1 /* ---------------------------------------------------------------------------
2 * $Id: Schedule.c,v 1.114 2002/01/31 11:18:07 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"
118 //@node Variables and Data structures, Prototypes, Includes, Main scheduling code
119 //@subsection Variables and Data structures
123 * These are the threads which clients have requested that we run.
125 * In a 'threaded' build, we might have several concurrent clients all
126 * waiting for results, and each one will wait on a condition variable
127 * until the result is available.
129 * In non-SMP, clients are strictly nested: the first client calls
130 * into the RTS, which might call out again to C with a _ccall_GC, and
131 * eventually re-enter the RTS.
133 * Main threads information is kept in a linked list:
135 //@cindex StgMainThread
136 typedef struct StgMainThread_ {
138 SchedulerStatus stat;
140 #if defined(RTS_SUPPORTS_THREADS)
143 struct StgMainThread_ *link;
146 /* Main thread queue.
147 * Locks required: sched_mutex.
149 static StgMainThread *main_threads;
152 * Locks required: sched_mutex.
156 StgTSO* ActiveTSO = NULL; /* for assigning system costs; GranSim-Light only */
157 /* rtsTime TimeOfNextEvent, EndOfTimeSlice; now in GranSim.c */
160 In GranSim we have a runable and a blocked queue for each processor.
161 In order to minimise code changes new arrays run_queue_hds/tls
162 are created. run_queue_hd is then a short cut (macro) for
163 run_queue_hds[CurrentProc] (see GranSim.h).
166 StgTSO *run_queue_hds[MAX_PROC], *run_queue_tls[MAX_PROC];
167 StgTSO *blocked_queue_hds[MAX_PROC], *blocked_queue_tls[MAX_PROC];
168 StgTSO *ccalling_threadss[MAX_PROC];
169 /* We use the same global list of threads (all_threads) in GranSim as in
170 the std RTS (i.e. we are cheating). However, we don't use this list in
171 the GranSim specific code at the moment (so we are only potentially
176 StgTSO *run_queue_hd, *run_queue_tl;
177 StgTSO *blocked_queue_hd, *blocked_queue_tl;
178 StgTSO *sleeping_queue; /* perhaps replace with a hash table? */
182 /* Linked list of all threads.
183 * Used for detecting garbage collected threads.
187 /* Threads suspended in _ccall_GC.
189 static StgTSO *suspended_ccalling_threads;
191 static StgTSO *threadStackOverflow(StgTSO *tso);
193 /* KH: The following two flags are shared memory locations. There is no need
194 to lock them, since they are only unset at the end of a scheduler
198 /* flag set by signal handler to precipitate a context switch */
199 //@cindex context_switch
202 /* if this flag is set as well, give up execution */
203 //@cindex interrupted
206 /* Next thread ID to allocate.
207 * Locks required: sched_mutex
209 //@cindex next_thread_id
210 StgThreadID next_thread_id = 1;
213 * Pointers to the state of the current thread.
214 * Rule of thumb: if CurrentTSO != NULL, then we're running a Haskell
215 * thread. If CurrentTSO == NULL, then we're at the scheduler level.
218 /* The smallest stack size that makes any sense is:
219 * RESERVED_STACK_WORDS (so we can get back from the stack overflow)
220 * + sizeofW(StgStopFrame) (the stg_stop_thread_info frame)
221 * + 1 (the realworld token for an IO thread)
222 * + 1 (the closure to enter)
224 * A thread with this stack will bomb immediately with a stack
225 * overflow, which will increase its stack size.
228 #define MIN_STACK_WORDS (RESERVED_STACK_WORDS + sizeofW(StgStopFrame) + 2)
232 Capability MainCapability; /* for non-SMP, we have one global capability */
239 /* This is used in `TSO.h' and gcc 2.96 insists that this variable actually
240 * exists - earlier gccs apparently didn't.
247 /* All our current task ids, saved in case we need to kill them later.
254 void addToBlockedQueue ( StgTSO *tso );
256 static void schedule ( void );
257 void interruptStgRts ( void );
259 static StgTSO * createThread_ ( nat size, rtsBool have_lock, StgInt pri );
261 static StgTSO * createThread_ ( nat size, rtsBool have_lock );
264 static void detectBlackHoles ( void );
267 static void sched_belch(char *s, ...);
270 #if defined(RTS_SUPPORTS_THREADS)
271 /* ToDo: carefully document the invariants that go together
272 * with these synchronisation objects.
274 MutexVar sched_mutex = INIT_MUTEX_VAR;
275 MutexVar term_mutex = INIT_MUTEX_VAR;
276 CondVar thread_ready_cond = INIT_COND_VAR;
277 CondVar gc_pending_cond = INIT_COND_VAR;
284 rtsTime TimeOfLastYield;
285 rtsBool emitSchedule = rtsTrue;
289 char *whatNext_strs[] = {
297 char *threadReturnCode_strs[] = {
298 "HeapOverflow", /* might also be StackOverflow */
307 StgTSO * createSparkThread(rtsSpark spark);
308 StgTSO * activateSpark (rtsSpark spark);
312 * The thread state for the main thread.
313 // ToDo: check whether not needed any more
317 //@node Main scheduling loop, Suspend and Resume, Prototypes, Main scheduling code
318 //@subsection Main scheduling loop
320 /* ---------------------------------------------------------------------------
321 Main scheduling loop.
323 We use round-robin scheduling, each thread returning to the
324 scheduler loop when one of these conditions is detected:
327 * timer expires (thread yields)
332 Locking notes: we acquire the scheduler lock once at the beginning
333 of the scheduler loop, and release it when
335 * running a thread, or
336 * waiting for work, or
337 * waiting for a GC to complete.
340 In a GranSim setup this loop iterates over the global event queue.
341 This revolves around the global event queue, which determines what
342 to do next. Therefore, it's more complicated than either the
343 concurrent or the parallel (GUM) setup.
346 GUM iterates over incoming messages.
347 It starts with nothing to do (thus CurrentTSO == END_TSO_QUEUE),
348 and sends out a fish whenever it has nothing to do; in-between
349 doing the actual reductions (shared code below) it processes the
350 incoming messages and deals with delayed operations
351 (see PendingFetches).
352 This is not the ugliest code you could imagine, but it's bloody close.
354 ------------------------------------------------------------------------ */
361 StgThreadReturnCode ret;
369 rtsBool receivedFinish = rtsFalse;
371 nat tp_size, sp_size; // stats only
374 rtsBool was_interrupted = rtsFalse;
376 ACQUIRE_LOCK(&sched_mutex);
380 /* set up first event to get things going */
381 /* ToDo: assign costs for system setup and init MainTSO ! */
382 new_event(CurrentProc, CurrentProc, CurrentTime[CurrentProc],
384 CurrentTSO, (StgClosure*)NULL, (rtsSpark*)NULL);
387 fprintf(stderr, "GRAN: Init CurrentTSO (in schedule) = %p\n", CurrentTSO);
388 G_TSO(CurrentTSO, 5));
390 if (RtsFlags.GranFlags.Light) {
391 /* Save current time; GranSim Light only */
392 CurrentTSO->gran.clock = CurrentTime[CurrentProc];
395 event = get_next_event();
397 while (event!=(rtsEvent*)NULL) {
398 /* Choose the processor with the next event */
399 CurrentProc = event->proc;
400 CurrentTSO = event->tso;
404 while (!receivedFinish) { /* set by processMessages */
405 /* when receiving PP_FINISH message */
412 IF_DEBUG(scheduler, printAllThreads());
414 /* If we're interrupted (the user pressed ^C, or some other
415 * termination condition occurred), kill all the currently running
419 IF_DEBUG(scheduler, sched_belch("interrupted"));
421 interrupted = rtsFalse;
422 was_interrupted = rtsTrue;
425 /* Go through the list of main threads and wake up any
426 * clients whose computations have finished. ToDo: this
427 * should be done more efficiently without a linear scan
428 * of the main threads list, somehow...
430 #if defined(RTS_SUPPORTS_THREADS)
432 StgMainThread *m, **prev;
433 prev = &main_threads;
434 for (m = main_threads; m != NULL; m = m->link) {
435 switch (m->tso->what_next) {
438 *(m->ret) = (StgClosure *)m->tso->sp[0];
442 broadcastCondVar(&m->wakeup);
445 if (m->ret) *(m->ret) = NULL;
447 if (was_interrupted) {
448 m->stat = Interrupted;
452 broadcastCondVar(&m->wakeup);
460 #else /* not threaded */
463 /* in GUM do this only on the Main PE */
466 /* If our main thread has finished or been killed, return.
469 StgMainThread *m = main_threads;
470 if (m->tso->what_next == ThreadComplete
471 || m->tso->what_next == ThreadKilled) {
472 main_threads = main_threads->link;
473 if (m->tso->what_next == ThreadComplete) {
474 /* we finished successfully, fill in the return value */
475 if (m->ret) { *(m->ret) = (StgClosure *)m->tso->sp[0]; };
479 if (m->ret) { *(m->ret) = NULL; };
480 if (was_interrupted) {
481 m->stat = Interrupted;
491 /* Top up the run queue from our spark pool. We try to make the
492 * number of threads in the run queue equal to the number of
495 * Disable spark support in SMP for now, non-essential & requires
496 * a little bit of work to make it compile cleanly. -- sof 1/02.
498 #if 0 /* defined(SMP) */
500 nat n = getFreeCapabilities();
501 StgTSO *tso = run_queue_hd;
503 /* Count the run queue */
504 while (n > 0 && tso != END_TSO_QUEUE) {
511 spark = findSpark(rtsFalse);
513 break; /* no more sparks in the pool */
515 /* I'd prefer this to be done in activateSpark -- HWL */
516 /* tricky - it needs to hold the scheduler lock and
517 * not try to re-acquire it -- SDM */
518 createSparkThread(spark);
520 sched_belch("==^^ turning spark of closure %p into a thread",
521 (StgClosure *)spark));
524 /* We need to wake up the other tasks if we just created some
527 if (getFreeCapabilities() - n > 1) {
528 signalCondVar ( &thread_ready_cond );
533 /* check for signals each time around the scheduler */
534 #ifndef mingw32_TARGET_OS
535 if (signals_pending()) {
536 startSignalHandlers();
540 /* Check whether any waiting threads need to be woken up. If the
541 * run queue is empty, and there are no other tasks running, we
542 * can wait indefinitely for something to happen.
543 * ToDo: what if another client comes along & requests another
546 if (blocked_queue_hd != END_TSO_QUEUE || sleeping_queue != END_TSO_QUEUE) {
548 (run_queue_hd == END_TSO_QUEUE)
550 && allFreeCapabilities()
554 /* we can be interrupted while waiting for I/O... */
555 if (interrupted) continue;
558 * Detect deadlock: when we have no threads to run, there are no
559 * threads waiting on I/O or sleeping, and all the other tasks are
560 * waiting for work, we must have a deadlock of some description.
562 * We first try to find threads blocked on themselves (ie. black
563 * holes), and generate NonTermination exceptions where necessary.
565 * If no threads are black holed, we have a deadlock situation, so
566 * inform all the main threads.
569 if (blocked_queue_hd == END_TSO_QUEUE
570 && run_queue_hd == END_TSO_QUEUE
571 && sleeping_queue == END_TSO_QUEUE
573 && allFreeCapabilities()
577 IF_DEBUG(scheduler, sched_belch("deadlocked, forcing major GC..."));
578 GarbageCollect(GetRoots,rtsTrue);
579 if (blocked_queue_hd == END_TSO_QUEUE
580 && run_queue_hd == END_TSO_QUEUE
581 && sleeping_queue == END_TSO_QUEUE) {
583 IF_DEBUG(scheduler, sched_belch("still deadlocked, checking for black holes..."));
586 /* No black holes, so probably a real deadlock. Send the
587 * current main thread the Deadlock exception (or in the SMP
588 * build, send *all* main threads the deadlock exception,
589 * since none of them can make progress).
