10 #include "snapshot-interface.h"
12 #include "clockvector.h"
13 #include "cyclegraph.h"
16 #include "threads-model.h"
19 #define INITIAL_THREAD_ID 0
24 bug_message(const char *str) {
25 const char *fmt = " [BUG] %s\n";
26 msg = (char *)snapshot_malloc(strlen(fmt) + strlen(str));
27 sprintf(msg, fmt, str);
29 ~bug_message() { if (msg) snapshot_free(msg); }
32 void print() { model_print("%s", msg); }
38 * Structure for holding small ModelChecker members that should be snapshotted
40 struct model_snapshot_members {
41 model_snapshot_members() :
43 /* First thread created will have id INITIAL_THREAD_ID */
44 next_thread_id(INITIAL_THREAD_ID),
45 used_sequence_numbers(0),
49 failed_promise(false),
50 too_many_reads(false),
51 no_valid_reads(false),
52 bad_synchronization(false),
56 ~model_snapshot_members() {
57 for (unsigned int i = 0; i < bugs.size(); i++)
62 ModelAction *current_action;
63 unsigned int next_thread_id;
64 modelclock_t used_sequence_numbers;
65 ModelAction *next_backtrack;
66 std::vector< bug_message *, SnapshotAlloc<bug_message *> > bugs;
67 struct execution_stats stats;
71 /** @brief Incorrectly-ordered synchronization was made */
72 bool bad_synchronization;
78 /** @brief Constructor */
79 ModelChecker::ModelChecker(struct model_params params) :
80 /* Initialize default scheduler */
82 scheduler(new Scheduler()),
84 earliest_diverge(NULL),
85 action_trace(new action_list_t()),
86 thread_map(new HashTable<int, Thread *, int>()),
87 obj_map(new HashTable<const void *, action_list_t *, uintptr_t, 4>()),
88 lock_waiters_map(new HashTable<const void *, action_list_t *, uintptr_t, 4>()),
89 condvar_waiters_map(new HashTable<const void *, action_list_t *, uintptr_t, 4>()),
90 obj_thrd_map(new HashTable<void *, std::vector<action_list_t> *, uintptr_t, 4 >()),
91 promises(new std::vector< Promise *, SnapshotAlloc<Promise *> >()),
92 futurevalues(new std::vector< struct PendingFutureValue, SnapshotAlloc<struct PendingFutureValue> >()),
93 pending_rel_seqs(new std::vector< struct release_seq *, SnapshotAlloc<struct release_seq *> >()),
94 thrd_last_action(new std::vector< ModelAction *, SnapshotAlloc<ModelAction *> >(1)),
95 thrd_last_fence_release(new std::vector< ModelAction *, SnapshotAlloc<ModelAction *> >()),
96 node_stack(new NodeStack()),
97 priv(new struct model_snapshot_members()),
98 mo_graph(new CycleGraph())
100 /* Initialize a model-checker thread, for special ModelActions */
101 model_thread = new Thread(get_next_id());
102 thread_map->put(id_to_int(model_thread->get_id()), model_thread);
105 /** @brief Destructor */
106 ModelChecker::~ModelChecker()
108 for (unsigned int i = 0; i < get_num_threads(); i++)
109 delete thread_map->get(i);
114 delete lock_waiters_map;
115 delete condvar_waiters_map;
118 for (unsigned int i = 0; i < promises->size(); i++)
119 delete (*promises)[i];
122 delete pending_rel_seqs;
124 delete thrd_last_action;
125 delete thrd_last_fence_release;
132 static action_list_t * get_safe_ptr_action(HashTable<const void *, action_list_t *, uintptr_t, 4> * hash, void * ptr)
134 action_list_t *tmp = hash->get(ptr);
136 tmp = new action_list_t();
142 static std::vector<action_list_t> * get_safe_ptr_vect_action(HashTable<void *, std::vector<action_list_t> *, uintptr_t, 4> * hash, void * ptr)
144 std::vector<action_list_t> *tmp = hash->get(ptr);
146 tmp = new std::vector<action_list_t>();
153 * Restores user program to initial state and resets all model-checker data
156 void ModelChecker::reset_to_initial_state()
158 DEBUG("+++ Resetting to initial state +++\n");
159 node_stack->reset_execution();
161 /* Print all model-checker output before rollback */
164 snapshot_backtrack_before(0);
167 /** @return a thread ID for a new Thread */
168 thread_id_t ModelChecker::get_next_id()
170 return priv->next_thread_id++;
173 /** @return the number of user threads created during this execution */
174 unsigned int ModelChecker::get_num_threads() const
176 return priv->next_thread_id;
180 * Must be called from user-thread context (e.g., through the global
181 * thread_current() interface)
183 * @return The currently executing Thread.
185 Thread * ModelChecker::get_current_thread() const
187 return scheduler->get_current_thread();
190 /** @return a sequence number for a new ModelAction */
191 modelclock_t ModelChecker::get_next_seq_num()
193 return ++priv->used_sequence_numbers;
196 Node * ModelChecker::get_curr_node() const
198 return node_stack->get_head();
202 * @brief Choose the next thread to execute.
204 * This function chooses the next thread that should execute. It can force the
205 * adjacency of read/write portions of a RMW action, force THREAD_CREATE to be
206 * followed by a THREAD_START, or it can enforce execution replay/backtracking.
207 * The model-checker may have no preference regarding the next thread (i.e.,
208 * when exploring a new execution ordering), in which case this will return
210 * @param curr The current ModelAction. This action might guide the choice of
212 * @return The next thread to run. If the model-checker has no preference, NULL.
214 Thread * ModelChecker::get_next_thread(ModelAction *curr)
219 /* Do not split atomic actions. */
221 return thread_current();
222 else if (curr->get_type() == THREAD_CREATE)
223 return curr->get_thread_operand();
226 /* Have we completed exploring the preselected path? */
230 /* Else, we are trying to replay an execution */
231 ModelAction *next = node_stack->get_next()->get_action();
233 if (next == diverge) {
234 if (earliest_diverge == NULL || *diverge < *earliest_diverge)
235 earliest_diverge = diverge;
237 Node *nextnode = next->get_node();
238 Node *prevnode = nextnode->get_parent();
239 scheduler->update_sleep_set(prevnode);
241 /* Reached divergence point */
242 if (nextnode->increment_misc()) {
243 /* The next node will try to satisfy a different misc_index values. */
244 tid = next->get_tid();
245 node_stack->pop_restofstack(2);
246 } else if (nextnode->increment_promise()) {
247 /* The next node will try to satisfy a different set of promises. */
248 tid = next->get_tid();
249 node_stack->pop_restofstack(2);
250 } else if (nextnode->increment_read_from()) {
251 /* The next node will read from a different value. */
252 tid = next->get_tid();
253 node_stack->pop_restofstack(2);
254 } else if (nextnode->increment_future_value()) {
255 /* The next node will try to read from a different future value. */
256 tid = next->get_tid();
257 node_stack->pop_restofstack(2);
258 } else if (nextnode->increment_relseq_break()) {
259 /* The next node will try to resolve a release sequence differently */
260 tid = next->get_tid();
261 node_stack->pop_restofstack(2);
264 /* Make a different thread execute for next step */
265 scheduler->add_sleep(get_thread(next->get_tid()));
266 tid = prevnode->get_next_backtrack();
267 /* Make sure the backtracked thread isn't sleeping. */
268 node_stack->pop_restofstack(1);
269 if (diverge == earliest_diverge) {
270 earliest_diverge = prevnode->get_action();
273 /* The correct sleep set is in the parent node. */
276 DEBUG("*** Divergence point ***\n");
280 tid = next->get_tid();
282 DEBUG("*** ModelChecker chose next thread = %d ***\n", id_to_int(tid));
283 ASSERT(tid != THREAD_ID_T_NONE);
284 return thread_map->get(id_to_int(tid));
288 * We need to know what the next actions of all threads in the sleep
289 * set will be. This method computes them and stores the actions at
290 * the corresponding thread object's pending action.
293 void ModelChecker::execute_sleep_set()
295 for (unsigned int i = 0; i < get_num_threads(); i++) {
296 thread_id_t tid = int_to_id(i);
297 Thread *thr = get_thread(tid);
298 if (scheduler->is_sleep_set(thr) && thr->get_pending() == NULL) {
299 thr->set_state(THREAD_RUNNING);
300 scheduler->next_thread(thr);
301 Thread::swap(&system_context, thr);
302 priv->current_action->set_sleep_flag();
303 thr->set_pending(priv->current_action);
308 void ModelChecker::wake_up_sleeping_actions(ModelAction *curr)
310 for (unsigned int i = 0; i < get_num_threads(); i++) {
311 Thread *thr = get_thread(int_to_id(i));
312 if (scheduler->is_sleep_set(thr)) {
313 ModelAction *pending_act = thr->get_pending();
314 if ((!curr->is_rmwr()) && pending_act->could_synchronize_with(curr))
315 //Remove this thread from sleep set
316 scheduler->remove_sleep(thr);
321 /** @brief Alert the model-checker that an incorrectly-ordered
322 * synchronization was made */
323 void ModelChecker::set_bad_synchronization()
325 priv->bad_synchronization = true;
328 bool ModelChecker::has_asserted() const
330 return priv->asserted;
333 void ModelChecker::set_assert()
335 priv->asserted = true;
339 * Check if we are in a deadlock. Should only be called at the end of an
340 * execution, although it should not give false positives in the middle of an
341 * execution (there should be some ENABLED thread).
343 * @return True if program is in a deadlock; false otherwise
345 bool ModelChecker::is_deadlocked() const
347 bool blocking_threads = false;
348 for (unsigned int i = 0; i < get_num_threads(); i++) {
349 thread_id_t tid = int_to_id(i);
352 Thread *t = get_thread(tid);
353 if (!t->is_model_thread() && t->get_pending())
354 blocking_threads = true;
356 return blocking_threads;
360 * Check if this is a complete execution. That is, have all thread completed
361 * execution (rather than exiting because sleep sets have forced a redundant
364 * @return True if the execution is complete.
366 bool ModelChecker::is_complete_execution() const
368 for (unsigned int i = 0; i < get_num_threads(); i++)
369 if (is_enabled(int_to_id(i)))
375 * @brief Assert a bug in the executing program.
377 * Use this function to assert any sort of bug in the user program. If the
378 * current trace is feasible (actually, a prefix of some feasible execution),
379 * then this execution will be aborted, printing the appropriate message. If
380 * the current trace is not yet feasible, the error message will be stashed and
381 * printed if the execution ever becomes feasible.
