1 /* -----------------------------------------------------------------------------
2 * $Id: Storage.c,v 1.75 2003/01/29 10:28:56 simonmar Exp $
4 * (c) The GHC Team, 1998-1999
6 * Storage manager front end
8 * ---------------------------------------------------------------------------*/
10 #include "PosixSource.h"
16 #include "BlockAlloc.h"
24 #include "OSThreads.h"
25 #include "StoragePriv.h"
27 #include "RetainerProfile.h" // for counting memory blocks (memInventory)
32 StgClosure *caf_list = NULL;
34 bdescr *small_alloc_list; /* allocate()d small objects */
35 bdescr *pinned_object_block; /* allocate pinned objects into this block */
36 nat alloc_blocks; /* number of allocate()d blocks since GC */
37 nat alloc_blocks_lim; /* approximate limit on alloc_blocks */
39 StgPtr alloc_Hp = NULL; /* next free byte in small_alloc_list */
40 StgPtr alloc_HpLim = NULL; /* end of block at small_alloc_list */
42 generation *generations = NULL; /* all the generations */
43 generation *g0 = NULL; /* generation 0, for convenience */
44 generation *oldest_gen = NULL; /* oldest generation, for convenience */
45 step *g0s0 = NULL; /* generation 0, step 0, for convenience */
47 lnat total_allocated = 0; /* total memory allocated during run */
50 * Storage manager mutex: protects all the above state from
51 * simultaneous access by two STG threads.
54 Mutex sm_mutex = INIT_MUTEX_VAR;
60 static void *stgAllocForGMP (size_t size_in_bytes);
61 static void *stgReallocForGMP (void *ptr, size_t old_size, size_t new_size);
62 static void stgDeallocForGMP (void *ptr, size_t size);
71 if (generations != NULL) {
72 // multi-init protection
76 /* Sanity check to make sure the LOOKS_LIKE_ macros appear to be
77 * doing something reasonable.
79 ASSERT(LOOKS_LIKE_INFO_PTR(&stg_BLACKHOLE_info));
80 ASSERT(LOOKS_LIKE_CLOSURE_PTR(&stg_dummy_ret_closure));
81 ASSERT(!HEAP_ALLOCED(&stg_dummy_ret_closure));
83 if (RtsFlags.GcFlags.maxHeapSize != 0 &&
84 RtsFlags.GcFlags.heapSizeSuggestion >
85 RtsFlags.GcFlags.maxHeapSize) {
86 RtsFlags.GcFlags.maxHeapSize = RtsFlags.GcFlags.heapSizeSuggestion;
89 if (RtsFlags.GcFlags.maxHeapSize != 0 &&
90 RtsFlags.GcFlags.minAllocAreaSize >
91 RtsFlags.GcFlags.maxHeapSize) {
92 prog_belch("maximum heap size (-M) is smaller than minimum alloc area size (-A)");
99 initCondition(&sm_mutex);
102 /* allocate generation info array */
103 generations = (generation *)stgMallocBytes(RtsFlags.GcFlags.generations
104 * sizeof(struct _generation),
105 "initStorage: gens");
107 /* Initialise all generations */
108 for(g = 0; g < RtsFlags.GcFlags.generations; g++) {
109 gen = &generations[g];
111 gen->mut_list = END_MUT_LIST;
112 gen->mut_once_list = END_MUT_LIST;
113 gen->collections = 0;
114 gen->failed_promotions = 0;
118 /* A couple of convenience pointers */
119 g0 = &generations[0];
120 oldest_gen = &generations[RtsFlags.GcFlags.generations-1];
122 /* Allocate step structures in each generation */
123 if (RtsFlags.GcFlags.generations > 1) {
124 /* Only for multiple-generations */
126 /* Oldest generation: one step */
127 oldest_gen->n_steps = 1;
129 stgMallocBytes(1 * sizeof(struct _step), "initStorage: last step");
131 /* set up all except the oldest generation with 2 steps */
132 for(g = 0; g < RtsFlags.GcFlags.generations-1; g++) {
133 generations[g].n_steps = RtsFlags.GcFlags.steps;
134 generations[g].steps =
135 stgMallocBytes (RtsFlags.GcFlags.steps * sizeof(struct _step),
136 "initStorage: steps");
140 /* single generation, i.e. a two-space collector */
142 g0->steps = stgMallocBytes (sizeof(struct _step), "initStorage: steps");
145 /* Initialise all steps */
146 for (g = 0; g < RtsFlags.GcFlags.generations; g++) {
