9 #include "snapshot-interface.h"
11 #include "clockvector.h"
12 #include "cyclegraph.h"
15 #include "threads-model.h"
18 #define INITIAL_THREAD_ID 0
23 bug_message(const char *str) {
24 const char *fmt = " [BUG] %s\n";
25 msg = (char *)snapshot_malloc(strlen(fmt) + strlen(str));
26 sprintf(msg, fmt, str);
28 ~bug_message() { if (msg) snapshot_free(msg); }
31 void print() { model_print("%s", msg); }
37 * Structure for holding small ModelChecker members that should be snapshotted
39 struct model_snapshot_members {
40 model_snapshot_members() :
42 /* First thread created will have id INITIAL_THREAD_ID */
43 next_thread_id(INITIAL_THREAD_ID),
44 used_sequence_numbers(0),
49 failed_promise(false),
50 too_many_reads(false),
51 bad_synchronization(false),
55 ~model_snapshot_members() {
56 for (unsigned int i = 0; i < bugs.size(); i++)
61 ModelAction *current_action;
62 unsigned int next_thread_id;
63 modelclock_t used_sequence_numbers;
65 ModelAction *next_backtrack;
66 std::vector< bug_message *, SnapshotAlloc<bug_message *> > bugs;
67 struct execution_stats stats;
70 /** @brief Incorrectly-ordered synchronization was made */
71 bool bad_synchronization;
77 /** @brief Constructor */
78 ModelChecker::ModelChecker(struct model_params params) :
79 /* Initialize default scheduler */
81 scheduler(new Scheduler()),
83 earliest_diverge(NULL),
84 action_trace(new action_list_t()),
85 thread_map(new HashTable<int, Thread *, int>()),
86 obj_map(new HashTable<const void *, action_list_t *, uintptr_t, 4>()),
87 lock_waiters_map(new HashTable<const void *, action_list_t *, uintptr_t, 4>()),
88 condvar_waiters_map(new HashTable<const void *, action_list_t *, uintptr_t, 4>()),
89 obj_thrd_map(new HashTable<void *, std::vector<action_list_t> *, uintptr_t, 4 >()),
90 promises(new std::vector< Promise *, SnapshotAlloc<Promise *> >()),
91 futurevalues(new std::vector< struct PendingFutureValue, SnapshotAlloc<struct PendingFutureValue> >()),
92 pending_rel_seqs(new std::vector< struct release_seq *, SnapshotAlloc<struct release_seq *> >()),
93 thrd_last_action(new std::vector< ModelAction *, SnapshotAlloc<ModelAction *> >(1)),
94 node_stack(new NodeStack()),
95 priv(new struct model_snapshot_members()),
96 mo_graph(new CycleGraph())
98 /* Initialize a model-checker thread, for special ModelActions */
99 model_thread = new Thread(get_next_id());
100 thread_map->put(id_to_int(model_thread->get_id()), model_thread);
103 /** @brief Destructor */
104 ModelChecker::~ModelChecker()
106 for (unsigned int i = 0; i < get_num_threads(); i++)
107 delete thread_map->get(i);
112 delete lock_waiters_map;
113 delete condvar_waiters_map;
116 for (unsigned int i = 0; i < promises->size(); i++)
117 delete (*promises)[i];
120 delete pending_rel_seqs;
122 delete thrd_last_action;
129 static action_list_t * get_safe_ptr_action(HashTable<const void *, action_list_t *, uintptr_t, 4> * hash, void * ptr) {
130 action_list_t * tmp=hash->get(ptr);
132 tmp=new action_list_t();
138 static std::vector<action_list_t> * get_safe_ptr_vect_action(HashTable<void *, std::vector<action_list_t> *, uintptr_t, 4> * hash, void * ptr) {
139 std::vector<action_list_t> * tmp=hash->get(ptr);
141 tmp=new std::vector<action_list_t>();
148 * Restores user program to initial state and resets all model-checker data
151 void ModelChecker::reset_to_initial_state()
153 DEBUG("+++ Resetting to initial state +++\n");
154 node_stack->reset_execution();
156 /* Print all model-checker output before rollback */
159 snapshotObject->backTrackBeforeStep(0);
162 /** @return a thread ID for a new Thread */
163 thread_id_t ModelChecker::get_next_id()
165 return priv->next_thread_id++;
168 /** @return the number of user threads created during this execution */
169 unsigned int ModelChecker::get_num_threads() const
171 return priv->next_thread_id;
174 /** @return The currently executing Thread. */
175 Thread * ModelChecker::get_current_thread() const
177 return scheduler->get_current_thread();
180 /** @return a sequence number for a new ModelAction */
181 modelclock_t ModelChecker::get_next_seq_num()
183 return ++priv->used_sequence_numbers;
186 Node * ModelChecker::get_curr_node() const
188 return node_stack->get_head();
192 * @brief Choose the next thread to execute.
194 * This function chooses the next thread that should execute. It can force the
195 * adjacency of read/write portions of a RMW action, force THREAD_CREATE to be
196 * followed by a THREAD_START, or it can enforce execution replay/backtracking.
197 * The model-checker may have no preference regarding the next thread (i.e.,
198 * when exploring a new execution ordering), in which case this will return
200 * @param curr The current ModelAction. This action might guide the choice of
202 * @return The next thread to run. If the model-checker has no preference, NULL.
204 Thread * ModelChecker::get_next_thread(ModelAction *curr)
209 /* Do not split atomic actions. */
211 return thread_current();
212 /* The THREAD_CREATE action points to the created Thread */
213 else if (curr->get_type() == THREAD_CREATE)
214 return (Thread *)curr->get_location();
217 /* Have we completed exploring the preselected path? */
221 /* Else, we are trying to replay an execution */
222 ModelAction *next = node_stack->get_next()->get_action();
224 if (next == diverge) {
225 if (earliest_diverge == NULL || *diverge < *earliest_diverge)
226 earliest_diverge=diverge;
228 Node *nextnode = next->get_node();
229 Node *prevnode = nextnode->get_parent();
230 scheduler->update_sleep_set(prevnode);
232 /* Reached divergence point */
233 if (nextnode->increment_misc()) {
234 /* The next node will try to satisfy a different misc_index values. */
235 tid = next->get_tid();
236 node_stack->pop_restofstack(2);
237 } else if (nextnode->increment_promise()) {
238 /* The next node will try to satisfy a different set of promises. */
239 tid = next->get_tid();
240 node_stack->pop_restofstack(2);
241 } else if (nextnode->increment_read_from()) {
242 /* The next node will read from a different value. */
243 tid = next->get_tid();
244 node_stack->pop_restofstack(2);
245 } else if (nextnode->increment_future_value()) {
246 /* The next node will try to read from a different future value. */
247 tid = next->get_tid();
248 node_stack->pop_restofstack(2);
249 } else if (nextnode->increment_relseq_break()) {
250 /* The next node will try to resolve a release sequence differently */
251 tid = next->get_tid();
252 node_stack->pop_restofstack(2);
254 /* Make a different thread execute for next step */
255 scheduler->add_sleep(thread_map->get(id_to_int(next->get_tid())));
256 tid = prevnode->get_next_backtrack();
257 /* Make sure the backtracked thread isn't sleeping. */
258 node_stack->pop_restofstack(1);
259 if (diverge==earliest_diverge) {
260 earliest_diverge=prevnode->get_action();
263 /* The correct sleep set is in the parent node. */
266 DEBUG("*** Divergence point ***\n");
270 tid = next->get_tid();
272 DEBUG("*** ModelChecker chose next thread = %d ***\n", id_to_int(tid));
273 ASSERT(tid != THREAD_ID_T_NONE);
274 return thread_map->get(id_to_int(tid));
278 * We need to know what the next actions of all threads in the sleep
279 * set will be. This method computes them and stores the actions at
280 * the corresponding thread object's pending action.
283 void ModelChecker::execute_sleep_set() {
284 for(unsigned int i=0;i<get_num_threads();i++) {
285 thread_id_t tid=int_to_id(i);
286 Thread *thr=get_thread(tid);
287 if ( scheduler->get_enabled(thr) == THREAD_SLEEP_SET &&
288 thr->get_pending() == NULL ) {
289 thr->set_state(THREAD_RUNNING);
290 scheduler->next_thread(thr);
291 Thread::swap(&system_context, thr);
292 priv->current_action->set_sleep_flag();
293 thr->set_pending(priv->current_action);
296 priv->current_action = NULL;
299 void ModelChecker::wake_up_sleeping_actions(ModelAction * curr) {
300 for(unsigned int i=0;i<get_num_threads();i++) {
301 thread_id_t tid=int_to_id(i);
302 Thread *thr=get_thread(tid);
303 if ( scheduler->get_enabled(thr) == THREAD_SLEEP_SET ) {
304 ModelAction *pending_act=thr->get_pending();
305 if ((!curr->is_rmwr())&&pending_act->could_synchronize_with(curr)) {
306 //Remove this thread from sleep set
307 scheduler->remove_sleep(thr);
313 /** @brief Alert the model-checker that an incorrectly-ordered
314 * synchronization was made */
315 void ModelChecker::set_bad_synchronization()
317 priv->bad_synchronization = true;
320 bool ModelChecker::has_asserted() const
322 return priv->asserted;
325 void ModelChecker::set_assert()
327 priv->asserted = true;
331 * Check if we are in a deadlock. Should only be called at the end of an
332 * execution, although it should not give false positives in the middle of an
333 * execution (there should be some ENABLED thread).
335 * @return True if program is in a deadlock; false otherwise
337 bool ModelChecker::is_deadlocked() const
339 bool blocking_threads = false;
340 for (unsigned int i = 0; i < get_num_threads(); i++) {
341 thread_id_t tid = int_to_id(i);
344 Thread *t = get_thread(tid);
345 if (!t->is_model_thread() && t->get_pending())
346 blocking_threads = true;
348 return blocking_threads;
352 * Check if this is a complete execution. That is, have all thread completed
353 * execution (rather than exiting because sleep sets have forced a redundant
356 * @return True if the execution is complete.