591 if (run_queue_hd == END_TSO_QUEUE) {
593 #if defined(RTS_SUPPORTS_THREADS)
594 for (m = main_threads; m != NULL; m = m->link) {
595 switch (m->tso->why_blocked) {
596 case BlockedOnBlackHole:
597 raiseAsync(m->tso, (StgClosure *)NonTermination_closure);
599 case BlockedOnException:
601 raiseAsync(m->tso, (StgClosure *)Deadlock_closure);
604 barf("deadlock: main thread blocked in a strange way");
609 switch (m->tso->why_blocked) {
610 case BlockedOnBlackHole:
611 raiseAsync(m->tso, (StgClosure *)NonTermination_closure);
613 case BlockedOnException:
615 raiseAsync(m->tso, (StgClosure *)Deadlock_closure);
618 barf("deadlock: main thread blocked in a strange way");
622 #if !defined(RTS_SUPPORTS_THREADS)
623 ASSERT( run_queue_hd != END_TSO_QUEUE );
628 /* ToDo: add deadlock detection in GUM (similar to SMP) -- HWL */
632 /* If there's a GC pending, don't do anything until it has
636 IF_DEBUG(scheduler,sched_belch("waiting for GC"));
637 waitCondVar ( &gc_pending_cond, &sched_mutex );
641 #if defined(RTS_SUPPORTS_THREADS)
642 /* block until we've got a thread on the run queue and a free
645 while ( run_queue_hd == END_TSO_QUEUE
647 || noFreeCapabilities()
650 IF_DEBUG(scheduler, sched_belch("waiting for work"));
651 waitCondVar ( &thread_ready_cond, &sched_mutex );
652 IF_DEBUG(scheduler, sched_belch("work now available"));
658 if (RtsFlags.GranFlags.Light)
659 GranSimLight_enter_system(event, &ActiveTSO); // adjust ActiveTSO etc
661 /* adjust time based on time-stamp */
662 if (event->time > CurrentTime[CurrentProc] &&
663 event->evttype != ContinueThread)
664 CurrentTime[CurrentProc] = event->time;
666 /* Deal with the idle PEs (may issue FindWork or MoveSpark events) */
667 if (!RtsFlags.GranFlags.Light)
670 IF_DEBUG(gran, fprintf(stderr, "GRAN: switch by event-type\n"));
672 /* main event dispatcher in GranSim */
673 switch (event->evttype) {
674 /* Should just be continuing execution */
676 IF_DEBUG(gran, fprintf(stderr, "GRAN: doing ContinueThread\n"));
677 /* ToDo: check assertion
678 ASSERT(run_queue_hd != (StgTSO*)NULL &&
679 run_queue_hd != END_TSO_QUEUE);
681 /* Ignore ContinueThreads for fetching threads (if synchr comm) */
682 if (!RtsFlags.GranFlags.DoAsyncFetch &&
683 procStatus[CurrentProc]==Fetching) {
684 belch("ghuH: Spurious ContinueThread while Fetching ignored; TSO %d (%p) [PE %d]",
685 CurrentTSO->id, CurrentTSO, CurrentProc);
688 /* Ignore ContinueThreads for completed threads */
689 if (CurrentTSO->what_next == ThreadComplete) {
690 belch("ghuH: found a ContinueThread event for completed thread %d (%p) [PE %d] (ignoring ContinueThread)",
691 CurrentTSO->id, CurrentTSO, CurrentProc);
694 /* Ignore ContinueThreads for threads that are being migrated */
695 if (PROCS(CurrentTSO)==Nowhere) {
696 belch("ghuH: trying to run the migrating TSO %d (%p) [PE %d] (ignoring ContinueThread)",
697 CurrentTSO->id, CurrentTSO, CurrentProc);
700 /* The thread should be at the beginning of the run queue */
701 if (CurrentTSO!=run_queue_hds[CurrentProc]) {
702 belch("ghuH: TSO %d (%p) [PE %d] is not at the start of the run_queue when doing a ContinueThread",
703 CurrentTSO->id, CurrentTSO, CurrentProc);
704 break; // run the thread anyway
707 new_event(proc, proc, CurrentTime[proc],
709 (StgTSO*)NULL, (StgClosure*)NULL, (rtsSpark*)NULL);
711 */ /* Catches superfluous CONTINUEs -- should be unnecessary */
712 break; // now actually run the thread; DaH Qu'vam yImuHbej
715 do_the_fetchnode(event);
716 goto next_thread; /* handle next event in event queue */
719 do_the_globalblock(event);
720 goto next_thread; /* handle next event in event queue */
723 do_the_fetchreply(event);
724 goto next_thread; /* handle next event in event queue */
726 case UnblockThread: /* Move from the blocked queue to the tail of */
727 do_the_unblock(event);
728 goto next_thread; /* handle next event in event queue */
730 case ResumeThread: /* Move from the blocked queue to the tail of */
731 /* the runnable queue ( i.e. Qu' SImqa'lu') */
732 event->tso->gran.blocktime +=
733 CurrentTime[CurrentProc] - event->tso->gran.blockedat;
734 do_the_startthread(event);
735 goto next_thread; /* handle next event in event queue */
738 do_the_startthread(event);
739 goto next_thread; /* handle next event in event queue */
742 do_the_movethread(event);
743 goto next_thread; /* handle next event in event queue */
746 do_the_movespark(event);
747 goto next_thread; /* handle next event in event queue */
750 do_the_findwork(event);
751 goto next_thread; /* handle next event in event queue */
754 barf("Illegal event type %u\n", event->evttype);
757 /* This point was scheduler_loop in the old RTS */
759 IF_DEBUG(gran, belch("GRAN: after main switch"));
761 TimeOfLastEvent = CurrentTime[CurrentProc];
762 TimeOfNextEvent = get_time_of_next_event();
763 IgnoreEvents=(TimeOfNextEvent==0); // HWL HACK
764 // CurrentTSO = ThreadQueueHd;
766 IF_DEBUG(gran, belch("GRAN: time of next event is: %ld",
769 if (RtsFlags.GranFlags.Light)
770 GranSimLight_leave_system(event, &ActiveTSO);
772 EndOfTimeSlice = CurrentTime[CurrentProc]+RtsFlags.GranFlags.time_slice;
775 belch("GRAN: end of time-slice is %#lx", EndOfTimeSlice));
777 /* in a GranSim setup the TSO stays on the run queue */
779 /* Take a thread from the run queue. */
780 t = POP_RUN_QUEUE(); // take_off_run_queue(t);
783 fprintf(stderr, "GRAN: About to run current thread, which is\n");
786 context_switch = 0; // turned on via GranYield, checking events and time slice
789 DumpGranEvent(GR_SCHEDULE, t));
791 procStatus[CurrentProc] = Busy;
794 if (PendingFetches != END_BF_QUEUE) {
798 /* ToDo: phps merge with spark activation above */
799 /* check whether we have local work and send requests if we have none */
800 if (EMPTY_RUN_QUEUE()) { /* no runnable threads */
801 /* :-[ no local threads => look out for local sparks */
802 /* the spark pool for the current PE */
803 pool = &(MainRegTable.rSparks); // generalise to cap = &MainRegTable
804 if (advisory_thread_count < RtsFlags.ParFlags.maxThreads &&
805 pool->hd < pool->tl) {
807 * ToDo: add GC code check that we really have enough heap afterwards!!
809 * If we're here (no runnable threads) and we have pending
810 * sparks, we must have a space problem. Get enough space
811 * to turn one of those pending sparks into a
815 spark = findSpark(rtsFalse); /* get a spark */
816 if (spark != (rtsSpark) NULL) {
817 tso = activateSpark(spark); /* turn the spark into a thread */
818 IF_PAR_DEBUG(schedule,
819 belch("==== schedule: Created TSO %d (%p); %d threads active",
820 tso->id, tso, advisory_thread_count));
822 if (tso==END_TSO_QUEUE) { /* failed to activate spark->back to loop */
823 belch("==^^ failed to activate spark");
825 } /* otherwise fall through & pick-up new tso */
827 IF_PAR_DEBUG(verbose,
828 belch("==^^ no local sparks (spark pool contains only NFs: %d)",
829 spark_queue_len(pool)));
834 /* If we still have no work we need to send a FISH to get a spark
837 if (EMPTY_RUN_QUEUE()) {
838 /* =8-[ no local sparks => look for work on other PEs */
840 * We really have absolutely no work. Send out a fish
841 * (there may be some out there already), and wait for
842 * something to arrive. We clearly can't run any threads
843 * until a SCHEDULE or RESUME arrives, and so that's what
844 * we're hoping to see. (Of course, we still have to
845 * respond to other types of messages.)
847 TIME now = msTime() /*CURRENT_TIME*/;
848 IF_PAR_DEBUG(verbose,
849 belch("-- now=%ld", now));
850 IF_PAR_DEBUG(verbose,
851 if (outstandingFishes < RtsFlags.ParFlags.maxFishes &&
852 (last_fish_arrived_at!=0 &&
853 last_fish_arrived_at+RtsFlags.ParFlags.fishDelay > now)) {
854 belch("--$$ delaying FISH until %ld (last fish %ld, delay %ld, now %ld)",
855 last_fish_arrived_at+RtsFlags.ParFlags.fishDelay,
856 last_fish_arrived_at,
857 RtsFlags.ParFlags.fishDelay, now);
860 if (outstandingFishes < RtsFlags.ParFlags.maxFishes &&
861 (last_fish_arrived_at==0 ||
862 (last_fish_arrived_at+RtsFlags.ParFlags.fishDelay <= now))) {
863 /* outstandingFishes is set in sendFish, processFish;
864 avoid flooding system with fishes via delay */
866 sendFish(pe, mytid, NEW_FISH_AGE, NEW_FISH_HISTORY,
869 // Global statistics: count no. of fishes
870 if (RtsFlags.ParFlags.ParStats.Global &&
871 RtsFlags.GcFlags.giveStats > NO_GC_STATS) {
872 globalParStats.tot_fish_mess++;
876 receivedFinish = processMessages();
879 } else if (PacketsWaiting()) { /* Look for incoming messages */
880 receivedFinish = processMessages();
883 /* Now we are sure that we have some work available */
884 ASSERT(run_queue_hd != END_TSO_QUEUE);
886 /* Take a thread from the run queue, if we have work */
887 t = POP_RUN_QUEUE(); // take_off_run_queue(END_TSO_QUEUE);
888 IF_DEBUG(sanity,checkTSO(t));
890 /* ToDo: write something to the log-file
891 if (RTSflags.ParFlags.granSimStats && !sameThread)
892 DumpGranEvent(GR_SCHEDULE, RunnableThreadsHd);
896 /* the spark pool for the current PE */
897 pool = &(MainRegTable.rSparks); // generalise to cap = &MainRegTable
900 belch("--=^ %d threads, %d sparks on [%#x]",
901 run_queue_len(), spark_queue_len(pool), CURRENT_PROC));
904 if (0 && RtsFlags.ParFlags.ParStats.Full &&
905 t && LastTSO && t->id != LastTSO->id &&
906 LastTSO->why_blocked == NotBlocked &&
907 LastTSO->what_next != ThreadComplete) {
908 // if previously scheduled TSO not blocked we have to record the context switch
909 DumpVeryRawGranEvent(TimeOfLastYield, CURRENT_PROC, CURRENT_PROC,
910 GR_DESCHEDULE, LastTSO, (StgClosure *)NULL, 0, 0);
913 if (RtsFlags.ParFlags.ParStats.Full &&
914 (emitSchedule /* forced emit */ ||
915 (t && LastTSO && t->id != LastTSO->id))) {
917 we are running a different TSO, so write a schedule event to log file
918 NB: If we use fair scheduling we also have to write a deschedule
919 event for LastTSO; with unfair scheduling we know that the
920 previous tso has blocked whenever we switch to another tso, so
921 we don't need it in GUM for now
923 DumpRawGranEvent(CURRENT_PROC, CURRENT_PROC,
924 GR_SCHEDULE, t, (StgClosure *)NULL, 0, 0);
925 emitSchedule = rtsFalse;
929 #else /* !GRAN && !PAR */
931 /* grab a thread from the run queue
933 ASSERT(run_queue_hd != END_TSO_QUEUE);
936 // Sanity check the thread we're about to run. This can be
937 // expensive if there is lots of thread switching going on...
938 IF_DEBUG(sanity,checkTSO(t));
943 grabCapability(&cap);
945 cap = &MainCapability;
948 cap->r.rCurrentTSO = t;
950 /* context switches are now initiated by the timer signal, unless
951 * the user specified "context switch as often as possible", with
956 RtsFlags.ProfFlags.profileInterval == 0 ||
958 (RtsFlags.ConcFlags.ctxtSwitchTicks == 0
959 && (run_queue_hd != END_TSO_QUEUE
960 || blocked_queue_hd != END_TSO_QUEUE
961 || sleeping_queue != END_TSO_QUEUE)))
966 RELEASE_LOCK(&sched_mutex);
968 IF_DEBUG(scheduler, sched_belch("-->> Running TSO %ld (%p) %s ...",
969 t->id, t, whatNext_strs[t->what_next]));
972 startHeapProfTimer();
975 /* +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ */
976 /* Run the current thread
978 switch (cap->r.rCurrentTSO->what_next) {
981 /* Thread already finished, return to scheduler. */
982 ret = ThreadFinished;
985 ret = StgRun((StgFunPtr) stg_enterStackTop, &cap->r);
988 ret = StgRun((StgFunPtr) stg_returnToStackTop, &cap->r);
990 case ThreadEnterInterp:
991 ret = interpretBCO(cap);
994 barf("schedule: invalid what_next field");
996 /* +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ */
998 /* Costs for the scheduler are assigned to CCS_SYSTEM */
1000 stopHeapProfTimer();
1004 ACQUIRE_LOCK(&sched_mutex);
1007 IF_DEBUG(scheduler,fprintf(stderr,"scheduler (task %ld): ", osThreadId()););
1008 #elif !defined(GRAN) && !defined(PAR)
1009 IF_DEBUG(scheduler,fprintf(stderr,"scheduler: "););
1011 t = cap->r.rCurrentTSO;
1014 /* HACK 675: if the last thread didn't yield, make sure to print a
1015 SCHEDULE event to the log file when StgRunning the next thread, even
1016 if it is the same one as before */
1018 TimeOfLastYield = CURRENT_TIME;
1024 IF_DEBUG(gran, DumpGranEvent(GR_DESCHEDULE, t));
1025 globalGranStats.tot_heapover++;
1027 globalParStats.tot_heapover++;
1030 // did the task ask for a large block?
1031 if (cap->r.rHpAlloc > BLOCK_SIZE_W) {
1032 // if so, get one and push it on the front of the nursery.