147 for (s = 0; s < generations[g].n_steps; s++) {
148 stp = &generations[g].steps[s];
152 stp->gen = &generations[g];
159 stp->large_objects = NULL;
160 stp->n_large_blocks = 0;
161 stp->new_large_objects = NULL;
162 stp->scavenged_large_objects = NULL;
163 stp->n_scavenged_large_blocks = 0;
164 stp->is_compacted = 0;
169 /* Set up the destination pointers in each younger gen. step */
170 for (g = 0; g < RtsFlags.GcFlags.generations-1; g++) {
171 for (s = 0; s < generations[g].n_steps-1; s++) {
172 generations[g].steps[s].to = &generations[g].steps[s+1];
174 generations[g].steps[s].to = &generations[g+1].steps[0];
177 /* The oldest generation has one step and it is compacted. */
178 if (RtsFlags.GcFlags.compact) {
179 if (RtsFlags.GcFlags.generations == 1) {
180 belch("WARNING: compaction is incompatible with -G1; disabled");
182 oldest_gen->steps[0].is_compacted = 1;
185 oldest_gen->steps[0].to = &oldest_gen->steps[0];
187 /* generation 0 is special: that's the nursery */
188 generations[0].max_blocks = 0;
190 /* G0S0: the allocation area. Policy: keep the allocation area
191 * small to begin with, even if we have a large suggested heap
192 * size. Reason: we're going to do a major collection first, and we
193 * don't want it to be a big one. This vague idea is borne out by
194 * rigorous experimental evidence.
196 g0s0 = &generations[0].steps[0];
200 weak_ptr_list = NULL;
203 /* initialise the allocate() interface */
204 small_alloc_list = NULL;
206 alloc_blocks_lim = RtsFlags.GcFlags.minAllocAreaSize;
208 /* Tell GNU multi-precision pkg about our custom alloc functions */
209 mp_set_memory_functions(stgAllocForGMP, stgReallocForGMP, stgDeallocForGMP);
212 initMutex(&sm_mutex);
215 IF_DEBUG(gc, statDescribeGens());
221 stat_exit(calcAllocated());
224 /* -----------------------------------------------------------------------------
227 The entry code for every CAF does the following:
229 - builds a CAF_BLACKHOLE in the heap
230 - pushes an update frame pointing to the CAF_BLACKHOLE
231 - invokes UPD_CAF(), which:
232 - calls newCaf, below
233 - updates the CAF with a static indirection to the CAF_BLACKHOLE
235 Why do we build a BLACKHOLE in the heap rather than just updating
236 the thunk directly? It's so that we only need one kind of update
237 frame - otherwise we'd need a static version of the update frame too.
239 newCaf() does the following:
241 - it puts the CAF on the oldest generation's mut-once list.
242 This is so that we can treat the CAF as a root when collecting
245 For GHCI, we have additional requirements when dealing with CAFs:
247 - we must *retain* all dynamically-loaded CAFs ever entered,
248 just in case we need them again.
249 - we must be able to *revert* CAFs that have been evaluated, to
250 their pre-evaluated form.
252 To do this, we use an additional CAF list. When newCaf() is
253 called on a dynamically-loaded CAF, we add it to the CAF list
254 instead of the old-generation mutable list, and save away its
255 old info pointer (in caf->saved_info) for later reversion.
257 To revert all the CAFs, we traverse the CAF list and reset the
258 info pointer to caf->saved_info, then throw away the CAF list.
259 (see GC.c:revertCAFs()).
263 -------------------------------------------------------------------------- */
266 newCAF(StgClosure* caf)
268 /* Put this CAF on the mutable list for the old generation.
269 * This is a HACK - the IND_STATIC closure doesn't really have
270 * a mut_link field, but we pretend it has - in fact we re-use
271 * the STATIC_LINK field for the time being, because when we
272 * come to do a major GC we won't need the mut_link field
273 * any more and can use it as a STATIC_LINK.