358 bool ModelChecker::is_complete_execution() const
360 for (unsigned int i = 0; i < get_num_threads(); i++)
361 if (is_enabled(int_to_id(i)))
367 * @brief Assert a bug in the executing program.
369 * Use this function to assert any sort of bug in the user program. If the
370 * current trace is feasible (actually, a prefix of some feasible execution),
371 * then this execution will be aborted, printing the appropriate message. If
372 * the current trace is not yet feasible, the error message will be stashed and
373 * printed if the execution ever becomes feasible.
375 * @param msg Descriptive message for the bug (do not include newline char)
376 * @return True if bug is immediately-feasible
378 bool ModelChecker::assert_bug(const char *msg)
380 priv->bugs.push_back(new bug_message(msg));
382 if (isfeasibleprefix()) {
390 * @brief Assert a bug in the executing program, asserted by a user thread
391 * @see ModelChecker::assert_bug
392 * @param msg Descriptive message for the bug (do not include newline char)
394 void ModelChecker::assert_user_bug(const char *msg)
396 /* If feasible bug, bail out now */
398 switch_to_master(NULL);
401 /** @return True, if any bugs have been reported for this execution */
402 bool ModelChecker::have_bug_reports() const
404 return priv->bugs.size() != 0;
407 /** @brief Print bug report listing for this execution (if any bugs exist) */
408 void ModelChecker::print_bugs() const
410 if (have_bug_reports()) {
411 model_print("Bug report: %zu bug%s detected\n",
413 priv->bugs.size() > 1 ? "s" : "");
414 for (unsigned int i = 0; i < priv->bugs.size(); i++)
415 priv->bugs[i]->print();
420 * @brief Record end-of-execution stats
422 * Must be run when exiting an execution. Records various stats.
423 * @see struct execution_stats
425 void ModelChecker::record_stats()
428 if (!isfeasibleprefix())
429 stats.num_infeasible++;
430 else if (have_bug_reports())
431 stats.num_buggy_executions++;
432 else if (is_complete_execution())
433 stats.num_complete++;
435 stats.num_redundant++;
438 /** @brief Print execution stats */
439 void ModelChecker::print_stats() const
441 model_print("Number of complete, bug-free executions: %d\n", stats.num_complete);
442 model_print("Number of redundant executions: %d\n", stats.num_redundant);
443 model_print("Number of buggy executions: %d\n", stats.num_buggy_executions);
444 model_print("Number of infeasible executions: %d\n", stats.num_infeasible);
445 model_print("Total executions: %d\n", stats.num_total);
446 model_print("Total nodes created: %d\n", node_stack->get_total_nodes());
450 * @brief End-of-exeuction print
451 * @param printbugs Should any existing bugs be printed?
453 void ModelChecker::print_execution(bool printbugs) const
455 print_program_output();
457 if (DBG_ENABLED() || params.verbose) {
458 model_print("Earliest divergence point since last feasible execution:\n");
459 if (earliest_diverge)
460 earliest_diverge->print();
462 model_print("(Not set)\n");
468 /* Don't print invalid bugs */
477 * Queries the model-checker for more executions to explore and, if one
478 * exists, resets the model-checker state to execute a new execution.
480 * @return If there are more executions to explore, return true. Otherwise,
483 bool ModelChecker::next_execution()
486 /* Is this execution a feasible execution that's worth bug-checking? */
487 bool complete = isfeasibleprefix() && (is_complete_execution() ||
490 /* End-of-execution bug checks */
493 assert_bug("Deadlock detected");
501 if (DBG_ENABLED() || params.verbose || have_bug_reports())
502 print_execution(complete);
504 clear_program_output();
507 earliest_diverge = NULL;
509 if ((diverge = get_next_backtrack()) == NULL)
513 model_print("Next execution will diverge at:\n");
517 reset_to_initial_state();
521 ModelAction * ModelChecker::get_last_conflict(ModelAction *act)
523 switch (act->get_type()) {
528 /* Optimization: relaxed operations don't need backtracking */
529 if (act->is_relaxed())
531 /* linear search: from most recent to oldest */
532 action_list_t *list = get_safe_ptr_action(obj_map, act->get_location());
533 action_list_t::reverse_iterator rit;
534 for (rit = list->rbegin(); rit != list->rend(); rit++) {
535 ModelAction *prev = *rit;
536 if (prev->could_synchronize_with(act))
542 case ATOMIC_TRYLOCK: {
543 /* linear search: from most recent to oldest */
544 action_list_t *list = get_safe_ptr_action(obj_map, act->get_location());
545 action_list_t::reverse_iterator rit;
546 for (rit = list->rbegin(); rit != list->rend(); rit++) {
547 ModelAction *prev = *rit;
548 if (act->is_conflicting_lock(prev))
553 case ATOMIC_UNLOCK: {
554 /* linear search: from most recent to oldest */
555 action_list_t *list = get_safe_ptr_action(obj_map, act->get_location());
556 action_list_t::reverse_iterator rit;
557 for (rit = list->rbegin(); rit != list->rend(); rit++) {
558 ModelAction *prev = *rit;
559 if (!act->same_thread(prev)&&prev->is_failed_trylock())
565 /* linear search: from most recent to oldest */
566 action_list_t *list = get_safe_ptr_action(obj_map, act->get_location());
567 action_list_t::reverse_iterator rit;
568 for (rit = list->rbegin(); rit != list->rend(); rit++) {
569 ModelAction *prev = *rit;
570 if (!act->same_thread(prev)&&prev->is_failed_trylock())
572 if (!act->same_thread(prev)&&prev->is_notify())
578 case ATOMIC_NOTIFY_ALL:
579 case ATOMIC_NOTIFY_ONE: {
580 /* linear search: from most recent to oldest */
581 action_list_t *list = get_safe_ptr_action(obj_map, act->get_location());
582 action_list_t::reverse_iterator rit;
583 for (rit = list->rbegin(); rit != list->rend(); rit++) {
584 ModelAction *prev = *rit;
585 if (!act->same_thread(prev)&&prev->is_wait())
596 /** This method finds backtracking points where we should try to
597 * reorder the parameter ModelAction against.
599 * @param the ModelAction to find backtracking points for.
601 void ModelChecker::set_backtracking(ModelAction *act)
603 Thread *t = get_thread(act);
604 ModelAction * prev = get_last_conflict(act);
608 Node * node = prev->get_node()->get_parent();
610 int low_tid, high_tid;
611 if (node->is_enabled(t)) {
612 low_tid = id_to_int(act->get_tid());
613 high_tid = low_tid+1;
616 high_tid = get_num_threads();
619 for(int i = low_tid; i < high_tid; i++) {
620 thread_id_t tid = int_to_id(i);
622 /* Make sure this thread can be enabled here. */
623 if (i >= node->get_num_threads())
626 /* Don't backtrack into a point where the thread is disabled or sleeping. */
627 if (node->enabled_status(tid)!=THREAD_ENABLED)
630 /* Check if this has been explored already */
631 if (node->has_been_explored(tid))
634 /* See if fairness allows */
635 if (model->params.fairwindow != 0 && !node->has_priority(tid)) {
637 for(int t=0;t<node->get_num_threads();t++) {
638 thread_id_t tother=int_to_id(t);
639 if (node->is_enabled(tother) && node->has_priority(tother)) {
647 /* Cache the latest backtracking point */
648 if (!priv->next_backtrack || *prev > *priv->next_backtrack)
649 priv->next_backtrack = prev;
651 /* If this is a new backtracking point, mark the tree */
652 if (!node->set_backtrack(tid))
654 DEBUG("Setting backtrack: conflict = %d, instead tid = %d\n",
655 id_to_int(prev->get_tid()),
656 id_to_int(t->get_id()));
665 * Returns last backtracking point. The model checker will explore a different
666 * path for this point in the next execution.
667 * @return The ModelAction at which the next execution should diverge.
669 ModelAction * ModelChecker::get_next_backtrack()
671 ModelAction *next = priv->next_backtrack;
672 priv->next_backtrack = NULL;
677 * Processes a read or rmw model action.
678 * @param curr is the read model action to process.
679 * @param second_part_of_rmw is boolean that is true is this is the second action of a rmw.
680 * @return True if processing this read updates the mo_graph.
682 bool ModelChecker::process_read(ModelAction *curr, bool second_part_of_rmw)
684 uint64_t value = VALUE_NONE;
685 bool updated = false;
687 const ModelAction *reads_from = curr->get_node()->get_read_from();
688 if (reads_from != NULL) {
689 mo_graph->startChanges();
691 value = reads_from->get_value();
692 bool r_status = false;
694 if (!second_part_of_rmw) {
695 check_recency(curr, reads_from);
696 r_status = r_modification_order(curr, reads_from);
700 if (!second_part_of_rmw&&is_infeasible()&&(curr->get_node()->increment_read_from()||curr->get_node()->increment_future_value())) {
701 mo_graph->rollbackChanges();
702 priv->too_many_reads = false;
706 curr->read_from(reads_from);
707 mo_graph->commitChanges();
708 mo_check_promises(curr->get_tid(), reads_from);
711 } else if (!second_part_of_rmw) {
712 /* Read from future value */
713 value = curr->get_node()->get_future_value();
714 modelclock_t expiration = curr->get_node()->get_future_value_expiration();
715 curr->read_from(NULL);
716 Promise *valuepromise = new Promise(curr, value, expiration);
717 promises->push_back(valuepromise);
719 get_thread(curr)->set_return_value(value);
725 * Processes a lock, trylock, or unlock model action. @param curr is
726 * the read model action to process.
728 * The try lock operation checks whether the lock is taken. If not,
729 * it falls to the normal lock operation case. If so, it returns
732 * The lock operation has already been checked that it is enabled, so
733 * it just grabs the lock and synchronizes with the previous unlock.
735 * The unlock operation has to re-enable all of the threads that are
736 * waiting on the lock.
