1 /* ---------------------------------------------------------------------------
3 * (c) The GHC Team, 1998-2006
5 * The scheduler and thread-related functionality
7 * --------------------------------------------------------------------------*/
9 #include "PosixSource.h"
10 #define KEEP_LOCKCLOSURE
15 #include "OSThreads.h"
20 #include "StgMiscClosures.h"
21 #include "Interpreter.h"
23 #include "RtsSignals.h"
29 #include "ThreadLabels.h"
30 #include "LdvProfile.h"
32 #include "Proftimer.h"
35 /* PARALLEL_HASKELL includes go here */
38 #include "Capability.h"
40 #include "AwaitEvent.h"
41 #if defined(mingw32_HOST_OS)
42 #include "win32/IOManager.h"
45 #include "RaiseAsync.h"
47 #include "ThrIOManager.h"
49 #ifdef HAVE_SYS_TYPES_H
50 #include <sys/types.h>
64 // Turn off inlining when debugging - it obfuscates things
67 # define STATIC_INLINE static
70 /* -----------------------------------------------------------------------------
72 * -------------------------------------------------------------------------- */
74 #if !defined(THREADED_RTS)
75 // Blocked/sleeping thrads
76 StgTSO *blocked_queue_hd = NULL;
77 StgTSO *blocked_queue_tl = NULL;
78 StgTSO *sleeping_queue = NULL; // perhaps replace with a hash table?
81 /* Threads blocked on blackholes.
82 * LOCK: sched_mutex+capability, or all capabilities
84 StgTSO *blackhole_queue = NULL;
86 /* The blackhole_queue should be checked for threads to wake up. See
87 * Schedule.h for more thorough comment.
88 * LOCK: none (doesn't matter if we miss an update)
90 rtsBool blackholes_need_checking = rtsFalse;
92 /* flag that tracks whether we have done any execution in this time slice.
93 * LOCK: currently none, perhaps we should lock (but needs to be
94 * updated in the fast path of the scheduler).
96 nat recent_activity = ACTIVITY_YES;
98 /* if this flag is set as well, give up execution
99 * LOCK: none (changes once, from false->true)
101 rtsBool sched_state = SCHED_RUNNING;
103 /* This is used in `TSO.h' and gcc 2.96 insists that this variable actually
104 * exists - earlier gccs apparently didn't.
110 * Set to TRUE when entering a shutdown state (via shutdownHaskellAndExit()) --
111 * in an MT setting, needed to signal that a worker thread shouldn't hang around
112 * in the scheduler when it is out of work.
114 rtsBool shutting_down_scheduler = rtsFalse;
117 * This mutex protects most of the global scheduler data in
118 * the THREADED_RTS runtime.
120 #if defined(THREADED_RTS)
124 #if !defined(mingw32_HOST_OS)
125 #define FORKPROCESS_PRIMOP_SUPPORTED
128 /* -----------------------------------------------------------------------------
129 * static function prototypes
130 * -------------------------------------------------------------------------- */
132 static Capability *schedule (Capability *initialCapability, Task *task);
135 // These function all encapsulate parts of the scheduler loop, and are
136 // abstracted only to make the structure and control flow of the
137 // scheduler clearer.
139 static void schedulePreLoop (void);
140 static void scheduleFindWork (Capability *cap);
141 #if defined(THREADED_RTS)
142 static void scheduleYield (Capability **pcap, Task *task);
144 static void scheduleStartSignalHandlers (Capability *cap);
145 static void scheduleCheckBlockedThreads (Capability *cap);
146 static void scheduleCheckWakeupThreads(Capability *cap USED_IF_NOT_THREADS);
147 static void scheduleCheckBlackHoles (Capability *cap);
148 static void scheduleDetectDeadlock (Capability *cap, Task *task);
149 static void schedulePushWork(Capability *cap, Task *task);
150 #if defined(PARALLEL_HASKELL)
151 static rtsBool scheduleGetRemoteWork(Capability *cap);
152 static void scheduleSendPendingMessages(void);
154 #if defined(PARALLEL_HASKELL) || defined(THREADED_RTS)
155 static void scheduleActivateSpark(Capability *cap);
157 static void schedulePostRunThread(Capability *cap, StgTSO *t);
158 static rtsBool scheduleHandleHeapOverflow( Capability *cap, StgTSO *t );
159 static void scheduleHandleStackOverflow( Capability *cap, Task *task,
161 static rtsBool scheduleHandleYield( Capability *cap, StgTSO *t,
162 nat prev_what_next );
163 static void scheduleHandleThreadBlocked( StgTSO *t );
164 static rtsBool scheduleHandleThreadFinished( Capability *cap, Task *task,
166 static rtsBool scheduleNeedHeapProfile(rtsBool ready_to_gc);
167 static Capability *scheduleDoGC(Capability *cap, Task *task,
168 rtsBool force_major);
170 static rtsBool checkBlackHoles(Capability *cap);
172 static StgTSO *threadStackOverflow(Capability *cap, StgTSO *tso);
173 static StgTSO *threadStackUnderflow(Task *task, StgTSO *tso);
175 static void deleteThread (Capability *cap, StgTSO *tso);
176 static void deleteAllThreads (Capability *cap);
178 #ifdef FORKPROCESS_PRIMOP_SUPPORTED
179 static void deleteThread_(Capability *cap, StgTSO *tso);
183 static char *whatNext_strs[] = {
193 /* -----------------------------------------------------------------------------
194 * Putting a thread on the run queue: different scheduling policies
195 * -------------------------------------------------------------------------- */
198 addToRunQueue( Capability *cap, StgTSO *t )
200 #if defined(PARALLEL_HASKELL)
201 if (RtsFlags.ParFlags.doFairScheduling) {
202 // this does round-robin scheduling; good for concurrency
203 appendToRunQueue(cap,t);
205 // this does unfair scheduling; good for parallelism
206 pushOnRunQueue(cap,t);
209 // this does round-robin scheduling; good for concurrency
210 appendToRunQueue(cap,t);
214 /* ---------------------------------------------------------------------------
215 Main scheduling loop.
217 We use round-robin scheduling, each thread returning to the
218 scheduler loop when one of these conditions is detected:
221 * timer expires (thread yields)
227 In a GranSim setup this loop iterates over the global event queue.
228 This revolves around the global event queue, which determines what
229 to do next. Therefore, it's more complicated than either the
230 concurrent or the parallel (GUM) setup.
231 This version has been entirely removed (JB 2008/08).
234 GUM iterates over incoming messages.
235 It starts with nothing to do (thus CurrentTSO == END_TSO_QUEUE),
236 and sends out a fish whenever it has nothing to do; in-between
237 doing the actual reductions (shared code below) it processes the
238 incoming messages and deals with delayed operations
239 (see PendingFetches).
240 This is not the ugliest code you could imagine, but it's bloody close.
242 (JB 2008/08) This version was formerly indicated by a PP-Flag PAR,
243 now by PP-flag PARALLEL_HASKELL. The Eden RTS (in GHC-6.x) uses it,
244 as well as future GUM versions. This file has been refurbished to
245 only contain valid code, which is however incomplete, refers to
246 invalid includes etc.
248 ------------------------------------------------------------------------ */
251 schedule (Capability *initialCapability, Task *task)
255 StgThreadReturnCode ret;
256 #if defined(PARALLEL_HASKELL)
257 rtsBool receivedFinish = rtsFalse;
261 #if defined(THREADED_RTS)
262 rtsBool first = rtsTrue;
265 cap = initialCapability;
267 // Pre-condition: this task owns initialCapability.
268 // The sched_mutex is *NOT* held
269 // NB. on return, we still hold a capability.
271 debugTrace (DEBUG_sched,
272 "### NEW SCHEDULER LOOP (task: %p, cap: %p)",
273 task, initialCapability);
277 // -----------------------------------------------------------
278 // Scheduler loop starts here:
280 #if defined(PARALLEL_HASKELL)
281 #define TERMINATION_CONDITION (!receivedFinish)
283 #define TERMINATION_CONDITION rtsTrue
286 while (TERMINATION_CONDITION) {
288 // Check whether we have re-entered the RTS from Haskell without
289 // going via suspendThread()/resumeThread (i.e. a 'safe' foreign
291 if (cap->in_haskell) {
292 errorBelch("schedule: re-entered unsafely.\n"
293 " Perhaps a 'foreign import unsafe' should be 'safe'?");
294 stg_exit(EXIT_FAILURE);
297 // The interruption / shutdown sequence.
299 // In order to cleanly shut down the runtime, we want to:
300 // * make sure that all main threads return to their callers
301 // with the state 'Interrupted'.
302 // * clean up all OS threads assocated with the runtime
303 // * free all memory etc.
305 // So the sequence for ^C goes like this:
307 // * ^C handler sets sched_state := SCHED_INTERRUPTING and
308 // arranges for some Capability to wake up
310 // * all threads in the system are halted, and the zombies are
311 // placed on the run queue for cleaning up. We acquire all
312 // the capabilities in order to delete the threads, this is
313 // done by scheduleDoGC() for convenience (because GC already
314 // needs to acquire all the capabilities). We can't kill
315 // threads involved in foreign calls.
317 // * somebody calls shutdownHaskell(), which calls exitScheduler()
319 // * sched_state := SCHED_SHUTTING_DOWN
321 // * all workers exit when the run queue on their capability
322 // drains. All main threads will also exit when their TSO
323 // reaches the head of the run queue and they can return.