277 ((StgIndStatic *)caf)->saved_info = NULL;
278 ((StgMutClosure *)caf)->mut_link = oldest_gen->mut_once_list;
279 oldest_gen->mut_once_list = (StgMutClosure *)caf;
284 /* If we are PAR or DIST then we never forget a CAF */
286 //belch("<##> Globalising CAF %08x %s",caf,info_type(caf));
287 newGA=makeGlobal(caf,rtsTrue); /*given full weight*/
293 // An alternate version of newCaf which is used for dynamically loaded
294 // object code in GHCi. In this case we want to retain *all* CAFs in
295 // the object code, because they might be demanded at any time from an
296 // expression evaluated on the command line.
298 // The linker hackily arranges that references to newCaf from dynamic
299 // code end up pointing to newDynCAF.
301 newDynCAF(StgClosure *caf)
305 ((StgIndStatic *)caf)->saved_info = (StgInfoTable *)caf->header.info;
306 ((StgIndStatic *)caf)->static_link = caf_list;
312 /* -----------------------------------------------------------------------------
314 -------------------------------------------------------------------------- */
317 allocNurseries( void )
326 for (cap = free_capabilities; cap != NULL; cap = cap->link) {
327 cap->r.rNursery = allocNursery(NULL, RtsFlags.GcFlags.minAllocAreaSize);
328 cap->r.rCurrentNursery = cap->r.rNursery;
329 for (bd = cap->r.rNursery; bd != NULL; bd = bd->link) {
330 bd->u.back = (bdescr *)cap;
333 /* Set the back links to be equal to the Capability,
334 * so we can do slightly better informed locking.
338 g0s0->blocks = allocNursery(NULL, RtsFlags.GcFlags.minAllocAreaSize);
339 g0s0->n_blocks = RtsFlags.GcFlags.minAllocAreaSize;
340 g0s0->to_blocks = NULL;
341 g0s0->n_to_blocks = 0;
342 MainCapability.r.rNursery = g0s0->blocks;
343 MainCapability.r.rCurrentNursery = g0s0->blocks;
344 /* hp, hpLim, hp_bd, to_space etc. aren't used in G0S0 */
349 resetNurseries( void )
355 /* All tasks must be stopped */
356 ASSERT(n_free_capabilities == RtsFlags.ParFlags.nNodes);
358 for (cap = free_capabilities; cap != NULL; cap = cap->link) {
359 for (bd = cap->r.rNursery; bd; bd = bd->link) {
360 bd->free = bd->start;
361 ASSERT(bd->gen_no == 0);
362 ASSERT(bd->step == g0s0);
363 IF_DEBUG(sanity,memset(bd->start, 0xaa, BLOCK_SIZE));
365 cap->r.rCurrentNursery = cap->r.rNursery;
368 for (bd = g0s0->blocks; bd; bd = bd->link) {
369 bd->free = bd->start;
370 ASSERT(bd->gen_no == 0);
371 ASSERT(bd->step == g0s0);
372 IF_DEBUG(sanity,memset(bd->start, 0xaa, BLOCK_SIZE));
374 MainCapability.r.rNursery = g0s0->blocks;
375 MainCapability.r.rCurrentNursery = g0s0->blocks;
380 allocNursery (bdescr *tail, nat blocks)
385 // Allocate a nursery: we allocate fresh blocks one at a time and
386 // cons them on to the front of the list, not forgetting to update
387 // the back pointer on the tail of the list to point to the new block.
388 for (i=0; i < blocks; i++) {
391 processNursery() in LdvProfile.c assumes that every block group in
392 the nursery contains only a single block. So, if a block group is
393 given multiple blocks, change processNursery() accordingly.
397 // double-link the nursery: we might need to insert blocks
404 bd->free = bd->start;
412 resizeNursery ( nat blocks )
418 barf("resizeNursery: can't resize in SMP mode");
421 nursery_blocks = g0s0->n_blocks;
422 if (nursery_blocks == blocks) {
426 else if (nursery_blocks < blocks) {
427 IF_DEBUG(gc, fprintf(stderr, "Increasing size of nursery to %d blocks\n",
429 g0s0->blocks = allocNursery(g0s0->blocks, blocks-nursery_blocks);
435 IF_DEBUG(gc, fprintf(stderr, "Decreasing size of nursery to %d blocks\n",
439 while (nursery_blocks > blocks) {
441 next_bd->u.back = NULL;
442 nursery_blocks -= bd->blocks; // might be a large block
447 // might have gone just under, by freeing a large block, so make
448 // up the difference.