738 * @return True if synchronization was updated; false otherwise
740 bool ModelChecker::process_mutex(ModelAction *curr) {
741 std::mutex *mutex=NULL;
742 struct std::mutex_state *state=NULL;
744 if (curr->is_trylock() || curr->is_lock() || curr->is_unlock()) {
745 mutex = (std::mutex *)curr->get_location();
746 state = mutex->get_state();
747 } else if(curr->is_wait()) {
748 mutex = (std::mutex *)curr->get_value();
749 state = mutex->get_state();
752 switch (curr->get_type()) {
753 case ATOMIC_TRYLOCK: {
754 bool success = !state->islocked;
755 curr->set_try_lock(success);
757 get_thread(curr)->set_return_value(0);
760 get_thread(curr)->set_return_value(1);
762 //otherwise fall into the lock case
764 if (curr->get_cv()->getClock(state->alloc_tid) <= state->alloc_clock)
765 assert_bug("Lock access before initialization");
766 state->islocked = true;
767 ModelAction *unlock = get_last_unlock(curr);
768 //synchronize with the previous unlock statement
769 if (unlock != NULL) {
770 curr->synchronize_with(unlock);
775 case ATOMIC_UNLOCK: {
777 state->islocked = false;
778 //wake up the other threads
779 action_list_t *waiters = get_safe_ptr_action(lock_waiters_map, curr->get_location());
780 //activate all the waiting threads
781 for (action_list_t::iterator rit = waiters->begin(); rit != waiters->end(); rit++) {
782 scheduler->wake(get_thread(*rit));
789 state->islocked = false;
790 //wake up the other threads
791 action_list_t *waiters = get_safe_ptr_action(lock_waiters_map, (void *) curr->get_value());
792 //activate all the waiting threads
793 for (action_list_t::iterator rit = waiters->begin(); rit != waiters->end(); rit++) {
794 scheduler->wake(get_thread(*rit));
797 //check whether we should go to sleep or not...simulate spurious failures
798 if (curr->get_node()->get_misc()==0) {
799 get_safe_ptr_action(condvar_waiters_map, curr->get_location())->push_back(curr);
801 scheduler->sleep(get_current_thread());
805 case ATOMIC_NOTIFY_ALL: {
806 action_list_t *waiters = get_safe_ptr_action(condvar_waiters_map, curr->get_location());
807 //activate all the waiting threads
808 for (action_list_t::iterator rit = waiters->begin(); rit != waiters->end(); rit++) {
809 scheduler->wake(get_thread(*rit));
814 case ATOMIC_NOTIFY_ONE: {
815 action_list_t *waiters = get_safe_ptr_action(condvar_waiters_map, curr->get_location());
816 int wakeupthread=curr->get_node()->get_misc();
817 action_list_t::iterator it = waiters->begin();
818 advance(it, wakeupthread);
819 scheduler->wake(get_thread(*it));
831 * Process a write ModelAction
832 * @param curr The ModelAction to process
833 * @return True if the mo_graph was updated or promises were resolved
835 bool ModelChecker::process_write(ModelAction *curr)
837 bool updated_mod_order = w_modification_order(curr);
838 bool updated_promises = resolve_promises(curr);
840 if (promises->size() == 0) {
841 for (unsigned int i = 0; i < futurevalues->size(); i++) {
842 struct PendingFutureValue pfv = (*futurevalues)[i];
843 //Do more ambitious checks now that mo is more complete
844 if (mo_may_allow(pfv.writer, pfv.act)&&
845 pfv.act->get_node()->add_future_value(pfv.writer->get_value(), pfv.writer->get_seq_number()+params.maxfuturedelay) &&
846 (!priv->next_backtrack || *pfv.act > *priv->next_backtrack))
847 priv->next_backtrack = pfv.act;
849 futurevalues->resize(0);
852 mo_graph->commitChanges();
853 mo_check_promises(curr->get_tid(), curr);
855 get_thread(curr)->set_return_value(VALUE_NONE);
856 return updated_mod_order || updated_promises;
860 * @brief Process the current action for thread-related activity
862 * Performs current-action processing for a THREAD_* ModelAction. Proccesses
863 * may include setting Thread status, completing THREAD_FINISH/THREAD_JOIN
864 * synchronization, etc. This function is a no-op for non-THREAD actions
865 * (e.g., ATOMIC_{READ,WRITE,RMW,LOCK}, etc.)
867 * @param curr The current action
868 * @return True if synchronization was updated or a thread completed
870 bool ModelChecker::process_thread_action(ModelAction *curr)
872 bool updated = false;
874 switch (curr->get_type()) {
875 case THREAD_CREATE: {
876 Thread *th = (Thread *)curr->get_location();
877 th->set_creation(curr);
881 Thread *blocking = (Thread *)curr->get_location();
882 ModelAction *act = get_last_action(blocking->get_id());
883 curr->synchronize_with(act);
884 updated = true; /* trigger rel-seq checks */
887 case THREAD_FINISH: {
888 Thread *th = get_thread(curr);
889 while (!th->wait_list_empty()) {
890 ModelAction *act = th->pop_wait_list();
891 scheduler->wake(get_thread(act));
894 updated = true; /* trigger rel-seq checks */
898 check_promises(curr->get_tid(), NULL, curr->get_cv());
909 * @brief Process the current action for release sequence fixup activity
911 * Performs model-checker release sequence fixups for the current action,
912 * forcing a single pending release sequence to break (with a given, potential
913 * "loose" write) or to complete (i.e., synchronize). If a pending release
914 * sequence forms a complete release sequence, then we must perform the fixup
915 * synchronization, mo_graph additions, etc.
917 * @param curr The current action; must be a release sequence fixup action
918 * @param work_queue The work queue to which to add work items as they are
921 void ModelChecker::process_relseq_fixup(ModelAction *curr, work_queue_t *work_queue)
923 const ModelAction *write = curr->get_node()->get_relseq_break();
924 struct release_seq *sequence = pending_rel_seqs->back();
925 pending_rel_seqs->pop_back();
927 ModelAction *acquire = sequence->acquire;
928 const ModelAction *rf = sequence->rf;
929 const ModelAction *release = sequence->release;
933 ASSERT(release->same_thread(rf));
937 * @todo Forcing a synchronization requires that we set
938 * modification order constraints. For instance, we can't allow
939 * a fixup sequence in which two separate read-acquire
940 * operations read from the same sequence, where the first one
941 * synchronizes and the other doesn't. Essentially, we can't
942 * allow any writes to insert themselves between 'release' and
946 /* Must synchronize */
947 if (!acquire->synchronize_with(release)) {
948 set_bad_synchronization();
951 /* Re-check all pending release sequences */
952 work_queue->push_back(CheckRelSeqWorkEntry(NULL));
953 /* Re-check act for mo_graph edges */
954 work_queue->push_back(MOEdgeWorkEntry(acquire));
956 /* propagate synchronization to later actions */
957 action_list_t::reverse_iterator rit = action_trace->rbegin();
958 for (; (*rit) != acquire; rit++) {
959 ModelAction *propagate = *rit;
960 if (acquire->happens_before(propagate)) {
961 propagate->synchronize_with(acquire);
962 /* Re-check 'propagate' for mo_graph edges */
963 work_queue->push_back(MOEdgeWorkEntry(propagate));
967 /* Break release sequence with new edges:
968 * release --mo--> write --mo--> rf */
969 mo_graph->addEdge(release, write);
970 mo_graph->addEdge(write, rf);
973 /* See if we have realized a data race */
978 * Initialize the current action by performing one or more of the following
979 * actions, as appropriate: merging RMWR and RMWC/RMW actions, stepping forward
980 * in the NodeStack, manipulating backtracking sets, allocating and
981 * initializing clock vectors, and computing the promises to fulfill.
983 * @param curr The current action, as passed from the user context; may be
984 * freed/invalidated after the execution of this function, with a different
985 * action "returned" its place (pass-by-reference)
986 * @return True if curr is a newly-explored action; false otherwise
988 bool ModelChecker::initialize_curr_action(ModelAction **curr)
990 ModelAction *newcurr;
992 if ((*curr)->is_rmwc() || (*curr)->is_rmw()) {
993 newcurr = process_rmw(*curr);
996 if (newcurr->is_rmw())
997 compute_promises(newcurr);
1003 (*curr)->set_seq_number(get_next_seq_num());
1005 newcurr = node_stack->explore_action(*curr, scheduler->get_enabled_array());
1007 /* First restore type and order in case of RMW operation */
1008 if ((*curr)->is_rmwr())
1009 newcurr->copy_typeandorder(*curr);
1011 ASSERT((*curr)->get_location() == newcurr->get_location());
1012 newcurr->copy_from_new(*curr);
1014 /* Discard duplicate ModelAction; use action from NodeStack */
1017 /* Always compute new clock vector */
1018 newcurr->create_cv(get_parent_action(newcurr->get_tid()));
1021 return false; /* Action was explored previously */
1025 /* Always compute new clock vector */
1026 newcurr->create_cv(get_parent_action(newcurr->get_tid()));
1028 * Perform one-time actions when pushing new ModelAction onto
1031 if (newcurr->is_write())
1032 compute_promises(newcurr);
1033 else if (newcurr->is_relseq_fixup())
1034 compute_relseq_breakwrites(newcurr);
1035 else if (newcurr->is_wait())
1036 newcurr->get_node()->set_misc_max(2);
1037 else if (newcurr->is_notify_one()) {
1038 newcurr->get_node()->set_misc_max(get_safe_ptr_action(condvar_waiters_map, newcurr->get_location())->size());
1040 return true; /* This was a new ModelAction */
1045 * @brief Check whether a model action is enabled.
1047 * Checks whether a lock or join operation would be successful (i.e., is the
1048 * lock already locked, or is the joined thread already complete). If not, put
1049 * the action in a waiter list.
1051 * @param curr is the ModelAction to check whether it is enabled.
1052 * @return a bool that indicates whether the action is enabled.