325 // * eventually all Capabilities will shut down, and the RTS can
328 // * We might be left with threads blocked in foreign calls,
329 // we should really attempt to kill these somehow (TODO);
331 switch (sched_state) {
334 case SCHED_INTERRUPTING:
335 debugTrace(DEBUG_sched, "SCHED_INTERRUPTING");
336 #if defined(THREADED_RTS)
337 discardSparksCap(cap);
339 /* scheduleDoGC() deletes all the threads */
340 cap = scheduleDoGC(cap,task,rtsFalse);
342 case SCHED_SHUTTING_DOWN:
343 debugTrace(DEBUG_sched, "SCHED_SHUTTING_DOWN");
344 // If we are a worker, just exit. If we're a bound thread
345 // then we will exit below when we've removed our TSO from
347 if (task->tso == NULL && emptyRunQueue(cap)) {
352 barf("sched_state: %d", sched_state);
355 scheduleFindWork(cap);
357 /* work pushing, currently relevant only for THREADED_RTS:
358 (pushes threads, wakes up idle capabilities for stealing) */
359 schedulePushWork(cap,task);
361 #if defined(PARALLEL_HASKELL)
362 /* since we perform a blocking receive and continue otherwise,
363 either we never reach here or we definitely have work! */
364 // from here: non-empty run queue
365 ASSERT(!emptyRunQueue(cap));
367 if (PacketsWaiting()) { /* now process incoming messages, if any
370 CAUTION: scheduleGetRemoteWork called
371 above, waits for messages as well! */
372 processMessages(cap, &receivedFinish);
374 #endif // PARALLEL_HASKELL: non-empty run queue!
376 scheduleDetectDeadlock(cap,task);
378 #if defined(THREADED_RTS)
379 cap = task->cap; // reload cap, it might have changed
382 // Normally, the only way we can get here with no threads to
383 // run is if a keyboard interrupt received during
384 // scheduleCheckBlockedThreads() or scheduleDetectDeadlock().
385 // Additionally, it is not fatal for the
386 // threaded RTS to reach here with no threads to run.
388 // win32: might be here due to awaitEvent() being abandoned
389 // as a result of a console event having been delivered.
391 #if defined(THREADED_RTS)
395 // // don't yield the first time, we want a chance to run this
396 // // thread for a bit, even if there are others banging at the
399 // ASSERT_FULL_CAPABILITY_INVARIANTS(cap,task);
403 scheduleYield(&cap,task);
404 if (emptyRunQueue(cap)) continue; // look for work again
407 #if !defined(THREADED_RTS) && !defined(mingw32_HOST_OS)
408 if ( emptyRunQueue(cap) ) {
409 ASSERT(sched_state >= SCHED_INTERRUPTING);
414 // Get a thread to run
416 t = popRunQueue(cap);
418 // Sanity check the thread we're about to run. This can be
419 // expensive if there is lots of thread switching going on...
420 IF_DEBUG(sanity,checkTSO(t));
422 #if defined(THREADED_RTS)
423 // Check whether we can run this thread in the current task.
424 // If not, we have to pass our capability to the right task.
426 Task *bound = t->bound;
430 debugTrace(DEBUG_sched,
431 "### Running thread %lu in bound thread", (unsigned long)t->id);
432 // yes, the Haskell thread is bound to the current native thread
434 debugTrace(DEBUG_sched,
435 "### thread %lu bound to another OS thread", (unsigned long)t->id);
436 // no, bound to a different Haskell thread: pass to that thread
437 pushOnRunQueue(cap,t);
441 // The thread we want to run is unbound.
443 debugTrace(DEBUG_sched,
444 "### this OS thread cannot run thread %lu", (unsigned long)t->id);
445 // no, the current native thread is bound to a different
446 // Haskell thread, so pass it to any worker thread
447 pushOnRunQueue(cap,t);
454 /* context switches are initiated by the timer signal, unless
455 * the user specified "context switch as often as possible", with
458 if (RtsFlags.ConcFlags.ctxtSwitchTicks == 0
459 && !emptyThreadQueues(cap)) {
460 cap->context_switch = 1;
465 // CurrentTSO is the thread to run. t might be different if we
466 // loop back to run_thread, so make sure to set CurrentTSO after
468 cap->r.rCurrentTSO = t;
470 debugTrace(DEBUG_sched, "-->> running thread %ld %s ...",
471 (long)t->id, whatNext_strs[t->what_next]);
473 startHeapProfTimer();
475 // Check for exceptions blocked on this thread
476 maybePerformBlockedException (cap, t);
478 // ----------------------------------------------------------------------
479 // Run the current thread
481 ASSERT_FULL_CAPABILITY_INVARIANTS(cap,task);
482 ASSERT(t->cap == cap);
483 ASSERT(t->bound ? t->bound->cap == cap : 1);
485 prev_what_next = t->what_next;
487 errno = t->saved_errno;
489 SetLastError(t->saved_winerror);
492 cap->in_haskell = rtsTrue;
496 #if defined(THREADED_RTS)
497 if (recent_activity == ACTIVITY_DONE_GC) {
498 // ACTIVITY_DONE_GC means we turned off the timer signal to
499 // conserve power (see #1623). Re-enable it here.
501 prev = xchg((P_)&recent_activity, ACTIVITY_YES);
502 if (prev == ACTIVITY_DONE_GC) {
506 recent_activity = ACTIVITY_YES;
510 switch (prev_what_next) {
514 /* Thread already finished, return to scheduler. */
515 ret = ThreadFinished;
521 r = StgRun((StgFunPtr) stg_returnToStackTop, &cap->r);
522 cap = regTableToCapability(r);
527 case ThreadInterpret:
528 cap = interpretBCO(cap);
533 barf("schedule: invalid what_next field");
536 cap->in_haskell = rtsFalse;
538 // The TSO might have moved, eg. if it re-entered the RTS and a GC
539 // happened. So find the new location:
540 t = cap->r.rCurrentTSO;
542 // We have run some Haskell code: there might be blackhole-blocked
543 // threads to wake up now.
544 // Lock-free test here should be ok, we're just setting a flag.
545 if ( blackhole_queue != END_TSO_QUEUE ) {
546 blackholes_need_checking = rtsTrue;
549 // And save the current errno in this thread.
550 // XXX: possibly bogus for SMP because this thread might already
551 // be running again, see code below.
552 t->saved_errno = errno;
554 // Similarly for Windows error code
555 t->saved_winerror = GetLastError();
558 #if defined(THREADED_RTS)
559 // If ret is ThreadBlocked, and this Task is bound to the TSO that
560 // blocked, we are in limbo - the TSO is now owned by whatever it
561 // is blocked on, and may in fact already have been woken up,
562 // perhaps even on a different Capability. It may be the case
563 // that task->cap != cap. We better yield this Capability
564 // immediately and return to normaility.
565 if (ret == ThreadBlocked) {
566 debugTrace(DEBUG_sched,
567 "--<< thread %lu (%s) stopped: blocked",
568 (unsigned long)t->id, whatNext_strs[t->what_next]);
573 ASSERT_FULL_CAPABILITY_INVARIANTS(cap,task);
574 ASSERT(t->cap == cap);
576 // ----------------------------------------------------------------------
578 // Costs for the scheduler are assigned to CCS_SYSTEM
580 #if defined(PROFILING)
584 schedulePostRunThread(cap,t);
586 t = threadStackUnderflow(task,t);
588 ready_to_gc = rtsFalse;
592 ready_to_gc = scheduleHandleHeapOverflow(cap,t);
596 scheduleHandleStackOverflow(cap,task,t);
600 if (scheduleHandleYield(cap, t, prev_what_next)) {
601 // shortcut for switching between compiler/interpreter:
607 scheduleHandleThreadBlocked(t);
611 if (scheduleHandleThreadFinished(cap, task, t)) return cap;
612 ASSERT_FULL_CAPABILITY_INVARIANTS(cap,task);
616 barf("schedule: invalid thread return code %d", (int)ret);
619 if (ready_to_gc || scheduleNeedHeapProfile(ready_to_gc)) {
620 cap = scheduleDoGC(cap,task,rtsFalse);
622 } /* end of while() */
625 /* ----------------------------------------------------------------------------
626 * Setting up the scheduler loop
627 * ------------------------------------------------------------------------- */
630 schedulePreLoop(void)
632 // initialisation for scheduler - what cannot go into initScheduler()
635 /* -----------------------------------------------------------------------------
638 * Search for work to do, and handle messages from elsewhere.
639 * -------------------------------------------------------------------------- */
642 scheduleFindWork (Capability *cap)
644 scheduleStartSignalHandlers(cap);
646 // Only check the black holes here if we've nothing else to do.
647 // During normal execution, the black hole list only gets checked
648 // at GC time, to avoid repeatedly traversing this possibly long
649 // list each time around the scheduler.
650 if (emptyRunQueue(cap)) { scheduleCheckBlackHoles(cap); }
652 scheduleCheckWakeupThreads(cap);
654 scheduleCheckBlockedThreads(cap);
656 #if defined(THREADED_RTS) || defined(PARALLEL_HASKELL)
657 // Try to activate one of our own sparks
658 if (emptyRunQueue(cap)) { scheduleActivateSpark(cap); }
661 #if defined(THREADED_RTS)
662 // Try to steak work if we don't have any
663 if (emptyRunQueue(cap)) { stealWork(cap); }
666 #if defined(PARALLEL_HASKELL)
667 // if messages have been buffered...
668 scheduleSendPendingMessages();
671 #if defined(PARALLEL_HASKELL)
672 if (emptyRunQueue(cap)) {
673 receivedFinish = scheduleGetRemoteWork(cap);
674 continue; // a new round, (hopefully) with new work
676 in GUM, this a) sends out a FISH and returns IF no fish is
678 b) (blocking) awaits and receives messages
680 in Eden, this is only the blocking receive, as b) in GUM.
686 #if defined(THREADED_RTS)
687 STATIC_INLINE rtsBool
688 shouldYieldCapability (Capability *cap, Task *task)
690 // we need to yield this capability to someone else if..
691 // - another thread is initiating a GC
692 // - another Task is returning from a foreign call
693 // - the thread at the head of the run queue cannot be run
694 // by this Task (it is bound to another Task, or it is unbound
695 // and this task it bound).
