1 /* -----------------------------------------------------------------------------
3 * (c) The GHC Team, 1998-2011
5 * Out-of-line primitive operations
7 * This file contains the implementations of all the primitive
8 * operations ("primops") which are not expanded inline. See
9 * ghc/compiler/prelude/primops.txt.pp for a list of all the primops;
10 * this file contains code for most of those with the attribute
13 * Entry convention: the entry convention for a primop is that all the
14 * args are in Stg registers (R1, R2, etc.). This is to make writing
15 * the primops easier. (see compiler/codeGen/CgCallConv.hs).
17 * Return convention: results from a primop are generally returned
18 * using the ordinary unboxed tuple return convention. The C-- parser
19 * implements the RET_xxxx() macros to perform unboxed-tuple returns
20 * based on the prevailing return convention.
22 * This file is written in a subset of C--, extended with various
23 * features specific to GHC. It is compiled by GHC directly. For the
24 * syntax of .cmm files, see the parser in ghc/compiler/cmm/CmmParse.y.
26 * ---------------------------------------------------------------------------*/
31 import pthread_mutex_lock;
32 import pthread_mutex_unlock;
34 import base_ControlziExceptionziBase_nestedAtomically_closure;
35 import EnterCriticalSection;
36 import LeaveCriticalSection;
37 import ghczmprim_GHCziTypes_False_closure;
38 #if !defined(mingw32_HOST_OS)
42 /*-----------------------------------------------------------------------------
45 Basically just new*Array - the others are all inline macros.
47 The size arg is always passed in R1, and the result returned in R1.
49 The slow entry point is for returning from a heap check, the saved
50 size argument must be re-loaded from the stack.
51 -------------------------------------------------------------------------- */
53 /* for objects that are *less* than the size of a word, make sure we
54 * round up to the nearest word for the size of the array.
59 W_ words, payload_words, n, p;
60 MAYBE_GC(NO_PTRS,stg_newByteArrayzh);
62 payload_words = ROUNDUP_BYTES_TO_WDS(n);
63 words = BYTES_TO_WDS(SIZEOF_StgArrWords) + payload_words;
64 ("ptr" p) = foreign "C" allocate(MyCapability() "ptr",words) [];
65 TICK_ALLOC_PRIM(SIZEOF_StgArrWords,WDS(payload_words),0);
66 SET_HDR(p, stg_ARR_WORDS_info, W_[CCCS]);
67 StgArrWords_bytes(p) = n;
72 #define BA_MASK (BA_ALIGN-1)
74 stg_newPinnedByteArrayzh
76 W_ words, n, bytes, payload_words, p;
78 MAYBE_GC(NO_PTRS,stg_newPinnedByteArrayzh);
81 /* payload_words is what we will tell the profiler we had to allocate */
82 payload_words = ROUNDUP_BYTES_TO_WDS(bytes);
83 /* When we actually allocate memory, we need to allow space for the
85 bytes = bytes + SIZEOF_StgArrWords;
86 /* And we want to align to BA_ALIGN bytes, so we need to allow space
87 to shift up to BA_ALIGN - 1 bytes: */
88 bytes = bytes + BA_ALIGN - 1;
89 /* Now we convert to a number of words: */
90 words = ROUNDUP_BYTES_TO_WDS(bytes);
92 ("ptr" p) = foreign "C" allocatePinned(MyCapability() "ptr", words) [];
93 TICK_ALLOC_PRIM(SIZEOF_StgArrWords,WDS(payload_words),0);
95 /* Now we need to move p forward so that the payload is aligned
97 p = p + ((-p - SIZEOF_StgArrWords) & BA_MASK);
99 SET_HDR(p, stg_ARR_WORDS_info, W_[CCCS]);
100 StgArrWords_bytes(p) = n;
104 stg_newAlignedPinnedByteArrayzh
106 W_ words, n, bytes, payload_words, p, alignment;
108 MAYBE_GC(NO_PTRS,stg_newAlignedPinnedByteArrayzh);
112 /* we always supply at least word-aligned memory, so there's no
113 need to allow extra space for alignment if the requirement is less
114 than a word. This also prevents mischief with alignment == 0. */
115 if (alignment <= SIZEOF_W) { alignment = 1; }
119 /* payload_words is what we will tell the profiler we had to allocate */
120 payload_words = ROUNDUP_BYTES_TO_WDS(bytes);
122 /* When we actually allocate memory, we need to allow space for the
124 bytes = bytes + SIZEOF_StgArrWords;
125 /* And we want to align to <alignment> bytes, so we need to allow space
126 to shift up to <alignment - 1> bytes: */
127 bytes = bytes + alignment - 1;
128 /* Now we convert to a number of words: */
129 words = ROUNDUP_BYTES_TO_WDS(bytes);
131 ("ptr" p) = foreign "C" allocatePinned(MyCapability() "ptr", words) [];
132 TICK_ALLOC_PRIM(SIZEOF_StgArrWords,WDS(payload_words),0);
134 /* Now we need to move p forward so that the payload is aligned
135 to <alignment> bytes. Note that we are assuming that
136 <alignment> is a power of 2, which is technically not guaranteed */
137 p = p + ((-p - SIZEOF_StgArrWords) & (alignment - 1));
139 SET_HDR(p, stg_ARR_WORDS_info, W_[CCCS]);
140 StgArrWords_bytes(p) = n;
146 W_ words, n, init, arr, p, size;
147 /* Args: R1 = words, R2 = initialisation value */
150 MAYBE_GC(R2_PTR,stg_newArrayzh);
152 // the mark area contains one byte for each 2^MUT_ARR_PTRS_CARD_BITS words
153 // in the array, making sure we round up, and then rounding up to a whole
155 size = n + mutArrPtrsCardWords(n);
156 words = BYTES_TO_WDS(SIZEOF_StgMutArrPtrs) + size;
157 ("ptr" arr) = foreign "C" allocate(MyCapability() "ptr",words) [R2];
158 TICK_ALLOC_PRIM(SIZEOF_StgMutArrPtrs, WDS(n), 0);
160 SET_HDR(arr, stg_MUT_ARR_PTRS_DIRTY_info, W_[CCCS]);
161 StgMutArrPtrs_ptrs(arr) = n;
162 StgMutArrPtrs_size(arr) = size;
164 // Initialise all elements of the the array with the value in R2
166 p = arr + SIZEOF_StgMutArrPtrs;
168 if (p < arr + WDS(words)) {
173 // Initialise the mark bits with 0
175 if (p < arr + WDS(size)) {
184 stg_unsafeThawArrayzh
186 // SUBTLETY TO DO WITH THE OLD GEN MUTABLE LIST
188 // A MUT_ARR_PTRS lives on the mutable list, but a MUT_ARR_PTRS_FROZEN
189 // normally doesn't. However, when we freeze a MUT_ARR_PTRS, we leave
190 // it on the mutable list for the GC to remove (removing something from
191 // the mutable list is not easy).
193 // So that we can tell whether a MUT_ARR_PTRS_FROZEN is on the mutable list,
194 // when we freeze it we set the info ptr to be MUT_ARR_PTRS_FROZEN0
195 // to indicate that it is still on the mutable list.
197 // So, when we thaw a MUT_ARR_PTRS_FROZEN, we must cope with two cases:
198 // either it is on a mut_list, or it isn't. We adopt the convention that
199 // the closure type is MUT_ARR_PTRS_FROZEN0 if it is on the mutable list,
200 // and MUT_ARR_PTRS_FROZEN otherwise. In fact it wouldn't matter if
201 // we put it on the mutable list more than once, but it would get scavenged
202 // multiple times during GC, which would be unnecessarily slow.
204 if (StgHeader_info(R1) != stg_MUT_ARR_PTRS_FROZEN0_info) {
205 SET_INFO(R1,stg_MUT_ARR_PTRS_DIRTY_info);
206 recordMutable(R1, R1);
207 // must be done after SET_INFO, because it ASSERTs closure_MUTABLE()
210 SET_INFO(R1,stg_MUT_ARR_PTRS_DIRTY_info);
215 /* -----------------------------------------------------------------------------
217 -------------------------------------------------------------------------- */
222 /* Args: R1 = initialisation value */
224 ALLOC_PRIM( SIZEOF_StgMutVar, R1_PTR, stg_newMutVarzh);
226 mv = Hp - SIZEOF_StgMutVar + WDS(1);
227 SET_HDR(mv,stg_MUT_VAR_DIRTY_info,W_[CCCS]);
228 StgMutVar_var(mv) = R1;
234 /* MutVar# s a -> a -> a -> State# s -> (# State#, Int#, a #) */
242 (h) = foreign "C" cas(mv + SIZEOF_StgHeader + OFFSET_StgMutVar_var,
252 stg_atomicModifyMutVarzh
254 W_ mv, f, z, x, y, r, h;
255 /* Args: R1 :: MutVar#, R2 :: a -> (a,b) */
257 /* If x is the current contents of the MutVar#, then
258 We want to make the new contents point to
262 and the return value is
266 obviously we can share (f x).
