1 /* ---------------------------------------------------------------------------
2 * $Id: Schedule.c,v 1.141 2002/04/26 22:35:54 sof Exp $
4 * (c) The GHC Team, 1998-2000
8 * Different GHC ways use this scheduler quite differently (see comments below)
9 * Here is the global picture:
11 * WAY Name CPP flag What's it for
12 * --------------------------------------
13 * mp GUM PAR Parallel execution on a distributed memory machine
14 * s SMP SMP Parallel execution on a shared memory machine
15 * mg GranSim GRAN Simulation of parallel execution
16 * md GUM/GdH DIST Distributed execution (based on GUM)
18 * --------------------------------------------------------------------------*/
20 //@node Main scheduling code, , ,
21 //@section Main scheduling code
24 * Version with scheduler monitor support for SMPs (WAY=s):
26 This design provides a high-level API to create and schedule threads etc.
27 as documented in the SMP design document.
29 It uses a monitor design controlled by a single mutex to exercise control
30 over accesses to shared data structures, and builds on the Posix threads
33 The majority of state is shared. In order to keep essential per-task state,
34 there is a Capability structure, which contains all the information
35 needed to run a thread: its STG registers, a pointer to its TSO, a
36 nursery etc. During STG execution, a pointer to the capability is
37 kept in a register (BaseReg).
39 In a non-SMP build, there is one global capability, namely MainRegTable.
43 * Version with support for distributed memory parallelism aka GUM (WAY=mp):
45 The main scheduling loop in GUM iterates until a finish message is received.
46 In that case a global flag @receivedFinish@ is set and this instance of
47 the RTS shuts down. See ghc/rts/parallel/HLComms.c:processMessages()
48 for the handling of incoming messages, such as PP_FINISH.
49 Note that in the parallel case we have a system manager that coordinates
50 different PEs, each of which are running one instance of the RTS.
51 See ghc/rts/parallel/SysMan.c for the main routine of the parallel program.
52 From this routine processes executing ghc/rts/Main.c are spawned. -- HWL
54 * Version with support for simulating parallel execution aka GranSim (WAY=mg):
56 The main scheduling code in GranSim is quite different from that in std
57 (concurrent) Haskell: while concurrent Haskell just iterates over the
58 threads in the runnable queue, GranSim is event driven, i.e. it iterates
59 over the events in the global event queue. -- HWL
64 //* Variables and Data structures::
65 //* Main scheduling loop::
66 //* Suspend and Resume::
68 //* Garbage Collextion Routines::
69 //* Blocking Queue Routines::
70 //* Exception Handling Routines::
71 //* Debugging Routines::
75 //@node Includes, Variables and Data structures, Main scheduling code, Main scheduling code
76 //@subsection Includes
78 #include "PosixSource.h"
85 #include "StgStartup.h"
88 #include "StgMiscClosures.h"
90 #include "Interpreter.h"
91 #include "Exception.h"
100 #include "Proftimer.h"
101 #include "ProfHeap.h"
103 #if defined(GRAN) || defined(PAR)
104 # include "GranSimRts.h"
105 # include "GranSim.h"
106 # include "ParallelRts.h"
107 # include "Parallel.h"
108 # include "ParallelDebug.h"
109 # include "FetchMe.h"
113 #include "Capability.h"
114 #include "OSThreads.h"
117 #ifdef HAVE_SYS_TYPES_H
118 #include <sys/types.h>
126 //@node Variables and Data structures, Prototypes, Includes, Main scheduling code
127 //@subsection Variables and Data structures
129 /* Main thread queue.
130 * Locks required: sched_mutex.
132 StgMainThread *main_threads;
135 * Locks required: sched_mutex.
139 StgTSO* ActiveTSO = NULL; /* for assigning system costs; GranSim-Light only */
140 /* rtsTime TimeOfNextEvent, EndOfTimeSlice; now in GranSim.c */
143 In GranSim we have a runnable and a blocked queue for each processor.
144 In order to minimise code changes new arrays run_queue_hds/tls
145 are created. run_queue_hd is then a short cut (macro) for
146 run_queue_hds[CurrentProc] (see GranSim.h).
149 StgTSO *run_queue_hds[MAX_PROC], *run_queue_tls[MAX_PROC];
150 StgTSO *blocked_queue_hds[MAX_PROC], *blocked_queue_tls[MAX_PROC];
151 StgTSO *ccalling_threadss[MAX_PROC];
152 /* We use the same global list of threads (all_threads) in GranSim as in
153 the std RTS (i.e. we are cheating). However, we don't use this list in
154 the GranSim specific code at the moment (so we are only potentially
159 StgTSO *run_queue_hd, *run_queue_tl;
160 StgTSO *blocked_queue_hd, *blocked_queue_tl;
161 StgTSO *sleeping_queue; /* perhaps replace with a hash table? */
165 /* Linked list of all threads.
166 * Used for detecting garbage collected threads.
170 /* When a thread performs a safe C call (_ccall_GC, using old
171 * terminology), it gets put on the suspended_ccalling_threads
172 * list. Used by the garbage collector.
174 static StgTSO *suspended_ccalling_threads;
176 static StgTSO *threadStackOverflow(StgTSO *tso);
178 /* KH: The following two flags are shared memory locations. There is no need
179 to lock them, since they are only unset at the end of a scheduler
183 /* flag set by signal handler to precipitate a context switch */
184 //@cindex context_switch
187 /* if this flag is set as well, give up execution */
188 //@cindex interrupted
191 /* Next thread ID to allocate.
192 * Locks required: sched_mutex
194 //@cindex next_thread_id
195 StgThreadID next_thread_id = 1;
198 * Pointers to the state of the current thread.
199 * Rule of thumb: if CurrentTSO != NULL, then we're running a Haskell
200 * thread. If CurrentTSO == NULL, then we're at the scheduler level.
203 /* The smallest stack size that makes any sense is:
204 * RESERVED_STACK_WORDS (so we can get back from the stack overflow)
205 * + sizeofW(StgStopFrame) (the stg_stop_thread_info frame)
206 * + 1 (the realworld token for an IO thread)
207 * + 1 (the closure to enter)
209 * A thread with this stack will bomb immediately with a stack
210 * overflow, which will increase its stack size.
213 #define MIN_STACK_WORDS (RESERVED_STACK_WORDS + sizeofW(StgStopFrame) + 2)
220 /* This is used in `TSO.h' and gcc 2.96 insists that this variable actually
221 * exists - earlier gccs apparently didn't.
228 void addToBlockedQueue ( StgTSO *tso );
230 static void schedule ( void );
231 void interruptStgRts ( void );
233 static StgTSO * createThread_ ( nat size, rtsBool have_lock, StgInt pri );
235 static StgTSO * createThread_ ( nat size, rtsBool have_lock );
238 static void detectBlackHoles ( void );
241 static void sched_belch(char *s, ...);
244 #if defined(RTS_SUPPORTS_THREADS)
245 /* ToDo: carefully document the invariants that go together
246 * with these synchronisation objects.
248 Mutex sched_mutex = INIT_MUTEX_VAR;
249 Mutex term_mutex = INIT_MUTEX_VAR;
252 static Condition gc_pending_cond = INIT_COND_VAR;
256 #endif /* RTS_SUPPORTS_THREADS */
260 rtsTime TimeOfLastYield;
261 rtsBool emitSchedule = rtsTrue;
265 char *whatNext_strs[] = {
273 char *threadReturnCode_strs[] = {
274 "HeapOverflow", /* might also be StackOverflow */
283 StgTSO * createSparkThread(rtsSpark spark);
284 StgTSO * activateSpark (rtsSpark spark);
288 * The thread state for the main thread.
289 // ToDo: check whether not needed any more
293 #if defined(PAR) || defined(RTS_SUPPORTS_THREADS)
294 static void taskStart(void);
305 //@node Main scheduling loop, Suspend and Resume, Prototypes, Main scheduling code
306 //@subsection Main scheduling loop
308 /* ---------------------------------------------------------------------------
309 Main scheduling loop.
311 We use round-robin scheduling, each thread returning to the
312 scheduler loop when one of these conditions is detected:
315 * timer expires (thread yields)
320 Locking notes: we acquire the scheduler lock once at the beginning
321 of the scheduler loop, and release it when
323 * running a thread, or
324 * waiting for work, or
325 * waiting for a GC to complete.
328 In a GranSim setup this loop iterates over the global event queue.
329 This revolves around the global event queue, which determines what
330 to do next. Therefore, it's more complicated than either the
331 concurrent or the parallel (GUM) setup.
334 GUM iterates over incoming messages.
335 It starts with nothing to do (thus CurrentTSO == END_TSO_QUEUE),
336 and sends out a fish whenever it has nothing to do; in-between
337 doing the actual reductions (shared code below) it processes the
338 incoming messages and deals with delayed operations
339 (see PendingFetches).
340 This is not the ugliest code you could imagine, but it's bloody close.
342 ------------------------------------------------------------------------ */
349 StgThreadReturnCode ret;
357 rtsBool receivedFinish = rtsFalse;
359 nat tp_size, sp_size; // stats only
362 rtsBool was_interrupted = rtsFalse;
364 ACQUIRE_LOCK(&sched_mutex);
366 #if defined(RTS_SUPPORTS_THREADS)
367 waitForWorkCapability(&sched_mutex, &cap, rtsFalse);
369 /* simply initialise it in the non-threaded case */
370 grabCapability(&cap);
374 /* set up first event to get things going */
375 /* ToDo: assign costs for system setup and init MainTSO ! */
376 new_event(CurrentProc, CurrentProc, CurrentTime[CurrentProc],
378 CurrentTSO, (StgClosure*)NULL, (rtsSpark*)NULL);
381 fprintf(stderr, "GRAN: Init CurrentTSO (in schedule) = %p\n", CurrentTSO);
382 G_TSO(CurrentTSO, 5));
384 if (RtsFlags.GranFlags.Light) {
385 /* Save current time; GranSim Light only */
386 CurrentTSO->gran.clock = CurrentTime[CurrentProc];
389 event = get_next_event();
391 while (event!=(rtsEvent*)NULL) {
392 /* Choose the processor with the next event */
393 CurrentProc = event->proc;
394 CurrentTSO = event->tso;
398 while (!receivedFinish) { /* set by processMessages */
399 /* when receiving PP_FINISH message */
406 IF_DEBUG(scheduler, printAllThreads());
408 #if defined(RTS_SUPPORTS_THREADS)
409 /* Check to see whether there are any worker threads
410 waiting to deposit external call results. If so,
411 yield our capability */
412 yieldToReturningWorker(&sched_mutex, &cap);
415 /* If we're interrupted (the user pressed ^C, or some other
416 * termination condition occurred), kill all the currently running
420 IF_DEBUG(scheduler, sched_belch("interrupted"));
422 interrupted = rtsFalse;
423 was_interrupted = rtsTrue;
426 /* Go through the list of main threads and wake up any
427 * clients whose computations have finished. ToDo: this
428 * should be done more efficiently without a linear scan
429 * of the main threads list, somehow...
431 #if defined(RTS_SUPPORTS_THREADS)
433 StgMainThread *m, **prev;
434 prev = &main_threads;
435 for (m = main_threads; m != NULL; m = m->link) {
436 switch (m->tso->what_next) {
439 *(m->ret) = (StgClosure *)m->tso->sp[0];
443 broadcastCondition(&m->wakeup);
446 m->tso->label = NULL;
450 if (m->ret) *(m->ret) = NULL;
452 if (was_interrupted) {
453 m->stat = Interrupted;
457 broadcastCondition(&m->wakeup);
460 m->tso->label = NULL;
469 #else /* not threaded */
472 /* in GUM do this only on the Main PE */
475 /* If our main thread has finished or been killed, return.
478 StgMainThread *m = main_threads;
479 if (m->tso->what_next == ThreadComplete
480 || m->tso->what_next == ThreadKilled) {
483 m->tso->label = NULL;
485 main_threads = main_threads->link;
486 if (m->tso->what_next == ThreadComplete) {
487 /* we finished successfully, fill in the return value */
488 if (m->ret) { *(m->ret) = (StgClosure *)m->tso->sp[0]; };
492 if (m->ret) { *(m->ret) = NULL; };
493 if (was_interrupted) {
494 m->stat = Interrupted;
504 /* Top up the run queue from our spark pool. We try to make the
505 * number of threads in the run queue equal to the number of
508 * Disable spark support in SMP for now, non-essential & requires
509 * a little bit of work to make it compile cleanly. -- sof 1/02.
511 #if 0 /* defined(SMP) */
513 nat n = getFreeCapabilities();
514 StgTSO *tso = run_queue_hd;
516 /* Count the run queue */
517 while (n > 0 && tso != END_TSO_QUEUE) {
524 spark = findSpark(rtsFalse);
526 break; /* no more sparks in the pool */
528 /* I'd prefer this to be done in activateSpark -- HWL */
529 /* tricky - it needs to hold the scheduler lock and
530 * not try to re-acquire it -- SDM */
531 createSparkThread(spark);
533 sched_belch("==^^ turning spark of closure %p into a thread",
534 (StgClosure *)spark));
537 /* We need to wake up the other tasks if we just created some
540 if (getFreeCapabilities() - n > 1) {
541 signalCondition( &thread_ready_cond );
546 /* check for signals each time around the scheduler */
547 #ifndef mingw32_TARGET_OS
548 if (signals_pending()) {
549 RELEASE_LOCK(&sched_mutex); /* ToDo: kill */
550 startSignalHandlers();
551 ACQUIRE_LOCK(&sched_mutex);
555 /* Check whether any waiting threads need to be woken up. If the
556 * run queue is empty, and there are no other tasks running, we
557 * can wait indefinitely for something to happen.
558 * ToDo: what if another client comes along & requests another
561 if ( !EMPTY_QUEUE(blocked_queue_hd) || !EMPTY_QUEUE(sleeping_queue) ) {
562 awaitEvent( EMPTY_RUN_QUEUE()
564 && allFreeCapabilities()
568 /* we can be interrupted while waiting for I/O... */
569 if (interrupted) continue;
572 * Detect deadlock: when we have no threads to run, there are no
573 * threads waiting on I/O or sleeping, and all the other tasks are
574 * waiting for work, we must have a deadlock of some description.
576 * We first try to find threads blocked on themselves (ie. black
577 * holes), and generate NonTermination exceptions where necessary.