1054 bool ModelChecker::check_action_enabled(ModelAction *curr) {
1055 if (curr->is_lock()) {
1056 std::mutex * lock = (std::mutex *)curr->get_location();
1057 struct std::mutex_state * state = lock->get_state();
1058 if (state->islocked) {
1059 //Stick the action in the appropriate waiting queue
1060 get_safe_ptr_action(lock_waiters_map, curr->get_location())->push_back(curr);
1063 } else if (curr->get_type() == THREAD_JOIN) {
1064 Thread *blocking = (Thread *)curr->get_location();
1065 if (!blocking->is_complete()) {
1066 blocking->push_wait_list(curr);
1075 * Stores the ModelAction for the current thread action. Call this
1076 * immediately before switching from user- to system-context to pass
1077 * data between them.
1078 * @param act The ModelAction created by the user-thread action
1080 void ModelChecker::set_current_action(ModelAction *act) {
1081 priv->current_action = act;
1085 * This is the heart of the model checker routine. It performs model-checking
1086 * actions corresponding to a given "current action." Among other processes, it
1087 * calculates reads-from relationships, updates synchronization clock vectors,
1088 * forms a memory_order constraints graph, and handles replay/backtrack
1089 * execution when running permutations of previously-observed executions.
1091 * @param curr The current action to process
1092 * @return The next Thread that must be executed. May be NULL if ModelChecker
1093 * makes no choice (e.g., according to replay execution, combining RMW actions,
1096 Thread * ModelChecker::check_current_action(ModelAction *curr)
1099 bool second_part_of_rmw = curr->is_rmwc() || curr->is_rmw();
1101 if (!check_action_enabled(curr)) {
1102 /* Make the execution look like we chose to run this action
1103 * much later, when a lock/join can succeed */
1104 get_current_thread()->set_pending(curr);
1105 scheduler->sleep(get_current_thread());
1106 return get_next_thread(NULL);
1109 bool newly_explored = initialize_curr_action(&curr);
1111 wake_up_sleeping_actions(curr);
1113 /* Add the action to lists before any other model-checking tasks */
1114 if (!second_part_of_rmw)
1115 add_action_to_lists(curr);
1117 /* Build may_read_from set for newly-created actions */
1118 if (newly_explored && curr->is_read())
1119 build_reads_from_past(curr);
1121 /* Initialize work_queue with the "current action" work */
1122 work_queue_t work_queue(1, CheckCurrWorkEntry(curr));
1123 while (!work_queue.empty() && !has_asserted()) {
1124 WorkQueueEntry work = work_queue.front();
1125 work_queue.pop_front();
1127 switch (work.type) {
1128 case WORK_CHECK_CURR_ACTION: {
1129 ModelAction *act = work.action;
1130 bool update = false; /* update this location's release seq's */
1131 bool update_all = false; /* update all release seq's */
1133 if (process_thread_action(curr))
1136 if (act->is_read() && process_read(act, second_part_of_rmw))
1139 if (act->is_write() && process_write(act))
1142 if (act->is_mutex_op() && process_mutex(act))
1145 if (act->is_relseq_fixup())
1146 process_relseq_fixup(curr, &work_queue);
1149 work_queue.push_back(CheckRelSeqWorkEntry(NULL));
1151 work_queue.push_back(CheckRelSeqWorkEntry(act->get_location()));
1154 case WORK_CHECK_RELEASE_SEQ:
1155 resolve_release_sequences(work.location, &work_queue);
1157 case WORK_CHECK_MO_EDGES: {
1158 /** @todo Complete verification of work_queue */
1159 ModelAction *act = work.action;
1160 bool updated = false;
1162 if (act->is_read()) {
1163 const ModelAction *rf = act->get_reads_from();
1164 if (rf != NULL && r_modification_order(act, rf))
1167 if (act->is_write()) {
1168 if (w_modification_order(act))
1171 mo_graph->commitChanges();
1174 work_queue.push_back(CheckRelSeqWorkEntry(act->get_location()));
1183 check_curr_backtracking(curr);
1184 set_backtracking(curr);
1185 return get_next_thread(curr);
1188 void ModelChecker::check_curr_backtracking(ModelAction * curr) {
1189 Node *currnode = curr->get_node();
1190 Node *parnode = currnode->get_parent();
1192 if ((!parnode->backtrack_empty() ||
1193 !currnode->misc_empty() ||
1194 !currnode->read_from_empty() ||
1195 !currnode->future_value_empty() ||
1196 !currnode->promise_empty() ||
1197 !currnode->relseq_break_empty())
1198 && (!priv->next_backtrack ||
1199 *curr > *priv->next_backtrack)) {
1200 priv->next_backtrack = curr;
1204 bool ModelChecker::promises_expired() const
1206 for (unsigned int promise_index = 0; promise_index < promises->size(); promise_index++) {
1207 Promise *promise = (*promises)[promise_index];
1208 if (promise->get_expiration()<priv->used_sequence_numbers) {
1216 * This is the strongest feasibility check available.
1217 * @return whether the current trace (partial or complete) must be a prefix of
1220 bool ModelChecker::isfeasibleprefix() const
1222 return pending_rel_seqs->size() == 0 && is_feasible_prefix_ignore_relseq();
1226 * Returns whether the current completed trace is feasible, except for pending
1227 * release sequences.
1229 bool ModelChecker::is_feasible_prefix_ignore_relseq() const
1231 if (DBG_ENABLED() && promises->size() != 0)
1232 DEBUG("Infeasible: unrevolved promises\n");
1234 return !is_infeasible() && promises->size() == 0;
1238 * Check if the current partial trace is infeasible. Does not check any
1239 * end-of-execution flags, which might rule out the execution. Thus, this is
1240 * useful only for ruling an execution as infeasible.
1241 * @return whether the current partial trace is infeasible.
1243 bool ModelChecker::is_infeasible() const
1245 if (DBG_ENABLED() && mo_graph->checkForRMWViolation())
1246 DEBUG("Infeasible: RMW violation\n");
1248 return mo_graph->checkForRMWViolation() || is_infeasible_ignoreRMW();
1252 * Check If the current partial trace is infeasible, while ignoring
1253 * infeasibility related to 2 RMW's reading from the same store. It does not
1254 * check end-of-execution feasibility.
1255 * @see ModelChecker::is_infeasible
1256 * @return whether the current partial trace is infeasible, ignoring multiple
1257 * RMWs reading from the same store.
1259 bool ModelChecker::is_infeasible_ignoreRMW() const
1261 if (DBG_ENABLED()) {
1262 if (mo_graph->checkForCycles())
1263 DEBUG("Infeasible: modification order cycles\n");
1264 if (priv->failed_promise)
1265 DEBUG("Infeasible: failed promise\n");
1266 if (priv->too_many_reads)
1267 DEBUG("Infeasible: too many reads\n");
1268 if (priv->bad_synchronization)
1269 DEBUG("Infeasible: bad synchronization ordering\n");
1270 if (promises_expired())
1271 DEBUG("Infeasible: promises expired\n");
1273 return mo_graph->checkForCycles() || priv->failed_promise ||
1274 priv->too_many_reads || priv->bad_synchronization ||
1278 /** Close out a RMWR by converting previous RMWR into a RMW or READ. */
1279 ModelAction * ModelChecker::process_rmw(ModelAction *act) {
1280 ModelAction *lastread = get_last_action(act->get_tid());
1281 lastread->process_rmw(act);
1282 if (act->is_rmw() && lastread->get_reads_from()!=NULL) {
1283 mo_graph->addRMWEdge(lastread->get_reads_from(), lastread);
1284 mo_graph->commitChanges();
1290 * Checks whether a thread has read from the same write for too many times
1291 * without seeing the effects of a later write.
1294 * 1) there must a different write that we could read from that would satisfy the modification order,
1295 * 2) we must have read from the same value in excess of maxreads times, and
1296 * 3) that other write must have been in the reads_from set for maxreads times.
1298 * If so, we decide that the execution is no longer feasible.
1300 void ModelChecker::check_recency(ModelAction *curr, const ModelAction *rf) {
1301 if (params.maxreads != 0) {
1303 if (curr->get_node()->get_read_from_size() <= 1)
1305 //Must make sure that execution is currently feasible... We could
1306 //accidentally clear by rolling back
1307 if (is_infeasible())
1309 std::vector<action_list_t> *thrd_lists = get_safe_ptr_vect_action(obj_thrd_map, curr->get_location());
1310 int tid = id_to_int(curr->get_tid());
1313 if ((int)thrd_lists->size() <= tid)
1315 action_list_t *list = &(*thrd_lists)[tid];
1317 action_list_t::reverse_iterator rit = list->rbegin();
1318 /* Skip past curr */
1319 for (; (*rit) != curr; rit++)
1321 /* go past curr now */
1324 action_list_t::reverse_iterator ritcopy = rit;
1325 //See if we have enough reads from the same value
1327 for (; count < params.maxreads; rit++,count++) {
1328 if (rit==list->rend())
1330 ModelAction *act = *rit;
1331 if (!act->is_read())
1334 if (act->get_reads_from() != rf)
1336 if (act->get_node()->get_read_from_size() <= 1)
1339 for (int i = 0; i<curr->get_node()->get_read_from_size(); i++) {
1341 const ModelAction * write = curr->get_node()->get_read_from_at(i);
1343 //Need a different write
1347 /* Test to see whether this is a feasible write to read from*/
1348 mo_graph->startChanges();
1349 r_modification_order(curr, write);
1350 bool feasiblereadfrom = !is_infeasible();
1351 mo_graph->rollbackChanges();
1353 if (!feasiblereadfrom)
1357 bool feasiblewrite = true;
1358 //new we need to see if this write works for everyone
1360 for (int loop = count; loop>0; loop--,rit++) {
1361 ModelAction *act=*rit;
1362 bool foundvalue = false;
1363 for (int j = 0; j<act->get_node()->get_read_from_size(); j++) {
1364 if (act->get_node()->get_read_from_at(j)==write) {
1370 feasiblewrite = false;
1374 if (feasiblewrite) {
1375 priv->too_many_reads = true;
1383 * Updates the mo_graph with the constraints imposed from the current
1386 * Basic idea is the following: Go through each other thread and find
1387 * the lastest action that happened before our read. Two cases:
1389 * (1) The action is a write => that write must either occur before
1390 * the write we read from or be the write we read from.
