1 /* ---------------------------------------------------------------------------
2 * $Id: Schedule.c,v 1.145 2002/06/19 20:45:15 sof Exp $
4 * (c) The GHC Team, 1998-2000
8 * Different GHC ways use this scheduler quite differently (see comments below)
9 * Here is the global picture:
11 * WAY Name CPP flag What's it for
12 * --------------------------------------
13 * mp GUM PAR Parallel execution on a distributed memory machine
14 * s SMP SMP Parallel execution on a shared memory machine
15 * mg GranSim GRAN Simulation of parallel execution
16 * md GUM/GdH DIST Distributed execution (based on GUM)
18 * --------------------------------------------------------------------------*/
20 //@node Main scheduling code, , ,
21 //@section Main scheduling code
24 * Version with scheduler monitor support for SMPs (WAY=s):
26 This design provides a high-level API to create and schedule threads etc.
27 as documented in the SMP design document.
29 It uses a monitor design controlled by a single mutex to exercise control
30 over accesses to shared data structures, and builds on the Posix threads
33 The majority of state is shared. In order to keep essential per-task state,
34 there is a Capability structure, which contains all the information
35 needed to run a thread: its STG registers, a pointer to its TSO, a
36 nursery etc. During STG execution, a pointer to the capability is
37 kept in a register (BaseReg).
39 In a non-SMP build, there is one global capability, namely MainRegTable.
43 * Version with support for distributed memory parallelism aka GUM (WAY=mp):
45 The main scheduling loop in GUM iterates until a finish message is received.
46 In that case a global flag @receivedFinish@ is set and this instance of
47 the RTS shuts down. See ghc/rts/parallel/HLComms.c:processMessages()
48 for the handling of incoming messages, such as PP_FINISH.
49 Note that in the parallel case we have a system manager that coordinates
50 different PEs, each of which are running one instance of the RTS.
51 See ghc/rts/parallel/SysMan.c for the main routine of the parallel program.
52 From this routine processes executing ghc/rts/Main.c are spawned. -- HWL
54 * Version with support for simulating parallel execution aka GranSim (WAY=mg):
56 The main scheduling code in GranSim is quite different from that in std
57 (concurrent) Haskell: while concurrent Haskell just iterates over the
58 threads in the runnable queue, GranSim is event driven, i.e. it iterates
59 over the events in the global event queue. -- HWL
64 //* Variables and Data structures::
65 //* Main scheduling loop::
66 //* Suspend and Resume::
68 //* Garbage Collextion Routines::
69 //* Blocking Queue Routines::
70 //* Exception Handling Routines::
71 //* Debugging Routines::
75 //@node Includes, Variables and Data structures, Main scheduling code, Main scheduling code
76 //@subsection Includes
78 #include "PosixSource.h"
85 #include "StgStartup.h"
88 #include "StgMiscClosures.h"
90 #include "Interpreter.h"
91 #include "Exception.h"
100 #include "Proftimer.h"
101 #include "ProfHeap.h"
103 #if defined(GRAN) || defined(PAR)
104 # include "GranSimRts.h"
105 # include "GranSim.h"
106 # include "ParallelRts.h"
107 # include "Parallel.h"
108 # include "ParallelDebug.h"
109 # include "FetchMe.h"
113 #include "Capability.h"
114 #include "OSThreads.h"
117 #ifdef HAVE_SYS_TYPES_H
118 #include <sys/types.h>
126 //@node Variables and Data structures, Prototypes, Includes, Main scheduling code
127 //@subsection Variables and Data structures
129 /* Main thread queue.
130 * Locks required: sched_mutex.
132 StgMainThread *main_threads;
135 * Locks required: sched_mutex.
139 StgTSO* ActiveTSO = NULL; /* for assigning system costs; GranSim-Light only */
140 /* rtsTime TimeOfNextEvent, EndOfTimeSlice; now in GranSim.c */
143 In GranSim we have a runnable and a blocked queue for each processor.
144 In order to minimise code changes new arrays run_queue_hds/tls
145 are created. run_queue_hd is then a short cut (macro) for
146 run_queue_hds[CurrentProc] (see GranSim.h).
149 StgTSO *run_queue_hds[MAX_PROC], *run_queue_tls[MAX_PROC];
150 StgTSO *blocked_queue_hds[MAX_PROC], *blocked_queue_tls[MAX_PROC];
151 StgTSO *ccalling_threadss[MAX_PROC];
152 /* We use the same global list of threads (all_threads) in GranSim as in
153 the std RTS (i.e. we are cheating). However, we don't use this list in
154 the GranSim specific code at the moment (so we are only potentially
159 StgTSO *run_queue_hd, *run_queue_tl;
160 StgTSO *blocked_queue_hd, *blocked_queue_tl;
161 StgTSO *sleeping_queue; /* perhaps replace with a hash table? */
165 /* Linked list of all threads.
166 * Used for detecting garbage collected threads.
170 /* When a thread performs a safe C call (_ccall_GC, using old
171 * terminology), it gets put on the suspended_ccalling_threads
172 * list. Used by the garbage collector.
174 static StgTSO *suspended_ccalling_threads;
176 static StgTSO *threadStackOverflow(StgTSO *tso);
178 /* KH: The following two flags are shared memory locations. There is no need
179 to lock them, since they are only unset at the end of a scheduler
183 /* flag set by signal handler to precipitate a context switch */
184 //@cindex context_switch
187 /* if this flag is set as well, give up execution */
188 //@cindex interrupted
191 /* Next thread ID to allocate.
192 * Locks required: thread_id_mutex
194 //@cindex next_thread_id
195 StgThreadID next_thread_id = 1;
198 * Pointers to the state of the current thread.
199 * Rule of thumb: if CurrentTSO != NULL, then we're running a Haskell
200 * thread. If CurrentTSO == NULL, then we're at the scheduler level.
203 /* The smallest stack size that makes any sense is:
204 * RESERVED_STACK_WORDS (so we can get back from the stack overflow)
205 * + sizeofW(StgStopFrame) (the stg_stop_thread_info frame)
206 * + 1 (the realworld token for an IO thread)
207 * + 1 (the closure to enter)
209 * A thread with this stack will bomb immediately with a stack
210 * overflow, which will increase its stack size.
213 #define MIN_STACK_WORDS (RESERVED_STACK_WORDS + sizeofW(StgStopFrame) + 2)
220 /* This is used in `TSO.h' and gcc 2.96 insists that this variable actually
221 * exists - earlier gccs apparently didn't.
229 * Set to TRUE when entering a shutdown state (via shutdownHaskellAndExit()) --
230 * in an MT setting, needed to signal that a worker thread shouldn't hang around
231 * in the scheduler when it is out of work.
233 static rtsBool shutting_down_scheduler = rtsFalse;
235 void addToBlockedQueue ( StgTSO *tso );
237 static void schedule ( void );
238 void interruptStgRts ( void );
240 static void detectBlackHoles ( void );
243 static void sched_belch(char *s, ...);
246 #if defined(RTS_SUPPORTS_THREADS)
247 /* ToDo: carefully document the invariants that go together
248 * with these synchronisation objects.
250 Mutex sched_mutex = INIT_MUTEX_VAR;
251 Mutex term_mutex = INIT_MUTEX_VAR;
254 * A heavyweight solution to the problem of protecting
255 * the thread_id from concurrent update.
257 Mutex thread_id_mutex = INIT_MUTEX_VAR;
261 static Condition gc_pending_cond = INIT_COND_VAR;
265 #endif /* RTS_SUPPORTS_THREADS */
269 rtsTime TimeOfLastYield;
270 rtsBool emitSchedule = rtsTrue;
274 char *whatNext_strs[] = {
282 char *threadReturnCode_strs[] = {
283 "HeapOverflow", /* might also be StackOverflow */
292 StgTSO * createSparkThread(rtsSpark spark);
293 StgTSO * activateSpark (rtsSpark spark);
297 * The thread state for the main thread.
298 // ToDo: check whether not needed any more
302 #if defined(PAR) || defined(RTS_SUPPORTS_THREADS)
303 static void taskStart(void);
314 //@node Main scheduling loop, Suspend and Resume, Prototypes, Main scheduling code
315 //@subsection Main scheduling loop
317 /* ---------------------------------------------------------------------------
318 Main scheduling loop.
320 We use round-robin scheduling, each thread returning to the
321 scheduler loop when one of these conditions is detected:
324 * timer expires (thread yields)
329 Locking notes: we acquire the scheduler lock once at the beginning
330 of the scheduler loop, and release it when
332 * running a thread, or
333 * waiting for work, or
334 * waiting for a GC to complete.
337 In a GranSim setup this loop iterates over the global event queue.
338 This revolves around the global event queue, which determines what
339 to do next. Therefore, it's more complicated than either the
340 concurrent or the parallel (GUM) setup.
343 GUM iterates over incoming messages.
344 It starts with nothing to do (thus CurrentTSO == END_TSO_QUEUE),
345 and sends out a fish whenever it has nothing to do; in-between
346 doing the actual reductions (shared code below) it processes the
347 incoming messages and deals with delayed operations
348 (see PendingFetches).
349 This is not the ugliest code you could imagine, but it's bloody close.
351 ------------------------------------------------------------------------ */
358 StgThreadReturnCode ret;
366 rtsBool receivedFinish = rtsFalse;
368 nat tp_size, sp_size; // stats only
371 rtsBool was_interrupted = rtsFalse;
373 ACQUIRE_LOCK(&sched_mutex);
375 #if defined(RTS_SUPPORTS_THREADS)
376 waitForWorkCapability(&sched_mutex, &cap, rtsFalse);
378 /* simply initialise it in the non-threaded case */
379 grabCapability(&cap);
383 /* set up first event to get things going */
384 /* ToDo: assign costs for system setup and init MainTSO ! */
385 new_event(CurrentProc, CurrentProc, CurrentTime[CurrentProc],
387 CurrentTSO, (StgClosure*)NULL, (rtsSpark*)NULL);
390 fprintf(stderr, "GRAN: Init CurrentTSO (in schedule) = %p\n", CurrentTSO);
391 G_TSO(CurrentTSO, 5));
393 if (RtsFlags.GranFlags.Light) {
394 /* Save current time; GranSim Light only */
395 CurrentTSO->gran.clock = CurrentTime[CurrentProc];
398 event = get_next_event();
400 while (event!=(rtsEvent*)NULL) {
401 /* Choose the processor with the next event */
402 CurrentProc = event->proc;
403 CurrentTSO = event->tso;
407 while (!receivedFinish) { /* set by processMessages */
408 /* when receiving PP_FINISH message */
415 IF_DEBUG(scheduler, printAllThreads());
417 #if defined(RTS_SUPPORTS_THREADS)
418 /* Check to see whether there are any worker threads
419 waiting to deposit external call results. If so,
420 yield our capability */
421 yieldToReturningWorker(&sched_mutex, &cap);
424 /* If we're interrupted (the user pressed ^C, or some other
425 * termination condition occurred), kill all the currently running
429 IF_DEBUG(scheduler, sched_belch("interrupted"));
431 interrupted = rtsFalse;
432 was_interrupted = rtsTrue;
435 /* Go through the list of main threads and wake up any
436 * clients whose computations have finished. ToDo: this
437 * should be done more efficiently without a linear scan
438 * of the main threads list, somehow...
440 #if defined(RTS_SUPPORTS_THREADS)
442 StgMainThread *m, **prev;
443 prev = &main_threads;
444 for (m = main_threads; m != NULL; m = m->link) {
445 switch (m->tso->what_next) {
448 *(m->ret) = (StgClosure *)m->tso->sp[0];
452 broadcastCondition(&m->wakeup);
455 m->tso->label = NULL;
459 if (m->ret) *(m->ret) = NULL;
461 if (was_interrupted) {
462 m->stat = Interrupted;
466 broadcastCondition(&m->wakeup);
469 m->tso->label = NULL;
478 #else /* not threaded */
481 /* in GUM do this only on the Main PE */
484 /* If our main thread has finished or been killed, return.
487 StgMainThread *m = main_threads;
488 if (m->tso->what_next == ThreadComplete
489 || m->tso->what_next == ThreadKilled) {
492 m->tso->label = NULL;
494 main_threads = main_threads->link;
495 if (m->tso->what_next == ThreadComplete) {
496 /* we finished successfully, fill in the return value */
497 if (m->ret) { *(m->ret) = (StgClosure *)m->tso->sp[0]; };
501 if (m->ret) { *(m->ret) = NULL; };
502 if (was_interrupted) {
503 m->stat = Interrupted;
513 /* Top up the run queue from our spark pool. We try to make the
514 * number of threads in the run queue equal to the number of
517 * Disable spark support in SMP for now, non-essential & requires
518 * a little bit of work to make it compile cleanly. -- sof 1/02.
520 #if 0 /* defined(SMP) */
522 nat n = getFreeCapabilities();
523 StgTSO *tso = run_queue_hd;
525 /* Count the run queue */
526 while (n > 0 && tso != END_TSO_QUEUE) {
533 spark = findSpark(rtsFalse);
535 break; /* no more sparks in the pool */
537 /* I'd prefer this to be done in activateSpark -- HWL */
538 /* tricky - it needs to hold the scheduler lock and
539 * not try to re-acquire it -- SDM */
540 createSparkThread(spark);
542 sched_belch("==^^ turning spark of closure %p into a thread",
543 (StgClosure *)spark));
546 /* We need to wake up the other tasks if we just created some
549 if (getFreeCapabilities() - n > 1) {
550 signalCondition( &thread_ready_cond );
555 /* check for signals each time around the scheduler */
556 #ifndef mingw32_TARGET_OS
557 if (signals_pending()) {
558 RELEASE_LOCK(&sched_mutex); /* ToDo: kill */
559 startSignalHandlers();
560 ACQUIRE_LOCK(&sched_mutex);
564 /* Check whether any waiting threads need to be woken up. If the
565 * run queue is empty, and there are no other tasks running, we
566 * can wait indefinitely for something to happen.
567 * ToDo: what if another client comes along & requests another
570 if ( !EMPTY_QUEUE(blocked_queue_hd) || !EMPTY_QUEUE(sleeping_queue) ) {
571 awaitEvent( EMPTY_RUN_QUEUE()
573 && allFreeCapabilities()
577 /* we can be interrupted while waiting for I/O... */
578 if (interrupted) continue;
581 * Detect deadlock: when we have no threads to run, there are no
582 * threads waiting on I/O or sleeping, and all the other tasks are
583 * waiting for work, we must have a deadlock of some description.