696 return (waiting_for_gc ||
697 cap->returning_tasks_hd != NULL ||
698 (!emptyRunQueue(cap) && (task->tso == NULL
699 ? cap->run_queue_hd->bound != NULL
700 : cap->run_queue_hd->bound != task)));
703 // This is the single place where a Task goes to sleep. There are
704 // two reasons it might need to sleep:
705 // - there are no threads to run
706 // - we need to yield this Capability to someone else
707 // (see shouldYieldCapability())
709 // The return value indicates whether
712 scheduleYield (Capability **pcap, Task *task)
714 Capability *cap = *pcap;
716 // if we have work, and we don't need to give up the Capability, continue.
717 if (!shouldYieldCapability(cap,task) &&
718 (!emptyRunQueue(cap) || blackholes_need_checking))
721 // otherwise yield (sleep), and keep yielding if necessary.
723 yieldCapability(&cap,task);
725 while (shouldYieldCapability(cap,task));
727 // note there may still be no threads on the run queue at this
728 // point, the caller has to check.
735 /* -----------------------------------------------------------------------------
738 * Push work to other Capabilities if we have some.
739 * -------------------------------------------------------------------------- */
742 schedulePushWork(Capability *cap USED_IF_THREADS,
743 Task *task USED_IF_THREADS)
745 /* following code not for PARALLEL_HASKELL. I kept the call general,
746 future GUM versions might use pushing in a distributed setup */
747 #if defined(THREADED_RTS)
749 Capability *free_caps[n_capabilities], *cap0;
752 // migration can be turned off with +RTS -qg
753 if (!RtsFlags.ParFlags.migrate) return;
755 // Check whether we have more threads on our run queue, or sparks
756 // in our pool, that we could hand to another Capability.
757 if ((emptyRunQueue(cap) || cap->run_queue_hd->_link == END_TSO_QUEUE)
758 && sparkPoolSizeCap(cap) < 2) {
762 // First grab as many free Capabilities as we can.
763 for (i=0, n_free_caps=0; i < n_capabilities; i++) {
764 cap0 = &capabilities[i];
765 if (cap != cap0 && tryGrabCapability(cap0,task)) {
766 if (!emptyRunQueue(cap0) || cap->returning_tasks_hd != NULL) {
767 // it already has some work, we just grabbed it at
768 // the wrong moment. Or maybe it's deadlocked!
769 releaseCapability(cap0);
771 free_caps[n_free_caps++] = cap0;
776 // we now have n_free_caps free capabilities stashed in
777 // free_caps[]. Share our run queue equally with them. This is
778 // probably the simplest thing we could do; improvements we might
779 // want to do include:
781 // - giving high priority to moving relatively new threads, on
782 // the gournds that they haven't had time to build up a
783 // working set in the cache on this CPU/Capability.
785 // - giving low priority to moving long-lived threads
787 if (n_free_caps > 0) {
788 StgTSO *prev, *t, *next;
789 rtsBool pushed_to_all;
791 debugTrace(DEBUG_sched,
792 "cap %d: %s and %d free capabilities, sharing...",
794 (!emptyRunQueue(cap) && cap->run_queue_hd->_link != END_TSO_QUEUE)?
795 "excess threads on run queue":"sparks to share (>=2)",
799 pushed_to_all = rtsFalse;
801 if (cap->run_queue_hd != END_TSO_QUEUE) {
802 prev = cap->run_queue_hd;
804 prev->_link = END_TSO_QUEUE;
805 for (; t != END_TSO_QUEUE; t = next) {
807 t->_link = END_TSO_QUEUE;
808 if (t->what_next == ThreadRelocated
809 || t->bound == task // don't move my bound thread
810 || tsoLocked(t)) { // don't move a locked thread
811 setTSOLink(cap, prev, t);
813 } else if (i == n_free_caps) {
814 pushed_to_all = rtsTrue;
817 setTSOLink(cap, prev, t);
820 debugTrace(DEBUG_sched, "pushing thread %lu to capability %d", (unsigned long)t->id, free_caps[i]->no);
821 appendToRunQueue(free_caps[i],t);
822 if (t->bound) { t->bound->cap = free_caps[i]; }
823 t->cap = free_caps[i];
827 cap->run_queue_tl = prev;
831 /* JB I left this code in place, it would work but is not necessary */
833 // If there are some free capabilities that we didn't push any
834 // threads to, then try to push a spark to each one.
835 if (!pushed_to_all) {
837 // i is the next free capability to push to
838 for (; i < n_free_caps; i++) {
839 if (emptySparkPoolCap(free_caps[i])) {
840 spark = tryStealSpark(cap->sparks);
842 debugTrace(DEBUG_sched, "pushing spark %p to capability %d", spark, free_caps[i]->no);
843 newSpark(&(free_caps[i]->r), spark);
848 #endif /* SPARK_PUSHING */
850 // release the capabilities
851 for (i = 0; i < n_free_caps; i++) {
852 task->cap = free_caps[i];
853 releaseAndWakeupCapability(free_caps[i]);
856 task->cap = cap; // reset to point to our Capability.
858 #endif /* THREADED_RTS */
862 /* ----------------------------------------------------------------------------
863 * Start any pending signal handlers
864 * ------------------------------------------------------------------------- */
866 #if defined(RTS_USER_SIGNALS) && !defined(THREADED_RTS)
868 scheduleStartSignalHandlers(Capability *cap)
870 if (RtsFlags.MiscFlags.install_signal_handlers && signals_pending()) {
871 // safe outside the lock
872 startSignalHandlers(cap);
877 scheduleStartSignalHandlers(Capability *cap STG_UNUSED)
882 /* ----------------------------------------------------------------------------
883 * Check for blocked threads that can be woken up.
884 * ------------------------------------------------------------------------- */
887 scheduleCheckBlockedThreads(Capability *cap USED_IF_NOT_THREADS)
889 #if !defined(THREADED_RTS)
891 // Check whether any waiting threads need to be woken up. If the
892 // run queue is empty, and there are no other tasks running, we
893 // can wait indefinitely for something to happen.
895 if ( !emptyQueue(blocked_queue_hd) || !emptyQueue(sleeping_queue) )
897 awaitEvent( emptyRunQueue(cap) && !blackholes_need_checking );
903 /* ----------------------------------------------------------------------------
904 * Check for threads woken up by other Capabilities
905 * ------------------------------------------------------------------------- */
908 scheduleCheckWakeupThreads(Capability *cap USED_IF_THREADS)
910 #if defined(THREADED_RTS)
911 // Any threads that were woken up by other Capabilities get
912 // appended to our run queue.
913 if (!emptyWakeupQueue(cap)) {
914 ACQUIRE_LOCK(&cap->lock);
915 if (emptyRunQueue(cap)) {
916 cap->run_queue_hd = cap->wakeup_queue_hd;
917 cap->run_queue_tl = cap->wakeup_queue_tl;
919 setTSOLink(cap, cap->run_queue_tl, cap->wakeup_queue_hd);
920 cap->run_queue_tl = cap->wakeup_queue_tl;
922 cap->wakeup_queue_hd = cap->wakeup_queue_tl = END_TSO_QUEUE;
923 RELEASE_LOCK(&cap->lock);
928 /* ----------------------------------------------------------------------------
929 * Check for threads blocked on BLACKHOLEs that can be woken up
930 * ------------------------------------------------------------------------- */
932 scheduleCheckBlackHoles (Capability *cap)
934 if ( blackholes_need_checking ) // check without the lock first
936 ACQUIRE_LOCK(&sched_mutex);
937 if ( blackholes_need_checking ) {
938 blackholes_need_checking = rtsFalse;
939 // important that we reset the flag *before* checking the
940 // blackhole queue, otherwise we could get deadlock. This
941 // happens as follows: we wake up a thread that
942 // immediately runs on another Capability, blocks on a
943 // blackhole, and then we reset the blackholes_need_checking flag.
944 checkBlackHoles(cap);
946 RELEASE_LOCK(&sched_mutex);
950 /* ----------------------------------------------------------------------------
951 * Detect deadlock conditions and attempt to resolve them.
952 * ------------------------------------------------------------------------- */
955 scheduleDetectDeadlock (Capability *cap, Task *task)
958 #if defined(PARALLEL_HASKELL)
959 // ToDo: add deadlock detection in GUM (similar to THREADED_RTS) -- HWL
964 * Detect deadlock: when we have no threads to run, there are no
965 * threads blocked, waiting for I/O, or sleeping, and all the
966 * other tasks are waiting for work, we must have a deadlock of
969 if ( emptyThreadQueues(cap) )
971 #if defined(THREADED_RTS)
973 * In the threaded RTS, we only check for deadlock if there
974 * has been no activity in a complete timeslice. This means
975 * we won't eagerly start a full GC just because we don't have
976 * any threads to run currently.
978 if (recent_activity != ACTIVITY_INACTIVE) return;
981 debugTrace(DEBUG_sched, "deadlocked, forcing major GC...");
983 // Garbage collection can release some new threads due to
984 // either (a) finalizers or (b) threads resurrected because
985 // they are unreachable and will therefore be sent an
986 // exception. Any threads thus released will be immediately
988 cap = scheduleDoGC (cap, task, rtsTrue/*force major GC*/);
990 recent_activity = ACTIVITY_DONE_GC;
991 // disable timer signals (see #1623)
994 if ( !emptyRunQueue(cap) ) return;
996 #if defined(RTS_USER_SIGNALS) && !defined(THREADED_RTS)
997 /* If we have user-installed signal handlers, then wait
998 * for signals to arrive rather then bombing out with a
1001 if ( RtsFlags.MiscFlags.install_signal_handlers && anyUserHandlers() ) {
1002 debugTrace(DEBUG_sched,
1003 "still deadlocked, waiting for signals...");
1007 if (signals_pending()) {
1008 startSignalHandlers(cap);
1011 // either we have threads to run, or we were interrupted:
1012 ASSERT(!emptyRunQueue(cap) || sched_state >= SCHED_INTERRUPTING);
1018 #if !defined(THREADED_RTS)
1019 /* Probably a real deadlock. Send the current main thread the
1020 * Deadlock exception.
