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 cdsc::mutex *mutex = curr->get_mutex();
691 struct cdsc::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();
758 if (it == waiters->end())
761 advance(it, wakeupthread);
762 scheduler->wake(get_thread(*it));
774 * @brief Check if the current pending promises allow a future value to be sent
776 * It is unsafe to pass a future value back if there exists a pending promise Pr
779 * reader --exec-> Pr --exec-> writer
781 * If such Pr exists, we must save the pending future value until Pr is
784 * @param writer The operation which sends the future value. Must be a write.
785 * @param reader The operation which will observe the value. Must be a read.
786 * @return True if the future value can be sent now; false if it must wait.
788 bool ModelExecution::promises_may_allow(const ModelAction *writer,
789 const ModelAction *reader) const
791 for (int i = promises.size() - 1; i >= 0; i--) {
792 ModelAction *pr = promises[i]->get_reader(0);
793 //reader is after promise...doesn't cross any promise
796 //writer is after promise, reader before...bad...
804 * @brief Add a future value to a reader
806 * This function performs a few additional checks to ensure that the future
807 * value can be feasibly observed by the reader
809 * @param writer The operation whose value is sent. Must be a write.
810 * @param reader The read operation which may read the future value. Must be a read.
812 void ModelExecution::add_future_value(const ModelAction *writer, ModelAction *reader)
814 /* Do more ambitious checks now that mo is more complete */
815 if (!mo_may_allow(writer, reader))
818 Node *node = reader->get_node();
820 /* Find an ancestor thread which exists at the time of the reader */
821 Thread *write_thread = get_thread(writer);
822 while (id_to_int(write_thread->get_id()) >= node->get_num_threads())
823 write_thread = write_thread->get_parent();
825 struct future_value fv = {
826 writer->get_write_value(),
827 writer->get_seq_number() + params->maxfuturedelay,
828 write_thread->get_id(),
830 if (node->add_future_value(fv))
831 set_latest_backtrack(reader);
835 * Process a write ModelAction
836 * @param curr The ModelAction to process
837 * @param work The work queue, for adding fixup work
838 * @return True if the mo_graph was updated or promises were resolved
840 bool ModelExecution::process_write(ModelAction *curr, work_queue_t *work)
842 /* Readers to which we may send our future value */
843 ModelVector<ModelAction *> send_fv;
845 const ModelAction *earliest_promise_reader;
846 bool updated_promises = false;
848 bool updated_mod_order = w_modification_order(curr, &send_fv);
849 Promise *promise = pop_promise_to_resolve(curr);
852 earliest_promise_reader = promise->get_reader(0);
853 updated_promises = resolve_promise(curr, promise, work);
855 earliest_promise_reader = NULL;
857 for (unsigned int i = 0; i < send_fv.size(); i++) {
858 ModelAction *read = send_fv[i];
860 /* Don't send future values to reads after the Promise we resolve */
861 if (!earliest_promise_reader || *read < *earliest_promise_reader) {
862 /* Check if future value can be sent immediately */
863 if (promises_may_allow(curr, read)) {
864 add_future_value(curr, read);
866 futurevalues.push_back(PendingFutureValue(curr, read));
871 /* Check the pending future values */
872 for (int i = (int)futurevalues.size() - 1; i >= 0; i--) {
873 struct PendingFutureValue pfv = futurevalues[i];
874 if (promises_may_allow(pfv.writer, pfv.reader)) {
875 add_future_value(pfv.writer, pfv.reader);
876 futurevalues.erase(futurevalues.begin() + i);
880 mo_graph->commitChanges();
881 mo_check_promises(curr, false);
883 get_thread(curr)->set_return_value(VALUE_NONE);
884 return updated_mod_order || updated_promises;
888 * Process a fence ModelAction
889 * @param curr The ModelAction to process
890 * @return True if synchronization was updated
892 bool ModelExecution::process_fence(ModelAction *curr)
895 * fence-relaxed: no-op
896 * fence-release: only log the occurence (not in this function), for
897 * use in later synchronization
898 * fence-acquire (this function): search for hypothetical release
900 * fence-seq-cst: MO constraints formed in {r,w}_modification_order
902 bool updated = false;
903 if (curr->is_acquire()) {
904 action_list_t *list = &action_trace;
905 action_list_t::reverse_iterator rit;
906 /* Find X : is_read(X) && X --sb-> curr */
907 for (rit = list->rbegin(); rit != list->rend(); rit++) {
908 ModelAction *act = *rit;
911 if (act->get_tid() != curr->get_tid())
913 /* Stop at the beginning of the thread */
914 if (act->is_thread_start())
916 /* Stop once we reach a prior fence-acquire */
917 if (act->is_fence() && act->is_acquire())
921 /* read-acquire will find its own release sequences */
922 if (act->is_acquire())
925 /* Establish hypothetical release sequences */
926 rel_heads_list_t release_heads;
927 get_release_seq_heads(curr, act, &release_heads);
928 for (unsigned int i = 0; i < release_heads.size(); i++)
929 synchronize(release_heads[i], curr);
930 if (release_heads.size() != 0)
938 * @brief Process the current action for thread-related activity
940 * Performs current-action processing for a THREAD_* ModelAction. Proccesses
941 * may include setting Thread status, completing THREAD_FINISH/THREAD_JOIN
942 * synchronization, etc. This function is a no-op for non-THREAD actions
943 * (e.g., ATOMIC_{READ,WRITE,RMW,LOCK}, etc.)
945 * @param curr The current action
946 * @return True if synchronization was updated or a thread completed
948 bool ModelExecution::process_thread_action(ModelAction *curr)
950 bool updated = false;
952 switch (curr->get_type()) {
953 case THREAD_CREATE: {
954 thrd_t *thrd = (thrd_t *)curr->get_location();
955 struct thread_params *params = (struct thread_params *)curr->get_value();
956 Thread *th = new Thread(get_next_id(), thrd, params->func, params->arg, get_thread(curr));
957 curr->set_thread_operand(th);
959 th->set_creation(curr);
960 /* Promises can be satisfied by children */
961 for (unsigned int i = 0; i < promises.size(); i++) {
962 Promise *promise = promises[i];
963 if (promise->thread_is_available(curr->get_tid()))
964 promise->add_thread(th->get_id());
968 case PTHREAD_CREATE: {
969 (*(pthread_t *)curr->get_location()) = pthread_counter++;
971 struct pthread_params *params = (struct pthread_params *)curr->get_value();
972 Thread *th = new Thread(get_next_id(), NULL, params->func, params->arg, get_thread(curr));
973 curr->set_thread_operand(th);
975 th->set_creation(curr);
977 if ( pthread_map.size() < pthread_counter )
978 pthread_map.resize( pthread_counter );
979 pthread_map[ pthread_counter-1 ] = th;
981 /* Promises can be satisfied by children */
982 for (unsigned int i = 0; i < promises.size(); i++) {
983 Promise *promise = promises[i];
984 if (promise->thread_is_available(curr->get_tid()))
985 promise->add_thread(th->get_id());
991 Thread *blocking = curr->get_thread_operand();
992 ModelAction *act = get_last_action(blocking->get_id());
993 synchronize(act, curr);
994 updated = true; /* trigger rel-seq checks */
998 Thread *blocking = curr->get_thread_operand();
999 ModelAction *act = get_last_action(blocking->get_id());
1000 synchronize(act, curr);
1001 updated = true; /* trigger rel-seq checks */
1002 break; // WL: to be add (modified)
1005 case THREAD_FINISH: {
1006 Thread *th = get_thread(curr);
1007 /* Wake up any joining threads */
1008 for (unsigned int i = 0; i < get_num_threads(); i++) {
1009 Thread *waiting = get_thread(int_to_id(i));
1010 if (waiting->waiting_on() == th &&
1011 waiting->get_pending()->is_thread_join())
1012 scheduler->wake(waiting);
1015 /* Completed thread can't satisfy promises */
1016 for (unsigned int i = 0; i < promises.size(); i++) {
1017 Promise *promise = promises[i];
1018 if (promise->thread_is_available(th->get_id()))
1019 if (promise->eliminate_thread(th->get_id()))
1020 priv->failed_promise = true;
1022 updated = true; /* trigger rel-seq checks */
1025 case THREAD_START: {
1026 check_promises(curr->get_tid(), NULL, curr->get_cv());
1037 * @brief Process the current action for release sequence fixup activity
1039 * Performs model-checker release sequence fixups for the current action,
1040 * forcing a single pending release sequence to break (with a given, potential
1041 * "loose" write) or to complete (i.e., synchronize). If a pending release
1042 * sequence forms a complete release sequence, then we must perform the fixup
1043 * synchronization, mo_graph additions, etc.