449 if (nursery_blocks < blocks) {
450 g0s0->blocks = allocNursery(g0s0->blocks, blocks-nursery_blocks);
454 g0s0->n_blocks = blocks;
455 ASSERT(countBlocks(g0s0->blocks) == g0s0->n_blocks);
458 /* -----------------------------------------------------------------------------
459 The allocate() interface
461 allocate(n) always succeeds, and returns a chunk of memory n words
462 long. n can be larger than the size of a block if necessary, in
463 which case a contiguous block group will be allocated.
464 -------------------------------------------------------------------------- */
474 TICK_ALLOC_HEAP_NOCTR(n);
477 /* big allocation (>LARGE_OBJECT_THRESHOLD) */
478 /* ToDo: allocate directly into generation 1 */
479 if (n >= LARGE_OBJECT_THRESHOLD/sizeof(W_)) {
480 nat req_blocks = (lnat)BLOCK_ROUND_UP(n*sizeof(W_)) / BLOCK_SIZE;
481 bd = allocGroup(req_blocks);
482 dbl_link_onto(bd, &g0s0->large_objects);
485 bd->flags = BF_LARGE;
486 bd->free = bd->start + n;
487 /* don't add these blocks to alloc_blocks, since we're assuming
488 * that large objects are likely to remain live for quite a while
489 * (eg. running threads), so garbage collecting early won't make
492 alloc_blocks += req_blocks;
496 /* small allocation (<LARGE_OBJECT_THRESHOLD) */
497 } else if (small_alloc_list == NULL || alloc_Hp + n > alloc_HpLim) {
498 if (small_alloc_list) {
499 small_alloc_list->free = alloc_Hp;
502 bd->link = small_alloc_list;
503 small_alloc_list = bd;
507 alloc_Hp = bd->start;
508 alloc_HpLim = bd->start + BLOCK_SIZE_W;
519 allocated_bytes( void )
523 allocated = alloc_blocks * BLOCK_SIZE_W - (alloc_HpLim - alloc_Hp);
524 if (pinned_object_block != NULL) {
525 allocated -= (pinned_object_block->start + BLOCK_SIZE_W) -
526 pinned_object_block->free;
533 tidyAllocateLists (void)
535 if (small_alloc_list != NULL) {
536 ASSERT(alloc_Hp >= small_alloc_list->start &&
537 alloc_Hp <= small_alloc_list->start + BLOCK_SIZE);
538 small_alloc_list->free = alloc_Hp;
542 /* ---------------------------------------------------------------------------
543 Allocate a fixed/pinned object.
545 We allocate small pinned objects into a single block, allocating a
546 new block when the current one overflows. The block is chained
547 onto the large_object_list of generation 0 step 0.
549 NOTE: The GC can't in general handle pinned objects. This
550 interface is only safe to use for ByteArrays, which have no
551 pointers and don't require scavenging. It works because the
552 block's descriptor has the BF_LARGE flag set, so the block is
553 treated as a large object and chained onto various lists, rather
554 than the individual objects being copied. However, when it comes
555 to scavenge the block, the GC will only scavenge the first object.
556 The reason is that the GC can't linearly scan a block of pinned
557 objects at the moment (doing so would require using the
558 mostly-copying techniques). But since we're restricting ourselves
559 to pinned ByteArrays, not scavenging is ok.
561 This function is called by newPinnedByteArray# which immediately
562 fills the allocated memory with a MutableByteArray#.
563 ------------------------------------------------------------------------- */
566 allocatePinned( nat n )
569 bdescr *bd = pinned_object_block;
573 TICK_ALLOC_HEAP_NOCTR(n);
576 // If the request is for a large object, then allocate()
577 // will give us a pinned object anyway.
578 if (n >= LARGE_OBJECT_THRESHOLD/sizeof(W_)) {
583 // we always return 8-byte aligned memory. bd->free must be
584 // 8-byte aligned to begin with, so we just round up n to
585 // the nearest multiple of 8 bytes.
586 if (sizeof(StgWord) == 4) {
590 // If we don't have a block of pinned objects yet, or the current
591 // one isn't large enough to hold the new object, allocate a new one.