1023 switch (task->tso->why_blocked) {
1025 case BlockedOnBlackHole:
1026 case BlockedOnException:
1028 throwToSingleThreaded(cap, task->tso,
1029 (StgClosure *)nonTermination_closure);
1032 barf("deadlock: main thread blocked in a strange way");
1041 /* ----------------------------------------------------------------------------
1042 * Send pending messages (PARALLEL_HASKELL only)
1043 * ------------------------------------------------------------------------- */
1045 #if defined(PARALLEL_HASKELL)
1047 scheduleSendPendingMessages(void)
1050 # if defined(PAR) // global Mem.Mgmt., omit for now
1051 if (PendingFetches != END_BF_QUEUE) {
1056 if (RtsFlags.ParFlags.BufferTime) {
1057 // if we use message buffering, we must send away all message
1058 // packets which have become too old...
1064 /* ----------------------------------------------------------------------------
1065 * Activate spark threads (PARALLEL_HASKELL and THREADED_RTS)
1066 * ------------------------------------------------------------------------- */
1068 #if defined(PARALLEL_HASKELL) || defined(THREADED_RTS)
1070 scheduleActivateSpark(Capability *cap)
1074 /* We only want to stay here if the run queue is empty and we want some
1075 work. We try to turn a spark into a thread, and add it to the run
1076 queue, from where it will be picked up in the next iteration of the
1079 if (!emptyRunQueue(cap))
1080 /* In the threaded RTS, another task might have pushed a thread
1081 on our run queue in the meantime ? But would need a lock.. */
1085 // Really we should be using reclaimSpark() here, but
1086 // experimentally it doesn't seem to perform as well as just
1087 // stealing from our own spark pool:
1088 // spark = reclaimSpark(cap->sparks);
1089 spark = tryStealSpark(cap->sparks); // defined in Sparks.c
1091 if (spark != NULL) {
1092 debugTrace(DEBUG_sched,
1093 "turning spark of closure %p into a thread",
1094 (StgClosure *)spark);
1095 createSparkThread(cap,spark); // defined in Sparks.c
1098 #endif // PARALLEL_HASKELL || THREADED_RTS
1100 /* ----------------------------------------------------------------------------
1101 * Get work from a remote node (PARALLEL_HASKELL only)
1102 * ------------------------------------------------------------------------- */
1104 #if defined(PARALLEL_HASKELL)
1105 static rtsBool /* return value used in PARALLEL_HASKELL only */
1106 scheduleGetRemoteWork (Capability *cap STG_UNUSED)
1108 #if defined(PARALLEL_HASKELL)
1109 rtsBool receivedFinish = rtsFalse;
1111 // idle() , i.e. send all buffers, wait for work
1112 if (RtsFlags.ParFlags.BufferTime) {
1113 IF_PAR_DEBUG(verbose,
1114 debugBelch("...send all pending data,"));
1117 for (i=1; i<=nPEs; i++)
1118 sendImmediately(i); // send all messages away immediately
1122 /* this would be the place for fishing in GUM...
1124 if (no-earlier-fish-around)
1125 sendFish(choosePe());
1128 // Eden:just look for incoming messages (blocking receive)
1129 IF_PAR_DEBUG(verbose,
1130 debugBelch("...wait for incoming messages...\n"));
1131 processMessages(cap, &receivedFinish); // blocking receive...
1134 return receivedFinish;
1135 // reenter scheduling look after having received something
1137 #else /* !PARALLEL_HASKELL, i.e. THREADED_RTS */
1139 return rtsFalse; /* return value unused in THREADED_RTS */
1141 #endif /* PARALLEL_HASKELL */
1143 #endif // PARALLEL_HASKELL || THREADED_RTS
1145 /* ----------------------------------------------------------------------------
1146 * After running a thread...
1147 * ------------------------------------------------------------------------- */
1150 schedulePostRunThread (Capability *cap, StgTSO *t)
1152 // We have to be able to catch transactions that are in an
1153 // infinite loop as a result of seeing an inconsistent view of
1157 // [a,b] <- mapM readTVar [ta,tb]
1158 // when (a == b) loop
1160 // and a is never equal to b given a consistent view of memory.
1162 if (t -> trec != NO_TREC && t -> why_blocked == NotBlocked) {
1163 if (!stmValidateNestOfTransactions (t -> trec)) {
1164 debugTrace(DEBUG_sched | DEBUG_stm,
1165 "trec %p found wasting its time", t);
1167 // strip the stack back to the
1168 // ATOMICALLY_FRAME, aborting the (nested)
1169 // transaction, and saving the stack of any
1170 // partially-evaluated thunks on the heap.
1171 throwToSingleThreaded_(cap, t, NULL, rtsTrue, NULL);
1173 ASSERT(get_itbl((StgClosure *)t->sp)->type == ATOMICALLY_FRAME);
1177 /* some statistics gathering in the parallel case */
1180 /* -----------------------------------------------------------------------------
1181 * Handle a thread that returned to the scheduler with ThreadHeepOverflow
1182 * -------------------------------------------------------------------------- */
1185 scheduleHandleHeapOverflow( Capability *cap, StgTSO *t )
1187 // did the task ask for a large block?
1188 if (cap->r.rHpAlloc > BLOCK_SIZE) {
1189 // if so, get one and push it on the front of the nursery.
1193 blocks = (lnat)BLOCK_ROUND_UP(cap->r.rHpAlloc) / BLOCK_SIZE;
1195 debugTrace(DEBUG_sched,
1196 "--<< thread %ld (%s) stopped: requesting a large block (size %ld)\n",
1197 (long)t->id, whatNext_strs[t->what_next], blocks);
1199 // don't do this if the nursery is (nearly) full, we'll GC first.
1200 if (cap->r.rCurrentNursery->link != NULL ||
1201 cap->r.rNursery->n_blocks == 1) { // paranoia to prevent infinite loop
1202 // if the nursery has only one block.
1205 bd = allocGroup( blocks );
1207 cap->r.rNursery->n_blocks += blocks;
1209 // link the new group into the list
1210 bd->link = cap->r.rCurrentNursery;
1211 bd->u.back = cap->r.rCurrentNursery->u.back;
1212 if (cap->r.rCurrentNursery->u.back != NULL) {
1213 cap->r.rCurrentNursery->u.back->link = bd;
1215 #if !defined(THREADED_RTS)
1216 ASSERT(g0s0->blocks == cap->r.rCurrentNursery &&
1217 g0s0 == cap->r.rNursery);
1219 cap->r.rNursery->blocks = bd;
1221 cap->r.rCurrentNursery->u.back = bd;
1223 // initialise it as a nursery block. We initialise the
1224 // step, gen_no, and flags field of *every* sub-block in
1225 // this large block, because this is easier than making
1226 // sure that we always find the block head of a large
1227 // block whenever we call Bdescr() (eg. evacuate() and
1228 // isAlive() in the GC would both have to do this, at
1232 for (x = bd; x < bd + blocks; x++) {
1233 x->step = cap->r.rNursery;
1239 // This assert can be a killer if the app is doing lots
1240 // of large block allocations.
1241 IF_DEBUG(sanity, checkNurserySanity(cap->r.rNursery));
1243 // now update the nursery to point to the new block
1244 cap->r.rCurrentNursery = bd;
1246 // we might be unlucky and have another thread get on the
1247 // run queue before us and steal the large block, but in that
1248 // case the thread will just end up requesting another large
1250 pushOnRunQueue(cap,t);
1251 return rtsFalse; /* not actually GC'ing */
1255 debugTrace(DEBUG_sched,
1256 "--<< thread %ld (%s) stopped: HeapOverflow",
1257 (long)t->id, whatNext_strs[t->what_next]);
1259 if (cap->context_switch) {
1260 // Sometimes we miss a context switch, e.g. when calling
1261 // primitives in a tight loop, MAYBE_GC() doesn't check the
1262 // context switch flag, and we end up waiting for a GC.
1263 // See #1984, and concurrent/should_run/1984
1264 cap->context_switch = 0;
1265 addToRunQueue(cap,t);
1267 pushOnRunQueue(cap,t);
1270 /* actual GC is done at the end of the while loop in schedule() */
1273 /* -----------------------------------------------------------------------------
1274 * Handle a thread that returned to the scheduler with ThreadStackOverflow
1275 * -------------------------------------------------------------------------- */
1278 scheduleHandleStackOverflow (Capability *cap, Task *task, StgTSO *t)
1280 debugTrace (DEBUG_sched,
1281 "--<< thread %ld (%s) stopped, StackOverflow",
1282 (long)t->id, whatNext_strs[t->what_next]);
1284 /* just adjust the stack for this thread, then pop it back
1288 /* enlarge the stack */
1289 StgTSO *new_t = threadStackOverflow(cap, t);
1291 /* The TSO attached to this Task may have moved, so update the
1294 if (task->tso == t) {
1297 pushOnRunQueue(cap,new_t);
1301 /* -----------------------------------------------------------------------------
1302 * Handle a thread that returned to the scheduler with ThreadYielding
1303 * -------------------------------------------------------------------------- */
1306 scheduleHandleYield( Capability *cap, StgTSO *t, nat prev_what_next )
1308 // Reset the context switch flag. We don't do this just before
1309 // running the thread, because that would mean we would lose ticks
1310 // during GC, which can lead to unfair scheduling (a thread hogs
1311 // the CPU because the tick always arrives during GC). This way
1312 // penalises threads that do a lot of allocation, but that seems
1313 // better than the alternative.
1314 cap->context_switch = 0;
1316 /* put the thread back on the run queue. Then, if we're ready to
1317 * GC, check whether this is the last task to stop. If so, wake
1318 * up the GC thread. getThread will block during a GC until the
1322 if (t->what_next != prev_what_next) {
1323 debugTrace(DEBUG_sched,
1324 "--<< thread %ld (%s) stopped to switch evaluators",
1325 (long)t->id, whatNext_strs[t->what_next]);
1327 debugTrace(DEBUG_sched,
1328 "--<< thread %ld (%s) stopped, yielding",
1329 (long)t->id, whatNext_strs[t->what_next]);
1334 //debugBelch("&& Doing sanity check on yielding TSO %ld.", t->id);
1336 ASSERT(t->_link == END_TSO_QUEUE);
1338 // Shortcut if we're just switching evaluators: don't bother
1339 // doing stack squeezing (which can be expensive), just run the
1341 if (t->what_next != prev_what_next) {
1345 addToRunQueue(cap,t);
1350 /* -----------------------------------------------------------------------------
1351 * Handle a thread that returned to the scheduler with ThreadBlocked
1352 * -------------------------------------------------------------------------- */
1355 scheduleHandleThreadBlocked( StgTSO *t
1356 #if !defined(GRAN) && !defined(DEBUG)
1362 // We don't need to do anything. The thread is blocked, and it
1363 // has tidied up its stack and placed itself on whatever queue
1364 // it needs to be on.