383 * @param msg Descriptive message for the bug (do not include newline char)
384 * @return True if bug is immediately-feasible
386 bool ModelChecker::assert_bug(const char *msg)
388 priv->bugs.push_back(new bug_message(msg));
390 if (isfeasibleprefix()) {
398 * @brief Assert a bug in the executing program, asserted by a user thread
399 * @see ModelChecker::assert_bug
400 * @param msg Descriptive message for the bug (do not include newline char)
402 void ModelChecker::assert_user_bug(const char *msg)
404 /* If feasible bug, bail out now */
406 switch_to_master(NULL);
409 /** @return True, if any bugs have been reported for this execution */
410 bool ModelChecker::have_bug_reports() const
412 return priv->bugs.size() != 0;
415 /** @brief Print bug report listing for this execution (if any bugs exist) */
416 void ModelChecker::print_bugs() const
418 if (have_bug_reports()) {
419 model_print("Bug report: %zu bug%s detected\n",
421 priv->bugs.size() > 1 ? "s" : "");
422 for (unsigned int i = 0; i < priv->bugs.size(); i++)
423 priv->bugs[i]->print();
428 * @brief Record end-of-execution stats
430 * Must be run when exiting an execution. Records various stats.
431 * @see struct execution_stats
433 void ModelChecker::record_stats()
436 if (!isfeasibleprefix())
437 stats.num_infeasible++;
438 else if (have_bug_reports())
439 stats.num_buggy_executions++;
440 else if (is_complete_execution())
441 stats.num_complete++;
443 stats.num_redundant++;
446 /** @brief Print execution stats */
447 void ModelChecker::print_stats() const
449 model_print("Number of complete, bug-free executions: %d\n", stats.num_complete);
450 model_print("Number of redundant executions: %d\n", stats.num_redundant);
451 model_print("Number of buggy executions: %d\n", stats.num_buggy_executions);
452 model_print("Number of infeasible executions: %d\n", stats.num_infeasible);
453 model_print("Total executions: %d\n", stats.num_total);
454 model_print("Total nodes created: %d\n", node_stack->get_total_nodes());
458 * @brief End-of-exeuction print
459 * @param printbugs Should any existing bugs be printed?
461 void ModelChecker::print_execution(bool printbugs) const
463 print_program_output();
465 if (DBG_ENABLED() || params.verbose) {
466 model_print("Earliest divergence point since last feasible execution:\n");
467 if (earliest_diverge)
468 earliest_diverge->print();
470 model_print("(Not set)\n");
476 /* Don't print invalid bugs */
485 * Queries the model-checker for more executions to explore and, if one
486 * exists, resets the model-checker state to execute a new execution.
488 * @return If there are more executions to explore, return true. Otherwise,
491 bool ModelChecker::next_execution()
494 /* Is this execution a feasible execution that's worth bug-checking? */
495 bool complete = isfeasibleprefix() && (is_complete_execution() ||
498 /* End-of-execution bug checks */
501 assert_bug("Deadlock detected");
509 if (DBG_ENABLED() || params.verbose || (complete && have_bug_reports()))
510 print_execution(complete);
512 clear_program_output();
515 earliest_diverge = NULL;
517 if ((diverge = get_next_backtrack()) == NULL)
521 model_print("Next execution will diverge at:\n");
525 reset_to_initial_state();
529 ModelAction * ModelChecker::get_last_conflict(ModelAction *act)
531 switch (act->get_type()) {
536 /* Optimization: relaxed operations don't need backtracking */
537 if (act->is_relaxed())
539 /* linear search: from most recent to oldest */
540 action_list_t *list = get_safe_ptr_action(obj_map, act->get_location());
541 action_list_t::reverse_iterator rit;
542 for (rit = list->rbegin(); rit != list->rend(); rit++) {
543 ModelAction *prev = *rit;
544 if (prev->could_synchronize_with(act))
550 case ATOMIC_TRYLOCK: {
551 /* linear search: from most recent to oldest */
552 action_list_t *list = get_safe_ptr_action(obj_map, act->get_location());
553 action_list_t::reverse_iterator rit;
554 for (rit = list->rbegin(); rit != list->rend(); rit++) {
555 ModelAction *prev = *rit;
556 if (act->is_conflicting_lock(prev))
561 case ATOMIC_UNLOCK: {
562 /* linear search: from most recent to oldest */
563 action_list_t *list = get_safe_ptr_action(obj_map, act->get_location());
564 action_list_t::reverse_iterator rit;
565 for (rit = list->rbegin(); rit != list->rend(); rit++) {
566 ModelAction *prev = *rit;
567 if (!act->same_thread(prev) && prev->is_failed_trylock())
573 /* linear search: from most recent to oldest */
574 action_list_t *list = get_safe_ptr_action(obj_map, act->get_location());
575 action_list_t::reverse_iterator rit;
576 for (rit = list->rbegin(); rit != list->rend(); rit++) {
577 ModelAction *prev = *rit;
578 if (!act->same_thread(prev) && prev->is_failed_trylock())
580 if (!act->same_thread(prev) && prev->is_notify())
586 case ATOMIC_NOTIFY_ALL:
587 case ATOMIC_NOTIFY_ONE: {
588 /* linear search: from most recent to oldest */
589 action_list_t *list = get_safe_ptr_action(obj_map, act->get_location());
590 action_list_t::reverse_iterator rit;
591 for (rit = list->rbegin(); rit != list->rend(); rit++) {
592 ModelAction *prev = *rit;
593 if (!act->same_thread(prev) && prev->is_wait())
604 /** This method finds backtracking points where we should try to
605 * reorder the parameter ModelAction against.
607 * @param the ModelAction to find backtracking points for.
609 void ModelChecker::set_backtracking(ModelAction *act)
611 Thread *t = get_thread(act);
612 ModelAction *prev = get_last_conflict(act);
616 Node *node = prev->get_node()->get_parent();
618 int low_tid, high_tid;
619 if (node->enabled_status(t->get_id()) == THREAD_ENABLED) {
620 low_tid = id_to_int(act->get_tid());
621 high_tid = low_tid + 1;
624 high_tid = get_num_threads();
627 for (int i = low_tid; i < high_tid; i++) {
628 thread_id_t tid = int_to_id(i);
630 /* Make sure this thread can be enabled here. */
631 if (i >= node->get_num_threads())
634 /* Don't backtrack into a point where the thread is disabled or sleeping. */
635 if (node->enabled_status(tid) != THREAD_ENABLED)
638 /* Check if this has been explored already */
639 if (node->has_been_explored(tid))
642 /* See if fairness allows */
643 if (model->params.fairwindow != 0 && !node->has_priority(tid)) {
645 for (int t = 0; t < node->get_num_threads(); t++) {
646 thread_id_t tother = int_to_id(t);
647 if (node->is_enabled(tother) && node->has_priority(tother)) {
655 /* Cache the latest backtracking point */
656 set_latest_backtrack(prev);
658 /* If this is a new backtracking point, mark the tree */
659 if (!node->set_backtrack(tid))
661 DEBUG("Setting backtrack: conflict = %d, instead tid = %d\n",
662 id_to_int(prev->get_tid()),
663 id_to_int(t->get_id()));
672 * @brief Cache the a backtracking point as the "most recent", if eligible
674 * Note that this does not prepare the NodeStack for this backtracking
675 * operation, it only caches the action on a per-execution basis
677 * @param act The operation at which we should explore a different next action
678 * (i.e., backtracking point)
679 * @return True, if this action is now the most recent backtracking point;
682 bool ModelChecker::set_latest_backtrack(ModelAction *act)
684 if (!priv->next_backtrack || *act > *priv->next_backtrack) {
685 priv->next_backtrack = act;
692 * Returns last backtracking point. The model checker will explore a different
693 * path for this point in the next execution.
694 * @return The ModelAction at which the next execution should diverge.
696 ModelAction * ModelChecker::get_next_backtrack()
698 ModelAction *next = priv->next_backtrack;
699 priv->next_backtrack = NULL;
704 * Processes a read or rmw model action.
705 * @param curr is the read model action to process.
706 * @param second_part_of_rmw is boolean that is true is this is the second action of a rmw.
707 * @return True if processing this read updates the mo_graph.
709 bool ModelChecker::process_read(ModelAction *curr, bool second_part_of_rmw)
711 uint64_t value = VALUE_NONE;
712 bool updated = false;
714 const ModelAction *reads_from = curr->get_node()->get_read_from();
715 if (reads_from != NULL) {
716 mo_graph->startChanges();
718 value = reads_from->get_value();
719 bool r_status = false;
721 if (!second_part_of_rmw) {
722 check_recency(curr, reads_from);
723 r_status = r_modification_order(curr, reads_from);
726 if (!second_part_of_rmw && is_infeasible() && (curr->get_node()->increment_read_from() || curr->get_node()->increment_future_value())) {
727 mo_graph->rollbackChanges();
728 priv->too_many_reads = false;
732 read_from(curr, reads_from);
733 mo_graph->commitChanges();
734 mo_check_promises(curr, true);
737 } else if (!second_part_of_rmw) {
738 /* Read from future value */
739 struct future_value fv = curr->get_node()->get_future_value();
740 Promise *promise = new Promise(curr, fv);
742 curr->set_read_from_promise(promise);
743 promises->push_back(promise);
744 mo_graph->startChanges();
745 updated = r_modification_order(curr, promise);
746 mo_graph->commitChanges();
748 get_thread(curr)->set_return_value(value);
754 * Processes a lock, trylock, or unlock model action. @param curr is
755 * the read model action to process.
757 * The try lock operation checks whether the lock is taken. If not,
758 * it falls to the normal lock operation case. If so, it returns
761 * The lock operation has already been checked that it is enabled, so
762 * it just grabs the lock and synchronizes with the previous unlock.
764 * The unlock operation has to re-enable all of the threads that are
765 * waiting on the lock.