268 z = [stg_ap_2 f x] (max (HS + 2) MIN_UPD_SIZE)
269 y = [stg_sel_0 z] (max (HS + 1) MIN_UPD_SIZE)
270 r = [stg_sel_1 z] (max (HS + 1) MIN_UPD_SIZE)
274 #define THUNK_1_SIZE (SIZEOF_StgThunkHeader + WDS(MIN_UPD_SIZE))
275 #define TICK_ALLOC_THUNK_1() TICK_ALLOC_UP_THK(WDS(1),WDS(MIN_UPD_SIZE-1))
277 #define THUNK_1_SIZE (SIZEOF_StgThunkHeader + WDS(1))
278 #define TICK_ALLOC_THUNK_1() TICK_ALLOC_UP_THK(WDS(1),0)
282 #define THUNK_2_SIZE (SIZEOF_StgThunkHeader + WDS(MIN_UPD_SIZE))
283 #define TICK_ALLOC_THUNK_2() TICK_ALLOC_UP_THK(WDS(2),WDS(MIN_UPD_SIZE-2))
285 #define THUNK_2_SIZE (SIZEOF_StgThunkHeader + WDS(2))
286 #define TICK_ALLOC_THUNK_2() TICK_ALLOC_UP_THK(WDS(2),0)
289 #define SIZE (THUNK_2_SIZE + THUNK_1_SIZE + THUNK_1_SIZE)
291 HP_CHK_GEN_TICKY(SIZE, R1_PTR & R2_PTR, stg_atomicModifyMutVarzh);
296 TICK_ALLOC_THUNK_2();
297 CCCS_ALLOC(THUNK_2_SIZE);
298 z = Hp - THUNK_2_SIZE + WDS(1);
299 SET_HDR(z, stg_ap_2_upd_info, W_[CCCS]);
300 LDV_RECORD_CREATE(z);
301 StgThunk_payload(z,0) = f;
303 TICK_ALLOC_THUNK_1();
304 CCCS_ALLOC(THUNK_1_SIZE);
305 y = z - THUNK_1_SIZE;
306 SET_HDR(y, stg_sel_0_upd_info, W_[CCCS]);
307 LDV_RECORD_CREATE(y);
308 StgThunk_payload(y,0) = z;
310 TICK_ALLOC_THUNK_1();
311 CCCS_ALLOC(THUNK_1_SIZE);
312 r = y - THUNK_1_SIZE;
313 SET_HDR(r, stg_sel_1_upd_info, W_[CCCS]);
314 LDV_RECORD_CREATE(r);
315 StgThunk_payload(r,0) = z;
318 x = StgMutVar_var(mv);
319 StgThunk_payload(z,1) = x;
321 (h) = foreign "C" cas(mv + SIZEOF_StgHeader + OFFSET_StgMutVar_var, x, y) [];
322 if (h != x) { goto retry; }
324 StgMutVar_var(mv) = y;
327 if (GET_INFO(mv) == stg_MUT_VAR_CLEAN_info) {
328 foreign "C" dirty_MUT_VAR(BaseReg "ptr", mv "ptr") [];
334 /* -----------------------------------------------------------------------------
335 Weak Pointer Primitives
336 -------------------------------------------------------------------------- */
338 STRING(stg_weak_msg,"New weak pointer at %p\n")
344 R3 = finalizer (or NULL)
349 R3 = stg_NO_FINALIZER_closure;
352 ALLOC_PRIM( SIZEOF_StgWeak, R1_PTR & R2_PTR & R3_PTR, stg_mkWeakzh );
354 w = Hp - SIZEOF_StgWeak + WDS(1);
355 SET_HDR(w, stg_WEAK_info, W_[CCCS]);
357 // We don't care about cfinalizer here.
358 // Should StgWeak_cfinalizer(w) be stg_NO_FINALIZER_closure or
362 StgWeak_value(w) = R2;
363 StgWeak_finalizer(w) = R3;
364 StgWeak_cfinalizer(w) = stg_NO_FINALIZER_closure;
366 ACQUIRE_LOCK(sm_mutex);
367 StgWeak_link(w) = W_[weak_ptr_list];
368 W_[weak_ptr_list] = w;
369 RELEASE_LOCK(sm_mutex);
371 IF_DEBUG(weak, foreign "C" debugBelch(stg_weak_msg,w) []);
376 stg_mkWeakForeignEnvzh
382 R5 = has environment (0 or 1)
385 W_ w, payload_words, words, p;
387 W_ key, val, fptr, ptr, flag, eptr;
396 ALLOC_PRIM( SIZEOF_StgWeak, R1_PTR & R2_PTR, stg_mkWeakForeignEnvzh );
398 w = Hp - SIZEOF_StgWeak + WDS(1);
399 SET_HDR(w, stg_WEAK_info, W_[CCCS]);
402 words = BYTES_TO_WDS(SIZEOF_StgArrWords) + payload_words;
403 ("ptr" p) = foreign "C" allocate(MyCapability() "ptr", words) [];
405 TICK_ALLOC_PRIM(SIZEOF_StgArrWords,WDS(payload_words),0);
406 SET_HDR(p, stg_ARR_WORDS_info, W_[CCCS]);
408 StgArrWords_bytes(p) = WDS(payload_words);
409 StgArrWords_payload(p,0) = fptr;
410 StgArrWords_payload(p,1) = ptr;
411 StgArrWords_payload(p,2) = eptr;
412 StgArrWords_payload(p,3) = flag;
414 // We don't care about the value here.
415 // Should StgWeak_value(w) be stg_NO_FINALIZER_closure or something else?
417 StgWeak_key(w) = key;
418 StgWeak_value(w) = val;
419 StgWeak_finalizer(w) = stg_NO_FINALIZER_closure;
420 StgWeak_cfinalizer(w) = p;
422 ACQUIRE_LOCK(sm_mutex);
423 StgWeak_link(w) = W_[weak_ptr_list];
424 W_[weak_ptr_list] = w;
425 RELEASE_LOCK(sm_mutex);
427 IF_DEBUG(weak, foreign "C" debugBelch(stg_weak_msg,w) []);
441 if (GET_INFO(w) == stg_DEAD_WEAK_info) {
442 RET_NP(0,stg_NO_FINALIZER_closure);
448 // A weak pointer is inherently used, so we do not need to call
449 // LDV_recordDead_FILL_SLOP_DYNAMIC():
450 // LDV_recordDead_FILL_SLOP_DYNAMIC((StgClosure *)w);
451 // or, LDV_recordDead():
452 // LDV_recordDead((StgClosure *)w, sizeofW(StgWeak) - sizeofW(StgProfHeader));
453 // Furthermore, when PROFILING is turned on, dead weak pointers are exactly as
454 // large as weak pointers, so there is no need to fill the slop, either.
455 // See stg_DEAD_WEAK_info in StgMiscClosures.hc.
459 // Todo: maybe use SET_HDR() and remove LDV_recordCreate()?
461 SET_INFO(w,stg_DEAD_WEAK_info);
462 LDV_RECORD_CREATE(w);
464 f = StgWeak_finalizer(w);
465 arr = StgWeak_cfinalizer(w);
467 StgDeadWeak_link(w) = StgWeak_link(w);
469 if (arr != stg_NO_FINALIZER_closure) {
470 foreign "C" runCFinalizer(StgArrWords_payload(arr,0),
471 StgArrWords_payload(arr,1),
472 StgArrWords_payload(arr,2),
473 StgArrWords_payload(arr,3)) [];
476 /* return the finalizer */
477 if (f == stg_NO_FINALIZER_closure) {
478 RET_NP(0,stg_NO_FINALIZER_closure);
490 if (GET_INFO(w) == stg_WEAK_info) {
492 val = StgWeak_value(w);
500 /* -----------------------------------------------------------------------------