1045 * @param curr The current action; must be a release sequence fixup action
1046 * @param work_queue The work queue to which to add work items as they are
1049 void ModelExecution::process_relseq_fixup(ModelAction *curr, work_queue_t *work_queue)
1051 const ModelAction *write = curr->get_node()->get_relseq_break();
1052 struct release_seq *sequence = pending_rel_seqs.back();
1053 pending_rel_seqs.pop_back();
1055 ModelAction *acquire = sequence->acquire;
1056 const ModelAction *rf = sequence->rf;
1057 const ModelAction *release = sequence->release;
1061 ASSERT(release->same_thread(rf));
1063 if (write == NULL) {
1065 * @todo Forcing a synchronization requires that we set
1066 * modification order constraints. For instance, we can't allow
1067 * a fixup sequence in which two separate read-acquire
1068 * operations read from the same sequence, where the first one
1069 * synchronizes and the other doesn't. Essentially, we can't
1070 * allow any writes to insert themselves between 'release' and
1074 /* Must synchronize */
1075 if (!synchronize(release, acquire))
1078 /* Propagate the changed clock vector */
1079 propagate_clockvector(acquire, work_queue);
1081 /* Break release sequence with new edges:
1082 * release --mo--> write --mo--> rf */
1083 mo_graph->addEdge(release, write);
1084 mo_graph->addEdge(write, rf);
1087 /* See if we have realized a data race */
1092 * Initialize the current action by performing one or more of the following
1093 * actions, as appropriate: merging RMWR and RMWC/RMW actions, stepping forward
1094 * in the NodeStack, manipulating backtracking sets, allocating and
1095 * initializing clock vectors, and computing the promises to fulfill.
1097 * @param curr The current action, as passed from the user context; may be
1098 * freed/invalidated after the execution of this function, with a different
1099 * action "returned" its place (pass-by-reference)
1100 * @return True if curr is a newly-explored action; false otherwise
1102 bool ModelExecution::initialize_curr_action(ModelAction **curr)
1104 ModelAction *newcurr;
1106 if ((*curr)->is_rmwc() || (*curr)->is_rmw()) {
1107 newcurr = process_rmw(*curr);
1110 if (newcurr->is_rmw())
1111 compute_promises(newcurr);
1117 (*curr)->set_seq_number(get_next_seq_num());
1119 newcurr = node_stack->explore_action(*curr, scheduler->get_enabled_array());
1121 /* First restore type and order in case of RMW operation */
1122 if ((*curr)->is_rmwr())
1123 newcurr->copy_typeandorder(*curr);
1125 ASSERT((*curr)->get_location() == newcurr->get_location());
1126 newcurr->copy_from_new(*curr);
1128 /* Discard duplicate ModelAction; use action from NodeStack */
1131 /* Always compute new clock vector */
1132 newcurr->create_cv(get_parent_action(newcurr->get_tid()));
1135 return false; /* Action was explored previously */
1139 /* Always compute new clock vector */
1140 newcurr->create_cv(get_parent_action(newcurr->get_tid()));
1142 /* Assign most recent release fence */
1143 newcurr->set_last_fence_release(get_last_fence_release(newcurr->get_tid()));
1146 * Perform one-time actions when pushing new ModelAction onto
1149 if (newcurr->is_write())
1150 compute_promises(newcurr);
1151 else if (newcurr->is_relseq_fixup())
1152 compute_relseq_breakwrites(newcurr);
1153 else if (newcurr->is_wait())
1154 newcurr->get_node()->set_misc_max(2);
1155 else if (newcurr->is_notify_one()) {
1156 newcurr->get_node()->set_misc_max(get_safe_ptr_action(&condvar_waiters_map, newcurr->get_location())->size());
1158 return true; /* This was a new ModelAction */
1163 * @brief Establish reads-from relation between two actions
1165 * Perform basic operations involved with establishing a concrete rf relation,
1166 * including setting the ModelAction data and checking for release sequences.
1168 * @param act The action that is reading (must be a read)
1169 * @param rf The action from which we are reading (must be a write)
1171 * @return True if this read established synchronization
1174 bool ModelExecution::read_from(ModelAction *act, const ModelAction *rf)
1177 ASSERT(rf->is_write());
1179 act->set_read_from(rf);
1180 if (act->is_acquire()) {
1181 rel_heads_list_t release_heads;
1182 get_release_seq_heads(act, act, &release_heads);
1183 int num_heads = release_heads.size();
1184 for (unsigned int i = 0; i < release_heads.size(); i++)
1185 if (!synchronize(release_heads[i], act))
1187 return num_heads > 0;
1193 * @brief Synchronizes two actions
1195 * When A synchronizes with B (or A --sw-> B), B inherits A's clock vector.
1196 * This function performs the synchronization as well as providing other hooks
1197 * for other checks along with synchronization.
1199 * @param first The left-hand side of the synchronizes-with relation
1200 * @param second The right-hand side of the synchronizes-with relation
1201 * @return True if the synchronization was successful (i.e., was consistent
1202 * with the execution order); false otherwise
1204 bool ModelExecution::synchronize(const ModelAction *first, ModelAction *second)
1206 if (*second < *first) {
1207 set_bad_synchronization();
1210 check_promises(first->get_tid(), second->get_cv(), first->get_cv());
1211 return second->synchronize_with(first);
1215 * Check promises and eliminate potentially-satisfying threads when a thread is
1216 * blocked (e.g., join, lock). A thread which is waiting on another thread can
1217 * no longer satisfy a promise generated from that thread.
1219 * @param blocker The thread on which a thread is waiting
1220 * @param waiting The waiting thread
1222 void ModelExecution::thread_blocking_check_promises(Thread *blocker, Thread *waiting)
1224 for (unsigned int i = 0; i < promises.size(); i++) {
1225 Promise *promise = promises[i];
1226 if (!promise->thread_is_available(waiting->get_id()))
1228 for (unsigned int j = 0; j < promise->get_num_readers(); j++) {
1229 ModelAction *reader = promise->get_reader(j);
1230 if (reader->get_tid() != blocker->get_id())
1232 if (promise->eliminate_thread(waiting->get_id())) {
1233 /* Promise has failed */
1234 priv->failed_promise = true;
1236 /* Only eliminate the 'waiting' thread once */
1244 * @brief Check whether a model action is enabled.
1246 * Checks whether an operation would be successful (i.e., is a lock already
1247 * locked, or is the joined thread already complete).
1249 * For yield-blocking, yields are never enabled.
1251 * @param curr is the ModelAction to check whether it is enabled.
1252 * @return a bool that indicates whether the action is enabled.
1254 bool ModelExecution::check_action_enabled(ModelAction *curr) {
1255 if (curr->is_lock()) {
1256 cdsc::mutex *lock = curr->get_mutex();
1257 struct cdsc::mutex_state *state = lock->get_state();
1260 } else if (curr->is_thread_join()) {
1261 Thread *blocking = curr->get_thread_operand();
1262 if (!blocking->is_complete()) {
1263 thread_blocking_check_promises(blocking, get_thread(curr));
1266 } else if (params->yieldblock && curr->is_yield()) {
1274 * This is the heart of the model checker routine. It performs model-checking
1275 * actions corresponding to a given "current action." Among other processes, it
1276 * calculates reads-from relationships, updates synchronization clock vectors,
1277 * forms a memory_order constraints graph, and handles replay/backtrack
1278 * execution when running permutations of previously-observed executions.
1280 * @param curr The current action to process
1281 * @return The ModelAction that is actually executed; may be different than
1284 ModelAction * ModelExecution::check_current_action(ModelAction *curr)
1287 bool second_part_of_rmw = curr->is_rmwc() || curr->is_rmw();
1288 bool newly_explored = initialize_curr_action(&curr);
1292 wake_up_sleeping_actions(curr);
1294 /* Compute fairness information for CHESS yield algorithm */
1295 if (params->yieldon) {
1296 curr->get_node()->update_yield(scheduler);
1299 /* Add the action to lists before any other model-checking tasks */
1300 if (!second_part_of_rmw)
1301 add_action_to_lists(curr);
1303 /* Build may_read_from set for newly-created actions */
1304 if (newly_explored && curr->is_read())
1305 build_may_read_from(curr);
1307 /* Initialize work_queue with the "current action" work */
1308 work_queue_t work_queue(1, CheckCurrWorkEntry(curr));
1309 while (!work_queue.empty() && !has_asserted()) {
1310 WorkQueueEntry work = work_queue.front();
1311 work_queue.pop_front();
1313 switch (work.type) {
1314 case WORK_CHECK_CURR_ACTION: {
1315 ModelAction *act = work.action;
1316 bool update = false; /* update this location's release seq's */
1317 bool update_all = false; /* update all release seq's */
1319 if (process_thread_action(curr))
1322 if (act->is_read() && !second_part_of_rmw && process_read(act))
1325 if (act->is_write() && process_write(act, &work_queue))
1328 if (act->is_fence() && process_fence(act))
1331 if (act->is_mutex_op() && process_mutex(act))
1334 if (act->is_relseq_fixup())
1335 process_relseq_fixup(curr, &work_queue);
1338 work_queue.push_back(CheckRelSeqWorkEntry(NULL));
1340 work_queue.push_back(CheckRelSeqWorkEntry(act->get_location()));
1343 case WORK_CHECK_RELEASE_SEQ:
1344 resolve_release_sequences(work.location, &work_queue);
1346 case WORK_CHECK_MO_EDGES: {
1347 /** @todo Complete verification of work_queue */
1348 ModelAction *act = work.action;
1349 bool updated = false;
1351 if (act->is_read()) {
1352 const ModelAction *rf = act->get_reads_from();
1353 const Promise *promise = act->get_reads_from_promise();
1355 if (r_modification_order(act, rf))
1357 if (act->is_seqcst()) {
1358 ModelAction *last_sc_write = get_last_seq_cst_write(act);
1359 if (last_sc_write != NULL && rf->happens_before(last_sc_write)) {
1363 } else if (promise) {
1364 if (r_modification_order(act, promise))
1368 if (act->is_write()) {
1369 if (w_modification_order(act, NULL))
1372 mo_graph->commitChanges();
1375 work_queue.push_back(CheckRelSeqWorkEntry(act->get_location()));
1384 check_curr_backtracking(curr);
1385 set_backtracking(curr);
1389 void ModelExecution::check_curr_backtracking(ModelAction *curr)
1391 Node *currnode = curr->get_node();
1392 Node *parnode = currnode->get_parent();
1394 if ((parnode && !parnode->backtrack_empty()) ||
1395 !currnode->misc_empty() ||
1396 !currnode->read_from_empty() ||
1397 !currnode->promise_empty() ||
1398 !currnode->relseq_break_empty()) {
1399 set_latest_backtrack(curr);
1403 bool ModelExecution::promises_expired() const
1405 for (unsigned int i = 0; i < promises.size(); i++) {
1406 Promise *promise = promises[i];
1407 if (promise->get_expiration() < priv->used_sequence_numbers)
1414 * This is the strongest feasibility check available.
