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"
36 /* PARALLEL_HASKELL includes go here */
39 #include "Capability.h"
41 #include "AwaitEvent.h"
42 #if defined(mingw32_HOST_OS)
43 #include "win32/IOManager.h"
46 #include "RaiseAsync.h"
48 #include "ThrIOManager.h"
50 #ifdef HAVE_SYS_TYPES_H
51 #include <sys/types.h>
65 // Turn off inlining when debugging - it obfuscates things
68 # define STATIC_INLINE static
71 /* -----------------------------------------------------------------------------
73 * -------------------------------------------------------------------------- */
75 #if !defined(THREADED_RTS)
76 // Blocked/sleeping thrads
77 StgTSO *blocked_queue_hd = NULL;
78 StgTSO *blocked_queue_tl = NULL;
79 StgTSO *sleeping_queue = NULL; // perhaps replace with a hash table?
82 /* Threads blocked on blackholes.
83 * LOCK: sched_mutex+capability, or all capabilities
85 StgTSO *blackhole_queue = NULL;
87 /* The blackhole_queue should be checked for threads to wake up. See
88 * Schedule.h for more thorough comment.
89 * LOCK: none (doesn't matter if we miss an update)
91 rtsBool blackholes_need_checking = rtsFalse;
93 /* Set to true when the latest garbage collection failed to reclaim
94 * enough space, and the runtime should proceed to shut itself down in
95 * an orderly fashion (emitting profiling info etc.)
97 rtsBool heap_overflow = rtsFalse;
99 /* flag that tracks whether we have done any execution in this time slice.
100 * LOCK: currently none, perhaps we should lock (but needs to be
101 * updated in the fast path of the scheduler).
103 * NB. must be StgWord, we do xchg() on it.
105 volatile StgWord recent_activity = ACTIVITY_YES;
107 /* if this flag is set as well, give up execution
108 * LOCK: none (changes monotonically)
110 volatile StgWord sched_state = SCHED_RUNNING;
112 /* This is used in `TSO.h' and gcc 2.96 insists that this variable actually
113 * exists - earlier gccs apparently didn't.
119 * Set to TRUE when entering a shutdown state (via shutdownHaskellAndExit()) --
120 * in an MT setting, needed to signal that a worker thread shouldn't hang around
121 * in the scheduler when it is out of work.
123 rtsBool shutting_down_scheduler = rtsFalse;
126 * This mutex protects most of the global scheduler data in
127 * the THREADED_RTS runtime.
129 #if defined(THREADED_RTS)
133 #if !defined(mingw32_HOST_OS)
134 #define FORKPROCESS_PRIMOP_SUPPORTED
137 /* -----------------------------------------------------------------------------
138 * static function prototypes
139 * -------------------------------------------------------------------------- */
141 static Capability *schedule (Capability *initialCapability, Task *task);
144 // These function all encapsulate parts of the scheduler loop, and are
145 // abstracted only to make the structure and control flow of the
146 // scheduler clearer.
148 static void schedulePreLoop (void);
149 static void scheduleFindWork (Capability *cap);
150 #if defined(THREADED_RTS)
151 static void scheduleYield (Capability **pcap, Task *task);
153 static void scheduleStartSignalHandlers (Capability *cap);
154 static void scheduleCheckBlockedThreads (Capability *cap);
155 static void scheduleCheckWakeupThreads(Capability *cap USED_IF_NOT_THREADS);
156 static void scheduleCheckBlackHoles (Capability *cap);
157 static void scheduleDetectDeadlock (Capability *cap, Task *task);
158 static void schedulePushWork(Capability *cap, Task *task);
159 #if defined(PARALLEL_HASKELL)
160 static rtsBool scheduleGetRemoteWork(Capability *cap);
161 static void scheduleSendPendingMessages(void);
163 #if defined(PARALLEL_HASKELL) || defined(THREADED_RTS)
164 static void scheduleActivateSpark(Capability *cap);
166 static void schedulePostRunThread(Capability *cap, StgTSO *t);
167 static rtsBool scheduleHandleHeapOverflow( Capability *cap, StgTSO *t );
168 static void scheduleHandleStackOverflow( Capability *cap, Task *task,
170 static rtsBool scheduleHandleYield( Capability *cap, StgTSO *t,
171 nat prev_what_next );
172 static void scheduleHandleThreadBlocked( StgTSO *t );
173 static rtsBool scheduleHandleThreadFinished( Capability *cap, Task *task,
175 static rtsBool scheduleNeedHeapProfile(rtsBool ready_to_gc);
176 static Capability *scheduleDoGC(Capability *cap, Task *task,
177 rtsBool force_major);
179 static rtsBool checkBlackHoles(Capability *cap);
181 static StgTSO *threadStackOverflow(Capability *cap, StgTSO *tso);
182 static StgTSO *threadStackUnderflow(Task *task, StgTSO *tso);
184 static void deleteThread (Capability *cap, StgTSO *tso);
185 static void deleteAllThreads (Capability *cap);
187 #ifdef FORKPROCESS_PRIMOP_SUPPORTED
188 static void deleteThread_(Capability *cap, StgTSO *tso);
192 static char *whatNext_strs[] = {
202 /* -----------------------------------------------------------------------------
203 * Putting a thread on the run queue: different scheduling policies
204 * -------------------------------------------------------------------------- */
207 addToRunQueue( Capability *cap, StgTSO *t )
209 #if defined(PARALLEL_HASKELL)
210 if (RtsFlags.ParFlags.doFairScheduling) {
211 // this does round-robin scheduling; good for concurrency
212 appendToRunQueue(cap,t);
214 // this does unfair scheduling; good for parallelism
215 pushOnRunQueue(cap,t);
218 // this does round-robin scheduling; good for concurrency
219 appendToRunQueue(cap,t);
223 /* ---------------------------------------------------------------------------
224 Main scheduling loop.
226 We use round-robin scheduling, each thread returning to the
227 scheduler loop when one of these conditions is detected:
230 * timer expires (thread yields)
236 In a GranSim setup this loop iterates over the global event queue.
237 This revolves around the global event queue, which determines what
238 to do next. Therefore, it's more complicated than either the
239 concurrent or the parallel (GUM) setup.
240 This version has been entirely removed (JB 2008/08).
243 GUM iterates over incoming messages.
244 It starts with nothing to do (thus CurrentTSO == END_TSO_QUEUE),
245 and sends out a fish whenever it has nothing to do; in-between
246 doing the actual reductions (shared code below) it processes the
247 incoming messages and deals with delayed operations
248 (see PendingFetches).
249 This is not the ugliest code you could imagine, but it's bloody close.
251 (JB 2008/08) This version was formerly indicated by a PP-Flag PAR,
252 now by PP-flag PARALLEL_HASKELL. The Eden RTS (in GHC-6.x) uses it,
253 as well as future GUM versions. This file has been refurbished to
254 only contain valid code, which is however incomplete, refers to
255 invalid includes etc.
257 ------------------------------------------------------------------------ */
260 schedule (Capability *initialCapability, Task *task)
264 StgThreadReturnCode ret;
265 #if defined(PARALLEL_HASKELL)
266 rtsBool receivedFinish = rtsFalse;
270 #if defined(THREADED_RTS)
271 rtsBool first = rtsTrue;
274 cap = initialCapability;
276 // Pre-condition: this task owns initialCapability.
277 // The sched_mutex is *NOT* held
278 // NB. on return, we still hold a capability.
280 debugTrace (DEBUG_sched,
281 "### NEW SCHEDULER LOOP (task: %p, cap: %p)",
282 task, initialCapability);
286 // -----------------------------------------------------------
287 // Scheduler loop starts here:
289 #if defined(PARALLEL_HASKELL)
290 #define TERMINATION_CONDITION (!receivedFinish)
292 #define TERMINATION_CONDITION rtsTrue
295 while (TERMINATION_CONDITION) {
297 // Check whether we have re-entered the RTS from Haskell without
298 // going via suspendThread()/resumeThread (i.e. a 'safe' foreign
300 if (cap->in_haskell) {
301 errorBelch("schedule: re-entered unsafely.\n"
302 " Perhaps a 'foreign import unsafe' should be 'safe'?");
303 stg_exit(EXIT_FAILURE);
306 // The interruption / shutdown sequence.
308 // In order to cleanly shut down the runtime, we want to:
309 // * make sure that all main threads return to their callers
310 // with the state 'Interrupted'.
311 // * clean up all OS threads assocated with the runtime
312 // * free all memory etc.
314 // So the sequence for ^C goes like this:
316 // * ^C handler sets sched_state := SCHED_INTERRUPTING and
317 // arranges for some Capability to wake up
319 // * all threads in the system are halted, and the zombies are
320 // placed on the run queue for cleaning up. We acquire all
321 // the capabilities in order to delete the threads, this is
322 // done by scheduleDoGC() for convenience (because GC already
323 // needs to acquire all the capabilities). We can't kill
324 // threads involved in foreign calls.
326 // * somebody calls shutdownHaskell(), which calls exitScheduler()
328 // * sched_state := SCHED_SHUTTING_DOWN
330 // * all workers exit when the run queue on their capability
331 // drains. All main threads will also exit when their TSO
332 // reaches the head of the run queue and they can return.
334 // * eventually all Capabilities will shut down, and the RTS can
337 // * We might be left with threads blocked in foreign calls,
338 // we should really attempt to kill these somehow (TODO);
340 switch (sched_state) {
343 case SCHED_INTERRUPTING:
344 debugTrace(DEBUG_sched, "SCHED_INTERRUPTING");
345 #if defined(THREADED_RTS)
346 discardSparksCap(cap);
348 /* scheduleDoGC() deletes all the threads */
349 cap = scheduleDoGC(cap,task,rtsFalse);
351 // after scheduleDoGC(), we must be shutting down. Either some
352 // other Capability did the final GC, or we did it above,
353 // either way we can fall through to the SCHED_SHUTTING_DOWN
355 ASSERT(sched_state == SCHED_SHUTTING_DOWN);
358 case SCHED_SHUTTING_DOWN:
359 debugTrace(DEBUG_sched, "SCHED_SHUTTING_DOWN");
360 // If we are a worker, just exit. If we're a bound thread
361 // then we will exit below when we've removed our TSO from
363 if (task->tso == NULL && emptyRunQueue(cap)) {
368 barf("sched_state: %d", sched_state);
371 scheduleFindWork(cap);
373 /* work pushing, currently relevant only for THREADED_RTS:
374 (pushes threads, wakes up idle capabilities for stealing) */
375 schedulePushWork(cap,task);
377 #if defined(PARALLEL_HASKELL)
378 /* since we perform a blocking receive and continue otherwise,
379 either we never reach here or we definitely have work! */
380 // from here: non-empty run queue
381 ASSERT(!emptyRunQueue(cap));
383 if (PacketsWaiting()) { /* now process incoming messages, if any
386 CAUTION: scheduleGetRemoteWork called
387 above, waits for messages as well! */
388 processMessages(cap, &receivedFinish);
390 #endif // PARALLEL_HASKELL: non-empty run queue!
