read != writer && mo_graph->checkReachable(write_after_read, writer))
1856 * Finds the head(s) of the release sequence(s) containing a given ModelAction.
1857 * The ModelAction under consideration is expected to be taking part in
1858 * release/acquire synchronization as an object of the "reads from" relation.
1859 * Note that this can only provide release sequence support for RMW chains
1860 * which do not read from the future, as those actions cannot be traced until
1861 * their "promise" is fulfilled. Similarly, we may not even establish the
1862 * presence of a release sequence with certainty, as some modification order
1863 * constraints may be decided further in the future. Thus, this function
1864 * "returns" two pieces of data: a pass-by-reference vector of @a release_heads
1865 * and a boolean representing certainty.
1867 * @param rf The action that might be part of a release sequence. Must be a
1869 * @param release_heads A pass-by-reference style return parameter. After
1870 * execution of this function, release_heads will contain the heads of all the
1871 * relevant release sequences, if any exists with certainty
1872 * @param pending A pass-by-reference style return parameter which is only used
1873 * when returning false (i.e., uncertain). Returns most information regarding
1874 * an uncertain release sequence, including any write operations that might
1875 * break the sequence.
1876 * @return true, if the ModelExecution is certain that release_heads is complete;
1879 bool ModelExecution::release_seq_heads(const ModelAction *rf,
1880 rel_heads_list_t *release_heads,
1881 struct release_seq *pending) const
1883 /* Only check for release sequences if there are no cycles */
1884 if (mo_graph->checkForCycles())
1887 for ( ; rf != NULL; rf = rf->get_reads_from()) {
1888 ASSERT(rf->is_write());
1890 if (rf->is_release())
1891 release_heads->push_back(rf);
1892 else if (rf->get_last_fence_release())
1893 release_heads->push_back(rf->get_last_fence_release());
1895 break; /* End of RMW chain */
1897 /** @todo Need to be smarter here... In the linux lock
1898 * example, this will run to the beginning of the program for
1900 /** @todo The way to be smarter here is to keep going until 1
1901 * thread has a release preceded by an acquire and you've seen
1904 /* acq_rel RMW is a sufficient stopping condition */
1905 if (rf->is_acquire() && rf->is_release())
1906 return true; /* complete */
1909 /* read from future: need to settle this later */
1911 return false; /* incomplete */
1914 if (rf->is_release())
1915 return true; /* complete */
1917 /* else relaxed write
1918 * - check for fence-release in the same thread (29.8, stmt. 3)
1919 * - check modification order for contiguous subsequence
1920 * -> rf must be same thread as release */
1922 const ModelAction *fence_release = rf->get_last_fence_release();
1923 /* Synchronize with a fence-release unconditionally; we don't need to
1924 * find any more "contiguous subsequence..." for it */
1926 release_heads->push_back(fence_release);
1928 int tid = id_to_int(rf->get_tid());
1929 SnapVector<action_list_t> *thrd_lists = obj_thrd_map.get(rf->get_location());
1930 action_list_t *list = &(*thrd_lists)[tid];
1931 action_list_t::const_reverse_iterator rit;
1933 /* Find rf in the thread list */
1934 rit = std::find(list->rbegin(), list->rend(), rf);
1935 ASSERT(rit != list->rend());
1937 /* Find the last {write,fence}-release */
1938 for (; rit != list->rend(); rit++) {
1939 if (fence_release && *(*rit) < *fence_release)
1941 if ((*rit)->is_release())
1944 if (rit == list->rend()) {
1945 /* No write-release in this thread */
1946 return true; /* complete */
1947 } else if (fence_release && *(*rit) < *fence_release) {