501 Floating point operations.
502 -------------------------------------------------------------------------- */
504 stg_decodeFloatzuIntzh
511 STK_CHK_GEN( WDS(2), NO_PTRS, stg_decodeFloatzuIntzh );
513 mp_tmp1 = Sp - WDS(1);
514 mp_tmp_w = Sp - WDS(2);
516 /* arguments: F1 = Float# */
519 /* Perform the operation */
520 foreign "C" __decodeFloat_Int(mp_tmp1 "ptr", mp_tmp_w "ptr", arg) [];
522 /* returns: (Int# (mantissa), Int# (exponent)) */
523 RET_NN(W_[mp_tmp1], W_[mp_tmp_w]);
526 stg_decodeDoublezu2Intzh
535 STK_CHK_GEN( WDS(4), NO_PTRS, stg_decodeDoublezu2Intzh );
537 mp_tmp1 = Sp - WDS(1);
538 mp_tmp2 = Sp - WDS(2);
539 mp_result1 = Sp - WDS(3);
540 mp_result2 = Sp - WDS(4);
542 /* arguments: D1 = Double# */
545 /* Perform the operation */
546 foreign "C" __decodeDouble_2Int(mp_tmp1 "ptr", mp_tmp2 "ptr",
547 mp_result1 "ptr", mp_result2 "ptr",
551 (Int# (mant sign), Word# (mant high), Word# (mant low), Int# (expn)) */
552 RET_NNNN(W_[mp_tmp1], W_[mp_tmp2], W_[mp_result1], W_[mp_result2]);
555 /* -----------------------------------------------------------------------------
556 * Concurrency primitives
557 * -------------------------------------------------------------------------- */
561 /* args: R1 = closure to spark */
563 MAYBE_GC(R1_PTR, stg_forkzh);
569 ("ptr" threadid) = foreign "C" createIOThread( MyCapability() "ptr",
570 RtsFlags_GcFlags_initialStkSize(RtsFlags),
573 /* start blocked if the current thread is blocked */
574 StgTSO_flags(threadid) = %lobits16(
575 TO_W_(StgTSO_flags(threadid)) |
576 TO_W_(StgTSO_flags(CurrentTSO)) & (TSO_BLOCKEX | TSO_INTERRUPTIBLE));
578 foreign "C" scheduleThread(MyCapability() "ptr", threadid "ptr") [];
580 // context switch soon, but not immediately: we don't want every
581 // forkIO to force a context-switch.
582 Capability_context_switch(MyCapability()) = 1 :: CInt;
589 /* args: R1 = cpu, R2 = closure to spark */
591 MAYBE_GC(R2_PTR, stg_forkOnzh);
599 ("ptr" threadid) = foreign "C" createIOThread( MyCapability() "ptr",
600 RtsFlags_GcFlags_initialStkSize(RtsFlags),
603 /* start blocked if the current thread is blocked */
604 StgTSO_flags(threadid) = %lobits16(
605 TO_W_(StgTSO_flags(threadid)) |
606 TO_W_(StgTSO_flags(CurrentTSO)) & (TSO_BLOCKEX | TSO_INTERRUPTIBLE));
608 foreign "C" scheduleThreadOn(MyCapability() "ptr", cpu, threadid "ptr") [];
610 // context switch soon, but not immediately: we don't want every
611 // forkIO to force a context-switch.
612 Capability_context_switch(MyCapability()) = 1 :: CInt;
619 jump stg_yield_noregs;
634 foreign "C" labelThread(R1 "ptr", R2 "ptr") [];
636 jump %ENTRY_CODE(Sp(0));
639 stg_isCurrentThreadBoundzh
643 (r) = foreign "C" isThreadBound(CurrentTSO) [];
649 /* args: R1 :: ThreadId# */
657 what_next = TO_W_(StgTSO_what_next(tso));
658 why_blocked = TO_W_(StgTSO_why_blocked(tso));
659 // Note: these two reads are not atomic, so they might end up
660 // being inconsistent. It doesn't matter, since we
661 // only return one or the other. If we wanted to return the
662 // contents of block_info too, then we'd have to do some synchronisation.
664 if (what_next == ThreadComplete) {
665 ret = 16; // NB. magic, matches up with GHC.Conc.threadStatus
667 if (what_next == ThreadKilled) {
674 cap = TO_W_(Capability_no(StgTSO_cap(tso)));
676 if ((TO_W_(StgTSO_flags(tso)) & TSO_LOCKED) != 0) {
682 RET_NNN(ret,cap,locked);
685 /* -----------------------------------------------------------------------------
687 * -------------------------------------------------------------------------- */
691 // Catch retry frame ------------------------------------------------------------
693 INFO_TABLE_RET(stg_catch_retry_frame, CATCH_RETRY_FRAME,
694 #if defined(PROFILING)
695 W_ unused1, W_ unused2,
697 W_ unused3, P_ unused4, P_ unused5)
699 W_ r, frame, trec, outer;
702 trec = StgTSO_trec(CurrentTSO);
703 outer = StgTRecHeader_enclosing_trec(trec);
704 (r) = foreign "C" stmCommitNestedTransaction(MyCapability() "ptr", trec "ptr") [];
706 /* Succeeded (either first branch or second branch) */
707 StgTSO_trec(CurrentTSO) = outer;
708 Sp = Sp + SIZEOF_StgCatchRetryFrame;
709 jump %ENTRY_CODE(Sp(SP_OFF));
711 /* Did not commit: re-execute */
713 ("ptr" new_trec) = foreign "C" stmStartTransaction(MyCapability() "ptr", outer "ptr") [];
714 StgTSO_trec(CurrentTSO) = new_trec;
715 if (StgCatchRetryFrame_running_alt_code(frame) != 0::I32) {
716 R1 = StgCatchRetryFrame_alt_code(frame);
718 R1 = StgCatchRetryFrame_first_code(frame);
725 // Atomically frame ------------------------------------------------------------
727 INFO_TABLE_RET(stg_atomically_frame, ATOMICALLY_FRAME,
728 #if defined(PROFILING)
729 W_ unused1, W_ unused2,
731 P_ code, P_ next_invariant_to_check, P_ result)
733 W_ frame, trec, valid, next_invariant, q, outer;
736 trec = StgTSO_trec(CurrentTSO);
738 outer = StgTRecHeader_enclosing_trec(trec);
740 if (outer == NO_TREC) {
741 /* First time back at the atomically frame -- pick up invariants */
742 ("ptr" q) = foreign "C" stmGetInvariantsToCheck(MyCapability() "ptr", trec "ptr") [];
743 StgAtomicallyFrame_next_invariant_to_check(frame) = q;
744 StgAtomicallyFrame_result(frame) = result;
747 /* Second/subsequent time back at the atomically frame -- abort the
748 * tx that's checking the invariant and move on to the next one */
749 StgTSO_trec(CurrentTSO) = outer;
750 q = StgAtomicallyFrame_next_invariant_to_check(frame);
751 StgInvariantCheckQueue_my_execution(q) = trec;
752 foreign "C" stmAbortTransaction(MyCapability() "ptr", trec "ptr") [];
753 /* Don't free trec -- it's linked from q and will be stashed in the
754 * invariant if we eventually commit. */
755 q = StgInvariantCheckQueue_next_queue_entry(q);
756 StgAtomicallyFrame_next_invariant_to_check(frame) = q;
760 q = StgAtomicallyFrame_next_invariant_to_check(frame);
762 if (q != END_INVARIANT_CHECK_QUEUE) {
763 /* We can't commit yet: another invariant to check */
764 ("ptr" trec) = foreign "C" stmStartTransaction(MyCapability() "ptr", trec "ptr") [];
765 StgTSO_trec(CurrentTSO) = trec;
767 next_invariant = StgInvariantCheckQueue_invariant(q);
768 R1 = StgAtomicInvariant_code(next_invariant);
773 /* We've got no more invariants to check, try to commit */
774 (valid) = foreign "C" stmCommitTransaction(MyCapability() "ptr", trec "ptr") [];
776 /* Transaction was valid: commit succeeded */
777 StgTSO_trec(CurrentTSO) = NO_TREC;
778 R1 = StgAtomicallyFrame_result(frame);
779 Sp = Sp + SIZEOF_StgAtomicallyFrame;
780 jump %ENTRY_CODE(Sp(SP_OFF));
782 /* Transaction was not valid: try again */
783 ("ptr" trec) = foreign "C" stmStartTransaction(MyCapability() "ptr", NO_TREC "ptr") [];
784 StgTSO_trec(CurrentTSO) = trec;
785 StgAtomicallyFrame_next_invariant_to_check(frame) = END_INVARIANT_CHECK_QUEUE;
786 R1 = StgAtomicallyFrame_code(frame);
792 INFO_TABLE_RET(stg_atomically_waiting_frame, ATOMICALLY_FRAME,
793 #if defined(PROFILING)
794 W_ unused1, W_ unused2,
796 P_ code, P_ next_invariant_to_check, P_ result)
798 W_ frame, trec, valid;
802 /* The TSO is currently waiting: should we stop waiting? */
803 (valid) = foreign "C" stmReWait(MyCapability() "ptr", CurrentTSO "ptr") [];
805 /* Previous attempt is still valid: no point trying again yet */
806 jump stg_block_noregs;
808 /* Previous attempt is no longer valid: try again */
809 ("ptr" trec) = foreign "C" stmStartTransaction(MyCapability() "ptr", NO_TREC "ptr") [];
810 StgTSO_trec(CurrentTSO) = trec;
811 StgHeader_info(frame) = stg_atomically_frame_info;
812 R1 = StgAtomicallyFrame_code(frame);
817 // STM catch frame --------------------------------------------------------------
821 /* Catch frames are very similar to update frames, but when entering
822 * one we just pop the frame off the stack and perform the correct
823 * kind of return to the activation record underneath us on the stack.