392 scheduleDetectDeadlock(cap,task);
394 #if defined(THREADED_RTS)
395 cap = task->cap; // reload cap, it might have changed
398 // Normally, the only way we can get here with no threads to
399 // run is if a keyboard interrupt received during
400 // scheduleCheckBlockedThreads() or scheduleDetectDeadlock().
401 // Additionally, it is not fatal for the
402 // threaded RTS to reach here with no threads to run.
404 // win32: might be here due to awaitEvent() being abandoned
405 // as a result of a console event having been delivered.
407 #if defined(THREADED_RTS)
411 // // don't yield the first time, we want a chance to run this
412 // // thread for a bit, even if there are others banging at the
415 // ASSERT_FULL_CAPABILITY_INVARIANTS(cap,task);
419 scheduleYield(&cap,task);
420 if (emptyRunQueue(cap)) continue; // look for work again
423 #if !defined(THREADED_RTS) && !defined(mingw32_HOST_OS)
424 if ( emptyRunQueue(cap) ) {
425 ASSERT(sched_state >= SCHED_INTERRUPTING);
430 // Get a thread to run
432 t = popRunQueue(cap);
434 // Sanity check the thread we're about to run. This can be
435 // expensive if there is lots of thread switching going on...
436 IF_DEBUG(sanity,checkTSO(t));
438 #if defined(THREADED_RTS)
439 // Check whether we can run this thread in the current task.
440 // If not, we have to pass our capability to the right task.
442 Task *bound = t->bound;
446 debugTrace(DEBUG_sched,
447 "### Running thread %lu in bound thread", (unsigned long)t->id);
448 // yes, the Haskell thread is bound to the current native thread
450 debugTrace(DEBUG_sched,
451 "### thread %lu bound to another OS thread", (unsigned long)t->id);
452 // no, bound to a different Haskell thread: pass to that thread
453 pushOnRunQueue(cap,t);
457 // The thread we want to run is unbound.
459 debugTrace(DEBUG_sched,
460 "### this OS thread cannot run thread %lu", (unsigned long)t->id);
461 // no, the current native thread is bound to a different
462 // Haskell thread, so pass it to any worker thread
463 pushOnRunQueue(cap,t);
470 // If we're shutting down, and this thread has not yet been
471 // killed, kill it now. This sometimes happens when a finalizer
472 // thread is created by the final GC, or a thread previously
473 // in a foreign call returns.
474 if (sched_state >= SCHED_INTERRUPTING &&
475 !(t->what_next == ThreadComplete || t->what_next == ThreadKilled)) {
479 /* context switches are initiated by the timer signal, unless
480 * the user specified "context switch as often as possible", with
483 if (RtsFlags.ConcFlags.ctxtSwitchTicks == 0
484 && !emptyThreadQueues(cap)) {
485 cap->context_switch = 1;
490 // CurrentTSO is the thread to run. t might be different if we
491 // loop back to run_thread, so make sure to set CurrentTSO after
493 cap->r.rCurrentTSO = t;
495 debugTrace(DEBUG_sched, "-->> running thread %ld %s ...",
496 (long)t->id, whatNext_strs[t->what_next]);
498 startHeapProfTimer();
500 // Check for exceptions blocked on this thread
501 maybePerformBlockedException (cap, t);
503 // ----------------------------------------------------------------------
504 // Run the current thread
506 ASSERT_FULL_CAPABILITY_INVARIANTS(cap,task);
507 ASSERT(t->cap == cap);
508 ASSERT(t->bound ? t->bound->cap == cap : 1);
510 prev_what_next = t->what_next;
512 errno = t->saved_errno;
514 SetLastError(t->saved_winerror);
517 cap->in_haskell = rtsTrue;
521 #if defined(THREADED_RTS)
522 if (recent_activity == ACTIVITY_DONE_GC) {
523 // ACTIVITY_DONE_GC means we turned off the timer signal to
524 // conserve power (see #1623). Re-enable it here.
526 prev = xchg((P_)&recent_activity, ACTIVITY_YES);
527 if (prev == ACTIVITY_DONE_GC) {
531 recent_activity = ACTIVITY_YES;
535 switch (prev_what_next) {
539 /* Thread already finished, return to scheduler. */
540 ret = ThreadFinished;
546 r = StgRun((StgFunPtr) stg_returnToStackTop, &cap->r);
547 cap = regTableToCapability(r);
552 case ThreadInterpret:
553 cap = interpretBCO(cap);
558 barf("schedule: invalid what_next field");
561 cap->in_haskell = rtsFalse;
563 // The TSO might have moved, eg. if it re-entered the RTS and a GC
564 // happened. So find the new location:
565 t = cap->r.rCurrentTSO;
567 // We have run some Haskell code: there might be blackhole-blocked
568 // threads to wake up now.
569 // Lock-free test here should be ok, we're just setting a flag.
570 if ( blackhole_queue != END_TSO_QUEUE ) {
571 blackholes_need_checking = rtsTrue;
574 // And save the current errno in this thread.
575 // XXX: possibly bogus for SMP because this thread might already
576 // be running again, see code below.
577 t->saved_errno = errno;
579 // Similarly for Windows error code
580 t->saved_winerror = GetLastError();
583 #if defined(THREADED_RTS)
584 // If ret is ThreadBlocked, and this Task is bound to the TSO that
585 // blocked, we are in limbo - the TSO is now owned by whatever it
586 // is blocked on, and may in fact already have been woken up,
587 // perhaps even on a different Capability. It may be the case
588 // that task->cap != cap. We better yield this Capability
589 // immediately and return to normaility.
590 if (ret == ThreadBlocked) {
591 debugTrace(DEBUG_sched,
592 "--<< thread %lu (%s) stopped: blocked",
593 (unsigned long)t->id, whatNext_strs[t->what_next]);
598 ASSERT_FULL_CAPABILITY_INVARIANTS(cap,task);
599 ASSERT(t->cap == cap);
601 // ----------------------------------------------------------------------
603 // Costs for the scheduler are assigned to CCS_SYSTEM
605 #if defined(PROFILING)
609 schedulePostRunThread(cap,t);
611 t = threadStackUnderflow(task,t);
613 ready_to_gc = rtsFalse;
617 ready_to_gc = scheduleHandleHeapOverflow(cap,t);
621 scheduleHandleStackOverflow(cap,task,t);
625 if (scheduleHandleYield(cap, t, prev_what_next)) {
626 // shortcut for switching between compiler/interpreter:
632 scheduleHandleThreadBlocked(t);
636 if (scheduleHandleThreadFinished(cap, task, t)) return cap;
637 ASSERT_FULL_CAPABILITY_INVARIANTS(cap,task);
641 barf("schedule: invalid thread return code %d", (int)ret);
644 if (ready_to_gc || scheduleNeedHeapProfile(ready_to_gc)) {
645 cap = scheduleDoGC(cap,task,rtsFalse);
647 } /* end of while() */
650 /* ----------------------------------------------------------------------------
651 * Setting up the scheduler loop
652 * ------------------------------------------------------------------------- */
655 schedulePreLoop(void)
657 // initialisation for scheduler - what cannot go into initScheduler()
660 /* -----------------------------------------------------------------------------
663 * Search for work to do, and handle messages from elsewhere.
664 * -------------------------------------------------------------------------- */
667 scheduleFindWork (Capability *cap)
669 scheduleStartSignalHandlers(cap);
671 // Only check the black holes here if we've nothing else to do.
672 // During normal execution, the black hole list only gets checked
673 // at GC time, to avoid repeatedly traversing this possibly long
674 // list each time around the scheduler.
675 if (emptyRunQueue(cap)) { scheduleCheckBlackHoles(cap); }
677 scheduleCheckWakeupThreads(cap);
679 scheduleCheckBlockedThreads(cap);
681 #if defined(THREADED_RTS) || defined(PARALLEL_HASKELL)
682 if (emptyRunQueue(cap)) { scheduleActivateSpark(cap); }
685 #if defined(PARALLEL_HASKELL)
686 // if messages have been buffered...
687 scheduleSendPendingMessages();
690 #if defined(PARALLEL_HASKELL)
691 if (emptyRunQueue(cap)) {
692 receivedFinish = scheduleGetRemoteWork(cap);
693 continue; // a new round, (hopefully) with new work
695 in GUM, this a) sends out a FISH and returns IF no fish is
697 b) (blocking) awaits and receives messages
699 in Eden, this is only the blocking receive, as b) in GUM.
705 #if defined(THREADED_RTS)
706 STATIC_INLINE rtsBool
707 shouldYieldCapability (Capability *cap, Task *task)
709 // we need to yield this capability to someone else if..
710 // - another thread is initiating a GC
711 // - another Task is returning from a foreign call
712 // - the thread at the head of the run queue cannot be run
713 // by this Task (it is bound to another Task, or it is unbound
714 // and this task it bound).
715 return (waiting_for_gc ||
716 cap->returning_tasks_hd != NULL ||
717 (!emptyRunQueue(cap) && (task->tso == NULL
718 ? cap->run_queue_hd->bound != NULL
719 : cap->run_queue_hd->bound != task)));
722 // This is the single place where a Task goes to sleep. There are
723 // two reasons it might need to sleep:
724 // - there are no threads to run
725 // - we need to yield this Capability to someone else
726 // (see shouldYieldCapability())
728 // Careful: the scheduler loop is quite delicate. Make sure you run
729 // the tests in testsuite/concurrent (all ways) after modifying this,
730 // and also check the benchmarks in nofib/parallel for regressions.
733 scheduleYield (Capability **pcap, Task *task)
735 Capability *cap = *pcap;
737 // if we have work, and we don't need to give up the Capability, continue.