1415 * @return whether the current trace (partial or complete) must be a prefix of
1418 bool ModelExecution::isfeasibleprefix() const
1420 return pending_rel_seqs.size() == 0 && is_feasible_prefix_ignore_relseq();
1424 * Print disagnostic information about an infeasible execution
1425 * @param prefix A string to prefix the output with; if NULL, then a default
1426 * message prefix will be provided
1428 void ModelExecution::print_infeasibility(const char *prefix) const
1432 if (mo_graph->checkForCycles())
1433 ptr += sprintf(ptr, "[mo cycle]");
1434 if (priv->failed_promise || priv->hard_failed_promise)
1435 ptr += sprintf(ptr, "[failed promise]");
1436 if (priv->too_many_reads)
1437 ptr += sprintf(ptr, "[too many reads]");
1438 if (priv->no_valid_reads)
1439 ptr += sprintf(ptr, "[no valid reads-from]");
1440 if (priv->bad_synchronization)
1441 ptr += sprintf(ptr, "[bad sw ordering]");
1442 if (priv->bad_sc_read)
1443 ptr += sprintf(ptr, "[bad sc read]");
1444 if (promises_expired())
1445 ptr += sprintf(ptr, "[promise expired]");
1446 if (promises.size() != 0)
1447 ptr += sprintf(ptr, "[unresolved promise]");
1449 model_print("%s: %s", prefix ? prefix : "Infeasible", buf);
1453 * Returns whether the current completed trace is feasible, except for pending
1454 * release sequences.
1456 bool ModelExecution::is_feasible_prefix_ignore_relseq() const
1458 return !is_infeasible() && promises.size() == 0 && ! priv->failed_promise;
1463 * Check if the current partial trace is infeasible. Does not check any
1464 * end-of-execution flags, which might rule out the execution. Thus, this is
1465 * useful only for ruling an execution as infeasible.
1466 * @return whether the current partial trace is infeasible.
1468 bool ModelExecution::is_infeasible() const
1470 return mo_graph->checkForCycles() ||
1471 priv->no_valid_reads ||
1472 priv->too_many_reads ||
1473 priv->bad_synchronization ||
1474 priv->bad_sc_read ||
1475 priv->hard_failed_promise ||
1479 /** Close out a RMWR by converting previous RMWR into a RMW or READ. */
1480 ModelAction * ModelExecution::process_rmw(ModelAction *act) {
1481 ModelAction *lastread = get_last_action(act->get_tid());
1482 lastread->process_rmw(act);
1483 if (act->is_rmw()) {
1484 if (lastread->get_reads_from())
1485 mo_graph->addRMWEdge(lastread->get_reads_from(), lastread);
1487 mo_graph->addRMWEdge(lastread->get_reads_from_promise(), lastread);
1488 mo_graph->commitChanges();
1494 * A helper function for ModelExecution::check_recency, to check if the current
1495 * thread is able to read from a different write/promise for 'params.maxreads'
1496 * number of steps and if that write/promise should become visible (i.e., is
1497 * ordered later in the modification order). This helps model memory liveness.
1499 * @param curr The current action. Must be a read.
1500 * @param rf The write/promise from which we plan to read
1501 * @param other_rf The write/promise from which we may read
1502 * @return True if we were able to read from other_rf for params.maxreads steps
1504 template <typename T, typename U>
1505 bool ModelExecution::should_read_instead(const ModelAction *curr, const T *rf, const U *other_rf) const
1507 /* Need a different write/promise */
1508 if (other_rf->equals(rf))
1511 /* Only look for "newer" writes/promises */
1512 if (!mo_graph->checkReachable(rf, other_rf))
1515 SnapVector<action_list_t> *thrd_lists = obj_thrd_map.get(curr->get_location());
1516 action_list_t *list = &(*thrd_lists)[id_to_int(curr->get_tid())];
1517 action_list_t::reverse_iterator rit = list->rbegin();
1518 ASSERT((*rit) == curr);
1519 /* Skip past curr */
1522 /* Does this write/promise work for everyone? */
1523 for (int i = 0; i < params->maxreads; i++, rit++) {
1524 ModelAction *act = *rit;
1525 if (!act->may_read_from(other_rf))
1532 * Checks whether a thread has read from the same write or Promise for too many
1533 * times without seeing the effects of a later write/Promise.
1536 * 1) there must a different write/promise that we could read from,
1537 * 2) we must have read from the same write/promise in excess of maxreads times,
1538 * 3) that other write/promise must have been in the reads_from set for maxreads times, and
1539 * 4) that other write/promise must be mod-ordered after the write/promise we are reading.
1541 * If so, we decide that the execution is no longer feasible.
1543 * @param curr The current action. Must be a read.
1544 * @param rf The ModelAction/Promise from which we might read.
1545 * @return True if the read should succeed; false otherwise
1547 template <typename T>
1548 bool ModelExecution::check_recency(ModelAction *curr, const T *rf) const
1550 if (!params->maxreads)
1553 //NOTE: Next check is just optimization, not really necessary....
1554 if (curr->get_node()->get_read_from_past_size() +
1555 curr->get_node()->get_read_from_promise_size() <= 1)
1558 SnapVector<action_list_t> *thrd_lists = obj_thrd_map.get(curr->get_location());
1559 int tid = id_to_int(curr->get_tid());
1560 ASSERT(tid < (int)thrd_lists->size());
1561 action_list_t *list = &(*thrd_lists)[tid];
1562 action_list_t::reverse_iterator rit = list->rbegin();
1563 ASSERT((*rit) == curr);
1564 /* Skip past curr */
1567 action_list_t::reverse_iterator ritcopy = rit;
1568 /* See if we have enough reads from the same value */
1569 for (int count = 0; count < params->maxreads; ritcopy++, count++) {
1570 if (ritcopy == list->rend())
1572 ModelAction *act = *ritcopy;
1573 if (!act->is_read())
1575 if (act->get_reads_from_promise() && !act->get_reads_from_promise()->equals(rf))
1577 if (act->get_reads_from() && !act->get_reads_from()->equals(rf))
1579 if (act->get_node()->get_read_from_past_size() +
1580 act->get_node()->get_read_from_promise_size() <= 1)
1583 for (int i = 0; i < curr->get_node()->get_read_from_past_size(); i++) {
1584 const ModelAction *write = curr->get_node()->get_read_from_past(i);
1585 if (should_read_instead(curr, rf, write))
1586 return false; /* liveness failure */
1588 for (int i = 0; i < curr->get_node()->get_read_from_promise_size(); i++) {
1589 const Promise *promise = curr->get_node()->get_read_from_promise(i);
1590 if (should_read_instead(curr, rf, promise))
1591 return false; /* liveness failure */
1597 * @brief Updates the mo_graph with the constraints imposed from the current
1600 * Basic idea is the following: Go through each other thread and find
1601 * the last action that happened before our read. Two cases:
1603 * -# The action is a write: that write must either occur before
1604 * the write we read from or be the write we read from.
1605 * -# The action is a read: the write that that action read from
1606 * must occur before the write we read from or be the same write.
1608 * @param curr The current action. Must be a read.
1609 * @param rf The ModelAction or Promise that curr reads from. Must be a write.