579 * If no threads are black holed, we have a deadlock situation, so
580 * inform all the main threads.
583 if ( EMPTY_THREAD_QUEUES()
584 #if defined(RTS_SUPPORTS_THREADS)
585 && EMPTY_QUEUE(suspended_ccalling_threads)
588 && allFreeCapabilities()
592 IF_DEBUG(scheduler, sched_belch("deadlocked, forcing major GC..."));
593 #if defined(THREADED_RTS)
594 /* and SMP mode ..? */
595 releaseCapability(cap);
597 // Garbage collection can release some new threads due to
598 // either (a) finalizers or (b) threads resurrected because
599 // they are about to be send BlockedOnDeadMVar. Any threads
600 // thus released will be immediately runnable.
601 GarbageCollect(GetRoots,rtsTrue);
603 if ( !EMPTY_RUN_QUEUE() ) { goto not_deadlocked; }
606 sched_belch("still deadlocked, checking for black holes..."));
609 if ( !EMPTY_RUN_QUEUE() ) { goto not_deadlocked; }
611 #ifndef mingw32_TARGET_OS
612 /* If we have user-installed signal handlers, then wait
613 * for signals to arrive rather then bombing out with a
616 #if defined(RTS_SUPPORTS_THREADS)
617 if ( 0 ) { /* hmm..what to do? Simply stop waiting for
618 a signal with no runnable threads (or I/O
619 suspended ones) leads nowhere quick.
620 For now, simply shut down when we reach this
623 ToDo: define precisely under what conditions
624 the Scheduler should shut down in an MT setting.
627 if ( anyUserHandlers() ) {
630 sched_belch("still deadlocked, waiting for signals..."));
634 // we might be interrupted...
635 if (interrupted) { continue; }
637 if (signals_pending()) {
638 RELEASE_LOCK(&sched_mutex);
639 startSignalHandlers();
640 ACQUIRE_LOCK(&sched_mutex);
642 ASSERT(!EMPTY_RUN_QUEUE());
647 /* Probably a real deadlock. Send the current main thread the
648 * Deadlock exception (or in the SMP build, send *all* main
649 * threads the deadlock exception, since none of them can make
654 #if defined(RTS_SUPPORTS_THREADS)
655 for (m = main_threads; m != NULL; m = m->link) {
656 switch (m->tso->why_blocked) {
657 case BlockedOnBlackHole:
658 raiseAsync(m->tso, (StgClosure *)NonTermination_closure);
660 case BlockedOnException:
662 raiseAsync(m->tso, (StgClosure *)Deadlock_closure);
665 barf("deadlock: main thread blocked in a strange way");
670 switch (m->tso->why_blocked) {
671 case BlockedOnBlackHole:
672 raiseAsync(m->tso, (StgClosure *)NonTermination_closure);
674 case BlockedOnException:
676 raiseAsync(m->tso, (StgClosure *)Deadlock_closure);
679 barf("deadlock: main thread blocked in a strange way");
684 #if defined(RTS_SUPPORTS_THREADS)
685 /* ToDo: revisit conditions (and mechanism) for shutting
686 down a multi-threaded world */
687 IF_DEBUG(scheduler, sched_belch("all done, i think...shutting down."));
688 RELEASE_LOCK(&sched_mutex);
696 /* ToDo: add deadlock detection in GUM (similar to SMP) -- HWL */
700 /* If there's a GC pending, don't do anything until it has
704 IF_DEBUG(scheduler,sched_belch("waiting for GC"));
705 waitCondition( &gc_pending_cond, &sched_mutex );
709 #if defined(RTS_SUPPORTS_THREADS)
710 /* block until we've got a thread on the run queue and a free
714 if ( EMPTY_RUN_QUEUE() ) {
715 /* Give up our capability */
716 releaseCapability(cap);
717 IF_DEBUG(scheduler, sched_belch("thread %d: waiting for work", osThreadId()));
718 waitForWorkCapability(&sched_mutex, &cap, rtsTrue);
719 IF_DEBUG(scheduler, sched_belch("thread %d: work now available", osThreadId()));
721 while ( EMPTY_RUN_QUEUE() ) {
722 waitForWorkCapability(&sched_mutex, &cap);
723 IF_DEBUG(scheduler, sched_belch("thread %d: work now available", osThreadId()));
730 if (RtsFlags.GranFlags.Light)
731 GranSimLight_enter_system(event, &ActiveTSO); // adjust ActiveTSO etc
733 /* adjust time based on time-stamp */
734 if (event->time > CurrentTime[CurrentProc] &&
735 event->evttype != ContinueThread)
736 CurrentTime[CurrentProc] = event->time;
738 /* Deal with the idle PEs (may issue FindWork or MoveSpark events) */
739 if (!RtsFlags.GranFlags.Light)
742 IF_DEBUG(gran, fprintf(stderr, "GRAN: switch by event-type\n"));
744 /* main event dispatcher in GranSim */
745 switch (event->evttype) {
746 /* Should just be continuing execution */
748 IF_DEBUG(gran, fprintf(stderr, "GRAN: doing ContinueThread\n"));
749 /* ToDo: check assertion
750 ASSERT(run_queue_hd != (StgTSO*)NULL &&
751 run_queue_hd != END_TSO_QUEUE);
753 /* Ignore ContinueThreads for fetching threads (if synchr comm) */
754 if (!RtsFlags.GranFlags.DoAsyncFetch &&
755 procStatus[CurrentProc]==Fetching) {
756 belch("ghuH: Spurious ContinueThread while Fetching ignored; TSO %d (%p) [PE %d]",
757 CurrentTSO->id, CurrentTSO, CurrentProc);
760 /* Ignore ContinueThreads for completed threads */
761 if (CurrentTSO->what_next == ThreadComplete) {
762 belch("ghuH: found a ContinueThread event for completed thread %d (%p) [PE %d] (ignoring ContinueThread)",
763 CurrentTSO->id, CurrentTSO, CurrentProc);
766 /* Ignore ContinueThreads for threads that are being migrated */
767 if (PROCS(CurrentTSO)==Nowhere) {
768 belch("ghuH: trying to run the migrating TSO %d (%p) [PE %d] (ignoring ContinueThread)",
769 CurrentTSO->id, CurrentTSO, CurrentProc);
772 /* The thread should be at the beginning of the run queue */
773 if (CurrentTSO!=run_queue_hds[CurrentProc]) {
774 belch("ghuH: TSO %d (%p) [PE %d] is not at the start of the run_queue when doing a ContinueThread",
775 CurrentTSO->id, CurrentTSO, CurrentProc);
776 break; // run the thread anyway
779 new_event(proc, proc, CurrentTime[proc],
781 (StgTSO*)NULL, (StgClosure*)NULL, (rtsSpark*)NULL);
783 */ /* Catches superfluous CONTINUEs -- should be unnecessary */
784 break; // now actually run the thread; DaH Qu'vam yImuHbej
787 do_the_fetchnode(event);
788 goto next_thread; /* handle next event in event queue */
791 do_the_globalblock(event);
792 goto next_thread; /* handle next event in event queue */
795 do_the_fetchreply(event);
796 goto next_thread; /* handle next event in event queue */
798 case UnblockThread: /* Move from the blocked queue to the tail of */
799 do_the_unblock(event);
800 goto next_thread; /* handle next event in event queue */
802 case ResumeThread: /* Move from the blocked queue to the tail of */
803 /* the runnable queue ( i.e. Qu' SImqa'lu') */
804 event->tso->gran.blocktime +=
805 CurrentTime[CurrentProc] - event->tso->gran.blockedat;
806 do_the_startthread(event);
807 goto next_thread; /* handle next event in event queue */
810 do_the_startthread(event);
811 goto next_thread; /* handle next event in event queue */
814 do_the_movethread(event);
815 goto next_thread; /* handle next event in event queue */
818 do_the_movespark(event);
819 goto next_thread; /* handle next event in event queue */
822 do_the_findwork(event);
823 goto next_thread; /* handle next event in event queue */
826 barf("Illegal event type %u\n", event->evttype);
829 /* This point was scheduler_loop in the old RTS */
831 IF_DEBUG(gran, belch("GRAN: after main switch"));
833 TimeOfLastEvent = CurrentTime[CurrentProc];
834 TimeOfNextEvent = get_time_of_next_event();
835 IgnoreEvents=(TimeOfNextEvent==0); // HWL HACK
836 // CurrentTSO = ThreadQueueHd;
838 IF_DEBUG(gran, belch("GRAN: time of next event is: %ld",
841 if (RtsFlags.GranFlags.Light)
842 GranSimLight_leave_system(event, &ActiveTSO);
844 EndOfTimeSlice = CurrentTime[CurrentProc]+RtsFlags.GranFlags.time_slice;
847 belch("GRAN: end of time-slice is %#lx", EndOfTimeSlice));
849 /* in a GranSim setup the TSO stays on the run queue */
851 /* Take a thread from the run queue. */
852 t = POP_RUN_QUEUE(); // take_off_run_queue(t);
855 fprintf(stderr, "GRAN: About to run current thread, which is\n");
858 context_switch = 0; // turned on via GranYield, checking events and time slice
861 DumpGranEvent(GR_SCHEDULE, t));
863 procStatus[CurrentProc] = Busy;
866 if (PendingFetches != END_BF_QUEUE) {
870 /* ToDo: phps merge with spark activation above */
871 /* check whether we have local work and send requests if we have none */
872 if (EMPTY_RUN_QUEUE()) { /* no runnable threads */
873 /* :-[ no local threads => look out for local sparks */
874 /* the spark pool for the current PE */
875 pool = &(MainRegTable.rSparks); // generalise to cap = &MainRegTable
876 if (advisory_thread_count < RtsFlags.ParFlags.maxThreads &&
877 pool->hd < pool->tl) {
879 * ToDo: add GC code check that we really have enough heap afterwards!!
881 * If we're here (no runnable threads) and we have pending
882 * sparks, we must have a space problem. Get enough space
883 * to turn one of those pending sparks into a
887 spark = findSpark(rtsFalse); /* get a spark */
888 if (spark != (rtsSpark) NULL) {
889 tso = activateSpark(spark); /* turn the spark into a thread */
890 IF_PAR_DEBUG(schedule,
891 belch("==== schedule: Created TSO %d (%p); %d threads active",
892 tso->id, tso, advisory_thread_count));
894 if (tso==END_TSO_QUEUE) { /* failed to activate spark->back to loop */
895 belch("==^^ failed to activate spark");
897 } /* otherwise fall through & pick-up new tso */
899 IF_PAR_DEBUG(verbose,
900 belch("==^^ no local sparks (spark pool contains only NFs: %d)",
901 spark_queue_len(pool)));
906 /* If we still have no work we need to send a FISH to get a spark
909 if (EMPTY_RUN_QUEUE()) {
910 /* =8-[ no local sparks => look for work on other PEs */
912 * We really have absolutely no work. Send out a fish
913 * (there may be some out there already), and wait for
914 * something to arrive. We clearly can't run any threads
915 * until a SCHEDULE or RESUME arrives, and so that's what
916 * we're hoping to see. (Of course, we still have to
917 * respond to other types of messages.)
919 TIME now = msTime() /*CURRENT_TIME*/;
920 IF_PAR_DEBUG(verbose,
921 belch("-- now=%ld", now));
922 IF_PAR_DEBUG(verbose,
923 if (outstandingFishes < RtsFlags.ParFlags.maxFishes &&
924 (last_fish_arrived_at!=0 &&
925 last_fish_arrived_at+RtsFlags.ParFlags.fishDelay > now)) {
926 belch("--$$ delaying FISH until %ld (last fish %ld, delay %ld, now %ld)",
927 last_fish_arrived_at+RtsFlags.ParFlags.fishDelay,
928 last_fish_arrived_at,
929 RtsFlags.ParFlags.fishDelay, now);
932 if (outstandingFishes < RtsFlags.ParFlags.maxFishes &&
933 (last_fish_arrived_at==0 ||
934 (last_fish_arrived_at+RtsFlags.ParFlags.fishDelay <= now))) {
935 /* outstandingFishes is set in sendFish, processFish;
936 avoid flooding system with fishes via delay */
938 sendFish(pe, mytid, NEW_FISH_AGE, NEW_FISH_HISTORY,
941 // Global statistics: count no. of fishes
942 if (RtsFlags.ParFlags.ParStats.Global &&
943 RtsFlags.GcFlags.giveStats > NO_GC_STATS) {
944 globalParStats.tot_fish_mess++;
948 receivedFinish = processMessages();
951 } else if (PacketsWaiting()) { /* Look for incoming messages */
952 receivedFinish = processMessages();
955 /* Now we are sure that we have some work available */
956 ASSERT(run_queue_hd != END_TSO_QUEUE);
958 /* Take a thread from the run queue, if we have work */
959 t = POP_RUN_QUEUE(); // take_off_run_queue(END_TSO_QUEUE);
960 IF_DEBUG(sanity,checkTSO(t));
962 /* ToDo: write something to the log-file
963 if (RTSflags.ParFlags.granSimStats && !sameThread)
964 DumpGranEvent(GR_SCHEDULE, RunnableThreadsHd);
968 /* the spark pool for the current PE */
969 pool = &(MainRegTable.rSparks); // generalise to cap = &MainRegTable
972 belch("--=^ %d threads, %d sparks on [%#x]",
973 run_queue_len(), spark_queue_len(pool), CURRENT_PROC));
976 if (0 && RtsFlags.ParFlags.ParStats.Full &&
977 t && LastTSO && t->id != LastTSO->id &&
978 LastTSO->why_blocked == NotBlocked &&
979 LastTSO->what_next != ThreadComplete) {
980 // if previously scheduled TSO not blocked we have to record the context switch
981 DumpVeryRawGranEvent(TimeOfLastYield, CURRENT_PROC, CURRENT_PROC,
982 GR_DESCHEDULE, LastTSO, (StgClosure *)NULL, 0, 0);
985 if (RtsFlags.ParFlags.ParStats.Full &&
986 (emitSchedule /* forced emit */ ||
987 (t && LastTSO && t->id != LastTSO->id))) {
989 we are running a different TSO, so write a schedule event to log file
990 NB: If we use fair scheduling we also have to write a deschedule
991 event for LastTSO; with unfair scheduling we know that the
992 previous tso has blocked whenever we switch to another tso, so
993 we don't need it in GUM for now
995 DumpRawGranEvent(CURRENT_PROC, CURRENT_PROC,
996 GR_SCHEDULE, t, (StgClosure *)NULL, 0, 0);
997 emitSchedule = rtsFalse;
1001 #else /* !GRAN && !PAR */
1003 /* grab a thread from the run queue */
1004 ASSERT(run_queue_hd != END_TSO_QUEUE);
1005 t = POP_RUN_QUEUE();
1006 // Sanity check the thread we're about to run. This can be
1007 // expensive if there is lots of thread switching going on...