826 INFO_TABLE_RET(stg_catch_stm_frame, CATCH_STM_FRAME,
827 #if defined(PROFILING)
828 W_ unused1, W_ unused2,
830 P_ unused3, P_ unused4)
832 W_ r, frame, trec, outer;
834 trec = StgTSO_trec(CurrentTSO);
835 outer = StgTRecHeader_enclosing_trec(trec);
836 (r) = foreign "C" stmCommitNestedTransaction(MyCapability() "ptr", trec "ptr") [];
838 /* Commit succeeded */
839 StgTSO_trec(CurrentTSO) = outer;
840 Sp = Sp + SIZEOF_StgCatchSTMFrame;
845 ("ptr" new_trec) = foreign "C" stmStartTransaction(MyCapability() "ptr", outer "ptr") [];
846 StgTSO_trec(CurrentTSO) = new_trec;
847 R1 = StgCatchSTMFrame_code(frame);
853 // Primop definition ------------------------------------------------------------
861 // stmStartTransaction may allocate
862 MAYBE_GC (R1_PTR, stg_atomicallyzh);
864 /* Args: R1 = m :: STM a */
865 STK_CHK_GEN(SIZEOF_StgAtomicallyFrame + WDS(1), R1_PTR, stg_atomicallyzh);
867 old_trec = StgTSO_trec(CurrentTSO);
869 /* Nested transactions are not allowed; raise an exception */
870 if (old_trec != NO_TREC) {
871 R1 = base_ControlziExceptionziBase_nestedAtomically_closure;
875 /* Set up the atomically frame */
876 Sp = Sp - SIZEOF_StgAtomicallyFrame;
879 SET_HDR(frame,stg_atomically_frame_info, W_[CCCS]);
880 StgAtomicallyFrame_code(frame) = R1;
881 StgAtomicallyFrame_result(frame) = NO_TREC;
882 StgAtomicallyFrame_next_invariant_to_check(frame) = END_INVARIANT_CHECK_QUEUE;
884 /* Start the memory transcation */
885 ("ptr" new_trec) = foreign "C" stmStartTransaction(MyCapability() "ptr", old_trec "ptr") [R1];
886 StgTSO_trec(CurrentTSO) = new_trec;
888 /* Apply R1 to the realworld token */
897 /* Args: R1 :: STM a */
898 /* Args: R2 :: Exception -> STM a */
899 STK_CHK_GEN(SIZEOF_StgCatchSTMFrame + WDS(1), R1_PTR & R2_PTR, stg_catchSTMzh);
901 /* Set up the catch frame */
902 Sp = Sp - SIZEOF_StgCatchSTMFrame;
905 SET_HDR(frame, stg_catch_stm_frame_info, W_[CCCS]);
906 StgCatchSTMFrame_handler(frame) = R2;
907 StgCatchSTMFrame_code(frame) = R1;
909 /* Start a nested transaction to run the body of the try block in */
912 cur_trec = StgTSO_trec(CurrentTSO);
913 ("ptr" new_trec) = foreign "C" stmStartTransaction(MyCapability() "ptr", cur_trec "ptr");
914 StgTSO_trec(CurrentTSO) = new_trec;
916 /* Apply R1 to the realworld token */
927 // stmStartTransaction may allocate
928 MAYBE_GC (R1_PTR & R2_PTR, stg_catchRetryzh);
930 /* Args: R1 :: STM a */
931 /* Args: R2 :: STM a */
932 STK_CHK_GEN(SIZEOF_StgCatchRetryFrame + WDS(1), R1_PTR & R2_PTR, stg_catchRetryzh);
934 /* Start a nested transaction within which to run the first code */
935 trec = StgTSO_trec(CurrentTSO);
936 ("ptr" new_trec) = foreign "C" stmStartTransaction(MyCapability() "ptr", trec "ptr") [R1,R2];
937 StgTSO_trec(CurrentTSO) = new_trec;
939 /* Set up the catch-retry frame */
940 Sp = Sp - SIZEOF_StgCatchRetryFrame;
943 SET_HDR(frame, stg_catch_retry_frame_info, W_[CCCS]);
944 StgCatchRetryFrame_running_alt_code(frame) = 0 :: CInt; // false;
945 StgCatchRetryFrame_first_code(frame) = R1;
946 StgCatchRetryFrame_alt_code(frame) = R2;
948 /* Apply R1 to the realworld token */
961 MAYBE_GC (NO_PTRS, stg_retryzh); // STM operations may allocate
963 // Find the enclosing ATOMICALLY_FRAME or CATCH_RETRY_FRAME
966 (frame_type) = foreign "C" findRetryFrameHelper(MyCapability(), CurrentTSO "ptr") [];
969 trec = StgTSO_trec(CurrentTSO);
970 outer = StgTRecHeader_enclosing_trec(trec);
972 if (frame_type == CATCH_RETRY_FRAME) {
973 // The retry reaches a CATCH_RETRY_FRAME before the atomic frame
974 ASSERT(outer != NO_TREC);
975 // Abort the transaction attempting the current branch
976 foreign "C" stmAbortTransaction(MyCapability() "ptr", trec "ptr") [];
977 foreign "C" stmFreeAbortedTRec(MyCapability() "ptr", trec "ptr") [];
978 if (!StgCatchRetryFrame_running_alt_code(frame) != 0::I32) {
979 // Retry in the first branch: try the alternative
980 ("ptr" trec) = foreign "C" stmStartTransaction(MyCapability() "ptr", outer "ptr") [];
981 StgTSO_trec(CurrentTSO) = trec;
982 StgCatchRetryFrame_running_alt_code(frame) = 1 :: CInt; // true;
983 R1 = StgCatchRetryFrame_alt_code(frame);
986 // Retry in the alternative code: propagate the retry
987 StgTSO_trec(CurrentTSO) = outer;
988 Sp = Sp + SIZEOF_StgCatchRetryFrame;
989 goto retry_pop_stack;
993 // We've reached the ATOMICALLY_FRAME: attempt to wait
994 ASSERT(frame_type == ATOMICALLY_FRAME);
995 if (outer != NO_TREC) {
996 // We called retry while checking invariants, so abort the current
997 // invariant check (merging its TVar accesses into the parents read
998 // set so we'll wait on them)
999 foreign "C" stmAbortTransaction(MyCapability() "ptr", trec "ptr") [];
1000 foreign "C" stmFreeAbortedTRec(MyCapability() "ptr", trec "ptr") [];
1002 StgTSO_trec(CurrentTSO) = trec;
1003 outer = StgTRecHeader_enclosing_trec(trec);
1005 ASSERT(outer == NO_TREC);
1007 (r) = foreign "C" stmWait(MyCapability() "ptr", CurrentTSO "ptr", trec "ptr") [];
1009 // Transaction was valid: stmWait put us on the TVars' queues, we now block
1010 StgHeader_info(frame) = stg_atomically_waiting_frame_info;
1012 // Fix up the stack in the unregisterised case: the return convention is different.