1610 * @return True if modification order edges were added; false otherwise
1612 template <typename rf_type>
1613 bool ModelExecution::r_modification_order(ModelAction *curr, const rf_type *rf)
1615 SnapVector<action_list_t> *thrd_lists = obj_thrd_map.get(curr->get_location());
1618 ASSERT(curr->is_read());
1620 /* Last SC fence in the current thread */
1621 ModelAction *last_sc_fence_local = get_last_seq_cst_fence(curr->get_tid(), NULL);
1622 ModelAction *last_sc_write = NULL;
1623 if (curr->is_seqcst())
1624 last_sc_write = get_last_seq_cst_write(curr);
1626 /* Iterate over all threads */
1627 for (i = 0; i < thrd_lists->size(); i++) {
1628 /* Last SC fence in thread i */
1629 ModelAction *last_sc_fence_thread_local = NULL;
1630 if (int_to_id((int)i) != curr->get_tid())
1631 last_sc_fence_thread_local = get_last_seq_cst_fence(int_to_id(i), NULL);
1633 /* Last SC fence in thread i, before last SC fence in current thread */
1634 ModelAction *last_sc_fence_thread_before = NULL;
1635 if (last_sc_fence_local)
1636 last_sc_fence_thread_before = get_last_seq_cst_fence(int_to_id(i), last_sc_fence_local);
1638 /* Iterate over actions in thread, starting from most recent */
1639 action_list_t *list = &(*thrd_lists)[i];
1640 action_list_t::reverse_iterator rit;
1641 for (rit = list->rbegin(); rit != list->rend(); rit++) {
1642 ModelAction *act = *rit;
1647 /* Don't want to add reflexive edges on 'rf' */
1648 if (act->equals(rf)) {
1649 if (act->happens_before(curr))
1655 if (act->is_write()) {
1656 /* C++, Section 29.3 statement 5 */
1657 if (curr->is_seqcst() && last_sc_fence_thread_local &&
1658 *act < *last_sc_fence_thread_local) {
1659 added = mo_graph->addEdge(act, rf) || added;
1662 /* C++, Section 29.3 statement 4 */
1663 else if (act->is_seqcst() && last_sc_fence_local &&
1664 *act < *last_sc_fence_local) {
1665 added = mo_graph->addEdge(act, rf) || added;
1668 /* C++, Section 29.3 statement 6 */
1669 else if (last_sc_fence_thread_before &&
1670 *act < *last_sc_fence_thread_before) {
1671 added = mo_graph->addEdge(act, rf) || added;
1677 * Include at most one act per-thread that "happens
1680 if (act->happens_before(curr)) {
1681 if (act->is_write()) {
1682 added = mo_graph->addEdge(act, rf) || added;
1684 const ModelAction *prevrf = act->get_reads_from();
1685 const Promise *prevrf_promise = act->get_reads_from_promise();
1687 if (!prevrf->equals(rf))
1688 added = mo_graph->addEdge(prevrf, rf) || added;
1689 } else if (!prevrf_promise->equals(rf)) {
1690 added = mo_graph->addEdge(prevrf_promise, rf) || added;
1699 * All compatible, thread-exclusive promises must be ordered after any
1700 * concrete loads from the same thread
1702 for (unsigned int i = 0; i < promises.size(); i++)
1703 if (promises[i]->is_compatible_exclusive(curr))
1704 added = mo_graph->addEdge(rf, promises[i]) || added;
1710 * Updates the mo_graph with the constraints imposed from the current write.
1712 * Basic idea is the following: Go through each other thread and find
1713 * the lastest action that happened before our write. Two cases:
1715 * (1) The action is a write => that write must occur before
1718 * (2) The action is a read => the write that that action read from
1719 * must occur before the current write.
1721 * This method also handles two other issues:
1723 * (I) Sequential Consistency: Making sure that if the current write is
1724 * seq_cst, that it occurs after the previous seq_cst write.
1726 * (II) Sending the write back to non-synchronizing reads.
1728 * @param curr The current action. Must be a write.
1729 * @param send_fv A vector for stashing reads to which we may pass our future
1730 * value. If NULL, then don't record any future values.
1731 * @return True if modification order edges were added; false otherwise
1733 bool ModelExecution::w_modification_order(ModelAction *curr, ModelVector<ModelAction *> *send_fv)
1735 SnapVector<action_list_t> *thrd_lists = obj_thrd_map.get(curr->get_location());
1738 ASSERT(curr->is_write());
1740 if (curr->is_seqcst()) {
1741 /* We have to at least see the last sequentially consistent write,
1742 so we are initialized. */
1743 ModelAction *last_seq_cst = get_last_seq_cst_write(curr);
1744 if (last_seq_cst != NULL) {
1745 added = mo_graph->addEdge(last_seq_cst, curr) || added;
1749 /* Last SC fence in the current thread */
1750 ModelAction *last_sc_fence_local = get_last_seq_cst_fence(curr->get_tid(), NULL);
1752 /* Iterate over all threads */
1753 for (i = 0; i < thrd_lists->size(); i++) {
1754 /* Last SC fence in thread i, before last SC fence in current thread */
1755 ModelAction *last_sc_fence_thread_before = NULL;
1756 if (last_sc_fence_local && int_to_id((int)i) != curr->get_tid())
1757 last_sc_fence_thread_before = get_last_seq_cst_fence(int_to_id(i), last_sc_fence_local);
1759 /* Iterate over actions in thread, starting from most recent */
1760 action_list_t *list = &(*thrd_lists)[i];
1761 action_list_t::reverse_iterator rit;
1762 for (rit = list->rbegin(); rit != list->rend(); rit++) {
1763 ModelAction *act = *rit;
1766 * 1) If RMW and it actually read from something, then we
1767 * already have all relevant edges, so just skip to next
1770 * 2) If RMW and it didn't read from anything, we should
1771 * whatever edge we can get to speed up convergence.
1773 * 3) If normal write, we need to look at earlier actions, so
1774 * continue processing list.
1776 if (curr->is_rmw()) {
1777 if (curr->get_reads_from() != NULL)
1785 /* C++, Section 29.3 statement 7 */
1786 if (last_sc_fence_thread_before && act->is_write() &&
1787 *act < *last_sc_fence_thread_before) {
1788 added = mo_graph->addEdge(act, curr) || added;
1793 * Include at most one act per-thread that "happens
1796 if (act->happens_before(curr)) {
1798 * Note: if act is RMW, just add edge:
1800 * The following edge should be handled elsewhere:
1801 * readfrom(act) --mo--> act
1803 if (act->is_write())
1804 added = mo_graph->addEdge(act, curr) || added;
1805 else if (act->is_read()) {
1806 //if previous read accessed a null, just keep going
1807 if (act->get_reads_from() == NULL) {
1808 added = mo_graph->addEdge(act->get_reads_from_promise(), curr) || added;
1810 added = mo_graph->addEdge(act->get_reads_from(), curr) || added;
1813 } else if (act->is_read() && !act->could_synchronize_with(curr) &&
1814 !act->same_thread(curr)) {
1815 /* We have an action that:
1816 (1) did not happen before us
1817 (2) is a read and we are a write
1818 (3) cannot synchronize with us
1819 (4) is in a different thread
1821 that read could potentially read from our write. Note that
1822 these checks are overly conservative at this point, we'll
1823 do more checks before actually removing the
1828 if (send_fv && thin_air_constraint_may_allow(curr, act) && check_coherence_promise(curr, act)) {
1829 if (!is_infeasible())
1830 send_fv->push_back(act);
1831 else if (curr->is_rmw() && act->is_rmw() && curr->get_reads_from() && curr->get_reads_from() == act->get_reads_from())
1832 add_future_value(curr, act);
1839 * All compatible, thread-exclusive promises must be ordered after any
1840 * concrete stores to the same thread, or else they can be merged with
1843 for (unsigned int i = 0; i < promises.size(); i++)
1844 if (promises[i]->is_compatible_exclusive(curr))
1845 added = mo_graph->addEdge(curr, promises[i]) || added;
1850 //This procedure uses cohere to prune future values that are
1851 //guaranteed to generate a coherence violation.