1366 // ASSERT(t->why_blocked != NotBlocked);
1367 // Not true: for example,
1368 // - in THREADED_RTS, the thread may already have been woken
1369 // up by another Capability. This actually happens: try
1370 // conc023 +RTS -N2.
1371 // - the thread may have woken itself up already, because
1372 // threadPaused() might have raised a blocked throwTo
1373 // exception, see maybePerformBlockedException().
1376 if (traceClass(DEBUG_sched)) {
1377 debugTraceBegin("--<< thread %lu (%s) stopped: ",
1378 (unsigned long)t->id, whatNext_strs[t->what_next]);
1379 printThreadBlockage(t);
1385 /* -----------------------------------------------------------------------------
1386 * Handle a thread that returned to the scheduler with ThreadFinished
1387 * -------------------------------------------------------------------------- */
1390 scheduleHandleThreadFinished (Capability *cap STG_UNUSED, Task *task, StgTSO *t)
1392 /* Need to check whether this was a main thread, and if so,
1393 * return with the return value.
1395 * We also end up here if the thread kills itself with an
1396 * uncaught exception, see Exception.cmm.
1398 debugTrace(DEBUG_sched, "--++ thread %lu (%s) finished",
1399 (unsigned long)t->id, whatNext_strs[t->what_next]);
1402 // Check whether the thread that just completed was a bound
1403 // thread, and if so return with the result.
1405 // There is an assumption here that all thread completion goes
1406 // through this point; we need to make sure that if a thread
1407 // ends up in the ThreadKilled state, that it stays on the run
1408 // queue so it can be dealt with here.
1413 if (t->bound != task) {
1414 #if !defined(THREADED_RTS)
1415 // Must be a bound thread that is not the topmost one. Leave
1416 // it on the run queue until the stack has unwound to the
1417 // point where we can deal with this. Leaving it on the run
1418 // queue also ensures that the garbage collector knows about
1419 // this thread and its return value (it gets dropped from the
1420 // step->threads list so there's no other way to find it).
1421 appendToRunQueue(cap,t);
1424 // this cannot happen in the threaded RTS, because a
1425 // bound thread can only be run by the appropriate Task.
1426 barf("finished bound thread that isn't mine");
1430 ASSERT(task->tso == t);
1432 if (t->what_next == ThreadComplete) {
1434 // NOTE: return val is tso->sp[1] (see StgStartup.hc)
1435 *(task->ret) = (StgClosure *)task->tso->sp[1];
1437 task->stat = Success;
1440 *(task->ret) = NULL;
1442 if (sched_state >= SCHED_INTERRUPTING) {
1443 task->stat = Interrupted;
1445 task->stat = Killed;
1449 removeThreadLabel((StgWord)task->tso->id);
1451 return rtsTrue; // tells schedule() to return
1457 /* -----------------------------------------------------------------------------
1458 * Perform a heap census
1459 * -------------------------------------------------------------------------- */
1462 scheduleNeedHeapProfile( rtsBool ready_to_gc STG_UNUSED )
1464 // When we have +RTS -i0 and we're heap profiling, do a census at
1465 // every GC. This lets us get repeatable runs for debugging.
1466 if (performHeapProfile ||
1467 (RtsFlags.ProfFlags.profileInterval==0 &&
1468 RtsFlags.ProfFlags.doHeapProfile && ready_to_gc)) {
1475 /* -----------------------------------------------------------------------------
1476 * Perform a garbage collection if necessary
1477 * -------------------------------------------------------------------------- */
1480 scheduleDoGC (Capability *cap, Task *task USED_IF_THREADS, rtsBool force_major)
1482 rtsBool heap_census;
1484 /* extern static volatile StgWord waiting_for_gc;
1485 lives inside capability.c */
1486 rtsBool was_waiting;
1491 // In order to GC, there must be no threads running Haskell code.
1492 // Therefore, the GC thread needs to hold *all* the capabilities,
1493 // and release them after the GC has completed.
1495 // This seems to be the simplest way: previous attempts involved
1496 // making all the threads with capabilities give up their
1497 // capabilities and sleep except for the *last* one, which
1498 // actually did the GC. But it's quite hard to arrange for all
1499 // the other tasks to sleep and stay asleep.
1502 /* Other capabilities are prevented from running yet more Haskell
1503 threads if waiting_for_gc is set. Tested inside
1504 yieldCapability() and releaseCapability() in Capability.c */
1506 was_waiting = cas(&waiting_for_gc, 0, 1);
1509 debugTrace(DEBUG_sched, "someone else is trying to GC...");
1510 if (cap) yieldCapability(&cap,task);
1511 } while (waiting_for_gc);
1512 return cap; // NOTE: task->cap might have changed here
1515 setContextSwitches();
1516 for (i=0; i < n_capabilities; i++) {
1517 debugTrace(DEBUG_sched, "ready_to_gc, grabbing all the capabilies (%d/%d)", i, n_capabilities);
1518 if (cap != &capabilities[i]) {
1519 Capability *pcap = &capabilities[i];
1520 // we better hope this task doesn't get migrated to
1521 // another Capability while we're waiting for this one.
1522 // It won't, because load balancing happens while we have
1523 // all the Capabilities, but even so it's a slightly
1524 // unsavoury invariant.
1526 waitForReturnCapability(&pcap, task);
1527 if (pcap != &capabilities[i]) {
1528 barf("scheduleDoGC: got the wrong capability");
1533 waiting_for_gc = rtsFalse;
1536 // so this happens periodically:
1537 if (cap) scheduleCheckBlackHoles(cap);
1539 IF_DEBUG(scheduler, printAllThreads());
1542 * We now have all the capabilities; if we're in an interrupting
1543 * state, then we should take the opportunity to delete all the
1544 * threads in the system.
1546 if (sched_state >= SCHED_INTERRUPTING) {
1547 deleteAllThreads(&capabilities[0]);
1548 sched_state = SCHED_SHUTTING_DOWN;
1551 heap_census = scheduleNeedHeapProfile(rtsTrue);
1553 /* everybody back, start the GC.
1554 * Could do it in this thread, or signal a condition var
1555 * to do it in another thread. Either way, we need to
1556 * broadcast on gc_pending_cond afterward.
1558 #if defined(THREADED_RTS)
1559 debugTrace(DEBUG_sched, "doing GC");
1561 GarbageCollect(force_major || heap_census);
1564 debugTrace(DEBUG_sched, "performing heap census");
1566 performHeapProfile = rtsFalse;
1571 Once we are all together... this would be the place to balance all
1572 spark pools. No concurrent stealing or adding of new sparks can
1573 occur. Should be defined in Sparks.c. */
1574 balanceSparkPoolsCaps(n_capabilities, capabilities);
1577 #if defined(THREADED_RTS)
1578 // release our stash of capabilities.
1579 for (i = 0; i < n_capabilities; i++) {
1580 if (cap != &capabilities[i]) {
1581 task->cap = &capabilities[i];
1582 releaseCapability(&capabilities[i]);
1595 /* ---------------------------------------------------------------------------
1596 * Singleton fork(). Do not copy any running threads.
1597 * ------------------------------------------------------------------------- */
1600 forkProcess(HsStablePtr *entry
1601 #ifndef FORKPROCESS_PRIMOP_SUPPORTED
1606 #ifdef FORKPROCESS_PRIMOP_SUPPORTED
1613 #if defined(THREADED_RTS)
1614 if (RtsFlags.ParFlags.nNodes > 1) {
1615 errorBelch("forking not supported with +RTS -N<n> greater than 1");
1616 stg_exit(EXIT_FAILURE);
1620 debugTrace(DEBUG_sched, "forking!");
1622 // ToDo: for SMP, we should probably acquire *all* the capabilities
1625 // no funny business: hold locks while we fork, otherwise if some
1626 // other thread is holding a lock when the fork happens, the data
1627 // structure protected by the lock will forever be in an
1628 // inconsistent state in the child. See also #1391.
1629 ACQUIRE_LOCK(&sched_mutex);
1630 ACQUIRE_LOCK(&cap->lock);
1631 ACQUIRE_LOCK(&cap->running_task->lock);
1635 if (pid) { // parent
1637 RELEASE_LOCK(&sched_mutex);
1638 RELEASE_LOCK(&cap->lock);
1639 RELEASE_LOCK(&cap->running_task->lock);
1641 // just return the pid
1647 #if defined(THREADED_RTS)
1648 initMutex(&sched_mutex);
1649 initMutex(&cap->lock);
1650 initMutex(&cap->running_task->lock);
1653 // Now, all OS threads except the thread that forked are
1654 // stopped. We need to stop all Haskell threads, including
1655 // those involved in foreign calls. Also we need to delete
1656 // all Tasks, because they correspond to OS threads that are
1659 for (s = 0; s < total_steps; s++) {
1660 for (t = all_steps[s].threads; t != END_TSO_QUEUE; t = next) {
1661 if (t->what_next == ThreadRelocated) {
1664 next = t->global_link;
1665 // don't allow threads to catch the ThreadKilled
1666 // exception, but we do want to raiseAsync() because these
1667 // threads may be evaluating thunks that we need later.
1668 deleteThread_(cap,t);
1673 // Empty the run queue. It seems tempting to let all the
1674 // killed threads stay on the run queue as zombies to be
1675 // cleaned up later, but some of them correspond to bound
1676 // threads for which the corresponding Task does not exist.