1008 IF_DEBUG(sanity,checkTSO(t));
1011 cap->r.rCurrentTSO = t;
1013 /* context switches are now initiated by the timer signal, unless
1014 * the user specified "context switch as often as possible", with
1019 RtsFlags.ProfFlags.profileInterval == 0 ||
1021 (RtsFlags.ConcFlags.ctxtSwitchTicks == 0
1022 && (run_queue_hd != END_TSO_QUEUE
1023 || blocked_queue_hd != END_TSO_QUEUE
1024 || sleeping_queue != END_TSO_QUEUE)))
1029 RELEASE_LOCK(&sched_mutex);
1031 IF_DEBUG(scheduler, sched_belch("-->> Running TSO %ld (%p) %s ...",
1032 t->id, t, whatNext_strs[t->what_next]));
1035 startHeapProfTimer();
1038 /* +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ */
1039 /* Run the current thread
1041 switch (cap->r.rCurrentTSO->what_next) {
1043 case ThreadComplete:
1044 /* Thread already finished, return to scheduler. */
1045 ret = ThreadFinished;
1047 case ThreadEnterGHC:
1048 ret = StgRun((StgFunPtr) stg_enterStackTop, &cap->r);
1051 ret = StgRun((StgFunPtr) stg_returnToStackTop, &cap->r);
1053 case ThreadEnterInterp:
1054 ret = interpretBCO(cap);
1057 barf("schedule: invalid what_next field");
1059 /* +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ */
1061 /* Costs for the scheduler are assigned to CCS_SYSTEM */
1063 stopHeapProfTimer();
1067 ACQUIRE_LOCK(&sched_mutex);
1070 IF_DEBUG(scheduler,fprintf(stderr,"scheduler (task %ld): ", osThreadId()););
1071 #elif !defined(GRAN) && !defined(PAR)
1072 IF_DEBUG(scheduler,fprintf(stderr,"scheduler: "););
1074 t = cap->r.rCurrentTSO;
1077 /* HACK 675: if the last thread didn't yield, make sure to print a
1078 SCHEDULE event to the log file when StgRunning the next thread, even
1079 if it is the same one as before */
1081 TimeOfLastYield = CURRENT_TIME;
1087 IF_DEBUG(gran, DumpGranEvent(GR_DESCHEDULE, t));
1088 globalGranStats.tot_heapover++;
1090 globalParStats.tot_heapover++;
1093 // did the task ask for a large block?
1094 if (cap->r.rHpAlloc > BLOCK_SIZE_W) {
1095 // if so, get one and push it on the front of the nursery.
1099 blocks = (nat)BLOCK_ROUND_UP(cap->r.rHpAlloc * sizeof(W_)) / BLOCK_SIZE;
1101 IF_DEBUG(scheduler,belch("--<< thread %ld (%p; %s) stopped: requesting a large block (size %d)",
1103 whatNext_strs[t->what_next], blocks));
1105 // don't do this if it would push us over the
1106 // alloc_blocks_lim limit; we'll GC first.
1107 if (alloc_blocks + blocks < alloc_blocks_lim) {
1109 alloc_blocks += blocks;
1110 bd = allocGroup( blocks );
1112 // link the new group into the list
1113 bd->link = cap->r.rCurrentNursery;
1114 bd->u.back = cap->r.rCurrentNursery->u.back;
1115 if (cap->r.rCurrentNursery->u.back != NULL) {
1116 cap->r.rCurrentNursery->u.back->link = bd;
1118 ASSERT(g0s0->blocks == cap->r.rCurrentNursery &&
1119 g0s0->blocks == cap->r.rNursery);
1120 cap->r.rNursery = g0s0->blocks = bd;
1122 cap->r.rCurrentNursery->u.back = bd;
1124 // initialise it as a nursery block
1128 bd->free = bd->start;
1130 // don't forget to update the block count in g0s0.
1131 g0s0->n_blocks += blocks;
1132 ASSERT(countBlocks(g0s0->blocks) == g0s0->n_blocks);
1134 // now update the nursery to point to the new block
1135 cap->r.rCurrentNursery = bd;
1137 // we might be unlucky and have another thread get on the
1138 // run queue before us and steal the large block, but in that
1139 // case the thread will just end up requesting another large
1141 PUSH_ON_RUN_QUEUE(t);
1146 /* make all the running tasks block on a condition variable,
1147 * maybe set context_switch and wait till they all pile in,
1148 * then have them wait on a GC condition variable.
1150 IF_DEBUG(scheduler,belch("--<< thread %ld (%p; %s) stopped: HeapOverflow",
1151 t->id, t, whatNext_strs[t->what_next]));
1154 ASSERT(!is_on_queue(t,CurrentProc));
1156 /* Currently we emit a DESCHEDULE event before GC in GUM.
1157 ToDo: either add separate event to distinguish SYSTEM time from rest
1158 or just nuke this DESCHEDULE (and the following SCHEDULE) */
1159 if (0 && RtsFlags.ParFlags.ParStats.Full) {
1160 DumpRawGranEvent(CURRENT_PROC, CURRENT_PROC,
1161 GR_DESCHEDULE, t, (StgClosure *)NULL, 0, 0);
1162 emitSchedule = rtsTrue;
1166 ready_to_gc = rtsTrue;
1167 context_switch = 1; /* stop other threads ASAP */
1168 PUSH_ON_RUN_QUEUE(t);
1169 /* actual GC is done at the end of the while loop */
1175 DumpGranEvent(GR_DESCHEDULE, t));
1176 globalGranStats.tot_stackover++;
1179 // DumpGranEvent(GR_DESCHEDULE, t);
1180 globalParStats.tot_stackover++;
1182 IF_DEBUG(scheduler,belch("--<< thread %ld (%p; %s) stopped, StackOverflow",
1183 t->id, t, whatNext_strs[t->what_next]));
1184 /* just adjust the stack for this thread, then pop it back
1190 /* enlarge the stack */
1191 StgTSO *new_t = threadStackOverflow(t);
1193 /* This TSO has moved, so update any pointers to it from the
1194 * main thread stack. It better not be on any other queues...
1195 * (it shouldn't be).
1197 for (m = main_threads; m != NULL; m = m->link) {
1202 threadPaused(new_t);
1203 PUSH_ON_RUN_QUEUE(new_t);
1207 case ThreadYielding:
1210 DumpGranEvent(GR_DESCHEDULE, t));
1211 globalGranStats.tot_yields++;
1214 // DumpGranEvent(GR_DESCHEDULE, t);
1215 globalParStats.tot_yields++;
1217 /* put the thread back on the run queue. Then, if we're ready to
1218 * GC, check whether this is the last task to stop. If so, wake
1219 * up the GC thread. getThread will block during a GC until the
1223 if (t->what_next == ThreadEnterInterp) {
1224 /* ToDo: or maybe a timer expired when we were in Hugs?
1225 * or maybe someone hit ctrl-C
1227 belch("--<< thread %ld (%p; %s) stopped to switch to Hugs",
1228 t->id, t, whatNext_strs[t->what_next]);
1230 belch("--<< thread %ld (%p; %s) stopped, yielding",
1231 t->id, t, whatNext_strs[t->what_next]);
1238 //belch("&& Doing sanity check on yielding TSO %ld.", t->id);
1240 ASSERT(t->link == END_TSO_QUEUE);
1242 ASSERT(!is_on_queue(t,CurrentProc));
1245 //belch("&& Doing sanity check on all ThreadQueues (and their TSOs).");
1246 checkThreadQsSanity(rtsTrue));
1249 if (RtsFlags.ParFlags.doFairScheduling) {
1250 /* this does round-robin scheduling; good for concurrency */
1251 APPEND_TO_RUN_QUEUE(t);
1253 /* this does unfair scheduling; good for parallelism */
1254 PUSH_ON_RUN_QUEUE(t);
1257 /* this does round-robin scheduling; good for concurrency */
1258 APPEND_TO_RUN_QUEUE(t);
1261 /* add a ContinueThread event to actually process the thread */
1262 new_event(CurrentProc, CurrentProc, CurrentTime[CurrentProc],
1264 t, (StgClosure*)NULL, (rtsSpark*)NULL);
1266 belch("GRAN: eventq and runnableq after adding yielded thread to queue again:");
1275 belch("--<< thread %ld (%p; %s) stopped, blocking on node %p [PE %d] with BQ: ",
1276 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)));
1277 if (t->block_info.closure!=(StgClosure*)NULL) print_bq(t->block_info.closure));
1279 // ??? needed; should emit block before
1281 DumpGranEvent(GR_DESCHEDULE, t));
1282 prune_eventq(t, (StgClosure *)NULL); // prune ContinueThreads for t
1285 ASSERT(procStatus[CurrentProc]==Busy ||
1286 ((procStatus[CurrentProc]==Fetching) &&
1287 (t->block_info.closure!=(StgClosure*)NULL)));
1288 if (run_queue_hds[CurrentProc] == END_TSO_QUEUE &&
1289 !(!RtsFlags.GranFlags.DoAsyncFetch &&
1290 procStatus[CurrentProc]==Fetching))
1291 procStatus[CurrentProc] = Idle;
1295 belch("--<< thread %ld (%p; %s) stopped, blocking on node %p with BQ: ",
1296 t->id, t, whatNext_strs[t->what_next], t->block_info.closure));
1299 if (t->block_info.closure!=(StgClosure*)NULL)
1300 print_bq(t->block_info.closure));
1302 /* Send a fetch (if BlockedOnGA) and dump event to log file */
1305 /* whatever we schedule next, we must log that schedule */
1306 emitSchedule = rtsTrue;
1309 /* don't need to do anything. Either the thread is blocked on
1310 * I/O, in which case we'll have called addToBlockedQueue
1311 * previously, or it's blocked on an MVar or Blackhole, in which
1312 * case it'll be on the relevant queue already.
1315 fprintf(stderr, "--<< thread %d (%p) stopped: ", t->id, t);
1316 printThreadBlockage(t);
1317 fprintf(stderr, "\n"));
1319 /* Only for dumping event to log file
1320 ToDo: do I need this in GranSim, too?
1327 case ThreadFinished:
1328 /* Need to check whether this was a main thread, and if so, signal
1329 * the task that started it with the return value. If we have no
1330 * more main threads, we probably need to stop all the tasks until
1333 /* We also end up here if the thread kills itself with an
1334 * uncaught exception, see Exception.hc.
1336 IF_DEBUG(scheduler,belch("--++ thread %d (%p) finished", t->id, t));
1338 endThread(t, CurrentProc); // clean-up the thread
1340 /* For now all are advisory -- HWL */
1341 //if(t->priority==AdvisoryPriority) ??
1342 advisory_thread_count--;
1345 if(t->dist.priority==RevalPriority)
1349 if (RtsFlags.ParFlags.ParStats.Full &&
1350 !RtsFlags.ParFlags.ParStats.Suppressed)
1351 DumpEndEvent(CURRENT_PROC, t, rtsFalse /* not mandatory */);
1356 barf("schedule: invalid thread return code %d", (int)ret);
1360 if (RtsFlags.ProfFlags.profileInterval==0 || performHeapProfile) {
1361 GarbageCollect(GetRoots, rtsTrue);
1363 performHeapProfile = rtsFalse;
1364 ready_to_gc = rtsFalse; // we already GC'd
1370 && allFreeCapabilities()
1373 /* everybody back, start the GC.
1374 * Could do it in this thread, or signal a condition var
1375 * to do it in another thread. Either way, we need to
1376 * broadcast on gc_pending_cond afterward.
1378 #if defined(RTS_SUPPORTS_THREADS)
1379 IF_DEBUG(scheduler,sched_belch("doing GC"));
1381 GarbageCollect(GetRoots,rtsFalse);
1382 ready_to_gc = rtsFalse;
1384 broadcastCondition(&gc_pending_cond);
1387 /* add a ContinueThread event to continue execution of current thread */
1388 new_event(CurrentProc, CurrentProc, CurrentTime[CurrentProc],
1390 t, (StgClosure*)NULL, (rtsSpark*)NULL);
1392 fprintf(stderr, "GRAN: eventq and runnableq after Garbage collection:\n");
1400 IF_GRAN_DEBUG(unused,
1401 print_eventq(EventHd));
1403 event = get_next_event();
1406 /* ToDo: wait for next message to arrive rather than busy wait */
1409 } /* end of while(1) */
1411 IF_PAR_DEBUG(verbose,
1412 belch("== Leaving schedule() after having received Finish"));
1415 /* ---------------------------------------------------------------------------
1416 * Singleton fork(). Do not copy any running threads.
1417 * ------------------------------------------------------------------------- */
1419 StgInt forkProcess(StgTSO* tso) {
1421 #ifndef mingw32_TARGET_OS
1425 IF_DEBUG(scheduler,sched_belch("forking!"));
1428 if (pid) { /* parent */
1430 /* just return the pid */
1432 } else { /* child */
1433 /* wipe all other threads */
1435 tso->link = END_TSO_QUEUE;
1437 /* DO NOT TOUCH THE QUEUES directly because most of the code around
1438 us is picky about finding the threat still in its queue when
1439 handling the deleteThread() */
1441 for (t = all_threads; t != END_TSO_QUEUE; t = next) {
1443 if (t->id != tso->id) {
1450 barf("forkProcess#: primop not implemented for mingw32, sorry! (%u)\n", tso->id);
1451 /* pointlessly printing out the TSOs 'id' to avoid CC unused warning. */
1453 #endif /* mingw32 */
1456 /* ---------------------------------------------------------------------------
1457 * deleteAllThreads(): kill all the live threads.
1459 * This is used when we catch a user interrupt (^C), before performing
1460 * any necessary cleanups and running finalizers.
1462 * Locks: sched_mutex held.