1948 /* The fence-release is more recent (and so, "stronger") than
1949 * the most recent write-release */
1950 return true; /* complete */
1951 } /* else, need to establish contiguous release sequence */
1952 ModelAction *release = *rit;
1954 ASSERT(rf->same_thread(release));
1956 pending->writes.clear();
1958 bool certain = true;
1959 for (unsigned int i = 0; i < thrd_lists->size(); i++) {
1960 if (id_to_int(rf->get_tid()) == (int)i)
1962 list = &(*thrd_lists)[i];
1964 /* Can we ensure no future writes from this thread may break
1965 * the release seq? */
1966 bool future_ordered = false;
1968 ModelAction *last = get_last_action(int_to_id(i));
1969 Thread *th = get_thread(int_to_id(i));
1970 if ((last && rf->happens_before(last)) ||
1973 future_ordered = true;
1975 ASSERT(!th->is_model_thread() || future_ordered);
1977 for (rit = list->rbegin(); rit != list->rend(); rit++) {
1978 const ModelAction *act = *rit;
1979 /* Reach synchronization -> this thread is complete */
1980 if (act->happens_before(release))
1982 if (rf->happens_before(act)) {
1983 future_ordered = true;
1987 /* Only non-RMW writes can break release sequences */
1988 if (!act->is_write() || act->is_rmw())
1991 /* Check modification order */
1992 if (mo_graph->checkReachable(rf, act)) {
1993 /* rf --mo--> act */
1994 future_ordered = true;
1997 if (mo_graph->checkReachable(act, release))
1998 /* act --mo--> release */
2000 if (mo_graph->checkReachable(release, act) &&
2001 mo_graph->checkReachable(act, rf)) {
2002 /* release --mo-> act --mo--> rf */
2003 return true; /* complete */
2005 /* act may break release sequence */
2006 pending->writes.push_back(act);
2009 if (!future_ordered)
2010 certain = false; /* This thread is uncertain */
2014 release_heads->push_back(release);
2015 pending->writes.clear();
2017 pending->release = release;
2024 * An interface for getting the release sequence head(s) with which a
2025 * given ModelAction must synchronize. This function only returns a non-empty
2026 * result when it can locate a release sequence head with certainty. Otherwise,
2027 * it may mark the internal state of the ModelExecution so that it will handle
2028 * the release sequence at a later time, causing @a acquire to update its
2029 * synchronization at some later point in execution.
2031 * @param acquire The 'acquire' action that may synchronize with a release
2033 * @param read The read action that may read from a release sequence; this may
2034 * be the same as acquire, or else an earlier action in the same thread (i.e.,
2035 * when 'acquire' is a fence-acquire)
2036 * @param release_heads A pass-by-reference return parameter. Will be filled
2037 * with the head(s) of the release sequence(s), if they exists with certainty.
2038 * @see ModelExecution::release_seq_heads
2040 void ModelExecution::get_release_seq_heads(ModelAction *acquire,
2041 ModelAction *read, rel_heads_list_t *release_heads)
2043 const ModelAction *rf = read->get_reads_from();
2044 struct release_seq *sequence = (struct release_seq *)snapshot_calloc(1, sizeof(struct release_seq));
2045 sequence->acquire = acquire;
2046 sequence->read = read;
2048 if (!release_seq_heads(rf, release_heads, sequence)) {
2049 /* add act to 'lazy checking' list */
2050 pending_rel_seqs.push_back(sequence);
2052 snapshot_free(sequence);
2057 * Attempt to resolve all stashed operations that might synchronize with a
2058 * release sequence for a given location. This implements the "lazy" portion of
2059 * determining whether or not a release sequence was contiguous, since not all
2060 * modification order information is present at the time an action occurs.
2062 * @param location The location/object that should be checked for release
2063 * sequence resolutions. A NULL value means to check all locations.