1013 R3 = trec; // passing to stmWaitUnblock()
1014 jump stg_block_stmwait;
1016 // Transaction was not valid: retry immediately
1017 ("ptr" trec) = foreign "C" stmStartTransaction(MyCapability() "ptr", outer "ptr") [];
1018 StgTSO_trec(CurrentTSO) = trec;
1019 R1 = StgAtomicallyFrame_code(frame);
1030 /* Args: R1 = invariant closure */
1031 MAYBE_GC (R1_PTR, stg_checkzh);
1033 trec = StgTSO_trec(CurrentTSO);
1035 foreign "C" stmAddInvariantToCheck(MyCapability() "ptr",
1039 jump %ENTRY_CODE(Sp(0));
1048 /* Args: R1 = initialisation value */
1050 MAYBE_GC (R1_PTR, stg_newTVarzh);
1052 ("ptr" tv) = foreign "C" stmNewTVar(MyCapability() "ptr", new_value "ptr") [];
1063 /* Args: R1 = TVar closure */
1065 MAYBE_GC (R1_PTR, stg_readTVarzh); // Call to stmReadTVar may allocate
1066 trec = StgTSO_trec(CurrentTSO);
1068 ("ptr" result) = foreign "C" stmReadTVar(MyCapability() "ptr", trec "ptr", tvar "ptr") [];
1078 result = StgTVar_current_value(R1);
1079 if (%INFO_PTR(result) == stg_TREC_HEADER_info) {
1091 /* Args: R1 = TVar closure */
1092 /* R2 = New value */
1094 MAYBE_GC (R1_PTR & R2_PTR, stg_writeTVarzh); // Call to stmWriteTVar may allocate
1095 trec = StgTSO_trec(CurrentTSO);
1098 foreign "C" stmWriteTVar(MyCapability() "ptr", trec "ptr", tvar "ptr", new_value "ptr") [];
1100 jump %ENTRY_CODE(Sp(0));
1104 /* -----------------------------------------------------------------------------
1107 * take & putMVar work as follows. Firstly, an important invariant:
1109 * If the MVar is full, then the blocking queue contains only
1110 * threads blocked on putMVar, and if the MVar is empty then the
1111 * blocking queue contains only threads blocked on takeMVar.
1114 * MVar empty : then add ourselves to the blocking queue
1115 * MVar full : remove the value from the MVar, and
1116 * blocking queue empty : return
1117 * blocking queue non-empty : perform the first blocked putMVar
1118 * from the queue, and wake up the
1119 * thread (MVar is now full again)
1121 * putMVar is just the dual of the above algorithm.
1123 * How do we "perform a putMVar"? Well, we have to fiddle around with
1124 * the stack of the thread waiting to do the putMVar. See
1125 * stg_block_putmvar and stg_block_takemvar in HeapStackCheck.c for
1126 * the stack layout, and the PerformPut and PerformTake macros below.
1128 * It is important that a blocked take or put is woken up with the
1129 * take/put already performed, because otherwise there would be a
1130 * small window of vulnerability where the thread could receive an
1131 * exception and never perform its take or put, and we'd end up with a
1134 * -------------------------------------------------------------------------- */
1138 /* args: R1 = MVar closure */
1140 if (StgMVar_value(R1) == stg_END_TSO_QUEUE_closure) {
1152 ALLOC_PRIM ( SIZEOF_StgMVar, NO_PTRS, stg_newMVarzh );
1154 mvar = Hp - SIZEOF_StgMVar + WDS(1);
1155 SET_HDR(mvar,stg_MVAR_DIRTY_info,W_[CCCS]);
1156 // MVARs start dirty: generation 0 has no mutable list
1157 StgMVar_head(mvar) = stg_END_TSO_QUEUE_closure;
1158 StgMVar_tail(mvar) = stg_END_TSO_QUEUE_closure;
1159 StgMVar_value(mvar) = stg_END_TSO_QUEUE_closure;
1164 #define PerformTake(stack, value) \
1166 sp = StgStack_sp(stack); \
1167 W_[sp + WDS(1)] = value; \
1168 W_[sp + WDS(0)] = stg_gc_unpt_r1_info;
1170 #define PerformPut(stack,lval) \
1172 sp = StgStack_sp(stack) + WDS(3); \
1173 StgStack_sp(stack) = sp; \
1174 lval = W_[sp - WDS(1)];
1178 W_ mvar, val, info, tso, q;
1180 /* args: R1 = MVar closure */
1183 #if defined(THREADED_RTS)
1184 ("ptr" info) = foreign "C" lockClosure(mvar "ptr") [];
1186 info = GET_INFO(mvar);
1189 if (info == stg_MVAR_CLEAN_info) {
1190 foreign "C" dirty_MVAR(BaseReg "ptr", mvar "ptr") [];
1193 /* If the MVar is empty, put ourselves on its blocking queue,
1194 * and wait until we're woken up.
1196 if (StgMVar_value(mvar) == stg_END_TSO_QUEUE_closure) {
1198 // Note [mvar-heap-check] We want to do the heap check in the
1199 // branch here, to avoid the conditional in the common case.
1200 // However, we've already locked the MVar above, so we better
1201 // be careful to unlock it again if the the heap check fails.
1202 // Unfortunately we don't have an easy way to inject any code
1203 // into the heap check generated by the code generator, so we
1204 // have to do it in stg_gc_gen (see HeapStackCheck.cmm).
1205 HP_CHK_GEN_TICKY(SIZEOF_StgMVarTSOQueue, R1_PTR, stg_takeMVarzh);
1207 q = Hp - SIZEOF_StgMVarTSOQueue + WDS(1);
1209 SET_HDR(q, stg_MVAR_TSO_QUEUE_info, CCS_SYSTEM);
1210 StgMVarTSOQueue_link(q) = END_TSO_QUEUE;
1211 StgMVarTSOQueue_tso(q) = CurrentTSO;
1213 if (StgMVar_head(mvar) == stg_END_TSO_QUEUE_closure) {
1214 StgMVar_head(mvar) = q;
1216 StgMVarTSOQueue_link(StgMVar_tail(mvar)) = q;
1217 foreign "C" recordClosureMutated(MyCapability() "ptr",
1218 StgMVar_tail(mvar)) [];
1220 StgTSO__link(CurrentTSO) = q;
1221 StgTSO_block_info(CurrentTSO) = mvar;
1222 StgTSO_why_blocked(CurrentTSO) = BlockedOnMVar::I16;
1223 StgMVar_tail(mvar) = q;
1226 jump stg_block_takemvar;
1229 /* we got the value... */
1230 val = StgMVar_value(mvar);
1232 q = StgMVar_head(mvar);
1234 if (q == stg_END_TSO_QUEUE_closure) {
1235 /* No further putMVars, MVar is now empty */
1236 StgMVar_value(mvar) = stg_END_TSO_QUEUE_closure;
1237 unlockClosure(mvar, stg_MVAR_DIRTY_info);
1240 if (StgHeader_info(q) == stg_IND_info ||
1241 StgHeader_info(q) == stg_MSG_NULL_info) {
1242 q = StgInd_indirectee(q);
1246 // There are putMVar(s) waiting... wake up the first thread on the queue
1248 tso = StgMVarTSOQueue_tso(q);
1249 StgMVar_head(mvar) = StgMVarTSOQueue_link(q);
1250 if (StgMVar_head(mvar) == stg_END_TSO_QUEUE_closure) {
1251 StgMVar_tail(mvar) = stg_END_TSO_QUEUE_closure;
1254 ASSERT(StgTSO_why_blocked(tso) == BlockedOnMVar::I16);
1255 ASSERT(StgTSO_block_info(tso) == mvar);
1257 // actually perform the putMVar for the thread that we just woke up
1259 stack = StgTSO_stackobj(tso);
1260 PerformPut(stack, StgMVar_value(mvar));
1262 // indicate that the MVar operation has now completed.
1263 StgTSO__link(tso) = stg_END_TSO_QUEUE_closure;
1265 // no need to mark the TSO dirty, we have only written END_TSO_QUEUE.
1267 foreign "C" tryWakeupThread(MyCapability() "ptr", tso) [];
1269 unlockClosure(mvar, stg_MVAR_DIRTY_info);
1276 W_ mvar, val, info, tso, q;
1278 /* args: R1 = MVar closure */
1281 #if defined(THREADED_RTS)
1282 ("ptr" info) = foreign "C" lockClosure(mvar "ptr") [];
1284 info = GET_INFO(mvar);
1287 /* If the MVar is empty, put ourselves on its blocking queue,
1288 * and wait until we're woken up.
1290 if (StgMVar_value(mvar) == stg_END_TSO_QUEUE_closure) {
1291 #if defined(THREADED_RTS)
1292 unlockClosure(mvar, info);
1294 /* HACK: we need a pointer to pass back,
1295 * so we abuse NO_FINALIZER_closure
1297 RET_NP(0, stg_NO_FINALIZER_closure);
1300 if (info == stg_MVAR_CLEAN_info) {
1301 foreign "C" dirty_MVAR(BaseReg "ptr", mvar "ptr") [];
1304 /* we got the value... */
1305 val = StgMVar_value(mvar);
1307 q = StgMVar_head(mvar);
1309 if (q == stg_END_TSO_QUEUE_closure) {
1310 /* No further putMVars, MVar is now empty */
1311 StgMVar_value(mvar) = stg_END_TSO_QUEUE_closure;
1312 unlockClosure(mvar, stg_MVAR_DIRTY_info);
1315 if (StgHeader_info(q) == stg_IND_info ||
1316 StgHeader_info(q) == stg_MSG_NULL_info) {
1317 q = StgInd_indirectee(q);
1321 // There are putMVar(s) waiting... wake up the first thread on the queue
1323 tso = StgMVarTSOQueue_tso(q);
1324 StgMVar_head(mvar) = StgMVarTSOQueue_link(q);
1325 if (StgMVar_head(mvar) == stg_END_TSO_QUEUE_closure) {
1326 StgMVar_tail(mvar) = stg_END_TSO_QUEUE_closure;
1329 ASSERT(StgTSO_why_blocked(tso) == BlockedOnMVar::I16);
1330 ASSERT(StgTSO_block_info(tso) == mvar);
1332 // actually perform the putMVar for the thread that we just woke up
1334 stack = StgTSO_stackobj(tso);
1335 PerformPut(stack, StgMVar_value(mvar));
1337 // indicate that the MVar operation has now completed.