585 * We first try to find threads blocked on themselves (ie. black
586 * holes), and generate NonTermination exceptions where necessary.
588 * If no threads are black holed, we have a deadlock situation, so
589 * inform all the main threads.
592 if ( EMPTY_THREAD_QUEUES()
593 #if defined(RTS_SUPPORTS_THREADS)
594 && EMPTY_QUEUE(suspended_ccalling_threads)
597 && allFreeCapabilities()
601 IF_DEBUG(scheduler, sched_belch("deadlocked, forcing major GC..."));
602 #if defined(THREADED_RTS)
603 /* and SMP mode ..? */
604 releaseCapability(cap);
606 // Garbage collection can release some new threads due to
607 // either (a) finalizers or (b) threads resurrected because
608 // they are about to be send BlockedOnDeadMVar. Any threads
609 // thus released will be immediately runnable.
610 GarbageCollect(GetRoots,rtsTrue);
612 if ( !EMPTY_RUN_QUEUE() ) { goto not_deadlocked; }
615 sched_belch("still deadlocked, checking for black holes..."));
618 if ( !EMPTY_RUN_QUEUE() ) { goto not_deadlocked; }
620 #ifndef mingw32_TARGET_OS
621 /* If we have user-installed signal handlers, then wait
622 * for signals to arrive rather then bombing out with a
625 #if defined(RTS_SUPPORTS_THREADS)
626 if ( 0 ) { /* hmm..what to do? Simply stop waiting for
627 a signal with no runnable threads (or I/O
628 suspended ones) leads nowhere quick.
629 For now, simply shut down when we reach this
632 ToDo: define precisely under what conditions
633 the Scheduler should shut down in an MT setting.
636 if ( anyUserHandlers() ) {
639 sched_belch("still deadlocked, waiting for signals..."));
643 // we might be interrupted...
644 if (interrupted) { continue; }
646 if (signals_pending()) {
647 RELEASE_LOCK(&sched_mutex);
648 startSignalHandlers();
649 ACQUIRE_LOCK(&sched_mutex);
651 ASSERT(!EMPTY_RUN_QUEUE());
656 /* Probably a real deadlock. Send the current main thread the
657 * Deadlock exception (or in the SMP build, send *all* main
658 * threads the deadlock exception, since none of them can make
663 #if defined(RTS_SUPPORTS_THREADS)
664 for (m = main_threads; m != NULL; m = m->link) {
665 switch (m->tso->why_blocked) {
666 case BlockedOnBlackHole:
667 raiseAsync(m->tso, (StgClosure *)NonTermination_closure);
669 case BlockedOnException:
671 raiseAsync(m->tso, (StgClosure *)Deadlock_closure);
674 barf("deadlock: main thread blocked in a strange way");
679 switch (m->tso->why_blocked) {
680 case BlockedOnBlackHole:
681 raiseAsync(m->tso, (StgClosure *)NonTermination_closure);
683 case BlockedOnException:
685 raiseAsync(m->tso, (StgClosure *)Deadlock_closure);
688 barf("deadlock: main thread blocked in a strange way");
693 #if defined(RTS_SUPPORTS_THREADS)
694 /* ToDo: revisit conditions (and mechanism) for shutting
695 down a multi-threaded world */
696 IF_DEBUG(scheduler, sched_belch("all done, i think...shutting down."));
697 RELEASE_LOCK(&sched_mutex);
705 /* ToDo: add deadlock detection in GUM (similar to SMP) -- HWL */
709 /* If there's a GC pending, don't do anything until it has
713 IF_DEBUG(scheduler,sched_belch("waiting for GC"));
714 waitCondition( &gc_pending_cond, &sched_mutex );
718 #if defined(RTS_SUPPORTS_THREADS)
719 /* block until we've got a thread on the run queue and a free
723 if ( EMPTY_RUN_QUEUE() ) {
724 /* Give up our capability */
725 releaseCapability(cap);
727 /* If we're in the process of shutting down (& running the
728 * a batch of finalisers), don't wait around.
730 if ( shutting_down_scheduler ) {
731 RELEASE_LOCK(&sched_mutex);
734 IF_DEBUG(scheduler, sched_belch("thread %d: waiting for work", osThreadId()));
735 waitForWorkCapability(&sched_mutex, &cap, rtsTrue);
736 IF_DEBUG(scheduler, sched_belch("thread %d: work now available", osThreadId()));
741 if (RtsFlags.GranFlags.Light)
742 GranSimLight_enter_system(event, &ActiveTSO); // adjust ActiveTSO etc
744 /* adjust time based on time-stamp */
745 if (event->time > CurrentTime[CurrentProc] &&
746 event->evttype != ContinueThread)
747 CurrentTime[CurrentProc] = event->time;
749 /* Deal with the idle PEs (may issue FindWork or MoveSpark events) */
750 if (!RtsFlags.GranFlags.Light)
753 IF_DEBUG(gran, fprintf(stderr, "GRAN: switch by event-type\n"));
755 /* main event dispatcher in GranSim */
756 switch (event->evttype) {
757 /* Should just be continuing execution */
759 IF_DEBUG(gran, fprintf(stderr, "GRAN: doing ContinueThread\n"));
760 /* ToDo: check assertion
761 ASSERT(run_queue_hd != (StgTSO*)NULL &&
762 run_queue_hd != END_TSO_QUEUE);
764 /* Ignore ContinueThreads for fetching threads (if synchr comm) */
765 if (!RtsFlags.GranFlags.DoAsyncFetch &&
766 procStatus[CurrentProc]==Fetching) {
767 belch("ghuH: Spurious ContinueThread while Fetching ignored; TSO %d (%p) [PE %d]",
768 CurrentTSO->id, CurrentTSO, CurrentProc);
771 /* Ignore ContinueThreads for completed threads */
772 if (CurrentTSO->what_next == ThreadComplete) {
773 belch("ghuH: found a ContinueThread event for completed thread %d (%p) [PE %d] (ignoring ContinueThread)",
774 CurrentTSO->id, CurrentTSO, CurrentProc);
777 /* Ignore ContinueThreads for threads that are being migrated */
778 if (PROCS(CurrentTSO)==Nowhere) {
779 belch("ghuH: trying to run the migrating TSO %d (%p) [PE %d] (ignoring ContinueThread)",
780 CurrentTSO->id, CurrentTSO, CurrentProc);
783 /* The thread should be at the beginning of the run queue */
784 if (CurrentTSO!=run_queue_hds[CurrentProc]) {
785 belch("ghuH: TSO %d (%p) [PE %d] is not at the start of the run_queue when doing a ContinueThread",
786 CurrentTSO->id, CurrentTSO, CurrentProc);
787 break; // run the thread anyway
790 new_event(proc, proc, CurrentTime[proc],
792 (StgTSO*)NULL, (StgClosure*)NULL, (rtsSpark*)NULL);
794 */ /* Catches superfluous CONTINUEs -- should be unnecessary */
795 break; // now actually run the thread; DaH Qu'vam yImuHbej
798 do_the_fetchnode(event);
799 goto next_thread; /* handle next event in event queue */
802 do_the_globalblock(event);
803 goto next_thread; /* handle next event in event queue */
806 do_the_fetchreply(event);
807 goto next_thread; /* handle next event in event queue */
809 case UnblockThread: /* Move from the blocked queue to the tail of */
810 do_the_unblock(event);
811 goto next_thread; /* handle next event in event queue */
813 case ResumeThread: /* Move from the blocked queue to the tail of */
814 /* the runnable queue ( i.e. Qu' SImqa'lu') */
815 event->tso->gran.blocktime +=
816 CurrentTime[CurrentProc] - event->tso->gran.blockedat;
817 do_the_startthread(event);
818 goto next_thread; /* handle next event in event queue */
821 do_the_startthread(event);
822 goto next_thread; /* handle next event in event queue */
825 do_the_movethread(event);
826 goto next_thread; /* handle next event in event queue */
829 do_the_movespark(event);
830 goto next_thread; /* handle next event in event queue */
833 do_the_findwork(event);
834 goto next_thread; /* handle next event in event queue */
837 barf("Illegal event type %u\n", event->evttype);
840 /* This point was scheduler_loop in the old RTS */
842 IF_DEBUG(gran, belch("GRAN: after main switch"));
844 TimeOfLastEvent = CurrentTime[CurrentProc];
845 TimeOfNextEvent = get_time_of_next_event();
846 IgnoreEvents=(TimeOfNextEvent==0); // HWL HACK
847 // CurrentTSO = ThreadQueueHd;
849 IF_DEBUG(gran, belch("GRAN: time of next event is: %ld",
852 if (RtsFlags.GranFlags.Light)
853 GranSimLight_leave_system(event, &ActiveTSO);
855 EndOfTimeSlice = CurrentTime[CurrentProc]+RtsFlags.GranFlags.time_slice;
858 belch("GRAN: end of time-slice is %#lx", EndOfTimeSlice));
860 /* in a GranSim setup the TSO stays on the run queue */
862 /* Take a thread from the run queue. */
863 t = POP_RUN_QUEUE(); // take_off_run_queue(t);
866 fprintf(stderr, "GRAN: About to run current thread, which is\n");
869 context_switch = 0; // turned on via GranYield, checking events and time slice
872 DumpGranEvent(GR_SCHEDULE, t));
874 procStatus[CurrentProc] = Busy;
877 if (PendingFetches != END_BF_QUEUE) {
881 /* ToDo: phps merge with spark activation above */
882 /* check whether we have local work and send requests if we have none */
883 if (EMPTY_RUN_QUEUE()) { /* no runnable threads */
884 /* :-[ no local threads => look out for local sparks */
885 /* the spark pool for the current PE */
886 pool = &(MainRegTable.rSparks); // generalise to cap = &MainRegTable
887 if (advisory_thread_count < RtsFlags.ParFlags.maxThreads &&
888 pool->hd < pool->tl) {
890 * ToDo: add GC code check that we really have enough heap afterwards!!
892 * If we're here (no runnable threads) and we have pending
893 * sparks, we must have a space problem. Get enough space
894 * to turn one of those pending sparks into a
898 spark = findSpark(rtsFalse); /* get a spark */
899 if (spark != (rtsSpark) NULL) {
900 tso = activateSpark(spark); /* turn the spark into a thread */
901 IF_PAR_DEBUG(schedule,
902 belch("==== schedule: Created TSO %d (%p); %d threads active",
903 tso->id, tso, advisory_thread_count));
905 if (tso==END_TSO_QUEUE) { /* failed to activate spark->back to loop */
906 belch("==^^ failed to activate spark");
908 } /* otherwise fall through & pick-up new tso */
910 IF_PAR_DEBUG(verbose,
911 belch("==^^ no local sparks (spark pool contains only NFs: %d)",
912 spark_queue_len(pool)));
917 /* If we still have no work we need to send a FISH to get a spark
920 if (EMPTY_RUN_QUEUE()) {
921 /* =8-[ no local sparks => look for work on other PEs */
923 * We really have absolutely no work. Send out a fish
924 * (there may be some out there already), and wait for
925 * something to arrive. We clearly can't run any threads
926 * until a SCHEDULE or RESUME arrives, and so that's what
927 * we're hoping to see. (Of course, we still have to
928 * respond to other types of messages.)
930 TIME now = msTime() /*CURRENT_TIME*/;
931 IF_PAR_DEBUG(verbose,
932 belch("-- now=%ld", now));
933 IF_PAR_DEBUG(verbose,
934 if (outstandingFishes < RtsFlags.ParFlags.maxFishes &&
935 (last_fish_arrived_at!=0 &&
936 last_fish_arrived_at+RtsFlags.ParFlags.fishDelay > now)) {
937 belch("--$$ delaying FISH until %ld (last fish %ld, delay %ld, now %ld)",
938 last_fish_arrived_at+RtsFlags.ParFlags.fishDelay,
939 last_fish_arrived_at,
940 RtsFlags.ParFlags.fishDelay, now);
943 if (outstandingFishes < RtsFlags.ParFlags.maxFishes &&
944 (last_fish_arrived_at==0 ||
945 (last_fish_arrived_at+RtsFlags.ParFlags.fishDelay <= now))) {
946 /* outstandingFishes is set in sendFish, processFish;
947 avoid flooding system with fishes via delay */
949 sendFish(pe, mytid, NEW_FISH_AGE, NEW_FISH_HISTORY,
952 // Global statistics: count no. of fishes
953 if (RtsFlags.ParFlags.ParStats.Global &&
954 RtsFlags.GcFlags.giveStats > NO_GC_STATS) {
955 globalParStats.tot_fish_mess++;
959 receivedFinish = processMessages();
962 } else if (PacketsWaiting()) { /* Look for incoming messages */
963 receivedFinish = processMessages();
966 /* Now we are sure that we have some work available */
967 ASSERT(run_queue_hd != END_TSO_QUEUE);
969 /* Take a thread from the run queue, if we have work */
970 t = POP_RUN_QUEUE(); // take_off_run_queue(END_TSO_QUEUE);
971 IF_DEBUG(sanity,checkTSO(t));
973 /* ToDo: write something to the log-file
974 if (RTSflags.ParFlags.granSimStats && !sameThread)
975 DumpGranEvent(GR_SCHEDULE, RunnableThreadsHd);
979 /* the spark pool for the current PE */
980 pool = &(MainRegTable.rSparks); // generalise to cap = &MainRegTable
983 belch("--=^ %d threads, %d sparks on [%#x]",
984 run_queue_len(), spark_queue_len(pool), CURRENT_PROC));
987 if (0 && RtsFlags.ParFlags.ParStats.Full &&
988 t && LastTSO && t->id != LastTSO->id &&
989 LastTSO->why_blocked == NotBlocked &&
990 LastTSO->what_next != ThreadComplete) {
991 // if previously scheduled TSO not blocked we have to record the context switch
992 DumpVeryRawGranEvent(TimeOfLastYield, CURRENT_PROC, CURRENT_PROC,
993 GR_DESCHEDULE, LastTSO, (StgClosure *)NULL, 0, 0);
996 if (RtsFlags.ParFlags.ParStats.Full &&
997 (emitSchedule /* forced emit */ ||
998 (t && LastTSO && t->id != LastTSO->id))) {
1000 we are running a different TSO, so write a schedule event to log file
1001 NB: If we use fair scheduling we also have to write a deschedule
1002 event for LastTSO; with unfair scheduling we know that the
1003 previous tso has blocked whenever we switch to another tso, so
1004 we don't need it in GUM for now
1006 DumpRawGranEvent(CURRENT_PROC, CURRENT_PROC,
1007 GR_SCHEDULE, t, (StgClosure *)NULL, 0, 0);
1008 emitSchedule = rtsFalse;
1012 #else /* !GRAN && !PAR */
1014 /* grab a thread from the run queue */
1015 ASSERT(run_queue_hd != END_TSO_QUEUE);
1016 t = POP_RUN_QUEUE();
1017 // Sanity check the thread we're about to run. This can be
1018 // expensive if there is lots of thread switching going on...