1036 blocks = (nat)BLOCK_ROUND_UP(cap->r.rHpAlloc * sizeof(W_)) / BLOCK_SIZE;
1038 IF_DEBUG(scheduler,belch("--<< thread %ld (%p; %s) stopped: requesting a large block (size %d)",
1040 whatNext_strs[t->what_next], blocks));
1042 // don't do this if it would push us over the
1043 // alloc_blocks_lim limit; we'll GC first.
1044 if (alloc_blocks + blocks < alloc_blocks_lim) {
1046 alloc_blocks += blocks;
1047 bd = allocGroup( blocks );
1049 // link the new group into the list
1050 bd->link = cap->r.rCurrentNursery;
1051 bd->u.back = cap->r.rCurrentNursery->u.back;
1052 if (cap->r.rCurrentNursery->u.back != NULL) {
1053 cap->r.rCurrentNursery->u.back->link = bd;
1055 ASSERT(g0s0->blocks == cap->r.rCurrentNursery &&
1056 g0s0->blocks == cap->r.rNursery);
1057 cap->r.rNursery = g0s0->blocks = bd;
1059 cap->r.rCurrentNursery->u.back = bd;
1061 // initialise it as a nursery block
1065 bd->free = bd->start;
1067 // don't forget to update the block count in g0s0.
1068 g0s0->n_blocks += blocks;
1069 ASSERT(countBlocks(g0s0->blocks) == g0s0->n_blocks);
1071 // now update the nursery to point to the new block
1072 cap->r.rCurrentNursery = bd;
1074 // we might be unlucky and have another thread get on the
1075 // run queue before us and steal the large block, but in that
1076 // case the thread will just end up requesting another large
1078 PUSH_ON_RUN_QUEUE(t);
1083 /* make all the running tasks block on a condition variable,
1084 * maybe set context_switch and wait till they all pile in,
1085 * then have them wait on a GC condition variable.
1087 IF_DEBUG(scheduler,belch("--<< thread %ld (%p; %s) stopped: HeapOverflow",
1088 t->id, t, whatNext_strs[t->what_next]));
1091 ASSERT(!is_on_queue(t,CurrentProc));
1093 /* Currently we emit a DESCHEDULE event before GC in GUM.
1094 ToDo: either add separate event to distinguish SYSTEM time from rest
1095 or just nuke this DESCHEDULE (and the following SCHEDULE) */
1096 if (0 && RtsFlags.ParFlags.ParStats.Full) {
1097 DumpRawGranEvent(CURRENT_PROC, CURRENT_PROC,
1098 GR_DESCHEDULE, t, (StgClosure *)NULL, 0, 0);
1099 emitSchedule = rtsTrue;
1103 ready_to_gc = rtsTrue;
1104 context_switch = 1; /* stop other threads ASAP */
1105 PUSH_ON_RUN_QUEUE(t);
1106 /* actual GC is done at the end of the while loop */
1112 DumpGranEvent(GR_DESCHEDULE, t));
1113 globalGranStats.tot_stackover++;
1116 // DumpGranEvent(GR_DESCHEDULE, t);
1117 globalParStats.tot_stackover++;
1119 IF_DEBUG(scheduler,belch("--<< thread %ld (%p; %s) stopped, StackOverflow",
1120 t->id, t, whatNext_strs[t->what_next]));
1121 /* just adjust the stack for this thread, then pop it back
1127 /* enlarge the stack */
1128 StgTSO *new_t = threadStackOverflow(t);
1130 /* This TSO has moved, so update any pointers to it from the
1131 * main thread stack. It better not be on any other queues...
1132 * (it shouldn't be).
1134 for (m = main_threads; m != NULL; m = m->link) {
1139 threadPaused(new_t);
1140 PUSH_ON_RUN_QUEUE(new_t);
1144 case ThreadYielding:
1147 DumpGranEvent(GR_DESCHEDULE, t));
1148 globalGranStats.tot_yields++;
1151 // DumpGranEvent(GR_DESCHEDULE, t);
1152 globalParStats.tot_yields++;
1154 /* put the thread back on the run queue. Then, if we're ready to
1155 * GC, check whether this is the last task to stop. If so, wake
1156 * up the GC thread. getThread will block during a GC until the
1160 if (t->what_next == ThreadEnterInterp) {
1161 /* ToDo: or maybe a timer expired when we were in Hugs?
1162 * or maybe someone hit ctrl-C
1164 belch("--<< thread %ld (%p; %s) stopped to switch to Hugs",
1165 t->id, t, whatNext_strs[t->what_next]);
1167 belch("--<< thread %ld (%p; %s) stopped, yielding",
1168 t->id, t, whatNext_strs[t->what_next]);
1175 //belch("&& Doing sanity check on yielding TSO %ld.", t->id);
1177 ASSERT(t->link == END_TSO_QUEUE);
1179 ASSERT(!is_on_queue(t,CurrentProc));
1182 //belch("&& Doing sanity check on all ThreadQueues (and their TSOs).");
1183 checkThreadQsSanity(rtsTrue));
1186 if (RtsFlags.ParFlags.doFairScheduling) {
1187 /* this does round-robin scheduling; good for concurrency */
1188 APPEND_TO_RUN_QUEUE(t);
1190 /* this does unfair scheduling; good for parallelism */
1191 PUSH_ON_RUN_QUEUE(t);
1194 /* this does round-robin scheduling; good for concurrency */
1195 APPEND_TO_RUN_QUEUE(t);
1198 /* add a ContinueThread event to actually process the thread */
1199 new_event(CurrentProc, CurrentProc, CurrentTime[CurrentProc],
1201 t, (StgClosure*)NULL, (rtsSpark*)NULL);
1203 belch("GRAN: eventq and runnableq after adding yielded thread to queue again:");
1212 belch("--<< thread %ld (%p; %s) stopped, blocking on node %p [PE %d] with BQ: ",
1213 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)));
1214 if (t->block_info.closure!=(StgClosure*)NULL) print_bq(t->block_info.closure));
1216 // ??? needed; should emit block before
1218 DumpGranEvent(GR_DESCHEDULE, t));
1219 prune_eventq(t, (StgClosure *)NULL); // prune ContinueThreads for t
1222 ASSERT(procStatus[CurrentProc]==Busy ||
1223 ((procStatus[CurrentProc]==Fetching) &&
1224 (t->block_info.closure!=(StgClosure*)NULL)));
1225 if (run_queue_hds[CurrentProc] == END_TSO_QUEUE &&
1226 !(!RtsFlags.GranFlags.DoAsyncFetch &&
1227 procStatus[CurrentProc]==Fetching))
1228 procStatus[CurrentProc] = Idle;
1232 belch("--<< thread %ld (%p; %s) stopped, blocking on node %p with BQ: ",
1233 t->id, t, whatNext_strs[t->what_next], t->block_info.closure));
1236 if (t->block_info.closure!=(StgClosure*)NULL)
1237 print_bq(t->block_info.closure));
1239 /* Send a fetch (if BlockedOnGA) and dump event to log file */
1242 /* whatever we schedule next, we must log that schedule */
1243 emitSchedule = rtsTrue;
1246 /* don't need to do anything. Either the thread is blocked on
1247 * I/O, in which case we'll have called addToBlockedQueue
1248 * previously, or it's blocked on an MVar or Blackhole, in which
1249 * case it'll be on the relevant queue already.
1252 fprintf(stderr, "--<< thread %d (%p) stopped: ", t->id, t);
1253 printThreadBlockage(t);
1254 fprintf(stderr, "\n"));
1256 /* Only for dumping event to log file
1257 ToDo: do I need this in GranSim, too?
1264 case ThreadFinished:
1265 /* Need to check whether this was a main thread, and if so, signal
1266 * the task that started it with the return value. If we have no
1267 * more main threads, we probably need to stop all the tasks until
1270 /* We also end up here if the thread kills itself with an
1271 * uncaught exception, see Exception.hc.
1273 IF_DEBUG(scheduler,belch("--++ thread %d (%p) finished", t->id, t));
1275 endThread(t, CurrentProc); // clean-up the thread
1277 /* For now all are advisory -- HWL */
1278 //if(t->priority==AdvisoryPriority) ??
1279 advisory_thread_count--;
1282 if(t->dist.priority==RevalPriority)
1286 if (RtsFlags.ParFlags.ParStats.Full &&
1287 !RtsFlags.ParFlags.ParStats.Suppressed)
1288 DumpEndEvent(CURRENT_PROC, t, rtsFalse /* not mandatory */);
1293 barf("schedule: invalid thread return code %d", (int)ret);
1297 grabCapability(&cap);
1301 if (RtsFlags.ProfFlags.profileInterval==0 || performHeapProfile) {
1302 GarbageCollect(GetRoots, rtsTrue);
1304 performHeapProfile = rtsFalse;
1305 ready_to_gc = rtsFalse; // we already GC'd
1310 if (ready_to_gc && allFreeCapabilities() )
1315 /* everybody back, start the GC.
1316 * Could do it in this thread, or signal a condition var
1317 * to do it in another thread. Either way, we need to
1318 * broadcast on gc_pending_cond afterward.
1320 #if defined(RTS_SUPPORTS_THREADS)
1321 IF_DEBUG(scheduler,sched_belch("doing GC"));
1323 GarbageCollect(GetRoots,rtsFalse);
1324 ready_to_gc = rtsFalse;
1326 broadcastCondVar(&gc_pending_cond);
1329 /* add a ContinueThread event to continue execution of current thread */
1330 new_event(CurrentProc, CurrentProc, CurrentTime[CurrentProc],
1332 t, (StgClosure*)NULL, (rtsSpark*)NULL);
1334 fprintf(stderr, "GRAN: eventq and runnableq after Garbage collection:\n");
1342 IF_GRAN_DEBUG(unused,
1343 print_eventq(EventHd));
1345 event = get_next_event();
1348 /* ToDo: wait for next message to arrive rather than busy wait */
1351 } /* end of while(1) */
1353 IF_PAR_DEBUG(verbose,
1354 belch("== Leaving schedule() after having received Finish"));
1357 /* ---------------------------------------------------------------------------
1358 * deleteAllThreads(): kill all the live threads.
1360 * This is used when we catch a user interrupt (^C), before performing
1361 * any necessary cleanups and running finalizers.
1362 * ------------------------------------------------------------------------- */
1364 void deleteAllThreads ( void )
1367 IF_DEBUG(scheduler,sched_belch("deleting all threads"));
1368 for (t = run_queue_hd; t != END_TSO_QUEUE; t = t->link) {
1371 for (t = blocked_queue_hd; t != END_TSO_QUEUE; t = t->link) {
1374 for (t = sleeping_queue; t != END_TSO_QUEUE; t = t->link) {
1377 run_queue_hd = run_queue_tl = END_TSO_QUEUE;
1378 blocked_queue_hd = blocked_queue_tl = END_TSO_QUEUE;
1379 sleeping_queue = END_TSO_QUEUE;
1382 /* startThread and insertThread are now in GranSim.c -- HWL */
1384 //@node Suspend and Resume, Run queue code, Main scheduling loop, Main scheduling code
1385 //@subsection Suspend and Resume
1387 /* ---------------------------------------------------------------------------
1388 * Suspending & resuming Haskell threads.
1390 * When making a "safe" call to C (aka _ccall_GC), the task gives back
1391 * its capability before calling the C function. This allows another
1392 * task to pick up the capability and carry on running Haskell
1393 * threads. It also means that if the C call blocks, it won't lock
1396 * The Haskell thread making the C call is put to sleep for the
1397 * duration of the call, on the susepended_ccalling_threads queue. We
1398 * give out a token to the task, which it can use to resume the thread
1399 * on return from the C function.
1400 * ------------------------------------------------------------------------- */
1403 suspendThread( StgRegTable *reg )
1408 // assume that *reg is a pointer to the StgRegTable part of a Capability
1409 cap = (Capability *)((void *)reg - sizeof(StgFunTable));
1411 ACQUIRE_LOCK(&sched_mutex);
1414 sched_belch("thread %d did a _ccall_gc", cap->r.rCurrentTSO->id));
1416 threadPaused(cap->r.rCurrentTSO);
1417 cap->r.rCurrentTSO->link = suspended_ccalling_threads;
1418 suspended_ccalling_threads = cap->r.rCurrentTSO;
1420 /* Use the thread ID as the token; it should be unique */
1421 tok = cap->r.rCurrentTSO->id;
1424 /* Hand back capability */
1425 releaseCapability(&cap);
1428 RELEASE_LOCK(&sched_mutex);
1433 resumeThread( StgInt tok )
1435 StgTSO *tso, **prev;
1438 ACQUIRE_LOCK(&sched_mutex);
1440 prev = &suspended_ccalling_threads;
1441 for (tso = suspended_ccalling_threads;
1442 tso != END_TSO_QUEUE;
1443 prev = &tso->link, tso = tso->link) {
1444 if (tso->id == (StgThreadID)tok) {
1449 if (tso == END_TSO_QUEUE) {
1450 barf("resumeThread: thread not found");
1452 tso->link = END_TSO_QUEUE;
1455 while ( noFreeCapabilities() ) {
1456 IF_DEBUG(scheduler, sched_belch("waiting to resume"));
1457 waitCondVar(&thread_ready_cond, &sched_mutex);
1458 IF_DEBUG(scheduler, sched_belch("resuming thread %d", tso->id));
1460 grabCapability(&cap);
1462 cap = &MainCapability;
1465 cap->r.rCurrentTSO = tso;
1467 RELEASE_LOCK(&sched_mutex);
1472 /* ---------------------------------------------------------------------------
1474 * ------------------------------------------------------------------------ */
1475 static void unblockThread(StgTSO *tso);
1477 /* ---------------------------------------------------------------------------
1478 * Comparing Thread ids.