2064 * @param work_queue The work queue to which to add work items as they are
2066 * @return True if any updates occurred (new synchronization, new mo_graph
2069 bool ModelExecution::resolve_release_sequences(void *location, work_queue_t *work_queue)
2071 bool updated = false;
2072 SnapVector<struct release_seq *>::iterator it = pending_rel_seqs.begin();
2073 while (it != pending_rel_seqs.end()) {
2074 struct release_seq *pending = *it;
2075 ModelAction *acquire = pending->acquire;
2076 const ModelAction *read = pending->read;
2078 /* Only resolve sequences on the given location, if provided */
2079 if (location && read->get_location() != location) {
2084 const ModelAction *rf = read->get_reads_from();
2085 rel_heads_list_t release_heads;
2087 complete = release_seq_heads(rf, &release_heads, pending);
2088 for (unsigned int i = 0; i < release_heads.size(); i++)
2089 if (!acquire->has_synchronized_with(release_heads[i]))
2090 if (synchronize(release_heads[i], acquire))
2094 /* Re-check all pending release sequences */
2095 work_queue->push_back(CheckRelSeqWorkEntry(NULL));
2096 /* Re-check read-acquire for mo_graph edges */
2097 if (acquire->is_read())
2098 work_queue->push_back(MOEdgeWorkEntry(acquire));
2100 /* propagate synchronization to later actions */
2101 action_list_t::reverse_iterator rit = action_trace.rbegin();
2102 for (; (*rit) != acquire; rit++) {
2103 ModelAction *propagate = *rit;
2104 if (acquire->happens_before(propagate)) {
2105 synchronize(acquire, propagate);
2106 /* Re-check 'propagate' for mo_graph edges */
2107 work_queue->push_back(MOEdgeWorkEntry(propagate));
2112 it = pending_rel_seqs.erase(it);
2113 snapshot_free(pending);
2119 // If we resolved promises or data races, see if we have realized a data race.
2126 * Performs various bookkeeping operations for the current ModelAction. For
2127 * instance, adds action to the per-object, per-thread action vector and to the
2128 * action trace list of all thread actions.
2130 * @param act is the ModelAction to add.
2132 void ModelExecution::add_action_to_lists(ModelAction *act)
2134 int tid = id_to_int(act->get_tid());
2135 ModelAction *uninit = NULL;
2137 action_list_t *list = get_safe_ptr_action(&obj_map, act->get_location());
2138 if (list->empty() && act->is_atomic_var()) {
2139 uninit = get_uninitialized_action(act);
2140 uninit_id = id_to_int(uninit->get_tid());
2141 list->push_front(uninit);
2143 list->push_back(act);
2145 action_trace.push_back(act);
2147 action_trace.push_front(uninit);
2149 SnapVector<action_list_t> *vec = get_safe_ptr_vect_action(&obj_thrd_map, act->get_location());
2150 if (tid >= (int)vec->size())
2151 vec->resize(priv->next_thread_id);
2152 (*vec)[tid].push_back(act);
2154 (*vec)[uninit_id].push_front(uninit);
2156 if ((int)thrd_last_action.size() <= tid)
2157 thrd_last_action.resize(get_num_threads());
2158 thrd_last_action[tid] = act;
2160 thrd_last_action[uninit_id] = uninit;
2162 if (act->is_fence() && act->is_release()) {
2163 if ((int)thrd_last_fence_release.size() <= tid)
2164 thrd_last_fence_release.resize(get_num_threads());
2165 thrd_last_fence_release[tid] = act;
2168 if (act->is_wait()) {
2169 void *mutex_loc = (void *) act->get_value();
2170 get_safe_ptr_action(&obj_map, mutex_loc)->push_back(act);
2172 SnapVector<action_list_t> *vec = get_safe_ptr_vect_action(&obj_thrd_map, mutex_loc);
2173 if (tid >= (int)vec->size())
2174 vec->resize(priv->next_thread_id);
2175 (*vec)[tid].push_back(act);
2180 * @brief Get the last action performed by a particular Thread
2181 * @param tid The thread ID of the Thread in question
2182 * @return The last action in the thread
2184 ModelAction * ModelExecution::get_last_action(thread_id_t tid) const
2186 int threadid = id_to_int(tid);
2187 if (threadid < (int)thrd_last_action.size())
2188 return thrd_last_action[id_to_int(tid)];
2194 * @brief Get the last fence release performed by a particular Thread
2195 * @param tid The thread ID of the Thread in question
2196 * @return The last fence release in the thread, if one exists; NULL otherwise
2198 ModelAction * ModelExecution::get_last_fence_release(thread_id_t tid) const
2200 int threadid = id_to_int(tid);
2201 if (threadid < (int)thrd_last_fence_release.size())
2202 return thrd_last_fence_release[id_to_int(tid)];
2208 * Gets the last memory_order_seq_cst write (in the total global sequence)
2209 * performed on a particular object (i.e., memory location), not including the
2211 * @param curr The current ModelAction; also denotes the object location to
2213 * @return The last seq_cst write
2215 ModelAction * ModelExecution::get_last_seq_cst_write(ModelAction *curr) const
2217 void *location = curr->get_location();
2218 action_list_t *list = obj_map.get(location);
2219 /* Find: max({i in dom(S) | seq_cst(t_i) && isWrite(t_i) && samevar(t_i, t)}) */
2220 action_list_t::reverse_iterator rit;
2221 for (rit = list->rbegin(); (*rit) != curr; rit++)
2223 rit++; /* Skip past curr */
2224 for ( ; rit != list->rend(); rit++)
2225 if ((*rit)->is_write() && (*rit)->is_seqcst())
2231 * Gets the last memory_order_seq_cst fence (in the total global sequence)
2232 * performed in a particular thread, prior to a particular fence.