767 * @return True if synchronization was updated; false otherwise
769 bool ModelChecker::process_mutex(ModelAction *curr)
771 std::mutex *mutex = NULL;
772 struct std::mutex_state *state = NULL;
774 if (curr->is_trylock() || curr->is_lock() || curr->is_unlock()) {
775 mutex = (std::mutex *)curr->get_location();
776 state = mutex->get_state();
777 } else if (curr->is_wait()) {
778 mutex = (std::mutex *)curr->get_value();
779 state = mutex->get_state();
782 switch (curr->get_type()) {
783 case ATOMIC_TRYLOCK: {
784 bool success = !state->islocked;
785 curr->set_try_lock(success);
787 get_thread(curr)->set_return_value(0);
790 get_thread(curr)->set_return_value(1);
792 //otherwise fall into the lock case
794 if (curr->get_cv()->getClock(state->alloc_tid) <= state->alloc_clock)
795 assert_bug("Lock access before initialization");
796 state->islocked = true;
797 ModelAction *unlock = get_last_unlock(curr);
798 //synchronize with the previous unlock statement
799 if (unlock != NULL) {
800 curr->synchronize_with(unlock);
805 case ATOMIC_UNLOCK: {
807 state->islocked = false;
808 //wake up the other threads
809 action_list_t *waiters = get_safe_ptr_action(lock_waiters_map, curr->get_location());
810 //activate all the waiting threads
811 for (action_list_t::iterator rit = waiters->begin(); rit != waiters->end(); rit++) {
812 scheduler->wake(get_thread(*rit));
819 state->islocked = false;
820 //wake up the other threads
821 action_list_t *waiters = get_safe_ptr_action(lock_waiters_map, (void *) curr->get_value());
822 //activate all the waiting threads
823 for (action_list_t::iterator rit = waiters->begin(); rit != waiters->end(); rit++) {
824 scheduler->wake(get_thread(*rit));
827 //check whether we should go to sleep or not...simulate spurious failures
828 if (curr->get_node()->get_misc() == 0) {
829 get_safe_ptr_action(condvar_waiters_map, curr->get_location())->push_back(curr);
831 scheduler->sleep(get_thread(curr));
835 case ATOMIC_NOTIFY_ALL: {
836 action_list_t *waiters = get_safe_ptr_action(condvar_waiters_map, curr->get_location());
837 //activate all the waiting threads
838 for (action_list_t::iterator rit = waiters->begin(); rit != waiters->end(); rit++) {
839 scheduler->wake(get_thread(*rit));
844 case ATOMIC_NOTIFY_ONE: {
845 action_list_t *waiters = get_safe_ptr_action(condvar_waiters_map, curr->get_location());
846 int wakeupthread = curr->get_node()->get_misc();
847 action_list_t::iterator it = waiters->begin();
848 advance(it, wakeupthread);
849 scheduler->wake(get_thread(*it));
860 void ModelChecker::add_future_value(const ModelAction *writer, ModelAction *reader)
862 /* Do more ambitious checks now that mo is more complete */
863 if (mo_may_allow(writer, reader)) {
864 Node *node = reader->get_node();
866 /* Find an ancestor thread which exists at the time of the reader */
867 Thread *write_thread = get_thread(writer);
868 while (id_to_int(write_thread->get_id()) >= node->get_num_threads())
869 write_thread = write_thread->get_parent();
871 struct future_value fv = {
873 writer->get_seq_number() + params.maxfuturedelay,
874 write_thread->get_id(),
876 if (node->add_future_value(fv))
877 set_latest_backtrack(reader);
882 * Process a write ModelAction
883 * @param curr The ModelAction to process
884 * @return True if the mo_graph was updated or promises were resolved
886 bool ModelChecker::process_write(ModelAction *curr)
888 bool updated_mod_order = w_modification_order(curr);
889 bool updated_promises = resolve_promises(curr);
891 if (promises->size() == 0) {
892 for (unsigned int i = 0; i < futurevalues->size(); i++) {
893 struct PendingFutureValue pfv = (*futurevalues)[i];
894 add_future_value(pfv.writer, pfv.act);
896 futurevalues->clear();
899 mo_graph->commitChanges();
900 mo_check_promises(curr, false);
902 get_thread(curr)->set_return_value(VALUE_NONE);
903 return updated_mod_order || updated_promises;
907 * Process a fence ModelAction
908 * @param curr The ModelAction to process
909 * @return True if synchronization was updated
911 bool ModelChecker::process_fence(ModelAction *curr)
914 * fence-relaxed: no-op
915 * fence-release: only log the occurence (not in this function), for
916 * use in later synchronization
917 * fence-acquire (this function): search for hypothetical release
920 bool updated = false;
921 if (curr->is_acquire()) {
922 action_list_t *list = action_trace;
923 action_list_t::reverse_iterator rit;
924 /* Find X : is_read(X) && X --sb-> curr */
925 for (rit = list->rbegin(); rit != list->rend(); rit++) {
926 ModelAction *act = *rit;
929 if (act->get_tid() != curr->get_tid())
931 /* Stop at the beginning of the thread */
932 if (act->is_thread_start())
934 /* Stop once we reach a prior fence-acquire */
935 if (act->is_fence() && act->is_acquire())
939 /* read-acquire will find its own release sequences */
940 if (act->is_acquire())
943 /* Establish hypothetical release sequences */
944 rel_heads_list_t release_heads;
945 get_release_seq_heads(curr, act, &release_heads);
946 for (unsigned int i = 0; i < release_heads.size(); i++)
947 if (!curr->synchronize_with(release_heads[i]))
948 set_bad_synchronization();
949 if (release_heads.size() != 0)
957 * @brief Process the current action for thread-related activity
959 * Performs current-action processing for a THREAD_* ModelAction. Proccesses
960 * may include setting Thread status, completing THREAD_FINISH/THREAD_JOIN
961 * synchronization, etc. This function is a no-op for non-THREAD actions
962 * (e.g., ATOMIC_{READ,WRITE,RMW,LOCK}, etc.)
964 * @param curr The current action
965 * @return True if synchronization was updated or a thread completed
967 bool ModelChecker::process_thread_action(ModelAction *curr)
969 bool updated = false;
971 switch (curr->get_type()) {
972 case THREAD_CREATE: {
973 Thread *th = curr->get_thread_operand();
974 th->set_creation(curr);
975 /* Promises can be satisfied by children */
976 for (unsigned int i = 0; i < promises->size(); i++) {
977 Promise *promise = (*promises)[i];
978 if (promise->thread_is_available(curr->get_tid()))
979 promise->add_thread(th->get_id());
984 Thread *blocking = curr->get_thread_operand();
985 ModelAction *act = get_last_action(blocking->get_id());
986 curr->synchronize_with(act);
987 updated = true; /* trigger rel-seq checks */
990 case THREAD_FINISH: {
991 Thread *th = get_thread(curr);
992 while (!th->wait_list_empty()) {
993 ModelAction *act = th->pop_wait_list();
994 scheduler->wake(get_thread(act));
997 /* Completed thread can't satisfy promises */
998 for (unsigned int i = 0; i < promises->size(); i++) {
999 Promise *promise = (*promises)[i];
1000 if (promise->thread_is_available(th->get_id()))
1001 if (promise->eliminate_thread(th->get_id()))
1002 priv->failed_promise = true;
1004 updated = true; /* trigger rel-seq checks */
1007 case THREAD_START: {
1008 check_promises(curr->get_tid(), NULL, curr->get_cv());
1019 * @brief Process the current action for release sequence fixup activity
1021 * Performs model-checker release sequence fixups for the current action,
1022 * forcing a single pending release sequence to break (with a given, potential
1023 * "loose" write) or to complete (i.e., synchronize). If a pending release
1024 * sequence forms a complete release sequence, then we must perform the fixup
1025 * synchronization, mo_graph additions, etc.
1027 * @param curr The current action; must be a release sequence fixup action
1028 * @param work_queue The work queue to which to add work items as they are
1031 void ModelChecker::process_relseq_fixup(ModelAction *curr, work_queue_t *work_queue)
1033 const ModelAction *write = curr->get_node()->get_relseq_break();
1034 struct release_seq *sequence = pending_rel_seqs->back();
1035 pending_rel_seqs->pop_back();
1037 ModelAction *acquire = sequence->acquire;
1038 const ModelAction *rf = sequence->rf;
1039 const ModelAction *release = sequence->release;
1043 ASSERT(release->same_thread(rf));
1045 if (write == NULL) {
1047 * @todo Forcing a synchronization requires that we set
1048 * modification order constraints. For instance, we can't allow
1049 * a fixup sequence in which two separate read-acquire
1050 * operations read from the same sequence, where the first one
1051 * synchronizes and the other doesn't. Essentially, we can't
1052 * allow any writes to insert themselves between 'release' and
1056 /* Must synchronize */
1057 if (!acquire->synchronize_with(release)) {
1058 set_bad_synchronization();
1061 /* Re-check all pending release sequences */
1062 work_queue->push_back(CheckRelSeqWorkEntry(NULL));
1063 /* Re-check act for mo_graph edges */
1064 work_queue->push_back(MOEdgeWorkEntry(acquire));
1066 /* propagate synchronization to later actions */
1067 action_list_t::reverse_iterator rit = action_trace->rbegin();
1068 for (; (*rit) != acquire; rit++) {
1069 ModelAction *propagate = *rit;
1070 if (acquire->happens_before(propagate)) {
1071 propagate->synchronize_with(acquire);
1072 /* Re-check 'propagate' for mo_graph edges */
1073 work_queue->push_back(MOEdgeWorkEntry(propagate));
1077 /* Break release sequence with new edges:
1078 * release --mo--> write --mo--> rf */
1079 mo_graph->addEdge(release, write);
1080 mo_graph->addEdge(write, rf);
1083 /* See if we have realized a data race */
1088 * Initialize the current action by performing one or more of the following
1089 * actions, as appropriate: merging RMWR and RMWC/RMW actions, stepping forward
1090 * in the NodeStack, manipulating backtracking sets, allocating and
1091 * initializing clock vectors, and computing the promises to fulfill.
1093 * @param curr The current action, as passed from the user context; may be
1094 * freed/invalidated after the execution of this function, with a different
1095 * action "returned" its place (pass-by-reference)
1096 * @return True if curr is a newly-explored action; false otherwise
1098 bool ModelChecker::initialize_curr_action(ModelAction **curr)
1100 ModelAction *newcurr;
1102 if ((*curr)->is_rmwc() || (*curr)->is_rmw()) {
1103 newcurr = process_rmw(*curr);
1106 if (newcurr->is_rmw())
1107 compute_promises(newcurr);
1113 (*curr)->set_seq_number(get_next_seq_num());
1115 newcurr = node_stack->explore_action(*curr, scheduler->get_enabled_array());
1117 /* First restore type and order in case of RMW operation */
1118 if ((*curr)->is_rmwr())
1119 newcurr->copy_typeandorder(*curr);
1121 ASSERT((*curr)->get_location() == newcurr->get_location());
1122 newcurr->copy_from_new(*curr);
1124 /* Discard duplicate ModelAction; use action from NodeStack */
1127 /* Always compute new clock vector */
1128 newcurr->create_cv(get_parent_action(newcurr->get_tid()));
1131 return false; /* Action was explored previously */
1135 /* Always compute new clock vector */
1136 newcurr->create_cv(get_parent_action(newcurr->get_tid()));
1138 /* Assign most recent release fence */
1139 newcurr->set_last_fence_release(get_last_fence_release(newcurr->get_tid()));
1142 * Perform one-time actions when pushing new ModelAction onto
1145 if (newcurr->is_write())
1146 compute_promises(newcurr);
1147 else if (newcurr->is_relseq_fixup())
1148 compute_relseq_breakwrites(newcurr);
1149 else if (newcurr->is_wait())
1150 newcurr->get_node()->set_misc_max(2);
1151 else if (newcurr->is_notify_one()) {
1152 newcurr->get_node()->set_misc_max(get_safe_ptr_action(condvar_waiters_map, newcurr->get_location())->size());
1154 return true; /* This was a new ModelAction */
1159 * @brief Establish reads-from relation between two actions
1161 * Perform basic operations involved with establishing a concrete rf relation,
1162 * including setting the ModelAction data and checking for release sequences.