738 if (!shouldYieldCapability(cap,task) &&
739 (!emptyRunQueue(cap) ||
740 blackholes_need_checking ||
741 sched_state >= SCHED_INTERRUPTING))
744 // otherwise yield (sleep), and keep yielding if necessary.
746 yieldCapability(&cap,task);
748 while (shouldYieldCapability(cap,task));
750 // note there may still be no threads on the run queue at this
751 // point, the caller has to check.
758 /* -----------------------------------------------------------------------------
761 * Push work to other Capabilities if we have some.
762 * -------------------------------------------------------------------------- */
765 schedulePushWork(Capability *cap USED_IF_THREADS,
766 Task *task USED_IF_THREADS)
768 /* following code not for PARALLEL_HASKELL. I kept the call general,
769 future GUM versions might use pushing in a distributed setup */
770 #if defined(THREADED_RTS)
772 Capability *free_caps[n_capabilities], *cap0;
775 // migration can be turned off with +RTS -qg
776 if (!RtsFlags.ParFlags.migrate) return;
778 // Check whether we have more threads on our run queue, or sparks
779 // in our pool, that we could hand to another Capability.
780 if ((emptyRunQueue(cap) || cap->run_queue_hd->_link == END_TSO_QUEUE)
781 && sparkPoolSizeCap(cap) < 2) {
785 // First grab as many free Capabilities as we can.
786 for (i=0, n_free_caps=0; i < n_capabilities; i++) {
787 cap0 = &capabilities[i];
788 if (cap != cap0 && tryGrabCapability(cap0,task)) {
789 if (!emptyRunQueue(cap0) || cap->returning_tasks_hd != NULL) {
790 // it already has some work, we just grabbed it at
791 // the wrong moment. Or maybe it's deadlocked!
792 releaseCapability(cap0);
794 free_caps[n_free_caps++] = cap0;
799 // we now have n_free_caps free capabilities stashed in
800 // free_caps[]. Share our run queue equally with them. This is
801 // probably the simplest thing we could do; improvements we might
802 // want to do include:
804 // - giving high priority to moving relatively new threads, on
805 // the gournds that they haven't had time to build up a
806 // working set in the cache on this CPU/Capability.
808 // - giving low priority to moving long-lived threads
810 if (n_free_caps > 0) {
811 StgTSO *prev, *t, *next;
812 rtsBool pushed_to_all;
814 debugTrace(DEBUG_sched,
815 "cap %d: %s and %d free capabilities, sharing...",
817 (!emptyRunQueue(cap) && cap->run_queue_hd->_link != END_TSO_QUEUE)?
818 "excess threads on run queue":"sparks to share (>=2)",
822 pushed_to_all = rtsFalse;
824 if (cap->run_queue_hd != END_TSO_QUEUE) {
825 prev = cap->run_queue_hd;
827 prev->_link = END_TSO_QUEUE;
828 for (; t != END_TSO_QUEUE; t = next) {
830 t->_link = END_TSO_QUEUE;
831 if (t->what_next == ThreadRelocated
832 || t->bound == task // don't move my bound thread
833 || tsoLocked(t)) { // don't move a locked thread
834 setTSOLink(cap, prev, t);
836 } else if (i == n_free_caps) {
837 pushed_to_all = rtsTrue;
840 setTSOLink(cap, prev, t);
843 debugTrace(DEBUG_sched, "pushing thread %lu to capability %d", (unsigned long)t->id, free_caps[i]->no);
844 appendToRunQueue(free_caps[i],t);
845 if (t->bound) { t->bound->cap = free_caps[i]; }
846 t->cap = free_caps[i];
850 cap->run_queue_tl = prev;
854 /* JB I left this code in place, it would work but is not necessary */
856 // If there are some free capabilities that we didn't push any
857 // threads to, then try to push a spark to each one.
858 if (!pushed_to_all) {
860 // i is the next free capability to push to
861 for (; i < n_free_caps; i++) {
862 if (emptySparkPoolCap(free_caps[i])) {
863 spark = tryStealSpark(cap->sparks);
865 debugTrace(DEBUG_sched, "pushing spark %p to capability %d", spark, free_caps[i]->no);
866 newSpark(&(free_caps[i]->r), spark);
871 #endif /* SPARK_PUSHING */
873 // release the capabilities
874 for (i = 0; i < n_free_caps; i++) {
875 task->cap = free_caps[i];
876 releaseAndWakeupCapability(free_caps[i]);
879 task->cap = cap; // reset to point to our Capability.
881 #endif /* THREADED_RTS */
885 /* ----------------------------------------------------------------------------
886 * Start any pending signal handlers
887 * ------------------------------------------------------------------------- */
889 #if defined(RTS_USER_SIGNALS) && !defined(THREADED_RTS)
891 scheduleStartSignalHandlers(Capability *cap)
893 if (RtsFlags.MiscFlags.install_signal_handlers && signals_pending()) {
894 // safe outside the lock
895 startSignalHandlers(cap);
900 scheduleStartSignalHandlers(Capability *cap STG_UNUSED)
905 /* ----------------------------------------------------------------------------
906 * Check for blocked threads that can be woken up.
907 * ------------------------------------------------------------------------- */
910 scheduleCheckBlockedThreads(Capability *cap USED_IF_NOT_THREADS)
912 #if !defined(THREADED_RTS)
914 // Check whether any waiting threads need to be woken up. If the
915 // run queue is empty, and there are no other tasks running, we
916 // can wait indefinitely for something to happen.
918 if ( !emptyQueue(blocked_queue_hd) || !emptyQueue(sleeping_queue) )
920 awaitEvent( emptyRunQueue(cap) && !blackholes_need_checking );
926 /* ----------------------------------------------------------------------------
927 * Check for threads woken up by other Capabilities
928 * ------------------------------------------------------------------------- */
931 scheduleCheckWakeupThreads(Capability *cap USED_IF_THREADS)
933 #if defined(THREADED_RTS)
934 // Any threads that were woken up by other Capabilities get
935 // appended to our run queue.
936 if (!emptyWakeupQueue(cap)) {
937 ACQUIRE_LOCK(&cap->lock);
938 if (emptyRunQueue(cap)) {
939 cap->run_queue_hd = cap->wakeup_queue_hd;
940 cap->run_queue_tl = cap->wakeup_queue_tl;
942 setTSOLink(cap, cap->run_queue_tl, cap->wakeup_queue_hd);
943 cap->run_queue_tl = cap->wakeup_queue_tl;
945 cap->wakeup_queue_hd = cap->wakeup_queue_tl = END_TSO_QUEUE;
946 RELEASE_LOCK(&cap->lock);
951 /* ----------------------------------------------------------------------------
952 * Check for threads blocked on BLACKHOLEs that can be woken up
953 * ------------------------------------------------------------------------- */
955 scheduleCheckBlackHoles (Capability *cap)
957 if ( blackholes_need_checking ) // check without the lock first
959 ACQUIRE_LOCK(&sched_mutex);
960 if ( blackholes_need_checking ) {
961 blackholes_need_checking = rtsFalse;
962 // important that we reset the flag *before* checking the
963 // blackhole queue, otherwise we could get deadlock. This
964 // happens as follows: we wake up a thread that
965 // immediately runs on another Capability, blocks on a
966 // blackhole, and then we reset the blackholes_need_checking flag.
967 checkBlackHoles(cap);
969 RELEASE_LOCK(&sched_mutex);
973 /* ----------------------------------------------------------------------------
974 * Detect deadlock conditions and attempt to resolve them.
975 * ------------------------------------------------------------------------- */
978 scheduleDetectDeadlock (Capability *cap, Task *task)
981 #if defined(PARALLEL_HASKELL)
982 // ToDo: add deadlock detection in GUM (similar to THREADED_RTS) -- HWL
987 * Detect deadlock: when we have no threads to run, there are no
988 * threads blocked, waiting for I/O, or sleeping, and all the
989 * other tasks are waiting for work, we must have a deadlock of
992 if ( emptyThreadQueues(cap) )
994 #if defined(THREADED_RTS)
996 * In the threaded RTS, we only check for deadlock if there
997 * has been no activity in a complete timeslice. This means
998 * we won't eagerly start a full GC just because we don't have
999 * any threads to run currently.
1001 if (recent_activity != ACTIVITY_INACTIVE) return;
1004 debugTrace(DEBUG_sched, "deadlocked, forcing major GC...");
1006 // Garbage collection can release some new threads due to
1007 // either (a) finalizers or (b) threads resurrected because
1008 // they are unreachable and will therefore be sent an
1009 // exception. Any threads thus released will be immediately
1011 cap = scheduleDoGC (cap, task, rtsTrue/*force major GC*/);
1012 // when force_major == rtsTrue. scheduleDoGC sets
1013 // recent_activity to ACTIVITY_DONE_GC and turns off the timer
1016 if ( !emptyRunQueue(cap) ) return;
1018 #if defined(RTS_USER_SIGNALS) && !defined(THREADED_RTS)
1019 /* If we have user-installed signal handlers, then wait
1020 * for signals to arrive rather then bombing out with a
1023 if ( RtsFlags.MiscFlags.install_signal_handlers && anyUserHandlers() ) {
1024 debugTrace(DEBUG_sched,
1025 "still deadlocked, waiting for signals...");
1029 if (signals_pending()) {
1030 startSignalHandlers(cap);
1033 // either we have threads to run, or we were interrupted:
1034 ASSERT(!emptyRunQueue(cap) || sched_state >= SCHED_INTERRUPTING);
1040 #if !defined(THREADED_RTS)
1041 /* Probably a real deadlock. Send the current main thread the
1042 * Deadlock exception.
1045 switch (task->tso->why_blocked) {
1047 case BlockedOnBlackHole:
1048 case BlockedOnException:
1050 throwToSingleThreaded(cap, task->tso,
1051 (StgClosure *)nonTermination_closure);
1054 barf("deadlock: main thread blocked in a strange way");
1063 /* ----------------------------------------------------------------------------
1064 * Send pending messages (PARALLEL_HASKELL only)
1065 * ------------------------------------------------------------------------- */
1067 #if defined(PARALLEL_HASKELL)
1069 scheduleSendPendingMessages(void)
1072 # if defined(PAR) // global Mem.Mgmt., omit for now
1073 if (PendingFetches != END_BF_QUEUE) {
1078 if (RtsFlags.ParFlags.BufferTime) {
1079 // if we use message buffering, we must send away all message
1080 // packets which have become too old...