2233 * @param tid The ID of the thread to check
2234 * @param before_fence The fence from which to begin the search; if NULL, then
2235 * search for the most recent fence in the thread.
2236 * @return The last prior seq_cst fence in the thread, if exists; otherwise, NULL
2238 ModelAction * ModelExecution::get_last_seq_cst_fence(thread_id_t tid, const ModelAction *before_fence) const
2240 /* All fences should have location FENCE_LOCATION */
2241 action_list_t *list = obj_map.get(FENCE_LOCATION);
2246 action_list_t::reverse_iterator rit = list->rbegin();
2249 for (; rit != list->rend(); rit++)
2250 if (*rit == before_fence)
2253 ASSERT(*rit == before_fence);
2257 for (; rit != list->rend(); rit++)
2258 if ((*rit)->is_fence() && (tid == (*rit)->get_tid()) && (*rit)->is_seqcst())
2264 * Gets the last unlock operation performed on a particular mutex (i.e., memory
2265 * location). This function identifies the mutex according to the current
2266 * action, which is presumed to perform on the same mutex.
2267 * @param curr The current ModelAction; also denotes the object location to
2269 * @return The last unlock operation
2271 ModelAction * ModelExecution::get_last_unlock(ModelAction *curr) const
2273 void *location = curr->get_location();
2274 action_list_t *list = obj_map.get(location);
2275 /* Find: max({i in dom(S) | isUnlock(t_i) && samevar(t_i, t)}) */
2276 action_list_t::reverse_iterator rit;
2277 for (rit = list->rbegin(); rit != list->rend(); rit++)
2278 if ((*rit)->is_unlock() || (*rit)->is_wait())
2283 ModelAction * ModelExecution::get_parent_action(thread_id_t tid) const
2285 ModelAction *parent = get_last_action(tid);
2287 parent = get_thread(tid)->get_creation();
2292 * Returns the clock vector for a given thread.
2293 * @param tid The thread whose clock vector we want
2294 * @return Desired clock vector
2296 ClockVector * ModelExecution::get_cv(thread_id_t tid) const
2298 return get_parent_action(tid)->get_cv();
2302 * @brief Find the promise (if any) to resolve for the current action and
2303 * remove it from the pending promise vector
2304 * @param curr The current ModelAction. Should be a write.
2305 * @return The Promise to resolve, if any; otherwise NULL
2307 Promise * ModelExecution::pop_promise_to_resolve(const ModelAction *curr)
2309 for (unsigned int i = 0; i < promises.size(); i++)
2310 if (curr->get_node()->get_promise(i)) {
2311 Promise *ret = promises[i];
2312 promises.erase(promises.begin() + i);
2319 * Resolve a Promise with a current write.