1164 * @param act The action that is reading (must be a read)
1165 * @param rf The action from which we are reading (must be a write)
1167 * @return True if this read established synchronization
1169 bool ModelChecker::read_from(ModelAction *act, const ModelAction *rf)
1171 act->set_read_from(rf);
1172 if (rf != NULL && act->is_acquire()) {
1173 rel_heads_list_t release_heads;
1174 get_release_seq_heads(act, act, &release_heads);
1175 int num_heads = release_heads.size();
1176 for (unsigned int i = 0; i < release_heads.size(); i++)
1177 if (!act->synchronize_with(release_heads[i])) {
1178 set_bad_synchronization();
1181 return num_heads > 0;
1187 * @brief Check whether a model action is enabled.
1189 * Checks whether a lock or join operation would be successful (i.e., is the
1190 * lock already locked, or is the joined thread already complete). If not, put
1191 * the action in a waiter list.
1193 * @param curr is the ModelAction to check whether it is enabled.
1194 * @return a bool that indicates whether the action is enabled.
1196 bool ModelChecker::check_action_enabled(ModelAction *curr) {
1197 if (curr->is_lock()) {
1198 std::mutex *lock = (std::mutex *)curr->get_location();
1199 struct std::mutex_state *state = lock->get_state();
1200 if (state->islocked) {
1201 //Stick the action in the appropriate waiting queue
1202 get_safe_ptr_action(lock_waiters_map, curr->get_location())->push_back(curr);
1205 } else if (curr->get_type() == THREAD_JOIN) {
1206 Thread *blocking = (Thread *)curr->get_location();
1207 if (!blocking->is_complete()) {
1208 blocking->push_wait_list(curr);
1217 * Stores the ModelAction for the current thread action. Call this
1218 * immediately before switching from user- to system-context to pass
1219 * data between them.
1220 * @param act The ModelAction created by the user-thread action
1222 void ModelChecker::set_current_action(ModelAction *act) {
1223 priv->current_action = act;
1227 * This is the heart of the model checker routine. It performs model-checking
1228 * actions corresponding to a given "current action." Among other processes, it
1229 * calculates reads-from relationships, updates synchronization clock vectors,
1230 * forms a memory_order constraints graph, and handles replay/backtrack
1231 * execution when running permutations of previously-observed executions.
1233 * @param curr The current action to process
1234 * @return The ModelAction that is actually executed; may be different than
1235 * curr; may be NULL, if the current action is not enabled to run
1237 ModelAction * ModelChecker::check_current_action(ModelAction *curr)
1240 bool second_part_of_rmw = curr->is_rmwc() || curr->is_rmw();
1242 if (!check_action_enabled(curr)) {
1243 /* Make the execution look like we chose to run this action
1244 * much later, when a lock/join can succeed */
1245 get_thread(curr)->set_pending(curr);
1246 scheduler->sleep(get_thread(curr));
1250 bool newly_explored = initialize_curr_action(&curr);
1256 wake_up_sleeping_actions(curr);
1258 /* Add the action to lists before any other model-checking tasks */
1259 if (!second_part_of_rmw)
1260 add_action_to_lists(curr);
1262 /* Build may_read_from set for newly-created actions */
1263 if (newly_explored && curr->is_read())
1264 build_reads_from_past(curr);
1266 /* Initialize work_queue with the "current action" work */
1267 work_queue_t work_queue(1, CheckCurrWorkEntry(curr));
1268 while (!work_queue.empty() && !has_asserted()) {
1269 WorkQueueEntry work = work_queue.front();
1270 work_queue.pop_front();
1272 switch (work.type) {
1273 case WORK_CHECK_CURR_ACTION: {
1274 ModelAction *act = work.action;
1275 bool update = false; /* update this location's release seq's */
1276 bool update_all = false; /* update all release seq's */
1278 if (process_thread_action(curr))
1281 if (act->is_read() && process_read(act, second_part_of_rmw))
1284 if (act->is_write() && process_write(act))
1287 if (act->is_fence() && process_fence(act))
1290 if (act->is_mutex_op() && process_mutex(act))
1293 if (act->is_relseq_fixup())
1294 process_relseq_fixup(curr, &work_queue);
1297 work_queue.push_back(CheckRelSeqWorkEntry(NULL));
1299 work_queue.push_back(CheckRelSeqWorkEntry(act->get_location()));
1302 case WORK_CHECK_RELEASE_SEQ:
1303 resolve_release_sequences(work.location, &work_queue);
1305 case WORK_CHECK_MO_EDGES: {
1306 /** @todo Complete verification of work_queue */
1307 ModelAction *act = work.action;
1308 bool updated = false;
1310 if (act->is_read()) {
1311 const ModelAction *rf = act->get_reads_from();
1312 const Promise *promise = act->get_reads_from_promise();
1314 if (r_modification_order(act, rf))
1316 } else if (promise) {
1317 if (r_modification_order(act, promise))
1321 if (act->is_write()) {
1322 if (w_modification_order(act))
1325 mo_graph->commitChanges();
1328 work_queue.push_back(CheckRelSeqWorkEntry(act->get_location()));
1337 check_curr_backtracking(curr);
1338 set_backtracking(curr);
1342 void ModelChecker::check_curr_backtracking(ModelAction *curr)
1344 Node *currnode = curr->get_node();
1345 Node *parnode = currnode->get_parent();
1347 if ((parnode && !parnode->backtrack_empty()) ||
1348 !currnode->misc_empty() ||
1349 !currnode->read_from_empty() ||
1350 !currnode->future_value_empty() ||
1351 !currnode->promise_empty() ||
1352 !currnode->relseq_break_empty()) {
1353 set_latest_backtrack(curr);
1357 bool ModelChecker::promises_expired() const
1359 for (unsigned int i = 0; i < promises->size(); i++) {
1360 Promise *promise = (*promises)[i];
1361 if (promise->get_expiration() < priv->used_sequence_numbers)
1368 * This is the strongest feasibility check available.
1369 * @return whether the current trace (partial or complete) must be a prefix of
1372 bool ModelChecker::isfeasibleprefix() const
1374 return pending_rel_seqs->size() == 0 && is_feasible_prefix_ignore_relseq();
1378 * Print disagnostic information about an infeasible execution
1379 * @param prefix A string to prefix the output with; if NULL, then a default
1380 * message prefix will be provided
1382 void ModelChecker::print_infeasibility(const char *prefix) const
1386 if (mo_graph->checkForCycles())
1387 ptr += sprintf(ptr, "[mo cycle]");
1388 if (priv->failed_promise)
1389 ptr += sprintf(ptr, "[failed promise]");
1390 if (priv->too_many_reads)
1391 ptr += sprintf(ptr, "[too many reads]");
1392 if (priv->no_valid_reads)
1393 ptr += sprintf(ptr, "[no valid reads-from]");
1394 if (priv->bad_synchronization)
1395 ptr += sprintf(ptr, "[bad sw ordering]");
1396 if (promises_expired())
1397 ptr += sprintf(ptr, "[promise expired]");
1398 if (promises->size() != 0)
1399 ptr += sprintf(ptr, "[unresolved promise]");
1401 model_print("%s: %s\n", prefix ? prefix : "Infeasible", buf);
1405 * Returns whether the current completed trace is feasible, except for pending
1406 * release sequences.
1408 bool ModelChecker::is_feasible_prefix_ignore_relseq() const
1410 return !is_infeasible() && promises->size() == 0;
1414 * Check if the current partial trace is infeasible. Does not check any
1415 * end-of-execution flags, which might rule out the execution. Thus, this is
1416 * useful only for ruling an execution as infeasible.
1417 * @return whether the current partial trace is infeasible.
1419 bool ModelChecker::is_infeasible() const
1421 return mo_graph->checkForCycles() ||
1422 priv->no_valid_reads ||
1423 priv->failed_promise ||
1424 priv->too_many_reads ||
1425 priv->bad_synchronization ||
1429 /** Close out a RMWR by converting previous RMWR into a RMW or READ. */
1430 ModelAction * ModelChecker::process_rmw(ModelAction *act) {
1431 ModelAction *lastread = get_last_action(act->get_tid());
1432 lastread->process_rmw(act);
1433 if (act->is_rmw()) {
1434 if (lastread->get_reads_from())
1435 mo_graph->addRMWEdge(lastread->get_reads_from(), lastread);
1437 mo_graph->addRMWEdge(lastread->get_reads_from_promise(), lastread);
1438 mo_graph->commitChanges();
1444 * Checks whether a thread has read from the same write for too many times
1445 * without seeing the effects of a later write.
1448 * 1) there must a different write that we could read from that would satisfy the modification order,
1449 * 2) we must have read from the same value in excess of maxreads times, and
1450 * 3) that other write must have been in the reads_from set for maxreads times.
1452 * If so, we decide that the execution is no longer feasible.