1480 * This is used from STG land in the implementation of the
1481 * instances of Eq/Ord for ThreadIds.
1482 * ------------------------------------------------------------------------ */
1484 int cmp_thread(const StgTSO *tso1, const StgTSO *tso2)
1486 StgThreadID id1 = tso1->id;
1487 StgThreadID id2 = tso2->id;
1489 if (id1 < id2) return (-1);
1490 if (id1 > id2) return 1;
1494 /* ---------------------------------------------------------------------------
1495 * Fetching the ThreadID from an StgTSO.
1497 * This is used in the implementation of Show for ThreadIds.
1498 * ------------------------------------------------------------------------ */
1499 int rts_getThreadId(const StgTSO *tso)
1504 /* ---------------------------------------------------------------------------
1505 Create a new thread.
1507 The new thread starts with the given stack size. Before the
1508 scheduler can run, however, this thread needs to have a closure
1509 (and possibly some arguments) pushed on its stack. See
1510 pushClosure() in Schedule.h.
1512 createGenThread() and createIOThread() (in SchedAPI.h) are
1513 convenient packaged versions of this function.
1515 currently pri (priority) is only used in a GRAN setup -- HWL
1516 ------------------------------------------------------------------------ */
1517 //@cindex createThread
1519 /* currently pri (priority) is only used in a GRAN setup -- HWL */
1521 createThread(nat stack_size, StgInt pri)
1523 return createThread_(stack_size, rtsFalse, pri);
1527 createThread_(nat size, rtsBool have_lock, StgInt pri)
1531 createThread(nat stack_size)
1533 return createThread_(stack_size, rtsFalse);
1537 createThread_(nat size, rtsBool have_lock)
1544 /* First check whether we should create a thread at all */
1546 /* check that no more than RtsFlags.ParFlags.maxThreads threads are created */
1547 if (advisory_thread_count >= RtsFlags.ParFlags.maxThreads) {
1549 belch("{createThread}Daq ghuH: refusing to create another thread; no more than %d threads allowed (currently %d)",
1550 RtsFlags.ParFlags.maxThreads, advisory_thread_count);
1551 return END_TSO_QUEUE;
1557 ASSERT(!RtsFlags.GranFlags.Light || CurrentProc==0);
1560 // ToDo: check whether size = stack_size - TSO_STRUCT_SIZEW
1562 /* catch ridiculously small stack sizes */
1563 if (size < MIN_STACK_WORDS + TSO_STRUCT_SIZEW) {
1564 size = MIN_STACK_WORDS + TSO_STRUCT_SIZEW;
1567 stack_size = size - TSO_STRUCT_SIZEW;
1569 tso = (StgTSO *)allocate(size);
1570 TICK_ALLOC_TSO(stack_size, 0);
1572 SET_HDR(tso, &stg_TSO_info, CCS_SYSTEM);
1574 SET_GRAN_HDR(tso, ThisPE);
1576 tso->what_next = ThreadEnterGHC;
1578 /* tso->id needs to be unique. For now we use a heavyweight mutex to
1579 * protect the increment operation on next_thread_id.
1580 * In future, we could use an atomic increment instead.
1582 if (!have_lock) { ACQUIRE_LOCK(&sched_mutex); }
1583 tso->id = next_thread_id++;
1584 if (!have_lock) { RELEASE_LOCK(&sched_mutex); }
1586 tso->why_blocked = NotBlocked;
1587 tso->blocked_exceptions = NULL;
1589 tso->stack_size = stack_size;
1590 tso->max_stack_size = round_to_mblocks(RtsFlags.GcFlags.maxStkSize)
1592 tso->sp = (P_)&(tso->stack) + stack_size;
1595 tso->prof.CCCS = CCS_MAIN;
1598 /* put a stop frame on the stack */
1599 tso->sp -= sizeofW(StgStopFrame);
1600 SET_HDR((StgClosure*)tso->sp,(StgInfoTable *)&stg_stop_thread_info,CCS_SYSTEM);
1601 tso->su = (StgUpdateFrame*)tso->sp;
1605 tso->link = END_TSO_QUEUE;
1606 /* uses more flexible routine in GranSim */
1607 insertThread(tso, CurrentProc);
1609 /* In a non-GranSim setup the pushing of a TSO onto the runq is separated
1615 if (RtsFlags.GranFlags.GranSimStats.Full)
1616 DumpGranEvent(GR_START,tso);
1618 if (RtsFlags.ParFlags.ParStats.Full)
1619 DumpGranEvent(GR_STARTQ,tso);
1620 /* HACk to avoid SCHEDULE
1624 /* Link the new thread on the global thread list.
1626 tso->global_link = all_threads;
1630 tso->dist.priority = MandatoryPriority; //by default that is...
1634 tso->gran.pri = pri;
1636 tso->gran.magic = TSO_MAGIC; // debugging only
1638 tso->gran.sparkname = 0;
1639 tso->gran.startedat = CURRENT_TIME;
1640 tso->gran.exported = 0;
1641 tso->gran.basicblocks = 0;
1642 tso->gran.allocs = 0;
1643 tso->gran.exectime = 0;
1644 tso->gran.fetchtime = 0;
1645 tso->gran.fetchcount = 0;
1646 tso->gran.blocktime = 0;
1647 tso->gran.blockcount = 0;
1648 tso->gran.blockedat = 0;
1649 tso->gran.globalsparks = 0;
1650 tso->gran.localsparks = 0;
1651 if (RtsFlags.GranFlags.Light)
1652 tso->gran.clock = Now; /* local clock */
1654 tso->gran.clock = 0;
1656 IF_DEBUG(gran,printTSO(tso));
1659 tso->par.magic = TSO_MAGIC; // debugging only
1661 tso->par.sparkname = 0;
1662 tso->par.startedat = CURRENT_TIME;
1663 tso->par.exported = 0;
1664 tso->par.basicblocks = 0;
1665 tso->par.allocs = 0;
1666 tso->par.exectime = 0;
1667 tso->par.fetchtime = 0;
1668 tso->par.fetchcount = 0;
1669 tso->par.blocktime = 0;
1670 tso->par.blockcount = 0;
1671 tso->par.blockedat = 0;
1672 tso->par.globalsparks = 0;
1673 tso->par.localsparks = 0;
1677 globalGranStats.tot_threads_created++;
1678 globalGranStats.threads_created_on_PE[CurrentProc]++;
1679 globalGranStats.tot_sq_len += spark_queue_len(CurrentProc);
1680 globalGranStats.tot_sq_probes++;
1682 // collect parallel global statistics (currently done together with GC stats)
1683 if (RtsFlags.ParFlags.ParStats.Global &&
1684 RtsFlags.GcFlags.giveStats > NO_GC_STATS) {
1685 //fprintf(stderr, "Creating thread %d @ %11.2f\n", tso->id, usertime());
1686 globalParStats.tot_threads_created++;
1692 belch("==__ schedule: Created TSO %d (%p);",
1693 CurrentProc, tso, tso->id));
1695 IF_PAR_DEBUG(verbose,
1696 belch("==__ schedule: Created TSO %d (%p); %d threads active",
1697 tso->id, tso, advisory_thread_count));
1699 IF_DEBUG(scheduler,sched_belch("created thread %ld, stack size = %lx words",
1700 tso->id, tso->stack_size));
1707 all parallel thread creation calls should fall through the following routine.
1710 createSparkThread(rtsSpark spark)
1712 ASSERT(spark != (rtsSpark)NULL);
1713 if (advisory_thread_count >= RtsFlags.ParFlags.maxThreads)
1715 barf("{createSparkThread}Daq ghuH: refusing to create another thread; no more than %d threads allowed (currently %d)",
1716 RtsFlags.ParFlags.maxThreads, advisory_thread_count);
1717 return END_TSO_QUEUE;
1721 tso = createThread_(RtsFlags.GcFlags.initialStkSize, rtsTrue);
1722 if (tso==END_TSO_QUEUE)
1723 barf("createSparkThread: Cannot create TSO");
1725 tso->priority = AdvisoryPriority;
1727 pushClosure(tso,spark);
1728 PUSH_ON_RUN_QUEUE(tso);
1729 advisory_thread_count++;
1736 Turn a spark into a thread.
1737 ToDo: fix for SMP (needs to acquire SCHED_MUTEX!)
1740 //@cindex activateSpark
1742 activateSpark (rtsSpark spark)
1746 tso = createSparkThread(spark);
1747 if (RtsFlags.ParFlags.ParStats.Full) {
1748 //ASSERT(run_queue_hd == END_TSO_QUEUE); // I think ...
1749 IF_PAR_DEBUG(verbose,
1750 belch("==^^ activateSpark: turning spark of closure %p (%s) into a thread",
1751 (StgClosure *)spark, info_type((StgClosure *)spark)));
1753 // ToDo: fwd info on local/global spark to thread -- HWL
1754 // tso->gran.exported = spark->exported;
1755 // tso->gran.locked = !spark->global;
1756 // tso->gran.sparkname = spark->name;
1762 /* ---------------------------------------------------------------------------
1765 * scheduleThread puts a thread on the head of the runnable queue.
1766 * This will usually be done immediately after a thread is created.
1767 * The caller of scheduleThread must create the thread using e.g.
1768 * createThread and push an appropriate closure
1769 * on this thread's stack before the scheduler is invoked.
1770 * ------------------------------------------------------------------------ */
1773 scheduleThread(StgTSO *tso)
1775 ACQUIRE_LOCK(&sched_mutex);
1777 /* Put the new thread on the head of the runnable queue. The caller
1778 * better push an appropriate closure on this thread's stack
1779 * beforehand. In the SMP case, the thread may start running as
1780 * soon as we release the scheduler lock below.
1782 PUSH_ON_RUN_QUEUE(tso);
1786 IF_DEBUG(scheduler,printTSO(tso));
1788 RELEASE_LOCK(&sched_mutex);
1791 /* ---------------------------------------------------------------------------
1794 * Start up Posix threads to run each of the scheduler tasks.
1795 * I believe the task ids are not needed in the system as defined.
1797 * ------------------------------------------------------------------------ */
1799 #if defined(PAR) || defined(SMP)
1801 taskStart(void) /* ( void *arg STG_UNUSED) */
1807 /* ---------------------------------------------------------------------------
1810 * Initialise the scheduler. This resets all the queues - if the
1811 * queues contained any threads, they'll be garbage collected at the
1814 * This now calls startTasks(), so should only be called once! KH @ 25/10/99
1815 * ------------------------------------------------------------------------ */
1819 term_handler(int sig STG_UNUSED)
1822 ACQUIRE_LOCK(&term_mutex);
1824 RELEASE_LOCK(&term_mutex);
1835 for (i=0; i<=MAX_PROC; i++) {
1836 run_queue_hds[i] = END_TSO_QUEUE;
1837 run_queue_tls[i] = END_TSO_QUEUE;
1838 blocked_queue_hds[i] = END_TSO_QUEUE;
1839 blocked_queue_tls[i] = END_TSO_QUEUE;
1840 ccalling_threadss[i] = END_TSO_QUEUE;
1841 sleeping_queue = END_TSO_QUEUE;
1844 run_queue_hd = END_TSO_QUEUE;
1845 run_queue_tl = END_TSO_QUEUE;
1846 blocked_queue_hd = END_TSO_QUEUE;
1847 blocked_queue_tl = END_TSO_QUEUE;
1848 sleeping_queue = END_TSO_QUEUE;
1851 suspended_ccalling_threads = END_TSO_QUEUE;
1853 main_threads = NULL;
1854 all_threads = END_TSO_QUEUE;
1859 RtsFlags.ConcFlags.ctxtSwitchTicks =
1860 RtsFlags.ConcFlags.ctxtSwitchTime / TICK_MILLISECS;
1862 /* Install the SIGHUP handler */
1865 struct sigaction action,oact;
1867 action.sa_handler = term_handler;
1868 sigemptyset(&action.sa_mask);
1869 action.sa_flags = 0;
1870 if (sigaction(SIGTERM, &action, &oact) != 0) {
1871 barf("can't install TERM handler");
1877 /* Allocate N Capabilities */
1878 initCapabilities(RtsFlags.ParFlags.nNodes);
1880 initCapability(&MainCapability);
1883 #if /* defined(SMP) ||*/ defined(PAR)
1896 /* make some space for saving all the thread ids */
1897 task_ids = stgMallocBytes(RtsFlags.ParFlags.nNodes * sizeof(task_info),
1898 "initScheduler:task_ids");
1900 /* and create all the threads */
1901 for (i = 0; i < RtsFlags.ParFlags.nNodes; i++) {
1902 r = createOSThread(&tid,taskStart);
1904 barf("startTasks: Can't create new Posix thread");
1906 task_ids[i].id = tid;
1907 task_ids[i].mut_time = 0.0;
1908 task_ids[i].mut_etime = 0.0;
1909 task_ids[i].gc_time = 0.0;
1910 task_ids[i].gc_etime = 0.0;
1911 task_ids[i].elapsedtimestart = stat_getElapsedTime();
1912 IF_DEBUG(scheduler,fprintf(stderr,"scheduler: Started task: %ld\n",tid););
1918 exitScheduler( void )
1923 /* Don't want to use pthread_cancel, since we'd have to install
1924 * these silly exception handlers (pthread_cleanup_{push,pop}) around
1928 /* Cancel all our tasks */
1929 for (i = 0; i < RtsFlags.ParFlags.nNodes; i++) {
1930 pthread_cancel(task_ids[i].id);
1933 /* Wait for all the tasks to terminate */
1934 for (i = 0; i < RtsFlags.ParFlags.nNodes; i++) {
1935 IF_DEBUG(scheduler,fprintf(stderr,"scheduler: waiting for task %ld\n",
1937 pthread_join(task_ids[i].id, NULL);
1941 /* Send 'em all a SIGHUP. That should shut 'em up.
