1 /* -----------------------------------------------------------------------------
2 * $Id: Storage.c,v 1.36 2001/02/11 17:51:08 simonmar Exp $
4 * (c) The GHC Team, 1998-1999
6 * Storage manager front end
8 * ---------------------------------------------------------------------------*/
15 #include "BlockAlloc.h"
22 #include "StoragePriv.h"
25 nat nursery_blocks; /* number of blocks in the nursery */
28 StgClosure *caf_list = NULL;
30 bdescr *small_alloc_list; /* allocate()d small objects */
31 bdescr *large_alloc_list; /* allocate()d large objects */
32 nat alloc_blocks; /* number of allocate()d blocks since GC */
33 nat alloc_blocks_lim; /* approximate limit on alloc_blocks */
35 StgPtr alloc_Hp = NULL; /* next free byte in small_alloc_list */
36 StgPtr alloc_HpLim = NULL; /* end of block at small_alloc_list */
38 generation *generations; /* all the generations */
39 generation *g0; /* generation 0, for convenience */
40 generation *oldest_gen; /* oldest generation, for convenience */
41 step *g0s0; /* generation 0, step 0, for convenience */
43 lnat total_allocated = 0; /* total memory allocated during run */
46 * Storage manager mutex: protects all the above state from
47 * simultaneous access by two STG threads.
50 pthread_mutex_t sm_mutex = PTHREAD_MUTEX_INITIALIZER;
56 static void *stgAllocForGMP (size_t size_in_bytes);
57 static void *stgReallocForGMP (void *ptr, size_t old_size, size_t new_size);
58 static void stgDeallocForGMP (void *ptr, size_t size);
67 /* If we're doing heap profiling, we want a two-space heap with a
68 * fixed-size allocation area so that we get roughly even-spaced
71 #if defined(PROFILING) || defined(DEBUG)
72 if (RtsFlags.ProfFlags.doHeapProfile) {
73 RtsFlags.GcFlags.generations = 1;
74 RtsFlags.GcFlags.steps = 1;
75 RtsFlags.GcFlags.oldGenFactor = 0;
76 RtsFlags.GcFlags.heapSizeSuggestion = 0;
80 if (RtsFlags.GcFlags.heapSizeSuggestion >
81 RtsFlags.GcFlags.maxHeapSize) {
82 RtsFlags.GcFlags.maxHeapSize = RtsFlags.GcFlags.heapSizeSuggestion;
87 /* allocate generation info array */
88 generations = (generation *)stgMallocBytes(RtsFlags.GcFlags.generations
89 * sizeof(struct _generation),
92 /* Initialise all generations */
93 for(g = 0; g < RtsFlags.GcFlags.generations; g++) {
94 gen = &generations[g];
96 gen->mut_list = END_MUT_LIST;
97 gen->mut_once_list = END_MUT_LIST;
99 gen->failed_promotions = 0;
103 /* A couple of convenience pointers */
104 g0 = &generations[0];
105 oldest_gen = &generations[RtsFlags.GcFlags.generations-1];
107 /* Allocate step structures in each generation */
108 if (RtsFlags.GcFlags.generations > 1) {
109 /* Only for multiple-generations */
111 /* Oldest generation: one step */
112 oldest_gen->n_steps = 1;
114 stgMallocBytes(1 * sizeof(struct _step), "initStorage: last step");
116 /* set up all except the oldest generation with 2 steps */
117 for(g = 0; g < RtsFlags.GcFlags.generations-1; g++) {
118 generations[g].n_steps = RtsFlags.GcFlags.steps;
119 generations[g].steps =
120 stgMallocBytes (RtsFlags.GcFlags.steps * sizeof(struct _step),
121 "initStorage: steps");
125 /* single generation, i.e. a two-space collector */
127 g0->steps = stgMallocBytes (sizeof(struct _step), "initStorage: steps");
130 /* Initialise all steps */
131 for (g = 0; g < RtsFlags.GcFlags.generations; g++) {
132 for (s = 0; s < generations[g].n_steps; s++) {
133 stp = &generations[g].steps[s];
137 stp->gen = &generations[g];
143 stp->large_objects = NULL;
144 stp->new_large_objects = NULL;
145 stp->scavenged_large_objects = NULL;
149 /* Set up the destination pointers in each younger gen. step */
150 for (g = 0; g < RtsFlags.GcFlags.generations-1; g++) {
151 for (s = 0; s < generations[g].n_steps-1; s++) {
152 generations[g].steps[s].to = &generations[g].steps[s+1];
154 generations[g].steps[s].to = &generations[g+1].steps[0];
157 /* The oldest generation has one step and its destination is the
159 oldest_gen->steps[0].to = &oldest_gen->steps[0];
161 /* generation 0 is special: that's the nursery */
162 generations[0].max_blocks = 0;
164 /* G0S0: the allocation area. Policy: keep the allocation area
165 * small to begin with, even if we have a large suggested heap
166 * size. Reason: we're going to do a major collection first, and we
167 * don't want it to be a big one. This vague idea is borne out by
168 * rigorous experimental evidence.