1392 * (2) The action is a read => the write that that action read from
1393 * must occur before the write we read from or be the same write.
1395 * @param curr The current action. Must be a read.
1396 * @param rf The action that curr reads from. Must be a write.
1397 * @return True if modification order edges were added; false otherwise
1399 bool ModelChecker::r_modification_order(ModelAction *curr, const ModelAction *rf)
1401 std::vector<action_list_t> *thrd_lists = get_safe_ptr_vect_action(obj_thrd_map, curr->get_location());
1404 ASSERT(curr->is_read());
1406 /* Last SC fence in the current thread */
1407 ModelAction *last_sc_fence_local = get_last_seq_cst_fence(curr->get_tid(), NULL);
1409 /* Iterate over all threads */
1410 for (i = 0; i < thrd_lists->size(); i++) {
1411 /* Last SC fence in thread i */
1412 ModelAction *last_sc_fence_thread_local = NULL;
1413 if (int_to_id((int)i) != curr->get_tid())
1414 last_sc_fence_thread_local = get_last_seq_cst_fence(int_to_id(i), NULL);
1416 /* Last SC fence in thread i, before last SC fence in current thread */
1417 ModelAction *last_sc_fence_thread_before = NULL;
1418 if (last_sc_fence_local)
1419 last_sc_fence_thread_before = get_last_seq_cst_fence(int_to_id(i), last_sc_fence_local);
1421 /* Iterate over actions in thread, starting from most recent */
1422 action_list_t *list = &(*thrd_lists)[i];
1423 action_list_t::reverse_iterator rit;
1424 for (rit = list->rbegin(); rit != list->rend(); rit++) {
1425 ModelAction *act = *rit;
1427 if (act->is_write() && act != rf && act != curr) {
1428 /* C++, Section 29.3 statement 5 */
1429 if (curr->is_seqcst() && last_sc_fence_thread_local &&
1430 *act < *last_sc_fence_thread_local) {
1431 mo_graph->addEdge(act, rf);
1434 /* C++, Section 29.3 statement 4 */
1435 else if (act->is_seqcst() && last_sc_fence_local &&
1436 *act < *last_sc_fence_local) {
1437 mo_graph->addEdge(act, rf);
1440 /* C++, Section 29.3 statement 6 */
1441 else if (last_sc_fence_thread_before &&
1442 *act < *last_sc_fence_thread_before) {
1443 mo_graph->addEdge(act, rf);
1449 * Include at most one act per-thread that "happens
1450 * before" curr. Don't consider reflexively.
1452 if (act->happens_before(curr) && act != curr) {
1453 if (act->is_write()) {
1455 mo_graph->addEdge(act, rf);
1459 const ModelAction *prevreadfrom = act->get_reads_from();
1460 //if the previous read is unresolved, keep going...
1461 if (prevreadfrom == NULL)
1464 if (rf != prevreadfrom) {
1465 mo_graph->addEdge(prevreadfrom, rf);
1477 /** This method fixes up the modification order when we resolve a
1478 * promises. The basic problem is that actions that occur after the
1479 * read curr could not property add items to the modification order
1482 * So for each thread, we find the earliest item that happens after
1483 * the read curr. This is the item we have to fix up with additional
1484 * constraints. If that action is write, we add a MO edge between
1485 * the Action rf and that action. If the action is a read, we add a
1486 * MO edge between the Action rf, and whatever the read accessed.
1488 * @param curr is the read ModelAction that we are fixing up MO edges for.
1489 * @param rf is the write ModelAction that curr reads from.
1492 void ModelChecker::post_r_modification_order(ModelAction *curr, const ModelAction *rf)
1494 std::vector<action_list_t> *thrd_lists = get_safe_ptr_vect_action(obj_thrd_map, curr->get_location());
1496 ASSERT(curr->is_read());
1498 /* Iterate over all threads */
1499 for (i = 0; i < thrd_lists->size(); i++) {
1500 /* Iterate over actions in thread, starting from most recent */
1501 action_list_t *list = &(*thrd_lists)[i];
1502 action_list_t::reverse_iterator rit;
1503 ModelAction *lastact = NULL;
1505 /* Find last action that happens after curr that is either not curr or a rmw */
1506 for (rit = list->rbegin(); rit != list->rend(); rit++) {
1507 ModelAction *act = *rit;
1508 if (curr->happens_before(act) && (curr != act || curr->is_rmw())) {
1514 /* Include at most one act per-thread that "happens before" curr */
1515 if (lastact != NULL) {
1516 if (lastact==curr) {
1517 //Case 1: The resolved read is a RMW, and we need to make sure
1518 //that the write portion of the RMW mod order after rf
1520 mo_graph->addEdge(rf, lastact);
1521 } else if (lastact->is_read()) {
1522 //Case 2: The resolved read is a normal read and the next
1523 //operation is a read, and we need to make sure the value read
1524 //is mod ordered after rf
1526 const ModelAction *postreadfrom = lastact->get_reads_from();
1527 if (postreadfrom != NULL&&rf != postreadfrom)
1528 mo_graph->addEdge(rf, postreadfrom);
1530 //Case 3: The resolved read is a normal read and the next
1531 //operation is a write, and we need to make sure that the
1532 //write is mod ordered after rf
1534 mo_graph->addEdge(rf, lastact);
1542 * Updates the mo_graph with the constraints imposed from the current write.
1544 * Basic idea is the following: Go through each other thread and find
1545 * the lastest action that happened before our write. Two cases:
1547 * (1) The action is a write => that write must occur before
1550 * (2) The action is a read => the write that that action read from
1551 * must occur before the current write.
1553 * This method also handles two other issues:
1555 * (I) Sequential Consistency: Making sure that if the current write is
1556 * seq_cst, that it occurs after the previous seq_cst write.
1558 * (II) Sending the write back to non-synchronizing reads.
1560 * @param curr The current action. Must be a write.
1561 * @return True if modification order edges were added; false otherwise
1563 bool ModelChecker::w_modification_order(ModelAction *curr)
1565 std::vector<action_list_t> *thrd_lists = get_safe_ptr_vect_action(obj_thrd_map, curr->get_location());
1568 ASSERT(curr->is_write());
1570 if (curr->is_seqcst()) {
1571 /* We have to at least see the last sequentially consistent write,
1572 so we are initialized. */
1573 ModelAction *last_seq_cst = get_last_seq_cst_write(curr);
1574 if (last_seq_cst != NULL) {
1575 mo_graph->addEdge(last_seq_cst, curr);
1580 /* Last SC fence in the current thread */
1581 ModelAction *last_sc_fence_local = get_last_seq_cst_fence(curr->get_tid(), NULL);
1583 /* Iterate over all threads */
1584 for (i = 0; i < thrd_lists->size(); i++) {
1585 /* Last SC fence in thread i, before last SC fence in current thread */
1586 ModelAction *last_sc_fence_thread_before = NULL;
1587 if (last_sc_fence_local && int_to_id((int)i) != curr->get_tid())
1588 last_sc_fence_thread_before = get_last_seq_cst_fence(int_to_id(i), last_sc_fence_local);
1590 /* Iterate over actions in thread, starting from most recent */
1591 action_list_t *list = &(*thrd_lists)[i];
1592 action_list_t::reverse_iterator rit;
1593 for (rit = list->rbegin(); rit != list->rend(); rit++) {
1594 ModelAction *act = *rit;
1597 * 1) If RMW and it actually read from something, then we
1598 * already have all relevant edges, so just skip to next
1601 * 2) If RMW and it didn't read from anything, we should
1602 * whatever edge we can get to speed up convergence.
1604 * 3) If normal write, we need to look at earlier actions, so
1605 * continue processing list.
1607 if (curr->is_rmw()) {
1608 if (curr->get_reads_from()!=NULL)
1616 /* C++, Section 29.3 statement 7 */
1617 if (last_sc_fence_thread_before && act->is_write() &&
1618 *act < *last_sc_fence_thread_before) {
1619 mo_graph->addEdge(act, curr);
1624 * Include at most one act per-thread that "happens
1627 if (act->happens_before(curr)) {
1629 * Note: if act is RMW, just add edge:
1631 * The following edge should be handled elsewhere:
1632 * readfrom(act) --mo--> act
1634 if (act->is_write())
1635 mo_graph->addEdge(act, curr);
1636 else if (act->is_read()) {
1637 //if previous read accessed a null, just keep going
1638 if (act->get_reads_from() == NULL)
1640 mo_graph->addEdge(act->get_reads_from(), curr);
1644 } else if (act->is_read() && !act->could_synchronize_with(curr) &&
1645 !act->same_thread(curr)) {
1646 /* We have an action that:
1647 (1) did not happen before us
1648 (2) is a read and we are a write
1649 (3) cannot synchronize with us
1650 (4) is in a different thread
1652 that read could potentially read from our write. Note that
1653 these checks are overly conservative at this point, we'll
1654 do more checks before actually removing the
1658 if (thin_air_constraint_may_allow(curr, act)) {
1659 if (!is_infeasible() ||
1660 (curr->is_rmw() && act->is_rmw() && curr->get_reads_from() == act->get_reads_from() && !is_infeasible_ignoreRMW())) {
1661 struct PendingFutureValue pfv = {curr,act};
1662 futurevalues->push_back(pfv);
1672 /** Arbitrary reads from the future are not allowed. Section 29.3
1673 * part 9 places some constraints. This method checks one result of constraint
1674 * constraint. Others require compiler support. */
1675 bool ModelChecker::thin_air_constraint_may_allow(const ModelAction * writer, const ModelAction *reader) {
1676 if (!writer->is_rmw())
1679 if (!reader->is_rmw())
1682 for (const ModelAction *search = writer->get_reads_from(); search != NULL; search = search->get_reads_from()) {
1683 if (search == reader)
1685 if (search->get_tid() == reader->get_tid() &&
1686 search->happens_before(reader))
1694 * Arbitrary reads from the future are not allowed. Section 29.3 part 9 places
1695 * some constraints. This method checks one the following constraint (others
1696 * require compiler support):
1698 * If X --hb-> Y --mo-> Z, then X should not read from Z.