1338 StgTSO__link(tso) = stg_END_TSO_QUEUE_closure;
1340 // no need to mark the TSO dirty, we have only written END_TSO_QUEUE.
1342 foreign "C" tryWakeupThread(MyCapability() "ptr", tso) [];
1344 unlockClosure(mvar, stg_MVAR_DIRTY_info);
1351 W_ mvar, val, info, tso, q;
1353 /* args: R1 = MVar, R2 = value */
1357 #if defined(THREADED_RTS)
1358 ("ptr" info) = foreign "C" lockClosure(mvar "ptr") [];
1360 info = GET_INFO(mvar);
1363 if (info == stg_MVAR_CLEAN_info) {
1364 foreign "C" dirty_MVAR(BaseReg "ptr", mvar "ptr");
1367 if (StgMVar_value(mvar) != stg_END_TSO_QUEUE_closure) {
1369 // see Note [mvar-heap-check] above
1370 HP_CHK_GEN_TICKY(SIZEOF_StgMVarTSOQueue, R1_PTR & R2_PTR, stg_putMVarzh);
1372 q = Hp - SIZEOF_StgMVarTSOQueue + WDS(1);
1374 SET_HDR(q, stg_MVAR_TSO_QUEUE_info, CCS_SYSTEM);
1375 StgMVarTSOQueue_link(q) = END_TSO_QUEUE;
1376 StgMVarTSOQueue_tso(q) = CurrentTSO;
1378 if (StgMVar_head(mvar) == stg_END_TSO_QUEUE_closure) {
1379 StgMVar_head(mvar) = q;
1381 StgMVarTSOQueue_link(StgMVar_tail(mvar)) = q;
1382 foreign "C" recordClosureMutated(MyCapability() "ptr",
1383 StgMVar_tail(mvar)) [];
1385 StgTSO__link(CurrentTSO) = q;
1386 StgTSO_block_info(CurrentTSO) = mvar;
1387 StgTSO_why_blocked(CurrentTSO) = BlockedOnMVar::I16;
1388 StgMVar_tail(mvar) = q;
1392 jump stg_block_putmvar;
1395 q = StgMVar_head(mvar);
1397 if (q == stg_END_TSO_QUEUE_closure) {
1398 /* No further takes, the MVar is now full. */
1399 StgMVar_value(mvar) = val;
1400 unlockClosure(mvar, stg_MVAR_DIRTY_info);
1401 jump %ENTRY_CODE(Sp(0));
1403 if (StgHeader_info(q) == stg_IND_info ||
1404 StgHeader_info(q) == stg_MSG_NULL_info) {
1405 q = StgInd_indirectee(q);
1409 // There are takeMVar(s) waiting: wake up the first one
1411 tso = StgMVarTSOQueue_tso(q);
1412 StgMVar_head(mvar) = StgMVarTSOQueue_link(q);
1413 if (StgMVar_head(mvar) == stg_END_TSO_QUEUE_closure) {
1414 StgMVar_tail(mvar) = stg_END_TSO_QUEUE_closure;
1417 ASSERT(StgTSO_why_blocked(tso) == BlockedOnMVar::I16);
1418 ASSERT(StgTSO_block_info(tso) == mvar);
1420 // actually perform the takeMVar
1422 stack = StgTSO_stackobj(tso);
1423 PerformTake(stack, val);
1425 // indicate that the MVar operation has now completed.
1426 StgTSO__link(tso) = stg_END_TSO_QUEUE_closure;
1428 if (TO_W_(StgStack_dirty(stack)) == 0) {
1429 foreign "C" dirty_STACK(MyCapability() "ptr", stack "ptr") [];
1432 foreign "C" tryWakeupThread(MyCapability() "ptr", tso) [];
1434 unlockClosure(mvar, stg_MVAR_DIRTY_info);
1435 jump %ENTRY_CODE(Sp(0));
1441 W_ mvar, val, info, tso, q;
1443 /* args: R1 = MVar, R2 = value */
1447 #if defined(THREADED_RTS)
1448 ("ptr" info) = foreign "C" lockClosure(mvar "ptr") [];
1450 info = GET_INFO(mvar);
1453 if (info == stg_MVAR_CLEAN_info) {
1454 foreign "C" dirty_MVAR(BaseReg "ptr", mvar "ptr");
1457 if (StgMVar_value(mvar) != stg_END_TSO_QUEUE_closure) {
1458 #if defined(THREADED_RTS)
1459 unlockClosure(mvar, info);
1464 q = StgMVar_head(mvar);
1466 if (q == stg_END_TSO_QUEUE_closure) {
1467 /* No further takes, the MVar is now full. */
1468 StgMVar_value(mvar) = val;
1469 unlockClosure(mvar, stg_MVAR_DIRTY_info);
1472 if (StgHeader_info(q) == stg_IND_info ||
1473 StgHeader_info(q) == stg_MSG_NULL_info) {
1474 q = StgInd_indirectee(q);
1478 // There are takeMVar(s) waiting: wake up the first one
1480 tso = StgMVarTSOQueue_tso(q);
1481 StgMVar_head(mvar) = StgMVarTSOQueue_link(q);
1482 if (StgMVar_head(mvar) == stg_END_TSO_QUEUE_closure) {
1483 StgMVar_tail(mvar) = stg_END_TSO_QUEUE_closure;
1486 ASSERT(StgTSO_why_blocked(tso) == BlockedOnMVar::I16);
1487 ASSERT(StgTSO_block_info(tso) == mvar);
1489 // actually perform the takeMVar
1491 stack = StgTSO_stackobj(tso);
1492 PerformTake(stack, val);
1494 // indicate that the MVar operation has now completed.
1495 StgTSO__link(tso) = stg_END_TSO_QUEUE_closure;
1497 if (TO_W_(StgStack_dirty(stack)) == 0) {
1498 foreign "C" dirty_STACK(MyCapability() "ptr", stack "ptr") [];
1501 foreign "C" tryWakeupThread(MyCapability() "ptr", tso) [];
1503 unlockClosure(mvar, stg_MVAR_DIRTY_info);
1508 /* -----------------------------------------------------------------------------
1509 Stable pointer primitives
1510 ------------------------------------------------------------------------- */
1512 stg_makeStableNamezh
1516 ALLOC_PRIM( SIZEOF_StgStableName, R1_PTR, stg_makeStableNamezh );
1518 (index) = foreign "C" lookupStableName(R1 "ptr") [];
1520 /* Is there already a StableName for this heap object?
1521 * stable_ptr_table is a pointer to an array of snEntry structs.
1523 if ( snEntry_sn_obj(W_[stable_ptr_table] + index*SIZEOF_snEntry) == NULL ) {
1524 sn_obj = Hp - SIZEOF_StgStableName + WDS(1);
1525 SET_HDR(sn_obj, stg_STABLE_NAME_info, W_[CCCS]);
1526 StgStableName_sn(sn_obj) = index;
1527 snEntry_sn_obj(W_[stable_ptr_table] + index*SIZEOF_snEntry) = sn_obj;
1529 sn_obj = snEntry_sn_obj(W_[stable_ptr_table] + index*SIZEOF_snEntry);
1540 MAYBE_GC(R1_PTR, stg_makeStablePtrzh);
1541 ("ptr" sp) = foreign "C" getStablePtr(R1 "ptr") [];
1545 stg_deRefStablePtrzh
1547 /* Args: R1 = the stable ptr */
1550 r = snEntry_addr(W_[stable_ptr_table] + sp*SIZEOF_snEntry);
1554 /* -----------------------------------------------------------------------------
1555 Bytecode object primitives
1556 ------------------------------------------------------------------------- */
1566 W_ bco, bitmap_arr, bytes, words;
1570 words = BYTES_TO_WDS(SIZEOF_StgBCO) + BYTE_ARR_WDS(bitmap_arr);
1573 ALLOC_PRIM( bytes, R1_PTR&R2_PTR&R3_PTR&R5_PTR, stg_newBCOzh );
1575 bco = Hp - bytes + WDS(1);
1576 SET_HDR(bco, stg_BCO_info, W_[CCCS]);
1578 StgBCO_instrs(bco) = R1;
1579 StgBCO_literals(bco) = R2;
1580 StgBCO_ptrs(bco) = R3;
1581 StgBCO_arity(bco) = HALF_W_(R4);
1582 StgBCO_size(bco) = HALF_W_(words);
1584 // Copy the arity/bitmap info into the BCO
1588 if (i < BYTE_ARR_WDS(bitmap_arr)) {
1589 StgBCO_bitmap(bco,i) = StgArrWords_payload(bitmap_arr,i);
1600 // R1 = the BCO# for the AP
1604 // This function is *only* used to wrap zero-arity BCOs in an
1605 // updatable wrapper (see ByteCodeLink.lhs). An AP thunk is always
1606 // saturated and always points directly to a FUN or BCO.