1943 await_death = RtsFlags.ParFlags.nNodes;
1944 for (i = 0; i < RtsFlags.ParFlags.nNodes; i++) {
1945 pthread_kill(task_ids[i].id,SIGTERM);
1947 while (await_death > 0) {
1953 /* -----------------------------------------------------------------------------
1954 Managing the per-task allocation areas.
1956 Each capability comes with an allocation area. These are
1957 fixed-length block lists into which allocation can be done.
1959 ToDo: no support for two-space collection at the moment???
1960 -------------------------------------------------------------------------- */
1962 /* -----------------------------------------------------------------------------
1963 * waitThread is the external interface for running a new computation
1964 * and waiting for the result.
1966 * In the non-SMP case, we create a new main thread, push it on the
1967 * main-thread stack, and invoke the scheduler to run it. The
1968 * scheduler will return when the top main thread on the stack has
1969 * completed or died, and fill in the necessary fields of the
1970 * main_thread structure.
1972 * In the SMP case, we create a main thread as before, but we then
1973 * create a new condition variable and sleep on it. When our new
1974 * main thread has completed, we'll be woken up and the status/result
1975 * will be in the main_thread struct.
1976 * -------------------------------------------------------------------------- */
1979 howManyThreadsAvail ( void )
1983 for (q = run_queue_hd; q != END_TSO_QUEUE; q = q->link)
1985 for (q = blocked_queue_hd; q != END_TSO_QUEUE; q = q->link)
1987 for (q = sleeping_queue; q != END_TSO_QUEUE; q = q->link)
1993 finishAllThreads ( void )
1996 while (run_queue_hd != END_TSO_QUEUE) {
1997 waitThread ( run_queue_hd, NULL );
1999 while (blocked_queue_hd != END_TSO_QUEUE) {
2000 waitThread ( blocked_queue_hd, NULL );
2002 while (sleeping_queue != END_TSO_QUEUE) {
2003 waitThread ( blocked_queue_hd, NULL );
2006 (blocked_queue_hd != END_TSO_QUEUE ||
2007 run_queue_hd != END_TSO_QUEUE ||
2008 sleeping_queue != END_TSO_QUEUE);
2012 waitThread(StgTSO *tso, /*out*/StgClosure **ret)
2015 SchedulerStatus stat;
2017 ACQUIRE_LOCK(&sched_mutex);
2019 m = stgMallocBytes(sizeof(StgMainThread), "waitThread");
2024 #if defined(RTS_SUPPORTS_THREADS)
2025 initCondVar(&m->wakeup);
2028 m->link = main_threads;
2031 IF_DEBUG(scheduler, fprintf(stderr, "== scheduler: new main thread (%d)\n",
2036 waitCondVar(&m->wakeup, &sched_mutex);
2037 } while (m->stat == NoStatus);
2039 /* GranSim specific init */
2040 CurrentTSO = m->tso; // the TSO to run
2041 procStatus[MainProc] = Busy; // status of main PE
2042 CurrentProc = MainProc; // PE to run it on
2047 ASSERT(m->stat != NoStatus);
2052 #if defined(RTS_SUPPORTS_THREADS)
2053 closeCondVar(&m->wakeup);
2056 IF_DEBUG(scheduler, fprintf(stderr, "== scheduler: main thread (%d) finished\n",
2060 RELEASE_LOCK(&sched_mutex);
2065 //@node Run queue code, Garbage Collextion Routines, Suspend and Resume, Main scheduling code
2066 //@subsection Run queue code
2070 NB: In GranSim we have many run queues; run_queue_hd is actually a macro
2071 unfolding to run_queue_hds[CurrentProc], thus CurrentProc is an
2072 implicit global variable that has to be correct when calling these
2076 /* Put the new thread on the head of the runnable queue.
2077 * The caller of createThread better push an appropriate closure
2078 * on this thread's stack before the scheduler is invoked.
2080 static /* inline */ void
2081 add_to_run_queue(tso)
2084 ASSERT(tso!=run_queue_hd && tso!=run_queue_tl);
2085 tso->link = run_queue_hd;
2087 if (run_queue_tl == END_TSO_QUEUE) {
2092 /* Put the new thread at the end of the runnable queue. */
2093 static /* inline */ void
2094 push_on_run_queue(tso)
2097 ASSERT(get_itbl((StgClosure *)tso)->type == TSO);
2098 ASSERT(run_queue_hd!=NULL && run_queue_tl!=NULL);
2099 ASSERT(tso!=run_queue_hd && tso!=run_queue_tl);
2100 if (run_queue_hd == END_TSO_QUEUE) {
2103 run_queue_tl->link = tso;
2109 Should be inlined because it's used very often in schedule. The tso
2110 argument is actually only needed in GranSim, where we want to have the
2111 possibility to schedule *any* TSO on the run queue, irrespective of the
2112 actual ordering. Therefore, if tso is not the nil TSO then we traverse
2113 the run queue and dequeue the tso, adjusting the links in the queue.
2115 //@cindex take_off_run_queue
2116 static /* inline */ StgTSO*
2117 take_off_run_queue(StgTSO *tso) {
2121 qetlaHbogh Qu' ngaSbogh ghomDaQ {tso} yIteq!
2123 if tso is specified, unlink that tso from the run_queue (doesn't have
2124 to be at the beginning of the queue); GranSim only
2126 if (tso!=END_TSO_QUEUE) {
2127 /* find tso in queue */
2128 for (t=run_queue_hd, prev=END_TSO_QUEUE;
2129 t!=END_TSO_QUEUE && t!=tso;
2133 /* now actually dequeue the tso */
2134 if (prev!=END_TSO_QUEUE) {
2135 ASSERT(run_queue_hd!=t);
2136 prev->link = t->link;
2138 /* t is at beginning of thread queue */
2139 ASSERT(run_queue_hd==t);
2140 run_queue_hd = t->link;
2142 /* t is at end of thread queue */
2143 if (t->link==END_TSO_QUEUE) {
2144 ASSERT(t==run_queue_tl);
2145 run_queue_tl = prev;
2147 ASSERT(run_queue_tl!=t);
2149 t->link = END_TSO_QUEUE;
2151 /* take tso from the beginning of the queue; std concurrent code */
2153 if (t != END_TSO_QUEUE) {
2154 run_queue_hd = t->link;
2155 t->link = END_TSO_QUEUE;
2156 if (run_queue_hd == END_TSO_QUEUE) {
2157 run_queue_tl = END_TSO_QUEUE;
2166 //@node Garbage Collextion Routines, Blocking Queue Routines, Run queue code, Main scheduling code
2167 //@subsection Garbage Collextion Routines
2169 /* ---------------------------------------------------------------------------
2170 Where are the roots that we know about?
2172 - all the threads on the runnable queue
2173 - all the threads on the blocked queue
2174 - all the threads on the sleeping queue
2175 - all the thread currently executing a _ccall_GC
2176 - all the "main threads"
2178 ------------------------------------------------------------------------ */
2180 /* This has to be protected either by the scheduler monitor, or by the
2181 garbage collection monitor (probably the latter).
2186 GetRoots(evac_fn evac)
2193 for (i=0; i<=RtsFlags.GranFlags.proc; i++) {
2194 if ((run_queue_hds[i] != END_TSO_QUEUE) && ((run_queue_hds[i] != NULL)))
2195 evac((StgClosure **)&run_queue_hds[i]);
2196 if ((run_queue_tls[i] != END_TSO_QUEUE) && ((run_queue_tls[i] != NULL)))
2197 evac((StgClosure **)&run_queue_tls[i]);
2199 if ((blocked_queue_hds[i] != END_TSO_QUEUE) && ((blocked_queue_hds[i] != NULL)))
2200 evac((StgClosure **)&blocked_queue_hds[i]);
2201 if ((blocked_queue_tls[i] != END_TSO_QUEUE) && ((blocked_queue_tls[i] != NULL)))
2202 evac((StgClosure **)&blocked_queue_tls[i]);
2203 if ((ccalling_threadss[i] != END_TSO_QUEUE) && ((ccalling_threadss[i] != NULL)))
2204 evac((StgClosure **)&ccalling_threads[i]);
2211 if (run_queue_hd != END_TSO_QUEUE) {
2212 ASSERT(run_queue_tl != END_TSO_QUEUE);
2213 evac((StgClosure **)&run_queue_hd);
2214 evac((StgClosure **)&run_queue_tl);
2217 if (blocked_queue_hd != END_TSO_QUEUE) {
2218 ASSERT(blocked_queue_tl != END_TSO_QUEUE);
2219 evac((StgClosure **)&blocked_queue_hd);
2220 evac((StgClosure **)&blocked_queue_tl);
2223 if (sleeping_queue != END_TSO_QUEUE) {
2224 evac((StgClosure **)&sleeping_queue);
2228 for (m = main_threads; m != NULL; m = m->link) {
2229 evac((StgClosure **)&m->tso);
2231 if (suspended_ccalling_threads != END_TSO_QUEUE) {
2232 evac((StgClosure **)&suspended_ccalling_threads);
2235 #if defined(PAR) || defined(GRAN)
2236 markSparkQueue(evac);
2240 /* -----------------------------------------------------------------------------
2243 This is the interface to the garbage collector from Haskell land.
2244 We provide this so that external C code can allocate and garbage
2245 collect when called from Haskell via _ccall_GC.
2247 It might be useful to provide an interface whereby the programmer
2248 can specify more roots (ToDo).
2250 This needs to be protected by the GC condition variable above. KH.
2251 -------------------------------------------------------------------------- */
2253 void (*extra_roots)(evac_fn);
2258 GarbageCollect(GetRoots,rtsFalse);
2262 performMajorGC(void)
2264 GarbageCollect(GetRoots,rtsTrue);
2268 AllRoots(evac_fn evac)
2270 GetRoots(evac); // the scheduler's roots
2271 extra_roots(evac); // the user's roots
2275 performGCWithRoots(void (*get_roots)(evac_fn))
2277 extra_roots = get_roots;
2278 GarbageCollect(AllRoots,rtsFalse);
2281 /* -----------------------------------------------------------------------------
2284 If the thread has reached its maximum stack size, then raise the
2285 StackOverflow exception in the offending thread. Otherwise
2286 relocate the TSO into a larger chunk of memory and adjust its stack
2288 -------------------------------------------------------------------------- */
2291 threadStackOverflow(StgTSO *tso)
2293 nat new_stack_size, new_tso_size, diff, stack_words;
2297 IF_DEBUG(sanity,checkTSO(tso));
2298 if (tso->stack_size >= tso->max_stack_size) {
2301 belch("@@ threadStackOverflow of TSO %d (%p): stack too large (now %ld; max is %ld",
2302 tso->id, tso, tso->stack_size, tso->max_stack_size);
2303 /* If we're debugging, just print out the top of the stack */
2304 printStackChunk(tso->sp, stg_min(tso->stack+tso->stack_size,
2307 /* Send this thread the StackOverflow exception */
2308 raiseAsync(tso, (StgClosure *)stackOverflow_closure);
2312 /* Try to double the current stack size. If that takes us over the
2313 * maximum stack size for this thread, then use the maximum instead.
2314 * Finally round up so the TSO ends up as a whole number of blocks.
2316 new_stack_size = stg_min(tso->stack_size * 2, tso->max_stack_size);
2317 new_tso_size = (nat)BLOCK_ROUND_UP(new_stack_size * sizeof(W_) +
2318 TSO_STRUCT_SIZE)/sizeof(W_);
2319 new_tso_size = round_to_mblocks(new_tso_size); /* Be MBLOCK-friendly */
2320 new_stack_size = new_tso_size - TSO_STRUCT_SIZEW;
2322 IF_DEBUG(scheduler, fprintf(stderr,"== scheduler: increasing stack size from %d words to %d.\n", tso->stack_size, new_stack_size));
2324 dest = (StgTSO *)allocate(new_tso_size);
2325 TICK_ALLOC_TSO(new_stack_size,0);
2327 /* copy the TSO block and the old stack into the new area */
2328 memcpy(dest,tso,TSO_STRUCT_SIZE);
2329 stack_words = tso->stack + tso->stack_size - tso->sp;
2330 new_sp = (P_)dest + new_tso_size - stack_words;
2331 memcpy(new_sp, tso->sp, stack_words * sizeof(W_));
2333 /* relocate the stack pointers... */
2334 diff = (P_)new_sp - (P_)tso->sp; /* In *words* */
2335 dest->su = (StgUpdateFrame *) ((P_)dest->su + diff);
2337 dest->stack_size = new_stack_size;
2339 /* and relocate the update frame list */
2340 relocate_stack(dest, diff);
2342 /* Mark the old TSO as relocated. We have to check for relocated
2343 * TSOs in the garbage collector and any primops that deal with TSOs.