1700 bool ModelChecker::mo_may_allow(const ModelAction *writer, const ModelAction *reader)
1702 std::vector<action_list_t> *thrd_lists = get_safe_ptr_vect_action(obj_thrd_map, reader->get_location());
1704 /* Iterate over all threads */
1705 for (i = 0; i < thrd_lists->size(); i++) {
1706 const ModelAction *write_after_read = NULL;
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;
1714 if (!reader->happens_before(act))
1716 else if (act->is_write())
1717 write_after_read = act;
1718 else if (act->is_read() && act->get_reads_from() != NULL && act != reader) {
1719 write_after_read = act->get_reads_from();
1723 if (write_after_read && write_after_read!=writer && mo_graph->checkReachable(write_after_read, writer))
1730 * Finds the head(s) of the release sequence(s) containing a given ModelAction.
1731 * The ModelAction under consideration is expected to be taking part in
1732 * release/acquire synchronization as an object of the "reads from" relation.
1733 * Note that this can only provide release sequence support for RMW chains
1734 * which do not read from the future, as those actions cannot be traced until
1735 * their "promise" is fulfilled. Similarly, we may not even establish the
1736 * presence of a release sequence with certainty, as some modification order
1737 * constraints may be decided further in the future. Thus, this function
1738 * "returns" two pieces of data: a pass-by-reference vector of @a release_heads
1739 * and a boolean representing certainty.
1741 * @param rf The action that might be part of a release sequence. Must be a
1743 * @param release_heads A pass-by-reference style return parameter. After
1744 * execution of this function, release_heads will contain the heads of all the
1745 * relevant release sequences, if any exists with certainty
1746 * @param pending A pass-by-reference style return parameter which is only used
1747 * when returning false (i.e., uncertain). Returns most information regarding
1748 * an uncertain release sequence, including any write operations that might
1749 * break the sequence.
1750 * @return true, if the ModelChecker is certain that release_heads is complete;
1753 bool ModelChecker::release_seq_heads(const ModelAction *rf,
1754 rel_heads_list_t *release_heads,
1755 struct release_seq *pending) const
1757 /* Only check for release sequences if there are no cycles */
1758 if (mo_graph->checkForCycles())
1762 ASSERT(rf->is_write());
1764 if (rf->is_release())
1765 release_heads->push_back(rf);
1767 break; /* End of RMW chain */
1769 /** @todo Need to be smarter here... In the linux lock
1770 * example, this will run to the beginning of the program for
1772 /** @todo The way to be smarter here is to keep going until 1
1773 * thread has a release preceded by an acquire and you've seen
1776 /* acq_rel RMW is a sufficient stopping condition */
1777 if (rf->is_acquire() && rf->is_release())
1778 return true; /* complete */
1780 rf = rf->get_reads_from();
1783 /* read from future: need to settle this later */
1785 return false; /* incomplete */
1788 if (rf->is_release())
1789 return true; /* complete */
1791 /* else relaxed write; check modification order for contiguous subsequence
1792 * -> rf must be same thread as release */
1793 int tid = id_to_int(rf->get_tid());
1794 std::vector<action_list_t> *thrd_lists = get_safe_ptr_vect_action(obj_thrd_map, rf->get_location());
1795 action_list_t *list = &(*thrd_lists)[tid];
1796 action_list_t::const_reverse_iterator rit;
1798 /* Find rf in the thread list */
1799 rit = std::find(list->rbegin(), list->rend(), rf);
1800 ASSERT(rit != list->rend());
1802 /* Find the last write/release */
1803 for (; rit != list->rend(); rit++)
1804 if ((*rit)->is_release())
1806 if (rit == list->rend()) {
1807 /* No write-release in this thread */
1808 return true; /* complete */
1810 ModelAction *release = *rit;
1812 ASSERT(rf->same_thread(release));
1814 pending->writes.clear();
1816 bool certain = true;
1817 for (unsigned int i = 0; i < thrd_lists->size(); i++) {
1818 if (id_to_int(rf->get_tid()) == (int)i)
1820 list = &(*thrd_lists)[i];
1822 /* Can we ensure no future writes from this thread may break
1823 * the release seq? */
1824 bool future_ordered = false;
1826 ModelAction *last = get_last_action(int_to_id(i));
1827 Thread *th = get_thread(int_to_id(i));
1828 if ((last && rf->happens_before(last)) ||
1831 future_ordered = true;
1833 ASSERT(!th->is_model_thread() || future_ordered);
1835 for (rit = list->rbegin(); rit != list->rend(); rit++) {
1836 const ModelAction *act = *rit;
1837 /* Reach synchronization -> this thread is complete */
1838 if (act->happens_before(release))
1840 if (rf->happens_before(act)) {
1841 future_ordered = true;
1845 /* Only non-RMW writes can break release sequences */
1846 if (!act->is_write() || act->is_rmw())
1849 /* Check modification order */
1850 if (mo_graph->checkReachable(rf, act)) {
1851 /* rf --mo--> act */
1852 future_ordered = true;
1855 if (mo_graph->checkReachable(act, release))
1856 /* act --mo--> release */
1858 if (mo_graph->checkReachable(release, act) &&
1859 mo_graph->checkReachable(act, rf)) {
1860 /* release --mo-> act --mo--> rf */
1861 return true; /* complete */
1863 /* act may break release sequence */
1864 pending->writes.push_back(act);
1867 if (!future_ordered)
1868 certain = false; /* This thread is uncertain */
1872 release_heads->push_back(release);
1873 pending->writes.clear();
1875 pending->release = release;
1882 * A public interface for getting the release sequence head(s) with which a
1883 * given ModelAction must synchronize. This function only returns a non-empty
1884 * result when it can locate a release sequence head with certainty. Otherwise,
1885 * it may mark the internal state of the ModelChecker so that it will handle
1886 * the release sequence at a later time, causing @a act to update its
1887 * synchronization at some later point in execution.
1888 * @param act The 'acquire' action that may read from a release sequence
1889 * @param release_heads A pass-by-reference return parameter. Will be filled
1890 * with the head(s) of the release sequence(s), if they exists with certainty.
1891 * @see ModelChecker::release_seq_heads
1893 void ModelChecker::get_release_seq_heads(ModelAction *act, rel_heads_list_t *release_heads)
1895 const ModelAction *rf = act->get_reads_from();
1896 struct release_seq *sequence = (struct release_seq *)snapshot_calloc(1, sizeof(struct release_seq));
1897 sequence->acquire = act;
1899 if (!release_seq_heads(rf, release_heads, sequence)) {
1900 /* add act to 'lazy checking' list */
1901 pending_rel_seqs->push_back(sequence);
1903 snapshot_free(sequence);
1908 * Attempt to resolve all stashed operations that might synchronize with a
1909 * release sequence for a given location. This implements the "lazy" portion of
1910 * determining whether or not a release sequence was contiguous, since not all
1911 * modification order information is present at the time an action occurs.
1913 * @param location The location/object that should be checked for release
1914 * sequence resolutions. A NULL value means to check all locations.
1915 * @param work_queue The work queue to which to add work items as they are
1917 * @return True if any updates occurred (new synchronization, new mo_graph
1920 bool ModelChecker::resolve_release_sequences(void *location, work_queue_t *work_queue)
1922 bool updated = false;
1923 std::vector< struct release_seq *, SnapshotAlloc<struct release_seq *> >::iterator it = pending_rel_seqs->begin();
1924 while (it != pending_rel_seqs->end()) {
1925 struct release_seq *pending = *it;
1926 ModelAction *act = pending->acquire;
1928 /* Only resolve sequences on the given location, if provided */
1929 if (location && act->get_location() != location) {
1934 const ModelAction *rf = act->get_reads_from();
1935 rel_heads_list_t release_heads;
1937 complete = release_seq_heads(rf, &release_heads, pending);
1938 for (unsigned int i = 0; i < release_heads.size(); i++) {
1939 if (!act->has_synchronized_with(release_heads[i])) {
1940 if (act->synchronize_with(release_heads[i]))
1943 set_bad_synchronization();
1948 /* Re-check all pending release sequences */
1949 work_queue->push_back(CheckRelSeqWorkEntry(NULL));
1950 /* Re-check act for mo_graph edges */
1951 work_queue->push_back(MOEdgeWorkEntry(act));
1953 /* propagate synchronization to later actions */
1954 action_list_t::reverse_iterator rit = action_trace->rbegin();
1955 for (; (*rit) != act; rit++) {
1956 ModelAction *propagate = *rit;
1957 if (act->happens_before(propagate)) {
1958 propagate->synchronize_with(act);
1959 /* Re-check 'propagate' for mo_graph edges */
1960 work_queue->push_back(MOEdgeWorkEntry(propagate));
1965 it = pending_rel_seqs->erase(it);
1966 snapshot_free(pending);
1972 // If we resolved promises or data races, see if we have realized a data race.
1979 * Performs various bookkeeping operations for the current ModelAction. For
1980 * instance, adds action to the per-object, per-thread action vector and to the
1981 * action trace list of all thread actions.
1983 * @param act is the ModelAction to add.