170 g0s0 = &generations[0].steps[0];
174 weak_ptr_list = NULL;
177 /* initialise the allocate() interface */
178 small_alloc_list = NULL;
179 large_alloc_list = NULL;
181 alloc_blocks_lim = RtsFlags.GcFlags.minAllocAreaSize;
183 /* Tell GNU multi-precision pkg about our custom alloc functions */
184 mp_set_memory_functions(stgAllocForGMP, stgReallocForGMP, stgDeallocForGMP);
187 pthread_mutex_init(&sm_mutex, NULL);
190 IF_DEBUG(gc, stat_describe_gens());
196 stat_exit(calcAllocated());
199 /* -----------------------------------------------------------------------------
202 The entry code for every CAF does the following:
204 - builds a CAF_BLACKHOLE in the heap
205 - pushes an update frame pointing to the CAF_BLACKHOLE
206 - invokes UPD_CAF(), which:
207 - calls newCaf, below
208 - updates the CAF with a static indirection to the CAF_BLACKHOLE
210 Why do we build a BLACKHOLE in the heap rather than just updating
211 the thunk directly? It's so that we only need one kind of update
212 frame - otherwise we'd need a static version of the update frame too.
214 newCaf() does the following:
216 - it puts the CAF on the oldest generation's mut-once list.
217 This is so that we can treat the CAF as a root when collecting
220 For GHCI, we have additional requirements when dealing with CAFs:
222 - we must *retain* all dynamically-loaded CAFs ever entered,
223 just in case we need them again.
224 - we must be able to *revert* CAFs that have been evaluated, to
225 their pre-evaluated form.
227 To do this, we use an additional CAF list. When newCaf() is
228 called on a dynamically-loaded CAF, we add it to the CAF list
229 instead of the old-generation mutable list, and save away its
230 old info pointer (in caf->saved_info) for later reversion.
232 To revert all the CAFs, we traverse the CAF list and reset the
233 info pointer to caf->saved_info, then throw away the CAF list.
234 (see GC.c:revertCAFs()).
238 -------------------------------------------------------------------------- */
241 newCAF(StgClosure* caf)
243 /* Put this CAF on the mutable list for the old generation.
244 * This is a HACK - the IND_STATIC closure doesn't really have
245 * a mut_link field, but we pretend it has - in fact we re-use
246 * the STATIC_LINK field for the time being, because when we
247 * come to do a major GC we won't need the mut_link field
248 * any more and can use it as a STATIC_LINK.
250 ACQUIRE_LOCK(&sm_mutex);
252 if (is_dynamically_loaded_rwdata_ptr((StgPtr)caf)) {
253 ((StgIndStatic *)caf)->saved_info = (StgInfoTable *)caf->header.info;
254 ((StgIndStatic *)caf)->static_link = caf_list;
257 ((StgIndStatic *)caf)->saved_info = NULL;
258 ((StgMutClosure *)caf)->mut_link = oldest_gen->mut_once_list;
259 oldest_gen->mut_once_list = (StgMutClosure *)caf;
262 RELEASE_LOCK(&sm_mutex);
265 /* -----------------------------------------------------------------------------
267 -------------------------------------------------------------------------- */
270 allocNurseries( void )
279 for (cap = free_capabilities; cap != NULL; cap = cap->link) {
280 cap->rNursery = allocNursery(NULL, RtsFlags.GcFlags.minAllocAreaSize);
281 cap->rCurrentNursery = cap->rNursery;
282 for (bd = cap->rNursery; bd != NULL; bd = bd->link) {
283 bd->back = (bdescr *)cap;
286 /* Set the back links to be equal to the Capability,
287 * so we can do slightly better informed locking.