1853 //need to see if there is (1) a promise for thread write, (2)
1854 //the promise is sb before write, (3) the promise can only be
1855 //resolved by the thread read, and (4) the promise has same
1856 //location as read/write
1858 bool ModelExecution::check_coherence_promise(const ModelAction * write, const ModelAction *read) {
1859 thread_id_t write_tid=write->get_tid();
1860 for(unsigned int i = promises.size(); i>0; i--) {
1861 Promise *pr=promises[i-1];
1862 if (!pr->same_location(write))
1864 //the reading thread is the only thread that can resolve the promise
1865 if (pr->get_num_was_available_threads()==1 && pr->thread_was_available(read->get_tid())) {
1866 for(unsigned int j=0;j<pr->get_num_readers();j++) {
1867 ModelAction *prreader=pr->get_reader(j);
1868 //the writing thread reads from the promise before the write
1869 if (prreader->get_tid()==write_tid &&
1870 (*prreader)<(*write)) {
1871 if ((*read)>(*prreader)) {
1872 //check that we don't have a read between the read and promise
1873 //from the same thread as read
1875 for(const ModelAction *tmp=read;tmp!=prreader;) {
1876 tmp=tmp->get_node()->get_parent()->get_action();
1877 if (tmp->is_read() && tmp->same_thread(read)) {
1894 /** Arbitrary reads from the future are not allowed. Section 29.3
1895 * part 9 places some constraints. This method checks one result of constraint
1896 * constraint. Others require compiler support. */
1897 bool ModelExecution::thin_air_constraint_may_allow(const ModelAction *writer, const ModelAction *reader) const
1899 if (!writer->is_rmw())
1902 if (!reader->is_rmw())
1905 for (const ModelAction *search = writer->get_reads_from(); search != NULL; search = search->get_reads_from()) {
1906 if (search == reader)
1908 if (search->get_tid() == reader->get_tid() &&
1909 search->happens_before(reader))
1917 * Arbitrary reads from the future are not allowed. Section 29.3 part 9 places
1918 * some constraints. This method checks one the following constraint (others
1919 * require compiler support):
1921 * If X --hb-> Y --mo-> Z, then X should not read from Z.
1922 * If X --hb-> Y, A --rf-> Y, and A --mo-> Z, then X should not read from Z.
1924 bool ModelExecution::mo_may_allow(const ModelAction *writer, const ModelAction *reader)
1926 SnapVector<action_list_t> *thrd_lists = obj_thrd_map.get(reader->get_location());
1928 /* Iterate over all threads */
1929 for (i = 0; i < thrd_lists->size(); i++) {
1930 const ModelAction *write_after_read = NULL;
1932 /* Iterate over actions in thread, starting from most recent */
1933 action_list_t *list = &(*thrd_lists)[i];
1934 action_list_t::reverse_iterator rit;
1935 for (rit = list->rbegin(); rit != list->rend(); rit++) {
1936 ModelAction *act = *rit;
1938 /* Don't disallow due to act == reader */
1939 if (!reader->happens_before(act) || reader == act)
1941 else if (act->is_write())
1942 write_after_read = act;
1943 else if (act->is_read() && act->get_reads_from() != NULL)
1944 write_after_read = act->get_reads_from();
1947 if (write_after_read && write_after_read != writer && mo_graph->checkReachable(write_after_read, writer))
1954 * Finds the head(s) of the release sequence(s) containing a given ModelAction.
1955 * The ModelAction under consideration is expected to be taking part in
1956 * release/acquire synchronization as an object of the "reads from" relation.
1957 * Note that this can only provide release sequence support for RMW chains
1958 * which do not read from the future, as those actions cannot be traced until
1959 * their "promise" is fulfilled. Similarly, we may not even establish the
1960 * presence of a release sequence with certainty, as some modification order
1961 * constraints may be decided further in the future. Thus, this function
1962 * "returns" two pieces of data: a pass-by-reference vector of @a release_heads
1963 * and a boolean representing certainty.
1965 * @param rf The action that might be part of a release sequence. Must be a
1967 * @param release_heads A pass-by-reference style return parameter. After
1968 * execution of this function, release_heads will contain the heads of all the
1969 * relevant release sequences, if any exists with certainty
1970 * @param pending A pass-by-reference style return parameter which is only used
1971 * when returning false (i.e., uncertain). Returns most information regarding
1972 * an uncertain release sequence, including any write operations that might
1973 * break the sequence.
1974 * @return true, if the ModelExecution is certain that release_heads is complete;
1977 bool ModelExecution::release_seq_heads(const ModelAction *rf,
1978 rel_heads_list_t *release_heads,
1979 struct release_seq *pending) const
1981 /* Only check for release sequences if there are no cycles */
1982 if (mo_graph->checkForCycles())
1985 for ( ; rf != NULL; rf = rf->get_reads_from()) {
1986 ASSERT(rf->is_write());
1988 if (rf->is_release())
1989 release_heads->push_back(rf);
1990 else if (rf->get_last_fence_release())
1991 release_heads->push_back(rf->get_last_fence_release());
1993 break; /* End of RMW chain */
1995 /** @todo Need to be smarter here... In the linux lock
1996 * example, this will run to the beginning of the program for
1998 /** @todo The way to be smarter here is to keep going until 1
1999 * thread has a release preceded by an acquire and you've seen
2002 /* acq_rel RMW is a sufficient stopping condition */
2003 if (rf->is_acquire() && rf->is_release())
2004 return true; /* complete */
2007 /* read from future: need to settle this later */
2009 return false; /* incomplete */
2012 if (rf->is_release())
2013 return true; /* complete */
2015 /* else relaxed write
2016 * - check for fence-release in the same thread (29.8, stmt. 3)
2017 * - check modification order for contiguous subsequence
2018 * -> rf must be same thread as release */
2020 const ModelAction *fence_release = rf->get_last_fence_release();
2021 /* Synchronize with a fence-release unconditionally; we don't need to
2022 * find any more "contiguous subsequence..." for it */
2024 release_heads->push_back(fence_release);
2026 int tid = id_to_int(rf->get_tid());
2027 SnapVector<action_list_t> *thrd_lists = obj_thrd_map.get(rf->get_location());
2028 action_list_t *list = &(*thrd_lists)[tid];
2029 action_list_t::const_reverse_iterator rit;
2031 /* Find rf in the thread list */
2032 rit = std::find(list->rbegin(), list->rend(), rf);
2033 ASSERT(rit != list->rend());
2035 /* Find the last {write,fence}-release */
2036 for (; rit != list->rend(); rit++) {
2037 if (fence_release && *(*rit) < *fence_release)
2039 if ((*rit)->is_release())
2042 if (rit == list->rend()) {
2043 /* No write-release in this thread */
2044 return true; /* complete */
2045 } else if (fence_release && *(*rit) < *fence_release) {
2046 /* The fence-release is more recent (and so, "stronger") than
2047 * the most recent write-release */
2048 return true; /* complete */
2049 } /* else, need to establish contiguous release sequence */
2050 ModelAction *release = *rit;
2052 ASSERT(rf->same_thread(release));
2054 pending->writes.clear();
2056 bool certain = true;
2057 for (unsigned int i = 0; i < thrd_lists->size(); i++) {
2058 if (id_to_int(rf->get_tid()) == (int)i)
2060 list = &(*thrd_lists)[i];
2062 /* Can we ensure no future writes from this thread may break
2063 * the release seq? */
2064 bool future_ordered = false;
2066 ModelAction *last = get_last_action(int_to_id(i));
2067 Thread *th = get_thread(int_to_id(i));
2068 if ((last && rf->happens_before(last)) ||
2071 future_ordered = true;
2073 ASSERT(!th->is_model_thread() || future_ordered);
2075 for (rit = list->rbegin(); rit != list->rend(); rit++) {
2076 const ModelAction *act = *rit;
2077 /* Reach synchronization -> this thread is complete */
2078 if (act->happens_before(release))
2080 if (rf->happens_before(act)) {
2081 future_ordered = true;
2085 /* Only non-RMW writes can break release sequences */
2086 if (!act->is_write() || act->is_rmw())
2089 /* Check modification order */
2090 if (mo_graph->checkReachable(rf, act)) {
2091 /* rf --mo--> act */
2092 future_ordered = true;
2095 if (mo_graph->checkReachable(act, release))
2096 /* act --mo--> release */
2098 if (mo_graph->checkReachable(release, act) &&
2099 mo_graph->checkReachable(act, rf)) {
2100 /* release --mo-> act --mo--> rf */
2101 return true; /* complete */
2103 /* act may break release sequence */
2104 pending->writes.push_back(act);
2107 if (!future_ordered)
2108 certain = false; /* This thread is uncertain */
2112 release_heads->push_back(release);
2113 pending->writes.clear();
2115 pending->release = release;
2122 * An interface for getting the release sequence head(s) with which a
2123 * given ModelAction must synchronize. This function only returns a non-empty
2124 * result when it can locate a release sequence head with certainty. Otherwise,
2125 * it may mark the internal state of the ModelExecution so that it will handle
2126 * the release sequence at a later time, causing @a acquire to update its
2127 * synchronization at some later point in execution.
2129 * @param acquire The 'acquire' action that may synchronize with a release
2131 * @param read The read action that may read from a release sequence; this may
2132 * be the same as acquire, or else an earlier action in the same thread (i.e.,
2133 * when 'acquire' is a fence-acquire)
2134 * @param release_heads A pass-by-reference return parameter. Will be filled
2135 * with the head(s) of the release sequence(s), if they exists with certainty.