1463 * ------------------------------------------------------------------------- */
1465 void deleteAllThreads ( void )
1468 IF_DEBUG(scheduler,sched_belch("deleting all threads"));
1469 for (t = all_threads; t != END_TSO_QUEUE; t = next) {
1470 next = t->global_link;
1473 run_queue_hd = run_queue_tl = END_TSO_QUEUE;
1474 blocked_queue_hd = blocked_queue_tl = END_TSO_QUEUE;
1475 sleeping_queue = END_TSO_QUEUE;
1478 /* startThread and insertThread are now in GranSim.c -- HWL */
1481 //@node Suspend and Resume, Run queue code, Main scheduling loop, Main scheduling code
1482 //@subsection Suspend and Resume
1484 /* ---------------------------------------------------------------------------
1485 * Suspending & resuming Haskell threads.
1487 * When making a "safe" call to C (aka _ccall_GC), the task gives back
1488 * its capability before calling the C function. This allows another
1489 * task to pick up the capability and carry on running Haskell
1490 * threads. It also means that if the C call blocks, it won't lock
1493 * The Haskell thread making the C call is put to sleep for the
1494 * duration of the call, on the susepended_ccalling_threads queue. We
1495 * give out a token to the task, which it can use to resume the thread
1496 * on return from the C function.
1497 * ------------------------------------------------------------------------- */
1500 suspendThread( StgRegTable *reg,
1502 #if !defined(RTS_SUPPORTS_THREADS) && !defined(DEBUG)
1510 /* assume that *reg is a pointer to the StgRegTable part
1513 cap = (Capability *)((void *)reg - sizeof(StgFunTable));
1515 ACQUIRE_LOCK(&sched_mutex);
1518 sched_belch("thread %d did a _ccall_gc (is_concurrent: %d)", cap->r.rCurrentTSO->id,concCall));
1520 threadPaused(cap->r.rCurrentTSO);
1521 cap->r.rCurrentTSO->link = suspended_ccalling_threads;
1522 suspended_ccalling_threads = cap->r.rCurrentTSO;
1524 #if defined(RTS_SUPPORTS_THREADS)
1525 cap->r.rCurrentTSO->why_blocked = BlockedOnCCall;
1528 /* Use the thread ID as the token; it should be unique */
1529 tok = cap->r.rCurrentTSO->id;
1531 /* Hand back capability */
1532 releaseCapability(cap);
1534 #if defined(RTS_SUPPORTS_THREADS)
1535 /* Preparing to leave the RTS, so ensure there's a native thread/task
1536 waiting to take over.
1538 ToDo: optimise this and only create a new task if there's a need
1539 for one (i.e., if there's only one Concurrent Haskell thread alive,
1540 there's no need to create a new task).
1542 IF_DEBUG(scheduler, sched_belch("worker thread (%d): leaving RTS", tok));
1544 startTask(taskStart);
1548 /* Other threads _might_ be available for execution; signal this */
1550 RELEASE_LOCK(&sched_mutex);
1555 resumeThread( StgInt tok,
1557 #if !defined(RTS_SUPPORTS_THREADS)
1562 StgTSO *tso, **prev;
1565 #if defined(RTS_SUPPORTS_THREADS)
1566 /* Wait for permission to re-enter the RTS with the result. */
1568 ACQUIRE_LOCK(&sched_mutex);
1569 grabReturnCapability(&sched_mutex, &cap);
1571 grabCapability(&cap);
1574 grabCapability(&cap);
1577 /* Remove the thread off of the suspended list */
1578 prev = &suspended_ccalling_threads;
1579 for (tso = suspended_ccalling_threads;
1580 tso != END_TSO_QUEUE;
1581 prev = &tso->link, tso = tso->link) {
1582 if (tso->id == (StgThreadID)tok) {
1587 if (tso == END_TSO_QUEUE) {
1588 barf("resumeThread: thread not found");
1590 tso->link = END_TSO_QUEUE;
1591 /* Reset blocking status */
1592 tso->why_blocked = NotBlocked;
1594 cap->r.rCurrentTSO = tso;
1595 RELEASE_LOCK(&sched_mutex);
1600 /* ---------------------------------------------------------------------------
1602 * ------------------------------------------------------------------------ */
1603 static void unblockThread(StgTSO *tso);
1605 /* ---------------------------------------------------------------------------
1606 * Comparing Thread ids.
1608 * This is used from STG land in the implementation of the
1609 * instances of Eq/Ord for ThreadIds.
1610 * ------------------------------------------------------------------------ */
1612 int cmp_thread(const StgTSO *tso1, const StgTSO *tso2)
1614 StgThreadID id1 = tso1->id;
1615 StgThreadID id2 = tso2->id;
1617 if (id1 < id2) return (-1);
1618 if (id1 > id2) return 1;
1622 /* ---------------------------------------------------------------------------
1623 * Fetching the ThreadID from an StgTSO.
1625 * This is used in the implementation of Show for ThreadIds.
1626 * ------------------------------------------------------------------------ */
1627 int rts_getThreadId(const StgTSO *tso)
1633 void labelThread(StgTSO *tso, char *label)
1638 /* Caveat: Once set, you can only set the thread name to "" */
1639 len = strlen(label)+1;
1640 buf = realloc(tso->label,len);
1642 fprintf(stderr,"insufficient memory for labelThread!\n");
1646 strncpy(buf,label,len);
1651 /* ---------------------------------------------------------------------------
1652 Create a new thread.
1654 The new thread starts with the given stack size. Before the
1655 scheduler can run, however, this thread needs to have a closure
1656 (and possibly some arguments) pushed on its stack. See
1657 pushClosure() in Schedule.h.
1659 createGenThread() and createIOThread() (in SchedAPI.h) are
1660 convenient packaged versions of this function.
1662 currently pri (priority) is only used in a GRAN setup -- HWL
1663 ------------------------------------------------------------------------ */
1664 //@cindex createThread
1666 /* currently pri (priority) is only used in a GRAN setup -- HWL */
1668 createThread(nat stack_size, StgInt pri)
1670 return createThread_(stack_size, rtsFalse, pri);
1674 createThread_(nat size, rtsBool have_lock, StgInt pri)
1678 createThread(nat stack_size)
1680 return createThread_(stack_size, rtsFalse);
1684 createThread_(nat size, rtsBool have_lock)
1691 /* First check whether we should create a thread at all */
1693 /* check that no more than RtsFlags.ParFlags.maxThreads threads are created */
1694 if (advisory_thread_count >= RtsFlags.ParFlags.maxThreads) {
1696 belch("{createThread}Daq ghuH: refusing to create another thread; no more than %d threads allowed (currently %d)",
1697 RtsFlags.ParFlags.maxThreads, advisory_thread_count);
1698 return END_TSO_QUEUE;
1704 ASSERT(!RtsFlags.GranFlags.Light || CurrentProc==0);
1707 // ToDo: check whether size = stack_size - TSO_STRUCT_SIZEW
1709 /* catch ridiculously small stack sizes */
1710 if (size < MIN_STACK_WORDS + TSO_STRUCT_SIZEW) {
1711 size = MIN_STACK_WORDS + TSO_STRUCT_SIZEW;
1714 stack_size = size - TSO_STRUCT_SIZEW;
1716 tso = (StgTSO *)allocate(size);
1717 TICK_ALLOC_TSO(stack_size, 0);
1719 SET_HDR(tso, &stg_TSO_info, CCS_SYSTEM);
1721 SET_GRAN_HDR(tso, ThisPE);
1723 tso->what_next = ThreadEnterGHC;
1729 /* tso->id needs to be unique. For now we use a heavyweight mutex to
1730 * protect the increment operation on next_thread_id.
1731 * In future, we could use an atomic increment instead.
1734 if (!have_lock) { ACQUIRE_LOCK(&sched_mutex); }
1736 tso->id = next_thread_id++;
1738 if (!have_lock) { RELEASE_LOCK(&sched_mutex); }
1741 tso->why_blocked = NotBlocked;
1742 tso->blocked_exceptions = NULL;
1744 tso->stack_size = stack_size;
1745 tso->max_stack_size = round_to_mblocks(RtsFlags.GcFlags.maxStkSize)
1747 tso->sp = (P_)&(tso->stack) + stack_size;
1750 tso->prof.CCCS = CCS_MAIN;
1753 /* put a stop frame on the stack */
1754 tso->sp -= sizeofW(StgStopFrame);
1755 SET_HDR((StgClosure*)tso->sp,(StgInfoTable *)&stg_stop_thread_info,CCS_SYSTEM);
1756 tso->su = (StgUpdateFrame*)tso->sp;
1760 tso->link = END_TSO_QUEUE;
1761 /* uses more flexible routine in GranSim */
1762 insertThread(tso, CurrentProc);
1764 /* In a non-GranSim setup the pushing of a TSO onto the runq is separated
1770 if (RtsFlags.GranFlags.GranSimStats.Full)
1771 DumpGranEvent(GR_START,tso);
1773 if (RtsFlags.ParFlags.ParStats.Full)
1774 DumpGranEvent(GR_STARTQ,tso);
1775 /* HACk to avoid SCHEDULE
1779 /* Link the new thread on the global thread list.
1781 tso->global_link = all_threads;
1785 tso->dist.priority = MandatoryPriority; //by default that is...
1789 tso->gran.pri = pri;
1791 tso->gran.magic = TSO_MAGIC; // debugging only
1793 tso->gran.sparkname = 0;
1794 tso->gran.startedat = CURRENT_TIME;
1795 tso->gran.exported = 0;
1796 tso->gran.basicblocks = 0;
1797 tso->gran.allocs = 0;
1798 tso->gran.exectime = 0;
1799 tso->gran.fetchtime = 0;
1800 tso->gran.fetchcount = 0;
1801 tso->gran.blocktime = 0;
1802 tso->gran.blockcount = 0;
1803 tso->gran.blockedat = 0;
1804 tso->gran.globalsparks = 0;
1805 tso->gran.localsparks = 0;
1806 if (RtsFlags.GranFlags.Light)
1807 tso->gran.clock = Now; /* local clock */
1809 tso->gran.clock = 0;
1811 IF_DEBUG(gran,printTSO(tso));
1814 tso->par.magic = TSO_MAGIC; // debugging only
1816 tso->par.sparkname = 0;
1817 tso->par.startedat = CURRENT_TIME;
1818 tso->par.exported = 0;
1819 tso->par.basicblocks = 0;
1820 tso->par.allocs = 0;
1821 tso->par.exectime = 0;
1822 tso->par.fetchtime = 0;
1823 tso->par.fetchcount = 0;
1824 tso->par.blocktime = 0;
1825 tso->par.blockcount = 0;
1826 tso->par.blockedat = 0;
1827 tso->par.globalsparks = 0;
1828 tso->par.localsparks = 0;
1832 globalGranStats.tot_threads_created++;
1833 globalGranStats.threads_created_on_PE[CurrentProc]++;
1834 globalGranStats.tot_sq_len += spark_queue_len(CurrentProc);
1835 globalGranStats.tot_sq_probes++;
1837 // collect parallel global statistics (currently done together with GC stats)
1838 if (RtsFlags.ParFlags.ParStats.Global &&
1839 RtsFlags.GcFlags.giveStats > NO_GC_STATS) {
1840 //fprintf(stderr, "Creating thread %d @ %11.2f\n", tso->id, usertime());
1841 globalParStats.tot_threads_created++;
1847 belch("==__ schedule: Created TSO %d (%p);",
1848 CurrentProc, tso, tso->id));
1850 IF_PAR_DEBUG(verbose,
1851 belch("==__ schedule: Created TSO %d (%p); %d threads active",
1852 tso->id, tso, advisory_thread_count));
1854 IF_DEBUG(scheduler,sched_belch("created thread %ld, stack size = %lx words",
1855 tso->id, tso->stack_size));
1862 all parallel thread creation calls should fall through the following routine.
1865 createSparkThread(rtsSpark spark)
1867 ASSERT(spark != (rtsSpark)NULL);
1868 if (advisory_thread_count >= RtsFlags.ParFlags.maxThreads)
1870 barf("{createSparkThread}Daq ghuH: refusing to create another thread; no more than %d threads allowed (currently %d)",
1871 RtsFlags.ParFlags.maxThreads, advisory_thread_count);
1872 return END_TSO_QUEUE;
1876 tso = createThread_(RtsFlags.GcFlags.initialStkSize, rtsTrue);
1877 if (tso==END_TSO_QUEUE)
1878 barf("createSparkThread: Cannot create TSO");
1880 tso->priority = AdvisoryPriority;
1882 pushClosure(tso,spark);
1883 PUSH_ON_RUN_QUEUE(tso);
1884 advisory_thread_count++;
1891 Turn a spark into a thread.
1892 ToDo: fix for SMP (needs to acquire SCHED_MUTEX!)
1895 //@cindex activateSpark
1897 activateSpark (rtsSpark spark)
1901 tso = createSparkThread(spark);
1902 if (RtsFlags.ParFlags.ParStats.Full) {
1903 //ASSERT(run_queue_hd == END_TSO_QUEUE); // I think ...
1904 IF_PAR_DEBUG(verbose,
1905 belch("==^^ activateSpark: turning spark of closure %p (%s) into a thread",
1906 (StgClosure *)spark, info_type((StgClosure *)spark)));
1908 // ToDo: fwd info on local/global spark to thread -- HWL
1909 // tso->gran.exported = spark->exported;
1910 // tso->gran.locked = !spark->global;
1911 // tso->gran.sparkname = spark->name;
1917 /* ---------------------------------------------------------------------------
1920 * scheduleThread puts a thread on the head of the runnable queue.
1921 * This will usually be done immediately after a thread is created.
1922 * The caller of scheduleThread must create the thread using e.g.
1923 * createThread and push an appropriate closure
1924 * on this thread's stack before the scheduler is invoked.
1925 * ------------------------------------------------------------------------ */
1927 static void scheduleThread_ (StgTSO* tso, rtsBool createTask);
1930 scheduleThread_(StgTSO *tso
1931 , rtsBool createTask
1932 #if !defined(THREADED_RTS)
1937 ACQUIRE_LOCK(&sched_mutex);
1939 /* Put the new thread on the head of the runnable queue. The caller
1940 * better push an appropriate closure on this thread's stack
1941 * beforehand. In the SMP case, the thread may start running as
1942 * soon as we release the scheduler lock below.