1607 ASSERT(%INFO_TYPE(%GET_STD_INFO(R1)) == HALF_W_(BCO) &&
1608 StgBCO_arity(R1) == HALF_W_(0));
1610 HP_CHK_GEN_TICKY(SIZEOF_StgAP, R1_PTR, stg_mkApUpd0zh);
1611 TICK_ALLOC_UP_THK(0, 0);
1612 CCCS_ALLOC(SIZEOF_StgAP);
1614 ap = Hp - SIZEOF_StgAP + WDS(1);
1615 SET_HDR(ap, stg_AP_info, W_[CCCS]);
1617 StgAP_n_args(ap) = HALF_W_(0);
1625 /* args: R1 = closure to analyze */
1626 // TODO: Consider the absence of ptrs or nonptrs as a special case ?
1628 W_ info, ptrs, nptrs, p, ptrs_arr, nptrs_arr;
1629 info = %GET_STD_INFO(UNTAG(R1));
1631 // Some closures have non-standard layout, so we omit those here.
1633 type = TO_W_(%INFO_TYPE(info));
1634 switch [0 .. N_CLOSURE_TYPES] type {
1635 case THUNK_SELECTOR : {
1640 case THUNK, THUNK_1_0, THUNK_0_1, THUNK_2_0, THUNK_1_1,
1641 THUNK_0_2, THUNK_STATIC, AP, PAP, AP_STACK, BCO : {
1647 ptrs = TO_W_(%INFO_PTRS(info));
1648 nptrs = TO_W_(%INFO_NPTRS(info));
1653 W_ ptrs_arr_sz, ptrs_arr_cards, nptrs_arr_sz;
1654 nptrs_arr_sz = SIZEOF_StgArrWords + WDS(nptrs);
1655 ptrs_arr_cards = mutArrPtrsCardWords(ptrs);
1656 ptrs_arr_sz = SIZEOF_StgMutArrPtrs + WDS(ptrs) + WDS(ptrs_arr_cards);
1658 ALLOC_PRIM (ptrs_arr_sz + nptrs_arr_sz, R1_PTR, stg_unpackClosurezh);
1663 ptrs_arr = Hp - nptrs_arr_sz - ptrs_arr_sz + WDS(1);
1664 nptrs_arr = Hp - nptrs_arr_sz + WDS(1);
1666 SET_HDR(ptrs_arr, stg_MUT_ARR_PTRS_FROZEN_info, W_[CCCS]);
1667 StgMutArrPtrs_ptrs(ptrs_arr) = ptrs;
1668 StgMutArrPtrs_size(ptrs_arr) = ptrs + ptrs_arr_cards;
1673 W_[ptrs_arr + SIZEOF_StgMutArrPtrs + WDS(p)] = StgClosure_payload(clos,p);
1677 /* We can leave the card table uninitialised, since the array is
1678 allocated in the nursery. The GC will fill it in if/when the array
1681 SET_HDR(nptrs_arr, stg_ARR_WORDS_info, W_[CCCS]);
1682 StgArrWords_bytes(nptrs_arr) = WDS(nptrs);
1686 W_[BYTE_ARR_CTS(nptrs_arr) + WDS(p)] = StgClosure_payload(clos, p+ptrs);
1690 RET_NPP(info, ptrs_arr, nptrs_arr);
1693 /* -----------------------------------------------------------------------------
1694 Thread I/O blocking primitives
1695 -------------------------------------------------------------------------- */
1697 /* Add a thread to the end of the blocked queue. (C-- version of the C
1698 * macro in Schedule.h).
1700 #define APPEND_TO_BLOCKED_QUEUE(tso) \
1701 ASSERT(StgTSO__link(tso) == END_TSO_QUEUE); \
1702 if (W_[blocked_queue_hd] == END_TSO_QUEUE) { \
1703 W_[blocked_queue_hd] = tso; \
1705 foreign "C" setTSOLink(MyCapability() "ptr", W_[blocked_queue_tl] "ptr", tso) []; \
1707 W_[blocked_queue_tl] = tso;
1713 foreign "C" barf("waitRead# on threaded RTS") never returns;
1716 ASSERT(StgTSO_why_blocked(CurrentTSO) == NotBlocked::I16);
1717 StgTSO_why_blocked(CurrentTSO) = BlockedOnRead::I16;
1718 StgTSO_block_info(CurrentTSO) = R1;
1719 // No locking - we're not going to use this interface in the
1720 // threaded RTS anyway.
1721 APPEND_TO_BLOCKED_QUEUE(CurrentTSO);
1722 jump stg_block_noregs;
1730 foreign "C" barf("waitWrite# on threaded RTS") never returns;
1733 ASSERT(StgTSO_why_blocked(CurrentTSO) == NotBlocked::I16);
1734 StgTSO_why_blocked(CurrentTSO) = BlockedOnWrite::I16;
1735 StgTSO_block_info(CurrentTSO) = R1;
1736 // No locking - we're not going to use this interface in the
1737 // threaded RTS anyway.
1738 APPEND_TO_BLOCKED_QUEUE(CurrentTSO);
1739 jump stg_block_noregs;
1744 STRING(stg_delayzh_malloc_str, "stg_delayzh")
1747 #ifdef mingw32_HOST_OS
1755 foreign "C" barf("delay# on threaded RTS") never returns;
1758 /* args: R1 (microsecond delay amount) */
1759 ASSERT(StgTSO_why_blocked(CurrentTSO) == NotBlocked::I16);
1760 StgTSO_why_blocked(CurrentTSO) = BlockedOnDelay::I16;
1762 #ifdef mingw32_HOST_OS
1764 /* could probably allocate this on the heap instead */
1765 ("ptr" ares) = foreign "C" stgMallocBytes(SIZEOF_StgAsyncIOResult,
1766 stg_delayzh_malloc_str);
1767 (reqID) = foreign "C" addDelayRequest(R1);
1768 StgAsyncIOResult_reqID(ares) = reqID;
1769 StgAsyncIOResult_len(ares) = 0;
1770 StgAsyncIOResult_errCode(ares) = 0;
1771 StgTSO_block_info(CurrentTSO) = ares;
1773 /* Having all async-blocked threads reside on the blocked_queue
1774 * simplifies matters, so change the status to OnDoProc put the
1775 * delayed thread on the blocked_queue.