1677 cap->run_queue_hd = END_TSO_QUEUE;
1678 cap->run_queue_tl = END_TSO_QUEUE;
1680 // Any suspended C-calling Tasks are no more, their OS threads
1682 cap->suspended_ccalling_tasks = NULL;
1684 // Empty the threads lists. Otherwise, the garbage
1685 // collector may attempt to resurrect some of these threads.
1686 for (s = 0; s < total_steps; s++) {
1687 all_steps[s].threads = END_TSO_QUEUE;
1690 // Wipe the task list, except the current Task.
1691 ACQUIRE_LOCK(&sched_mutex);
1692 for (task = all_tasks; task != NULL; task=task->all_link) {
1693 if (task != cap->running_task) {
1694 #if defined(THREADED_RTS)
1695 initMutex(&task->lock); // see #1391
1700 RELEASE_LOCK(&sched_mutex);
1702 #if defined(THREADED_RTS)
1703 // Wipe our spare workers list, they no longer exist. New
1704 // workers will be created if necessary.
1705 cap->spare_workers = NULL;
1706 cap->returning_tasks_hd = NULL;
1707 cap->returning_tasks_tl = NULL;
1710 // On Unix, all timers are reset in the child, so we need to start
1715 cap = rts_evalStableIO(cap, entry, NULL); // run the action
1716 rts_checkSchedStatus("forkProcess",cap);
1719 hs_exit(); // clean up and exit
1720 stg_exit(EXIT_SUCCESS);
1722 #else /* !FORKPROCESS_PRIMOP_SUPPORTED */
1723 barf("forkProcess#: primop not supported on this platform, sorry!\n");
1728 /* ---------------------------------------------------------------------------
1729 * Delete all the threads in the system
1730 * ------------------------------------------------------------------------- */
1733 deleteAllThreads ( Capability *cap )
1735 // NOTE: only safe to call if we own all capabilities.
1740 debugTrace(DEBUG_sched,"deleting all threads");
1741 for (s = 0; s < total_steps; s++) {
1742 for (t = all_steps[s].threads; t != END_TSO_QUEUE; t = next) {
1743 if (t->what_next == ThreadRelocated) {
1746 next = t->global_link;
1747 deleteThread(cap,t);
1752 // The run queue now contains a bunch of ThreadKilled threads. We
1753 // must not throw these away: the main thread(s) will be in there
1754 // somewhere, and the main scheduler loop has to deal with it.
1755 // Also, the run queue is the only thing keeping these threads from
1756 // being GC'd, and we don't want the "main thread has been GC'd" panic.
1758 #if !defined(THREADED_RTS)
1759 ASSERT(blocked_queue_hd == END_TSO_QUEUE);
1760 ASSERT(sleeping_queue == END_TSO_QUEUE);
1764 /* -----------------------------------------------------------------------------
1765 Managing the suspended_ccalling_tasks list.
1766 Locks required: sched_mutex
1767 -------------------------------------------------------------------------- */
1770 suspendTask (Capability *cap, Task *task)
1772 ASSERT(task->next == NULL && task->prev == NULL);
1773 task->next = cap->suspended_ccalling_tasks;
1775 if (cap->suspended_ccalling_tasks) {
1776 cap->suspended_ccalling_tasks->prev = task;
1778 cap->suspended_ccalling_tasks = task;
1782 recoverSuspendedTask (Capability *cap, Task *task)
1785 task->prev->next = task->next;
1787 ASSERT(cap->suspended_ccalling_tasks == task);
1788 cap->suspended_ccalling_tasks = task->next;
1791 task->next->prev = task->prev;
1793 task->next = task->prev = NULL;
1796 /* ---------------------------------------------------------------------------
1797 * Suspending & resuming Haskell threads.
1799 * When making a "safe" call to C (aka _ccall_GC), the task gives back
1800 * its capability before calling the C function. This allows another
1801 * task to pick up the capability and carry on running Haskell
1802 * threads. It also means that if the C call blocks, it won't lock
1805 * The Haskell thread making the C call is put to sleep for the
1806 * duration of the call, on the susepended_ccalling_threads queue. We
1807 * give out a token to the task, which it can use to resume the thread
1808 * on return from the C function.
1809 * ------------------------------------------------------------------------- */
1812 suspendThread (StgRegTable *reg)
1819 StgWord32 saved_winerror;
1822 saved_errno = errno;
1824 saved_winerror = GetLastError();
1827 /* assume that *reg is a pointer to the StgRegTable part of a Capability.
1829 cap = regTableToCapability(reg);
1831 task = cap->running_task;
1832 tso = cap->r.rCurrentTSO;
1834 debugTrace(DEBUG_sched,
1835 "thread %lu did a safe foreign call",
1836 (unsigned long)cap->r.rCurrentTSO->id);
1838 // XXX this might not be necessary --SDM
1839 tso->what_next = ThreadRunGHC;
1841 threadPaused(cap,tso);
1843 if ((tso->flags & TSO_BLOCKEX) == 0) {
1844 tso->why_blocked = BlockedOnCCall;
1845 tso->flags |= TSO_BLOCKEX;
1846 tso->flags &= ~TSO_INTERRUPTIBLE;
1848 tso->why_blocked = BlockedOnCCall_NoUnblockExc;
1851 // Hand back capability
1852 task->suspended_tso = tso;
1854 ACQUIRE_LOCK(&cap->lock);
1856 suspendTask(cap,task);
1857 cap->in_haskell = rtsFalse;
1858 releaseCapability_(cap,rtsFalse);
1860 RELEASE_LOCK(&cap->lock);
1862 #if defined(THREADED_RTS)
1863 /* Preparing to leave the RTS, so ensure there's a native thread/task
1864 waiting to take over.
1866 debugTrace(DEBUG_sched, "thread %lu: leaving RTS", (unsigned long)tso->id);
1869 errno = saved_errno;
1871 SetLastError(saved_winerror);
1877 resumeThread (void *task_)
1884 StgWord32 saved_winerror;
1887 saved_errno = errno;
1889 saved_winerror = GetLastError();
1893 // Wait for permission to re-enter the RTS with the result.
1894 waitForReturnCapability(&cap,task);
1895 // we might be on a different capability now... but if so, our
1896 // entry on the suspended_ccalling_tasks list will also have been
1899 // Remove the thread from the suspended list
1900 recoverSuspendedTask(cap,task);
1902 tso = task->suspended_tso;
1903 task->suspended_tso = NULL;
1904 tso->_link = END_TSO_QUEUE; // no write barrier reqd
1905 debugTrace(DEBUG_sched, "thread %lu: re-entering RTS", (unsigned long)tso->id);
1907 if (tso->why_blocked == BlockedOnCCall) {
1908 awakenBlockedExceptionQueue(cap,tso);
1909 tso->flags &= ~(TSO_BLOCKEX | TSO_INTERRUPTIBLE);
1912 /* Reset blocking status */
1913 tso->why_blocked = NotBlocked;
1915 cap->r.rCurrentTSO = tso;
1916 cap->in_haskell = rtsTrue;
1917 errno = saved_errno;
1919 SetLastError(saved_winerror);
1922 /* We might have GC'd, mark the TSO dirty again */
1925 IF_DEBUG(sanity, checkTSO(tso));
1930 /* ---------------------------------------------------------------------------
1933 * scheduleThread puts a thread on the end of the runnable queue.
1934 * This will usually be done immediately after a thread is created.
1935 * The caller of scheduleThread must create the thread using e.g.
1936 * createThread and push an appropriate closure
1937 * on this thread's stack before the scheduler is invoked.
1938 * ------------------------------------------------------------------------ */
1941 scheduleThread(Capability *cap, StgTSO *tso)
1943 // The thread goes at the *end* of the run-queue, to avoid possible
1944 // starvation of any threads already on the queue.
1945 appendToRunQueue(cap,tso);
1949 scheduleThreadOn(Capability *cap, StgWord cpu USED_IF_THREADS, StgTSO *tso)
1951 #if defined(THREADED_RTS)
1952 tso->flags |= TSO_LOCKED; // we requested explicit affinity; don't
1953 // move this thread from now on.
1954 cpu %= RtsFlags.ParFlags.nNodes;
1955 if (cpu == cap->no) {
1956 appendToRunQueue(cap,tso);
1958 wakeupThreadOnCapability(cap, &capabilities[cpu], tso);
1961 appendToRunQueue(cap,tso);
1966 scheduleWaitThread (StgTSO* tso, /*[out]*/HaskellObj* ret, Capability *cap)
1970 // We already created/initialised the Task
1971 task = cap->running_task;
1973 // This TSO is now a bound thread; make the Task and TSO
1974 // point to each other.
1980 task->stat = NoStatus;
1982 appendToRunQueue(cap,tso);
1984 debugTrace(DEBUG_sched, "new bound thread (%lu)", (unsigned long)tso->id);
1986 cap = schedule(cap,task);
1988 ASSERT(task->stat != NoStatus);
1989 ASSERT_FULL_CAPABILITY_INVARIANTS(cap,task);
1991 debugTrace(DEBUG_sched, "bound thread (%lu) finished", (unsigned long)task->tso->id);
1995 /* ----------------------------------------------------------------------------
1997 * ------------------------------------------------------------------------- */
1999 #if defined(THREADED_RTS)
2000 void OSThreadProcAttr
2001 workerStart(Task *task)
2005 // See startWorkerTask().
2006 ACQUIRE_LOCK(&task->lock);
2008 RELEASE_LOCK(&task->lock);
2010 // set the thread-local pointer to the Task:
2013 // schedule() runs without a lock.
2014 cap = schedule(cap,task);
2016 // On exit from schedule(), we have a Capability.