1086 /* ----------------------------------------------------------------------------
1087 * Activate spark threads (PARALLEL_HASKELL and THREADED_RTS)
1088 * ------------------------------------------------------------------------- */
1090 #if defined(PARALLEL_HASKELL) || defined(THREADED_RTS)
1092 scheduleActivateSpark(Capability *cap)
1096 createSparkThread(cap);
1097 debugTrace(DEBUG_sched, "creating a spark thread");
1100 #endif // PARALLEL_HASKELL || THREADED_RTS
1102 /* ----------------------------------------------------------------------------
1103 * Get work from a remote node (PARALLEL_HASKELL only)
1104 * ------------------------------------------------------------------------- */
1106 #if defined(PARALLEL_HASKELL)
1107 static rtsBool /* return value used in PARALLEL_HASKELL only */
1108 scheduleGetRemoteWork (Capability *cap STG_UNUSED)
1110 #if defined(PARALLEL_HASKELL)
1111 rtsBool receivedFinish = rtsFalse;
1113 // idle() , i.e. send all buffers, wait for work
1114 if (RtsFlags.ParFlags.BufferTime) {
1115 IF_PAR_DEBUG(verbose,
1116 debugBelch("...send all pending data,"));
1119 for (i=1; i<=nPEs; i++)
1120 sendImmediately(i); // send all messages away immediately
1124 /* this would be the place for fishing in GUM...
1126 if (no-earlier-fish-around)
1127 sendFish(choosePe());
1130 // Eden:just look for incoming messages (blocking receive)
1131 IF_PAR_DEBUG(verbose,
1132 debugBelch("...wait for incoming messages...\n"));
1133 processMessages(cap, &receivedFinish); // blocking receive...
1136 return receivedFinish;
1137 // reenter scheduling look after having received something
1139 #else /* !PARALLEL_HASKELL, i.e. THREADED_RTS */
1141 return rtsFalse; /* return value unused in THREADED_RTS */
1143 #endif /* PARALLEL_HASKELL */
1145 #endif // PARALLEL_HASKELL || THREADED_RTS
1147 /* ----------------------------------------------------------------------------
1148 * After running a thread...
1149 * ------------------------------------------------------------------------- */
1152 schedulePostRunThread (Capability *cap, StgTSO *t)
1154 // We have to be able to catch transactions that are in an
1155 // infinite loop as a result of seeing an inconsistent view of
1159 // [a,b] <- mapM readTVar [ta,tb]
1160 // when (a == b) loop
1162 // and a is never equal to b given a consistent view of memory.
1164 if (t -> trec != NO_TREC && t -> why_blocked == NotBlocked) {
1165 if (!stmValidateNestOfTransactions (t -> trec)) {
1166 debugTrace(DEBUG_sched | DEBUG_stm,
1167 "trec %p found wasting its time", t);
1169 // strip the stack back to the
1170 // ATOMICALLY_FRAME, aborting the (nested)
1171 // transaction, and saving the stack of any
1172 // partially-evaluated thunks on the heap.
1173 throwToSingleThreaded_(cap, t, NULL, rtsTrue);
1175 ASSERT(get_itbl((StgClosure *)t->sp)->type == ATOMICALLY_FRAME);
1179 /* some statistics gathering in the parallel case */
1182 /* -----------------------------------------------------------------------------
1183 * Handle a thread that returned to the scheduler with ThreadHeepOverflow
1184 * -------------------------------------------------------------------------- */
1187 scheduleHandleHeapOverflow( Capability *cap, StgTSO *t )
1189 // did the task ask for a large block?
1190 if (cap->r.rHpAlloc > BLOCK_SIZE) {
1191 // if so, get one and push it on the front of the nursery.
1195 blocks = (lnat)BLOCK_ROUND_UP(cap->r.rHpAlloc) / BLOCK_SIZE;
1197 debugTrace(DEBUG_sched,
1198 "--<< thread %ld (%s) stopped: requesting a large block (size %ld)\n",
1199 (long)t->id, whatNext_strs[t->what_next], blocks);
1201 // don't do this if the nursery is (nearly) full, we'll GC first.
1202 if (cap->r.rCurrentNursery->link != NULL ||
1203 cap->r.rNursery->n_blocks == 1) { // paranoia to prevent infinite loop
1204 // if the nursery has only one block.
1207 bd = allocGroup( blocks );
1209 cap->r.rNursery->n_blocks += blocks;
1211 // link the new group into the list
1212 bd->link = cap->r.rCurrentNursery;
1213 bd->u.back = cap->r.rCurrentNursery->u.back;
1214 if (cap->r.rCurrentNursery->u.back != NULL) {
1215 cap->r.rCurrentNursery->u.back->link = bd;
1217 #if !defined(THREADED_RTS)
1218 ASSERT(g0s0->blocks == cap->r.rCurrentNursery &&
1219 g0s0 == cap->r.rNursery);
1221 cap->r.rNursery->blocks = bd;
1223 cap->r.rCurrentNursery->u.back = bd;
1225 // initialise it as a nursery block. We initialise the
1226 // step, gen_no, and flags field of *every* sub-block in
1227 // this large block, because this is easier than making
1228 // sure that we always find the block head of a large
1229 // block whenever we call Bdescr() (eg. evacuate() and
1230 // isAlive() in the GC would both have to do this, at
1234 for (x = bd; x < bd + blocks; x++) {
1235 x->step = cap->r.rNursery;
1241 // This assert can be a killer if the app is doing lots
1242 // of large block allocations.
1243 IF_DEBUG(sanity, checkNurserySanity(cap->r.rNursery));
1245 // now update the nursery to point to the new block
1246 cap->r.rCurrentNursery = bd;
1248 // we might be unlucky and have another thread get on the
1249 // run queue before us and steal the large block, but in that
1250 // case the thread will just end up requesting another large
1252 pushOnRunQueue(cap,t);
1253 return rtsFalse; /* not actually GC'ing */
1257 debugTrace(DEBUG_sched,
1258 "--<< thread %ld (%s) stopped: HeapOverflow",
1259 (long)t->id, whatNext_strs[t->what_next]);
1261 if (cap->context_switch) {
1262 // Sometimes we miss a context switch, e.g. when calling
1263 // primitives in a tight loop, MAYBE_GC() doesn't check the
1264 // context switch flag, and we end up waiting for a GC.
1265 // See #1984, and concurrent/should_run/1984
1266 cap->context_switch = 0;
1267 addToRunQueue(cap,t);
1269 pushOnRunQueue(cap,t);
1272 /* actual GC is done at the end of the while loop in schedule() */
1275 /* -----------------------------------------------------------------------------
1276 * Handle a thread that returned to the scheduler with ThreadStackOverflow
1277 * -------------------------------------------------------------------------- */
1280 scheduleHandleStackOverflow (Capability *cap, Task *task, StgTSO *t)
1282 debugTrace (DEBUG_sched,
1283 "--<< thread %ld (%s) stopped, StackOverflow",
1284 (long)t->id, whatNext_strs[t->what_next]);
1286 /* just adjust the stack for this thread, then pop it back
1290 /* enlarge the stack */
1291 StgTSO *new_t = threadStackOverflow(cap, t);
1293 /* The TSO attached to this Task may have moved, so update the
1296 if (task->tso == t) {
1299 pushOnRunQueue(cap,new_t);
1303 /* -----------------------------------------------------------------------------
1304 * Handle a thread that returned to the scheduler with ThreadYielding
1305 * -------------------------------------------------------------------------- */
1308 scheduleHandleYield( Capability *cap, StgTSO *t, nat prev_what_next )
1310 // Reset the context switch flag. We don't do this just before
1311 // running the thread, because that would mean we would lose ticks
1312 // during GC, which can lead to unfair scheduling (a thread hogs
1313 // the CPU because the tick always arrives during GC). This way
1314 // penalises threads that do a lot of allocation, but that seems
1315 // better than the alternative.
1316 cap->context_switch = 0;
1318 /* put the thread back on the run queue. Then, if we're ready to
1319 * GC, check whether this is the last task to stop. If so, wake
1320 * up the GC thread. getThread will block during a GC until the
1324 if (t->what_next != prev_what_next) {
1325 debugTrace(DEBUG_sched,
1326 "--<< thread %ld (%s) stopped to switch evaluators",
1327 (long)t->id, whatNext_strs[t->what_next]);
1329 debugTrace(DEBUG_sched,
1330 "--<< thread %ld (%s) stopped, yielding",
1331 (long)t->id, whatNext_strs[t->what_next]);
1336 //debugBelch("&& Doing sanity check on yielding TSO %ld.", t->id);
1338 ASSERT(t->_link == END_TSO_QUEUE);
1340 // Shortcut if we're just switching evaluators: don't bother
1341 // doing stack squeezing (which can be expensive), just run the
1343 if (t->what_next != prev_what_next) {
1347 addToRunQueue(cap,t);
1352 /* -----------------------------------------------------------------------------
1353 * Handle a thread that returned to the scheduler with ThreadBlocked
1354 * -------------------------------------------------------------------------- */
1357 scheduleHandleThreadBlocked( StgTSO *t
1358 #if !defined(GRAN) && !defined(DEBUG)
1364 // We don't need to do anything. The thread is blocked, and it
1365 // has tidied up its stack and placed itself on whatever queue
1366 // it needs to be on.
1368 // ASSERT(t->why_blocked != NotBlocked);
1369 // Not true: for example,
1370 // - in THREADED_RTS, the thread may already have been woken
1371 // up by another Capability. This actually happens: try
1372 // conc023 +RTS -N2.
1373 // - the thread may have woken itself up already, because
1374 // threadPaused() might have raised a blocked throwTo
1375 // exception, see maybePerformBlockedException().
1378 if (traceClass(DEBUG_sched)) {
1379 debugTraceBegin("--<< thread %lu (%s) stopped: ",
1380 (unsigned long)t->id, whatNext_strs[t->what_next]);
1381 printThreadBlockage(t);
1387 /* -----------------------------------------------------------------------------
1388 * Handle a thread that returned to the scheduler with ThreadFinished
1389 * -------------------------------------------------------------------------- */
1392 scheduleHandleThreadFinished (Capability *cap STG_UNUSED, Task *task, StgTSO *t)
1394 /* Need to check whether this was a main thread, and if so,
1395 * return with the return value.
1397 * We also end up here if the thread kills itself with an
1398 * uncaught exception, see Exception.cmm.