2345 * It's important to set the sp and su values to just beyond the end
2346 * of the stack, so we don't attempt to scavenge any part of the
2349 tso->what_next = ThreadRelocated;
2351 tso->sp = (P_)&(tso->stack[tso->stack_size]);
2352 tso->su = (StgUpdateFrame *)tso->sp;
2353 tso->why_blocked = NotBlocked;
2354 dest->mut_link = NULL;
2356 IF_PAR_DEBUG(verbose,
2357 belch("@@ threadStackOverflow of TSO %d (now at %p): stack size increased to %ld",
2358 tso->id, tso, tso->stack_size);
2359 /* If we're debugging, just print out the top of the stack */
2360 printStackChunk(tso->sp, stg_min(tso->stack+tso->stack_size,
2363 IF_DEBUG(sanity,checkTSO(tso));
2365 IF_DEBUG(scheduler,printTSO(dest));
2371 //@node Blocking Queue Routines, Exception Handling Routines, Garbage Collextion Routines, Main scheduling code
2372 //@subsection Blocking Queue Routines
2374 /* ---------------------------------------------------------------------------
2375 Wake up a queue that was blocked on some resource.
2376 ------------------------------------------------------------------------ */
2380 unblockCount ( StgBlockingQueueElement *bqe, StgClosure *node )
2385 unblockCount ( StgBlockingQueueElement *bqe, StgClosure *node )
2387 /* write RESUME events to log file and
2388 update blocked and fetch time (depending on type of the orig closure) */
2389 if (RtsFlags.ParFlags.ParStats.Full) {
2390 DumpRawGranEvent(CURRENT_PROC, CURRENT_PROC,
2391 GR_RESUMEQ, ((StgTSO *)bqe), ((StgTSO *)bqe)->block_info.closure,
2392 0, 0 /* spark_queue_len(ADVISORY_POOL) */);
2393 if (EMPTY_RUN_QUEUE())
2394 emitSchedule = rtsTrue;
2396 switch (get_itbl(node)->type) {
2398 ((StgTSO *)bqe)->par.fetchtime += CURRENT_TIME-((StgTSO *)bqe)->par.blockedat;
2403 ((StgTSO *)bqe)->par.blocktime += CURRENT_TIME-((StgTSO *)bqe)->par.blockedat;
2410 barf("{unblockOneLocked}Daq Qagh: unexpected closure in blocking queue");
2417 static StgBlockingQueueElement *
2418 unblockOneLocked(StgBlockingQueueElement *bqe, StgClosure *node)
2421 PEs node_loc, tso_loc;
2423 node_loc = where_is(node); // should be lifted out of loop
2424 tso = (StgTSO *)bqe; // wastes an assignment to get the type right
2425 tso_loc = where_is((StgClosure *)tso);
2426 if (IS_LOCAL_TO(PROCS(node),tso_loc)) { // TSO is local
2427 /* !fake_fetch => TSO is on CurrentProc is same as IS_LOCAL_TO */
2428 ASSERT(CurrentProc!=node_loc || tso_loc==CurrentProc);
2429 CurrentTime[CurrentProc] += RtsFlags.GranFlags.Costs.lunblocktime;
2430 // insertThread(tso, node_loc);
2431 new_event(tso_loc, tso_loc, CurrentTime[CurrentProc],
2433 tso, node, (rtsSpark*)NULL);
2434 tso->link = END_TSO_QUEUE; // overwrite link just to be sure
2437 } else { // TSO is remote (actually should be FMBQ)
2438 CurrentTime[CurrentProc] += RtsFlags.GranFlags.Costs.mpacktime +
2439 RtsFlags.GranFlags.Costs.gunblocktime +
2440 RtsFlags.GranFlags.Costs.latency;
2441 new_event(tso_loc, CurrentProc, CurrentTime[CurrentProc],
2443 tso, node, (rtsSpark*)NULL);
2444 tso->link = END_TSO_QUEUE; // overwrite link just to be sure
2447 /* the thread-queue-overhead is accounted for in either Resume or UnblockThread */
2449 fprintf(stderr," %s TSO %d (%p) [PE %d] (block_info.closure=%p) (next=%p) ,",
2450 (node_loc==tso_loc ? "Local" : "Global"),
2451 tso->id, tso, CurrentProc, tso->block_info.closure, tso->link));
2452 tso->block_info.closure = NULL;
2453 IF_DEBUG(scheduler,belch("-- Waking up thread %ld (%p)",
2457 static StgBlockingQueueElement *
2458 unblockOneLocked(StgBlockingQueueElement *bqe, StgClosure *node)
2460 StgBlockingQueueElement *next;
2462 switch (get_itbl(bqe)->type) {
2464 ASSERT(((StgTSO *)bqe)->why_blocked != NotBlocked);
2465 /* if it's a TSO just push it onto the run_queue */
2467 // ((StgTSO *)bqe)->link = END_TSO_QUEUE; // debugging?
2468 PUSH_ON_RUN_QUEUE((StgTSO *)bqe);
2470 unblockCount(bqe, node);
2471 /* reset blocking status after dumping event */
2472 ((StgTSO *)bqe)->why_blocked = NotBlocked;
2476 /* if it's a BLOCKED_FETCH put it on the PendingFetches list */
2478 bqe->link = (StgBlockingQueueElement *)PendingFetches;
2479 PendingFetches = (StgBlockedFetch *)bqe;
2483 /* can ignore this case in a non-debugging setup;
2484 see comments on RBHSave closures above */
2486 /* check that the closure is an RBHSave closure */
2487 ASSERT(get_itbl((StgClosure *)bqe) == &stg_RBH_Save_0_info ||
2488 get_itbl((StgClosure *)bqe) == &stg_RBH_Save_1_info ||
2489 get_itbl((StgClosure *)bqe) == &stg_RBH_Save_2_info);
2493 barf("{unblockOneLocked}Daq Qagh: Unexpected IP (%#lx; %s) in blocking queue at %#lx\n",
2494 get_itbl((StgClosure *)bqe), info_type((StgClosure *)bqe),
2498 IF_PAR_DEBUG(bq, fprintf(stderr, ", %p (%s)", bqe, info_type((StgClosure*)bqe)));
2502 #else /* !GRAN && !PAR */
2504 unblockOneLocked(StgTSO *tso)
2508 ASSERT(get_itbl(tso)->type == TSO);
2509 ASSERT(tso->why_blocked != NotBlocked);
2510 tso->why_blocked = NotBlocked;
2512 PUSH_ON_RUN_QUEUE(tso);
2514 IF_DEBUG(scheduler,sched_belch("waking up thread %ld", tso->id));
2519 #if defined(GRAN) || defined(PAR)
2520 inline StgBlockingQueueElement *
2521 unblockOne(StgBlockingQueueElement *bqe, StgClosure *node)
2523 ACQUIRE_LOCK(&sched_mutex);
2524 bqe = unblockOneLocked(bqe, node);
2525 RELEASE_LOCK(&sched_mutex);
2530 unblockOne(StgTSO *tso)
2532 ACQUIRE_LOCK(&sched_mutex);
2533 tso = unblockOneLocked(tso);
2534 RELEASE_LOCK(&sched_mutex);
2541 awakenBlockedQueue(StgBlockingQueueElement *q, StgClosure *node)
2543 StgBlockingQueueElement *bqe;
2548 belch("##-_ AwBQ for node %p on PE %d @ %ld by TSO %d (%p): ", \
2549 node, CurrentProc, CurrentTime[CurrentProc],
2550 CurrentTSO->id, CurrentTSO));
2552 node_loc = where_is(node);
2554 ASSERT(q == END_BQ_QUEUE ||
2555 get_itbl(q)->type == TSO || // q is either a TSO or an RBHSave
2556 get_itbl(q)->type == CONSTR); // closure (type constructor)
2557 ASSERT(is_unique(node));
2559 /* FAKE FETCH: magically copy the node to the tso's proc;
2560 no Fetch necessary because in reality the node should not have been
2561 moved to the other PE in the first place
2563 if (CurrentProc!=node_loc) {
2565 belch("## node %p is on PE %d but CurrentProc is %d (TSO %d); assuming fake fetch and adjusting bitmask (old: %#x)",
2566 node, node_loc, CurrentProc, CurrentTSO->id,
2567 // CurrentTSO, where_is(CurrentTSO),
2568 node->header.gran.procs));
2569 node->header.gran.procs = (node->header.gran.procs) | PE_NUMBER(CurrentProc);
2571 belch("## new bitmask of node %p is %#x",
2572 node, node->header.gran.procs));
2573 if (RtsFlags.GranFlags.GranSimStats.Global) {
2574 globalGranStats.tot_fake_fetches++;
2579 // ToDo: check: ASSERT(CurrentProc==node_loc);
2580 while (get_itbl(bqe)->type==TSO) { // q != END_TSO_QUEUE) {
2583 bqe points to the current element in the queue
2584 next points to the next element in the queue
2586 //tso = (StgTSO *)bqe; // wastes an assignment to get the type right
2587 //tso_loc = where_is(tso);
2589 bqe = unblockOneLocked(bqe, node);
2592 /* if this is the BQ of an RBH, we have to put back the info ripped out of
2593 the closure to make room for the anchor of the BQ */
2594 if (bqe!=END_BQ_QUEUE) {
2595 ASSERT(get_itbl(node)->type == RBH && get_itbl(bqe)->type == CONSTR);
2597 ASSERT((info_ptr==&RBH_Save_0_info) ||
2598 (info_ptr==&RBH_Save_1_info) ||
2599 (info_ptr==&RBH_Save_2_info));
2601 /* cf. convertToRBH in RBH.c for writing the RBHSave closure */
2602 ((StgRBH *)node)->blocking_queue = (StgBlockingQueueElement *)((StgRBHSave *)bqe)->payload[0];
2603 ((StgRBH *)node)->mut_link = (StgMutClosure *)((StgRBHSave *)bqe)->payload[1];
2606 belch("## Filled in RBH_Save for %p (%s) at end of AwBQ",
2607 node, info_type(node)));
2610 /* statistics gathering */
2611 if (RtsFlags.GranFlags.GranSimStats.Global) {
2612 // globalGranStats.tot_bq_processing_time += bq_processing_time;
2613 globalGranStats.tot_bq_len += len; // total length of all bqs awakened
2614 // globalGranStats.tot_bq_len_local += len_local; // same for local TSOs only
2615 globalGranStats.tot_awbq++; // total no. of bqs awakened
2618 fprintf(stderr,"## BQ Stats of %p: [%d entries] %s\n",
2619 node, len, (bqe!=END_BQ_QUEUE) ? "RBH" : ""));
2623 awakenBlockedQueue(StgBlockingQueueElement *q, StgClosure *node)
2625 StgBlockingQueueElement *bqe;
2627 ACQUIRE_LOCK(&sched_mutex);
2629 IF_PAR_DEBUG(verbose,
2630 belch("##-_ AwBQ for node %p on [%x]: ",
2634 if(get_itbl(q)->type == CONSTR || q==END_BQ_QUEUE) {
2635 IF_PAR_DEBUG(verbose, belch("## ... nothing to unblock so lets just return. RFP (BUG?)"));
2640 ASSERT(q == END_BQ_QUEUE ||
2641 get_itbl(q)->type == TSO ||
2642 get_itbl(q)->type == BLOCKED_FETCH ||
2643 get_itbl(q)->type == CONSTR);
2646 while (get_itbl(bqe)->type==TSO ||
2647 get_itbl(bqe)->type==BLOCKED_FETCH) {
2648 bqe = unblockOneLocked(bqe, node);
2650 RELEASE_LOCK(&sched_mutex);
2653 #else /* !GRAN && !PAR */
2655 awakenBlockedQueue(StgTSO *tso)
2657 ACQUIRE_LOCK(&sched_mutex);
2658 while (tso != END_TSO_QUEUE) {
2659 tso = unblockOneLocked(tso);
2661 RELEASE_LOCK(&sched_mutex);
2665 //@node Exception Handling Routines, Debugging Routines, Blocking Queue Routines, Main scheduling code
2666 //@subsection Exception Handling Routines
2668 /* ---------------------------------------------------------------------------
2670 - usually called inside a signal handler so it mustn't do anything fancy.
2671 ------------------------------------------------------------------------ */
2674 interruptStgRts(void)
2680 /* -----------------------------------------------------------------------------
2683 This is for use when we raise an exception in another thread, which
2685 This has nothing to do with the UnblockThread event in GranSim. -- HWL
2686 -------------------------------------------------------------------------- */
2688 #if defined(GRAN) || defined(PAR)
2690 NB: only the type of the blocking queue is different in GranSim and GUM
2691 the operations on the queue-elements are the same
2692 long live polymorphism!