1019 IF_DEBUG(sanity,checkTSO(t));
1022 cap->r.rCurrentTSO = t;
1024 /* context switches are now initiated by the timer signal, unless
1025 * the user specified "context switch as often as possible", with
1030 RtsFlags.ProfFlags.profileInterval == 0 ||
1032 (RtsFlags.ConcFlags.ctxtSwitchTicks == 0
1033 && (run_queue_hd != END_TSO_QUEUE
1034 || blocked_queue_hd != END_TSO_QUEUE
1035 || sleeping_queue != END_TSO_QUEUE)))
1040 RELEASE_LOCK(&sched_mutex);
1042 IF_DEBUG(scheduler, sched_belch("-->> Running TSO %ld (%p) %s ...",
1043 t->id, t, whatNext_strs[t->what_next]));
1046 startHeapProfTimer();
1049 /* +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ */
1050 /* Run the current thread
1052 switch (cap->r.rCurrentTSO->what_next) {
1054 case ThreadComplete:
1055 /* Thread already finished, return to scheduler. */
1056 ret = ThreadFinished;
1058 case ThreadEnterGHC:
1059 ret = StgRun((StgFunPtr) stg_enterStackTop, &cap->r);
1062 ret = StgRun((StgFunPtr) stg_returnToStackTop, &cap->r);
1064 case ThreadEnterInterp:
1065 ret = interpretBCO(cap);
1068 barf("schedule: invalid what_next field");
1070 /* +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ */
1072 /* Costs for the scheduler are assigned to CCS_SYSTEM */
1074 stopHeapProfTimer();
1078 ACQUIRE_LOCK(&sched_mutex);
1081 IF_DEBUG(scheduler,fprintf(stderr,"scheduler (task %ld): ", osThreadId()););
1082 #elif !defined(GRAN) && !defined(PAR)
1083 IF_DEBUG(scheduler,fprintf(stderr,"scheduler: "););
1085 t = cap->r.rCurrentTSO;
1088 /* HACK 675: if the last thread didn't yield, make sure to print a
1089 SCHEDULE event to the log file when StgRunning the next thread, even
1090 if it is the same one as before */
1092 TimeOfLastYield = CURRENT_TIME;
1098 IF_DEBUG(gran, DumpGranEvent(GR_DESCHEDULE, t));
1099 globalGranStats.tot_heapover++;
1101 globalParStats.tot_heapover++;
1104 // did the task ask for a large block?
1105 if (cap->r.rHpAlloc > BLOCK_SIZE_W) {
1106 // if so, get one and push it on the front of the nursery.
1110 blocks = (nat)BLOCK_ROUND_UP(cap->r.rHpAlloc * sizeof(W_)) / BLOCK_SIZE;
1112 IF_DEBUG(scheduler,belch("--<< thread %ld (%p; %s) stopped: requesting a large block (size %d)",
1114 whatNext_strs[t->what_next], blocks));
1116 // don't do this if it would push us over the
1117 // alloc_blocks_lim limit; we'll GC first.
1118 if (alloc_blocks + blocks < alloc_blocks_lim) {
1120 alloc_blocks += blocks;
1121 bd = allocGroup( blocks );
1123 // link the new group into the list
1124 bd->link = cap->r.rCurrentNursery;
1125 bd->u.back = cap->r.rCurrentNursery->u.back;
1126 if (cap->r.rCurrentNursery->u.back != NULL) {
1127 cap->r.rCurrentNursery->u.back->link = bd;
1129 ASSERT(g0s0->blocks == cap->r.rCurrentNursery &&
1130 g0s0->blocks == cap->r.rNursery);
1131 cap->r.rNursery = g0s0->blocks = bd;
1133 cap->r.rCurrentNursery->u.back = bd;
1135 // initialise it as a nursery block
1139 bd->free = bd->start;
1141 // don't forget to update the block count in g0s0.
1142 g0s0->n_blocks += blocks;
1143 ASSERT(countBlocks(g0s0->blocks) == g0s0->n_blocks);
1145 // now update the nursery to point to the new block
1146 cap->r.rCurrentNursery = bd;
1148 // we might be unlucky and have another thread get on the
1149 // run queue before us and steal the large block, but in that
1150 // case the thread will just end up requesting another large
1152 PUSH_ON_RUN_QUEUE(t);
1157 /* make all the running tasks block on a condition variable,
1158 * maybe set context_switch and wait till they all pile in,
1159 * then have them wait on a GC condition variable.
1161 IF_DEBUG(scheduler,belch("--<< thread %ld (%p; %s) stopped: HeapOverflow",
1162 t->id, t, whatNext_strs[t->what_next]));
1165 ASSERT(!is_on_queue(t,CurrentProc));
1167 /* Currently we emit a DESCHEDULE event before GC in GUM.
1168 ToDo: either add separate event to distinguish SYSTEM time from rest
1169 or just nuke this DESCHEDULE (and the following SCHEDULE) */
1170 if (0 && RtsFlags.ParFlags.ParStats.Full) {
1171 DumpRawGranEvent(CURRENT_PROC, CURRENT_PROC,
1172 GR_DESCHEDULE, t, (StgClosure *)NULL, 0, 0);
1173 emitSchedule = rtsTrue;
1177 ready_to_gc = rtsTrue;
1178 context_switch = 1; /* stop other threads ASAP */
1179 PUSH_ON_RUN_QUEUE(t);
1180 /* actual GC is done at the end of the while loop */
1186 DumpGranEvent(GR_DESCHEDULE, t));
1187 globalGranStats.tot_stackover++;
1190 // DumpGranEvent(GR_DESCHEDULE, t);
1191 globalParStats.tot_stackover++;
1193 IF_DEBUG(scheduler,belch("--<< thread %ld (%p; %s) stopped, StackOverflow",
1194 t->id, t, whatNext_strs[t->what_next]));
1195 /* just adjust the stack for this thread, then pop it back
1201 /* enlarge the stack */
1202 StgTSO *new_t = threadStackOverflow(t);
1204 /* This TSO has moved, so update any pointers to it from the
1205 * main thread stack. It better not be on any other queues...
1206 * (it shouldn't be).
1208 for (m = main_threads; m != NULL; m = m->link) {
1213 threadPaused(new_t);
1214 PUSH_ON_RUN_QUEUE(new_t);
1218 case ThreadYielding:
1221 DumpGranEvent(GR_DESCHEDULE, t));
1222 globalGranStats.tot_yields++;
1225 // DumpGranEvent(GR_DESCHEDULE, t);
1226 globalParStats.tot_yields++;
1228 /* put the thread back on the run queue. Then, if we're ready to
1229 * GC, check whether this is the last task to stop. If so, wake
1230 * up the GC thread. getThread will block during a GC until the
1234 if (t->what_next == ThreadEnterInterp) {
1235 /* ToDo: or maybe a timer expired when we were in Hugs?
1236 * or maybe someone hit ctrl-C
1238 belch("--<< thread %ld (%p; %s) stopped to switch to Hugs",
1239 t->id, t, whatNext_strs[t->what_next]);
1241 belch("--<< thread %ld (%p; %s) stopped, yielding",
1242 t->id, t, whatNext_strs[t->what_next]);
1249 //belch("&& Doing sanity check on yielding TSO %ld.", t->id);
1251 ASSERT(t->link == END_TSO_QUEUE);
1253 ASSERT(!is_on_queue(t,CurrentProc));
1256 //belch("&& Doing sanity check on all ThreadQueues (and their TSOs).");
1257 checkThreadQsSanity(rtsTrue));
1260 if (RtsFlags.ParFlags.doFairScheduling) {
1261 /* this does round-robin scheduling; good for concurrency */
1262 APPEND_TO_RUN_QUEUE(t);
1264 /* this does unfair scheduling; good for parallelism */
1265 PUSH_ON_RUN_QUEUE(t);
1268 /* this does round-robin scheduling; good for concurrency */
1269 APPEND_TO_RUN_QUEUE(t);
1272 /* add a ContinueThread event to actually process the thread */
1273 new_event(CurrentProc, CurrentProc, CurrentTime[CurrentProc],
1275 t, (StgClosure*)NULL, (rtsSpark*)NULL);
1277 belch("GRAN: eventq and runnableq after adding yielded thread to queue again:");
1286 belch("--<< thread %ld (%p; %s) stopped, blocking on node %p [PE %d] with BQ: ",
1287 t->id, t, whatNext_strs[t->what_next], t->block_info.closure, (t->block_info.closure==(StgClosure*)NULL ? 99 : where_is(t->block_info.closure)));
1288 if (t->block_info.closure!=(StgClosure*)NULL) print_bq(t->block_info.closure));
1290 // ??? needed; should emit block before
1292 DumpGranEvent(GR_DESCHEDULE, t));
1293 prune_eventq(t, (StgClosure *)NULL); // prune ContinueThreads for t
1296 ASSERT(procStatus[CurrentProc]==Busy ||
1297 ((procStatus[CurrentProc]==Fetching) &&
1298 (t->block_info.closure!=(StgClosure*)NULL)));
1299 if (run_queue_hds[CurrentProc] == END_TSO_QUEUE &&
1300 !(!RtsFlags.GranFlags.DoAsyncFetch &&
1301 procStatus[CurrentProc]==Fetching))
1302 procStatus[CurrentProc] = Idle;
1306 belch("--<< thread %ld (%p; %s) stopped, blocking on node %p with BQ: ",
1307 t->id, t, whatNext_strs[t->what_next], t->block_info.closure));
1310 if (t->block_info.closure!=(StgClosure*)NULL)
1311 print_bq(t->block_info.closure));
1313 /* Send a fetch (if BlockedOnGA) and dump event to log file */
1316 /* whatever we schedule next, we must log that schedule */
1317 emitSchedule = rtsTrue;
1320 /* don't need to do anything. Either the thread is blocked on
1321 * I/O, in which case we'll have called addToBlockedQueue
1322 * previously, or it's blocked on an MVar or Blackhole, in which
1323 * case it'll be on the relevant queue already.
1326 fprintf(stderr, "--<< thread %d (%p) stopped: ", t->id, t);
1327 printThreadBlockage(t);
1328 fprintf(stderr, "\n"));
1330 /* Only for dumping event to log file
1331 ToDo: do I need this in GranSim, too?
1338 case ThreadFinished:
1339 /* Need to check whether this was a main thread, and if so, signal
1340 * the task that started it with the return value. If we have no
1341 * more main threads, we probably need to stop all the tasks until
1344 /* We also end up here if the thread kills itself with an
1345 * uncaught exception, see Exception.hc.
1347 IF_DEBUG(scheduler,belch("--++ thread %d (%p) finished", t->id, t));
1349 endThread(t, CurrentProc); // clean-up the thread
1351 /* For now all are advisory -- HWL */
1352 //if(t->priority==AdvisoryPriority) ??
1353 advisory_thread_count--;
1356 if(t->dist.priority==RevalPriority)
1360 if (RtsFlags.ParFlags.ParStats.Full &&
1361 !RtsFlags.ParFlags.ParStats.Suppressed)
1362 DumpEndEvent(CURRENT_PROC, t, rtsFalse /* not mandatory */);
1367 barf("schedule: invalid thread return code %d", (int)ret);
1371 if (RtsFlags.ProfFlags.profileInterval==0 || performHeapProfile) {
1372 GarbageCollect(GetRoots, rtsTrue);
1374 performHeapProfile = rtsFalse;
1375 ready_to_gc = rtsFalse; // we already GC'd
1381 && allFreeCapabilities()
1384 /* everybody back, start the GC.
1385 * Could do it in this thread, or signal a condition var
1386 * to do it in another thread. Either way, we need to
1387 * broadcast on gc_pending_cond afterward.
1389 #if defined(RTS_SUPPORTS_THREADS)
1390 IF_DEBUG(scheduler,sched_belch("doing GC"));
1392 GarbageCollect(GetRoots,rtsFalse);
1393 ready_to_gc = rtsFalse;
1395 broadcastCondition(&gc_pending_cond);
1398 /* add a ContinueThread event to continue execution of current thread */
1399 new_event(CurrentProc, CurrentProc, CurrentTime[CurrentProc],
1401 t, (StgClosure*)NULL, (rtsSpark*)NULL);
1403 fprintf(stderr, "GRAN: eventq and runnableq after Garbage collection:\n");
1411 IF_GRAN_DEBUG(unused,
1412 print_eventq(EventHd));
1414 event = get_next_event();
1417 /* ToDo: wait for next message to arrive rather than busy wait */
1420 } /* end of while(1) */
1422 IF_PAR_DEBUG(verbose,
1423 belch("== Leaving schedule() after having received Finish"));
1426 /* ---------------------------------------------------------------------------
1427 * Singleton fork(). Do not copy any running threads.
1428 * ------------------------------------------------------------------------- */
1430 StgInt forkProcess(StgTSO* tso) {
1432 #ifndef mingw32_TARGET_OS
1436 IF_DEBUG(scheduler,sched_belch("forking!"));
1439 if (pid) { /* parent */
1441 /* just return the pid */
1443 } else { /* child */
1444 /* wipe all other threads */
1446 tso->link = END_TSO_QUEUE;
1448 /* DO NOT TOUCH THE QUEUES directly because most of the code around
1449 us is picky about finding the threat still in its queue when
1450 handling the deleteThread() */
1452 for (t = all_threads; t != END_TSO_QUEUE; t = next) {
1454 if (t->id != tso->id) {
1461 barf("forkProcess#: primop not implemented for mingw32, sorry! (%u)\n", tso->id);
1462 /* pointlessly printing out the TSOs 'id' to avoid CC unused warning. */
1464 #endif /* mingw32 */
1467 /* ---------------------------------------------------------------------------
1468 * deleteAllThreads(): kill all the live threads.
1470 * This is used when we catch a user interrupt (^C), before performing
1471 * any necessary cleanups and running finalizers.
1473 * Locks: sched_mutex held.