2320 * @param write The ModelAction that is fulfilling Promises
2321 * @param promise The Promise to resolve
2322 * @return True if the Promise was successfully resolved; false otherwise
2324 bool ModelExecution::resolve_promise(ModelAction *write, Promise *promise)
2326 ModelVector<ModelAction *> actions_to_check;
2328 for (unsigned int i = 0; i < promise->get_num_readers(); i++) {
2329 ModelAction *read = promise->get_reader(i);
2330 read_from(read, write);
2331 actions_to_check.push_back(read);
2333 /* Make sure the promise's value matches the write's value */
2334 ASSERT(promise->is_compatible(write) && promise->same_value(write));
2335 if (!mo_graph->resolvePromise(promise, write))
2336 priv->failed_promise = true;
2339 * @todo It is possible to end up in an inconsistent state, where a
2340 * "resolved" promise may still be referenced if
2341 * CycleGraph::resolvePromise() failed, so don't delete 'promise'.
2343 * Note that the inconsistency only matters when dumping mo_graph to
2349 //Check whether reading these writes has made threads unable to
2351 for (unsigned int i = 0; i < actions_to_check.size(); i++) {
2352 ModelAction *read = actions_to_check[i];
2353 mo_check_promises(read, true);
2360 * Compute the set of promises that could potentially be satisfied by this
2361 * action. Note that the set computation actually appears in the Node, not in
2363 * @param curr The ModelAction that may satisfy promises
2365 void ModelExecution::compute_promises(ModelAction *curr)
2367 for (unsigned int i = 0; i < promises.size(); i++) {
2368 Promise *promise = promises[i];
2369 if (!promise->is_compatible(curr) || !promise->same_value(curr))
2372 bool satisfy = true;
2373 for (unsigned int j = 0; j < promise->get_num_readers(); j++) {
2374 const ModelAction *act = promise->get_reader(j);
2375 if (act->happens_before(curr) ||
2376 act->could_synchronize_with(curr)) {
2382 curr->get_node()->set_promise(i);
2386 /** Checks promises in response to change in ClockVector Threads. */
2387 void ModelExecution::check_promises(thread_id_t tid, ClockVector *old_cv, ClockVector *merge_cv)
2389 for (unsigned int i = 0; i < promises.size(); i++) {
2390 Promise *promise = promises[i];
2391 if (!promise->thread_is_available(tid))
2393 for (unsigned int j = 0; j < promise->get_num_readers(); j++) {
2394 const ModelAction *act = promise->get_reader(j);
2395 if ((!old_cv || !old_cv->synchronized_since(act)) &&
2396 merge_cv->synchronized_since(act)) {
2397 if (promise->eliminate_thread(tid)) {
2398 /* Promise has failed */
2399 priv->failed_promise = true;
2407 void ModelExecution::check_promises_thread_disabled()
2409 for (unsigned int i = 0; i < promises.size(); i++) {
2410 Promise *promise = promises[i];
2411 if (promise->has_failed()) {
2412 priv->failed_promise = true;
2419 * @brief Checks promises in response to addition to modification order for
2422 * We test whether threads are still available for satisfying promises after an
2423 * addition to our modification order constraints. Those that are unavailable
2424 * are "eliminated". Once all threads are eliminated from satisfying a promise,
2425 * that promise has failed.
2427 * @param act The ModelAction which updated the modification order
2428 * @param is_read_check Should be true if act is a read and we must check for
2429 * updates to the store from which it read (there is a distinction here for
2430 * RMW's, which are both a load and a store)
2432 void ModelExecution::mo_check_promises(const ModelAction *act, bool is_read_check)
2434 const ModelAction *write = is_read_check ? act->get_reads_from() : act;
2436 for (unsigned int i = 0; i < promises.size(); i++) {
2437 Promise *promise = promises[i];
2439 // Is this promise on the same location?