592 if (bd == NULL || (bd->free + n) > (bd->start + BLOCK_SIZE_W)) {
593 pinned_object_block = bd = allocBlock();
594 dbl_link_onto(bd, &g0s0->large_objects);
597 bd->flags = BF_LARGE;
598 bd->free = bd->start;
608 /* -----------------------------------------------------------------------------
609 Allocation functions for GMP.
611 These all use the allocate() interface - we can't have any garbage
612 collection going on during a gmp operation, so we use allocate()
613 which always succeeds. The gmp operations which might need to
614 allocate will ask the storage manager (via doYouWantToGC()) whether
615 a garbage collection is required, in case we get into a loop doing
616 only allocate() style allocation.
617 -------------------------------------------------------------------------- */
620 stgAllocForGMP (size_t size_in_bytes)
623 nat data_size_in_words, total_size_in_words;
625 /* round up to a whole number of words */
626 data_size_in_words = (size_in_bytes + sizeof(W_) + 1) / sizeof(W_);
627 total_size_in_words = sizeofW(StgArrWords) + data_size_in_words;
629 /* allocate and fill it in. */
630 arr = (StgArrWords *)allocate(total_size_in_words);
631 SET_ARR_HDR(arr, &stg_ARR_WORDS_info, CCCS, data_size_in_words);
633 /* and return a ptr to the goods inside the array */
634 return(BYTE_ARR_CTS(arr));
638 stgReallocForGMP (void *ptr, size_t old_size, size_t new_size)
640 void *new_stuff_ptr = stgAllocForGMP(new_size);
642 char *p = (char *) ptr;
643 char *q = (char *) new_stuff_ptr;
645 for (; i < old_size; i++, p++, q++) {
649 return(new_stuff_ptr);
653 stgDeallocForGMP (void *ptr STG_UNUSED,
654 size_t size STG_UNUSED)
656 /* easy for us: the garbage collector does the dealloc'n */
659 /* -----------------------------------------------------------------------------
661 * -------------------------------------------------------------------------- */
663 /* -----------------------------------------------------------------------------
666 * Approximate how much we've allocated: number of blocks in the
667 * nursery + blocks allocated via allocate() - unused nusery blocks.
668 * This leaves a little slop at the end of each block, and doesn't
669 * take into account large objects (ToDo).
670 * -------------------------------------------------------------------------- */
673 calcAllocated( void )
681 /* All tasks must be stopped. Can't assert that all the
682 capabilities are owned by the scheduler, though: one or more
683 tasks might have been stopped while they were running (non-main)
685 /* ASSERT(n_free_capabilities == RtsFlags.ParFlags.nNodes); */
688 n_free_capabilities * RtsFlags.GcFlags.minAllocAreaSize * BLOCK_SIZE_W
691 for (cap = free_capabilities; cap != NULL; cap = cap->link) {
692 for ( bd = cap->r.rCurrentNursery->link; bd != NULL; bd = bd->link ) {
693 allocated -= BLOCK_SIZE_W;
695 if (cap->r.rCurrentNursery->free < cap->r.rCurrentNursery->start
697 allocated -= (cap->r.rCurrentNursery->start + BLOCK_SIZE_W)
698 - cap->r.rCurrentNursery->free;
703 bdescr *current_nursery = MainCapability.r.rCurrentNursery;
705 allocated = (g0s0->n_blocks * BLOCK_SIZE_W) + allocated_bytes();
706 for ( bd = current_nursery->link; bd != NULL; bd = bd->link ) {
707 allocated -= BLOCK_SIZE_W;
709 if (current_nursery->free < current_nursery->start + BLOCK_SIZE_W) {
710 allocated -= (current_nursery->start + BLOCK_SIZE_W)
711 - current_nursery->free;
715 total_allocated += allocated;
719 /* Approximate the amount of live data in the heap. To be called just
720 * after garbage collection (see GarbageCollect()).
729 if (RtsFlags.GcFlags.generations == 1) {
730 live = (g0s0->n_to_blocks - 1) * BLOCK_SIZE_W +
731 ((lnat)g0s0->hp_bd->free - (lnat)g0s0->hp_bd->start) / sizeof(W_);
735 for (g = 0; g < RtsFlags.GcFlags.generations; g++) {
736 for (s = 0; s < generations[g].n_steps; s++) {
737 /* approximate amount of live data (doesn't take into account slop
738 * at end of each block).