1474 * ------------------------------------------------------------------------- */
1476 void deleteAllThreads ( void )
1479 IF_DEBUG(scheduler,sched_belch("deleting all threads"));
1480 for (t = all_threads; t != END_TSO_QUEUE; t = next) {
1481 next = t->global_link;
1484 run_queue_hd = run_queue_tl = END_TSO_QUEUE;
1485 blocked_queue_hd = blocked_queue_tl = END_TSO_QUEUE;
1486 sleeping_queue = END_TSO_QUEUE;
1489 /* startThread and insertThread are now in GranSim.c -- HWL */
1492 //@node Suspend and Resume, Run queue code, Main scheduling loop, Main scheduling code
1493 //@subsection Suspend and Resume
1495 /* ---------------------------------------------------------------------------
1496 * Suspending & resuming Haskell threads.
1498 * When making a "safe" call to C (aka _ccall_GC), the task gives back
1499 * its capability before calling the C function. This allows another
1500 * task to pick up the capability and carry on running Haskell
1501 * threads. It also means that if the C call blocks, it won't lock
1504 * The Haskell thread making the C call is put to sleep for the
1505 * duration of the call, on the susepended_ccalling_threads queue. We
1506 * give out a token to the task, which it can use to resume the thread
1507 * on return from the C function.
1508 * ------------------------------------------------------------------------- */
1511 suspendThread( StgRegTable *reg,
1513 #if !defined(RTS_SUPPORTS_THREADS) && !defined(DEBUG)
1521 /* assume that *reg is a pointer to the StgRegTable part
1524 cap = (Capability *)((void *)reg - sizeof(StgFunTable));
1526 ACQUIRE_LOCK(&sched_mutex);
1529 sched_belch("thread %d did a _ccall_gc (is_concurrent: %d)", cap->r.rCurrentTSO->id,concCall));
1531 threadPaused(cap->r.rCurrentTSO);
1532 cap->r.rCurrentTSO->link = suspended_ccalling_threads;
1533 suspended_ccalling_threads = cap->r.rCurrentTSO;
1535 #if defined(RTS_SUPPORTS_THREADS)
1536 cap->r.rCurrentTSO->why_blocked = BlockedOnCCall;
1539 /* Use the thread ID as the token; it should be unique */
1540 tok = cap->r.rCurrentTSO->id;
1542 /* Hand back capability */
1543 releaseCapability(cap);
1545 #if defined(RTS_SUPPORTS_THREADS)
1546 /* Preparing to leave the RTS, so ensure there's a native thread/task
1547 waiting to take over.
1549 ToDo: optimise this and only create a new task if there's a need
1550 for one (i.e., if there's only one Concurrent Haskell thread alive,
1551 there's no need to create a new task).
1553 IF_DEBUG(scheduler, sched_belch("worker thread (%d): leaving RTS", tok));
1555 startTask(taskStart);
1559 /* Other threads _might_ be available for execution; signal this */
1561 RELEASE_LOCK(&sched_mutex);
1566 resumeThread( StgInt tok,
1568 #if !defined(RTS_SUPPORTS_THREADS)
1573 StgTSO *tso, **prev;
1576 #if defined(RTS_SUPPORTS_THREADS)
1577 /* Wait for permission to re-enter the RTS with the result. */
1579 ACQUIRE_LOCK(&sched_mutex);
1580 grabReturnCapability(&sched_mutex, &cap);
1582 grabCapability(&cap);
1585 grabCapability(&cap);
1588 /* Remove the thread off of the suspended list */
1589 prev = &suspended_ccalling_threads;
1590 for (tso = suspended_ccalling_threads;
1591 tso != END_TSO_QUEUE;
1592 prev = &tso->link, tso = tso->link) {
1593 if (tso->id == (StgThreadID)tok) {
1598 if (tso == END_TSO_QUEUE) {
1599 barf("resumeThread: thread not found");
1601 tso->link = END_TSO_QUEUE;
1602 /* Reset blocking status */
1603 tso->why_blocked = NotBlocked;
1605 cap->r.rCurrentTSO = tso;
1606 RELEASE_LOCK(&sched_mutex);
1611 /* ---------------------------------------------------------------------------
1613 * ------------------------------------------------------------------------ */
1614 static void unblockThread(StgTSO *tso);
1616 /* ---------------------------------------------------------------------------
1617 * Comparing Thread ids.
1619 * This is used from STG land in the implementation of the
1620 * instances of Eq/Ord for ThreadIds.
1621 * ------------------------------------------------------------------------ */
1623 int cmp_thread(const StgTSO *tso1, const StgTSO *tso2)
1625 StgThreadID id1 = tso1->id;
1626 StgThreadID id2 = tso2->id;
1628 if (id1 < id2) return (-1);
1629 if (id1 > id2) return 1;
1633 /* ---------------------------------------------------------------------------
1634 * Fetching the ThreadID from an StgTSO.
1636 * This is used in the implementation of Show for ThreadIds.
1637 * ------------------------------------------------------------------------ */
1638 int rts_getThreadId(const StgTSO *tso)
1644 void labelThread(StgTSO *tso, char *label)
1649 /* Caveat: Once set, you can only set the thread name to "" */
1650 len = strlen(label)+1;
1651 buf = realloc(tso->label,len);
1653 fprintf(stderr,"insufficient memory for labelThread!\n");
1657 strncpy(buf,label,len);
1662 /* ---------------------------------------------------------------------------
1663 Create a new thread.
1665 The new thread starts with the given stack size. Before the
1666 scheduler can run, however, this thread needs to have a closure
1667 (and possibly some arguments) pushed on its stack. See
1668 pushClosure() in Schedule.h.
1670 createGenThread() and createIOThread() (in SchedAPI.h) are
1671 convenient packaged versions of this function.
1673 currently pri (priority) is only used in a GRAN setup -- HWL
1674 ------------------------------------------------------------------------ */
1675 //@cindex createThread
1677 /* currently pri (priority) is only used in a GRAN setup -- HWL */
1679 createThread(nat size, StgInt pri)
1682 createThread(nat size)
1689 /* First check whether we should create a thread at all */
1691 /* check that no more than RtsFlags.ParFlags.maxThreads threads are created */
1692 if (advisory_thread_count >= RtsFlags.ParFlags.maxThreads) {
1694 belch("{createThread}Daq ghuH: refusing to create another thread; no more than %d threads allowed (currently %d)",
1695 RtsFlags.ParFlags.maxThreads, advisory_thread_count);
1696 return END_TSO_QUEUE;
1702 ASSERT(!RtsFlags.GranFlags.Light || CurrentProc==0);
1705 // ToDo: check whether size = stack_size - TSO_STRUCT_SIZEW
1707 /* catch ridiculously small stack sizes */
1708 if (size < MIN_STACK_WORDS + TSO_STRUCT_SIZEW) {
1709 size = MIN_STACK_WORDS + TSO_STRUCT_SIZEW;
1712 stack_size = size - TSO_STRUCT_SIZEW;
1714 tso = (StgTSO *)allocate(size);
1715 TICK_ALLOC_TSO(stack_size, 0);
1717 SET_HDR(tso, &stg_TSO_info, CCS_SYSTEM);
1719 SET_GRAN_HDR(tso, ThisPE);
1721 tso->what_next = ThreadEnterGHC;
1727 /* tso->id needs to be unique. For now we use a heavyweight mutex to
1728 * protect the increment operation on next_thread_id.
1729 * In future, we could use an atomic increment instead.
1731 ACQUIRE_LOCK(&thread_id_mutex);
1732 tso->id = next_thread_id++;
1733 RELEASE_LOCK(&thread_id_mutex);
1735 tso->why_blocked = NotBlocked;
1736 tso->blocked_exceptions = NULL;
1738 tso->stack_size = stack_size;
1739 tso->max_stack_size = round_to_mblocks(RtsFlags.GcFlags.maxStkSize)
1741 tso->sp = (P_)&(tso->stack) + stack_size;
1744 tso->prof.CCCS = CCS_MAIN;
1747 /* put a stop frame on the stack */
1748 tso->sp -= sizeofW(StgStopFrame);
1749 SET_HDR((StgClosure*)tso->sp,(StgInfoTable *)&stg_stop_thread_info,CCS_SYSTEM);
1750 tso->su = (StgUpdateFrame*)tso->sp;
1754 tso->link = END_TSO_QUEUE;
1755 /* uses more flexible routine in GranSim */
1756 insertThread(tso, CurrentProc);
1758 /* In a non-GranSim setup the pushing of a TSO onto the runq is separated
1764 if (RtsFlags.GranFlags.GranSimStats.Full)
1765 DumpGranEvent(GR_START,tso);
1767 if (RtsFlags.ParFlags.ParStats.Full)
1768 DumpGranEvent(GR_STARTQ,tso);
1769 /* HACk to avoid SCHEDULE
1773 /* Link the new thread on the global thread list.
1775 tso->global_link = all_threads;
1779 tso->dist.priority = MandatoryPriority; //by default that is...
1783 tso->gran.pri = pri;
1785 tso->gran.magic = TSO_MAGIC; // debugging only
1787 tso->gran.sparkname = 0;
1788 tso->gran.startedat = CURRENT_TIME;
1789 tso->gran.exported = 0;
1790 tso->gran.basicblocks = 0;
1791 tso->gran.allocs = 0;
1792 tso->gran.exectime = 0;
1793 tso->gran.fetchtime = 0;
1794 tso->gran.fetchcount = 0;
1795 tso->gran.blocktime = 0;
1796 tso->gran.blockcount = 0;
1797 tso->gran.blockedat = 0;
1798 tso->gran.globalsparks = 0;
1799 tso->gran.localsparks = 0;
1800 if (RtsFlags.GranFlags.Light)
1801 tso->gran.clock = Now; /* local clock */
1803 tso->gran.clock = 0;
1805 IF_DEBUG(gran,printTSO(tso));
1808 tso->par.magic = TSO_MAGIC; // debugging only
1810 tso->par.sparkname = 0;
1811 tso->par.startedat = CURRENT_TIME;
1812 tso->par.exported = 0;
1813 tso->par.basicblocks = 0;
1814 tso->par.allocs = 0;
1815 tso->par.exectime = 0;
1816 tso->par.fetchtime = 0;
1817 tso->par.fetchcount = 0;
1818 tso->par.blocktime = 0;
1819 tso->par.blockcount = 0;
1820 tso->par.blockedat = 0;
1821 tso->par.globalsparks = 0;
1822 tso->par.localsparks = 0;
1826 globalGranStats.tot_threads_created++;
1827 globalGranStats.threads_created_on_PE[CurrentProc]++;
1828 globalGranStats.tot_sq_len += spark_queue_len(CurrentProc);
1829 globalGranStats.tot_sq_probes++;
1831 // collect parallel global statistics (currently done together with GC stats)
1832 if (RtsFlags.ParFlags.ParStats.Global &&
1833 RtsFlags.GcFlags.giveStats > NO_GC_STATS) {
1834 //fprintf(stderr, "Creating thread %d @ %11.2f\n", tso->id, usertime());
1835 globalParStats.tot_threads_created++;
1841 belch("==__ schedule: Created TSO %d (%p);",
1842 CurrentProc, tso, tso->id));
1844 IF_PAR_DEBUG(verbose,
1845 belch("==__ schedule: Created TSO %d (%p); %d threads active",
1846 tso->id, tso, advisory_thread_count));
1848 IF_DEBUG(scheduler,sched_belch("created thread %ld, stack size = %lx words",
1849 tso->id, tso->stack_size));
1856 all parallel thread creation calls should fall through the following routine.
1859 createSparkThread(rtsSpark spark)
1861 ASSERT(spark != (rtsSpark)NULL);
1862 if (advisory_thread_count >= RtsFlags.ParFlags.maxThreads)
1864 barf("{createSparkThread}Daq ghuH: refusing to create another thread; no more than %d threads allowed (currently %d)",
1865 RtsFlags.ParFlags.maxThreads, advisory_thread_count);
1866 return END_TSO_QUEUE;
1870 tso = createThread(RtsFlags.GcFlags.initialStkSize);
1871 if (tso==END_TSO_QUEUE)
1872 barf("createSparkThread: Cannot create TSO");
1874 tso->priority = AdvisoryPriority;
1876 pushClosure(tso,spark);
1877 PUSH_ON_RUN_QUEUE(tso);
1878 advisory_thread_count++;
1885 Turn a spark into a thread.
1886 ToDo: fix for SMP (needs to acquire SCHED_MUTEX!)
1889 //@cindex activateSpark
1891 activateSpark (rtsSpark spark)
1895 tso = createSparkThread(spark);
1896 if (RtsFlags.ParFlags.ParStats.Full) {
1897 //ASSERT(run_queue_hd == END_TSO_QUEUE); // I think ...
1898 IF_PAR_DEBUG(verbose,
1899 belch("==^^ activateSpark: turning spark of closure %p (%s) into a thread",
1900 (StgClosure *)spark, info_type((StgClosure *)spark)));
1902 // ToDo: fwd info on local/global spark to thread -- HWL
1903 // tso->gran.exported = spark->exported;
1904 // tso->gran.locked = !spark->global;
1905 // tso->gran.sparkname = spark->name;
1911 static SchedulerStatus waitThread_(/*out*/StgMainThread* m
1912 #if defined(THREADED_RTS)
1913 , rtsBool blockWaiting
1918 /* ---------------------------------------------------------------------------
1921 * scheduleThread puts a thread on the head of the runnable queue.
1922 * This will usually be done immediately after a thread is created.
1923 * The caller of scheduleThread must create the thread using e.g.
1924 * createThread and push an appropriate closure
1925 * on this thread's stack before the scheduler is invoked.
1926 * ------------------------------------------------------------------------ */
1928 static void scheduleThread_ (StgTSO* tso, rtsBool createTask);
1931 scheduleThread_(StgTSO *tso
1932 , rtsBool createTask
1933 #if !defined(THREADED_RTS)
1938 ACQUIRE_LOCK(&sched_mutex);
1940 /* Put the new thread on the head of the runnable queue. The caller
1941 * better push an appropriate closure on this thread's stack
1942 * beforehand. In the SMP case, the thread may start running as
1943 * soon as we release the scheduler lock below.
1945 PUSH_ON_RUN_QUEUE(tso);
1946 #if defined(THREADED_RTS)
1947 /* If main() is scheduling a thread, don't bother creating a
1951 startTask(taskStart);
1957 IF_DEBUG(scheduler,printTSO(tso));
1959 RELEASE_LOCK(&sched_mutex);
1962 void scheduleThread(StgTSO* tso)
1964 scheduleThread_(tso, rtsFalse);
1968 scheduleWaitThread(StgTSO* tso, /*[out]*/HaskellObj* ret)
1972 m = stgMallocBytes(sizeof(StgMainThread), "waitThread");
1976 #if defined(RTS_SUPPORTS_THREADS)
1977 initCondition(&m->wakeup);
1980 /* Put the thread on the main-threads list prior to scheduling the TSO.