2136 * @see ModelExecution::release_seq_heads
2138 void ModelExecution::get_release_seq_heads(ModelAction *acquire,
2139 ModelAction *read, rel_heads_list_t *release_heads)
2141 const ModelAction *rf = read->get_reads_from();
2142 struct release_seq *sequence = (struct release_seq *)snapshot_calloc(1, sizeof(struct release_seq));
2143 sequence->acquire = acquire;
2144 sequence->read = read;
2146 if (!release_seq_heads(rf, release_heads, sequence)) {
2147 /* add act to 'lazy checking' list */
2148 pending_rel_seqs.push_back(sequence);
2150 snapshot_free(sequence);
2155 * @brief Propagate a modified clock vector to actions later in the execution
2158 * After an acquire operation lazily completes a release-sequence
2159 * synchronization, we must update all clock vectors for operations later than
2160 * the acquire in the execution order.
2162 * @param acquire The ModelAction whose clock vector must be propagated
2163 * @param work The work queue to which we can add work items, if this
2164 * propagation triggers more updates (e.g., to the modification order)
2166 void ModelExecution::propagate_clockvector(ModelAction *acquire, work_queue_t *work)
2168 /* Re-check all pending release sequences */
2169 work->push_back(CheckRelSeqWorkEntry(NULL));
2170 /* Re-check read-acquire for mo_graph edges */
2171 work->push_back(MOEdgeWorkEntry(acquire));
2173 /* propagate synchronization to later actions */
2174 action_list_t::reverse_iterator rit = action_trace.rbegin();
2175 for (; (*rit) != acquire; rit++) {
2176 ModelAction *propagate = *rit;
2177 if (acquire->happens_before(propagate)) {
2178 synchronize(acquire, propagate);
2179 /* Re-check 'propagate' for mo_graph edges */
2180 work->push_back(MOEdgeWorkEntry(propagate));
2186 * Attempt to resolve all stashed operations that might synchronize with a
2187 * release sequence for a given location. This implements the "lazy" portion of
2188 * determining whether or not a release sequence was contiguous, since not all
2189 * modification order information is present at the time an action occurs.
2191 * @param location The location/object that should be checked for release
2192 * sequence resolutions. A NULL value means to check all locations.
2193 * @param work_queue The work queue to which to add work items as they are
2195 * @return True if any updates occurred (new synchronization, new mo_graph
2198 bool ModelExecution::resolve_release_sequences(void *location, work_queue_t *work_queue)
2200 bool updated = false;
2201 SnapVector<struct release_seq *>::iterator it = pending_rel_seqs.begin();
2202 while (it != pending_rel_seqs.end()) {
2203 struct release_seq *pending = *it;
2204 ModelAction *acquire = pending->acquire;
2205 const ModelAction *read = pending->read;
2207 /* Only resolve sequences on the given location, if provided */
2208 if (location && read->get_location() != location) {
2213 const ModelAction *rf = read->get_reads_from();
2214 rel_heads_list_t release_heads;
2216 complete = release_seq_heads(rf, &release_heads, pending);
2217 for (unsigned int i = 0; i < release_heads.size(); i++)
2218 if (!acquire->has_synchronized_with(release_heads[i]))
2219 if (synchronize(release_heads[i], acquire))
2223 /* Propagate the changed clock vector */
2224 propagate_clockvector(acquire, work_queue);
2227 it = pending_rel_seqs.erase(it);
2228 snapshot_free(pending);
2234 // If we resolved promises or data races, see if we have realized a data race.
2241 * Performs various bookkeeping operations for the current ModelAction. For
2242 * instance, adds action to the per-object, per-thread action vector and to the
2243 * action trace list of all thread actions.
2245 * @param act is the ModelAction to add.
2247 void ModelExecution::add_action_to_lists(ModelAction *act)
2249 int tid = id_to_int(act->get_tid());
2250 ModelAction *uninit = NULL;
2252 action_list_t *list = get_safe_ptr_action(&obj_map, act->get_location());
2253 if (list->empty() && act->is_atomic_var()) {
2254 uninit = get_uninitialized_action(act);
2255 uninit_id = id_to_int(uninit->get_tid());
2256 list->push_front(uninit);
2258 list->push_back(act);
2260 action_trace.push_back(act);
2262 action_trace.push_front(uninit);
2264 SnapVector<action_list_t> *vec = get_safe_ptr_vect_action(&obj_thrd_map, act->get_location());
2265 if (tid >= (int)vec->size())
2266 vec->resize(priv->next_thread_id);
2267 (*vec)[tid].push_back(act);
2269 (*vec)[uninit_id].push_front(uninit);
2271 if ((int)thrd_last_action.size() <= tid)
2272 thrd_last_action.resize(get_num_threads());
2273 thrd_last_action[tid] = act;
2275 thrd_last_action[uninit_id] = uninit;
2277 if (act->is_fence() && act->is_release()) {
2278 if ((int)thrd_last_fence_release.size() <= tid)
2279 thrd_last_fence_release.resize(get_num_threads());
2280 thrd_last_fence_release[tid] = act;
2283 if (act->is_wait()) {
2284 void *mutex_loc = (void *) act->get_value();
2285 get_safe_ptr_action(&obj_map, mutex_loc)->push_back(act);
2287 SnapVector<action_list_t> *vec = get_safe_ptr_vect_action(&obj_thrd_map, mutex_loc);
2288 if (tid >= (int)vec->size())
2289 vec->resize(priv->next_thread_id);
2290 (*vec)[tid].push_back(act);
2295 * @brief Get the last action performed by a particular Thread
2296 * @param tid The thread ID of the Thread in question
2297 * @return The last action in the thread
2299 ModelAction * ModelExecution::get_last_action(thread_id_t tid) const
2301 int threadid = id_to_int(tid);
2302 if (threadid < (int)thrd_last_action.size())
2303 return thrd_last_action[id_to_int(tid)];
2309 * @brief Get the last fence release performed by a particular Thread
2310 * @param tid The thread ID of the Thread in question
2311 * @return The last fence release in the thread, if one exists; NULL otherwise
2313 ModelAction * ModelExecution::get_last_fence_release(thread_id_t tid) const
2315 int threadid = id_to_int(tid);
2316 if (threadid < (int)thrd_last_fence_release.size())
2317 return thrd_last_fence_release[id_to_int(tid)];
2323 * Gets the last memory_order_seq_cst write (in the total global sequence)
2324 * performed on a particular object (i.e., memory location), not including the
2326 * @param curr The current ModelAction; also denotes the object location to
2328 * @return The last seq_cst write
2330 ModelAction * ModelExecution::get_last_seq_cst_write(ModelAction *curr) const
2332 void *location = curr->get_location();
2333 action_list_t *list = obj_map.get(location);
2334 /* Find: max({i in dom(S) | seq_cst(t_i) && isWrite(t_i) && samevar(t_i, t)}) */
2335 action_list_t::reverse_iterator rit;
2336 for (rit = list->rbegin(); (*rit) != curr; rit++)
2338 rit++; /* Skip past curr */
2339 for ( ; rit != list->rend(); rit++)
2340 if ((*rit)->is_write() && (*rit)->is_seqcst())
2346 * Gets the last memory_order_seq_cst fence (in the total global sequence)
2347 * performed in a particular thread, prior to a particular fence.
2348 * @param tid The ID of the thread to check
2349 * @param before_fence The fence from which to begin the search; if NULL, then
2350 * search for the most recent fence in the thread.
2351 * @return The last prior seq_cst fence in the thread, if exists; otherwise, NULL
2353 ModelAction * ModelExecution::get_last_seq_cst_fence(thread_id_t tid, const ModelAction *before_fence) const
2355 /* All fences should have location FENCE_LOCATION */
2356 action_list_t *list = obj_map.get(FENCE_LOCATION);
2361 action_list_t::reverse_iterator rit = list->rbegin();
2364 for (; rit != list->rend(); rit++)
2365 if (*rit == before_fence)
2368 ASSERT(*rit == before_fence);
2372 for (; rit != list->rend(); rit++)
2373 if ((*rit)->is_fence() && (tid == (*rit)->get_tid()) && (*rit)->is_seqcst())
2379 * Gets the last unlock operation performed on a particular mutex (i.e., memory
2380 * location). This function identifies the mutex according to the current
2381 * action, which is presumed to perform on the same mutex.
2382 * @param curr The current ModelAction; also denotes the object location to
2384 * @return The last unlock operation
2386 ModelAction * ModelExecution::get_last_unlock(ModelAction *curr) const
2388 void *location = curr->get_location();
2390 action_list_t *list = obj_map.get(location);
2391 /* Find: max({i in dom(S) | isUnlock(t_i) && samevar(t_i, t)}) */
2392 action_list_t::reverse_iterator rit;
2393 for (rit = list->rbegin(); rit != list->rend(); rit++)
2394 if ((*rit)->is_unlock() || (*rit)->is_wait())
2399 ModelAction * ModelExecution::get_parent_action(thread_id_t tid) const
2401 ModelAction *parent = get_last_action(tid);
2403 parent = get_thread(tid)->get_creation();
2408 * Returns the clock vector for a given thread.