1454 void ModelChecker::check_recency(ModelAction *curr, const ModelAction *rf)
1456 if (params.maxreads != 0) {
1457 if (curr->get_node()->get_read_from_size() <= 1)
1459 //Must make sure that execution is currently feasible... We could
1460 //accidentally clear by rolling back
1461 if (is_infeasible())
1463 std::vector<action_list_t> *thrd_lists = get_safe_ptr_vect_action(obj_thrd_map, curr->get_location());
1464 int tid = id_to_int(curr->get_tid());
1467 if ((int)thrd_lists->size() <= tid)
1469 action_list_t *list = &(*thrd_lists)[tid];
1471 action_list_t::reverse_iterator rit = list->rbegin();
1472 /* Skip past curr */
1473 for (; (*rit) != curr; rit++)
1475 /* go past curr now */
1478 action_list_t::reverse_iterator ritcopy = rit;
1479 //See if we have enough reads from the same value
1481 for (; count < params.maxreads; rit++, count++) {
1482 if (rit == list->rend())
1484 ModelAction *act = *rit;
1485 if (!act->is_read())
1488 if (act->get_reads_from() != rf)
1490 if (act->get_node()->get_read_from_size() <= 1)
1493 for (int i = 0; i < curr->get_node()->get_read_from_size(); i++) {
1495 const ModelAction *write = curr->get_node()->get_read_from_at(i);
1497 /* Need a different write */
1501 /* Test to see whether this is a feasible write to read from */
1502 /** NOTE: all members of read-from set should be
1503 * feasible, so we no longer check it here **/
1507 bool feasiblewrite = true;
1508 //new we need to see if this write works for everyone
1510 for (int loop = count; loop > 0; loop--, rit++) {
1511 ModelAction *act = *rit;
1512 bool foundvalue = false;
1513 for (int j = 0; j < act->get_node()->get_read_from_size(); j++) {
1514 if (act->get_node()->get_read_from_at(j) == write) {
1520 feasiblewrite = false;
1524 if (feasiblewrite) {
1525 priv->too_many_reads = true;
1533 * Updates the mo_graph with the constraints imposed from the current
1536 * Basic idea is the following: Go through each other thread and find
1537 * the last action that happened before our read. Two cases:
1539 * (1) The action is a write => that write must either occur before
1540 * the write we read from or be the write we read from.
1542 * (2) The action is a read => the write that that action read from
1543 * must occur before the write we read from or be the same write.
1545 * @param curr The current action. Must be a read.
1546 * @param rf The ModelAction or Promise that curr reads from. Must be a write.
1547 * @return True if modification order edges were added; false otherwise
1549 template <typename rf_type>
1550 bool ModelChecker::r_modification_order(ModelAction *curr, const rf_type *rf)
1552 std::vector<action_list_t> *thrd_lists = get_safe_ptr_vect_action(obj_thrd_map, curr->get_location());
1555 ASSERT(curr->is_read());
1557 /* Last SC fence in the current thread */
1558 ModelAction *last_sc_fence_local = get_last_seq_cst_fence(curr->get_tid(), NULL);
1560 /* Iterate over all threads */
1561 for (i = 0; i < thrd_lists->size(); i++) {
1562 /* Last SC fence in thread i */
1563 ModelAction *last_sc_fence_thread_local = NULL;
1564 if (int_to_id((int)i) != curr->get_tid())
1565 last_sc_fence_thread_local = get_last_seq_cst_fence(int_to_id(i), NULL);
1567 /* Last SC fence in thread i, before last SC fence in current thread */
1568 ModelAction *last_sc_fence_thread_before = NULL;
1569 if (last_sc_fence_local)
1570 last_sc_fence_thread_before = get_last_seq_cst_fence(int_to_id(i), last_sc_fence_local);
1572 /* Iterate over actions in thread, starting from most recent */
1573 action_list_t *list = &(*thrd_lists)[i];
1574 action_list_t::reverse_iterator rit;
1575 for (rit = list->rbegin(); rit != list->rend(); rit++) {
1576 ModelAction *act = *rit;
1578 if (act->is_write() && !act->equals(rf) && act != curr) {
1579 /* C++, Section 29.3 statement 5 */
1580 if (curr->is_seqcst() && last_sc_fence_thread_local &&
1581 *act < *last_sc_fence_thread_local) {
1582 added = mo_graph->addEdge(act, rf) || added;
1585 /* C++, Section 29.3 statement 4 */
1586 else if (act->is_seqcst() && last_sc_fence_local &&
1587 *act < *last_sc_fence_local) {
1588 added = mo_graph->addEdge(act, rf) || added;
1591 /* C++, Section 29.3 statement 6 */
1592 else if (last_sc_fence_thread_before &&
1593 *act < *last_sc_fence_thread_before) {
1594 added = mo_graph->addEdge(act, rf) || added;
1600 * Include at most one act per-thread that "happens
1601 * before" curr. Don't consider reflexively.
1603 if (act->happens_before(curr) && act != curr) {
1604 if (act->is_write()) {
1605 if (!act->equals(rf)) {
1606 added = mo_graph->addEdge(act, rf) || added;
1609 const ModelAction *prevreadfrom = act->get_reads_from();
1610 //if the previous read is unresolved, keep going...
1611 if (prevreadfrom == NULL)
1614 if (!prevreadfrom->equals(rf)) {
1615 added = mo_graph->addEdge(prevreadfrom, rf) || added;
1624 * All compatible, thread-exclusive promises must be ordered after any
1625 * concrete loads from the same thread
1627 for (unsigned int i = 0; i < promises->size(); i++)
1628 if ((*promises)[i]->is_compatible_exclusive(curr))
1629 added = mo_graph->addEdge(rf, (*promises)[i]) || added;
1635 * Updates the mo_graph with the constraints imposed from the current write.
1637 * Basic idea is the following: Go through each other thread and find
1638 * the lastest action that happened before our write. Two cases:
1640 * (1) The action is a write => that write must occur before
1643 * (2) The action is a read => the write that that action read from
1644 * must occur before the current write.
1646 * This method also handles two other issues:
1648 * (I) Sequential Consistency: Making sure that if the current write is
1649 * seq_cst, that it occurs after the previous seq_cst write.
1651 * (II) Sending the write back to non-synchronizing reads.
1653 * @param curr The current action. Must be a write.
1654 * @return True if modification order edges were added; false otherwise
1656 bool ModelChecker::w_modification_order(ModelAction *curr)
1658 std::vector<action_list_t> *thrd_lists = get_safe_ptr_vect_action(obj_thrd_map, curr->get_location());
1661 ASSERT(curr->is_write());
1663 if (curr->is_seqcst()) {
1664 /* We have to at least see the last sequentially consistent write,
1665 so we are initialized. */
1666 ModelAction *last_seq_cst = get_last_seq_cst_write(curr);
1667 if (last_seq_cst != NULL) {
1668 added = mo_graph->addEdge(last_seq_cst, curr) || added;
1672 /* Last SC fence in the current thread */
1673 ModelAction *last_sc_fence_local = get_last_seq_cst_fence(curr->get_tid(), NULL);
1675 /* Iterate over all threads */
1676 for (i = 0; i < thrd_lists->size(); i++) {
1677 /* Last SC fence in thread i, before last SC fence in current thread */
1678 ModelAction *last_sc_fence_thread_before = NULL;
1679 if (last_sc_fence_local && int_to_id((int)i) != curr->get_tid())
1680 last_sc_fence_thread_before = get_last_seq_cst_fence(int_to_id(i), last_sc_fence_local);
1682 /* Iterate over actions in thread, starting from most recent */
1683 action_list_t *list = &(*thrd_lists)[i];
1684 action_list_t::reverse_iterator rit;
1685 for (rit = list->rbegin(); rit != list->rend(); rit++) {
1686 ModelAction *act = *rit;
1689 * 1) If RMW and it actually read from something, then we
1690 * already have all relevant edges, so just skip to next
1693 * 2) If RMW and it didn't read from anything, we should
1694 * whatever edge we can get to speed up convergence.
1696 * 3) If normal write, we need to look at earlier actions, so
1697 * continue processing list.
1699 if (curr->is_rmw()) {
1700 if (curr->get_reads_from() != NULL)
1708 /* C++, Section 29.3 statement 7 */
1709 if (last_sc_fence_thread_before && act->is_write() &&
1710 *act < *last_sc_fence_thread_before) {
1711 added = mo_graph->addEdge(act, curr) || added;
1716 * Include at most one act per-thread that "happens
1719 if (act->happens_before(curr)) {
1721 * Note: if act is RMW, just add edge:
1723 * The following edge should be handled elsewhere:
1724 * readfrom(act) --mo--> act
1726 if (act->is_write())
1727 added = mo_graph->addEdge(act, curr) || added;
1728 else if (act->is_read()) {
1729 //if previous read accessed a null, just keep going
1730 if (act->get_reads_from() == NULL)
1732 added = mo_graph->addEdge(act->get_reads_from(), curr) || added;
1735 } else if (act->is_read() && !act->could_synchronize_with(curr) &&
1736 !act->same_thread(curr)) {
1737 /* We have an action that:
1738 (1) did not happen before us
1739 (2) is a read and we are a write
1740 (3) cannot synchronize with us
1741 (4) is in a different thread
1743 that read could potentially read from our write. Note that
1744 these checks are overly conservative at this point, we'll
1745 do more checks before actually removing the
1749 if (thin_air_constraint_may_allow(curr, act)) {
1750 if (!is_infeasible())
1751 futurevalues->push_back(PendingFutureValue(curr, act));
1752 else if (curr->is_rmw() && act->is_rmw() && curr->get_reads_from() && curr->get_reads_from() == act->get_reads_from())
1753 add_future_value(curr, act);
1760 * All compatible, thread-exclusive promises must be ordered after any
1761 * concrete stores to the same thread, or else they can be merged with
1764 for (unsigned int i = 0; i < promises->size(); i++)
1765 if ((*promises)[i]->is_compatible_exclusive(curr))
1766 added = mo_graph->addEdge(curr, (*promises)[i]) || added;
1771 /** Arbitrary reads from the future are not allowed. Section 29.3
1772 * part 9 places some constraints. This method checks one result of constraint
1773 * constraint. Others require compiler support. */
1774 bool ModelChecker::thin_air_constraint_may_allow(const ModelAction *writer, const ModelAction *reader)
1776 if (!writer->is_rmw())
1779 if (!reader->is_rmw())
1782 for (const ModelAction *search = writer->get_reads_from(); search != NULL; search = search->get_reads_from()) {
1783 if (search == reader)
1785 if (search->get_tid() == reader->get_tid() &&
1786 search->happens_before(reader))
1794 * Arbitrary reads from the future are not allowed. Section 29.3 part 9 places
1795 * some constraints. This method checks one the following constraint (others
1796 * require compiler support):
1798 * If X --hb-> Y --mo-> Z, then X should not read from Z.