1400 debugTrace(DEBUG_sched, "--++ thread %lu (%s) finished",
1401 (unsigned long)t->id, whatNext_strs[t->what_next]);
1404 // Check whether the thread that just completed was a bound
1405 // thread, and if so return with the result.
1407 // There is an assumption here that all thread completion goes
1408 // through this point; we need to make sure that if a thread
1409 // ends up in the ThreadKilled state, that it stays on the run
1410 // queue so it can be dealt with here.
1415 if (t->bound != task) {
1416 #if !defined(THREADED_RTS)
1417 // Must be a bound thread that is not the topmost one. Leave
1418 // it on the run queue until the stack has unwound to the
1419 // point where we can deal with this. Leaving it on the run
1420 // queue also ensures that the garbage collector knows about
1421 // this thread and its return value (it gets dropped from the
1422 // step->threads list so there's no other way to find it).
1423 appendToRunQueue(cap,t);
1426 // this cannot happen in the threaded RTS, because a
1427 // bound thread can only be run by the appropriate Task.
1428 barf("finished bound thread that isn't mine");
1432 ASSERT(task->tso == t);
1434 if (t->what_next == ThreadComplete) {
1436 // NOTE: return val is tso->sp[1] (see StgStartup.hc)
1437 *(task->ret) = (StgClosure *)task->tso->sp[1];
1439 task->stat = Success;
1442 *(task->ret) = NULL;
1444 if (sched_state >= SCHED_INTERRUPTING) {
1445 if (heap_overflow) {
1446 task->stat = HeapExhausted;
1448 task->stat = Interrupted;
1451 task->stat = Killed;
1455 removeThreadLabel((StgWord)task->tso->id);
1457 return rtsTrue; // tells schedule() to return
1463 /* -----------------------------------------------------------------------------
1464 * Perform a heap census
1465 * -------------------------------------------------------------------------- */
1468 scheduleNeedHeapProfile( rtsBool ready_to_gc STG_UNUSED )
1470 // When we have +RTS -i0 and we're heap profiling, do a census at
1471 // every GC. This lets us get repeatable runs for debugging.
1472 if (performHeapProfile ||
1473 (RtsFlags.ProfFlags.profileInterval==0 &&
1474 RtsFlags.ProfFlags.doHeapProfile && ready_to_gc)) {
1481 /* -----------------------------------------------------------------------------
1482 * Perform a garbage collection if necessary
1483 * -------------------------------------------------------------------------- */
1486 scheduleDoGC (Capability *cap, Task *task USED_IF_THREADS, rtsBool force_major)
1488 rtsBool heap_census;
1490 /* extern static volatile StgWord waiting_for_gc;
1491 lives inside capability.c */
1492 rtsBool gc_type, prev_pending_gc;
1496 if (sched_state == SCHED_SHUTTING_DOWN) {
1497 // The final GC has already been done, and the system is
1498 // shutting down. We'll probably deadlock if we try to GC
1504 if (sched_state < SCHED_INTERRUPTING
1505 && RtsFlags.ParFlags.parGcEnabled
1506 && N >= RtsFlags.ParFlags.parGcGen
1507 && ! oldest_gen->steps[0].mark)
1509 gc_type = PENDING_GC_PAR;
1511 gc_type = PENDING_GC_SEQ;
1514 // In order to GC, there must be no threads running Haskell code.
1515 // Therefore, the GC thread needs to hold *all* the capabilities,
1516 // and release them after the GC has completed.
1518 // This seems to be the simplest way: previous attempts involved
1519 // making all the threads with capabilities give up their
1520 // capabilities and sleep except for the *last* one, which
1521 // actually did the GC. But it's quite hard to arrange for all
1522 // the other tasks to sleep and stay asleep.
1525 /* Other capabilities are prevented from running yet more Haskell
1526 threads if waiting_for_gc is set. Tested inside
1527 yieldCapability() and releaseCapability() in Capability.c */
1529 prev_pending_gc = cas(&waiting_for_gc, 0, gc_type);
1530 if (prev_pending_gc) {
1532 debugTrace(DEBUG_sched, "someone else is trying to GC (%d)...",
1535 yieldCapability(&cap,task);
1536 } while (waiting_for_gc);
1537 return cap; // NOTE: task->cap might have changed here
1540 setContextSwitches();
1542 // The final shutdown GC is always single-threaded, because it's
1543 // possible that some of the Capabilities have no worker threads.
1545 if (gc_type == PENDING_GC_SEQ)
1547 // single-threaded GC: grab all the capabilities
1548 for (i=0; i < n_capabilities; i++) {
1549 debugTrace(DEBUG_sched, "ready_to_gc, grabbing all the capabilies (%d/%d)", i, n_capabilities);
1550 if (cap != &capabilities[i]) {
1551 Capability *pcap = &capabilities[i];
1552 // we better hope this task doesn't get migrated to
1553 // another Capability while we're waiting for this one.
1554 // It won't, because load balancing happens while we have
1555 // all the Capabilities, but even so it's a slightly
1556 // unsavoury invariant.
1558 waitForReturnCapability(&pcap, task);
1559 if (pcap != &capabilities[i]) {
1560 barf("scheduleDoGC: got the wrong capability");
1567 // multi-threaded GC: make sure all the Capabilities donate one
1569 debugTrace(DEBUG_sched, "ready_to_gc, grabbing GC threads");
1571 waitForGcThreads(cap);
1575 // so this happens periodically:
1576 if (cap) scheduleCheckBlackHoles(cap);
1578 IF_DEBUG(scheduler, printAllThreads());
1580 delete_threads_and_gc:
1582 * We now have all the capabilities; if we're in an interrupting
1583 * state, then we should take the opportunity to delete all the
1584 * threads in the system.
1586 if (sched_state == SCHED_INTERRUPTING) {
1587 deleteAllThreads(cap);
1588 sched_state = SCHED_SHUTTING_DOWN;
1591 heap_census = scheduleNeedHeapProfile(rtsTrue);
1593 #if defined(THREADED_RTS)
1594 debugTrace(DEBUG_sched, "doing GC");
1595 // reset waiting_for_gc *before* GC, so that when the GC threads
1596 // emerge they don't immediately re-enter the GC.
1598 GarbageCollect(force_major || heap_census, gc_type, cap);
1600 GarbageCollect(force_major || heap_census, 0, cap);
1604 debugTrace(DEBUG_sched, "performing heap census");
1606 performHeapProfile = rtsFalse;
1609 if (heap_overflow && sched_state < SCHED_INTERRUPTING) {
1610 // GC set the heap_overflow flag, so we should proceed with
1611 // an orderly shutdown now. Ultimately we want the main
1612 // thread to return to its caller with HeapExhausted, at which
1613 // point the caller should call hs_exit(). The first step is
1614 // to delete all the threads.
1616 // Another way to do this would be to raise an exception in
1617 // the main thread, which we really should do because it gives
1618 // the program a chance to clean up. But how do we find the
1619 // main thread? It should presumably be the same one that
1620 // gets ^C exceptions, but that's all done on the Haskell side
1621 // (GHC.TopHandler).
1622 sched_state = SCHED_INTERRUPTING;
1623 goto delete_threads_and_gc;
1628 Once we are all together... this would be the place to balance all
1629 spark pools. No concurrent stealing or adding of new sparks can
1630 occur. Should be defined in Sparks.c. */
1631 balanceSparkPoolsCaps(n_capabilities, capabilities);
1636 // We've just done a major GC and we don't need the timer
1637 // signal turned on any more (#1623).
1638 // NB. do this *before* releasing the Capabilities, to avoid
1640 recent_activity = ACTIVITY_DONE_GC;
1644 #if defined(THREADED_RTS)
1645 if (gc_type == PENDING_GC_SEQ) {
1646 // release our stash of capabilities.
1647 for (i = 0; i < n_capabilities; i++) {
1648 if (cap != &capabilities[i]) {
1649 task->cap = &capabilities[i];
1650 releaseCapability(&capabilities[i]);
1664 /* ---------------------------------------------------------------------------
1665 * Singleton fork(). Do not copy any running threads.
1666 * ------------------------------------------------------------------------- */
1669 forkProcess(HsStablePtr *entry
1670 #ifndef FORKPROCESS_PRIMOP_SUPPORTED
1675 #ifdef FORKPROCESS_PRIMOP_SUPPORTED
1682 #if defined(THREADED_RTS)
1683 if (RtsFlags.ParFlags.nNodes > 1) {
1684 errorBelch("forking not supported with +RTS -N<n> greater than 1");
1685 stg_exit(EXIT_FAILURE);
1689 debugTrace(DEBUG_sched, "forking!");
1691 // ToDo: for SMP, we should probably acquire *all* the capabilities
1694 // no funny business: hold locks while we fork, otherwise if some
1695 // other thread is holding a lock when the fork happens, the data
1696 // structure protected by the lock will forever be in an
1697 // inconsistent state in the child. See also #1391.
1698 ACQUIRE_LOCK(&sched_mutex);
1699 ACQUIRE_LOCK(&cap->lock);
1700 ACQUIRE_LOCK(&cap->running_task->lock);
1704 if (pid) { // parent
1706 RELEASE_LOCK(&sched_mutex);
1707 RELEASE_LOCK(&cap->lock);
1708 RELEASE_LOCK(&cap->running_task->lock);
1710 // just return the pid
1716 #if defined(THREADED_RTS)
1717 initMutex(&sched_mutex);
1718 initMutex(&cap->lock);
1719 initMutex(&cap->running_task->lock);
1722 // Now, all OS threads except the thread that forked are
1723 // stopped. We need to stop all Haskell threads, including
1724 // those involved in foreign calls. Also we need to delete
1725 // all Tasks, because they correspond to OS threads that are
1728 for (s = 0; s < total_steps; s++) {
1729 for (t = all_steps[s].threads; t != END_TSO_QUEUE; t = next) {
1730 if (t->what_next == ThreadRelocated) {
1733 next = t->global_link;
1734 // don't allow threads to catch the ThreadKilled
1735 // exception, but we do want to raiseAsync() because these
1736 // threads may be evaluating thunks that we need later.
1737 deleteThread_(cap,t);
1742 // Empty the run queue. It seems tempting to let all the
1743 // killed threads stay on the run queue as zombies to be
1744 // cleaned up later, but some of them correspond to bound
1745 // threads for which the corresponding Task does not exist.