1981 Failure to do so introduces a race condition in the MT case (as
1982 identified by Wolfgang Thaller), whereby the new task/OS thread
1983 created by scheduleThread_() would complete prior to the thread
1984 that spawned it managed to put 'itself' on the main-threads list.
1985 The upshot of it all being that the worker thread wouldn't get to
1986 signal the completion of the its work item for the main thread to
1987 see (==> it got stuck waiting.) -- sof 6/02.
1989 ACQUIRE_LOCK(&sched_mutex);
1990 IF_DEBUG(scheduler, sched_belch("== scheduler: waiting for thread (%d)\n", tso->id));
1992 m->link = main_threads;
1995 /* Inefficient (scheduleThread_() acquires it again right away),
1996 * but obviously correct.
1998 RELEASE_LOCK(&sched_mutex);
2000 scheduleThread_(tso, rtsTrue);
2001 #if defined(THREADED_RTS)
2002 return waitThread_(m, rtsTrue);
2004 return waitThread_(m);
2008 /* ---------------------------------------------------------------------------
2011 * Initialise the scheduler. This resets all the queues - if the
2012 * queues contained any threads, they'll be garbage collected at the
2015 * ------------------------------------------------------------------------ */
2019 term_handler(int sig STG_UNUSED)
2022 ACQUIRE_LOCK(&term_mutex);
2024 RELEASE_LOCK(&term_mutex);
2035 for (i=0; i<=MAX_PROC; i++) {
2036 run_queue_hds[i] = END_TSO_QUEUE;
2037 run_queue_tls[i] = END_TSO_QUEUE;
2038 blocked_queue_hds[i] = END_TSO_QUEUE;
2039 blocked_queue_tls[i] = END_TSO_QUEUE;
2040 ccalling_threadss[i] = END_TSO_QUEUE;
2041 sleeping_queue = END_TSO_QUEUE;
2044 run_queue_hd = END_TSO_QUEUE;
2045 run_queue_tl = END_TSO_QUEUE;
2046 blocked_queue_hd = END_TSO_QUEUE;
2047 blocked_queue_tl = END_TSO_QUEUE;
2048 sleeping_queue = END_TSO_QUEUE;
2051 suspended_ccalling_threads = END_TSO_QUEUE;
2053 main_threads = NULL;
2054 all_threads = END_TSO_QUEUE;
2059 RtsFlags.ConcFlags.ctxtSwitchTicks =
2060 RtsFlags.ConcFlags.ctxtSwitchTime / TICK_MILLISECS;
2062 #if defined(RTS_SUPPORTS_THREADS)
2063 /* Initialise the mutex and condition variables used by
2065 initMutex(&sched_mutex);
2066 initMutex(&term_mutex);
2067 initMutex(&thread_id_mutex);
2069 initCondition(&thread_ready_cond);
2073 initCondition(&gc_pending_cond);
2076 #if defined(RTS_SUPPORTS_THREADS)
2077 ACQUIRE_LOCK(&sched_mutex);
2080 /* Install the SIGHUP handler */
2083 struct sigaction action,oact;
2085 action.sa_handler = term_handler;
2086 sigemptyset(&action.sa_mask);
2087 action.sa_flags = 0;
2088 if (sigaction(SIGTERM, &action, &oact) != 0) {
2089 barf("can't install TERM handler");
2094 /* A capability holds the state a native thread needs in
2095 * order to execute STG code. At least one capability is
2096 * floating around (only SMP builds have more than one).
2100 #if defined(RTS_SUPPORTS_THREADS)
2101 /* start our haskell execution tasks */
2103 startTaskManager(RtsFlags.ParFlags.nNodes, taskStart);
2105 startTaskManager(0,taskStart);
2109 #if /* defined(SMP) ||*/ defined(PAR)
2113 #if defined(RTS_SUPPORTS_THREADS)
2114 RELEASE_LOCK(&sched_mutex);
2120 exitScheduler( void )
2122 #if defined(RTS_SUPPORTS_THREADS)
2125 shutting_down_scheduler = rtsTrue;
2128 /* -----------------------------------------------------------------------------
2129 Managing the per-task allocation areas.
2131 Each capability comes with an allocation area. These are
2132 fixed-length block lists into which allocation can be done.
2134 ToDo: no support for two-space collection at the moment???
2135 -------------------------------------------------------------------------- */
2137 /* -----------------------------------------------------------------------------
2138 * waitThread is the external interface for running a new computation
2139 * and waiting for the result.
2141 * In the non-SMP case, we create a new main thread, push it on the
2142 * main-thread stack, and invoke the scheduler to run it. The
2143 * scheduler will return when the top main thread on the stack has
2144 * completed or died, and fill in the necessary fields of the
2145 * main_thread structure.
2147 * In the SMP case, we create a main thread as before, but we then
2148 * create a new condition variable and sleep on it. When our new
2149 * main thread has completed, we'll be woken up and the status/result
2150 * will be in the main_thread struct.
2151 * -------------------------------------------------------------------------- */
2154 howManyThreadsAvail ( void )
2158 for (q = run_queue_hd; q != END_TSO_QUEUE; q = q->link)
2160 for (q = blocked_queue_hd; q != END_TSO_QUEUE; q = q->link)
2162 for (q = sleeping_queue; q != END_TSO_QUEUE; q = q->link)
2168 finishAllThreads ( void )
2171 while (run_queue_hd != END_TSO_QUEUE) {
2172 waitThread ( run_queue_hd, NULL);
2174 while (blocked_queue_hd != END_TSO_QUEUE) {
2175 waitThread ( blocked_queue_hd, NULL);
2177 while (sleeping_queue != END_TSO_QUEUE) {
2178 waitThread ( blocked_queue_hd, NULL);
2181 (blocked_queue_hd != END_TSO_QUEUE ||
2182 run_queue_hd != END_TSO_QUEUE ||
2183 sleeping_queue != END_TSO_QUEUE);
2187 waitThread(StgTSO *tso, /*out*/StgClosure **ret)
2191 m = stgMallocBytes(sizeof(StgMainThread), "waitThread");
2195 #if defined(RTS_SUPPORTS_THREADS)
2196 initCondition(&m->wakeup);
2199 /* see scheduleWaitThread() comment */
2200 ACQUIRE_LOCK(&sched_mutex);
2201 IF_DEBUG(scheduler, sched_belch("== scheduler: waiting for thread (%d)\n", tso->id));
2202 m->link = main_threads;
2204 RELEASE_LOCK(&sched_mutex);
2206 IF_DEBUG(scheduler, sched_belch("== scheduler: waiting for thread (%d)\n", tso->id));
2207 #if defined(THREADED_RTS)
2208 return waitThread_(m, rtsFalse);
2210 return waitThread_(m);
2216 waitThread_(StgMainThread* m
2217 #if defined(THREADED_RTS)
2218 , rtsBool blockWaiting
2222 SchedulerStatus stat;
2224 IF_DEBUG(scheduler, sched_belch("== scheduler: new main thread (%d)\n", m->tso->id));
2226 #if defined(RTS_SUPPORTS_THREADS)
2228 # if defined(THREADED_RTS)
2229 if (!blockWaiting) {
2230 /* In the threaded case, the OS thread that called main()
2231 * gets to enter the RTS directly without going via another
2235 ASSERT(m->stat != NoStatus);
2239 ACQUIRE_LOCK(&sched_mutex);
2241 waitCondition(&m->wakeup, &sched_mutex);
2242 } while (m->stat == NoStatus);
2245 /* GranSim specific init */
2246 CurrentTSO = m->tso; // the TSO to run
2247 procStatus[MainProc] = Busy; // status of main PE
2248 CurrentProc = MainProc; // PE to run it on
2252 RELEASE_LOCK(&sched_mutex);
2254 ASSERT(m->stat != NoStatus);
2259 #if defined(RTS_SUPPORTS_THREADS)
2260 closeCondition(&m->wakeup);
2263 IF_DEBUG(scheduler, fprintf(stderr, "== scheduler: main thread (%d) finished\n",
2267 #if defined(THREADED_RTS)
2270 RELEASE_LOCK(&sched_mutex);
2275 //@node Run queue code, Garbage Collextion Routines, Suspend and Resume, Main scheduling code
2276 //@subsection Run queue code
2280 NB: In GranSim we have many run queues; run_queue_hd is actually a macro
2281 unfolding to run_queue_hds[CurrentProc], thus CurrentProc is an
2282 implicit global variable that has to be correct when calling these
2286 /* Put the new thread on the head of the runnable queue.
2287 * The caller of createThread better push an appropriate closure
2288 * on this thread's stack before the scheduler is invoked.
2290 static /* inline */ void
2291 add_to_run_queue(tso)
2294 ASSERT(tso!=run_queue_hd && tso!=run_queue_tl);
2295 tso->link = run_queue_hd;
2297 if (run_queue_tl == END_TSO_QUEUE) {
2302 /* Put the new thread at the end of the runnable queue. */
2303 static /* inline */ void
2304 push_on_run_queue(tso)
2307 ASSERT(get_itbl((StgClosure *)tso)->type == TSO);
2308 ASSERT(run_queue_hd!=NULL && run_queue_tl!=NULL);
2309 ASSERT(tso!=run_queue_hd && tso!=run_queue_tl);
2310 if (run_queue_hd == END_TSO_QUEUE) {
2313 run_queue_tl->link = tso;
2319 Should be inlined because it's used very often in schedule. The tso
2320 argument is actually only needed in GranSim, where we want to have the
2321 possibility to schedule *any* TSO on the run queue, irrespective of the
2322 actual ordering. Therefore, if tso is not the nil TSO then we traverse
2323 the run queue and dequeue the tso, adjusting the links in the queue.
2325 //@cindex take_off_run_queue
2326 static /* inline */ StgTSO*
2327 take_off_run_queue(StgTSO *tso) {
2331 qetlaHbogh Qu' ngaSbogh ghomDaQ {tso} yIteq!
2333 if tso is specified, unlink that tso from the run_queue (doesn't have
2334 to be at the beginning of the queue); GranSim only
2336 if (tso!=END_TSO_QUEUE) {
2337 /* find tso in queue */
2338 for (t=run_queue_hd, prev=END_TSO_QUEUE;
2339 t!=END_TSO_QUEUE && t!=tso;
2343 /* now actually dequeue the tso */
2344 if (prev!=END_TSO_QUEUE) {
2345 ASSERT(run_queue_hd!=t);
2346 prev->link = t->link;
2348 /* t is at beginning of thread queue */
2349 ASSERT(run_queue_hd==t);
2350 run_queue_hd = t->link;
2352 /* t is at end of thread queue */
2353 if (t->link==END_TSO_QUEUE) {
2354 ASSERT(t==run_queue_tl);
2355 run_queue_tl = prev;
2357 ASSERT(run_queue_tl!=t);
2359 t->link = END_TSO_QUEUE;
2361 /* take tso from the beginning of the queue; std concurrent code */
2363 if (t != END_TSO_QUEUE) {
2364 run_queue_hd = t->link;
2365 t->link = END_TSO_QUEUE;
2366 if (run_queue_hd == END_TSO_QUEUE) {
2367 run_queue_tl = END_TSO_QUEUE;
2376 //@node Garbage Collextion Routines, Blocking Queue Routines, Run queue code, Main scheduling code
2377 //@subsection Garbage Collextion Routines
2379 /* ---------------------------------------------------------------------------
2380 Where are the roots that we know about?
2382 - all the threads on the runnable queue
2383 - all the threads on the blocked queue
2384 - all the threads on the sleeping queue
2385 - all the thread currently executing a _ccall_GC
2386 - all the "main threads"
2388 ------------------------------------------------------------------------ */
2390 /* This has to be protected either by the scheduler monitor, or by the
2391 garbage collection monitor (probably the latter).
2396 GetRoots(evac_fn evac)
2401 for (i=0; i<=RtsFlags.GranFlags.proc; i++) {
2402 if ((run_queue_hds[i] != END_TSO_QUEUE) && ((run_queue_hds[i] != NULL)))
2403 evac((StgClosure **)&run_queue_hds[i]);
2404 if ((run_queue_tls[i] != END_TSO_QUEUE) && ((run_queue_tls[i] != NULL)))
2405 evac((StgClosure **)&run_queue_tls[i]);
2407 if ((blocked_queue_hds[i] != END_TSO_QUEUE) && ((blocked_queue_hds[i] != NULL)))
2408 evac((StgClosure **)&blocked_queue_hds[i]);
2409 if ((blocked_queue_tls[i] != END_TSO_QUEUE) && ((blocked_queue_tls[i] != NULL)))
2410 evac((StgClosure **)&blocked_queue_tls[i]);
2411 if ((ccalling_threadss[i] != END_TSO_QUEUE) && ((ccalling_threadss[i] != NULL)))
2412 evac((StgClosure **)&ccalling_threads[i]);
2419 if (run_queue_hd != END_TSO_QUEUE) {
2420 ASSERT(run_queue_tl != END_TSO_QUEUE);
2421 evac((StgClosure **)&run_queue_hd);
2422 evac((StgClosure **)&run_queue_tl);
2425 if (blocked_queue_hd != END_TSO_QUEUE) {
2426 ASSERT(blocked_queue_tl != END_TSO_QUEUE);
2427 evac((StgClosure **)&blocked_queue_hd);
2428 evac((StgClosure **)&blocked_queue_tl);
2431 if (sleeping_queue != END_TSO_QUEUE) {
2432 evac((StgClosure **)&sleeping_queue);
2436 if (suspended_ccalling_threads != END_TSO_QUEUE) {
2437 evac((StgClosure **)&suspended_ccalling_threads);
2440 #if defined(PAR) || defined(GRAN)
2441 markSparkQueue(evac);
2445 /* -----------------------------------------------------------------------------
2448 This is the interface to the garbage collector from Haskell land.
2449 We provide this so that external C code can allocate and garbage
2450 collect when called from Haskell via _ccall_GC.
2452 It might be useful to provide an interface whereby the programmer
2453 can specify more roots (ToDo).