1800 bool ModelChecker::mo_may_allow(const ModelAction *writer, const ModelAction *reader)
1802 std::vector<action_list_t> *thrd_lists = get_safe_ptr_vect_action(obj_thrd_map, reader->get_location());
1804 /* Iterate over all threads */
1805 for (i = 0; i < thrd_lists->size(); i++) {
1806 const ModelAction *write_after_read = NULL;
1808 /* Iterate over actions in thread, starting from most recent */
1809 action_list_t *list = &(*thrd_lists)[i];
1810 action_list_t::reverse_iterator rit;
1811 for (rit = list->rbegin(); rit != list->rend(); rit++) {
1812 ModelAction *act = *rit;
1814 /* Don't disallow due to act == reader */
1815 if (!reader->happens_before(act) || reader == act)
1817 else if (act->is_write())
1818 write_after_read = act;
1819 else if (act->is_read() && act->get_reads_from() != NULL)
1820 write_after_read = act->get_reads_from();
1823 if (write_after_read && write_after_read != writer && mo_graph->checkReachable(write_after_read, writer))
1830 * Finds the head(s) of the release sequence(s) containing a given ModelAction.
1831 * The ModelAction under consideration is expected to be taking part in
1832 * release/acquire synchronization as an object of the "reads from" relation.
1833 * Note that this can only provide release sequence support for RMW chains
1834 * which do not read from the future, as those actions cannot be traced until
1835 * their "promise" is fulfilled. Similarly, we may not even establish the
1836 * presence of a release sequence with certainty, as some modification order
1837 * constraints may be decided further in the future. Thus, this function
1838 * "returns" two pieces of data: a pass-by-reference vector of @a release_heads
1839 * and a boolean representing certainty.
1841 * @param rf The action that might be part of a release sequence. Must be a
1843 * @param release_heads A pass-by-reference style return parameter. After
1844 * execution of this function, release_heads will contain the heads of all the
1845 * relevant release sequences, if any exists with certainty
1846 * @param pending A pass-by-reference style return parameter which is only used
1847 * when returning false (i.e., uncertain). Returns most information regarding
1848 * an uncertain release sequence, including any write operations that might
1849 * break the sequence.
1850 * @return true, if the ModelChecker is certain that release_heads is complete;
1853 bool ModelChecker::release_seq_heads(const ModelAction *rf,
1854 rel_heads_list_t *release_heads,
1855 struct release_seq *pending) const
1857 /* Only check for release sequences if there are no cycles */
1858 if (mo_graph->checkForCycles())
1862 ASSERT(rf->is_write());
1864 if (rf->is_release())
1865 release_heads->push_back(rf);
1866 else if (rf->get_last_fence_release())
1867 release_heads->push_back(rf->get_last_fence_release());
1869 break; /* End of RMW chain */
1871 /** @todo Need to be smarter here... In the linux lock
1872 * example, this will run to the beginning of the program for
1874 /** @todo The way to be smarter here is to keep going until 1
1875 * thread has a release preceded by an acquire and you've seen
1878 /* acq_rel RMW is a sufficient stopping condition */
1879 if (rf->is_acquire() && rf->is_release())
1880 return true; /* complete */
1882 rf = rf->get_reads_from();
1885 /* read from future: need to settle this later */
1887 return false; /* incomplete */
1890 if (rf->is_release())
1891 return true; /* complete */
1893 /* else relaxed write
1894 * - check for fence-release in the same thread (29.8, stmt. 3)
1895 * - check modification order for contiguous subsequence
1896 * -> rf must be same thread as release */
1898 const ModelAction *fence_release = rf->get_last_fence_release();
1899 /* Synchronize with a fence-release unconditionally; we don't need to
1900 * find any more "contiguous subsequence..." for it */
1902 release_heads->push_back(fence_release);
1904 int tid = id_to_int(rf->get_tid());
1905 std::vector<action_list_t> *thrd_lists = get_safe_ptr_vect_action(obj_thrd_map, rf->get_location());
1906 action_list_t *list = &(*thrd_lists)[tid];
1907 action_list_t::const_reverse_iterator rit;
1909 /* Find rf in the thread list */
1910 rit = std::find(list->rbegin(), list->rend(), rf);
1911 ASSERT(rit != list->rend());
1913 /* Find the last {write,fence}-release */
1914 for (; rit != list->rend(); rit++) {
1915 if (fence_release && *(*rit) < *fence_release)
1917 if ((*rit)->is_release())
1920 if (rit == list->rend()) {
1921 /* No write-release in this thread */
1922 return true; /* complete */
1923 } else if (fence_release && *(*rit) < *fence_release) {
1924 /* The fence-release is more recent (and so, "stronger") than
1925 * the most recent write-release */
1926 return true; /* complete */
1927 } /* else, need to establish contiguous release sequence */
1928 ModelAction *release = *rit;
1930 ASSERT(rf->same_thread(release));
1932 pending->writes.clear();
1934 bool certain = true;
1935 for (unsigned int i = 0; i < thrd_lists->size(); i++) {
1936 if (id_to_int(rf->get_tid()) == (int)i)
1938 list = &(*thrd_lists)[i];
1940 /* Can we ensure no future writes from this thread may break
1941 * the release seq? */
1942 bool future_ordered = false;
1944 ModelAction *last = get_last_action(int_to_id(i));
1945 Thread *th = get_thread(int_to_id(i));
1946 if ((last && rf->happens_before(last)) ||
1949 future_ordered = true;
1951 ASSERT(!th->is_model_thread() || future_ordered);
1953 for (rit = list->rbegin(); rit != list->rend(); rit++) {
1954 const ModelAction *act = *rit;
1955 /* Reach synchronization -> this thread is complete */
1956 if (act->happens_before(release))
1958 if (rf->happens_before(act)) {
1959 future_ordered = true;
1963 /* Only non-RMW writes can break release sequences */
1964 if (!act->is_write() || act->is_rmw())
1967 /* Check modification order */
1968 if (mo_graph->checkReachable(rf, act)) {
1969 /* rf --mo--> act */
1970 future_ordered = true;
1973 if (mo_graph->checkReachable(act, release))
1974 /* act --mo--> release */
1976 if (mo_graph->checkReachable(release, act) &&
1977 mo_graph->checkReachable(act, rf)) {
1978 /* release --mo-> act --mo--> rf */
1979 return true; /* complete */
1981 /* act may break release sequence */
1982 pending->writes.push_back(act);
1985 if (!future_ordered)
1986 certain = false; /* This thread is uncertain */
1990 release_heads->push_back(release);
1991 pending->writes.clear();
1993 pending->release = release;
2000 * An interface for getting the release sequence head(s) with which a
2001 * given ModelAction must synchronize. This function only returns a non-empty
2002 * result when it can locate a release sequence head with certainty. Otherwise,
2003 * it may mark the internal state of the ModelChecker so that it will handle
2004 * the release sequence at a later time, causing @a acquire to update its
2005 * synchronization at some later point in execution.
2007 * @param acquire The 'acquire' action that may synchronize with a release
2009 * @param read The read action that may read from a release sequence; this may
2010 * be the same as acquire, or else an earlier action in the same thread (i.e.,
2011 * when 'acquire' is a fence-acquire)
2012 * @param release_heads A pass-by-reference return parameter. Will be filled
2013 * with the head(s) of the release sequence(s), if they exists with certainty.
2014 * @see ModelChecker::release_seq_heads
2016 void ModelChecker::get_release_seq_heads(ModelAction *acquire,
2017 ModelAction *read, rel_heads_list_t *release_heads)
2019 const ModelAction *rf = read->get_reads_from();
2020 struct release_seq *sequence = (struct release_seq *)snapshot_calloc(1, sizeof(struct release_seq));
2021 sequence->acquire = acquire;
2022 sequence->read = read;
2024 if (!release_seq_heads(rf, release_heads, sequence)) {
2025 /* add act to 'lazy checking' list */
2026 pending_rel_seqs->push_back(sequence);
2028 snapshot_free(sequence);
2033 * Attempt to resolve all stashed operations that might synchronize with a
2034 * release sequence for a given location. This implements the "lazy" portion of
2035 * determining whether or not a release sequence was contiguous, since not all
2036 * modification order information is present at the time an action occurs.
2038 * @param location The location/object that should be checked for release
2039 * sequence resolutions. A NULL value means to check all locations.
2040 * @param work_queue The work queue to which to add work items as they are
2042 * @return True if any updates occurred (new synchronization, new mo_graph
2045 bool ModelChecker::resolve_release_sequences(void *location, work_queue_t *work_queue)
2047 bool updated = false;
2048 std::vector< struct release_seq *, SnapshotAlloc<struct release_seq *> >::iterator it = pending_rel_seqs->begin();
2049 while (it != pending_rel_seqs->end()) {
2050 struct release_seq *pending = *it;
2051 ModelAction *acquire = pending->acquire;
2052 const ModelAction *read = pending->read;
2054 /* Only resolve sequences on the given location, if provided */
2055 if (location && read->get_location() != location) {
2060 const ModelAction *rf = read->get_reads_from();
2061 rel_heads_list_t release_heads;
2063 complete = release_seq_heads(rf, &release_heads, pending);
2064 for (unsigned int i = 0; i < release_heads.size(); i++) {
2065 if (!acquire->has_synchronized_with(release_heads[i])) {
2066 if (acquire->synchronize_with(release_heads[i]))
2069 set_bad_synchronization();
2074 /* Re-check all pending release sequences */
2075 work_queue->push_back(CheckRelSeqWorkEntry(NULL));
2076 /* Re-check read-acquire for mo_graph edges */
2077 if (acquire->is_read())
2078 work_queue->push_back(MOEdgeWorkEntry(acquire));
2080 /* propagate synchronization to later actions */
2081 action_list_t::reverse_iterator rit = action_trace->rbegin();
2082 for (; (*rit) != acquire; rit++) {
2083 ModelAction *propagate = *rit;
2084 if (acquire->happens_before(propagate)) {
2085 propagate->synchronize_with(acquire);
2086 /* Re-check 'propagate' for mo_graph edges */
2087 work_queue->push_back(MOEdgeWorkEntry(propagate));
2092 it = pending_rel_seqs->erase(it);
2093 snapshot_free(pending);
2099 // If we resolved promises or data races, see if we have realized a data race.
2106 * Performs various bookkeeping operations for the current ModelAction. For
2107 * instance, adds action to the per-object, per-thread action vector and to the
2108 * action trace list of all thread actions.
2110 * @param act is the ModelAction to add.