2440 if (!promise->same_location(write))
2443 for (unsigned int j = 0; j < promise->get_num_readers(); j++) {
2444 const ModelAction *pread = promise->get_reader(j);
2445 if (!pread->happens_before(act))
2447 if (mo_graph->checkPromise(write, promise)) {
2448 priv->failed_promise = true;
2454 // Don't do any lookups twice for the same thread
2455 if (!promise->thread_is_available(act->get_tid()))
2458 if (mo_graph->checkReachable(promise, write)) {
2459 if (mo_graph->checkPromise(write, promise)) {
2460 priv->failed_promise = true;
2468 * Compute the set of writes that may break the current pending release
2469 * sequence. This information is extracted from previou release sequence
2472 * @param curr The current ModelAction. Must be a release sequence fixup
2475 void ModelExecution::compute_relseq_breakwrites(ModelAction *curr)
2477 if (pending_rel_seqs.empty())
2480 struct release_seq *pending = pending_rel_seqs.back();
2481 for (unsigned int i = 0; i < pending->writes.size(); i++) {
2482 const ModelAction *write = pending->writes[i];
2483 curr->get_node()->add_relseq_break(write);
2486 /* NULL means don't break the sequence; just synchronize */
2487 curr->get_node()->add_relseq_break(NULL);
2491 * Build up an initial set of all past writes that this 'read' action may read
2492 * from, as well as any previously-observed future values that must still be valid.
2494 * @param curr is the current ModelAction that we are exploring; it must be a
2497 void ModelExecution::build_may_read_from(ModelAction *curr)
2499 SnapVector<action_list_t> *thrd_lists = obj_thrd_map.get(curr->get_location());
2501 ASSERT(curr->is_read());
2503 ModelAction *last_sc_write = NULL;
2505 if (curr->is_seqcst())
2506 last_sc_write = get_last_seq_cst_write(curr);
2508 /* Iterate over all threads */
2509 for (i = 0; i < thrd_lists->size(); i++) {
2510 /* Iterate over actions in thread, starting from most recent */
2511 action_list_t *list = &(*thrd_lists)[i];
2512 action_list_t::reverse_iterator rit;
2513 for (rit = list->rbegin(); rit != list->rend(); rit++) {
2514 ModelAction *act = *rit;
2516 /* Only consider 'write' actions */
2517 if (!act->is_write() || act == curr)
2520 /* Don't consider more than one seq_cst write if we are a seq_cst read. */
2521 bool allow_read = true;
2523 if (curr->is_seqcst() && (act->is_seqcst() || (last_sc_write != NULL && act->happens_before(last_sc_write))) && act != last_sc_write)
2525 else if (curr->get_sleep_flag() && !curr->is_seqcst() && !sleep_can_read_from(curr, act))
2529 /* Only add feasible reads */
2530 mo_graph->startChanges();
2531 r_modification_order(curr, act);
2532 if (!is_infeasible())
2533 curr->get_node()->add_read_from_past(act);
2534 mo_graph->rollbackChanges();
2537 /* Include at most one act per-thread that "happens before" curr */
2538 if (act->happens_before(curr))
2543 /* Inherit existing, promised future values */
2544 for (i = 0; i < promises.size(); i++) {
2545 const Promise *promise = promises[i];
2546 const ModelAction *promise_read = promise->get_reader(0);
2547 if (promise_read->same_var(curr)) {
2548 /* Only add feasible future-values */
2549 mo_graph->startChanges();
2550 r_modification_order(curr, promise);
2551 if (!is_infeasible())
2552 curr->get_node()->add_read_from_promise(promise_read);
2553 mo_graph->rollbackChanges();
2557 /* We may find no valid may-read-from only if the execution is doomed */
2558 if (!curr->get_node()->read_from_size()) {
2559 priv->no_valid_reads = true;
2563 if (DBG_ENABLED()) {
2564 model_print("Reached read action:\n");
2566 model_print("Printing read_from_past\n");
2567 curr->get_node()->print_read_from_past();
2568 model_print("End printing read_from_past\n");
2572 bool ModelExecution::sleep_can_read_from(ModelAction *curr, const ModelAction *write)
2574 for ( ; write != NULL; write = write->get_reads_from()) {