1944 PUSH_ON_RUN_QUEUE(tso);
1945 #if defined(THREADED_RTS)
1946 /* If main() is scheduling a thread, don't bother creating a
1950 startTask(taskStart);
1956 IF_DEBUG(scheduler,printTSO(tso));
1958 RELEASE_LOCK(&sched_mutex);
1961 void scheduleThread(StgTSO* tso)
1963 return scheduleThread_(tso, rtsFalse);
1966 void scheduleExtThread(StgTSO* tso)
1968 return scheduleThread_(tso, rtsTrue);
1971 /* ---------------------------------------------------------------------------
1974 * Initialise the scheduler. This resets all the queues - if the
1975 * queues contained any threads, they'll be garbage collected at the
1978 * ------------------------------------------------------------------------ */
1982 term_handler(int sig STG_UNUSED)
1985 ACQUIRE_LOCK(&term_mutex);
1987 RELEASE_LOCK(&term_mutex);
1998 for (i=0; i<=MAX_PROC; i++) {
1999 run_queue_hds[i] = END_TSO_QUEUE;
2000 run_queue_tls[i] = END_TSO_QUEUE;
2001 blocked_queue_hds[i] = END_TSO_QUEUE;
2002 blocked_queue_tls[i] = END_TSO_QUEUE;
2003 ccalling_threadss[i] = END_TSO_QUEUE;
2004 sleeping_queue = END_TSO_QUEUE;
2007 run_queue_hd = END_TSO_QUEUE;
2008 run_queue_tl = END_TSO_QUEUE;
2009 blocked_queue_hd = END_TSO_QUEUE;
2010 blocked_queue_tl = END_TSO_QUEUE;
2011 sleeping_queue = END_TSO_QUEUE;
2014 suspended_ccalling_threads = END_TSO_QUEUE;
2016 main_threads = NULL;
2017 all_threads = END_TSO_QUEUE;
2022 RtsFlags.ConcFlags.ctxtSwitchTicks =
2023 RtsFlags.ConcFlags.ctxtSwitchTime / TICK_MILLISECS;
2025 #if defined(RTS_SUPPORTS_THREADS)
2026 /* Initialise the mutex and condition variables used by
2028 initMutex(&sched_mutex);
2029 initMutex(&term_mutex);
2031 initCondition(&thread_ready_cond);
2035 initCondition(&gc_pending_cond);
2038 #if defined(RTS_SUPPORTS_THREADS)
2039 ACQUIRE_LOCK(&sched_mutex);
2042 /* Install the SIGHUP handler */
2045 struct sigaction action,oact;
2047 action.sa_handler = term_handler;
2048 sigemptyset(&action.sa_mask);
2049 action.sa_flags = 0;
2050 if (sigaction(SIGTERM, &action, &oact) != 0) {
2051 barf("can't install TERM handler");
2056 /* A capability holds the state a native thread needs in
2057 * order to execute STG code. At least one capability is
2058 * floating around (only SMP builds have more than one).
2062 #if defined(RTS_SUPPORTS_THREADS)
2063 /* start our haskell execution tasks */
2065 startTaskManager(RtsFlags.ParFlags.nNodes, taskStart);
2067 startTaskManager(0,taskStart);
2071 #if /* defined(SMP) ||*/ defined(PAR)
2075 #if defined(RTS_SUPPORTS_THREADS)
2076 RELEASE_LOCK(&sched_mutex);
2082 exitScheduler( void )
2084 #if defined(RTS_SUPPORTS_THREADS)
2089 /* -----------------------------------------------------------------------------
2090 Managing the per-task allocation areas.
2092 Each capability comes with an allocation area. These are
2093 fixed-length block lists into which allocation can be done.
2095 ToDo: no support for two-space collection at the moment???
2096 -------------------------------------------------------------------------- */
2098 /* -----------------------------------------------------------------------------
2099 * waitThread is the external interface for running a new computation
2100 * and waiting for the result.
2102 * In the non-SMP case, we create a new main thread, push it on the
2103 * main-thread stack, and invoke the scheduler to run it. The
2104 * scheduler will return when the top main thread on the stack has
2105 * completed or died, and fill in the necessary fields of the
2106 * main_thread structure.
2108 * In the SMP case, we create a main thread as before, but we then
2109 * create a new condition variable and sleep on it. When our new
2110 * main thread has completed, we'll be woken up and the status/result
2111 * will be in the main_thread struct.
2112 * -------------------------------------------------------------------------- */
2115 howManyThreadsAvail ( void )
2119 for (q = run_queue_hd; q != END_TSO_QUEUE; q = q->link)
2121 for (q = blocked_queue_hd; q != END_TSO_QUEUE; q = q->link)
2123 for (q = sleeping_queue; q != END_TSO_QUEUE; q = q->link)
2129 finishAllThreads ( void )
2132 while (run_queue_hd != END_TSO_QUEUE) {
2133 waitThread ( run_queue_hd, NULL);
2135 while (blocked_queue_hd != END_TSO_QUEUE) {
2136 waitThread ( blocked_queue_hd, NULL);
2138 while (sleeping_queue != END_TSO_QUEUE) {
2139 waitThread ( blocked_queue_hd, NULL);
2142 (blocked_queue_hd != END_TSO_QUEUE ||
2143 run_queue_hd != END_TSO_QUEUE ||
2144 sleeping_queue != END_TSO_QUEUE);
2148 waitThread(StgTSO *tso, /*out*/StgClosure **ret)
2150 IF_DEBUG(scheduler, sched_belch("== scheduler: waiting for thread (%d)\n", tso->id));
2151 #if defined(THREADED_RTS)
2152 return waitThread_(tso,ret, rtsFalse);
2154 return waitThread_(tso,ret);
2159 waitThread_(StgTSO *tso,
2160 /*out*/StgClosure **ret
2161 #if defined(THREADED_RTS)
2162 , rtsBool blockWaiting
2167 SchedulerStatus stat;
2169 ACQUIRE_LOCK(&sched_mutex);
2170 IF_DEBUG(scheduler, sched_belch("== scheduler: waiting for thread (%d)\n", tso->id));
2172 m = stgMallocBytes(sizeof(StgMainThread), "waitThread");
2177 #if defined(RTS_SUPPORTS_THREADS)
2178 initCondition(&m->wakeup);
2181 m->link = main_threads;
2184 IF_DEBUG(scheduler, sched_belch("== scheduler: new main thread (%d)\n", m->tso->id));
2186 #if defined(RTS_SUPPORTS_THREADS)
2188 # if defined(THREADED_RTS)
2189 if (!blockWaiting) {
2190 /* In the threaded case, the OS thread that called main()
2191 * gets to enter the RTS directly without going via another
2194 RELEASE_LOCK(&sched_mutex);
2196 ASSERT(m->stat != NoStatus);
2201 waitCondition(&m->wakeup, &sched_mutex);
2202 } while (m->stat == NoStatus);
2205 /* GranSim specific init */
2206 CurrentTSO = m->tso; // the TSO to run
2207 procStatus[MainProc] = Busy; // status of main PE
2208 CurrentProc = MainProc; // PE to run it on
2212 RELEASE_LOCK(&sched_mutex);
2214 ASSERT(m->stat != NoStatus);
2219 #if defined(RTS_SUPPORTS_THREADS)
2220 closeCondition(&m->wakeup);
2223 IF_DEBUG(scheduler, fprintf(stderr, "== scheduler: main thread (%d) finished\n",
2227 #if defined(THREADED_RTS)
2230 RELEASE_LOCK(&sched_mutex);
2235 //@node Run queue code, Garbage Collextion Routines, Suspend and Resume, Main scheduling code
2236 //@subsection Run queue code
2240 NB: In GranSim we have many run queues; run_queue_hd is actually a macro
2241 unfolding to run_queue_hds[CurrentProc], thus CurrentProc is an
2242 implicit global variable that has to be correct when calling these
2246 /* Put the new thread on the head of the runnable queue.
2247 * The caller of createThread better push an appropriate closure
2248 * on this thread's stack before the scheduler is invoked.
2250 static /* inline */ void
2251 add_to_run_queue(tso)
2254 ASSERT(tso!=run_queue_hd && tso!=run_queue_tl);
2255 tso->link = run_queue_hd;
2257 if (run_queue_tl == END_TSO_QUEUE) {
2262 /* Put the new thread at the end of the runnable queue. */
2263 static /* inline */ void
2264 push_on_run_queue(tso)
2267 ASSERT(get_itbl((StgClosure *)tso)->type == TSO);
2268 ASSERT(run_queue_hd!=NULL && run_queue_tl!=NULL);
2269 ASSERT(tso!=run_queue_hd && tso!=run_queue_tl);
2270 if (run_queue_hd == END_TSO_QUEUE) {
2273 run_queue_tl->link = tso;
2279 Should be inlined because it's used very often in schedule. The tso
2280 argument is actually only needed in GranSim, where we want to have the
2281 possibility to schedule *any* TSO on the run queue, irrespective of the
2282 actual ordering. Therefore, if tso is not the nil TSO then we traverse
2283 the run queue and dequeue the tso, adjusting the links in the queue.
2285 //@cindex take_off_run_queue
2286 static /* inline */ StgTSO*
2287 take_off_run_queue(StgTSO *tso) {
2291 qetlaHbogh Qu' ngaSbogh ghomDaQ {tso} yIteq!
2293 if tso is specified, unlink that tso from the run_queue (doesn't have
2294 to be at the beginning of the queue); GranSim only
2296 if (tso!=END_TSO_QUEUE) {
2297 /* find tso in queue */
2298 for (t=run_queue_hd, prev=END_TSO_QUEUE;
2299 t!=END_TSO_QUEUE && t!=tso;
2303 /* now actually dequeue the tso */
2304 if (prev!=END_TSO_QUEUE) {
2305 ASSERT(run_queue_hd!=t);
2306 prev->link = t->link;
2308 /* t is at beginning of thread queue */
2309 ASSERT(run_queue_hd==t);
2310 run_queue_hd = t->link;
2312 /* t is at end of thread queue */
2313 if (t->link==END_TSO_QUEUE) {
2314 ASSERT(t==run_queue_tl);
2315 run_queue_tl = prev;
2317 ASSERT(run_queue_tl!=t);
2319 t->link = END_TSO_QUEUE;
2321 /* take tso from the beginning of the queue; std concurrent code */
2323 if (t != END_TSO_QUEUE) {
2324 run_queue_hd = t->link;
2325 t->link = END_TSO_QUEUE;
2326 if (run_queue_hd == END_TSO_QUEUE) {
2327 run_queue_tl = END_TSO_QUEUE;
2336 //@node Garbage Collextion Routines, Blocking Queue Routines, Run queue code, Main scheduling code
2337 //@subsection Garbage Collextion Routines
2339 /* ---------------------------------------------------------------------------
2340 Where are the roots that we know about?
2342 - all the threads on the runnable queue
2343 - all the threads on the blocked queue
2344 - all the threads on the sleeping queue
2345 - all the thread currently executing a _ccall_GC
2346 - all the "main threads"
2348 ------------------------------------------------------------------------ */
2350 /* This has to be protected either by the scheduler monitor, or by the
2351 garbage collection monitor (probably the latter).
2356 GetRoots(evac_fn evac)
2361 for (i=0; i<=RtsFlags.GranFlags.proc; i++) {
2362 if ((run_queue_hds[i] != END_TSO_QUEUE) && ((run_queue_hds[i] != NULL)))
2363 evac((StgClosure **)&run_queue_hds[i]);
2364 if ((run_queue_tls[i] != END_TSO_QUEUE) && ((run_queue_tls[i] != NULL)))
2365 evac((StgClosure **)&run_queue_tls[i]);
2367 if ((blocked_queue_hds[i] != END_TSO_QUEUE) && ((blocked_queue_hds[i] != NULL)))
2368 evac((StgClosure **)&blocked_queue_hds[i]);
2369 if ((blocked_queue_tls[i] != END_TSO_QUEUE) && ((blocked_queue_tls[i] != NULL)))
2370 evac((StgClosure **)&blocked_queue_tls[i]);
2371 if ((ccalling_threadss[i] != END_TSO_QUEUE) && ((ccalling_threadss[i] != NULL)))
2372 evac((StgClosure **)&ccalling_threads[i]);
2379 if (run_queue_hd != END_TSO_QUEUE) {
2380 ASSERT(run_queue_tl != END_TSO_QUEUE);
2381 evac((StgClosure **)&run_queue_hd);
2382 evac((StgClosure **)&run_queue_tl);
2385 if (blocked_queue_hd != END_TSO_QUEUE) {
2386 ASSERT(blocked_queue_tl != END_TSO_QUEUE);
2387 evac((StgClosure **)&blocked_queue_hd);
2388 evac((StgClosure **)&blocked_queue_tl);
2391 if (sleeping_queue != END_TSO_QUEUE) {
2392 evac((StgClosure **)&sleeping_queue);
2396 if (suspended_ccalling_threads != END_TSO_QUEUE) {
2397 evac((StgClosure **)&suspended_ccalling_threads);
2400 #if defined(PAR) || defined(GRAN)
2401 markSparkQueue(evac);
2405 /* -----------------------------------------------------------------------------
2408 This is the interface to the garbage collector from Haskell land.
2409 We provide this so that external C code can allocate and garbage
2410 collect when called from Haskell via _ccall_GC.
2412 It might be useful to provide an interface whereby the programmer
2413 can specify more roots (ToDo).
2415 This needs to be protected by the GC condition variable above. KH.
2416 -------------------------------------------------------------------------- */
2418 void (*extra_roots)(evac_fn);
2423 /* Obligated to hold this lock upon entry */
2424 ACQUIRE_LOCK(&sched_mutex);
2425 GarbageCollect(GetRoots,rtsFalse);
2426 RELEASE_LOCK(&sched_mutex);
2430 performMajorGC(void)
2432 ACQUIRE_LOCK(&sched_mutex);
2433 GarbageCollect(GetRoots,rtsTrue);
2434 RELEASE_LOCK(&sched_mutex);
2438 AllRoots(evac_fn evac)
2440 GetRoots(evac); // the scheduler's roots
2441 extra_roots(evac); // the user's roots
2445 performGCWithRoots(void (*get_roots)(evac_fn))
2447 ACQUIRE_LOCK(&sched_mutex);
2448 extra_roots = get_roots;
2449 GarbageCollect(AllRoots,rtsFalse);
2450 RELEASE_LOCK(&sched_mutex);
2453 /* -----------------------------------------------------------------------------
2456 If the thread has reached its maximum stack size, then raise the
2457 StackOverflow exception in the offending thread. Otherwise
2458 relocate the TSO into a larger chunk of memory and adjust its stack
2460 -------------------------------------------------------------------------- */
2463 threadStackOverflow(StgTSO *tso)
2465 nat new_stack_size, new_tso_size, diff, stack_words;
2469 IF_DEBUG(sanity,checkTSO(tso));
2470 if (tso->stack_size >= tso->max_stack_size) {
2473 belch("@@ threadStackOverflow of TSO %d (%p): stack too large (now %ld; max is %ld",
2474 tso->id, tso, tso->stack_size, tso->max_stack_size);
2475 /* If we're debugging, just print out the top of the stack */
2476 printStackChunk(tso->sp, stg_min(tso->stack+tso->stack_size,
2479 /* Send this thread the StackOverflow exception */
2480 raiseAsync(tso, (StgClosure *)stackOverflow_closure);
2484 /* Try to double the current stack size. If that takes us over the
2485 * maximum stack size for this thread, then use the maximum instead.