2409 * @param tid The thread whose clock vector we want
2410 * @return Desired clock vector
2412 ClockVector * ModelExecution::get_cv(thread_id_t tid) const
2414 return get_parent_action(tid)->get_cv();
2418 * @brief Find the promise (if any) to resolve for the current action and
2419 * remove it from the pending promise vector
2420 * @param curr The current ModelAction. Should be a write.
2421 * @return The Promise to resolve, if any; otherwise NULL
2423 Promise * ModelExecution::pop_promise_to_resolve(const ModelAction *curr)
2425 for (unsigned int i = 0; i < promises.size(); i++)
2426 if (curr->get_node()->get_promise(i)) {
2427 Promise *ret = promises[i];
2428 promises.erase(promises.begin() + i);
2435 * Resolve a Promise with a current write.
2436 * @param write The ModelAction that is fulfilling Promises
2437 * @param promise The Promise to resolve
2438 * @param work The work queue, for adding new fixup work
2439 * @return True if the Promise was successfully resolved; false otherwise
2441 bool ModelExecution::resolve_promise(ModelAction *write, Promise *promise,
2444 ModelVector<ModelAction *> actions_to_check;
2446 for (unsigned int i = 0; i < promise->get_num_readers(); i++) {
2447 ModelAction *read = promise->get_reader(i);
2448 if (read_from(read, write)) {
2449 /* Propagate the changed clock vector */
2450 propagate_clockvector(read, work);
2452 actions_to_check.push_back(read);
2454 /* Make sure the promise's value matches the write's value */
2455 ASSERT(promise->is_compatible(write) && promise->same_value(write));
2456 if (!mo_graph->resolvePromise(promise, write))
2457 priv->hard_failed_promise = true;
2460 * @todo It is possible to end up in an inconsistent state, where a
2461 * "resolved" promise may still be referenced if
2462 * CycleGraph::resolvePromise() failed, so don't delete 'promise'.
2464 * Note that the inconsistency only matters when dumping mo_graph to
2470 //Check whether reading these writes has made threads unable to
2472 for (unsigned int i = 0; i < actions_to_check.size(); i++) {
2473 ModelAction *read = actions_to_check[i];
2474 mo_check_promises(read, true);
2481 * Compute the set of promises that could potentially be satisfied by this
2482 * action. Note that the set computation actually appears in the Node, not in
2484 * @param curr The ModelAction that may satisfy promises
2486 void ModelExecution::compute_promises(ModelAction *curr)
2488 for (unsigned int i = 0; i < promises.size(); i++) {
2489 Promise *promise = promises[i];
2490 if (!promise->is_compatible(curr) || !promise->same_value(curr))
2493 bool satisfy = true;
2494 for (unsigned int j = 0; j < promise->get_num_readers(); j++) {
2495 const ModelAction *act = promise->get_reader(j);
2496 if (act->happens_before(curr) ||
2497 act->could_synchronize_with(curr)) {
2503 curr->get_node()->set_promise(i);
2507 /** Checks promises in response to change in ClockVector Threads. */
2508 void ModelExecution::check_promises(thread_id_t tid, ClockVector *old_cv, ClockVector *merge_cv)
2510 for (unsigned int i = 0; i < promises.size(); i++) {
2511 Promise *promise = promises[i];
2512 if (!promise->thread_is_available(tid))
2514 for (unsigned int j = 0; j < promise->get_num_readers(); j++) {
2515 const ModelAction *act = promise->get_reader(j);
2516 if ((!old_cv || !old_cv->synchronized_since(act)) &&
2517 merge_cv->synchronized_since(act)) {
2518 if (promise->eliminate_thread(tid)) {
2519 /* Promise has failed */
2520 priv->failed_promise = true;
2528 void ModelExecution::check_promises_thread_disabled()
2530 for (unsigned int i = 0; i < promises.size(); i++) {
2531 Promise *promise = promises[i];
2532 if (promise->has_failed()) {
2533 priv->failed_promise = true;
2540 * @brief Checks promises in response to addition to modification order for
2543 * We test whether threads are still available for satisfying promises after an
2544 * addition to our modification order constraints. Those that are unavailable
2545 * are "eliminated". Once all threads are eliminated from satisfying a promise,
2546 * that promise has failed.
2548 * @param act The ModelAction which updated the modification order
2549 * @param is_read_check Should be true if act is a read and we must check for
2550 * updates to the store from which it read (there is a distinction here for
2551 * RMW's, which are both a load and a store)
2553 void ModelExecution::mo_check_promises(const ModelAction *act, bool is_read_check)
2555 const ModelAction *write = is_read_check ? act->get_reads_from() : act;
2557 for (unsigned int i = 0; i < promises.size(); i++) {
2558 Promise *promise = promises[i];
2560 // Is this promise on the same location?
2561 if (!promise->same_location(write))
2564 for (unsigned int j = 0; j < promise->get_num_readers(); j++) {
2565 const ModelAction *pread = promise->get_reader(j);
2566 if (!pread->happens_before(act))
2568 if (mo_graph->checkPromise(write, promise)) {
2569 priv->hard_failed_promise = true;
2575 // Don't do any lookups twice for the same thread
2576 if (!promise->thread_is_available(act->get_tid()))
2579 if (mo_graph->checkReachable(promise, write)) {
2580 if (mo_graph->checkPromise(write, promise)) {
2581 priv->hard_failed_promise = true;
2589 * Compute the set of writes that may break the current pending release
2590 * sequence. This information is extracted from previou release sequence
2593 * @param curr The current ModelAction. Must be a release sequence fixup
2596 void ModelExecution::compute_relseq_breakwrites(ModelAction *curr)
2598 if (pending_rel_seqs.empty())
2601 struct release_seq *pending = pending_rel_seqs.back();
2602 for (unsigned int i = 0; i < pending->writes.size(); i++) {
2603 const ModelAction *write = pending->writes[i];
2604 curr->get_node()->add_relseq_break(write);
2607 /* NULL means don't break the sequence; just synchronize */
2608 curr->get_node()->add_relseq_break(NULL);
2612 * Build up an initial set of all past writes that this 'read' action may read
2613 * from, as well as any previously-observed future values that must still be valid.
2615 * @param curr is the current ModelAction that we are exploring; it must be a
2618 void ModelExecution::build_may_read_from(ModelAction *curr)
2620 SnapVector<action_list_t> *thrd_lists = obj_thrd_map.get(curr->get_location());
2622 ASSERT(curr->is_read());
2624 ModelAction *last_sc_write = NULL;
2626 if (curr->is_seqcst())
2627 last_sc_write = get_last_seq_cst_write(curr);
2629 /* Iterate over all threads */
2630 for (i = 0; i < thrd_lists->size(); i++) {
2631 /* Iterate over actions in thread, starting from most recent */
2632 action_list_t *list = &(*thrd_lists)[i];
2633 action_list_t::reverse_iterator rit;
2634 for (rit = list->rbegin(); rit != list->rend(); rit++) {
2635 ModelAction *act = *rit;
2637 /* Only consider 'write' actions */
2638 if (!act->is_write() || act == curr)
2641 /* Don't consider more than one seq_cst write if we are a seq_cst read. */
2642 bool allow_read = true;
2644 if (curr->is_seqcst() && (act->is_seqcst() || (last_sc_write != NULL && act->happens_before(last_sc_write))) && act != last_sc_write)
2646 else if (curr->get_sleep_flag() && !curr->is_seqcst() && !sleep_can_read_from(curr, act))
2650 /* Only add feasible reads */
2651 mo_graph->startChanges();
2652 r_modification_order(curr, act);
2653 if (!is_infeasible())
2654 curr->get_node()->add_read_from_past(act);
2655 mo_graph->rollbackChanges();
2658 /* Include at most one act per-thread that "happens before" curr */
2659 if (act->happens_before(curr))
2664 /* Inherit existing, promised future values */
2665 for (i = 0; i < promises.size(); i++) {
2666 const Promise *promise = promises[i];
2667 const ModelAction *promise_read = promise->get_reader(0);
2668 if (promise_read->same_var(curr)) {
2669 /* Only add feasible future-values */
2670 mo_graph->startChanges();
2671 r_modification_order(curr, promise);
2672 if (!is_infeasible())
2673 curr->get_node()->add_read_from_promise(promise_read);
2674 mo_graph->rollbackChanges();
2678 /* We may find no valid may-read-from only if the execution is doomed */
2679 if (!curr->get_node()->read_from_size()) {
2680 priv->no_valid_reads = true;
2684 if (DBG_ENABLED()) {
2685 model_print("Reached read action:\n");
2687 model_print("Printing read_from_past\n");
2688 curr->get_node()->print_read_from_past();
2689 model_print("End printing read_from_past\n");
2693 bool ModelExecution::sleep_can_read_from(ModelAction *curr, const ModelAction *write)
2695 for ( ; write != NULL; write = write->get_reads_from()) {