1746 cap->run_queue_hd = END_TSO_QUEUE;
1747 cap->run_queue_tl = END_TSO_QUEUE;
1749 // Any suspended C-calling Tasks are no more, their OS threads
1751 cap->suspended_ccalling_tasks = NULL;
1753 // Empty the threads lists. Otherwise, the garbage
1754 // collector may attempt to resurrect some of these threads.
1755 for (s = 0; s < total_steps; s++) {
1756 all_steps[s].threads = END_TSO_QUEUE;
1759 // Wipe the task list, except the current Task.
1760 ACQUIRE_LOCK(&sched_mutex);
1761 for (task = all_tasks; task != NULL; task=task->all_link) {
1762 if (task != cap->running_task) {
1763 #if defined(THREADED_RTS)
1764 initMutex(&task->lock); // see #1391
1769 RELEASE_LOCK(&sched_mutex);
1771 #if defined(THREADED_RTS)
1772 // Wipe our spare workers list, they no longer exist. New
1773 // workers will be created if necessary.
1774 cap->spare_workers = NULL;
1775 cap->returning_tasks_hd = NULL;
1776 cap->returning_tasks_tl = NULL;
1779 // On Unix, all timers are reset in the child, so we need to start
1784 cap = rts_evalStableIO(cap, entry, NULL); // run the action
1785 rts_checkSchedStatus("forkProcess",cap);
1788 hs_exit(); // clean up and exit
1789 stg_exit(EXIT_SUCCESS);
1791 #else /* !FORKPROCESS_PRIMOP_SUPPORTED */
1792 barf("forkProcess#: primop not supported on this platform, sorry!\n");
1797 /* ---------------------------------------------------------------------------
1798 * Delete all the threads in the system
1799 * ------------------------------------------------------------------------- */
1802 deleteAllThreads ( Capability *cap )
1804 // NOTE: only safe to call if we own all capabilities.
1809 debugTrace(DEBUG_sched,"deleting all threads");
1810 for (s = 0; s < total_steps; s++) {
1811 for (t = all_steps[s].threads; t != END_TSO_QUEUE; t = next) {
1812 if (t->what_next == ThreadRelocated) {
1815 next = t->global_link;
1816 deleteThread(cap,t);
1821 // The run queue now contains a bunch of ThreadKilled threads. We
1822 // must not throw these away: the main thread(s) will be in there
1823 // somewhere, and the main scheduler loop has to deal with it.
1824 // Also, the run queue is the only thing keeping these threads from
1825 // being GC'd, and we don't want the "main thread has been GC'd" panic.
1827 #if !defined(THREADED_RTS)
1828 ASSERT(blocked_queue_hd == END_TSO_QUEUE);
1829 ASSERT(sleeping_queue == END_TSO_QUEUE);
1833 /* -----------------------------------------------------------------------------
1834 Managing the suspended_ccalling_tasks list.
1835 Locks required: sched_mutex
1836 -------------------------------------------------------------------------- */
1839 suspendTask (Capability *cap, Task *task)
1841 ASSERT(task->next == NULL && task->prev == NULL);
1842 task->next = cap->suspended_ccalling_tasks;
1844 if (cap->suspended_ccalling_tasks) {
1845 cap->suspended_ccalling_tasks->prev = task;
1847 cap->suspended_ccalling_tasks = task;
1851 recoverSuspendedTask (Capability *cap, Task *task)
1854 task->prev->next = task->next;
1856 ASSERT(cap->suspended_ccalling_tasks == task);
1857 cap->suspended_ccalling_tasks = task->next;
1860 task->next->prev = task->prev;
1862 task->next = task->prev = NULL;
1865 /* ---------------------------------------------------------------------------
1866 * Suspending & resuming Haskell threads.
1868 * When making a "safe" call to C (aka _ccall_GC), the task gives back
1869 * its capability before calling the C function. This allows another
1870 * task to pick up the capability and carry on running Haskell
1871 * threads. It also means that if the C call blocks, it won't lock
1874 * The Haskell thread making the C call is put to sleep for the
1875 * duration of the call, on the susepended_ccalling_threads queue. We
1876 * give out a token to the task, which it can use to resume the thread
1877 * on return from the C function.
1878 * ------------------------------------------------------------------------- */
1881 suspendThread (StgRegTable *reg)
1888 StgWord32 saved_winerror;
1891 saved_errno = errno;
1893 saved_winerror = GetLastError();
1896 /* assume that *reg is a pointer to the StgRegTable part of a Capability.
1898 cap = regTableToCapability(reg);
1900 task = cap->running_task;
1901 tso = cap->r.rCurrentTSO;
1903 debugTrace(DEBUG_sched,
1904 "thread %lu did a safe foreign call",
1905 (unsigned long)cap->r.rCurrentTSO->id);
1907 // XXX this might not be necessary --SDM
1908 tso->what_next = ThreadRunGHC;
1910 threadPaused(cap,tso);
1912 if ((tso->flags & TSO_BLOCKEX) == 0) {
1913 tso->why_blocked = BlockedOnCCall;
1914 tso->flags |= TSO_BLOCKEX;
1915 tso->flags &= ~TSO_INTERRUPTIBLE;
1917 tso->why_blocked = BlockedOnCCall_NoUnblockExc;
1920 // Hand back capability
1921 task->suspended_tso = tso;
1923 ACQUIRE_LOCK(&cap->lock);
1925 suspendTask(cap,task);
1926 cap->in_haskell = rtsFalse;
1927 releaseCapability_(cap,rtsFalse);
1929 RELEASE_LOCK(&cap->lock);
1931 #if defined(THREADED_RTS)
1932 /* Preparing to leave the RTS, so ensure there's a native thread/task
1933 waiting to take over.
1935 debugTrace(DEBUG_sched, "thread %lu: leaving RTS", (unsigned long)tso->id);
1938 errno = saved_errno;
1940 SetLastError(saved_winerror);
1946 resumeThread (void *task_)
1953 StgWord32 saved_winerror;
1956 saved_errno = errno;
1958 saved_winerror = GetLastError();
1962 // Wait for permission to re-enter the RTS with the result.
1963 waitForReturnCapability(&cap,task);
1964 // we might be on a different capability now... but if so, our
1965 // entry on the suspended_ccalling_tasks list will also have been
1968 // Remove the thread from the suspended list
1969 recoverSuspendedTask(cap,task);
1971 tso = task->suspended_tso;
1972 task->suspended_tso = NULL;
1973 tso->_link = END_TSO_QUEUE; // no write barrier reqd
1974 debugTrace(DEBUG_sched, "thread %lu: re-entering RTS", (unsigned long)tso->id);
1976 if (tso->why_blocked == BlockedOnCCall) {
1977 awakenBlockedExceptionQueue(cap,tso);
1978 tso->flags &= ~(TSO_BLOCKEX | TSO_INTERRUPTIBLE);
1981 /* Reset blocking status */
1982 tso->why_blocked = NotBlocked;
1984 cap->r.rCurrentTSO = tso;
1985 cap->in_haskell = rtsTrue;
1986 errno = saved_errno;
1988 SetLastError(saved_winerror);
1991 /* We might have GC'd, mark the TSO dirty again */
1994 IF_DEBUG(sanity, checkTSO(tso));
1999 /* ---------------------------------------------------------------------------
2002 * scheduleThread puts a thread on the end of the runnable queue.
2003 * This will usually be done immediately after a thread is created.
2004 * The caller of scheduleThread must create the thread using e.g.
2005 * createThread and push an appropriate closure
2006 * on this thread's stack before the scheduler is invoked.
2007 * ------------------------------------------------------------------------ */
2010 scheduleThread(Capability *cap, StgTSO *tso)
2012 // The thread goes at the *end* of the run-queue, to avoid possible
2013 // starvation of any threads already on the queue.
2014 appendToRunQueue(cap,tso);
2018 scheduleThreadOn(Capability *cap, StgWord cpu USED_IF_THREADS, StgTSO *tso)
2020 #if defined(THREADED_RTS)
2021 tso->flags |= TSO_LOCKED; // we requested explicit affinity; don't
2022 // move this thread from now on.
2023 cpu %= RtsFlags.ParFlags.nNodes;
2024 if (cpu == cap->no) {
2025 appendToRunQueue(cap,tso);
2027 wakeupThreadOnCapability(cap, &capabilities[cpu], tso);
2030 appendToRunQueue(cap,tso);
2035 scheduleWaitThread (StgTSO* tso, /*[out]*/HaskellObj* ret, Capability *cap)
2039 // We already created/initialised the Task
2040 task = cap->running_task;
2042 // This TSO is now a bound thread; make the Task and TSO
2043 // point to each other.
2049 task->stat = NoStatus;
2051 appendToRunQueue(cap,tso);
2053 debugTrace(DEBUG_sched, "new bound thread (%lu)", (unsigned long)tso->id);
2055 cap = schedule(cap,task);
2057 ASSERT(task->stat != NoStatus);
2058 ASSERT_FULL_CAPABILITY_INVARIANTS(cap,task);
2060 debugTrace(DEBUG_sched, "bound thread (%lu) finished", (unsigned long)task->tso->id);
2064 /* ----------------------------------------------------------------------------
2066 * ------------------------------------------------------------------------- */
2068 #if defined(THREADED_RTS)
2069 void OSThreadProcAttr
2070 workerStart(Task *task)
2074 // See startWorkerTask().
2075 ACQUIRE_LOCK(&task->lock);
2077 RELEASE_LOCK(&task->lock);
2079 // set the thread-local pointer to the Task:
2082 // schedule() runs without a lock.
2083 cap = schedule(cap,task);
2085 // On exit from schedule(), we have a Capability, but possibly not
2086 // the same one we started with.
2088 // During shutdown, the requirement is that after all the
2089 // Capabilities are shut down, all workers that are shutting down
2090 // have finished workerTaskStop(). This is why we hold on to
2091 // cap->lock until we've finished workerTaskStop() below.
2093 // There may be workers still involved in foreign calls; those
2094 // will just block in waitForReturnCapability() because the
2095 // Capability has been shut down.