2486 * Finally round up so the TSO ends up as a whole number of blocks.
2488 new_stack_size = stg_min(tso->stack_size * 2, tso->max_stack_size);
2489 new_tso_size = (nat)BLOCK_ROUND_UP(new_stack_size * sizeof(W_) +
2490 TSO_STRUCT_SIZE)/sizeof(W_);
2491 new_tso_size = round_to_mblocks(new_tso_size); /* Be MBLOCK-friendly */
2492 new_stack_size = new_tso_size - TSO_STRUCT_SIZEW;
2494 IF_DEBUG(scheduler, fprintf(stderr,"== scheduler: increasing stack size from %d words to %d.\n", tso->stack_size, new_stack_size));
2496 dest = (StgTSO *)allocate(new_tso_size);
2497 TICK_ALLOC_TSO(new_stack_size,0);
2499 /* copy the TSO block and the old stack into the new area */
2500 memcpy(dest,tso,TSO_STRUCT_SIZE);
2501 stack_words = tso->stack + tso->stack_size - tso->sp;
2502 new_sp = (P_)dest + new_tso_size - stack_words;
2503 memcpy(new_sp, tso->sp, stack_words * sizeof(W_));
2505 /* relocate the stack pointers... */
2506 diff = (P_)new_sp - (P_)tso->sp; /* In *words* */
2507 dest->su = (StgUpdateFrame *) ((P_)dest->su + diff);
2509 dest->stack_size = new_stack_size;
2511 /* and relocate the update frame list */
2512 relocate_stack(dest, diff);
2514 /* Mark the old TSO as relocated. We have to check for relocated
2515 * TSOs in the garbage collector and any primops that deal with TSOs.
2517 * It's important to set the sp and su values to just beyond the end
2518 * of the stack, so we don't attempt to scavenge any part of the
2521 tso->what_next = ThreadRelocated;
2523 tso->sp = (P_)&(tso->stack[tso->stack_size]);
2524 tso->su = (StgUpdateFrame *)tso->sp;
2525 tso->why_blocked = NotBlocked;
2526 dest->mut_link = NULL;
2528 IF_PAR_DEBUG(verbose,
2529 belch("@@ threadStackOverflow of TSO %d (now at %p): stack size increased to %ld",
2530 tso->id, tso, tso->stack_size);
2531 /* If we're debugging, just print out the top of the stack */
2532 printStackChunk(tso->sp, stg_min(tso->stack+tso->stack_size,
2535 IF_DEBUG(sanity,checkTSO(tso));
2537 IF_DEBUG(scheduler,printTSO(dest));
2543 //@node Blocking Queue Routines, Exception Handling Routines, Garbage Collextion Routines, Main scheduling code
2544 //@subsection Blocking Queue Routines
2546 /* ---------------------------------------------------------------------------
2547 Wake up a queue that was blocked on some resource.
2548 ------------------------------------------------------------------------ */
2552 unblockCount ( StgBlockingQueueElement *bqe, StgClosure *node )
2557 unblockCount ( StgBlockingQueueElement *bqe, StgClosure *node )
2559 /* write RESUME events to log file and
2560 update blocked and fetch time (depending on type of the orig closure) */
2561 if (RtsFlags.ParFlags.ParStats.Full) {
2562 DumpRawGranEvent(CURRENT_PROC, CURRENT_PROC,
2563 GR_RESUMEQ, ((StgTSO *)bqe), ((StgTSO *)bqe)->block_info.closure,
2564 0, 0 /* spark_queue_len(ADVISORY_POOL) */);
2565 if (EMPTY_RUN_QUEUE())
2566 emitSchedule = rtsTrue;
2568 switch (get_itbl(node)->type) {
2570 ((StgTSO *)bqe)->par.fetchtime += CURRENT_TIME-((StgTSO *)bqe)->par.blockedat;
2575 ((StgTSO *)bqe)->par.blocktime += CURRENT_TIME-((StgTSO *)bqe)->par.blockedat;
2582 barf("{unblockOneLocked}Daq Qagh: unexpected closure in blocking queue");
2589 static StgBlockingQueueElement *
2590 unblockOneLocked(StgBlockingQueueElement *bqe, StgClosure *node)
2593 PEs node_loc, tso_loc;
2595 node_loc = where_is(node); // should be lifted out of loop
2596 tso = (StgTSO *)bqe; // wastes an assignment to get the type right
2597 tso_loc = where_is((StgClosure *)tso);
2598 if (IS_LOCAL_TO(PROCS(node),tso_loc)) { // TSO is local
2599 /* !fake_fetch => TSO is on CurrentProc is same as IS_LOCAL_TO */
2600 ASSERT(CurrentProc!=node_loc || tso_loc==CurrentProc);
2601 CurrentTime[CurrentProc] += RtsFlags.GranFlags.Costs.lunblocktime;
2602 // insertThread(tso, node_loc);
2603 new_event(tso_loc, tso_loc, CurrentTime[CurrentProc],
2605 tso, node, (rtsSpark*)NULL);
2606 tso->link = END_TSO_QUEUE; // overwrite link just to be sure
2609 } else { // TSO is remote (actually should be FMBQ)
2610 CurrentTime[CurrentProc] += RtsFlags.GranFlags.Costs.mpacktime +
2611 RtsFlags.GranFlags.Costs.gunblocktime +
2612 RtsFlags.GranFlags.Costs.latency;
2613 new_event(tso_loc, CurrentProc, CurrentTime[CurrentProc],
2615 tso, node, (rtsSpark*)NULL);
2616 tso->link = END_TSO_QUEUE; // overwrite link just to be sure
2619 /* the thread-queue-overhead is accounted for in either Resume or UnblockThread */
2621 fprintf(stderr," %s TSO %d (%p) [PE %d] (block_info.closure=%p) (next=%p) ,",
2622 (node_loc==tso_loc ? "Local" : "Global"),
2623 tso->id, tso, CurrentProc, tso->block_info.closure, tso->link));
2624 tso->block_info.closure = NULL;
2625 IF_DEBUG(scheduler,belch("-- Waking up thread %ld (%p)",
2629 static StgBlockingQueueElement *
2630 unblockOneLocked(StgBlockingQueueElement *bqe, StgClosure *node)
2632 StgBlockingQueueElement *next;
2634 switch (get_itbl(bqe)->type) {
2636 ASSERT(((StgTSO *)bqe)->why_blocked != NotBlocked);
2637 /* if it's a TSO just push it onto the run_queue */
2639 // ((StgTSO *)bqe)->link = END_TSO_QUEUE; // debugging?
2640 PUSH_ON_RUN_QUEUE((StgTSO *)bqe);
2642 unblockCount(bqe, node);
2643 /* reset blocking status after dumping event */
2644 ((StgTSO *)bqe)->why_blocked = NotBlocked;
2648 /* if it's a BLOCKED_FETCH put it on the PendingFetches list */
2650 bqe->link = (StgBlockingQueueElement *)PendingFetches;
2651 PendingFetches = (StgBlockedFetch *)bqe;
2655 /* can ignore this case in a non-debugging setup;
2656 see comments on RBHSave closures above */
2658 /* check that the closure is an RBHSave closure */
2659 ASSERT(get_itbl((StgClosure *)bqe) == &stg_RBH_Save_0_info ||
2660 get_itbl((StgClosure *)bqe) == &stg_RBH_Save_1_info ||
2661 get_itbl((StgClosure *)bqe) == &stg_RBH_Save_2_info);
2665 barf("{unblockOneLocked}Daq Qagh: Unexpected IP (%#lx; %s) in blocking queue at %#lx\n",
2666 get_itbl((StgClosure *)bqe), info_type((StgClosure *)bqe),
2670 IF_PAR_DEBUG(bq, fprintf(stderr, ", %p (%s)", bqe, info_type((StgClosure*)bqe)));
2674 #else /* !GRAN && !PAR */
2676 unblockOneLocked(StgTSO *tso)
2680 ASSERT(get_itbl(tso)->type == TSO);
2681 ASSERT(tso->why_blocked != NotBlocked);
2682 tso->why_blocked = NotBlocked;
2684 PUSH_ON_RUN_QUEUE(tso);
2686 IF_DEBUG(scheduler,sched_belch("waking up thread %ld", tso->id));
2691 #if defined(GRAN) || defined(PAR)
2692 inline StgBlockingQueueElement *
2693 unblockOne(StgBlockingQueueElement *bqe, StgClosure *node)
2695 ACQUIRE_LOCK(&sched_mutex);
2696 bqe = unblockOneLocked(bqe, node);
2697 RELEASE_LOCK(&sched_mutex);
2702 unblockOne(StgTSO *tso)
2704 ACQUIRE_LOCK(&sched_mutex);
2705 tso = unblockOneLocked(tso);
2706 RELEASE_LOCK(&sched_mutex);
2713 awakenBlockedQueue(StgBlockingQueueElement *q, StgClosure *node)
2715 StgBlockingQueueElement *bqe;
2720 belch("##-_ AwBQ for node %p on PE %d @ %ld by TSO %d (%p): ", \
2721 node, CurrentProc, CurrentTime[CurrentProc],
2722 CurrentTSO->id, CurrentTSO));
2724 node_loc = where_is(node);
2726 ASSERT(q == END_BQ_QUEUE ||
2727 get_itbl(q)->type == TSO || // q is either a TSO or an RBHSave
2728 get_itbl(q)->type == CONSTR); // closure (type constructor)
2729 ASSERT(is_unique(node));
2731 /* FAKE FETCH: magically copy the node to the tso's proc;
2732 no Fetch necessary because in reality the node should not have been
2733 moved to the other PE in the first place
2735 if (CurrentProc!=node_loc) {
2737 belch("## node %p is on PE %d but CurrentProc is %d (TSO %d); assuming fake fetch and adjusting bitmask (old: %#x)",
2738 node, node_loc, CurrentProc, CurrentTSO->id,
2739 // CurrentTSO, where_is(CurrentTSO),
2740 node->header.gran.procs));
2741 node->header.gran.procs = (node->header.gran.procs) | PE_NUMBER(CurrentProc);
2743 belch("## new bitmask of node %p is %#x",
2744 node, node->header.gran.procs));
2745 if (RtsFlags.GranFlags.GranSimStats.Global) {
2746 globalGranStats.tot_fake_fetches++;
2751 // ToDo: check: ASSERT(CurrentProc==node_loc);
2752 while (get_itbl(bqe)->type==TSO) { // q != END_TSO_QUEUE) {
2755 bqe points to the current element in the queue
2756 next points to the next element in the queue
2758 //tso = (StgTSO *)bqe; // wastes an assignment to get the type right
2759 //tso_loc = where_is(tso);
2761 bqe = unblockOneLocked(bqe, node);
2764 /* if this is the BQ of an RBH, we have to put back the info ripped out of
2765 the closure to make room for the anchor of the BQ */
2766 if (bqe!=END_BQ_QUEUE) {
2767 ASSERT(get_itbl(node)->type == RBH && get_itbl(bqe)->type == CONSTR);
2769 ASSERT((info_ptr==&RBH_Save_0_info) ||
2770 (info_ptr==&RBH_Save_1_info) ||
2771 (info_ptr==&RBH_Save_2_info));
2773 /* cf. convertToRBH in RBH.c for writing the RBHSave closure */
2774 ((StgRBH *)node)->blocking_queue = (StgBlockingQueueElement *)((StgRBHSave *)bqe)->payload[0];
2775 ((StgRBH *)node)->mut_link = (StgMutClosure *)((StgRBHSave *)bqe)->payload[1];
2778 belch("## Filled in RBH_Save for %p (%s) at end of AwBQ",
2779 node, info_type(node)));
2782 /* statistics gathering */
2783 if (RtsFlags.GranFlags.GranSimStats.Global) {
2784 // globalGranStats.tot_bq_processing_time += bq_processing_time;
2785 globalGranStats.tot_bq_len += len; // total length of all bqs awakened
2786 // globalGranStats.tot_bq_len_local += len_local; // same for local TSOs only
2787 globalGranStats.tot_awbq++; // total no. of bqs awakened
2790 fprintf(stderr,"## BQ Stats of %p: [%d entries] %s\n",
2791 node, len, (bqe!=END_BQ_QUEUE) ? "RBH" : ""));
2795 awakenBlockedQueue(StgBlockingQueueElement *q, StgClosure *node)
2797 StgBlockingQueueElement *bqe;
2799 ACQUIRE_LOCK(&sched_mutex);
2801 IF_PAR_DEBUG(verbose,
2802 belch("##-_ AwBQ for node %p on [%x]: ",
2806 if(get_itbl(q)->type == CONSTR || q==END_BQ_QUEUE) {
2807 IF_PAR_DEBUG(verbose, belch("## ... nothing to unblock so lets just return. RFP (BUG?)"));
2812 ASSERT(q == END_BQ_QUEUE ||
2813 get_itbl(q)->type == TSO ||
2814 get_itbl(q)->type == BLOCKED_FETCH ||
2815 get_itbl(q)->type == CONSTR);
2818 while (get_itbl(bqe)->type==TSO ||
2819 get_itbl(bqe)->type==BLOCKED_FETCH) {
2820 bqe = unblockOneLocked(bqe, node);
2822 RELEASE_LOCK(&sched_mutex);
2825 #else /* !GRAN && !PAR */
2827 awakenBlockedQueue(StgTSO *tso)
2829 ACQUIRE_LOCK(&sched_mutex);
2830 while (tso != END_TSO_QUEUE) {
2831 tso = unblockOneLocked(tso);
2833 RELEASE_LOCK(&sched_mutex);
2837 //@node Exception Handling Routines, Debugging Routines, Blocking Queue Routines, Main scheduling code
2838 //@subsection Exception Handling Routines
2840 /* ---------------------------------------------------------------------------
2842 - usually called inside a signal handler so it mustn't do anything fancy.