2455 This needs to be protected by the GC condition variable above. KH.
2456 -------------------------------------------------------------------------- */
2458 void (*extra_roots)(evac_fn);
2463 /* Obligated to hold this lock upon entry */
2464 ACQUIRE_LOCK(&sched_mutex);
2465 GarbageCollect(GetRoots,rtsFalse);
2466 RELEASE_LOCK(&sched_mutex);
2470 performMajorGC(void)
2472 ACQUIRE_LOCK(&sched_mutex);
2473 GarbageCollect(GetRoots,rtsTrue);
2474 RELEASE_LOCK(&sched_mutex);
2478 AllRoots(evac_fn evac)
2480 GetRoots(evac); // the scheduler's roots
2481 extra_roots(evac); // the user's roots
2485 performGCWithRoots(void (*get_roots)(evac_fn))
2487 ACQUIRE_LOCK(&sched_mutex);
2488 extra_roots = get_roots;
2489 GarbageCollect(AllRoots,rtsFalse);
2490 RELEASE_LOCK(&sched_mutex);
2493 /* -----------------------------------------------------------------------------
2496 If the thread has reached its maximum stack size, then raise the
2497 StackOverflow exception in the offending thread. Otherwise
2498 relocate the TSO into a larger chunk of memory and adjust its stack
2500 -------------------------------------------------------------------------- */
2503 threadStackOverflow(StgTSO *tso)
2505 nat new_stack_size, new_tso_size, diff, stack_words;
2509 IF_DEBUG(sanity,checkTSO(tso));
2510 if (tso->stack_size >= tso->max_stack_size) {
2513 belch("@@ threadStackOverflow of TSO %d (%p): stack too large (now %ld; max is %ld",
2514 tso->id, tso, tso->stack_size, tso->max_stack_size);
2515 /* If we're debugging, just print out the top of the stack */
2516 printStackChunk(tso->sp, stg_min(tso->stack+tso->stack_size,
2519 /* Send this thread the StackOverflow exception */
2520 raiseAsync(tso, (StgClosure *)stackOverflow_closure);
2524 /* Try to double the current stack size. If that takes us over the
2525 * maximum stack size for this thread, then use the maximum instead.
2526 * Finally round up so the TSO ends up as a whole number of blocks.
2528 new_stack_size = stg_min(tso->stack_size * 2, tso->max_stack_size);
2529 new_tso_size = (nat)BLOCK_ROUND_UP(new_stack_size * sizeof(W_) +
2530 TSO_STRUCT_SIZE)/sizeof(W_);
2531 new_tso_size = round_to_mblocks(new_tso_size); /* Be MBLOCK-friendly */
2532 new_stack_size = new_tso_size - TSO_STRUCT_SIZEW;
2534 IF_DEBUG(scheduler, fprintf(stderr,"== scheduler: increasing stack size from %d words to %d.\n", tso->stack_size, new_stack_size));
2536 dest = (StgTSO *)allocate(new_tso_size);
2537 TICK_ALLOC_TSO(new_stack_size,0);
2539 /* copy the TSO block and the old stack into the new area */
2540 memcpy(dest,tso,TSO_STRUCT_SIZE);
2541 stack_words = tso->stack + tso->stack_size - tso->sp;
2542 new_sp = (P_)dest + new_tso_size - stack_words;
2543 memcpy(new_sp, tso->sp, stack_words * sizeof(W_));
2545 /* relocate the stack pointers... */
2546 diff = (P_)new_sp - (P_)tso->sp; /* In *words* */
2547 dest->su = (StgUpdateFrame *) ((P_)dest->su + diff);
2549 dest->stack_size = new_stack_size;
2551 /* and relocate the update frame list */
2552 relocate_stack(dest, diff);
2554 /* Mark the old TSO as relocated. We have to check for relocated
2555 * TSOs in the garbage collector and any primops that deal with TSOs.
2557 * It's important to set the sp and su values to just beyond the end
2558 * of the stack, so we don't attempt to scavenge any part of the
2561 tso->what_next = ThreadRelocated;
2563 tso->sp = (P_)&(tso->stack[tso->stack_size]);
2564 tso->su = (StgUpdateFrame *)tso->sp;
2565 tso->why_blocked = NotBlocked;
2566 dest->mut_link = NULL;
2568 IF_PAR_DEBUG(verbose,
2569 belch("@@ threadStackOverflow of TSO %d (now at %p): stack size increased to %ld",
2570 tso->id, tso, tso->stack_size);
2571 /* If we're debugging, just print out the top of the stack */
2572 printStackChunk(tso->sp, stg_min(tso->stack+tso->stack_size,
2575 IF_DEBUG(sanity,checkTSO(tso));
2577 IF_DEBUG(scheduler,printTSO(dest));
2583 //@node Blocking Queue Routines, Exception Handling Routines, Garbage Collextion Routines, Main scheduling code
2584 //@subsection Blocking Queue Routines
2586 /* ---------------------------------------------------------------------------
2587 Wake up a queue that was blocked on some resource.
2588 ------------------------------------------------------------------------ */
2592 unblockCount ( StgBlockingQueueElement *bqe, StgClosure *node )
2597 unblockCount ( StgBlockingQueueElement *bqe, StgClosure *node )
2599 /* write RESUME events to log file and
2600 update blocked and fetch time (depending on type of the orig closure) */
2601 if (RtsFlags.ParFlags.ParStats.Full) {
2602 DumpRawGranEvent(CURRENT_PROC, CURRENT_PROC,
2603 GR_RESUMEQ, ((StgTSO *)bqe), ((StgTSO *)bqe)->block_info.closure,
2604 0, 0 /* spark_queue_len(ADVISORY_POOL) */);
2605 if (EMPTY_RUN_QUEUE())
2606 emitSchedule = rtsTrue;
2608 switch (get_itbl(node)->type) {
2610 ((StgTSO *)bqe)->par.fetchtime += CURRENT_TIME-((StgTSO *)bqe)->par.blockedat;
2615 ((StgTSO *)bqe)->par.blocktime += CURRENT_TIME-((StgTSO *)bqe)->par.blockedat;
2622 barf("{unblockOneLocked}Daq Qagh: unexpected closure in blocking queue");
2629 static StgBlockingQueueElement *
2630 unblockOneLocked(StgBlockingQueueElement *bqe, StgClosure *node)
2633 PEs node_loc, tso_loc;
2635 node_loc = where_is(node); // should be lifted out of loop
2636 tso = (StgTSO *)bqe; // wastes an assignment to get the type right
2637 tso_loc = where_is((StgClosure *)tso);
2638 if (IS_LOCAL_TO(PROCS(node),tso_loc)) { // TSO is local
2639 /* !fake_fetch => TSO is on CurrentProc is same as IS_LOCAL_TO */
2640 ASSERT(CurrentProc!=node_loc || tso_loc==CurrentProc);
2641 CurrentTime[CurrentProc] += RtsFlags.GranFlags.Costs.lunblocktime;
2642 // insertThread(tso, node_loc);
2643 new_event(tso_loc, tso_loc, CurrentTime[CurrentProc],
2645 tso, node, (rtsSpark*)NULL);
2646 tso->link = END_TSO_QUEUE; // overwrite link just to be sure
2649 } else { // TSO is remote (actually should be FMBQ)
2650 CurrentTime[CurrentProc] += RtsFlags.GranFlags.Costs.mpacktime +
2651 RtsFlags.GranFlags.Costs.gunblocktime +
2652 RtsFlags.GranFlags.Costs.latency;
2653 new_event(tso_loc, CurrentProc, CurrentTime[CurrentProc],
2655 tso, node, (rtsSpark*)NULL);
2656 tso->link = END_TSO_QUEUE; // overwrite link just to be sure
2659 /* the thread-queue-overhead is accounted for in either Resume or UnblockThread */
2661 fprintf(stderr," %s TSO %d (%p) [PE %d] (block_info.closure=%p) (next=%p) ,",
2662 (node_loc==tso_loc ? "Local" : "Global"),
2663 tso->id, tso, CurrentProc, tso->block_info.closure, tso->link));
2664 tso->block_info.closure = NULL;
2665 IF_DEBUG(scheduler,belch("-- Waking up thread %ld (%p)",
2669 static StgBlockingQueueElement *
2670 unblockOneLocked(StgBlockingQueueElement *bqe, StgClosure *node)
2672 StgBlockingQueueElement *next;
2674 switch (get_itbl(bqe)->type) {
2676 ASSERT(((StgTSO *)bqe)->why_blocked != NotBlocked);
2677 /* if it's a TSO just push it onto the run_queue */
2679 // ((StgTSO *)bqe)->link = END_TSO_QUEUE; // debugging?
2680 PUSH_ON_RUN_QUEUE((StgTSO *)bqe);
2682 unblockCount(bqe, node);
2683 /* reset blocking status after dumping event */
2684 ((StgTSO *)bqe)->why_blocked = NotBlocked;
2688 /* if it's a BLOCKED_FETCH put it on the PendingFetches list */
2690 bqe->link = (StgBlockingQueueElement *)PendingFetches;
2691 PendingFetches = (StgBlockedFetch *)bqe;
2695 /* can ignore this case in a non-debugging setup;
2696 see comments on RBHSave closures above */
2698 /* check that the closure is an RBHSave closure */
2699 ASSERT(get_itbl((StgClosure *)bqe) == &stg_RBH_Save_0_info ||
2700 get_itbl((StgClosure *)bqe) == &stg_RBH_Save_1_info ||
2701 get_itbl((StgClosure *)bqe) == &stg_RBH_Save_2_info);
2705 barf("{unblockOneLocked}Daq Qagh: Unexpected IP (%#lx; %s) in blocking queue at %#lx\n",
2706 get_itbl((StgClosure *)bqe), info_type((StgClosure *)bqe),
2710 IF_PAR_DEBUG(bq, fprintf(stderr, ", %p (%s)", bqe, info_type((StgClosure*)bqe)));
2714 #else /* !GRAN && !PAR */
2716 unblockOneLocked(StgTSO *tso)
2720 ASSERT(get_itbl(tso)->type == TSO);
2721 ASSERT(tso->why_blocked != NotBlocked);
2722 tso->why_blocked = NotBlocked;
2724 PUSH_ON_RUN_QUEUE(tso);
2726 IF_DEBUG(scheduler,sched_belch("waking up thread %ld", tso->id));
2731 #if defined(GRAN) || defined(PAR)
2732 inline StgBlockingQueueElement *
2733 unblockOne(StgBlockingQueueElement *bqe, StgClosure *node)
2735 ACQUIRE_LOCK(&sched_mutex);
2736 bqe = unblockOneLocked(bqe, node);
2737 RELEASE_LOCK(&sched_mutex);
2742 unblockOne(StgTSO *tso)
2744 ACQUIRE_LOCK(&sched_mutex);
2745 tso = unblockOneLocked(tso);
2746 RELEASE_LOCK(&sched_mutex);
2753 awakenBlockedQueue(StgBlockingQueueElement *q, StgClosure *node)
2755 StgBlockingQueueElement *bqe;
2760 belch("##-_ AwBQ for node %p on PE %d @ %ld by TSO %d (%p): ", \
2761 node, CurrentProc, CurrentTime[CurrentProc],
2762 CurrentTSO->id, CurrentTSO));
2764 node_loc = where_is(node);
2766 ASSERT(q == END_BQ_QUEUE ||
2767 get_itbl(q)->type == TSO || // q is either a TSO or an RBHSave
2768 get_itbl(q)->type == CONSTR); // closure (type constructor)
2769 ASSERT(is_unique(node));
2771 /* FAKE FETCH: magically copy the node to the tso's proc;
2772 no Fetch necessary because in reality the node should not have been
2773 moved to the other PE in the first place
2775 if (CurrentProc!=node_loc) {
2777 belch("## node %p is on PE %d but CurrentProc is %d (TSO %d); assuming fake fetch and adjusting bitmask (old: %#x)",
2778 node, node_loc, CurrentProc, CurrentTSO->id,
2779 // CurrentTSO, where_is(CurrentTSO),
2780 node->header.gran.procs));
2781 node->header.gran.procs = (node->header.gran.procs) | PE_NUMBER(CurrentProc);
2783 belch("## new bitmask of node %p is %#x",
2784 node, node->header.gran.procs));
2785 if (RtsFlags.GranFlags.GranSimStats.Global) {
2786 globalGranStats.tot_fake_fetches++;
2791 // ToDo: check: ASSERT(CurrentProc==node_loc);
2792 while (get_itbl(bqe)->type==TSO) { // q != END_TSO_QUEUE) {
2795 bqe points to the current element in the queue
2796 next points to the next element in the queue
2798 //tso = (StgTSO *)bqe; // wastes an assignment to get the type right
2799 //tso_loc = where_is(tso);
2801 bqe = unblockOneLocked(bqe, node);
2804 /* if this is the BQ of an RBH, we have to put back the info ripped out of
2805 the closure to make room for the anchor of the BQ */
2806 if (bqe!=END_BQ_QUEUE) {
2807 ASSERT(get_itbl(node)->type == RBH && get_itbl(bqe)->type == CONSTR);
2809 ASSERT((info_ptr==&RBH_Save_0_info) ||
2810 (info_ptr==&RBH_Save_1_info) ||
2811 (info_ptr==&RBH_Save_2_info));
2813 /* cf. convertToRBH in RBH.c for writing the RBHSave closure */
2814 ((StgRBH *)node)->blocking_queue = (StgBlockingQueueElement *)((StgRBHSave *)bqe)->payload[0];
2815 ((StgRBH *)node)->mut_link = (StgMutClosure *)((StgRBHSave *)bqe)->payload[1];
2818 belch("## Filled in RBH_Save for %p (%s) at end of AwBQ",
2819 node, info_type(node)));
2822 /* statistics gathering */
2823 if (RtsFlags.GranFlags.GranSimStats.Global) {
2824 // globalGranStats.tot_bq_processing_time += bq_processing_time;
2825 globalGranStats.tot_bq_len += len; // total length of all bqs awakened
2826 // globalGranStats.tot_bq_len_local += len_local; // same for local TSOs only
2827 globalGranStats.tot_awbq++; // total no. of bqs awakened
2830 fprintf(stderr,"## BQ Stats of %p: [%d entries] %s\n",
2831 node, len, (bqe!=END_BQ_QUEUE) ? "RBH" : ""));
2835 awakenBlockedQueue(StgBlockingQueueElement *q, StgClosure *node)
2837 StgBlockingQueueElement *bqe;
2839 ACQUIRE_LOCK(&sched_mutex);
2841 IF_PAR_DEBUG(verbose,
2842 belch("##-_ AwBQ for node %p on [%x]: ",
2846 if(get_itbl(q)->type == CONSTR || q==END_BQ_QUEUE) {
2847 IF_PAR_DEBUG(verbose, belch("## ... nothing to unblock so lets just return. RFP (BUG?)"));
2852 ASSERT(q == END_BQ_QUEUE ||
2853 get_itbl(q)->type == TSO ||
2854 get_itbl(q)->type == BLOCKED_FETCH ||
2855 get_itbl(q)->type == CONSTR);
2858 while (get_itbl(bqe)->type==TSO ||
2859 get_itbl(bqe)->type==BLOCKED_FETCH) {
2860 bqe = unblockOneLocked(bqe, node);
2862 RELEASE_LOCK(&sched_mutex);
2865 #else /* !GRAN && !PAR */
2867 awakenBlockedQueue(StgTSO *tso)
2869 ACQUIRE_LOCK(&sched_mutex);
2870 while (tso != END_TSO_QUEUE) {
2871 tso = unblockOneLocked(tso);
2873 RELEASE_LOCK(&sched_mutex);
2877 //@node Exception Handling Routines, Debugging Routines, Blocking Queue Routines, Main scheduling code
2878 //@subsection Exception Handling Routines
2880 /* ---------------------------------------------------------------------------
2882 - usually called inside a signal handler so it mustn't do anything fancy.