2696 /* UNINIT actions don't have a Node, and they never sleep */
2697 if (write->is_uninitialized())
2699 Node *prevnode = write->get_node()->get_parent();
2701 bool thread_sleep = prevnode->enabled_status(curr->get_tid()) == THREAD_SLEEP_SET;
2702 if (write->is_release() && thread_sleep)
2704 if (!write->is_rmw())
2711 * @brief Get an action representing an uninitialized atomic
2713 * This function may create a new one or try to retrieve one from the NodeStack
2715 * @param curr The current action, which prompts the creation of an UNINIT action
2716 * @return A pointer to the UNINIT ModelAction
2718 ModelAction * ModelExecution::get_uninitialized_action(const ModelAction *curr) const
2720 Node *node = curr->get_node();
2721 ModelAction *act = node->get_uninit_action();
2723 act = new ModelAction(ATOMIC_UNINIT, std::memory_order_relaxed, curr->get_location(), params->uninitvalue, model_thread);
2724 node->set_uninit_action(act);
2726 act->create_cv(NULL);
2730 static void print_list(const action_list_t *list)
2732 action_list_t::const_iterator it;
2734 model_print("------------------------------------------------------------------------------------\n");
2735 model_print("# t Action type MO Location Value Rf CV\n");
2736 model_print("------------------------------------------------------------------------------------\n");
2738 unsigned int hash = 0;
2740 for (it = list->begin(); it != list->end(); it++) {
2741 const ModelAction *act = *it;
2742 if (act->get_seq_number() > 0)
2744 hash = hash^(hash<<3)^((*it)->hash());
2746 model_print("HASH %u\n", hash);
2747 model_print("------------------------------------------------------------------------------------\n");
2750 #if SUPPORT_MOD_ORDER_DUMP
2751 void ModelExecution::dumpGraph(char *filename) const
2754 sprintf(buffer, "%s.dot", filename);
2755 FILE *file = fopen(buffer, "w");
2756 fprintf(file, "digraph %s {\n", filename);
2757 mo_graph->dumpNodes(file);
2758 ModelAction **thread_array = (ModelAction **)model_calloc(1, sizeof(ModelAction *) * get_num_threads());
2760 for (action_list_t::const_iterator it = action_trace.begin(); it != action_trace.end(); it++) {
2761 ModelAction *act = *it;
2762 if (act->is_read()) {
2763 mo_graph->dot_print_node(file, act);
2764 if (act->get_reads_from())
2765 mo_graph->dot_print_edge(file,
2766 act->get_reads_from(),
2768 "label=\"rf\", color=red, weight=2");
2770 mo_graph->dot_print_edge(file,
2771 act->get_reads_from_promise(),
2773 "label=\"rf\", color=red");
2775 if (thread_array[act->get_tid()]) {
2776 mo_graph->dot_print_edge(file,
2777 thread_array[id_to_int(act->get_tid())],
2779 "label=\"sb\", color=blue, weight=400");
2782 thread_array[act->get_tid()] = act;
2784 fprintf(file, "}\n");
2785 model_free(thread_array);
2790 /** @brief Prints an execution trace summary. */
2791 void ModelExecution::print_summary() const
2793 #if SUPPORT_MOD_ORDER_DUMP
2794 char buffername[100];
2795 sprintf(buffername, "exec%04u", get_execution_number());
2796 mo_graph->dumpGraphToFile(buffername);
2797 sprintf(buffername, "graph%04u", get_execution_number());
2798 dumpGraph(buffername);
2801 model_print("Execution trace %d:", get_execution_number());
2802 if (isfeasibleprefix()) {
2803 if (is_yieldblocked())
2804 model_print(" YIELD BLOCKED");
2805 if (scheduler->all_threads_sleeping())
2806 model_print(" SLEEP-SET REDUNDANT");
2807 if (have_bug_reports())
2808 model_print(" DETECTED BUG(S)");
2810 print_infeasibility(" INFEASIBLE");
2813 print_list(&action_trace);
2816 if (!promises.empty()) {
2817 model_print("Pending promises:\n");
2818 for (unsigned int i = 0; i < promises.size(); i++) {
2819 model_print(" [P%u] ", i);
2820 promises[i]->print();
2827 * Add a Thread to the system for the first time. Should only be called once
2829 * @param t The Thread to add
2831 void ModelExecution::add_thread(Thread *t)
2833 unsigned int i = id_to_int(t->get_id());
2834 if (i >= thread_map.size())
2835 thread_map.resize(i + 1);
2837 if (!t->is_model_thread())
2838 scheduler->add_thread(t);
2842 * @brief Get a Thread reference by its ID
2843 * @param tid The Thread's ID
2844 * @return A Thread reference
2846 Thread * ModelExecution::get_thread(thread_id_t tid) const
2848 unsigned int i = id_to_int(tid);
2849 if (i < thread_map.size())
2850 return thread_map[i];
2855 * @brief Get a reference to the Thread in which a ModelAction was executed
2856 * @param act The ModelAction
2857 * @return A Thread reference
2859 Thread * ModelExecution::get_thread(const ModelAction *act) const
2861 return get_thread(act->get_tid());
2865 * @brief Get a Thread reference by its pthread ID
2866 * @param index The pthread's ID
2867 * @return A Thread reference
2869 Thread * ModelExecution::get_pthread(pthread_t pid) {
2870 if (pid < pthread_counter + 1) return pthread_map[pid];
2875 * @brief Get a Promise's "promise number"
2877 * A "promise number" is an index number that is unique to a promise, valid
2878 * only for a specific snapshot of an execution trace. Promises may come and go
2879 * as they are generated an resolved, so an index only retains meaning for the
2882 * @param promise The Promise to check
2883 * @return The promise index, if the promise still is valid; otherwise -1
2885 int ModelExecution::get_promise_number(const Promise *promise) const
2887 for (unsigned int i = 0; i < promises.size(); i++)
2888 if (promises[i] == promise)
2895 * @brief Check if a Thread is currently enabled
2896 * @param t The Thread to check
2897 * @return True if the Thread is currently enabled
2899 bool ModelExecution::is_enabled(Thread *t) const
2901 return scheduler->is_enabled(t);
2905 * @brief Check if a Thread is currently enabled
2906 * @param tid The ID of the Thread to check
2907 * @return True if the Thread is currently enabled
2909 bool ModelExecution::is_enabled(thread_id_t tid) const
2911 return scheduler->is_enabled(tid);
2915 * @brief Select the next thread to execute based on the curren action
2917 * RMW actions occur in two parts, and we cannot split them. And THREAD_CREATE
2918 * actions should be followed by the execution of their child thread. In either
2919 * case, the current action should determine the next thread schedule.
2921 * @param curr The current action
2922 * @return The next thread to run, if the current action will determine this
2923 * selection; otherwise NULL
2925 Thread * ModelExecution::action_select_next_thread(const ModelAction *curr) const
2927 /* Do not split atomic RMW */
2928 if (curr->is_rmwr())
2929 return get_thread(curr);
2930 if (curr->is_write()) {
2931 // std::memory_order order = curr->get_mo();
2933 // case std::memory_order_relaxed:
2934 // return get_thread(curr);
2935 // case std::memory_order_release:
2936 // return get_thread(curr);
2943 /* Follow CREATE with the created thread */
2944 /* which is not needed, because model.cc takes care of this */
2945 if (curr->get_type() == THREAD_CREATE)
2946 return curr->get_thread_operand();
2947 if (curr->get_type() == PTHREAD_CREATE) {
2948 return curr->get_thread_operand();
2953 /** @return True if the execution has taken too many steps */
2954 bool ModelExecution::too_many_steps() const
2956 return params->bound != 0 && priv->used_sequence_numbers > params->bound;
2960 * Takes the next step in the execution, if possible.
2961 * @param curr The current step to take
2962 * @return Returns the next Thread to run, if any; NULL if this execution
2965 Thread * ModelExecution::take_step(ModelAction *curr)
2967 Thread *curr_thrd = get_thread(curr);
2968 ASSERT(curr_thrd->get_state() == THREAD_READY);
2970 ASSERT(check_action_enabled(curr)); /* May have side effects? */
2971 curr = check_current_action(curr);
2974 if (curr_thrd->is_blocked() || curr_thrd->is_complete())
2975 scheduler->remove_thread(curr_thrd);
2977 return action_select_next_thread(curr);
2981 * Launch end-of-execution release sequence fixups only when
2982 * the execution is otherwise feasible AND there are:
2984 * (1) pending release sequences
2985 * (2) pending assertions that could be invalidated by a change
2986 * in clock vectors (i.e., data races)
2987 * (3) no pending promises
2989 void ModelExecution::fixup_release_sequences()
2991 while (!pending_rel_seqs.empty() &&
2992 is_feasible_prefix_ignore_relseq() &&
2993 haveUnrealizedRaces()) {
2994 model_print("*** WARNING: release sequence fixup action "
2995 "(%zu pending release seuqence(s)) ***\n",
2996 pending_rel_seqs.size());
2997 ModelAction *fixup = new ModelAction(MODEL_FIXUP_RELSEQ,
2998 std::memory_order_seq_cst, NULL, VALUE_NONE,