2843 ------------------------------------------------------------------------ */
2846 interruptStgRts(void)
2852 /* -----------------------------------------------------------------------------
2855 This is for use when we raise an exception in another thread, which
2857 This has nothing to do with the UnblockThread event in GranSim. -- HWL
2858 -------------------------------------------------------------------------- */
2860 #if defined(GRAN) || defined(PAR)
2862 NB: only the type of the blocking queue is different in GranSim and GUM
2863 the operations on the queue-elements are the same
2864 long live polymorphism!
2866 Locks: sched_mutex is held upon entry and exit.
2870 unblockThread(StgTSO *tso)
2872 StgBlockingQueueElement *t, **last;
2874 switch (tso->why_blocked) {
2877 return; /* not blocked */
2880 ASSERT(get_itbl(tso->block_info.closure)->type == MVAR);
2882 StgBlockingQueueElement *last_tso = END_BQ_QUEUE;
2883 StgMVar *mvar = (StgMVar *)(tso->block_info.closure);
2885 last = (StgBlockingQueueElement **)&mvar->head;
2886 for (t = (StgBlockingQueueElement *)mvar->head;
2888 last = &t->link, last_tso = t, t = t->link) {
2889 if (t == (StgBlockingQueueElement *)tso) {
2890 *last = (StgBlockingQueueElement *)tso->link;
2891 if (mvar->tail == tso) {
2892 mvar->tail = (StgTSO *)last_tso;
2897 barf("unblockThread (MVAR): TSO not found");
2900 case BlockedOnBlackHole:
2901 ASSERT(get_itbl(tso->block_info.closure)->type == BLACKHOLE_BQ);
2903 StgBlockingQueue *bq = (StgBlockingQueue *)(tso->block_info.closure);
2905 last = &bq->blocking_queue;
2906 for (t = bq->blocking_queue;
2908 last = &t->link, t = t->link) {
2909 if (t == (StgBlockingQueueElement *)tso) {
2910 *last = (StgBlockingQueueElement *)tso->link;
2914 barf("unblockThread (BLACKHOLE): TSO not found");
2917 case BlockedOnException:
2919 StgTSO *target = tso->block_info.tso;
2921 ASSERT(get_itbl(target)->type == TSO);
2923 if (target->what_next == ThreadRelocated) {
2924 target = target->link;
2925 ASSERT(get_itbl(target)->type == TSO);
2928 ASSERT(target->blocked_exceptions != NULL);
2930 last = (StgBlockingQueueElement **)&target->blocked_exceptions;
2931 for (t = (StgBlockingQueueElement *)target->blocked_exceptions;
2933 last = &t->link, t = t->link) {
2934 ASSERT(get_itbl(t)->type == TSO);
2935 if (t == (StgBlockingQueueElement *)tso) {
2936 *last = (StgBlockingQueueElement *)tso->link;
2940 barf("unblockThread (Exception): TSO not found");
2944 case BlockedOnWrite:
2946 /* take TSO off blocked_queue */
2947 StgBlockingQueueElement *prev = NULL;
2948 for (t = (StgBlockingQueueElement *)blocked_queue_hd; t != END_BQ_QUEUE;
2949 prev = t, t = t->link) {
2950 if (t == (StgBlockingQueueElement *)tso) {
2952 blocked_queue_hd = (StgTSO *)t->link;
2953 if ((StgBlockingQueueElement *)blocked_queue_tl == t) {
2954 blocked_queue_tl = END_TSO_QUEUE;
2957 prev->link = t->link;
2958 if ((StgBlockingQueueElement *)blocked_queue_tl == t) {
2959 blocked_queue_tl = (StgTSO *)prev;
2965 barf("unblockThread (I/O): TSO not found");
2968 case BlockedOnDelay:
2970 /* take TSO off sleeping_queue */
2971 StgBlockingQueueElement *prev = NULL;
2972 for (t = (StgBlockingQueueElement *)sleeping_queue; t != END_BQ_QUEUE;
2973 prev = t, t = t->link) {
2974 if (t == (StgBlockingQueueElement *)tso) {
2976 sleeping_queue = (StgTSO *)t->link;
2978 prev->link = t->link;
2983 barf("unblockThread (I/O): TSO not found");
2987 barf("unblockThread");
2991 tso->link = END_TSO_QUEUE;
2992 tso->why_blocked = NotBlocked;
2993 tso->block_info.closure = NULL;
2994 PUSH_ON_RUN_QUEUE(tso);
2998 unblockThread(StgTSO *tso)
3002 /* To avoid locking unnecessarily. */
3003 if (tso->why_blocked == NotBlocked) {
3007 switch (tso->why_blocked) {
3010 ASSERT(get_itbl(tso->block_info.closure)->type == MVAR);
3012 StgTSO *last_tso = END_TSO_QUEUE;
3013 StgMVar *mvar = (StgMVar *)(tso->block_info.closure);
3016 for (t = mvar->head; t != END_TSO_QUEUE;
3017 last = &t->link, last_tso = t, t = t->link) {
3020 if (mvar->tail == tso) {
3021 mvar->tail = last_tso;
3026 barf("unblockThread (MVAR): TSO not found");
3029 case BlockedOnBlackHole:
3030 ASSERT(get_itbl(tso->block_info.closure)->type == BLACKHOLE_BQ);
3032 StgBlockingQueue *bq = (StgBlockingQueue *)(tso->block_info.closure);
3034 last = &bq->blocking_queue;
3035 for (t = bq->blocking_queue; t != END_TSO_QUEUE;
3036 last = &t->link, t = t->link) {
3042 barf("unblockThread (BLACKHOLE): TSO not found");
3045 case BlockedOnException:
3047 StgTSO *target = tso->block_info.tso;
3049 ASSERT(get_itbl(target)->type == TSO);
3051 while (target->what_next == ThreadRelocated) {
3052 target = target->link;
3053 ASSERT(get_itbl(target)->type == TSO);
3056 ASSERT(target->blocked_exceptions != NULL);
3058 last = &target->blocked_exceptions;
3059 for (t = target->blocked_exceptions; t != END_TSO_QUEUE;
3060 last = &t->link, t = t->link) {
3061 ASSERT(get_itbl(t)->type == TSO);
3067 barf("unblockThread (Exception): TSO not found");
3071 case BlockedOnWrite:
3073 StgTSO *prev = NULL;
3074 for (t = blocked_queue_hd; t != END_TSO_QUEUE;
3075 prev = t, t = t->link) {
3078 blocked_queue_hd = t->link;
3079 if (blocked_queue_tl == t) {
3080 blocked_queue_tl = END_TSO_QUEUE;
3083 prev->link = t->link;
3084 if (blocked_queue_tl == t) {
3085 blocked_queue_tl = prev;
3091 barf("unblockThread (I/O): TSO not found");
3094 case BlockedOnDelay:
3096 StgTSO *prev = NULL;
3097 for (t = sleeping_queue; t != END_TSO_QUEUE;
3098 prev = t, t = t->link) {
3101 sleeping_queue = t->link;
3103 prev->link = t->link;
3108 barf("unblockThread (I/O): TSO not found");
3112 barf("unblockThread");
3116 tso->link = END_TSO_QUEUE;
3117 tso->why_blocked = NotBlocked;
3118 tso->block_info.closure = NULL;
3119 PUSH_ON_RUN_QUEUE(tso);
3123 /* -----------------------------------------------------------------------------
3126 * The following function implements the magic for raising an
3127 * asynchronous exception in an existing thread.
3129 * We first remove the thread from any queue on which it might be
3130 * blocked. The possible blockages are MVARs and BLACKHOLE_BQs.
3132 * We strip the stack down to the innermost CATCH_FRAME, building
3133 * thunks in the heap for all the active computations, so they can
3134 * be restarted if necessary. When we reach a CATCH_FRAME, we build
3135 * an application of the handler to the exception, and push it on
3136 * the top of the stack.
3138 * How exactly do we save all the active computations? We create an
3139 * AP_UPD for every UpdateFrame on the stack. Entering one of these
3140 * AP_UPDs pushes everything from the corresponding update frame
3141 * upwards onto the stack. (Actually, it pushes everything up to the
3142 * next update frame plus a pointer to the next AP_UPD object.
3143 * Entering the next AP_UPD object pushes more onto the stack until we
3144 * reach the last AP_UPD object - at which point the stack should look
3145 * exactly as it did when we killed the TSO and we can continue
3146 * execution by entering the closure on top of the stack.
3148 * We can also kill a thread entirely - this happens if either (a) the
3149 * exception passed to raiseAsync is NULL, or (b) there's no
3150 * CATCH_FRAME on the stack. In either case, we strip the entire
3151 * stack and replace the thread with a zombie.
3153 * Locks: sched_mutex held upon entry nor exit.
3155 * -------------------------------------------------------------------------- */
3158 deleteThread(StgTSO *tso)
3160 raiseAsync(tso,NULL);
3164 raiseAsyncWithLock(StgTSO *tso, StgClosure *exception)
3166 /* When raising async exs from contexts where sched_mutex isn't held;
3167 use raiseAsyncWithLock(). */
3168 ACQUIRE_LOCK(&sched_mutex);
3169 raiseAsync(tso,exception);
3170 RELEASE_LOCK(&sched_mutex);
3174 raiseAsync(StgTSO *tso, StgClosure *exception)
3176 StgUpdateFrame* su = tso->su;
3177 StgPtr sp = tso->sp;
3179 /* Thread already dead? */
3180 if (tso->what_next == ThreadComplete || tso->what_next == ThreadKilled) {
3184 IF_DEBUG(scheduler, sched_belch("raising exception in thread %ld.", tso->id));
3186 /* Remove it from any blocking queues */
3189 IF_DEBUG(scheduler, sched_belch("raising exception in thread %ld.", tso->id));
3190 /* The stack freezing code assumes there's a closure pointer on
3191 * the top of the stack. This isn't always the case with compiled
3192 * code, so we have to push a dummy closure on the top which just
3193 * returns to the next return address on the stack.
3195 if ( LOOKS_LIKE_GHC_INFO((void*)*sp) ) {
3196 *(--sp) = (W_)&stg_dummy_ret_closure;
3200 nat words = ((P_)su - (P_)sp) - 1;
3204 /* If we find a CATCH_FRAME, and we've got an exception to raise,
3205 * then build the THUNK raise(exception), and leave it on
3206 * top of the CATCH_FRAME ready to enter.
3208 if (get_itbl(su)->type == CATCH_FRAME && exception != NULL) {
3210 StgCatchFrame *cf = (StgCatchFrame *)su;
3214 /* we've got an exception to raise, so let's pass it to the
3215 * handler in this frame.
3217 raise = (StgClosure *)allocate(sizeofW(StgClosure)+1);
3218 TICK_ALLOC_SE_THK(1,0);
3219 SET_HDR(raise,&stg_raise_info,cf->header.prof.ccs);
3220 raise->payload[0] = exception;
3222 /* throw away the stack from Sp up to the CATCH_FRAME.
3226 /* Ensure that async excpetions are blocked now, so we don't get
3227 * a surprise exception before we get around to executing the
3230 if (tso->blocked_exceptions == NULL) {
3231 tso->blocked_exceptions = END_TSO_QUEUE;
3234 /* Put the newly-built THUNK on top of the stack, ready to execute
3235 * when the thread restarts.
3240 tso->what_next = ThreadEnterGHC;
3241 IF_DEBUG(sanity, checkTSO(tso));
3245 /* First build an AP_UPD consisting of the stack chunk above the
3246 * current update frame, with the top word on the stack as the
3249 ap = (StgAP_UPD *)allocate(AP_sizeW(words));
3254 ap->fun = (StgClosure *)sp[0];
3256 for(i=0; i < (nat)words; ++i) {
3257 ap->payload[i] = (StgClosure *)*sp++;
3260 switch (get_itbl(su)->type) {
3264 SET_HDR(ap,&stg_AP_UPD_info,su->header.prof.ccs /* ToDo */);
3265 TICK_ALLOC_UP_THK(words+1,0);
3268 fprintf(stderr, "scheduler: Updating ");
3269 printPtr((P_)su->updatee);
3270 fprintf(stderr, " with ");
3271 printObj((StgClosure *)ap);
3274 /* Replace the updatee with an indirection - happily
3275 * this will also wake up any threads currently
3276 * waiting on the result.
3278 * Warning: if we're in a loop, more than one update frame on
3279 * the stack may point to the same object. Be careful not to
3280 * overwrite an IND_OLDGEN in this case, because we'll screw
3281 * up the mutable lists. To be on the safe side, don't
3282 * overwrite any kind of indirection at all. See also
3283 * threadSqueezeStack in GC.c, where we have to make a similar
3286 if (!closure_IND(su->updatee)) {
3287 UPD_IND_NOLOCK(su->updatee,ap); /* revert the black hole */
3290 sp += sizeofW(StgUpdateFrame) -1;
3291 sp[0] = (W_)ap; /* push onto stack */
3297 StgCatchFrame *cf = (StgCatchFrame *)su;
3300 /* We want a PAP, not an AP_UPD. Fortunately, the
3301 * layout's the same.