1777 StgTSO_why_blocked(CurrentTSO) = BlockedOnDoProc::I16;
1778 APPEND_TO_BLOCKED_QUEUE(CurrentTSO);
1779 jump stg_block_async_void;
1785 (time) = foreign "C" getourtimeofday() [R1];
1786 divisor = TO_W_(RtsFlags_MiscFlags_tickInterval(RtsFlags));
1790 divisor = divisor * 1000;
1791 target = ((R1 + divisor - 1) / divisor) /* divide rounding up */
1792 + time + 1; /* Add 1 as getourtimeofday rounds down */
1793 StgTSO_block_info(CurrentTSO) = target;
1795 /* Insert the new thread in the sleeping queue. */
1797 t = W_[sleeping_queue];
1799 if (t != END_TSO_QUEUE && StgTSO_block_info(t) < target) {
1801 t = StgTSO__link(t);
1805 StgTSO__link(CurrentTSO) = t;
1807 W_[sleeping_queue] = CurrentTSO;
1809 foreign "C" setTSOLink(MyCapability() "ptr", prev "ptr", CurrentTSO) [];
1811 jump stg_block_noregs;
1813 #endif /* !THREADED_RTS */
1817 #ifdef mingw32_HOST_OS
1818 STRING(stg_asyncReadzh_malloc_str, "stg_asyncReadzh")
1825 foreign "C" barf("asyncRead# on threaded RTS") never returns;
1828 /* args: R1 = fd, R2 = isSock, R3 = len, R4 = buf */
1829 ASSERT(StgTSO_why_blocked(CurrentTSO) == NotBlocked::I16);
1830 StgTSO_why_blocked(CurrentTSO) = BlockedOnRead::I16;
1832 /* could probably allocate this on the heap instead */
1833 ("ptr" ares) = foreign "C" stgMallocBytes(SIZEOF_StgAsyncIOResult,
1834 stg_asyncReadzh_malloc_str)
1836 (reqID) = foreign "C" addIORequest(R1, 0/*FALSE*/,R2,R3,R4 "ptr") [];
1837 StgAsyncIOResult_reqID(ares) = reqID;
1838 StgAsyncIOResult_len(ares) = 0;
1839 StgAsyncIOResult_errCode(ares) = 0;
1840 StgTSO_block_info(CurrentTSO) = ares;
1841 APPEND_TO_BLOCKED_QUEUE(CurrentTSO);
1842 jump stg_block_async;
1846 STRING(stg_asyncWritezh_malloc_str, "stg_asyncWritezh")
1853 foreign "C" barf("asyncWrite# on threaded RTS") never returns;
1856 /* args: R1 = fd, R2 = isSock, R3 = len, R4 = buf */
1857 ASSERT(StgTSO_why_blocked(CurrentTSO) == NotBlocked::I16);
1858 StgTSO_why_blocked(CurrentTSO) = BlockedOnWrite::I16;
1860 ("ptr" ares) = foreign "C" stgMallocBytes(SIZEOF_StgAsyncIOResult,
1861 stg_asyncWritezh_malloc_str)
1863 (reqID) = foreign "C" addIORequest(R1, 1/*TRUE*/,R2,R3,R4 "ptr") [];
1865 StgAsyncIOResult_reqID(ares) = reqID;
1866 StgAsyncIOResult_len(ares) = 0;
1867 StgAsyncIOResult_errCode(ares) = 0;
1868 StgTSO_block_info(CurrentTSO) = ares;
1869 APPEND_TO_BLOCKED_QUEUE(CurrentTSO);
1870 jump stg_block_async;
1874 STRING(stg_asyncDoProczh_malloc_str, "stg_asyncDoProczh")
1881 foreign "C" barf("asyncDoProc# on threaded RTS") never returns;
1884 /* args: R1 = proc, R2 = param */
1885 ASSERT(StgTSO_why_blocked(CurrentTSO) == NotBlocked::I16);
1886 StgTSO_why_blocked(CurrentTSO) = BlockedOnDoProc::I16;
1888 /* could probably allocate this on the heap instead */
1889 ("ptr" ares) = foreign "C" stgMallocBytes(SIZEOF_StgAsyncIOResult,
1890 stg_asyncDoProczh_malloc_str)
1892 (reqID) = foreign "C" addDoProcRequest(R1 "ptr",R2 "ptr") [];
1893 StgAsyncIOResult_reqID(ares) = reqID;
1894 StgAsyncIOResult_len(ares) = 0;
1895 StgAsyncIOResult_errCode(ares) = 0;
1896 StgTSO_block_info(CurrentTSO) = ares;
1897 APPEND_TO_BLOCKED_QUEUE(CurrentTSO);
1898 jump stg_block_async;
1903 /* -----------------------------------------------------------------------------
1906 * noDuplicate# tries to ensure that none of the thunks under
1907 * evaluation by the current thread are also under evaluation by
1908 * another thread. It relies on *both* threads doing noDuplicate#;
1909 * the second one will get blocked if they are duplicating some work.
1911 * The idea is that noDuplicate# is used within unsafePerformIO to
1912 * ensure that the IO operation is performed at most once.
1913 * noDuplicate# calls threadPaused which acquires an exclusive lock on
1914 * all the thunks currently under evaluation by the current thread.
1916 * Consider the following scenario. There is a thunk A, whose
1917 * evaluation requires evaluating thunk B, where thunk B is an
1918 * unsafePerformIO. Two threads, 1 and 2, bother enter A. Thread 2
1919 * is pre-empted before it enters B, and claims A by blackholing it
1920 * (in threadPaused). Thread 1 now enters B, and calls noDuplicate#.
1923 * +-----------+ +---------------+
1924 * | -------+-----> A <-------+------- |
1925 * | update | BLACKHOLE | marked_update |
1926 * +-----------+ +---------------+
1929 * | | +---------------+
1932 * | update | BLACKHOLE
1935 * At this point: A is a blackhole, owned by thread 2. noDuplicate#
1936 * calls threadPaused, which walks up the stack and
1937 * - claims B on behalf of thread 1
1938 * - then it reaches the update frame for A, which it sees is already
1939 * a BLACKHOLE and is therefore owned by another thread. Since
1940 * thread 1 is duplicating work, the computation up to the update
1941 * frame for A is suspended, including thunk B.
1942 * - thunk B, which is an unsafePerformIO, has now been reverted to
1943 * an AP_STACK which could be duplicated - BAD!
1944 * - The solution is as follows: before calling threadPaused, we
1945 * leave a frame on the stack (stg_noDuplicate_info) that will call
1946 * noDuplicate# again if the current computation is suspended and
1949 * See the test program in concurrent/prog003 for a way to demonstrate
1950 * this. It needs to be run with +RTS -N3 or greater, and the bug
1951 * only manifests occasionally (once very 10 runs or so).
1952 * -------------------------------------------------------------------------- */
1954 INFO_TABLE_RET(stg_noDuplicate, RET_SMALL)
1957 jump stg_noDuplicatezh;
1962 STK_CHK_GEN( WDS(1), NO_PTRS, stg_noDuplicatezh );
1963 // leave noDuplicate frame in case the current
1964 // computation is suspended and restarted (see above).
1966 Sp(0) = stg_noDuplicate_info;
1968 SAVE_THREAD_STATE();
1969 ASSERT(StgTSO_what_next(CurrentTSO) == ThreadRunGHC::I16);
1970 foreign "C" threadPaused (MyCapability() "ptr", CurrentTSO "ptr") [];
1972 if (StgTSO_what_next(CurrentTSO) == ThreadKilled::I16) {
1973 jump stg_threadFinished;
1975 LOAD_THREAD_STATE();
1976 ASSERT(StgTSO_what_next(CurrentTSO) == ThreadRunGHC::I16);
1977 // remove the stg_noDuplicate frame if it is still there.
1978 if (Sp(0) == stg_noDuplicate_info) {
1981 jump %ENTRY_CODE(Sp(0));
1985 /* -----------------------------------------------------------------------------
1987 -------------------------------------------------------------------------- */
1991 W_ ap_stack, offset, val, ok;
1993 /* args: R1 = AP_STACK, R2 = offset */
1997 if (%INFO_PTR(ap_stack) == stg_AP_STACK_info) {
1999 val = StgAP_STACK_payload(ap_stack,offset);
2007 // Write the cost center stack of the first argument on stderr; return
2008 // the second. Possibly only makes sense for already evaluated
2015 ccs = StgHeader_ccs(UNTAG(R1));
2016 foreign "C" fprintCCS_stderr(ccs "ptr") [R2];
2027 #ifndef THREADED_RTS
2028 RET_NP(0,ghczmprim_GHCziTypes_False_closure);
2030 (spark) = foreign "C" findSpark(MyCapability());
2034 RET_NP(0,ghczmprim_GHCziTypes_False_closure);
2043 (n) = foreign "C" dequeElements(Capability_sparks(MyCapability()));
2055 #if defined(TRACING) || defined(DEBUG)
2057 foreign "C" traceUserMsg(MyCapability() "ptr", msg "ptr") [];
2059 #elif defined(DTRACE)
2063 // We should go through the macro HASKELLEVENT_USER_MSG_ENABLED from
2064 // RtsProbes.h, but that header file includes unistd.h, which doesn't
2066 #if !defined(solaris2_TARGET_OS)
2067 (enabled) = foreign "C" __dtrace_isenabled$HaskellEvent$user__msg$v1() [];
2069 // Solaris' DTrace can't handle the
2070 // __dtrace_isenabled$HaskellEvent$user__msg$v1
2071 // call above. This call is just for testing whether the user__msg
2072 // probe is enabled, and is here for just performance optimization.
2073 // Since preparation for the probe is not that complex I disable usage of
2074 // this test above for Solaris and enable the probe usage manually
2075 // here. Please note that this does not mean that the probe will be
2076 // used during the runtime! You still need to enable it by consumption
2077 // in your dtrace script as you do with any other probe.
2081 foreign "C" dtraceUserMsgWrapper(MyCapability() "ptr", msg "ptr") [];
2085 jump %ENTRY_CODE(Sp(0));