291 nursery_blocks = RtsFlags.GcFlags.minAllocAreaSize;
292 g0s0->blocks = allocNursery(NULL, nursery_blocks);
293 g0s0->n_blocks = nursery_blocks;
294 g0s0->to_space = NULL;
295 MainRegTable.rNursery = g0s0->blocks;
296 MainRegTable.rCurrentNursery = g0s0->blocks;
297 /* hp, hpLim, hp_bd, to_space etc. aren't used in G0S0 */
302 resetNurseries( void )
308 /* All tasks must be stopped */
309 ASSERT(n_free_capabilities == RtsFlags.ParFlags.nNodes);
311 for (cap = free_capabilities; cap != NULL; cap = cap->link) {
312 for (bd = cap->rNursery; bd; bd = bd->link) {
313 bd->free = bd->start;
314 ASSERT(bd->gen == g0);
315 ASSERT(bd->step == g0s0);
316 IF_DEBUG(sanity,memset(bd->start, 0xaa, BLOCK_SIZE));
318 cap->rCurrentNursery = cap->rNursery;
321 for (bd = g0s0->blocks; bd; bd = bd->link) {
322 bd->free = bd->start;
323 ASSERT(bd->gen == g0);
324 ASSERT(bd->step == g0s0);
325 IF_DEBUG(sanity,memset(bd->start, 0xaa, BLOCK_SIZE));
327 MainRegTable.rNursery = g0s0->blocks;
328 MainRegTable.rCurrentNursery = g0s0->blocks;
333 allocNursery (bdescr *last_bd, nat blocks)
338 /* Allocate a nursery */
339 for (i=0; i < blocks; i++) {
345 bd->free = bd->start;
352 resizeNursery ( nat blocks )
357 barf("resizeNursery: can't resize in SMP mode");
360 if (nursery_blocks == blocks) {
361 ASSERT(g0s0->n_blocks == blocks);
365 else if (nursery_blocks < blocks) {
366 IF_DEBUG(gc, fprintf(stderr, "Increasing size of nursery to %d blocks\n",
368 g0s0->blocks = allocNursery(g0s0->blocks, blocks-nursery_blocks);
374 IF_DEBUG(gc, fprintf(stderr, "Decreasing size of nursery to %d blocks\n",
376 for (bd = g0s0->blocks; nursery_blocks > blocks; nursery_blocks--) {
384 g0s0->n_blocks = nursery_blocks = blocks;
387 /* -----------------------------------------------------------------------------
388 The allocate() interface
390 allocate(n) always succeeds, and returns a chunk of memory n words
391 long. n can be larger than the size of a block if necessary, in
392 which case a contiguous block group will be allocated.
393 -------------------------------------------------------------------------- */
401 ACQUIRE_LOCK(&sm_mutex);
403 TICK_ALLOC_HEAP_NOCTR(n);
406 /* big allocation (>LARGE_OBJECT_THRESHOLD) */
407 /* ToDo: allocate directly into generation 1 */
408 if (n >= LARGE_OBJECT_THRESHOLD/sizeof(W_)) {
409 nat req_blocks = (lnat)BLOCK_ROUND_UP(n*sizeof(W_)) / BLOCK_SIZE;
410 bd = allocGroup(req_blocks);
411 dbl_link_onto(bd, &g0s0->large_objects);
415 bd->free = bd->start;
416 /* don't add these blocks to alloc_blocks, since we're assuming
417 * that large objects are likely to remain live for quite a while
418 * (eg. running threads), so garbage collecting early won't make
421 alloc_blocks += req_blocks;
422 RELEASE_LOCK(&sm_mutex);
425 /* small allocation (<LARGE_OBJECT_THRESHOLD) */
426 } else if (small_alloc_list == NULL || alloc_Hp + n > alloc_HpLim) {
427 if (small_alloc_list) {
428 small_alloc_list->free = alloc_Hp;
431 bd->link = small_alloc_list;
432 small_alloc_list = bd;
436 alloc_Hp = bd->start;
437 alloc_HpLim = bd->start + BLOCK_SIZE_W;
443 RELEASE_LOCK(&sm_mutex);
447 lnat allocated_bytes(void)
449 return (alloc_blocks * BLOCK_SIZE_W - (alloc_HpLim - alloc_Hp));
452 /* -----------------------------------------------------------------------------