2575 /* UNINIT actions don't have a Node, and they never sleep */
2576 if (write->is_uninitialized())
2578 Node *prevnode = write->get_node()->get_parent();
2580 bool thread_sleep = prevnode->enabled_status(curr->get_tid()) == THREAD_SLEEP_SET;
2581 if (write->is_release() && thread_sleep)
2583 if (!write->is_rmw())
2590 * @brief Get an action representing an uninitialized atomic
2592 * This function may create a new one or try to retrieve one from the NodeStack
2594 * @param curr The current action, which prompts the creation of an UNINIT action
2595 * @return A pointer to the UNINIT ModelAction
2597 ModelAction * ModelExecution::get_uninitialized_action(const ModelAction *curr) const
2599 Node *node = curr->get_node();
2600 ModelAction *act = node->get_uninit_action();
2602 act = new ModelAction(ATOMIC_UNINIT, std::memory_order_relaxed, curr->get_location(), params->uninitvalue, model_thread);
2603 node->set_uninit_action(act);
2605 act->create_cv(NULL);
2609 static void print_list(const action_list_t *list)
2611 action_list_t::const_iterator it;
2613 model_print("---------------------------------------------------------------------\n");
2615 unsigned int hash = 0;
2617 for (it = list->begin(); it != list->end(); it++) {
2618 const ModelAction *act = *it;
2619 if (act->get_seq_number() > 0)
2621 hash = hash^(hash<<3)^((*it)->hash());
2623 model_print("HASH %u\n", hash);
2624 model_print("---------------------------------------------------------------------\n");
2627 #if SUPPORT_MOD_ORDER_DUMP
2628 void ModelExecution::dumpGraph(char *filename) const
2631 sprintf(buffer, "%s.dot", filename);
2632 FILE *file = fopen(buffer, "w");
2633 fprintf(file, "digraph %s {\n", filename);
2634 mo_graph->dumpNodes(file);
2635 ModelAction **thread_array = (ModelAction **)model_calloc(1, sizeof(ModelAction *) * get_num_threads());
2637 for (action_list_t::iterator it = action_trace.begin(); it != action_trace.end(); it++) {
2638 ModelAction *act = *it;
2639 if (act->is_read()) {
2640 mo_graph->dot_print_node(file, act);
2641 if (act->get_reads_from())
2642 mo_graph->dot_print_edge(file,
2643 act->get_reads_from(),
2645 "label=\"rf\", color=red, weight=2");
2647 mo_graph->dot_print_edge(file,
2648 act->get_reads_from_promise(),
2650 "label=\"rf\", color=red");
2652 if (thread_array[act->get_tid()]) {
2653 mo_graph->dot_print_edge(file,
2654 thread_array[id_to_int(act->get_tid())],
2656 "label=\"sb\", color=blue, weight=400");
2659 thread_array[act->get_tid()] = act;
2661 fprintf(file, "}\n");
2662 model_free(thread_array);
2667 /** @brief Prints an execution trace summary. */
2668 void ModelExecution::print_summary() const
2670 #if SUPPORT_MOD_ORDER_DUMP
2671 char buffername[100];
2672 sprintf(buffername, "exec%04u", get_execution_number());
2673 mo_graph->dumpGraphToFile(buffername);
2674 sprintf(buffername, "graph%04u", get_execution_number());
2675 dumpGraph(buffername);
2678 model_print("Execution %d:", get_execution_number());
2679 if (isfeasibleprefix()) {
2680 if (is_yieldblocked())
2681 model_print(" YIELD BLOCKED");
2682 if (scheduler->all_threads_sleeping())
2683 model_print(" SLEEP-SET REDUNDANT");
2686 print_infeasibility(" INFEASIBLE");
2687 print_list(&action_trace);
2689 if (!promises.empty()) {
2690 model_print("Pending promises:\n");
2691 for (unsigned int i = 0; i < promises.size(); i++) {
2692 model_print(" [P%u] ", i);
2693 promises[i]->print();
2700 * Add a Thread to the system for the first time. Should only be called once
2702 * @param t The Thread to add
2704 void ModelExecution::add_thread(Thread *t)
2706 unsigned int i = id_to_int(t->get_id());
2707 if (i >= thread_map.size())
2708 thread_map.resize(i + 1);
2710 if (!t->is_model_thread())
2711 scheduler->add_thread(t);
2715 * @brief Get a Thread reference by its ID