2097 ACQUIRE_LOCK(&cap->lock);
2098 releaseCapability_(cap,rtsFalse);
2099 workerTaskStop(task);
2100 RELEASE_LOCK(&cap->lock);
2104 /* ---------------------------------------------------------------------------
2107 * Initialise the scheduler. This resets all the queues - if the
2108 * queues contained any threads, they'll be garbage collected at the
2111 * ------------------------------------------------------------------------ */
2116 #if !defined(THREADED_RTS)
2117 blocked_queue_hd = END_TSO_QUEUE;
2118 blocked_queue_tl = END_TSO_QUEUE;
2119 sleeping_queue = END_TSO_QUEUE;
2122 blackhole_queue = END_TSO_QUEUE;
2124 sched_state = SCHED_RUNNING;
2125 recent_activity = ACTIVITY_YES;
2127 #if defined(THREADED_RTS)
2128 /* Initialise the mutex and condition variables used by
2130 initMutex(&sched_mutex);
2133 ACQUIRE_LOCK(&sched_mutex);
2135 /* A capability holds the state a native thread needs in
2136 * order to execute STG code. At least one capability is
2137 * floating around (only THREADED_RTS builds have more than one).
2143 #if defined(THREADED_RTS) || defined(PARALLEL_HASKELL)
2147 #if defined(THREADED_RTS)
2149 * Eagerly start one worker to run each Capability, except for
2150 * Capability 0. The idea is that we're probably going to start a
2151 * bound thread on Capability 0 pretty soon, so we don't want a
2152 * worker task hogging it.
2157 for (i = 1; i < n_capabilities; i++) {
2158 cap = &capabilities[i];
2159 ACQUIRE_LOCK(&cap->lock);
2160 startWorkerTask(cap, workerStart);
2161 RELEASE_LOCK(&cap->lock);
2166 trace(TRACE_sched, "start: %d capabilities", n_capabilities);
2168 RELEASE_LOCK(&sched_mutex);
2173 rtsBool wait_foreign
2174 #if !defined(THREADED_RTS)
2175 __attribute__((unused))
2178 /* see Capability.c, shutdownCapability() */
2182 #if defined(THREADED_RTS)
2183 ACQUIRE_LOCK(&sched_mutex);
2184 task = newBoundTask();
2185 RELEASE_LOCK(&sched_mutex);
2188 // If we haven't killed all the threads yet, do it now.
2189 if (sched_state < SCHED_SHUTTING_DOWN) {
2190 sched_state = SCHED_INTERRUPTING;
2191 #if defined(THREADED_RTS)
2192 waitForReturnCapability(&task->cap,task);
2193 scheduleDoGC(task->cap,task,rtsFalse);
2194 releaseCapability(task->cap);
2196 scheduleDoGC(&MainCapability,task,rtsFalse);
2199 sched_state = SCHED_SHUTTING_DOWN;
2201 #if defined(THREADED_RTS)
2205 for (i = 0; i < n_capabilities; i++) {
2206 shutdownCapability(&capabilities[i], task, wait_foreign);
2208 boundTaskExiting(task);
2214 freeScheduler( void )
2218 ACQUIRE_LOCK(&sched_mutex);
2219 still_running = freeTaskManager();
2220 // We can only free the Capabilities if there are no Tasks still
2221 // running. We might have a Task about to return from a foreign
2222 // call into waitForReturnCapability(), for example (actually,
2223 // this should be the *only* thing that a still-running Task can
2224 // do at this point, and it will block waiting for the
2226 if (still_running == 0) {
2228 if (n_capabilities != 1) {
2229 stgFree(capabilities);
2232 RELEASE_LOCK(&sched_mutex);
2233 #if defined(THREADED_RTS)
2234 closeMutex(&sched_mutex);
2238 /* -----------------------------------------------------------------------------
2241 This is the interface to the garbage collector from Haskell land.
2242 We provide this so that external C code can allocate and garbage
2243 collect when called from Haskell via _ccall_GC.
2244 -------------------------------------------------------------------------- */
2247 performGC_(rtsBool force_major)
2251 // We must grab a new Task here, because the existing Task may be
2252 // associated with a particular Capability, and chained onto the
2253 // suspended_ccalling_tasks queue.
2254 ACQUIRE_LOCK(&sched_mutex);
2255 task = newBoundTask();
2256 RELEASE_LOCK(&sched_mutex);
2258 waitForReturnCapability(&task->cap,task);
2259 scheduleDoGC(task->cap,task,force_major);
2260 releaseCapability(task->cap);
2261 boundTaskExiting(task);
2267 performGC_(rtsFalse);
2271 performMajorGC(void)
2273 performGC_(rtsTrue);
2276 /* -----------------------------------------------------------------------------
2279 If the thread has reached its maximum stack size, then raise the
2280 StackOverflow exception in the offending thread. Otherwise
2281 relocate the TSO into a larger chunk of memory and adjust its stack
2283 -------------------------------------------------------------------------- */
2286 threadStackOverflow(Capability *cap, StgTSO *tso)
2288 nat new_stack_size, stack_words;
2293 IF_DEBUG(sanity,checkTSO(tso));
2295 // don't allow throwTo() to modify the blocked_exceptions queue
2296 // while we are moving the TSO:
2297 lockClosure((StgClosure *)tso);
2299 if (tso->stack_size >= tso->max_stack_size && !(tso->flags & TSO_BLOCKEX)) {
2300 // NB. never raise a StackOverflow exception if the thread is
2301 // inside Control.Exceptino.block. It is impractical to protect
2302 // against stack overflow exceptions, since virtually anything
2303 // can raise one (even 'catch'), so this is the only sensible
2304 // thing to do here. See bug #767.
2306 debugTrace(DEBUG_gc,
2307 "threadStackOverflow of TSO %ld (%p): stack too large (now %ld; max is %ld)",
2308 (long)tso->id, tso, (long)tso->stack_size, (long)tso->max_stack_size);
2310 /* If we're debugging, just print out the top of the stack */
2311 printStackChunk(tso->sp, stg_min(tso->stack+tso->stack_size,
2314 // Send this thread the StackOverflow exception
2316 throwToSingleThreaded(cap, tso, (StgClosure *)stackOverflow_closure);
2320 /* Try to double the current stack size. If that takes us over the
2321 * maximum stack size for this thread, then use the maximum instead
2322 * (that is, unless we're already at or over the max size and we
2323 * can't raise the StackOverflow exception (see above), in which
2324 * case just double the size). Finally round up so the TSO ends up as
2325 * a whole number of blocks.
2327 if (tso->stack_size >= tso->max_stack_size) {
2328 new_stack_size = tso->stack_size * 2;
2330 new_stack_size = stg_min(tso->stack_size * 2, tso->max_stack_size);
2332 new_tso_size = (lnat)BLOCK_ROUND_UP(new_stack_size * sizeof(W_) +
2333 TSO_STRUCT_SIZE)/sizeof(W_);
2334 new_tso_size = round_to_mblocks(new_tso_size); /* Be MBLOCK-friendly */
2335 new_stack_size = new_tso_size - TSO_STRUCT_SIZEW;
2337 debugTrace(DEBUG_sched,
2338 "increasing stack size from %ld words to %d.",
2339 (long)tso->stack_size, new_stack_size);
2341 dest = (StgTSO *)allocateLocal(cap,new_tso_size);
2342 TICK_ALLOC_TSO(new_stack_size,0);
2344 /* copy the TSO block and the old stack into the new area */
2345 memcpy(dest,tso,TSO_STRUCT_SIZE);
2346 stack_words = tso->stack + tso->stack_size - tso->sp;
2347 new_sp = (P_)dest + new_tso_size - stack_words;
2348 memcpy(new_sp, tso->sp, stack_words * sizeof(W_));
2350 /* relocate the stack pointers... */
2352 dest->stack_size = new_stack_size;
2354 /* Mark the old TSO as relocated. We have to check for relocated
2355 * TSOs in the garbage collector and any primops that deal with TSOs.
2357 * It's important to set the sp value to just beyond the end
2358 * of the stack, so we don't attempt to scavenge any part of the
2361 tso->what_next = ThreadRelocated;
2362 setTSOLink(cap,tso,dest);
2363 tso->sp = (P_)&(tso->stack[tso->stack_size]);
2364 tso->why_blocked = NotBlocked;
2366 IF_PAR_DEBUG(verbose,
2367 debugBelch("@@ threadStackOverflow of TSO %d (now at %p): stack size increased to %ld\n",
2368 tso->id, tso, tso->stack_size);
2369 /* If we're debugging, just print out the top of the stack */
2370 printStackChunk(tso->sp, stg_min(tso->stack+tso->stack_size,
2376 IF_DEBUG(sanity,checkTSO(dest));
2378 IF_DEBUG(scheduler,printTSO(dest));
2385 threadStackUnderflow (Task *task STG_UNUSED, StgTSO *tso)
2387 bdescr *bd, *new_bd;
2388 lnat free_w, tso_size_w;
2391 tso_size_w = tso_sizeW(tso);
2393 if (tso_size_w < MBLOCK_SIZE_W ||
2394 (nat)(tso->stack + tso->stack_size - tso->sp) > tso->stack_size / 4)
2399 // don't allow throwTo() to modify the blocked_exceptions queue
2400 // while we are moving the TSO:
2401 lockClosure((StgClosure *)tso);
2403 // this is the number of words we'll free
2404 free_w = round_to_mblocks(tso_size_w/2);
2406 bd = Bdescr((StgPtr)tso);
2407 new_bd = splitLargeBlock(bd, free_w / BLOCK_SIZE_W);
2408 bd->free = bd->start + TSO_STRUCT_SIZEW;
2410 new_tso = (StgTSO *)new_bd->start;
2411 memcpy(new_tso,tso,TSO_STRUCT_SIZE);
2412 new_tso->stack_size = new_bd->free - new_tso->stack;
2414 debugTrace(DEBUG_sched, "thread %ld: reducing TSO size from %lu words to %lu",
2415 (long)tso->id, tso_size_w, tso_sizeW(new_tso));
2417 tso->what_next = ThreadRelocated;
2418 tso->_link = new_tso; // no write barrier reqd: same generation
2420 // The TSO attached to this Task may have moved, so update the
2422 if (task->tso == tso) {
2423 task->tso = new_tso;
2429 IF_DEBUG(sanity,checkTSO(new_tso));
2434 /* ---------------------------------------------------------------------------
2436 - usually called inside a signal handler so it mustn't do anything fancy.
2437 ------------------------------------------------------------------------ */
2440 interruptStgRts(void)
2442 sched_state = SCHED_INTERRUPTING;
2443 setContextSwitches();
2447 /* -----------------------------------------------------------------------------
2450 This function causes at least one OS thread to wake up and run the
2451 scheduler loop. It is invoked when the RTS might be deadlocked, or
2452 an external event has arrived that may need servicing (eg. a
2453 keyboard interrupt).