453 Allocation functions for GMP.
455 These all use the allocate() interface - we can't have any garbage
456 collection going on during a gmp operation, so we use allocate()
457 which always succeeds. The gmp operations which might need to
458 allocate will ask the storage manager (via doYouWantToGC()) whether
459 a garbage collection is required, in case we get into a loop doing
460 only allocate() style allocation.
461 -------------------------------------------------------------------------- */
464 stgAllocForGMP (size_t size_in_bytes)
467 nat data_size_in_words, total_size_in_words;
469 /* should be a multiple of sizeof(StgWord) (whole no. of limbs) */
470 ASSERT(size_in_bytes % sizeof(W_) == 0);
472 data_size_in_words = size_in_bytes / sizeof(W_);
473 total_size_in_words = sizeofW(StgArrWords) + data_size_in_words;
475 /* allocate and fill it in. */
476 arr = (StgArrWords *)allocate(total_size_in_words);
477 SET_ARR_HDR(arr, &stg_ARR_WORDS_info, CCCS, data_size_in_words);
479 /* and return a ptr to the goods inside the array */
480 return(BYTE_ARR_CTS(arr));
484 stgReallocForGMP (void *ptr, size_t old_size, size_t new_size)
486 void *new_stuff_ptr = stgAllocForGMP(new_size);
488 char *p = (char *) ptr;
489 char *q = (char *) new_stuff_ptr;
491 for (; i < old_size; i++, p++, q++) {
495 return(new_stuff_ptr);
499 stgDeallocForGMP (void *ptr STG_UNUSED,
500 size_t size STG_UNUSED)
502 /* easy for us: the garbage collector does the dealloc'n */
505 /* -----------------------------------------------------------------------------
507 * -------------------------------------------------------------------------- */
509 /* -----------------------------------------------------------------------------
512 * Approximate how much we've allocated: number of blocks in the
513 * nursery + blocks allocated via allocate() - unused nusery blocks.
514 * This leaves a little slop at the end of each block, and doesn't
515 * take into account large objects (ToDo).
516 * -------------------------------------------------------------------------- */
519 calcAllocated( void )
527 /* All tasks must be stopped. Can't assert that all the
528 capabilities are owned by the scheduler, though: one or more
529 tasks might have been stopped while they were running (non-main)
531 /* ASSERT(n_free_capabilities == RtsFlags.ParFlags.nNodes); */
534 n_free_capabilities * RtsFlags.GcFlags.minAllocAreaSize * BLOCK_SIZE_W
537 for (cap = free_capabilities; cap != NULL; cap = cap->link) {
538 for ( bd = cap->rCurrentNursery->link; bd != NULL; bd = bd->link ) {
539 allocated -= BLOCK_SIZE_W;
541 if (cap->rCurrentNursery->free < cap->rCurrentNursery->start
543 allocated -= (cap->rCurrentNursery->start + BLOCK_SIZE_W)
544 - cap->rCurrentNursery->free;
549 bdescr *current_nursery = MainRegTable.rCurrentNursery;
551 allocated = (nursery_blocks * BLOCK_SIZE_W) + allocated_bytes();
552 for ( bd = current_nursery->link; bd != NULL; bd = bd->link ) {
553 allocated -= BLOCK_SIZE_W;
555 if (current_nursery->free < current_nursery->start + BLOCK_SIZE_W) {
556 allocated -= (current_nursery->start + BLOCK_SIZE_W)
557 - current_nursery->free;
561 total_allocated += allocated;
565 /* Approximate the amount of live data in the heap. To be called just
566 * after garbage collection (see GarbageCollect()).