2716 * @param tid The Thread's ID
2717 * @return A Thread reference
2719 Thread * ModelExecution::get_thread(thread_id_t tid) const
2721 unsigned int i = id_to_int(tid);
2722 if (i < thread_map.size())
2723 return thread_map[i];
2728 * @brief Get a reference to the Thread in which a ModelAction was executed
2729 * @param act The ModelAction
2730 * @return A Thread reference
2732 Thread * ModelExecution::get_thread(const ModelAction *act) const
2734 return get_thread(act->get_tid());
2738 * @brief Get a Promise's "promise number"
2740 * A "promise number" is an index number that is unique to a promise, valid
2741 * only for a specific snapshot of an execution trace. Promises may come and go
2742 * as they are generated an resolved, so an index only retains meaning for the
2745 * @param promise The Promise to check
2746 * @return The promise index, if the promise still is valid; otherwise -1
2748 int ModelExecution::get_promise_number(const Promise *promise) const
2750 for (unsigned int i = 0; i < promises.size(); i++)
2751 if (promises[i] == promise)
2758 * @brief Check if a Thread is currently enabled
2759 * @param t The Thread to check
2760 * @return True if the Thread is currently enabled
2762 bool ModelExecution::is_enabled(Thread *t) const
2764 return scheduler->is_enabled(t);
2768 * @brief Check if a Thread is currently enabled
2769 * @param tid The ID of the Thread to check
2770 * @return True if the Thread is currently enabled
2772 bool ModelExecution::is_enabled(thread_id_t tid) const
2774 return scheduler->is_enabled(tid);
2778 * @brief Select the next thread to execute based on the curren action
2780 * RMW actions occur in two parts, and we cannot split them. And THREAD_CREATE
2781 * actions should be followed by the execution of their child thread. In either
2782 * case, the current action should determine the next thread schedule.
2784 * @param curr The current action
2785 * @return The next thread to run, if the current action will determine this
2786 * selection; otherwise NULL
2788 Thread * ModelExecution::action_select_next_thread(const ModelAction *curr) const
2790 /* Do not split atomic RMW */
2791 if (curr->is_rmwr())
2792 return get_thread(curr);
2793 /* Follow CREATE with the created thread */
2794 if (curr->get_type() == THREAD_CREATE)
2795 return curr->get_thread_operand();
2799 /** @return True if the execution has taken too many steps */
2800 bool ModelExecution::too_many_steps() const
2802 return params->bound != 0 && priv->used_sequence_numbers > params->bound;
2806 * Takes the next step in the execution, if possible.
2807 * @param curr The current step to take
2808 * @return Returns the next Thread to run, if any; NULL if this execution
2811 Thread * ModelExecution::take_step(ModelAction *curr)
2813 Thread *curr_thrd = get_thread(curr);
2814 ASSERT(curr_thrd->get_state() == THREAD_READY);
2816 ASSERT(check_action_enabled(curr)); /* May have side effects? */
2817 curr = check_current_action(curr);
2820 if (curr_thrd->is_blocked() || curr_thrd->is_complete())
2821 scheduler->remove_thread(curr_thrd);
2823 return action_select_next_thread(curr);
2827 * Launch end-of-execution release sequence fixups only when
2828 * the execution is otherwise feasible AND there are:
2830 * (1) pending release sequences
2831 * (2) pending assertions that could be invalidated by a change
2832 * in clock vectors (i.e., data races)
2833 * (3) no pending promises
2835 void ModelExecution::fixup_release_sequences()
2837 while (!pending_rel_seqs.empty() &&
2838 is_feasible_prefix_ignore_relseq() &&
2839 haveUnrealizedRaces()) {
2840 model_print("*** WARNING: release sequence fixup action "
2841 "(%zu pending release seuqence(s)) ***\n",
2842 pending_rel_seqs.size());
2843 ModelAction *fixup = new ModelAction(MODEL_FIXUP_RELSEQ,
2844 std::memory_order_seq_cst, NULL, VALUE_NONE,