2112 void ModelChecker::add_action_to_lists(ModelAction *act)
2114 int tid = id_to_int(act->get_tid());
2115 ModelAction *uninit = NULL;
2117 action_list_t *list = get_safe_ptr_action(obj_map, act->get_location());
2118 if (list->empty() && act->is_atomic_var()) {
2119 uninit = new_uninitialized_action(act->get_location());
2120 uninit_id = id_to_int(uninit->get_tid());
2121 list->push_back(uninit);
2123 list->push_back(act);
2125 action_trace->push_back(act);
2127 action_trace->push_front(uninit);
2129 std::vector<action_list_t> *vec = get_safe_ptr_vect_action(obj_thrd_map, act->get_location());
2130 if (tid >= (int)vec->size())
2131 vec->resize(priv->next_thread_id);
2132 (*vec)[tid].push_back(act);
2134 (*vec)[uninit_id].push_front(uninit);
2136 if ((int)thrd_last_action->size() <= tid)
2137 thrd_last_action->resize(get_num_threads());
2138 (*thrd_last_action)[tid] = act;
2140 (*thrd_last_action)[uninit_id] = uninit;
2142 if (act->is_fence() && act->is_release()) {
2143 if ((int)thrd_last_fence_release->size() <= tid)
2144 thrd_last_fence_release->resize(get_num_threads());
2145 (*thrd_last_fence_release)[tid] = act;
2148 if (act->is_wait()) {
2149 void *mutex_loc = (void *) act->get_value();
2150 get_safe_ptr_action(obj_map, mutex_loc)->push_back(act);
2152 std::vector<action_list_t> *vec = get_safe_ptr_vect_action(obj_thrd_map, mutex_loc);
2153 if (tid >= (int)vec->size())
2154 vec->resize(priv->next_thread_id);
2155 (*vec)[tid].push_back(act);
2160 * @brief Get the last action performed by a particular Thread
2161 * @param tid The thread ID of the Thread in question
2162 * @return The last action in the thread
2164 ModelAction * ModelChecker::get_last_action(thread_id_t tid) const
2166 int threadid = id_to_int(tid);
2167 if (threadid < (int)thrd_last_action->size())
2168 return (*thrd_last_action)[id_to_int(tid)];
2174 * @brief Get the last fence release performed by a particular Thread
2175 * @param tid The thread ID of the Thread in question
2176 * @return The last fence release in the thread, if one exists; NULL otherwise
2178 ModelAction * ModelChecker::get_last_fence_release(thread_id_t tid) const
2180 int threadid = id_to_int(tid);
2181 if (threadid < (int)thrd_last_fence_release->size())
2182 return (*thrd_last_fence_release)[id_to_int(tid)];
2188 * Gets the last memory_order_seq_cst write (in the total global sequence)
2189 * performed on a particular object (i.e., memory location), not including the
2191 * @param curr The current ModelAction; also denotes the object location to
2193 * @return The last seq_cst write
2195 ModelAction * ModelChecker::get_last_seq_cst_write(ModelAction *curr) const
2197 void *location = curr->get_location();
2198 action_list_t *list = get_safe_ptr_action(obj_map, location);
2199 /* Find: max({i in dom(S) | seq_cst(t_i) && isWrite(t_i) && samevar(t_i, t)}) */
2200 action_list_t::reverse_iterator rit;
2201 for (rit = list->rbegin(); rit != list->rend(); rit++)
2202 if ((*rit)->is_write() && (*rit)->is_seqcst() && (*rit) != curr)
2208 * Gets the last memory_order_seq_cst fence (in the total global sequence)
2209 * performed in a particular thread, prior to a particular fence.
2210 * @param tid The ID of the thread to check
2211 * @param before_fence The fence from which to begin the search; if NULL, then
2212 * search for the most recent fence in the thread.
2213 * @return The last prior seq_cst fence in the thread, if exists; otherwise, NULL
2215 ModelAction * ModelChecker::get_last_seq_cst_fence(thread_id_t tid, const ModelAction *before_fence) const
2217 /* All fences should have NULL location */
2218 action_list_t *list = get_safe_ptr_action(obj_map, NULL);
2219 action_list_t::reverse_iterator rit = list->rbegin();
2222 for (; rit != list->rend(); rit++)
2223 if (*rit == before_fence)
2226 ASSERT(*rit == before_fence);
2230 for (; rit != list->rend(); rit++)
2231 if ((*rit)->is_fence() && (tid == (*rit)->get_tid()) && (*rit)->is_seqcst())
2237 * Gets the last unlock operation performed on a particular mutex (i.e., memory
2238 * location). This function identifies the mutex according to the current
2239 * action, which is presumed to perform on the same mutex.
2240 * @param curr The current ModelAction; also denotes the object location to
2242 * @return The last unlock operation
2244 ModelAction * ModelChecker::get_last_unlock(ModelAction *curr) const
2246 void *location = curr->get_location();
2247 action_list_t *list = get_safe_ptr_action(obj_map, location);
2248 /* Find: max({i in dom(S) | isUnlock(t_i) && samevar(t_i, t)}) */
2249 action_list_t::reverse_iterator rit;
2250 for (rit = list->rbegin(); rit != list->rend(); rit++)
2251 if ((*rit)->is_unlock() || (*rit)->is_wait())
2256 ModelAction * ModelChecker::get_parent_action(thread_id_t tid) const
2258 ModelAction *parent = get_last_action(tid);
2260 parent = get_thread(tid)->get_creation();
2265 * Returns the clock vector for a given thread.
2266 * @param tid The thread whose clock vector we want
2267 * @return Desired clock vector
2269 ClockVector * ModelChecker::get_cv(thread_id_t tid) const
2271 return get_parent_action(tid)->get_cv();
2275 * Resolve a set of Promises with a current write. The set is provided in the
2276 * Node corresponding to @a write.
2277 * @param write The ModelAction that is fulfilling Promises
2278 * @return True if promises were resolved; false otherwise
2280 bool ModelChecker::resolve_promises(ModelAction *write)
2282 bool haveResolved = false;
2283 std::vector< ModelAction *, ModelAlloc<ModelAction *> > actions_to_check;
2284 promise_list_t mustResolve, resolved;
2286 for (unsigned int i = 0, promise_index = 0; promise_index < promises->size(); i++) {
2287 Promise *promise = (*promises)[promise_index];
2288 if (write->get_node()->get_promise(i)) {
2289 ModelAction *read = promise->get_action();
2290 read_from(read, write);
2291 //Make sure the promise's value matches the write's value
2292 ASSERT(promise->is_compatible(write));
2293 mo_graph->resolvePromise(read, write, &mustResolve);
2295 resolved.push_back(promise);
2296 promises->erase(promises->begin() + promise_index);
2297 actions_to_check.push_back(read);
2299 haveResolved = true;
2304 for (unsigned int i = 0; i < mustResolve.size(); i++) {
2305 if (std::find(resolved.begin(), resolved.end(), mustResolve[i])
2307 priv->failed_promise = true;
2309 for (unsigned int i = 0; i < resolved.size(); i++)
2311 //Check whether reading these writes has made threads unable to
2314 for (unsigned int i = 0; i < actions_to_check.size(); i++) {
2315 ModelAction *read = actions_to_check[i];
2316 mo_check_promises(read, true);
2319 return haveResolved;
2323 * Compute the set of promises that could potentially be satisfied by this
2324 * action. Note that the set computation actually appears in the Node, not in
2326 * @param curr The ModelAction that may satisfy promises
2328 void ModelChecker::compute_promises(ModelAction *curr)
2330 for (unsigned int i = 0; i < promises->size(); i++) {
2331 Promise *promise = (*promises)[i];
2332 const ModelAction *act = promise->get_action();
2333 if (!act->happens_before(curr) &&
2335 !act->could_synchronize_with(curr) &&
2336 !act->same_thread(curr) &&
2337 act->get_location() == curr->get_location() &&
2338 promise->get_value() == curr->get_value()) {
2339 curr->get_node()->set_promise(i, act->is_rmw());
2344 /** Checks promises in response to change in ClockVector Threads. */
2345 void ModelChecker::check_promises(thread_id_t tid, ClockVector *old_cv, ClockVector *merge_cv)
2347 for (unsigned int i = 0; i < promises->size(); i++) {
2348 Promise *promise = (*promises)[i];
2349 const ModelAction *act = promise->get_action();
2350 if ((old_cv == NULL || !old_cv->synchronized_since(act)) &&
2351 merge_cv->synchronized_since(act)) {
2352 if (promise->eliminate_thread(tid)) {
2353 //Promise has failed
2354 priv->failed_promise = true;
2361 void ModelChecker::check_promises_thread_disabled()
2363 for (unsigned int i = 0; i < promises->size(); i++) {
2364 Promise *promise = (*promises)[i];
2365 if (promise->has_failed()) {
2366 priv->failed_promise = true;
2373 * @brief Checks promises in response to addition to modification order for
2376 * We test whether threads are still available for satisfying promises after an
2377 * addition to our modification order constraints. Those that are unavailable
2378 * are "eliminated". Once all threads are eliminated from satisfying a promise,
2379 * that promise has failed.
2381 * @param act The ModelAction which updated the modification order
2382 * @param is_read_check Should be true if act is a read and we must check for
2383 * updates to the store from which it read (there is a distinction here for
2384 * RMW's, which are both a load and a store)
2386 void ModelChecker::mo_check_promises(const ModelAction *act, bool is_read_check)
2388 const ModelAction *write = is_read_check ? act->get_reads_from() : act;
2390 for (unsigned int i = 0; i < promises->size(); i++) {
2391 Promise *promise = (*promises)[i];
2392 const ModelAction *pread = promise->get_action();
2394 // Is this promise on the same location?