1985 void ModelChecker::add_action_to_lists(ModelAction *act)
1987 int tid = id_to_int(act->get_tid());
1988 action_trace->push_back(act);
1990 get_safe_ptr_action(obj_map, act->get_location())->push_back(act);
1992 std::vector<action_list_t> *vec = get_safe_ptr_vect_action(obj_thrd_map, act->get_location());
1993 if (tid >= (int)vec->size())
1994 vec->resize(priv->next_thread_id);
1995 (*vec)[tid].push_back(act);
1997 if ((int)thrd_last_action->size() <= tid)
1998 thrd_last_action->resize(get_num_threads());
1999 (*thrd_last_action)[tid] = act;
2001 if (act->is_wait()) {
2002 void *mutex_loc=(void *) act->get_value();
2003 get_safe_ptr_action(obj_map, mutex_loc)->push_back(act);
2005 std::vector<action_list_t> *vec = get_safe_ptr_vect_action(obj_thrd_map, mutex_loc);
2006 if (tid >= (int)vec->size())
2007 vec->resize(priv->next_thread_id);
2008 (*vec)[tid].push_back(act);
2010 if ((int)thrd_last_action->size() <= tid)
2011 thrd_last_action->resize(get_num_threads());
2012 (*thrd_last_action)[tid] = act;
2017 * @brief Get the last action performed by a particular Thread
2018 * @param tid The thread ID of the Thread in question
2019 * @return The last action in the thread
2021 ModelAction * ModelChecker::get_last_action(thread_id_t tid) const
2023 int threadid = id_to_int(tid);
2024 if (threadid < (int)thrd_last_action->size())
2025 return (*thrd_last_action)[id_to_int(tid)];
2031 * Gets the last memory_order_seq_cst write (in the total global sequence)
2032 * performed on a particular object (i.e., memory location), not including the
2034 * @param curr The current ModelAction; also denotes the object location to
2036 * @return The last seq_cst write
2038 ModelAction * ModelChecker::get_last_seq_cst_write(ModelAction *curr) const
2040 void *location = curr->get_location();
2041 action_list_t *list = get_safe_ptr_action(obj_map, location);
2042 /* Find: max({i in dom(S) | seq_cst(t_i) && isWrite(t_i) && samevar(t_i, t)}) */
2043 action_list_t::reverse_iterator rit;
2044 for (rit = list->rbegin(); rit != list->rend(); rit++)
2045 if ((*rit)->is_write() && (*rit)->is_seqcst() && (*rit) != curr)
2051 * Gets the last memory_order_seq_cst fence (in the total global sequence)
2052 * performed in a particular thread, prior to a particular fence.
2053 * @param tid The ID of the thread to check
2054 * @param before_fence The fence from which to begin the search; if NULL, then
2055 * search for the most recent fence in the thread.
2056 * @return The last prior seq_cst fence in the thread, if exists; otherwise, NULL
2058 ModelAction * ModelChecker::get_last_seq_cst_fence(thread_id_t tid, const ModelAction *before_fence) const
2060 /* All fences should have NULL location */
2061 action_list_t *list = get_safe_ptr_action(obj_map, NULL);
2062 action_list_t::reverse_iterator rit = list->rbegin();
2065 for (; rit != list->rend(); rit++)
2066 if (*rit == before_fence)
2069 ASSERT(*rit == before_fence);
2073 for (; rit != list->rend(); rit++)
2074 if ((*rit)->is_fence() && (tid == (*rit)->get_tid()) && (*rit)->is_seqcst())
2080 * Gets the last unlock operation performed on a particular mutex (i.e., memory
2081 * location). This function identifies the mutex according to the current
2082 * action, which is presumed to perform on the same mutex.
2083 * @param curr The current ModelAction; also denotes the object location to
2085 * @return The last unlock operation
2087 ModelAction * ModelChecker::get_last_unlock(ModelAction *curr) const
2089 void *location = curr->get_location();
2090 action_list_t *list = get_safe_ptr_action(obj_map, location);
2091 /* Find: max({i in dom(S) | isUnlock(t_i) && samevar(t_i, t)}) */
2092 action_list_t::reverse_iterator rit;
2093 for (rit = list->rbegin(); rit != list->rend(); rit++)
2094 if ((*rit)->is_unlock() || (*rit)->is_wait())
2099 ModelAction * ModelChecker::get_parent_action(thread_id_t tid) const
2101 ModelAction *parent = get_last_action(tid);
2103 parent = get_thread(tid)->get_creation();
2108 * Returns the clock vector for a given thread.
2109 * @param tid The thread whose clock vector we want
2110 * @return Desired clock vector
2112 ClockVector * ModelChecker::get_cv(thread_id_t tid) const
2114 return get_parent_action(tid)->get_cv();
2118 * Resolve a set of Promises with a current write. The set is provided in the
2119 * Node corresponding to @a write.
2120 * @param write The ModelAction that is fulfilling Promises
2121 * @return True if promises were resolved; false otherwise
2123 bool ModelChecker::resolve_promises(ModelAction *write)
2125 bool resolved = false;
2126 std::vector< thread_id_t, ModelAlloc<thread_id_t> > threads_to_check;
2128 for (unsigned int i = 0, promise_index = 0; promise_index < promises->size(); i++) {
2129 Promise *promise = (*promises)[promise_index];
2130 if (write->get_node()->get_promise(i)) {
2131 ModelAction *read = promise->get_action();
2132 if (read->is_rmw()) {
2133 mo_graph->addRMWEdge(write, read);
2135 read->read_from(write);
2136 //First fix up the modification order for actions that happened
2138 r_modification_order(read, write);
2139 //Next fix up the modification order for actions that happened
2141 post_r_modification_order(read, write);
2142 //Make sure the promise's value matches the write's value
2143 ASSERT(promise->get_value() == write->get_value());
2146 promises->erase(promises->begin() + promise_index);
2147 threads_to_check.push_back(read->get_tid());
2154 //Check whether reading these writes has made threads unable to
2157 for(unsigned int i=0;i<threads_to_check.size();i++)
2158 mo_check_promises(threads_to_check[i], write);
2164 * Compute the set of promises that could potentially be satisfied by this
2165 * action. Note that the set computation actually appears in the Node, not in
2167 * @param curr The ModelAction that may satisfy promises
2169 void ModelChecker::compute_promises(ModelAction *curr)
2171 for (unsigned int i = 0; i < promises->size(); i++) {
2172 Promise *promise = (*promises)[i];
2173 const ModelAction *act = promise->get_action();
2174 if (!act->happens_before(curr) &&
2176 !act->could_synchronize_with(curr) &&
2177 !act->same_thread(curr) &&
2178 act->get_location() == curr->get_location() &&
2179 promise->get_value() == curr->get_value()) {
2180 curr->get_node()->set_promise(i, act->is_rmw());
2185 /** Checks promises in response to change in ClockVector Threads. */
2186 void ModelChecker::check_promises(thread_id_t tid, ClockVector *old_cv, ClockVector *merge_cv)
2188 for (unsigned int i = 0; i < promises->size(); i++) {
2189 Promise *promise = (*promises)[i];
2190 const ModelAction *act = promise->get_action();
2191 if ((old_cv == NULL || !old_cv->synchronized_since(act)) &&
2192 merge_cv->synchronized_since(act)) {
2193 if (promise->increment_threads(tid)) {
2194 //Promise has failed
2195 priv->failed_promise = true;
2202 void ModelChecker::check_promises_thread_disabled() {
2203 for (unsigned int i = 0; i < promises->size(); i++) {
2204 Promise *promise = (*promises)[i];
2205 if (promise->check_promise()) {
2206 priv->failed_promise = true;
2212 /** Checks promises in response to addition to modification order for threads.
2214 * pthread is the thread that performed the read that created the promise
2216 * pread is the read that created the promise
2218 * pwrite is either the first write to same location as pread by
2219 * pthread that is sequenced after pread or the value read by the
2220 * first read to the same lcoation as pread by pthread that is
2221 * sequenced after pread..
2223 * 1. If tid=pthread, then we check what other threads are reachable
2224 * through the mode order starting with pwrite. Those threads cannot
2225 * perform a write that will resolve the promise due to modification
2226 * order constraints.
2228 * 2. If the tid is not pthread, we check whether pwrite can reach the
2229 * action write through the modification order. If so, that thread
2230 * cannot perform a future write that will resolve the promise due to
2231 * modificatin order constraints.
2233 * @parem tid The thread that either read from the model action
2234 * write, or actually did the model action write.
2236 * @parem write The ModelAction representing the relevant write.
2239 void ModelChecker::mo_check_promises(thread_id_t tid, const ModelAction *write) {
2240 void * location = write->get_location();
2241 for (unsigned int i = 0; i < promises->size(); i++) {
2242 Promise *promise = (*promises)[i];
2243 const ModelAction *act = promise->get_action();
2245 //Is this promise on the same location?