2883 ------------------------------------------------------------------------ */
2886 interruptStgRts(void)
2892 /* -----------------------------------------------------------------------------
2895 This is for use when we raise an exception in another thread, which
2897 This has nothing to do with the UnblockThread event in GranSim. -- HWL
2898 -------------------------------------------------------------------------- */
2900 #if defined(GRAN) || defined(PAR)
2902 NB: only the type of the blocking queue is different in GranSim and GUM
2903 the operations on the queue-elements are the same
2904 long live polymorphism!
2906 Locks: sched_mutex is held upon entry and exit.
2910 unblockThread(StgTSO *tso)
2912 StgBlockingQueueElement *t, **last;
2914 switch (tso->why_blocked) {
2917 return; /* not blocked */
2920 ASSERT(get_itbl(tso->block_info.closure)->type == MVAR);
2922 StgBlockingQueueElement *last_tso = END_BQ_QUEUE;
2923 StgMVar *mvar = (StgMVar *)(tso->block_info.closure);
2925 last = (StgBlockingQueueElement **)&mvar->head;
2926 for (t = (StgBlockingQueueElement *)mvar->head;
2928 last = &t->link, last_tso = t, t = t->link) {
2929 if (t == (StgBlockingQueueElement *)tso) {
2930 *last = (StgBlockingQueueElement *)tso->link;
2931 if (mvar->tail == tso) {
2932 mvar->tail = (StgTSO *)last_tso;
2937 barf("unblockThread (MVAR): TSO not found");
2940 case BlockedOnBlackHole:
2941 ASSERT(get_itbl(tso->block_info.closure)->type == BLACKHOLE_BQ);
2943 StgBlockingQueue *bq = (StgBlockingQueue *)(tso->block_info.closure);
2945 last = &bq->blocking_queue;
2946 for (t = bq->blocking_queue;
2948 last = &t->link, t = t->link) {
2949 if (t == (StgBlockingQueueElement *)tso) {
2950 *last = (StgBlockingQueueElement *)tso->link;
2954 barf("unblockThread (BLACKHOLE): TSO not found");
2957 case BlockedOnException:
2959 StgTSO *target = tso->block_info.tso;
2961 ASSERT(get_itbl(target)->type == TSO);
2963 if (target->what_next == ThreadRelocated) {
2964 target = target->link;
2965 ASSERT(get_itbl(target)->type == TSO);
2968 ASSERT(target->blocked_exceptions != NULL);
2970 last = (StgBlockingQueueElement **)&target->blocked_exceptions;
2971 for (t = (StgBlockingQueueElement *)target->blocked_exceptions;
2973 last = &t->link, t = t->link) {
2974 ASSERT(get_itbl(t)->type == TSO);
2975 if (t == (StgBlockingQueueElement *)tso) {
2976 *last = (StgBlockingQueueElement *)tso->link;
2980 barf("unblockThread (Exception): TSO not found");
2984 case BlockedOnWrite:
2986 /* take TSO off blocked_queue */
2987 StgBlockingQueueElement *prev = NULL;
2988 for (t = (StgBlockingQueueElement *)blocked_queue_hd; t != END_BQ_QUEUE;
2989 prev = t, t = t->link) {
2990 if (t == (StgBlockingQueueElement *)tso) {
2992 blocked_queue_hd = (StgTSO *)t->link;
2993 if ((StgBlockingQueueElement *)blocked_queue_tl == t) {
2994 blocked_queue_tl = END_TSO_QUEUE;
2997 prev->link = t->link;
2998 if ((StgBlockingQueueElement *)blocked_queue_tl == t) {
2999 blocked_queue_tl = (StgTSO *)prev;
3005 barf("unblockThread (I/O): TSO not found");
3008 case BlockedOnDelay:
3010 /* take TSO off sleeping_queue */
3011 StgBlockingQueueElement *prev = NULL;
3012 for (t = (StgBlockingQueueElement *)sleeping_queue; t != END_BQ_QUEUE;
3013 prev = t, t = t->link) {
3014 if (t == (StgBlockingQueueElement *)tso) {
3016 sleeping_queue = (StgTSO *)t->link;
3018 prev->link = t->link;
3023 barf("unblockThread (I/O): TSO not found");
3027 barf("unblockThread");
3031 tso->link = END_TSO_QUEUE;
3032 tso->why_blocked = NotBlocked;
3033 tso->block_info.closure = NULL;
3034 PUSH_ON_RUN_QUEUE(tso);
3038 unblockThread(StgTSO *tso)
3042 /* To avoid locking unnecessarily. */
3043 if (tso->why_blocked == NotBlocked) {
3047 switch (tso->why_blocked) {
3050 ASSERT(get_itbl(tso->block_info.closure)->type == MVAR);
3052 StgTSO *last_tso = END_TSO_QUEUE;
3053 StgMVar *mvar = (StgMVar *)(tso->block_info.closure);
3056 for (t = mvar->head; t != END_TSO_QUEUE;
3057 last = &t->link, last_tso = t, t = t->link) {
3060 if (mvar->tail == tso) {
3061 mvar->tail = last_tso;
3066 barf("unblockThread (MVAR): TSO not found");
3069 case BlockedOnBlackHole:
3070 ASSERT(get_itbl(tso->block_info.closure)->type == BLACKHOLE_BQ);
3072 StgBlockingQueue *bq = (StgBlockingQueue *)(tso->block_info.closure);
3074 last = &bq->blocking_queue;
3075 for (t = bq->blocking_queue; t != END_TSO_QUEUE;
3076 last = &t->link, t = t->link) {
3082 barf("unblockThread (BLACKHOLE): TSO not found");
3085 case BlockedOnException:
3087 StgTSO *target = tso->block_info.tso;
3089 ASSERT(get_itbl(target)->type == TSO);
3091 while (target->what_next == ThreadRelocated) {
3092 target = target->link;
3093 ASSERT(get_itbl(target)->type == TSO);
3096 ASSERT(target->blocked_exceptions != NULL);
3098 last = &target->blocked_exceptions;
3099 for (t = target->blocked_exceptions; t != END_TSO_QUEUE;
3100 last = &t->link, t = t->link) {
3101 ASSERT(get_itbl(t)->type == TSO);
3107 barf("unblockThread (Exception): TSO not found");
3111 case BlockedOnWrite:
3113 StgTSO *prev = NULL;
3114 for (t = blocked_queue_hd; t != END_TSO_QUEUE;
3115 prev = t, t = t->link) {
3118 blocked_queue_hd = t->link;
3119 if (blocked_queue_tl == t) {
3120 blocked_queue_tl = END_TSO_QUEUE;
3123 prev->link = t->link;
3124 if (blocked_queue_tl == t) {
3125 blocked_queue_tl = prev;
3131 barf("unblockThread (I/O): TSO not found");
3134 case BlockedOnDelay:
3136 StgTSO *prev = NULL;
3137 for (t = sleeping_queue; t != END_TSO_QUEUE;
3138 prev = t, t = t->link) {
3141 sleeping_queue = t->link;
3143 prev->link = t->link;
3148 barf("unblockThread (I/O): TSO not found");
3152 barf("unblockThread");
3156 tso->link = END_TSO_QUEUE;
3157 tso->why_blocked = NotBlocked;
3158 tso->block_info.closure = NULL;
3159 PUSH_ON_RUN_QUEUE(tso);
3163 /* -----------------------------------------------------------------------------
3166 * The following function implements the magic for raising an
3167 * asynchronous exception in an existing thread.
3169 * We first remove the thread from any queue on which it might be
3170 * blocked. The possible blockages are MVARs and BLACKHOLE_BQs.
3172 * We strip the stack down to the innermost CATCH_FRAME, building
3173 * thunks in the heap for all the active computations, so they can
3174 * be restarted if necessary. When we reach a CATCH_FRAME, we build
3175 * an application of the handler to the exception, and push it on
3176 * the top of the stack.
3178 * How exactly do we save all the active computations? We create an
3179 * AP_UPD for every UpdateFrame on the stack. Entering one of these
3180 * AP_UPDs pushes everything from the corresponding update frame
3181 * upwards onto the stack. (Actually, it pushes everything up to the
3182 * next update frame plus a pointer to the next AP_UPD object.
3183 * Entering the next AP_UPD object pushes more onto the stack until we
3184 * reach the last AP_UPD object - at which point the stack should look
3185 * exactly as it did when we killed the TSO and we can continue
3186 * execution by entering the closure on top of the stack.
3188 * We can also kill a thread entirely - this happens if either (a) the
3189 * exception passed to raiseAsync is NULL, or (b) there's no
3190 * CATCH_FRAME on the stack. In either case, we strip the entire
3191 * stack and replace the thread with a zombie.
3193 * Locks: sched_mutex held upon entry nor exit.
3195 * -------------------------------------------------------------------------- */
3198 deleteThread(StgTSO *tso)
3200 raiseAsync(tso,NULL);
3204 raiseAsyncWithLock(StgTSO *tso, StgClosure *exception)
3206 /* When raising async exs from contexts where sched_mutex isn't held;
3207 use raiseAsyncWithLock(). */
3208 ACQUIRE_LOCK(&sched_mutex);
3209 raiseAsync(tso,exception);
3210 RELEASE_LOCK(&sched_mutex);
3214 raiseAsync(StgTSO *tso, StgClosure *exception)
3216 StgUpdateFrame* su = tso->su;
3217 StgPtr sp = tso->sp;
3219 /* Thread already dead? */
3220 if (tso->what_next == ThreadComplete || tso->what_next == ThreadKilled) {
3224 IF_DEBUG(scheduler, sched_belch("raising exception in thread %ld.", tso->id));
3226 /* Remove it from any blocking queues */
3229 IF_DEBUG(scheduler, sched_belch("raising exception in thread %ld.", tso->id));
3230 /* The stack freezing code assumes there's a closure pointer on
3231 * the top of the stack. This isn't always the case with compiled
3232 * code, so we have to push a dummy closure on the top which just
3233 * returns to the next return address on the stack.
3235 if ( LOOKS_LIKE_GHC_INFO((void*)*sp) ) {
3236 *(--sp) = (W_)&stg_dummy_ret_closure;
3240 nat words = ((P_)su - (P_)sp) - 1;
3244 /* If we find a CATCH_FRAME, and we've got an exception to raise,
3245 * then build the THUNK raise(exception), and leave it on
3246 * top of the CATCH_FRAME ready to enter.
3248 if (get_itbl(su)->type == CATCH_FRAME && exception != NULL) {
3250 StgCatchFrame *cf = (StgCatchFrame *)su;
3254 /* we've got an exception to raise, so let's pass it to the
3255 * handler in this frame.
3257 raise = (StgClosure *)allocate(sizeofW(StgClosure)+1);
3258 TICK_ALLOC_SE_THK(1,0);
3259 SET_HDR(raise,&stg_raise_info,cf->header.prof.ccs);
3260 raise->payload[0] = exception;
3262 /* throw away the stack from Sp up to the CATCH_FRAME.
3266 /* Ensure that async excpetions are blocked now, so we don't get
3267 * a surprise exception before we get around to executing the
3270 if (tso->blocked_exceptions == NULL) {
3271 tso->blocked_exceptions = END_TSO_QUEUE;
3274 /* Put the newly-built THUNK on top of the stack, ready to execute
3275 * when the thread restarts.
3280 tso->what_next = ThreadEnterGHC;
3281 IF_DEBUG(sanity, checkTSO(tso));
3285 /* First build an AP_UPD consisting of the stack chunk above the
3286 * current update frame, with the top word on the stack as the
3289 ap = (StgAP_UPD *)allocate(AP_sizeW(words));
3294 ap->fun = (StgClosure *)sp[0];
3296 for(i=0; i < (nat)words; ++i) {
3297 ap->payload[i] = (StgClosure *)*sp++;
3300 switch (get_itbl(su)->type) {
3304 SET_HDR(ap,&stg_AP_UPD_info,su->header.prof.ccs /* ToDo */);
3305 TICK_ALLOC_UP_THK(words+1,0);
3308 fprintf(stderr, "scheduler: Updating ");
3309 printPtr((P_)su->updatee);
3310 fprintf(stderr, " with ");
3311 printObj((StgClosure *)ap);
3314 /* Replace the updatee with an indirection - happily
3315 * this will also wake up any threads currently
3316 * waiting on the result.
3318 * Warning: if we're in a loop, more than one update frame on
3319 * the stack may point to the same object. Be careful not to
3320 * overwrite an IND_OLDGEN in this case, because we'll screw
3321 * up the mutable lists. To be on the safe side, don't
3322 * overwrite any kind of indirection at all. See also
3323 * threadSqueezeStack in GC.c, where we have to make a similar
3326 if (!closure_IND(su->updatee)) {
3327 UPD_IND_NOLOCK(su->updatee,ap); /* revert the black hole */
3330 sp += sizeofW(StgUpdateFrame) -1;
3331 sp[0] = (W_)ap; /* push onto stack */
3337 StgCatchFrame *cf = (StgCatchFrame *)su;
3340 /* We want a PAP, not an AP_UPD. Fortunately, the
3341 * layout's the same.