2695 unblockThread(StgTSO *tso)
2697 StgBlockingQueueElement *t, **last;
2699 ACQUIRE_LOCK(&sched_mutex);
2700 switch (tso->why_blocked) {
2703 return; /* not blocked */
2706 ASSERT(get_itbl(tso->block_info.closure)->type == MVAR);
2708 StgBlockingQueueElement *last_tso = END_BQ_QUEUE;
2709 StgMVar *mvar = (StgMVar *)(tso->block_info.closure);
2711 last = (StgBlockingQueueElement **)&mvar->head;
2712 for (t = (StgBlockingQueueElement *)mvar->head;
2714 last = &t->link, last_tso = t, t = t->link) {
2715 if (t == (StgBlockingQueueElement *)tso) {
2716 *last = (StgBlockingQueueElement *)tso->link;
2717 if (mvar->tail == tso) {
2718 mvar->tail = (StgTSO *)last_tso;
2723 barf("unblockThread (MVAR): TSO not found");
2726 case BlockedOnBlackHole:
2727 ASSERT(get_itbl(tso->block_info.closure)->type == BLACKHOLE_BQ);
2729 StgBlockingQueue *bq = (StgBlockingQueue *)(tso->block_info.closure);
2731 last = &bq->blocking_queue;
2732 for (t = bq->blocking_queue;
2734 last = &t->link, t = t->link) {
2735 if (t == (StgBlockingQueueElement *)tso) {
2736 *last = (StgBlockingQueueElement *)tso->link;
2740 barf("unblockThread (BLACKHOLE): TSO not found");
2743 case BlockedOnException:
2745 StgTSO *target = tso->block_info.tso;
2747 ASSERT(get_itbl(target)->type == TSO);
2749 if (target->what_next == ThreadRelocated) {
2750 target = target->link;
2751 ASSERT(get_itbl(target)->type == TSO);
2754 ASSERT(target->blocked_exceptions != NULL);
2756 last = (StgBlockingQueueElement **)&target->blocked_exceptions;
2757 for (t = (StgBlockingQueueElement *)target->blocked_exceptions;
2759 last = &t->link, t = t->link) {
2760 ASSERT(get_itbl(t)->type == TSO);
2761 if (t == (StgBlockingQueueElement *)tso) {
2762 *last = (StgBlockingQueueElement *)tso->link;
2766 barf("unblockThread (Exception): TSO not found");
2770 case BlockedOnWrite:
2772 /* take TSO off blocked_queue */
2773 StgBlockingQueueElement *prev = NULL;
2774 for (t = (StgBlockingQueueElement *)blocked_queue_hd; t != END_BQ_QUEUE;
2775 prev = t, t = t->link) {
2776 if (t == (StgBlockingQueueElement *)tso) {
2778 blocked_queue_hd = (StgTSO *)t->link;
2779 if ((StgBlockingQueueElement *)blocked_queue_tl == t) {
2780 blocked_queue_tl = END_TSO_QUEUE;
2783 prev->link = t->link;
2784 if ((StgBlockingQueueElement *)blocked_queue_tl == t) {
2785 blocked_queue_tl = (StgTSO *)prev;
2791 barf("unblockThread (I/O): TSO not found");
2794 case BlockedOnDelay:
2796 /* take TSO off sleeping_queue */
2797 StgBlockingQueueElement *prev = NULL;
2798 for (t = (StgBlockingQueueElement *)sleeping_queue; t != END_BQ_QUEUE;
2799 prev = t, t = t->link) {
2800 if (t == (StgBlockingQueueElement *)tso) {
2802 sleeping_queue = (StgTSO *)t->link;
2804 prev->link = t->link;
2809 barf("unblockThread (I/O): TSO not found");
2813 barf("unblockThread");
2817 tso->link = END_TSO_QUEUE;
2818 tso->why_blocked = NotBlocked;
2819 tso->block_info.closure = NULL;
2820 PUSH_ON_RUN_QUEUE(tso);
2821 RELEASE_LOCK(&sched_mutex);
2825 unblockThread(StgTSO *tso)
2829 ACQUIRE_LOCK(&sched_mutex);
2830 switch (tso->why_blocked) {
2833 return; /* not blocked */
2836 ASSERT(get_itbl(tso->block_info.closure)->type == MVAR);
2838 StgTSO *last_tso = END_TSO_QUEUE;
2839 StgMVar *mvar = (StgMVar *)(tso->block_info.closure);
2842 for (t = mvar->head; t != END_TSO_QUEUE;
2843 last = &t->link, last_tso = t, t = t->link) {
2846 if (mvar->tail == tso) {
2847 mvar->tail = last_tso;
2852 barf("unblockThread (MVAR): TSO not found");
2855 case BlockedOnBlackHole:
2856 ASSERT(get_itbl(tso->block_info.closure)->type == BLACKHOLE_BQ);
2858 StgBlockingQueue *bq = (StgBlockingQueue *)(tso->block_info.closure);
2860 last = &bq->blocking_queue;
2861 for (t = bq->blocking_queue; t != END_TSO_QUEUE;
2862 last = &t->link, t = t->link) {
2868 barf("unblockThread (BLACKHOLE): TSO not found");
2871 case BlockedOnException:
2873 StgTSO *target = tso->block_info.tso;
2875 ASSERT(get_itbl(target)->type == TSO);
2877 while (target->what_next == ThreadRelocated) {
2878 target = target->link;
2879 ASSERT(get_itbl(target)->type == TSO);
2882 ASSERT(target->blocked_exceptions != NULL);
2884 last = &target->blocked_exceptions;
2885 for (t = target->blocked_exceptions; t != END_TSO_QUEUE;
2886 last = &t->link, t = t->link) {
2887 ASSERT(get_itbl(t)->type == TSO);
2893 barf("unblockThread (Exception): TSO not found");
2897 case BlockedOnWrite:
2899 StgTSO *prev = NULL;
2900 for (t = blocked_queue_hd; t != END_TSO_QUEUE;
2901 prev = t, t = t->link) {
2904 blocked_queue_hd = t->link;
2905 if (blocked_queue_tl == t) {
2906 blocked_queue_tl = END_TSO_QUEUE;
2909 prev->link = t->link;
2910 if (blocked_queue_tl == t) {
2911 blocked_queue_tl = prev;
2917 barf("unblockThread (I/O): TSO not found");
2920 case BlockedOnDelay:
2922 StgTSO *prev = NULL;
2923 for (t = sleeping_queue; t != END_TSO_QUEUE;
2924 prev = t, t = t->link) {
2927 sleeping_queue = t->link;
2929 prev->link = t->link;
2934 barf("unblockThread (I/O): TSO not found");
2938 barf("unblockThread");
2942 tso->link = END_TSO_QUEUE;
2943 tso->why_blocked = NotBlocked;
2944 tso->block_info.closure = NULL;
2945 PUSH_ON_RUN_QUEUE(tso);
2946 RELEASE_LOCK(&sched_mutex);
2950 /* -----------------------------------------------------------------------------
2953 * The following function implements the magic for raising an
2954 * asynchronous exception in an existing thread.
2956 * We first remove the thread from any queue on which it might be
2957 * blocked. The possible blockages are MVARs and BLACKHOLE_BQs.
2959 * We strip the stack down to the innermost CATCH_FRAME, building
2960 * thunks in the heap for all the active computations, so they can
2961 * be restarted if necessary. When we reach a CATCH_FRAME, we build
2962 * an application of the handler to the exception, and push it on
2963 * the top of the stack.
2965 * How exactly do we save all the active computations? We create an
2966 * AP_UPD for every UpdateFrame on the stack. Entering one of these
2967 * AP_UPDs pushes everything from the corresponding update frame
2968 * upwards onto the stack. (Actually, it pushes everything up to the
2969 * next update frame plus a pointer to the next AP_UPD object.
2970 * Entering the next AP_UPD object pushes more onto the stack until we
2971 * reach the last AP_UPD object - at which point the stack should look
2972 * exactly as it did when we killed the TSO and we can continue
2973 * execution by entering the closure on top of the stack.
2975 * We can also kill a thread entirely - this happens if either (a) the
2976 * exception passed to raiseAsync is NULL, or (b) there's no
2977 * CATCH_FRAME on the stack. In either case, we strip the entire
2978 * stack and replace the thread with a zombie.
2980 * -------------------------------------------------------------------------- */
2983 deleteThread(StgTSO *tso)
2985 raiseAsync(tso,NULL);
2989 raiseAsync(StgTSO *tso, StgClosure *exception)
2991 StgUpdateFrame* su = tso->su;
2992 StgPtr sp = tso->sp;
2994 /* Thread already dead? */
2995 if (tso->what_next == ThreadComplete || tso->what_next == ThreadKilled) {
2999 IF_DEBUG(scheduler, sched_belch("raising exception in thread %ld.", tso->id));
3001 /* Remove it from any blocking queues */
3004 /* The stack freezing code assumes there's a closure pointer on
3005 * the top of the stack. This isn't always the case with compiled
3006 * code, so we have to push a dummy closure on the top which just
3007 * returns to the next return address on the stack.
3009 if ( LOOKS_LIKE_GHC_INFO((void*)*sp) ) {
3010 *(--sp) = (W_)&stg_dummy_ret_closure;
3014 nat words = ((P_)su - (P_)sp) - 1;
3018 /* If we find a CATCH_FRAME, and we've got an exception to raise,
3019 * then build PAP(handler,exception,realworld#), and leave it on
3020 * top of the stack ready to enter.
3022 if (get_itbl(su)->type == CATCH_FRAME && exception != NULL) {
3023 StgCatchFrame *cf = (StgCatchFrame *)su;
3024 /* we've got an exception to raise, so let's pass it to the
3025 * handler in this frame.
3027 ap = (StgAP_UPD *)allocate(sizeofW(StgPAP) + 2);
3028 TICK_ALLOC_UPD_PAP(3,0);
3029 SET_HDR(ap,&stg_PAP_info,cf->header.prof.ccs);
3032 ap->fun = cf->handler; /* :: Exception -> IO a */
3033 ap->payload[0] = exception;
3034 ap->payload[1] = ARG_TAG(0); /* realworld token */
3036 /* throw away the stack from Sp up to and including the
3039 sp = (P_)su + sizeofW(StgCatchFrame) - 1;
3042 /* Restore the blocked/unblocked state for asynchronous exceptions
3043 * at the CATCH_FRAME.
3045 * If exceptions were unblocked at the catch, arrange that they
3046 * are unblocked again after executing the handler by pushing an
3047 * unblockAsyncExceptions_ret stack frame.
3049 if (!cf->exceptions_blocked) {
3050 *(sp--) = (W_)&stg_unblockAsyncExceptionszh_ret_info;
3053 /* Ensure that async exceptions are blocked when running the handler.
3055 if (tso->blocked_exceptions == NULL) {
3056 tso->blocked_exceptions = END_TSO_QUEUE;
3059 /* Put the newly-built PAP on top of the stack, ready to execute
3060 * when the thread restarts.
3064 tso->what_next = ThreadEnterGHC;
3065 IF_DEBUG(sanity, checkTSO(tso));
3069 /* First build an AP_UPD consisting of the stack chunk above the
3070 * current update frame, with the top word on the stack as the
3073 ap = (StgAP_UPD *)allocate(AP_sizeW(words));
3078 ap->fun = (StgClosure *)sp[0];
3080 for(i=0; i < (nat)words; ++i) {
3081 ap->payload[i] = (StgClosure *)*sp++;
3084 switch (get_itbl(su)->type) {
3088 SET_HDR(ap,&stg_AP_UPD_info,su->header.prof.ccs /* ToDo */);
3089 TICK_ALLOC_UP_THK(words+1,0);
3092 fprintf(stderr, "scheduler: Updating ");
3093 printPtr((P_)su->updatee);
3094 fprintf(stderr, " with ");
3095 printObj((StgClosure *)ap);
3098 /* Replace the updatee with an indirection - happily
3099 * this will also wake up any threads currently
3100 * waiting on the result.
3102 * Warning: if we're in a loop, more than one update frame on
3103 * the stack may point to the same object. Be careful not to
3104 * overwrite an IND_OLDGEN in this case, because we'll screw
3105 * up the mutable lists. To be on the safe side, don't
3106 * overwrite any kind of indirection at all. See also
3107 * threadSqueezeStack in GC.c, where we have to make a similar
3110 if (!closure_IND(su->updatee)) {
3111 UPD_IND_NOLOCK(su->updatee,ap); /* revert the black hole */
3114 sp += sizeofW(StgUpdateFrame) -1;
3115 sp[0] = (W_)ap; /* push onto stack */
3121 StgCatchFrame *cf = (StgCatchFrame *)su;
3124 /* We want a PAP, not an AP_UPD. Fortunately, the
3125 * layout's the same.
3127 SET_HDR(ap,&stg_PAP_info,su->header.prof.ccs /* ToDo */);
3128 TICK_ALLOC_UPD_PAP(words+1,0);
3130 /* now build o = FUN(catch,ap,handler) */
3131 o = (StgClosure *)allocate(sizeofW(StgClosure)+2);
3132 TICK_ALLOC_FUN(2,0);
3133 SET_HDR(o,&stg_catch_info,su->header.prof.ccs /* ToDo */);
3134 o->payload[0] = (StgClosure *)ap;
3135 o->payload[1] = cf->handler;
3138 fprintf(stderr, "scheduler: Built ");
3139 printObj((StgClosure *)o);
3142 /* pop the old handler and put o on the stack */
3144 sp += sizeofW(StgCatchFrame) - 1;
3151 StgSeqFrame *sf = (StgSeqFrame *)su;
3154 SET_HDR(ap,&stg_PAP_info,su->header.prof.ccs /* ToDo */);
3155 TICK_ALLOC_UPD_PAP(words+1,0);
3157 /* now build o = FUN(seq,ap) */
3158 o = (StgClosure *)allocate(sizeofW(StgClosure)+1);
3159 TICK_ALLOC_SE_THK(1,0);
3160 SET_HDR(o,&stg_seq_info,su->header.prof.ccs /* ToDo */);
3161 o->payload[0] = (StgClosure *)ap;
3164 fprintf(stderr, "scheduler: Built ");
3165 printObj((StgClosure *)o);
3168 /* pop the old handler and put o on the stack */
3170 sp += sizeofW(StgSeqFrame) - 1;
3176 /* We've stripped the entire stack, the thread is now dead. */
3177 sp += sizeofW(StgStopFrame) - 1;
3178 sp[0] = (W_)exception; /* save the exception */
3179 tso->what_next = ThreadKilled;
3180 tso->su = (StgUpdateFrame *)(sp+1);
3191 /* -----------------------------------------------------------------------------
3192 resurrectThreads is called after garbage collection on the list of
3193 threads found to be garbage. Each of these threads will be woken
3194 up and sent a signal: BlockedOnDeadMVar if the thread was blocked
3195 on an MVar, or NonTermination if the thread was blocked on a Black
3197 -------------------------------------------------------------------------- */
3200 resurrectThreads( StgTSO *threads )
3204 for (tso = threads; tso != END_TSO_QUEUE; tso = next) {
3205 next = tso->global_link;
3206 tso->global_link = all_threads;
3208 IF_DEBUG(scheduler, sched_belch("resurrecting thread %d", tso->id));
3210 switch (tso->why_blocked) {
3212 case BlockedOnException:
3213 raiseAsync(tso,(StgClosure *)BlockedOnDeadMVar_closure);
3215 case BlockedOnBlackHole:
3216 raiseAsync(tso,(StgClosure *)NonTermination_closure);
3219 /* This might happen if the thread was blocked on a black hole
3220 * belonging to a thread that we've just woken up (raiseAsync
3221 * can wake up threads, remember...).