2455 In the single-threaded RTS we don't do anything here; we only have
2456 one thread anyway, and the event that caused us to want to wake up
2457 will have interrupted any blocking system call in progress anyway.
2458 -------------------------------------------------------------------------- */
2463 #if defined(THREADED_RTS)
2464 // This forces the IO Manager thread to wakeup, which will
2465 // in turn ensure that some OS thread wakes up and runs the
2466 // scheduler loop, which will cause a GC and deadlock check.
2471 /* -----------------------------------------------------------------------------
2474 * Check the blackhole_queue for threads that can be woken up. We do
2475 * this periodically: before every GC, and whenever the run queue is
2478 * An elegant solution might be to just wake up all the blocked
2479 * threads with awakenBlockedQueue occasionally: they'll go back to
2480 * sleep again if the object is still a BLACKHOLE. Unfortunately this
2481 * doesn't give us a way to tell whether we've actually managed to
2482 * wake up any threads, so we would be busy-waiting.
2484 * -------------------------------------------------------------------------- */
2487 checkBlackHoles (Capability *cap)
2490 rtsBool any_woke_up = rtsFalse;
2493 // blackhole_queue is global:
2494 ASSERT_LOCK_HELD(&sched_mutex);
2496 debugTrace(DEBUG_sched, "checking threads blocked on black holes");
2498 // ASSUMES: sched_mutex
2499 prev = &blackhole_queue;
2500 t = blackhole_queue;
2501 while (t != END_TSO_QUEUE) {
2502 ASSERT(t->why_blocked == BlockedOnBlackHole);
2503 type = get_itbl(UNTAG_CLOSURE(t->block_info.closure))->type;
2504 if (type != BLACKHOLE && type != CAF_BLACKHOLE) {
2505 IF_DEBUG(sanity,checkTSO(t));
2506 t = unblockOne(cap, t);
2508 any_woke_up = rtsTrue;
2518 /* -----------------------------------------------------------------------------
2521 This is used for interruption (^C) and forking, and corresponds to
2522 raising an exception but without letting the thread catch the
2524 -------------------------------------------------------------------------- */
2527 deleteThread (Capability *cap, StgTSO *tso)
2529 // NOTE: must only be called on a TSO that we have exclusive
2530 // access to, because we will call throwToSingleThreaded() below.
2531 // The TSO must be on the run queue of the Capability we own, or
2532 // we must own all Capabilities.
2534 if (tso->why_blocked != BlockedOnCCall &&
2535 tso->why_blocked != BlockedOnCCall_NoUnblockExc) {
2536 throwToSingleThreaded(cap,tso,NULL);
2540 #ifdef FORKPROCESS_PRIMOP_SUPPORTED
2542 deleteThread_(Capability *cap, StgTSO *tso)
2543 { // for forkProcess only:
2544 // like deleteThread(), but we delete threads in foreign calls, too.
2546 if (tso->why_blocked == BlockedOnCCall ||
2547 tso->why_blocked == BlockedOnCCall_NoUnblockExc) {
2548 unblockOne(cap,tso);
2549 tso->what_next = ThreadKilled;
2551 deleteThread(cap,tso);
2556 /* -----------------------------------------------------------------------------
2557 raiseExceptionHelper
2559 This function is called by the raise# primitve, just so that we can
2560 move some of the tricky bits of raising an exception from C-- into
2561 C. Who knows, it might be a useful re-useable thing here too.
2562 -------------------------------------------------------------------------- */
2565 raiseExceptionHelper (StgRegTable *reg, StgTSO *tso, StgClosure *exception)
2567 Capability *cap = regTableToCapability(reg);
2568 StgThunk *raise_closure = NULL;
2570 StgRetInfoTable *info;
2572 // This closure represents the expression 'raise# E' where E
2573 // is the exception raise. It is used to overwrite all the
2574 // thunks which are currently under evaluataion.
2577 // OLD COMMENT (we don't have MIN_UPD_SIZE now):
2578 // LDV profiling: stg_raise_info has THUNK as its closure
2579 // type. Since a THUNK takes at least MIN_UPD_SIZE words in its
2580 // payload, MIN_UPD_SIZE is more approprate than 1. It seems that
2581 // 1 does not cause any problem unless profiling is performed.
2582 // However, when LDV profiling goes on, we need to linearly scan
2583 // small object pool, where raise_closure is stored, so we should
2584 // use MIN_UPD_SIZE.
2586 // raise_closure = (StgClosure *)RET_STGCALL1(P_,allocate,
2587 // sizeofW(StgClosure)+1);
2591 // Walk up the stack, looking for the catch frame. On the way,
2592 // we update any closures pointed to from update frames with the
2593 // raise closure that we just built.
2597 info = get_ret_itbl((StgClosure *)p);
2598 next = p + stack_frame_sizeW((StgClosure *)p);
2599 switch (info->i.type) {
2602 // Only create raise_closure if we need to.
2603 if (raise_closure == NULL) {
2605 (StgThunk *)allocateLocal(cap,sizeofW(StgThunk)+1);
2606 SET_HDR(raise_closure, &stg_raise_info, CCCS);
2607 raise_closure->payload[0] = exception;
2609 UPD_IND(((StgUpdateFrame *)p)->updatee,(StgClosure *)raise_closure);
2613 case ATOMICALLY_FRAME:
2614 debugTrace(DEBUG_stm, "found ATOMICALLY_FRAME at %p", p);
2616 return ATOMICALLY_FRAME;
2622 case CATCH_STM_FRAME:
2623 debugTrace(DEBUG_stm, "found CATCH_STM_FRAME at %p", p);
2625 return CATCH_STM_FRAME;
2631 case CATCH_RETRY_FRAME:
2640 /* -----------------------------------------------------------------------------
2641 findRetryFrameHelper
2643 This function is called by the retry# primitive. It traverses the stack
2644 leaving tso->sp referring to the frame which should handle the retry.
2646 This should either be a CATCH_RETRY_FRAME (if the retry# is within an orElse#)
2647 or should be a ATOMICALLY_FRAME (if the retry# reaches the top level).
2649 We skip CATCH_STM_FRAMEs (aborting and rolling back the nested tx that they
2650 create) because retries are not considered to be exceptions, despite the
2651 similar implementation.
2653 We should not expect to see CATCH_FRAME or STOP_FRAME because those should
2654 not be created within memory transactions.
2655 -------------------------------------------------------------------------- */
2658 findRetryFrameHelper (StgTSO *tso)
2661 StgRetInfoTable *info;
2665 info = get_ret_itbl((StgClosure *)p);
2666 next = p + stack_frame_sizeW((StgClosure *)p);
2667 switch (info->i.type) {
2669 case ATOMICALLY_FRAME:
2670 debugTrace(DEBUG_stm,
2671 "found ATOMICALLY_FRAME at %p during retry", p);
2673 return ATOMICALLY_FRAME;
2675 case CATCH_RETRY_FRAME:
2676 debugTrace(DEBUG_stm,
2677 "found CATCH_RETRY_FRAME at %p during retrry", p);
2679 return CATCH_RETRY_FRAME;
2681 case CATCH_STM_FRAME: {
2682 StgTRecHeader *trec = tso -> trec;
2683 StgTRecHeader *outer = stmGetEnclosingTRec(trec);
2684 debugTrace(DEBUG_stm,
2685 "found CATCH_STM_FRAME at %p during retry", p);
2686 debugTrace(DEBUG_stm, "trec=%p outer=%p", trec, outer);
2687 stmAbortTransaction(tso -> cap, trec);
2688 stmFreeAbortedTRec(tso -> cap, trec);
2689 tso -> trec = outer;
2696 ASSERT(info->i.type != CATCH_FRAME);
2697 ASSERT(info->i.type != STOP_FRAME);
2704 /* -----------------------------------------------------------------------------
2705 resurrectThreads is called after garbage collection on the list of
2706 threads found to be garbage. Each of these threads will be woken
2707 up and sent a signal: BlockedOnDeadMVar if the thread was blocked
2708 on an MVar, or NonTermination if the thread was blocked on a Black
2711 Locks: assumes we hold *all* the capabilities.
2712 -------------------------------------------------------------------------- */
2715 resurrectThreads (StgTSO *threads)
2721 for (tso = threads; tso != END_TSO_QUEUE; tso = next) {
2722 next = tso->global_link;
2724 step = Bdescr((P_)tso)->step;
2725 tso->global_link = step->threads;
2726 step->threads = tso;
2728 debugTrace(DEBUG_sched, "resurrecting thread %lu", (unsigned long)tso->id);
2730 // Wake up the thread on the Capability it was last on
2733 switch (tso->why_blocked) {
2735 case BlockedOnException:
2736 /* Called by GC - sched_mutex lock is currently held. */
2737 throwToSingleThreaded(cap, tso,
2738 (StgClosure *)blockedOnDeadMVar_closure);
2740 case BlockedOnBlackHole:
2741 throwToSingleThreaded(cap, tso,
2742 (StgClosure *)nonTermination_closure);
2745 throwToSingleThreaded(cap, tso,
2746 (StgClosure *)blockedIndefinitely_closure);
2749 /* This might happen if the thread was blocked on a black hole
2750 * belonging to a thread that we've just woken up (raiseAsync
2751 * can wake up threads, remember...).
2755 barf("resurrectThreads: thread blocked in a strange way");
2760 /* -----------------------------------------------------------------------------
2761 performPendingThrowTos is called after garbage collection, and
2762 passed a list of threads that were found to have pending throwTos
2763 (tso->blocked_exceptions was not empty), and were blocked.
2764 Normally this doesn't happen, because we would deliver the
2765 exception directly if the target thread is blocked, but there are
2766 small windows where it might occur on a multiprocessor (see
2769 NB. we must be holding all the capabilities at this point, just
2770 like resurrectThreads().
2771 -------------------------------------------------------------------------- */
2774 performPendingThrowTos (StgTSO *threads)
2780 for (tso = threads; tso != END_TSO_QUEUE; tso = next) {
2781 next = tso->global_link;
2783 step = Bdescr((P_)tso)->step;
2784 tso->global_link = step->threads;
2785 step->threads = tso;
2787 debugTrace(DEBUG_sched, "performing blocked throwTo to thread %lu", (unsigned long)tso->id);
2790 maybePerformBlockedException(cap, tso);