3303 SET_HDR(ap,&stg_PAP_info,su->header.prof.ccs /* ToDo */);
3304 TICK_ALLOC_UPD_PAP(words+1,0);
3306 /* now build o = FUN(catch,ap,handler) */
3307 o = (StgClosure *)allocate(sizeofW(StgClosure)+2);
3308 TICK_ALLOC_FUN(2,0);
3309 SET_HDR(o,&stg_catch_info,su->header.prof.ccs /* ToDo */);
3310 o->payload[0] = (StgClosure *)ap;
3311 o->payload[1] = cf->handler;
3314 fprintf(stderr, "scheduler: Built ");
3315 printObj((StgClosure *)o);
3318 /* pop the old handler and put o on the stack */
3320 sp += sizeofW(StgCatchFrame) - 1;
3327 StgSeqFrame *sf = (StgSeqFrame *)su;
3330 SET_HDR(ap,&stg_PAP_info,su->header.prof.ccs /* ToDo */);
3331 TICK_ALLOC_UPD_PAP(words+1,0);
3333 /* now build o = FUN(seq,ap) */
3334 o = (StgClosure *)allocate(sizeofW(StgClosure)+1);
3335 TICK_ALLOC_SE_THK(1,0);
3336 SET_HDR(o,&stg_seq_info,su->header.prof.ccs /* ToDo */);
3337 o->payload[0] = (StgClosure *)ap;
3340 fprintf(stderr, "scheduler: Built ");
3341 printObj((StgClosure *)o);
3344 /* pop the old handler and put o on the stack */
3346 sp += sizeofW(StgSeqFrame) - 1;
3352 /* We've stripped the entire stack, the thread is now dead. */
3353 sp += sizeofW(StgStopFrame) - 1;
3354 sp[0] = (W_)exception; /* save the exception */
3355 tso->what_next = ThreadKilled;
3356 tso->su = (StgUpdateFrame *)(sp+1);
3367 /* -----------------------------------------------------------------------------
3368 resurrectThreads is called after garbage collection on the list of
3369 threads found to be garbage. Each of these threads will be woken
3370 up and sent a signal: BlockedOnDeadMVar if the thread was blocked
3371 on an MVar, or NonTermination if the thread was blocked on a Black
3374 Locks: sched_mutex isn't held upon entry nor exit.
3375 -------------------------------------------------------------------------- */
3378 resurrectThreads( StgTSO *threads )
3382 for (tso = threads; tso != END_TSO_QUEUE; tso = next) {
3383 next = tso->global_link;
3384 tso->global_link = all_threads;
3386 IF_DEBUG(scheduler, sched_belch("resurrecting thread %d", tso->id));
3388 switch (tso->why_blocked) {
3390 case BlockedOnException:
3391 /* Called by GC - sched_mutex lock is currently held. */
3392 raiseAsync(tso,(StgClosure *)BlockedOnDeadMVar_closure);
3394 case BlockedOnBlackHole:
3395 raiseAsync(tso,(StgClosure *)NonTermination_closure);
3398 /* This might happen if the thread was blocked on a black hole
3399 * belonging to a thread that we've just woken up (raiseAsync
3400 * can wake up threads, remember...).
3404 barf("resurrectThreads: thread blocked in a strange way");
3409 /* -----------------------------------------------------------------------------
3410 * Blackhole detection: if we reach a deadlock, test whether any
3411 * threads are blocked on themselves. Any threads which are found to
3412 * be self-blocked get sent a NonTermination exception.
3414 * This is only done in a deadlock situation in order to avoid
3415 * performance overhead in the normal case.
3417 * Locks: sched_mutex is held upon entry and exit.
3418 * -------------------------------------------------------------------------- */
3421 detectBlackHoles( void )
3423 StgTSO *t = all_threads;
3424 StgUpdateFrame *frame;
3425 StgClosure *blocked_on;
3427 for (t = all_threads; t != END_TSO_QUEUE; t = t->global_link) {
3429 while (t->what_next == ThreadRelocated) {
3431 ASSERT(get_itbl(t)->type == TSO);
3434 if (t->why_blocked != BlockedOnBlackHole) {
3438 blocked_on = t->block_info.closure;
3440 for (frame = t->su; ; frame = frame->link) {
3441 switch (get_itbl(frame)->type) {
3444 if (frame->updatee == blocked_on) {
3445 /* We are blocking on one of our own computations, so
3446 * send this thread the NonTermination exception.
3449 sched_belch("thread %d is blocked on itself", t->id));
3450 raiseAsync(t, (StgClosure *)NonTermination_closure);
3471 //@node Debugging Routines, Index, Exception Handling Routines, Main scheduling code
3472 //@subsection Debugging Routines
3474 /* -----------------------------------------------------------------------------
3475 Debugging: why is a thread blocked
3476 -------------------------------------------------------------------------- */
3481 printThreadBlockage(StgTSO *tso)
3483 switch (tso->why_blocked) {
3485 fprintf(stderr,"is blocked on read from fd %d", tso->block_info.fd);
3487 case BlockedOnWrite:
3488 fprintf(stderr,"is blocked on write to fd %d", tso->block_info.fd);
3490 case BlockedOnDelay:
3491 fprintf(stderr,"is blocked until %d", tso->block_info.target);
3494 fprintf(stderr,"is blocked on an MVar");
3496 case BlockedOnException:
3497 fprintf(stderr,"is blocked on delivering an exception to thread %d",
3498 tso->block_info.tso->id);
3500 case BlockedOnBlackHole:
3501 fprintf(stderr,"is blocked on a black hole");
3504 fprintf(stderr,"is not blocked");
3508 fprintf(stderr,"is blocked on global address; local FM_BQ is %p (%s)",
3509 tso->block_info.closure, info_type(tso->block_info.closure));
3511 case BlockedOnGA_NoSend:
3512 fprintf(stderr,"is blocked on global address (no send); local FM_BQ is %p (%s)",
3513 tso->block_info.closure, info_type(tso->block_info.closure));
3516 #if defined(RTS_SUPPORTS_THREADS)
3517 case BlockedOnCCall:
3518 fprintf(stderr,"is blocked on an external call");
3522 barf("printThreadBlockage: strange tso->why_blocked: %d for TSO %d (%d)",
3523 tso->why_blocked, tso->id, tso);
3528 printThreadStatus(StgTSO *tso)
3530 switch (tso->what_next) {
3532 fprintf(stderr,"has been killed");
3534 case ThreadComplete:
3535 fprintf(stderr,"has completed");
3538 printThreadBlockage(tso);
3543 printAllThreads(void)
3548 char time_string[TIME_STR_LEN], node_str[NODE_STR_LEN];
3549 ullong_format_string(TIME_ON_PROC(CurrentProc),
3550 time_string, rtsFalse/*no commas!*/);
3552 sched_belch("all threads at [%s]:", time_string);
3554 char time_string[TIME_STR_LEN], node_str[NODE_STR_LEN];
3555 ullong_format_string(CURRENT_TIME,
3556 time_string, rtsFalse/*no commas!*/);
3558 sched_belch("all threads at [%s]:", time_string);
3560 sched_belch("all threads:");
3563 for (t = all_threads; t != END_TSO_QUEUE; t = t->global_link) {
3564 fprintf(stderr, "\tthread %d ", t->id);
3565 if (t->label) fprintf(stderr,"[\"%s\"] ",t->label);
3566 printThreadStatus(t);
3567 fprintf(stderr,"\n");
3572 Print a whole blocking queue attached to node (debugging only).
3577 print_bq (StgClosure *node)
3579 StgBlockingQueueElement *bqe;
3583 fprintf(stderr,"## BQ of closure %p (%s): ",
3584 node, info_type(node));
3586 /* should cover all closures that may have a blocking queue */
3587 ASSERT(get_itbl(node)->type == BLACKHOLE_BQ ||
3588 get_itbl(node)->type == FETCH_ME_BQ ||
3589 get_itbl(node)->type == RBH ||
3590 get_itbl(node)->type == MVAR);
3592 ASSERT(node!=(StgClosure*)NULL); // sanity check
3594 print_bqe(((StgBlockingQueue*)node)->blocking_queue);
3598 Print a whole blocking queue starting with the element bqe.
3601 print_bqe (StgBlockingQueueElement *bqe)
3606 NB: In a parallel setup a BQ of an RBH must end with an RBH_Save closure;
3608 for (end = (bqe==END_BQ_QUEUE);
3609 !end; // iterate until bqe points to a CONSTR
3610 end = (get_itbl(bqe)->type == CONSTR) || (bqe->link==END_BQ_QUEUE),
3611 bqe = end ? END_BQ_QUEUE : bqe->link) {
3612 ASSERT(bqe != END_BQ_QUEUE); // sanity check
3613 ASSERT(bqe != (StgBlockingQueueElement *)NULL); // sanity check
3614 /* types of closures that may appear in a blocking queue */
3615 ASSERT(get_itbl(bqe)->type == TSO ||
3616 get_itbl(bqe)->type == BLOCKED_FETCH ||
3617 get_itbl(bqe)->type == CONSTR);
3618 /* only BQs of an RBH end with an RBH_Save closure */
3619 //ASSERT(get_itbl(bqe)->type != CONSTR || get_itbl(node)->type == RBH);
3621 switch (get_itbl(bqe)->type) {
3623 fprintf(stderr," TSO %u (%x),",
3624 ((StgTSO *)bqe)->id, ((StgTSO *)bqe));
3627 fprintf(stderr," BF (node=%p, ga=((%x, %d, %x)),",
3628 ((StgBlockedFetch *)bqe)->node,
3629 ((StgBlockedFetch *)bqe)->ga.payload.gc.gtid,
3630 ((StgBlockedFetch *)bqe)->ga.payload.gc.slot,
3631 ((StgBlockedFetch *)bqe)->ga.weight);
3634 fprintf(stderr," %s (IP %p),",
3635 (get_itbl(bqe) == &stg_RBH_Save_0_info ? "RBH_Save_0" :
3636 get_itbl(bqe) == &stg_RBH_Save_1_info ? "RBH_Save_1" :
3637 get_itbl(bqe) == &stg_RBH_Save_2_info ? "RBH_Save_2" :
3638 "RBH_Save_?"), get_itbl(bqe));
3641 barf("Unexpected closure type %s in blocking queue", // of %p (%s)",
3642 info_type((StgClosure *)bqe)); // , node, info_type(node));
3646 fputc('\n', stderr);
3648 # elif defined(GRAN)
3650 print_bq (StgClosure *node)
3652 StgBlockingQueueElement *bqe;
3653 PEs node_loc, tso_loc;
3656 /* should cover all closures that may have a blocking queue */
3657 ASSERT(get_itbl(node)->type == BLACKHOLE_BQ ||
3658 get_itbl(node)->type == FETCH_ME_BQ ||
3659 get_itbl(node)->type == RBH);
3661 ASSERT(node!=(StgClosure*)NULL); // sanity check
3662 node_loc = where_is(node);
3664 fprintf(stderr,"## BQ of closure %p (%s) on [PE %d]: ",
3665 node, info_type(node), node_loc);
3668 NB: In a parallel setup a BQ of an RBH must end with an RBH_Save closure;
3670 for (bqe = ((StgBlockingQueue*)node)->blocking_queue, end = (bqe==END_BQ_QUEUE);
3671 !end; // iterate until bqe points to a CONSTR
3672 end = (get_itbl(bqe)->type == CONSTR) || (bqe->link==END_BQ_QUEUE), bqe = end ? END_BQ_QUEUE : bqe->link) {
3673 ASSERT(bqe != END_BQ_QUEUE); // sanity check
3674 ASSERT(bqe != (StgBlockingQueueElement *)NULL); // sanity check
3675 /* types of closures that may appear in a blocking queue */
3676 ASSERT(get_itbl(bqe)->type == TSO ||
3677 get_itbl(bqe)->type == CONSTR);
3678 /* only BQs of an RBH end with an RBH_Save closure */
3679 ASSERT(get_itbl(bqe)->type != CONSTR || get_itbl(node)->type == RBH);
3681 tso_loc = where_is((StgClosure *)bqe);
3682 switch (get_itbl(bqe)->type) {
3684 fprintf(stderr," TSO %d (%p) on [PE %d],",
3685 ((StgTSO *)bqe)->id, (StgTSO *)bqe, tso_loc);
3688 fprintf(stderr," %s (IP %p),",
3689 (get_itbl(bqe) == &stg_RBH_Save_0_info ? "RBH_Save_0" :
3690 get_itbl(bqe) == &stg_RBH_Save_1_info ? "RBH_Save_1" :
3691 get_itbl(bqe) == &stg_RBH_Save_2_info ? "RBH_Save_2" :
3692 "RBH_Save_?"), get_itbl(bqe));
3695 barf("Unexpected closure type %s in blocking queue of %p (%s)",
3696 info_type((StgClosure *)bqe), node, info_type(node));
3700 fputc('\n', stderr);
3704 Nice and easy: only TSOs on the blocking queue
3707 print_bq (StgClosure *node)
3711 ASSERT(node!=(StgClosure*)NULL); // sanity check
3712 for (tso = ((StgBlockingQueue*)node)->blocking_queue;
3713 tso != END_TSO_QUEUE;
3715 ASSERT(tso!=NULL && tso!=END_TSO_QUEUE); // sanity check
3716 ASSERT(get_itbl(tso)->type == TSO); // guess what, sanity check
3717 fprintf(stderr," TSO %d (%p),", tso->id, tso);
3719 fputc('\n', stderr);
3730 for (i=0, tso=run_queue_hd;
3731 tso != END_TSO_QUEUE;
3740 sched_belch(char *s, ...)
3745 fprintf(stderr, "scheduler (task %ld): ", osThreadId());
3747 fprintf(stderr, "== ");
3749 fprintf(stderr, "scheduler: ");
3751 vfprintf(stderr, s, ap);
3752 fprintf(stderr, "\n");
3758 //@node Index, , Debugging Routines, Main scheduling code
3762 //* StgMainThread:: @cindex\s-+StgMainThread
3763 //* awaken_blocked_queue:: @cindex\s-+awaken_blocked_queue
3764 //* blocked_queue_hd:: @cindex\s-+blocked_queue_hd
3765 //* blocked_queue_tl:: @cindex\s-+blocked_queue_tl
3766 //* context_switch:: @cindex\s-+context_switch
3767 //* createThread:: @cindex\s-+createThread
3768 //* gc_pending_cond:: @cindex\s-+gc_pending_cond
3769 //* initScheduler:: @cindex\s-+initScheduler
3770 //* interrupted:: @cindex\s-+interrupted
3771 //* next_thread_id:: @cindex\s-+next_thread_id
3772 //* print_bq:: @cindex\s-+print_bq
3773 //* run_queue_hd:: @cindex\s-+run_queue_hd
3774 //* run_queue_tl:: @cindex\s-+run_queue_tl
3775 //* sched_mutex:: @cindex\s-+sched_mutex
3776 //* schedule:: @cindex\s-+schedule
3777 //* take_off_run_queue:: @cindex\s-+take_off_run_queue
3778 //* term_mutex:: @cindex\s-+term_mutex