2246 if ( act->get_location() != location )
2249 //same thread as the promise
2250 if ( act->get_tid()==tid ) {
2252 //do we have a pwrite for the promise, if not, set it
2253 if (promise->get_write() == NULL ) {
2254 promise->set_write(write);
2255 //The pwrite cannot happen before the promise
2256 if (write->happens_before(act) && (write != act)) {
2257 priv->failed_promise = true;
2261 if (mo_graph->checkPromise(write, promise)) {
2262 priv->failed_promise = true;
2267 //Don't do any lookups twice for the same thread
2268 if (promise->has_sync_thread(tid))
2271 if (promise->get_write()&&mo_graph->checkReachable(promise->get_write(), write)) {
2272 if (promise->increment_threads(tid)) {
2273 priv->failed_promise = true;
2281 * Compute the set of writes that may break the current pending release
2282 * sequence. This information is extracted from previou release sequence
2285 * @param curr The current ModelAction. Must be a release sequence fixup
2288 void ModelChecker::compute_relseq_breakwrites(ModelAction *curr)
2290 if (pending_rel_seqs->empty())
2293 struct release_seq *pending = pending_rel_seqs->back();
2294 for (unsigned int i = 0; i < pending->writes.size(); i++) {
2295 const ModelAction *write = pending->writes[i];
2296 curr->get_node()->add_relseq_break(write);
2299 /* NULL means don't break the sequence; just synchronize */
2300 curr->get_node()->add_relseq_break(NULL);
2304 * Build up an initial set of all past writes that this 'read' action may read
2305 * from. This set is determined by the clock vector's "happens before"
2307 * @param curr is the current ModelAction that we are exploring; it must be a
2310 void ModelChecker::build_reads_from_past(ModelAction *curr)
2312 std::vector<action_list_t> *thrd_lists = get_safe_ptr_vect_action(obj_thrd_map, curr->get_location());
2314 ASSERT(curr->is_read());
2316 ModelAction *last_sc_write = NULL;
2318 /* Track whether this object has been initialized */
2319 bool initialized = false;
2321 if (curr->is_seqcst()) {
2322 last_sc_write = get_last_seq_cst_write(curr);
2323 /* We have to at least see the last sequentially consistent write,
2324 so we are initialized. */
2325 if (last_sc_write != NULL)
2329 /* Iterate over all threads */
2330 for (i = 0; i < thrd_lists->size(); i++) {
2331 /* Iterate over actions in thread, starting from most recent */
2332 action_list_t *list = &(*thrd_lists)[i];
2333 action_list_t::reverse_iterator rit;
2334 for (rit = list->rbegin(); rit != list->rend(); rit++) {
2335 ModelAction *act = *rit;
2337 /* Only consider 'write' actions */
2338 if (!act->is_write() || act == curr)
2341 /* Don't consider more than one seq_cst write if we are a seq_cst read. */
2342 bool allow_read = true;
2344 if (curr->is_seqcst() && (act->is_seqcst() || (last_sc_write != NULL && act->happens_before(last_sc_write))) && act != last_sc_write)
2346 else if (curr->get_sleep_flag() && !curr->is_seqcst() && !sleep_can_read_from(curr, act))
2350 DEBUG("Adding action to may_read_from:\n");
2351 if (DBG_ENABLED()) {
2355 curr->get_node()->add_read_from(act);
2358 /* Include at most one act per-thread that "happens before" curr */
2359 if (act->happens_before(curr)) {
2367 assert_bug("May read from uninitialized atomic");
2369 if (DBG_ENABLED() || !initialized) {
2370 model_print("Reached read action:\n");
2372 model_print("Printing may_read_from\n");
2373 curr->get_node()->print_may_read_from();
2374 model_print("End printing may_read_from\n");
2378 bool ModelChecker::sleep_can_read_from(ModelAction * curr, const ModelAction *write) {
2380 Node *prevnode=write->get_node()->get_parent();
2382 bool thread_sleep=prevnode->enabled_status(curr->get_tid())==THREAD_SLEEP_SET;
2383 if (write->is_release()&&thread_sleep)
2385 if (!write->is_rmw()) {
2388 if (write->get_reads_from()==NULL)
2390 write=write->get_reads_from();
2394 static void print_list(action_list_t *list, int exec_num = -1)
2396 action_list_t::iterator it;
2398 model_print("---------------------------------------------------------------------\n");
2400 model_print("Execution %d:\n", exec_num);
2402 unsigned int hash=0;
2404 for (it = list->begin(); it != list->end(); it++) {
2406 hash=hash^(hash<<3)^((*it)->hash());
2408 model_print("HASH %u\n", hash);
2409 model_print("---------------------------------------------------------------------\n");
2412 #if SUPPORT_MOD_ORDER_DUMP
2413 void ModelChecker::dumpGraph(char *filename) {
2415 sprintf(buffer, "%s.dot",filename);
2416 FILE *file=fopen(buffer, "w");
2417 fprintf(file, "digraph %s {\n",filename);
2418 mo_graph->dumpNodes(file);
2419 ModelAction ** thread_array=(ModelAction **)model_calloc(1, sizeof(ModelAction *)*get_num_threads());
2421 for (action_list_t::iterator it = action_trace->begin(); it != action_trace->end(); it++) {
2422 ModelAction *action=*it;
2423 if (action->is_read()) {
2424 fprintf(file, "N%u [label=\"%u, T%u\"];\n", action->get_seq_number(),action->get_seq_number(), action->get_tid());
2425 if (action->get_reads_from()!=NULL)
2426 fprintf(file, "N%u -> N%u[label=\"rf\", color=red];\n", action->get_seq_number(), action->get_reads_from()->get_seq_number());
2428 if (thread_array[action->get_tid()] != NULL) {
2429 fprintf(file, "N%u -> N%u[label=\"sb\", color=blue];\n", thread_array[action->get_tid()]->get_seq_number(), action->get_seq_number());
2432 thread_array[action->get_tid()]=action;
2434 fprintf(file,"}\n");
2435 model_free(thread_array);
2440 /** @brief Prints an execution trace summary. */
2441 void ModelChecker::print_summary() const
2443 #if SUPPORT_MOD_ORDER_DUMP
2445 char buffername[100];
2446 sprintf(buffername, "exec%04u", stats.num_total);
2447 mo_graph->dumpGraphToFile(buffername);
2448 sprintf(buffername, "graph%04u", stats.num_total);
2449 dumpGraph(buffername);
2452 if (!isfeasibleprefix())
2453 model_print("INFEASIBLE EXECUTION!\n");
2454 print_list(action_trace, stats.num_total);
2459 * Add a Thread to the system for the first time. Should only be called once
2461 * @param t The Thread to add
2463 void ModelChecker::add_thread(Thread *t)
2465 thread_map->put(id_to_int(t->get_id()), t);
2466 scheduler->add_thread(t);
2470 * Removes a thread from the scheduler.
2471 * @param the thread to remove.
2473 void ModelChecker::remove_thread(Thread *t)
2475 scheduler->remove_thread(t);
2479 * @brief Get a Thread reference by its ID
2480 * @param tid The Thread's ID
2481 * @return A Thread reference
2483 Thread * ModelChecker::get_thread(thread_id_t tid) const
2485 return thread_map->get(id_to_int(tid));
2489 * @brief Get a reference to the Thread in which a ModelAction was executed
2490 * @param act The ModelAction
2491 * @return A Thread reference
2493 Thread * ModelChecker::get_thread(ModelAction *act) const
2495 return get_thread(act->get_tid());
2499 * @brief Check if a Thread is currently enabled
2500 * @param t The Thread to check
2501 * @return True if the Thread is currently enabled
2503 bool ModelChecker::is_enabled(Thread *t) const
2505 return scheduler->is_enabled(t);
2509 * @brief Check if a Thread is currently enabled
2510 * @param tid The ID of the Thread to check
2511 * @return True if the Thread is currently enabled
2513 bool ModelChecker::is_enabled(thread_id_t tid) const
2515 return scheduler->is_enabled(tid);
2519 * Switch from a user-context to the "master thread" context (a.k.a. system
2520 * context). This switch is made with the intention of exploring a particular
2521 * model-checking action (described by a ModelAction object). Must be called
2522 * from a user-thread context.
2524 * @param act The current action that will be explored. May be NULL only if
2525 * trace is exiting via an assertion (see ModelChecker::set_assert and
2526 * ModelChecker::has_asserted).
2527 * @return Return status from the 'swap' call (i.e., success/fail, 0/-1)
2529 int ModelChecker::switch_to_master(ModelAction *act)
2532 Thread *old = thread_current();
2533 set_current_action(act);
2534 old->set_state(THREAD_READY);
2535 return Thread::swap(old, &system_context);
2539 * Takes the next step in the execution, if possible.
2540 * @return Returns true (success) if a step was taken and false otherwise.
2542 bool ModelChecker::take_step() {
2546 Thread *curr = priv->current_action ? get_thread(priv->current_action) : NULL;
2548 if (curr->get_state() == THREAD_READY) {
2549 ASSERT(priv->current_action);
2551 priv->nextThread = check_current_action(priv->current_action);
2552 priv->current_action = NULL;
2554 if (curr->is_blocked() || curr->is_complete())
2555 scheduler->remove_thread(curr);
2560 Thread *next = scheduler->next_thread(priv->nextThread);
2562 /* Infeasible -> don't take any more steps */
2563 if (is_infeasible())
2565 else if (isfeasibleprefix() && have_bug_reports()) {
2570 if (params.bound != 0) {
2571 if (priv->used_sequence_numbers > params.bound) {
2576 DEBUG("(%d, %d)\n", curr ? id_to_int(curr->get_id()) : -1,
2577 next ? id_to_int(next->get_id()) : -1);
2580 * Launch end-of-execution release sequence fixups only when there are:
2582 * (1) no more user threads to run (or when execution replay chooses
2583 * the 'model_thread')
2584 * (2) pending release sequences
2585 * (3) pending assertions (i.e., data races)
2586 * (4) no pending promises
2588 if (!pending_rel_seqs->empty() && (!next || next->is_model_thread()) &&
2589 is_feasible_prefix_ignore_relseq() && !unrealizedraces.empty()) {
2590 model_print("*** WARNING: release sequence fixup action (%zu pending release seuqences) ***\n",
2591 pending_rel_seqs->size());
2592 ModelAction *fixup = new ModelAction(MODEL_FIXUP_RELSEQ,
2593 std::memory_order_seq_cst, NULL, VALUE_NONE,
2595 set_current_action(fixup);
2599 /* next == NULL -> don't take any more steps */
2603 next->set_state(THREAD_RUNNING);
2605 if (next->get_pending() != NULL) {
2606 /* restart a pending action */
2607 set_current_action(next->get_pending());
2608 next->set_pending(NULL);
2609 next->set_state(THREAD_READY);
2613 /* Return false only if swap fails with an error */
2614 return (Thread::swap(&system_context, next) == 0);
2617 /** Wrapper to run the user's main function, with appropriate arguments */
2618 void user_main_wrapper(void *)
2620 user_main(model->params.argc, model->params.argv);
2623 /** @brief Run ModelChecker for the user program */
2624 void ModelChecker::run()
2629 /* Start user program */
2630 add_thread(new Thread(&user_thread, &user_main_wrapper, NULL));
2632 /* Wait for all threads to complete */
2633 while (take_step());
2634 } while (next_execution());