740 if (g == 0 && s == 0) {
743 stp = &generations[g].steps[s];
744 live += (stp->n_large_blocks + stp->n_blocks - 1) * BLOCK_SIZE_W;
745 if (stp->hp_bd != NULL) {
746 live += ((lnat)stp->hp_bd->free - (lnat)stp->hp_bd->start)
754 /* Approximate the number of blocks that will be needed at the next
755 * garbage collection.
757 * Assume: all data currently live will remain live. Steps that will
758 * be collected next time will therefore need twice as many blocks
759 * since all the data will be copied.
768 for (g = 0; g < RtsFlags.GcFlags.generations; g++) {
769 for (s = 0; s < generations[g].n_steps; s++) {
770 if (g == 0 && s == 0) { continue; }
771 stp = &generations[g].steps[s];
772 if (generations[g].steps[0].n_blocks +
773 generations[g].steps[0].n_large_blocks
774 > generations[g].max_blocks
775 && stp->is_compacted == 0) {
776 needed += 2 * stp->n_blocks;
778 needed += stp->n_blocks;
785 /* -----------------------------------------------------------------------------
788 memInventory() checks for memory leaks by counting up all the
789 blocks we know about and comparing that to the number of blocks
790 allegedly floating around in the system.
791 -------------------------------------------------------------------------- */
801 lnat total_blocks = 0, free_blocks = 0;
803 /* count the blocks we current have */
805 for (g = 0; g < RtsFlags.GcFlags.generations; g++) {
806 for (s = 0; s < generations[g].n_steps; s++) {
807 stp = &generations[g].steps[s];
808 total_blocks += stp->n_blocks;
809 if (RtsFlags.GcFlags.generations == 1) {
810 /* two-space collector has a to-space too :-) */
811 total_blocks += g0s0->n_to_blocks;
813 for (bd = stp->large_objects; bd; bd = bd->link) {
814 total_blocks += bd->blocks;
815 /* hack for megablock groups: they have an extra block or two in
816 the second and subsequent megablocks where the block
817 descriptors would normally go.
819 if (bd->blocks > BLOCKS_PER_MBLOCK) {
820 total_blocks -= (MBLOCK_SIZE / BLOCK_SIZE - BLOCKS_PER_MBLOCK)
821 * (bd->blocks/(MBLOCK_SIZE/BLOCK_SIZE));
827 /* any blocks held by allocate() */
828 for (bd = small_alloc_list; bd; bd = bd->link) {
829 total_blocks += bd->blocks;
833 if (RtsFlags.ProfFlags.doHeapProfile == HEAP_BY_RETAINER) {
834 for (bd = firstStack; bd != NULL; bd = bd->link)
835 total_blocks += bd->blocks;
839 // count the blocks allocated by the arena allocator
840 total_blocks += arenaBlocks();
842 /* count the blocks on the free list */
843 free_blocks = countFreeList();
845 if (total_blocks + free_blocks != mblocks_allocated *
847 fprintf(stderr, "Blocks: %ld live + %ld free = %ld total (%ld around)\n",
848 total_blocks, free_blocks, total_blocks + free_blocks,
849 mblocks_allocated * BLOCKS_PER_MBLOCK);
852 ASSERT(total_blocks + free_blocks == mblocks_allocated * BLOCKS_PER_MBLOCK);
857 countBlocks(bdescr *bd)
860 for (n=0; bd != NULL; bd=bd->link) {
866 /* Full heap sanity check. */
872 if (RtsFlags.GcFlags.generations == 1) {
873 checkHeap(g0s0->to_blocks);
874 checkChain(g0s0->large_objects);
877 for (g = 0; g < RtsFlags.GcFlags.generations; g++) {
878 for (s = 0; s < generations[g].n_steps; s++) {
879 ASSERT(countBlocks(generations[g].steps[s].blocks)
880 == generations[g].steps[s].n_blocks);
881 ASSERT(countBlocks(generations[g].steps[s].large_objects)
882 == generations[g].steps[s].n_large_blocks);
883 if (g == 0 && s == 0) { continue; }
884 checkHeap(generations[g].steps[s].blocks);
885 checkChain(generations[g].steps[s].large_objects);
887 checkMutableList(generations[g].mut_list, g);
888 checkMutOnceList(generations[g].mut_once_list, g);
892 checkFreeListSanity();
896 // handy function for use in gdb, because Bdescr() is inlined.
897 extern bdescr *_bdescr( StgPtr p );