3225 barf("resurrectThreads: thread blocked in a strange way");
3230 /* -----------------------------------------------------------------------------
3231 * Blackhole detection: if we reach a deadlock, test whether any
3232 * threads are blocked on themselves. Any threads which are found to
3233 * be self-blocked get sent a NonTermination exception.
3235 * This is only done in a deadlock situation in order to avoid
3236 * performance overhead in the normal case.
3237 * -------------------------------------------------------------------------- */
3240 detectBlackHoles( void )
3242 StgTSO *t = all_threads;
3243 StgUpdateFrame *frame;
3244 StgClosure *blocked_on;
3246 for (t = all_threads; t != END_TSO_QUEUE; t = t->global_link) {
3248 while (t->what_next == ThreadRelocated) {
3250 ASSERT(get_itbl(t)->type == TSO);
3253 if (t->why_blocked != BlockedOnBlackHole) {
3257 blocked_on = t->block_info.closure;
3259 for (frame = t->su; ; frame = frame->link) {
3260 switch (get_itbl(frame)->type) {
3263 if (frame->updatee == blocked_on) {
3264 /* We are blocking on one of our own computations, so
3265 * send this thread the NonTermination exception.
3268 sched_belch("thread %d is blocked on itself", t->id));
3269 raiseAsync(t, (StgClosure *)NonTermination_closure);
3290 //@node Debugging Routines, Index, Exception Handling Routines, Main scheduling code
3291 //@subsection Debugging Routines
3293 /* -----------------------------------------------------------------------------
3294 Debugging: why is a thread blocked
3295 -------------------------------------------------------------------------- */
3300 printThreadBlockage(StgTSO *tso)
3302 switch (tso->why_blocked) {
3304 fprintf(stderr,"is blocked on read from fd %d", tso->block_info.fd);
3306 case BlockedOnWrite:
3307 fprintf(stderr,"is blocked on write to fd %d", tso->block_info.fd);
3309 case BlockedOnDelay:
3310 fprintf(stderr,"is blocked until %d", tso->block_info.target);
3313 fprintf(stderr,"is blocked on an MVar");
3315 case BlockedOnException:
3316 fprintf(stderr,"is blocked on delivering an exception to thread %d",
3317 tso->block_info.tso->id);
3319 case BlockedOnBlackHole:
3320 fprintf(stderr,"is blocked on a black hole");
3323 fprintf(stderr,"is not blocked");
3327 fprintf(stderr,"is blocked on global address; local FM_BQ is %p (%s)",
3328 tso->block_info.closure, info_type(tso->block_info.closure));
3330 case BlockedOnGA_NoSend:
3331 fprintf(stderr,"is blocked on global address (no send); local FM_BQ is %p (%s)",
3332 tso->block_info.closure, info_type(tso->block_info.closure));
3336 barf("printThreadBlockage: strange tso->why_blocked: %d for TSO %d (%d)",
3337 tso->why_blocked, tso->id, tso);
3342 printThreadStatus(StgTSO *tso)
3344 switch (tso->what_next) {
3346 fprintf(stderr,"has been killed");
3348 case ThreadComplete:
3349 fprintf(stderr,"has completed");
3352 printThreadBlockage(tso);
3357 printAllThreads(void)
3362 char time_string[TIME_STR_LEN], node_str[NODE_STR_LEN];
3363 ullong_format_string(TIME_ON_PROC(CurrentProc),
3364 time_string, rtsFalse/*no commas!*/);
3366 sched_belch("all threads at [%s]:", time_string);
3368 char time_string[TIME_STR_LEN], node_str[NODE_STR_LEN];
3369 ullong_format_string(CURRENT_TIME,
3370 time_string, rtsFalse/*no commas!*/);
3372 sched_belch("all threads at [%s]:", time_string);
3374 sched_belch("all threads:");
3377 for (t = all_threads; t != END_TSO_QUEUE; t = t->global_link) {
3378 fprintf(stderr, "\tthread %d ", t->id);
3379 printThreadStatus(t);
3380 fprintf(stderr,"\n");
3385 Print a whole blocking queue attached to node (debugging only).
3390 print_bq (StgClosure *node)
3392 StgBlockingQueueElement *bqe;
3396 fprintf(stderr,"## BQ of closure %p (%s): ",
3397 node, info_type(node));
3399 /* should cover all closures that may have a blocking queue */
3400 ASSERT(get_itbl(node)->type == BLACKHOLE_BQ ||
3401 get_itbl(node)->type == FETCH_ME_BQ ||
3402 get_itbl(node)->type == RBH ||
3403 get_itbl(node)->type == MVAR);
3405 ASSERT(node!=(StgClosure*)NULL); // sanity check
3407 print_bqe(((StgBlockingQueue*)node)->blocking_queue);
3411 Print a whole blocking queue starting with the element bqe.
3414 print_bqe (StgBlockingQueueElement *bqe)
3419 NB: In a parallel setup a BQ of an RBH must end with an RBH_Save closure;
3421 for (end = (bqe==END_BQ_QUEUE);
3422 !end; // iterate until bqe points to a CONSTR
3423 end = (get_itbl(bqe)->type == CONSTR) || (bqe->link==END_BQ_QUEUE),
3424 bqe = end ? END_BQ_QUEUE : bqe->link) {
3425 ASSERT(bqe != END_BQ_QUEUE); // sanity check
3426 ASSERT(bqe != (StgBlockingQueueElement *)NULL); // sanity check
3427 /* types of closures that may appear in a blocking queue */
3428 ASSERT(get_itbl(bqe)->type == TSO ||
3429 get_itbl(bqe)->type == BLOCKED_FETCH ||
3430 get_itbl(bqe)->type == CONSTR);
3431 /* only BQs of an RBH end with an RBH_Save closure */
3432 //ASSERT(get_itbl(bqe)->type != CONSTR || get_itbl(node)->type == RBH);
3434 switch (get_itbl(bqe)->type) {
3436 fprintf(stderr," TSO %u (%x),",
3437 ((StgTSO *)bqe)->id, ((StgTSO *)bqe));
3440 fprintf(stderr," BF (node=%p, ga=((%x, %d, %x)),",
3441 ((StgBlockedFetch *)bqe)->node,
3442 ((StgBlockedFetch *)bqe)->ga.payload.gc.gtid,
3443 ((StgBlockedFetch *)bqe)->ga.payload.gc.slot,
3444 ((StgBlockedFetch *)bqe)->ga.weight);
3447 fprintf(stderr," %s (IP %p),",
3448 (get_itbl(bqe) == &stg_RBH_Save_0_info ? "RBH_Save_0" :
3449 get_itbl(bqe) == &stg_RBH_Save_1_info ? "RBH_Save_1" :
3450 get_itbl(bqe) == &stg_RBH_Save_2_info ? "RBH_Save_2" :
3451 "RBH_Save_?"), get_itbl(bqe));
3454 barf("Unexpected closure type %s in blocking queue", // of %p (%s)",
3455 info_type((StgClosure *)bqe)); // , node, info_type(node));
3459 fputc('\n', stderr);
3461 # elif defined(GRAN)
3463 print_bq (StgClosure *node)
3465 StgBlockingQueueElement *bqe;
3466 PEs node_loc, tso_loc;
3469 /* should cover all closures that may have a blocking queue */
3470 ASSERT(get_itbl(node)->type == BLACKHOLE_BQ ||
3471 get_itbl(node)->type == FETCH_ME_BQ ||
3472 get_itbl(node)->type == RBH);
3474 ASSERT(node!=(StgClosure*)NULL); // sanity check
3475 node_loc = where_is(node);
3477 fprintf(stderr,"## BQ of closure %p (%s) on [PE %d]: ",
3478 node, info_type(node), node_loc);
3481 NB: In a parallel setup a BQ of an RBH must end with an RBH_Save closure;
3483 for (bqe = ((StgBlockingQueue*)node)->blocking_queue, end = (bqe==END_BQ_QUEUE);
3484 !end; // iterate until bqe points to a CONSTR
3485 end = (get_itbl(bqe)->type == CONSTR) || (bqe->link==END_BQ_QUEUE), bqe = end ? END_BQ_QUEUE : bqe->link) {
3486 ASSERT(bqe != END_BQ_QUEUE); // sanity check
3487 ASSERT(bqe != (StgBlockingQueueElement *)NULL); // sanity check
3488 /* types of closures that may appear in a blocking queue */
3489 ASSERT(get_itbl(bqe)->type == TSO ||
3490 get_itbl(bqe)->type == CONSTR);
3491 /* only BQs of an RBH end with an RBH_Save closure */
3492 ASSERT(get_itbl(bqe)->type != CONSTR || get_itbl(node)->type == RBH);
3494 tso_loc = where_is((StgClosure *)bqe);
3495 switch (get_itbl(bqe)->type) {
3497 fprintf(stderr," TSO %d (%p) on [PE %d],",
3498 ((StgTSO *)bqe)->id, (StgTSO *)bqe, tso_loc);
3501 fprintf(stderr," %s (IP %p),",
3502 (get_itbl(bqe) == &stg_RBH_Save_0_info ? "RBH_Save_0" :
3503 get_itbl(bqe) == &stg_RBH_Save_1_info ? "RBH_Save_1" :
3504 get_itbl(bqe) == &stg_RBH_Save_2_info ? "RBH_Save_2" :
3505 "RBH_Save_?"), get_itbl(bqe));
3508 barf("Unexpected closure type %s in blocking queue of %p (%s)",
3509 info_type((StgClosure *)bqe), node, info_type(node));
3513 fputc('\n', stderr);
3517 Nice and easy: only TSOs on the blocking queue
3520 print_bq (StgClosure *node)
3524 ASSERT(node!=(StgClosure*)NULL); // sanity check
3525 for (tso = ((StgBlockingQueue*)node)->blocking_queue;
3526 tso != END_TSO_QUEUE;
3528 ASSERT(tso!=NULL && tso!=END_TSO_QUEUE); // sanity check
3529 ASSERT(get_itbl(tso)->type == TSO); // guess what, sanity check
3530 fprintf(stderr," TSO %d (%p),", tso->id, tso);
3532 fputc('\n', stderr);
3543 for (i=0, tso=run_queue_hd;
3544 tso != END_TSO_QUEUE;
3553 sched_belch(char *s, ...)
3558 fprintf(stderr, "scheduler (task %ld): ", osThreadId());
3560 fprintf(stderr, "== ");
3562 fprintf(stderr, "scheduler: ");
3564 vfprintf(stderr, s, ap);
3565 fprintf(stderr, "\n");
3571 //@node Index, , Debugging Routines, Main scheduling code
3575 //* MainRegTable:: @cindex\s-+MainRegTable
3576 //* StgMainThread:: @cindex\s-+StgMainThread
3577 //* awaken_blocked_queue:: @cindex\s-+awaken_blocked_queue
3578 //* blocked_queue_hd:: @cindex\s-+blocked_queue_hd
3579 //* blocked_queue_tl:: @cindex\s-+blocked_queue_tl
3580 //* context_switch:: @cindex\s-+context_switch
3581 //* createThread:: @cindex\s-+createThread
3582 //* gc_pending_cond:: @cindex\s-+gc_pending_cond
3583 //* initScheduler:: @cindex\s-+initScheduler
3584 //* interrupted:: @cindex\s-+interrupted
3585 //* next_thread_id:: @cindex\s-+next_thread_id
3586 //* print_bq:: @cindex\s-+print_bq
3587 //* run_queue_hd:: @cindex\s-+run_queue_hd
3588 //* run_queue_tl:: @cindex\s-+run_queue_tl
3589 //* sched_mutex:: @cindex\s-+sched_mutex
3590 //* schedule:: @cindex\s-+schedule
3591 //* take_off_run_queue:: @cindex\s-+take_off_run_queue
3592 //* task_ids:: @cindex\s-+task_ids
3593 //* term_mutex:: @cindex\s-+term_mutex
3594 //* thread_ready_cond:: @cindex\s-+thread_ready_cond