2395 if (!pread->same_var(write))
2398 if (pread->happens_before(act) && mo_graph->checkPromise(write, promise)) {
2399 priv->failed_promise = true;
2403 // Don't do any lookups twice for the same thread
2404 if (!promise->thread_is_available(act->get_tid()))
2407 if (mo_graph->checkReachable(promise, write)) {
2408 if (mo_graph->checkPromise(write, promise)) {
2409 priv->failed_promise = true;
2417 * Compute the set of writes that may break the current pending release
2418 * sequence. This information is extracted from previou release sequence
2421 * @param curr The current ModelAction. Must be a release sequence fixup
2424 void ModelChecker::compute_relseq_breakwrites(ModelAction *curr)
2426 if (pending_rel_seqs->empty())
2429 struct release_seq *pending = pending_rel_seqs->back();
2430 for (unsigned int i = 0; i < pending->writes.size(); i++) {
2431 const ModelAction *write = pending->writes[i];
2432 curr->get_node()->add_relseq_break(write);
2435 /* NULL means don't break the sequence; just synchronize */
2436 curr->get_node()->add_relseq_break(NULL);
2440 * Build up an initial set of all past writes that this 'read' action may read
2441 * from. This set is determined by the clock vector's "happens before"
2443 * @param curr is the current ModelAction that we are exploring; it must be a
2446 void ModelChecker::build_reads_from_past(ModelAction *curr)
2448 std::vector<action_list_t> *thrd_lists = get_safe_ptr_vect_action(obj_thrd_map, curr->get_location());
2450 ASSERT(curr->is_read());
2452 ModelAction *last_sc_write = NULL;
2454 if (curr->is_seqcst())
2455 last_sc_write = get_last_seq_cst_write(curr);
2457 /* Iterate over all threads */
2458 for (i = 0; i < thrd_lists->size(); i++) {
2459 /* Iterate over actions in thread, starting from most recent */
2460 action_list_t *list = &(*thrd_lists)[i];
2461 action_list_t::reverse_iterator rit;
2462 for (rit = list->rbegin(); rit != list->rend(); rit++) {
2463 ModelAction *act = *rit;
2465 /* Only consider 'write' actions */
2466 if (!act->is_write() || act == curr)
2469 /* Don't consider more than one seq_cst write if we are a seq_cst read. */
2470 bool allow_read = true;
2472 if (curr->is_seqcst() && (act->is_seqcst() || (last_sc_write != NULL && act->happens_before(last_sc_write))) && act != last_sc_write)
2474 else if (curr->get_sleep_flag() && !curr->is_seqcst() && !sleep_can_read_from(curr, act))
2478 /* Only add feasible reads */
2479 mo_graph->startChanges();
2480 r_modification_order(curr, act);
2481 if (!is_infeasible())
2482 curr->get_node()->add_read_from(act);
2483 mo_graph->rollbackChanges();
2486 /* Include at most one act per-thread that "happens before" curr */
2487 if (act->happens_before(curr))
2491 /* We may find no valid may-read-from only if the execution is doomed */
2492 if (!curr->get_node()->get_read_from_size()) {
2493 priv->no_valid_reads = true;
2497 if (DBG_ENABLED()) {
2498 model_print("Reached read action:\n");
2500 model_print("Printing may_read_from\n");
2501 curr->get_node()->print_may_read_from();
2502 model_print("End printing may_read_from\n");
2506 bool ModelChecker::sleep_can_read_from(ModelAction *curr, const ModelAction *write)
2509 /* UNINIT actions don't have a Node, and they never sleep */
2510 if (write->is_uninitialized())
2512 Node *prevnode = write->get_node()->get_parent();
2514 bool thread_sleep = prevnode->enabled_status(curr->get_tid()) == THREAD_SLEEP_SET;
2515 if (write->is_release() && thread_sleep)
2517 if (!write->is_rmw()) {
2520 if (write->get_reads_from() == NULL)
2522 write = write->get_reads_from();
2527 * @brief Create a new action representing an uninitialized atomic
2528 * @param location The memory location of the atomic object
2529 * @return A pointer to a new ModelAction
2531 ModelAction * ModelChecker::new_uninitialized_action(void *location) const
2533 ModelAction *act = (ModelAction *)snapshot_malloc(sizeof(class ModelAction));
2534 act = new (act) ModelAction(ATOMIC_UNINIT, std::memory_order_relaxed, location, 0, model_thread);
2535 act->create_cv(NULL);
2539 static void print_list(action_list_t *list)
2541 action_list_t::iterator it;
2543 model_print("---------------------------------------------------------------------\n");
2545 unsigned int hash = 0;
2547 for (it = list->begin(); it != list->end(); it++) {
2549 hash = hash^(hash<<3)^((*it)->hash());
2551 model_print("HASH %u\n", hash);
2552 model_print("---------------------------------------------------------------------\n");
2555 #if SUPPORT_MOD_ORDER_DUMP
2556 void ModelChecker::dumpGraph(char *filename) const
2559 sprintf(buffer, "%s.dot", filename);
2560 FILE *file = fopen(buffer, "w");
2561 fprintf(file, "digraph %s {\n", filename);
2562 mo_graph->dumpNodes(file);
2563 ModelAction **thread_array = (ModelAction **)model_calloc(1, sizeof(ModelAction *) * get_num_threads());
2565 for (action_list_t::iterator it = action_trace->begin(); it != action_trace->end(); it++) {
2566 ModelAction *action = *it;
2567 if (action->is_read()) {
2568 fprintf(file, "N%u [label=\"N%u, T%u\"];\n", action->get_seq_number(), action->get_seq_number(), action->get_tid());
2569 if (action->get_reads_from() != NULL)
2570 fprintf(file, "N%u -> N%u[label=\"rf\", color=red];\n", action->get_seq_number(), action->get_reads_from()->get_seq_number());
2572 if (thread_array[action->get_tid()] != NULL) {
2573 fprintf(file, "N%u -> N%u[label=\"sb\", color=blue];\n", thread_array[action->get_tid()]->get_seq_number(), action->get_seq_number());
2576 thread_array[action->get_tid()] = action;
2578 fprintf(file, "}\n");
2579 model_free(thread_array);
2584 /** @brief Prints an execution trace summary. */
2585 void ModelChecker::print_summary() const
2587 #if SUPPORT_MOD_ORDER_DUMP
2588 char buffername[100];
2589 sprintf(buffername, "exec%04u", stats.num_total);
2590 mo_graph->dumpGraphToFile(buffername);
2591 sprintf(buffername, "graph%04u", stats.num_total);
2592 dumpGraph(buffername);
2595 model_print("Execution %d:", stats.num_total);
2596 if (isfeasibleprefix())
2599 print_infeasibility(" INFEASIBLE");
2600 print_list(action_trace);
2605 * Add a Thread to the system for the first time. Should only be called once
2607 * @param t The Thread to add
2609 void ModelChecker::add_thread(Thread *t)
2611 thread_map->put(id_to_int(t->get_id()), t);
2612 scheduler->add_thread(t);
2616 * Removes a thread from the scheduler.
2617 * @param the thread to remove.
2619 void ModelChecker::remove_thread(Thread *t)
2621 scheduler->remove_thread(t);
2625 * @brief Get a Thread reference by its ID
2626 * @param tid The Thread's ID
2627 * @return A Thread reference
2629 Thread * ModelChecker::get_thread(thread_id_t tid) const
2631 return thread_map->get(id_to_int(tid));
2635 * @brief Get a reference to the Thread in which a ModelAction was executed
2636 * @param act The ModelAction
2637 * @return A Thread reference
2639 Thread * ModelChecker::get_thread(const ModelAction *act) const
2641 return get_thread(act->get_tid());
2645 * @brief Check if a Thread is currently enabled
2646 * @param t The Thread to check
2647 * @return True if the Thread is currently enabled
2649 bool ModelChecker::is_enabled(Thread *t) const
2651 return scheduler->is_enabled(t);
2655 * @brief Check if a Thread is currently enabled
2656 * @param tid The ID of the Thread to check
2657 * @return True if the Thread is currently enabled
2659 bool ModelChecker::is_enabled(thread_id_t tid) const
2661 return scheduler->is_enabled(tid);
2665 * Switch from a user-context to the "master thread" context (a.k.a. system
2666 * context). This switch is made with the intention of exploring a particular
2667 * model-checking action (described by a ModelAction object). Must be called
2668 * from a user-thread context.
2670 * @param act The current action that will be explored. May be NULL only if
2671 * trace is exiting via an assertion (see ModelChecker::set_assert and
2672 * ModelChecker::has_asserted).
2673 * @return Return the value returned by the current action
2675 uint64_t ModelChecker::switch_to_master(ModelAction *act)
2678 Thread *old = thread_current();
2679 set_current_action(act);
2680 old->set_state(THREAD_READY);
2681 if (Thread::swap(old, &system_context) < 0) {
2682 perror("swap threads");
2685 return old->get_return_value();
2689 * Takes the next step in the execution, if possible.
2690 * @param curr The current step to take
2691 * @return Returns true (success) if a step was taken and false otherwise.
2693 bool ModelChecker::take_step(ModelAction *curr)
2698 Thread *curr_thrd = get_thread(curr);
2699 ASSERT(curr_thrd->get_state() == THREAD_READY);
2701 curr = check_current_action(curr);
2703 /* Infeasible -> don't take any more steps */
2704 if (is_infeasible())
2706 else if (isfeasibleprefix() && have_bug_reports()) {
2711 if (params.bound != 0)
2712 if (priv->used_sequence_numbers > params.bound)
2715 if (curr_thrd->is_blocked() || curr_thrd->is_complete())
2716 scheduler->remove_thread(curr_thrd);
2718 Thread *next_thrd = get_next_thread(curr);
2719 next_thrd = scheduler->next_thread(next_thrd);
2721 DEBUG("(%d, %d)\n", curr_thrd ? id_to_int(curr_thrd->get_id()) : -1,
2722 next_thrd ? id_to_int(next_thrd->get_id()) : -1);
2725 * Launch end-of-execution release sequence fixups only when there are:
2727 * (1) no more user threads to run (or when execution replay chooses
2728 * the 'model_thread')
2729 * (2) pending release sequences
2730 * (3) pending assertions (i.e., data races)
2731 * (4) no pending promises
2733 if (!pending_rel_seqs->empty() && (!next_thrd || next_thrd->is_model_thread()) &&
2734 is_feasible_prefix_ignore_relseq() && !unrealizedraces.empty()) {
2735 model_print("*** WARNING: release sequence fixup action (%zu pending release seuqences) ***\n",
2736 pending_rel_seqs->size());
2737 ModelAction *fixup = new ModelAction(MODEL_FIXUP_RELSEQ,
2738 std::memory_order_seq_cst, NULL, VALUE_NONE,
2740 set_current_action(fixup);
2744 /* next_thrd == NULL -> don't take any more steps */
2748 next_thrd->set_state(THREAD_RUNNING);
2750 if (next_thrd->get_pending() != NULL) {
2751 /* restart a pending action */
2752 set_current_action(next_thrd->get_pending());
2753 next_thrd->set_pending(NULL);
2754 next_thrd->set_state(THREAD_READY);
2758 /* Return false only if swap fails with an error */
2759 return (Thread::swap(&system_context, next_thrd) == 0);
2762 /** Wrapper to run the user's main function, with appropriate arguments */
2763 void user_main_wrapper(void *)
2765 user_main(model->params.argc, model->params.argv);
2768 /** @brief Run ModelChecker for the user program */
2769 void ModelChecker::run()
2773 Thread *t = new Thread(&user_thread, &user_main_wrapper, NULL);
2777 /* Run user thread up to its first action */
2778 scheduler->next_thread(t);
2779 Thread::swap(&system_context, t);
2781 /* Wait for all threads to complete */
2782 while (take_step(priv->current_action));
2783 } while (next_execution());