575 if (RtsFlags.GcFlags.generations == 1) {
576 live = (g0s0->to_blocks - 1) * BLOCK_SIZE_W +
577 ((lnat)g0s0->hp_bd->free - (lnat)g0s0->hp_bd->start) / sizeof(W_);
581 for (g = 0; g < RtsFlags.GcFlags.generations; g++) {
582 for (s = 0; s < generations[g].n_steps; s++) {
583 /* approximate amount of live data (doesn't take into account slop
584 * at end of each block).
586 if (g == 0 && s == 0) {
589 stp = &generations[g].steps[s];
590 live += (stp->n_blocks - 1) * BLOCK_SIZE_W +
591 ((lnat)stp->hp_bd->free - (lnat)stp->hp_bd->start) / sizeof(W_);
597 /* Approximate the number of blocks that will be needed at the next
598 * garbage collection.
600 * Assume: all data currently live will remain live. Steps that will
601 * be collected next time will therefore need twice as many blocks
602 * since all the data will be copied.
611 for (g = 0; g < RtsFlags.GcFlags.generations; g++) {
612 for (s = 0; s < generations[g].n_steps; s++) {
613 if (g == 0 && s == 0) { continue; }
614 stp = &generations[g].steps[s];
615 if (generations[g].steps[0].n_blocks > generations[g].max_blocks) {
616 needed += 2 * stp->n_blocks;
618 needed += stp->n_blocks;
625 /* -----------------------------------------------------------------------------
628 memInventory() checks for memory leaks by counting up all the
629 blocks we know about and comparing that to the number of blocks
630 allegedly floating around in the system.
631 -------------------------------------------------------------------------- */
641 lnat total_blocks = 0, free_blocks = 0;
643 /* count the blocks we current have */
645 for (g = 0; g < RtsFlags.GcFlags.generations; g++) {
646 for (s = 0; s < generations[g].n_steps; s++) {
647 stp = &generations[g].steps[s];
648 total_blocks += stp->n_blocks;
649 if (RtsFlags.GcFlags.generations == 1) {
650 /* two-space collector has a to-space too :-) */
651 total_blocks += g0s0->to_blocks;
653 for (bd = stp->large_objects; bd; bd = bd->link) {
654 total_blocks += bd->blocks;
655 /* hack for megablock groups: they have an extra block or two in
656 the second and subsequent megablocks where the block
657 descriptors would normally go.
659 if (bd->blocks > BLOCKS_PER_MBLOCK) {
660 total_blocks -= (MBLOCK_SIZE / BLOCK_SIZE - BLOCKS_PER_MBLOCK)
661 * (bd->blocks/(MBLOCK_SIZE/BLOCK_SIZE));
667 /* any blocks held by allocate() */
668 for (bd = small_alloc_list; bd; bd = bd->link) {
669 total_blocks += bd->blocks;
671 for (bd = large_alloc_list; bd; bd = bd->link) {
672 total_blocks += bd->blocks;
675 /* count the blocks on the free list */
676 free_blocks = countFreeList();
678 ASSERT(total_blocks + free_blocks == mblocks_allocated * BLOCKS_PER_MBLOCK);
681 if (total_blocks + free_blocks != mblocks_allocated *
683 fprintf(stderr, "Blocks: %ld live + %ld free = %ld total (%ld around)\n",
684 total_blocks, free_blocks, total_blocks + free_blocks,
685 mblocks_allocated * BLOCKS_PER_MBLOCK);
690 /* Full heap sanity check. */
697 if (RtsFlags.GcFlags.generations == 1) {
698 checkHeap(g0s0->to_space, NULL);
699 checkChain(g0s0->large_objects);
702 for (g = 0; g <= N; g++) {
703 for (s = 0; s < generations[g].n_steps; s++) {
704 if (g == 0 && s == 0) { continue; }
705 checkHeap(generations[g].steps[s].blocks, NULL);
708 for (g = N+1; g < RtsFlags.GcFlags.generations; g++) {
709 for (s = 0; s < generations[g].n_steps; s++) {
710 checkHeap(generations[g].steps[s].blocks,
711 generations[g].steps[s].blocks->start);
712 checkChain(generations[g].steps[s].large_objects);
715 checkFreeListSanity();