2017 releaseCapability(cap);
2018 workerTaskStop(task);
2022 /* ---------------------------------------------------------------------------
2025 * Initialise the scheduler. This resets all the queues - if the
2026 * queues contained any threads, they'll be garbage collected at the
2029 * ------------------------------------------------------------------------ */
2034 #if !defined(THREADED_RTS)
2035 blocked_queue_hd = END_TSO_QUEUE;
2036 blocked_queue_tl = END_TSO_QUEUE;
2037 sleeping_queue = END_TSO_QUEUE;
2040 blackhole_queue = END_TSO_QUEUE;
2042 sched_state = SCHED_RUNNING;
2043 recent_activity = ACTIVITY_YES;
2045 #if defined(THREADED_RTS)
2046 /* Initialise the mutex and condition variables used by
2048 initMutex(&sched_mutex);
2051 ACQUIRE_LOCK(&sched_mutex);
2053 /* A capability holds the state a native thread needs in
2054 * order to execute STG code. At least one capability is
2055 * floating around (only THREADED_RTS builds have more than one).
2061 #if defined(THREADED_RTS) || defined(PARALLEL_HASKELL)
2065 #if defined(THREADED_RTS)
2067 * Eagerly start one worker to run each Capability, except for
2068 * Capability 0. The idea is that we're probably going to start a
2069 * bound thread on Capability 0 pretty soon, so we don't want a
2070 * worker task hogging it.
2075 for (i = 1; i < n_capabilities; i++) {
2076 cap = &capabilities[i];
2077 ACQUIRE_LOCK(&cap->lock);
2078 startWorkerTask(cap, workerStart);
2079 RELEASE_LOCK(&cap->lock);
2084 trace(TRACE_sched, "start: %d capabilities", n_capabilities);
2086 RELEASE_LOCK(&sched_mutex);
2091 rtsBool wait_foreign
2092 #if !defined(THREADED_RTS)
2093 __attribute__((unused))
2096 /* see Capability.c, shutdownCapability() */
2100 #if defined(THREADED_RTS)
2101 ACQUIRE_LOCK(&sched_mutex);
2102 task = newBoundTask();
2103 RELEASE_LOCK(&sched_mutex);
2106 // If we haven't killed all the threads yet, do it now.
2107 if (sched_state < SCHED_SHUTTING_DOWN) {
2108 sched_state = SCHED_INTERRUPTING;
2109 scheduleDoGC(NULL,task,rtsFalse);
2111 sched_state = SCHED_SHUTTING_DOWN;
2113 #if defined(THREADED_RTS)
2117 for (i = 0; i < n_capabilities; i++) {
2118 shutdownCapability(&capabilities[i], task, wait_foreign);
2120 boundTaskExiting(task);
2127 freeScheduler( void )
2131 if (n_capabilities != 1) {
2132 stgFree(capabilities);
2134 #if defined(THREADED_RTS)
2135 closeMutex(&sched_mutex);
2139 /* -----------------------------------------------------------------------------
2142 This is the interface to the garbage collector from Haskell land.
2143 We provide this so that external C code can allocate and garbage
2144 collect when called from Haskell via _ccall_GC.
2145 -------------------------------------------------------------------------- */
2148 performGC_(rtsBool force_major)
2151 // We must grab a new Task here, because the existing Task may be
2152 // associated with a particular Capability, and chained onto the
2153 // suspended_ccalling_tasks queue.
2154 ACQUIRE_LOCK(&sched_mutex);
2155 task = newBoundTask();
2156 RELEASE_LOCK(&sched_mutex);
2157 scheduleDoGC(NULL,task,force_major);
2158 boundTaskExiting(task);
2164 performGC_(rtsFalse);
2168 performMajorGC(void)
2170 performGC_(rtsTrue);
2173 /* -----------------------------------------------------------------------------
2176 If the thread has reached its maximum stack size, then raise the
2177 StackOverflow exception in the offending thread. Otherwise
2178 relocate the TSO into a larger chunk of memory and adjust its stack
2180 -------------------------------------------------------------------------- */
2183 threadStackOverflow(Capability *cap, StgTSO *tso)
2185 nat new_stack_size, stack_words;
2190 IF_DEBUG(sanity,checkTSO(tso));
2192 // don't allow throwTo() to modify the blocked_exceptions queue
2193 // while we are moving the TSO:
2194 lockClosure((StgClosure *)tso);
2196 if (tso->stack_size >= tso->max_stack_size && !(tso->flags & TSO_BLOCKEX)) {
2197 // NB. never raise a StackOverflow exception if the thread is
2198 // inside Control.Exceptino.block. It is impractical to protect
2199 // against stack overflow exceptions, since virtually anything
2200 // can raise one (even 'catch'), so this is the only sensible
2201 // thing to do here. See bug #767.
2203 debugTrace(DEBUG_gc,
2204 "threadStackOverflow of TSO %ld (%p): stack too large (now %ld; max is %ld)",
2205 (long)tso->id, tso, (long)tso->stack_size, (long)tso->max_stack_size);
2207 /* If we're debugging, just print out the top of the stack */
2208 printStackChunk(tso->sp, stg_min(tso->stack+tso->stack_size,
2211 // Send this thread the StackOverflow exception
2213 throwToSingleThreaded(cap, tso, (StgClosure *)stackOverflow_closure);
2217 /* Try to double the current stack size. If that takes us over the
2218 * maximum stack size for this thread, then use the maximum instead
2219 * (that is, unless we're already at or over the max size and we
2220 * can't raise the StackOverflow exception (see above), in which
2221 * case just double the size). Finally round up so the TSO ends up as
2222 * a whole number of blocks.
2224 if (tso->stack_size >= tso->max_stack_size) {
2225 new_stack_size = tso->stack_size * 2;
2227 new_stack_size = stg_min(tso->stack_size * 2, tso->max_stack_size);
2229 new_tso_size = (lnat)BLOCK_ROUND_UP(new_stack_size * sizeof(W_) +
2230 TSO_STRUCT_SIZE)/sizeof(W_);
2231 new_tso_size = round_to_mblocks(new_tso_size); /* Be MBLOCK-friendly */
2232 new_stack_size = new_tso_size - TSO_STRUCT_SIZEW;
2234 debugTrace(DEBUG_sched,
2235 "increasing stack size from %ld words to %d.",
2236 (long)tso->stack_size, new_stack_size);
2238 dest = (StgTSO *)allocateLocal(cap,new_tso_size);
2239 TICK_ALLOC_TSO(new_stack_size,0);
2241 /* copy the TSO block and the old stack into the new area */
2242 memcpy(dest,tso,TSO_STRUCT_SIZE);
2243 stack_words = tso->stack + tso->stack_size - tso->sp;
2244 new_sp = (P_)dest + new_tso_size - stack_words;
2245 memcpy(new_sp, tso->sp, stack_words * sizeof(W_));
2247 /* relocate the stack pointers... */
2249 dest->stack_size = new_stack_size;
2251 /* Mark the old TSO as relocated. We have to check for relocated
2252 * TSOs in the garbage collector and any primops that deal with TSOs.
2254 * It's important to set the sp value to just beyond the end
2255 * of the stack, so we don't attempt to scavenge any part of the
2258 tso->what_next = ThreadRelocated;
2259 setTSOLink(cap,tso,dest);
2260 tso->sp = (P_)&(tso->stack[tso->stack_size]);
2261 tso->why_blocked = NotBlocked;
2263 IF_PAR_DEBUG(verbose,
2264 debugBelch("@@ threadStackOverflow of TSO %d (now at %p): stack size increased to %ld\n",
2265 tso->id, tso, tso->stack_size);
2266 /* If we're debugging, just print out the top of the stack */
2267 printStackChunk(tso->sp, stg_min(tso->stack+tso->stack_size,
2273 IF_DEBUG(sanity,checkTSO(dest));
2275 IF_DEBUG(scheduler,printTSO(dest));
2282 threadStackUnderflow (Task *task STG_UNUSED, StgTSO *tso)
2284 bdescr *bd, *new_bd;
2285 lnat free_w, tso_size_w;
2288 tso_size_w = tso_sizeW(tso);
2290 if (tso_size_w < MBLOCK_SIZE_W ||
2291 (nat)(tso->stack + tso->stack_size - tso->sp) > tso->stack_size / 4)
2296 // don't allow throwTo() to modify the blocked_exceptions queue
2297 // while we are moving the TSO:
2298 lockClosure((StgClosure *)tso);
2300 // this is the number of words we'll free
2301 free_w = round_to_mblocks(tso_size_w/2);
2303 bd = Bdescr((StgPtr)tso);
2304 new_bd = splitLargeBlock(bd, free_w / BLOCK_SIZE_W);
2305 bd->free = bd->start + TSO_STRUCT_SIZEW;
2307 new_tso = (StgTSO *)new_bd->start;
2308 memcpy(new_tso,tso,TSO_STRUCT_SIZE);
2309 new_tso->stack_size = new_bd->free - new_tso->stack;
2311 debugTrace(DEBUG_sched, "thread %ld: reducing TSO size from %lu words to %lu",
2312 (long)tso->id, tso_size_w, tso_sizeW(new_tso));
2314 tso->what_next = ThreadRelocated;
2315 tso->_link = new_tso; // no write barrier reqd: same generation
2317 // The TSO attached to this Task may have moved, so update the
2319 if (task->tso == tso) {
2320 task->tso = new_tso;
2326 IF_DEBUG(sanity,checkTSO(new_tso));
2331 /* ---------------------------------------------------------------------------
2333 - usually called inside a signal handler so it mustn't do anything fancy.
2334 ------------------------------------------------------------------------ */
2337 interruptStgRts(void)
2339 sched_state = SCHED_INTERRUPTING;
2340 setContextSwitches();
2344 /* -----------------------------------------------------------------------------
2347 This function causes at least one OS thread to wake up and run the
2348 scheduler loop. It is invoked when the RTS might be deadlocked, or
2349 an external event has arrived that may need servicing (eg. a
2350 keyboard interrupt).
2352 In the single-threaded RTS we don't do anything here; we only have
2353 one thread anyway, and the event that caused us to want to wake up
2354 will have interrupted any blocking system call in progress anyway.