3343 SET_HDR(ap,&stg_PAP_info,su->header.prof.ccs /* ToDo */);
3344 TICK_ALLOC_UPD_PAP(words+1,0);
3346 /* now build o = FUN(catch,ap,handler) */
3347 o = (StgClosure *)allocate(sizeofW(StgClosure)+2);
3348 TICK_ALLOC_FUN(2,0);
3349 SET_HDR(o,&stg_catch_info,su->header.prof.ccs /* ToDo */);
3350 o->payload[0] = (StgClosure *)ap;
3351 o->payload[1] = cf->handler;
3354 fprintf(stderr, "scheduler: Built ");
3355 printObj((StgClosure *)o);
3358 /* pop the old handler and put o on the stack */
3360 sp += sizeofW(StgCatchFrame) - 1;
3367 StgSeqFrame *sf = (StgSeqFrame *)su;
3370 SET_HDR(ap,&stg_PAP_info,su->header.prof.ccs /* ToDo */);
3371 TICK_ALLOC_UPD_PAP(words+1,0);
3373 /* now build o = FUN(seq,ap) */
3374 o = (StgClosure *)allocate(sizeofW(StgClosure)+1);
3375 TICK_ALLOC_SE_THK(1,0);
3376 SET_HDR(o,&stg_seq_info,su->header.prof.ccs /* ToDo */);
3377 o->payload[0] = (StgClosure *)ap;
3380 fprintf(stderr, "scheduler: Built ");
3381 printObj((StgClosure *)o);
3384 /* pop the old handler and put o on the stack */
3386 sp += sizeofW(StgSeqFrame) - 1;
3392 /* We've stripped the entire stack, the thread is now dead. */
3393 sp += sizeofW(StgStopFrame) - 1;
3394 sp[0] = (W_)exception; /* save the exception */
3395 tso->what_next = ThreadKilled;
3396 tso->su = (StgUpdateFrame *)(sp+1);
3407 /* -----------------------------------------------------------------------------
3408 resurrectThreads is called after garbage collection on the list of
3409 threads found to be garbage. Each of these threads will be woken
3410 up and sent a signal: BlockedOnDeadMVar if the thread was blocked
3411 on an MVar, or NonTermination if the thread was blocked on a Black
3414 Locks: sched_mutex isn't held upon entry nor exit.
3415 -------------------------------------------------------------------------- */
3418 resurrectThreads( StgTSO *threads )
3422 for (tso = threads; tso != END_TSO_QUEUE; tso = next) {
3423 next = tso->global_link;
3424 tso->global_link = all_threads;
3426 IF_DEBUG(scheduler, sched_belch("resurrecting thread %d", tso->id));
3428 switch (tso->why_blocked) {
3430 case BlockedOnException:
3431 /* Called by GC - sched_mutex lock is currently held. */
3432 raiseAsync(tso,(StgClosure *)BlockedOnDeadMVar_closure);
3434 case BlockedOnBlackHole:
3435 raiseAsync(tso,(StgClosure *)NonTermination_closure);
3438 /* This might happen if the thread was blocked on a black hole
3439 * belonging to a thread that we've just woken up (raiseAsync
3440 * can wake up threads, remember...).
3444 barf("resurrectThreads: thread blocked in a strange way");
3449 /* -----------------------------------------------------------------------------
3450 * Blackhole detection: if we reach a deadlock, test whether any
3451 * threads are blocked on themselves. Any threads which are found to
3452 * be self-blocked get sent a NonTermination exception.
3454 * This is only done in a deadlock situation in order to avoid
3455 * performance overhead in the normal case.
3457 * Locks: sched_mutex is held upon entry and exit.
3458 * -------------------------------------------------------------------------- */
3461 detectBlackHoles( void )
3463 StgTSO *t = all_threads;
3464 StgUpdateFrame *frame;
3465 StgClosure *blocked_on;
3467 for (t = all_threads; t != END_TSO_QUEUE; t = t->global_link) {
3469 while (t->what_next == ThreadRelocated) {
3471 ASSERT(get_itbl(t)->type == TSO);
3474 if (t->why_blocked != BlockedOnBlackHole) {
3478 blocked_on = t->block_info.closure;
3480 for (frame = t->su; ; frame = frame->link) {
3481 switch (get_itbl(frame)->type) {
3484 if (frame->updatee == blocked_on) {
3485 /* We are blocking on one of our own computations, so
3486 * send this thread the NonTermination exception.
3489 sched_belch("thread %d is blocked on itself", t->id));
3490 raiseAsync(t, (StgClosure *)NonTermination_closure);
3511 //@node Debugging Routines, Index, Exception Handling Routines, Main scheduling code
3512 //@subsection Debugging Routines
3514 /* -----------------------------------------------------------------------------
3515 Debugging: why is a thread blocked
3516 -------------------------------------------------------------------------- */
3521 printThreadBlockage(StgTSO *tso)
3523 switch (tso->why_blocked) {
3525 fprintf(stderr,"is blocked on read from fd %d", tso->block_info.fd);
3527 case BlockedOnWrite:
3528 fprintf(stderr,"is blocked on write to fd %d", tso->block_info.fd);
3530 case BlockedOnDelay:
3531 fprintf(stderr,"is blocked until %d", tso->block_info.target);
3534 fprintf(stderr,"is blocked on an MVar");
3536 case BlockedOnException:
3537 fprintf(stderr,"is blocked on delivering an exception to thread %d",
3538 tso->block_info.tso->id);
3540 case BlockedOnBlackHole:
3541 fprintf(stderr,"is blocked on a black hole");
3544 fprintf(stderr,"is not blocked");
3548 fprintf(stderr,"is blocked on global address; local FM_BQ is %p (%s)",
3549 tso->block_info.closure, info_type(tso->block_info.closure));
3551 case BlockedOnGA_NoSend:
3552 fprintf(stderr,"is blocked on global address (no send); local FM_BQ is %p (%s)",
3553 tso->block_info.closure, info_type(tso->block_info.closure));
3556 #if defined(RTS_SUPPORTS_THREADS)
3557 case BlockedOnCCall:
3558 fprintf(stderr,"is blocked on an external call");
3562 barf("printThreadBlockage: strange tso->why_blocked: %d for TSO %d (%d)",
3563 tso->why_blocked, tso->id, tso);
3568 printThreadStatus(StgTSO *tso)
3570 switch (tso->what_next) {
3572 fprintf(stderr,"has been killed");
3574 case ThreadComplete:
3575 fprintf(stderr,"has completed");
3578 printThreadBlockage(tso);
3583 printAllThreads(void)
3588 char time_string[TIME_STR_LEN], node_str[NODE_STR_LEN];
3589 ullong_format_string(TIME_ON_PROC(CurrentProc),
3590 time_string, rtsFalse/*no commas!*/);
3592 sched_belch("all threads at [%s]:", time_string);
3594 char time_string[TIME_STR_LEN], node_str[NODE_STR_LEN];
3595 ullong_format_string(CURRENT_TIME,
3596 time_string, rtsFalse/*no commas!*/);
3598 sched_belch("all threads at [%s]:", time_string);
3600 sched_belch("all threads:");
3603 for (t = all_threads; t != END_TSO_QUEUE; t = t->global_link) {
3604 fprintf(stderr, "\tthread %d ", t->id);
3605 if (t->label) fprintf(stderr,"[\"%s\"] ",t->label);
3606 printThreadStatus(t);
3607 fprintf(stderr,"\n");
3612 Print a whole blocking queue attached to node (debugging only).
3617 print_bq (StgClosure *node)
3619 StgBlockingQueueElement *bqe;
3623 fprintf(stderr,"## BQ of closure %p (%s): ",
3624 node, info_type(node));
3626 /* should cover all closures that may have a blocking queue */
3627 ASSERT(get_itbl(node)->type == BLACKHOLE_BQ ||
3628 get_itbl(node)->type == FETCH_ME_BQ ||
3629 get_itbl(node)->type == RBH ||
3630 get_itbl(node)->type == MVAR);
3632 ASSERT(node!=(StgClosure*)NULL); // sanity check
3634 print_bqe(((StgBlockingQueue*)node)->blocking_queue);
3638 Print a whole blocking queue starting with the element bqe.
3641 print_bqe (StgBlockingQueueElement *bqe)
3646 NB: In a parallel setup a BQ of an RBH must end with an RBH_Save closure;
3648 for (end = (bqe==END_BQ_QUEUE);
3649 !end; // iterate until bqe points to a CONSTR
3650 end = (get_itbl(bqe)->type == CONSTR) || (bqe->link==END_BQ_QUEUE),
3651 bqe = end ? END_BQ_QUEUE : bqe->link) {
3652 ASSERT(bqe != END_BQ_QUEUE); // sanity check
3653 ASSERT(bqe != (StgBlockingQueueElement *)NULL); // sanity check
3654 /* types of closures that may appear in a blocking queue */
3655 ASSERT(get_itbl(bqe)->type == TSO ||
3656 get_itbl(bqe)->type == BLOCKED_FETCH ||
3657 get_itbl(bqe)->type == CONSTR);
3658 /* only BQs of an RBH end with an RBH_Save closure */
3659 //ASSERT(get_itbl(bqe)->type != CONSTR || get_itbl(node)->type == RBH);
3661 switch (get_itbl(bqe)->type) {
3663 fprintf(stderr," TSO %u (%x),",
3664 ((StgTSO *)bqe)->id, ((StgTSO *)bqe));
3667 fprintf(stderr," BF (node=%p, ga=((%x, %d, %x)),",
3668 ((StgBlockedFetch *)bqe)->node,
3669 ((StgBlockedFetch *)bqe)->ga.payload.gc.gtid,
3670 ((StgBlockedFetch *)bqe)->ga.payload.gc.slot,
3671 ((StgBlockedFetch *)bqe)->ga.weight);
3674 fprintf(stderr," %s (IP %p),",
3675 (get_itbl(bqe) == &stg_RBH_Save_0_info ? "RBH_Save_0" :
3676 get_itbl(bqe) == &stg_RBH_Save_1_info ? "RBH_Save_1" :
3677 get_itbl(bqe) == &stg_RBH_Save_2_info ? "RBH_Save_2" :
3678 "RBH_Save_?"), get_itbl(bqe));
3681 barf("Unexpected closure type %s in blocking queue", // of %p (%s)",
3682 info_type((StgClosure *)bqe)); // , node, info_type(node));
3686 fputc('\n', stderr);
3688 # elif defined(GRAN)
3690 print_bq (StgClosure *node)
3692 StgBlockingQueueElement *bqe;
3693 PEs node_loc, tso_loc;
3696 /* should cover all closures that may have a blocking queue */
3697 ASSERT(get_itbl(node)->type == BLACKHOLE_BQ ||
3698 get_itbl(node)->type == FETCH_ME_BQ ||
3699 get_itbl(node)->type == RBH);
3701 ASSERT(node!=(StgClosure*)NULL); // sanity check
3702 node_loc = where_is(node);
3704 fprintf(stderr,"## BQ of closure %p (%s) on [PE %d]: ",
3705 node, info_type(node), node_loc);
3708 NB: In a parallel setup a BQ of an RBH must end with an RBH_Save closure;
3710 for (bqe = ((StgBlockingQueue*)node)->blocking_queue, end = (bqe==END_BQ_QUEUE);
3711 !end; // iterate until bqe points to a CONSTR
3712 end = (get_itbl(bqe)->type == CONSTR) || (bqe->link==END_BQ_QUEUE), bqe = end ? END_BQ_QUEUE : bqe->link) {
3713 ASSERT(bqe != END_BQ_QUEUE); // sanity check
3714 ASSERT(bqe != (StgBlockingQueueElement *)NULL); // sanity check
3715 /* types of closures that may appear in a blocking queue */
3716 ASSERT(get_itbl(bqe)->type == TSO ||
3717 get_itbl(bqe)->type == CONSTR);
3718 /* only BQs of an RBH end with an RBH_Save closure */
3719 ASSERT(get_itbl(bqe)->type != CONSTR || get_itbl(node)->type == RBH);
3721 tso_loc = where_is((StgClosure *)bqe);
3722 switch (get_itbl(bqe)->type) {
3724 fprintf(stderr," TSO %d (%p) on [PE %d],",
3725 ((StgTSO *)bqe)->id, (StgTSO *)bqe, tso_loc);
3728 fprintf(stderr," %s (IP %p),",
3729 (get_itbl(bqe) == &stg_RBH_Save_0_info ? "RBH_Save_0" :
3730 get_itbl(bqe) == &stg_RBH_Save_1_info ? "RBH_Save_1" :
3731 get_itbl(bqe) == &stg_RBH_Save_2_info ? "RBH_Save_2" :
3732 "RBH_Save_?"), get_itbl(bqe));
3735 barf("Unexpected closure type %s in blocking queue of %p (%s)",
3736 info_type((StgClosure *)bqe), node, info_type(node));
3740 fputc('\n', stderr);
3744 Nice and easy: only TSOs on the blocking queue
3747 print_bq (StgClosure *node)
3751 ASSERT(node!=(StgClosure*)NULL); // sanity check
3752 for (tso = ((StgBlockingQueue*)node)->blocking_queue;
3753 tso != END_TSO_QUEUE;
3755 ASSERT(tso!=NULL && tso!=END_TSO_QUEUE); // sanity check
3756 ASSERT(get_itbl(tso)->type == TSO); // guess what, sanity check
3757 fprintf(stderr," TSO %d (%p),", tso->id, tso);
3759 fputc('\n', stderr);
3770 for (i=0, tso=run_queue_hd;
3771 tso != END_TSO_QUEUE;
3780 sched_belch(char *s, ...)
3785 fprintf(stderr, "scheduler (task %ld): ", osThreadId());
3787 fprintf(stderr, "== ");
3789 fprintf(stderr, "scheduler: ");
3791 vfprintf(stderr, s, ap);
3792 fprintf(stderr, "\n");
3799 //@node Index, , Debugging Routines, Main scheduling code
3803 //* StgMainThread:: @cindex\s-+StgMainThread
3804 //* awaken_blocked_queue:: @cindex\s-+awaken_blocked_queue
3805 //* blocked_queue_hd:: @cindex\s-+blocked_queue_hd
3806 //* blocked_queue_tl:: @cindex\s-+blocked_queue_tl
3807 //* context_switch:: @cindex\s-+context_switch
3808 //* createThread:: @cindex\s-+createThread
3809 //* gc_pending_cond:: @cindex\s-+gc_pending_cond
3810 //* initScheduler:: @cindex\s-+initScheduler
3811 //* interrupted:: @cindex\s-+interrupted
3812 //* next_thread_id:: @cindex\s-+next_thread_id
3813 //* print_bq:: @cindex\s-+print_bq
3814 //* run_queue_hd:: @cindex\s-+run_queue_hd
3815 //* run_queue_tl:: @cindex\s-+run_queue_tl
3816 //* sched_mutex:: @cindex\s-+sched_mutex
3817 //* schedule:: @cindex\s-+schedule
3818 //* take_off_run_queue:: @cindex\s-+take_off_run_queue
3819 //* term_mutex:: @cindex\s-+term_mutex