2355 -------------------------------------------------------------------------- */
2360 #if defined(THREADED_RTS)
2361 // This forces the IO Manager thread to wakeup, which will
2362 // in turn ensure that some OS thread wakes up and runs the
2363 // scheduler loop, which will cause a GC and deadlock check.
2368 /* -----------------------------------------------------------------------------
2371 * Check the blackhole_queue for threads that can be woken up. We do
2372 * this periodically: before every GC, and whenever the run queue is
2375 * An elegant solution might be to just wake up all the blocked
2376 * threads with awakenBlockedQueue occasionally: they'll go back to
2377 * sleep again if the object is still a BLACKHOLE. Unfortunately this
2378 * doesn't give us a way to tell whether we've actually managed to
2379 * wake up any threads, so we would be busy-waiting.
2381 * -------------------------------------------------------------------------- */
2384 checkBlackHoles (Capability *cap)
2387 rtsBool any_woke_up = rtsFalse;
2390 // blackhole_queue is global:
2391 ASSERT_LOCK_HELD(&sched_mutex);
2393 debugTrace(DEBUG_sched, "checking threads blocked on black holes");
2395 // ASSUMES: sched_mutex
2396 prev = &blackhole_queue;
2397 t = blackhole_queue;
2398 while (t != END_TSO_QUEUE) {
2399 ASSERT(t->why_blocked == BlockedOnBlackHole);
2400 type = get_itbl(UNTAG_CLOSURE(t->block_info.closure))->type;
2401 if (type != BLACKHOLE && type != CAF_BLACKHOLE) {
2402 IF_DEBUG(sanity,checkTSO(t));
2403 t = unblockOne(cap, t);
2405 any_woke_up = rtsTrue;
2415 /* -----------------------------------------------------------------------------
2418 This is used for interruption (^C) and forking, and corresponds to
2419 raising an exception but without letting the thread catch the
2421 -------------------------------------------------------------------------- */
2424 deleteThread (Capability *cap, StgTSO *tso)
2426 // NOTE: must only be called on a TSO that we have exclusive
2427 // access to, because we will call throwToSingleThreaded() below.
2428 // The TSO must be on the run queue of the Capability we own, or
2429 // we must own all Capabilities.
2431 if (tso->why_blocked != BlockedOnCCall &&
2432 tso->why_blocked != BlockedOnCCall_NoUnblockExc) {
2433 throwToSingleThreaded(cap,tso,NULL);
2437 #ifdef FORKPROCESS_PRIMOP_SUPPORTED
2439 deleteThread_(Capability *cap, StgTSO *tso)
2440 { // for forkProcess only:
2441 // like deleteThread(), but we delete threads in foreign calls, too.
2443 if (tso->why_blocked == BlockedOnCCall ||
2444 tso->why_blocked == BlockedOnCCall_NoUnblockExc) {
2445 unblockOne(cap,tso);
2446 tso->what_next = ThreadKilled;
2448 deleteThread(cap,tso);
2453 /* -----------------------------------------------------------------------------
2454 raiseExceptionHelper
2456 This function is called by the raise# primitve, just so that we can
2457 move some of the tricky bits of raising an exception from C-- into
2458 C. Who knows, it might be a useful re-useable thing here too.
2459 -------------------------------------------------------------------------- */
2462 raiseExceptionHelper (StgRegTable *reg, StgTSO *tso, StgClosure *exception)
2464 Capability *cap = regTableToCapability(reg);
2465 StgThunk *raise_closure = NULL;
2467 StgRetInfoTable *info;
2469 // This closure represents the expression 'raise# E' where E
2470 // is the exception raise. It is used to overwrite all the
2471 // thunks which are currently under evaluataion.
2474 // OLD COMMENT (we don't have MIN_UPD_SIZE now):
2475 // LDV profiling: stg_raise_info has THUNK as its closure
2476 // type. Since a THUNK takes at least MIN_UPD_SIZE words in its
2477 // payload, MIN_UPD_SIZE is more approprate than 1. It seems that
2478 // 1 does not cause any problem unless profiling is performed.
2479 // However, when LDV profiling goes on, we need to linearly scan
2480 // small object pool, where raise_closure is stored, so we should
2481 // use MIN_UPD_SIZE.
2483 // raise_closure = (StgClosure *)RET_STGCALL1(P_,allocate,
2484 // sizeofW(StgClosure)+1);
2488 // Walk up the stack, looking for the catch frame. On the way,
2489 // we update any closures pointed to from update frames with the
2490 // raise closure that we just built.
2494 info = get_ret_itbl((StgClosure *)p);
2495 next = p + stack_frame_sizeW((StgClosure *)p);
2496 switch (info->i.type) {
2499 // Only create raise_closure if we need to.
2500 if (raise_closure == NULL) {
2502 (StgThunk *)allocateLocal(cap,sizeofW(StgThunk)+1);
2503 SET_HDR(raise_closure, &stg_raise_info, CCCS);
2504 raise_closure->payload[0] = exception;
2506 UPD_IND(((StgUpdateFrame *)p)->updatee,(StgClosure *)raise_closure);
2510 case ATOMICALLY_FRAME:
2511 debugTrace(DEBUG_stm, "found ATOMICALLY_FRAME at %p", p);
2513 return ATOMICALLY_FRAME;
2519 case CATCH_STM_FRAME:
2520 debugTrace(DEBUG_stm, "found CATCH_STM_FRAME at %p", p);
2522 return CATCH_STM_FRAME;
2528 case CATCH_RETRY_FRAME:
2537 /* -----------------------------------------------------------------------------
2538 findRetryFrameHelper
2540 This function is called by the retry# primitive. It traverses the stack
2541 leaving tso->sp referring to the frame which should handle the retry.
2543 This should either be a CATCH_RETRY_FRAME (if the retry# is within an orElse#)
2544 or should be a ATOMICALLY_FRAME (if the retry# reaches the top level).
2546 We skip CATCH_STM_FRAMEs (aborting and rolling back the nested tx that they
2547 create) because retries are not considered to be exceptions, despite the
2548 similar implementation.
2550 We should not expect to see CATCH_FRAME or STOP_FRAME because those should
2551 not be created within memory transactions.
2552 -------------------------------------------------------------------------- */
2555 findRetryFrameHelper (StgTSO *tso)
2558 StgRetInfoTable *info;
2562 info = get_ret_itbl((StgClosure *)p);
2563 next = p + stack_frame_sizeW((StgClosure *)p);
2564 switch (info->i.type) {
2566 case ATOMICALLY_FRAME:
2567 debugTrace(DEBUG_stm,
2568 "found ATOMICALLY_FRAME at %p during retry", p);
2570 return ATOMICALLY_FRAME;
2572 case CATCH_RETRY_FRAME:
2573 debugTrace(DEBUG_stm,
2574 "found CATCH_RETRY_FRAME at %p during retrry", p);
2576 return CATCH_RETRY_FRAME;
2578 case CATCH_STM_FRAME: {
2579 StgTRecHeader *trec = tso -> trec;
2580 StgTRecHeader *outer = stmGetEnclosingTRec(trec);
2581 debugTrace(DEBUG_stm,
2582 "found CATCH_STM_FRAME at %p during retry", p);
2583 debugTrace(DEBUG_stm, "trec=%p outer=%p", trec, outer);
2584 stmAbortTransaction(tso -> cap, trec);
2585 stmFreeAbortedTRec(tso -> cap, trec);
2586 tso -> trec = outer;
2593 ASSERT(info->i.type != CATCH_FRAME);
2594 ASSERT(info->i.type != STOP_FRAME);
2601 /* -----------------------------------------------------------------------------
2602 resurrectThreads is called after garbage collection on the list of
2603 threads found to be garbage. Each of these threads will be woken
2604 up and sent a signal: BlockedOnDeadMVar if the thread was blocked
2605 on an MVar, or NonTermination if the thread was blocked on a Black
2608 Locks: assumes we hold *all* the capabilities.
2609 -------------------------------------------------------------------------- */
2612 resurrectThreads (StgTSO *threads)
2618 for (tso = threads; tso != END_TSO_QUEUE; tso = next) {
2619 next = tso->global_link;
2621 step = Bdescr((P_)tso)->step;
2622 tso->global_link = step->threads;
2623 step->threads = tso;
2625 debugTrace(DEBUG_sched, "resurrecting thread %lu", (unsigned long)tso->id);
2627 // Wake up the thread on the Capability it was last on
2630 switch (tso->why_blocked) {
2632 case BlockedOnException:
2633 /* Called by GC - sched_mutex lock is currently held. */
2634 throwToSingleThreaded(cap, tso,
2635 (StgClosure *)blockedOnDeadMVar_closure);
2637 case BlockedOnBlackHole:
2638 throwToSingleThreaded(cap, tso,
2639 (StgClosure *)nonTermination_closure);
2642 throwToSingleThreaded(cap, tso,
2643 (StgClosure *)blockedIndefinitely_closure);
2646 /* This might happen if the thread was blocked on a black hole
2647 * belonging to a thread that we've just woken up (raiseAsync
2648 * can wake up threads, remember...).
2652 barf("resurrectThreads: thread blocked in a strange way");
2657 /* -----------------------------------------------------------------------------
2658 performPendingThrowTos is called after garbage collection, and
2659 passed a list of threads that were found to have pending throwTos
2660 (tso->blocked_exceptions was not empty), and were blocked.
2661 Normally this doesn't happen, because we would deliver the
2662 exception directly if the target thread is blocked, but there are
2663 small windows where it might occur on a multiprocessor (see
2666 NB. we must be holding all the capabilities at this point, just
2667 like resurrectThreads().
2668 -------------------------------------------------------------------------- */
2671 performPendingThrowTos (StgTSO *threads)
2677 for (tso = threads; tso != END_TSO_QUEUE; tso = next) {
2678 next = tso->global_link;
2680 step = Bdescr((P_)tso)->step;
2681 tso->global_link = step->threads;
2682 step->threads = tso;
2684 debugTrace(DEBUG_sched, "performing blocked throwTo to thread %lu", (unsigned long)tso->id);
2687 maybePerformBlockedException(cap, tso);