/* -----------------------------------------------------------------------------
- * $Id: GC.c,v 1.7 1999/01/14 10:49:01 simonm Exp $
+ * $Id: GC.c,v 1.17 1999/01/20 16:07:40 simonm Exp $
*
* Two-space garbage collector
*
*/
static rtsBool failed_to_evac;
+/* Old to-space (used for two-space collector only)
+ */
+bdescr *old_to_space;
+
/* -----------------------------------------------------------------------------
Static function declarations
-------------------------------------------------------------------------- */
/* Figure out which generation to collect
*/
+ N = 0;
for (g = 0; g < RtsFlags.GcFlags.generations; g++) {
if (generations[g].steps[0].n_blocks >= generations[g].max_blocks) {
N = g;
zeroMutableList(generations[RtsFlags.GcFlags.generations-1].mut_list);
}
+ /* Save the old to-space if we're doing a two-space collection
+ */
+ if (RtsFlags.GcFlags.generations == 1) {
+ old_to_space = g0s0->to_space;
+ g0s0->to_space = NULL;
+ }
+
/* Initialise to-space in all the generations/steps that we're
* collecting.
*/
generations[g].mut_list = END_MUT_LIST;
for (s = 0; s < generations[g].n_steps; s++) {
+
/* generation 0, step 0 doesn't need to-space */
- if (g == 0 && s == 0) { continue; }
+ if (g == 0 && s == 0 && RtsFlags.GcFlags.generations > 1) {
+ continue;
+ }
+
/* Get a free block for to-space. Extra blocks will be chained on
* as necessary.
*/
step->to_blocks = 0;
step->new_large_objects = NULL;
step->scavenged_large_objects = NULL;
-#ifdef DEBUG
- /* retain these so we can sanity-check later on */
- step->old_scan = step->scan;
- step->old_scan_bd = step->scan_bd;
-#endif
}
}
/* -----------------------------------------------------------------------
- * follow all the roots that the application knows about.
- */
- evac_gen = 0;
- get_roots();
-
- /* follow all the roots that we know about:
+ * follow all the roots that we know about:
* - mutable lists from each generation > N
* we want to *scavenge* these roots, not evacuate them: they're not
* going to move in this GC.
*/
{
StgMutClosure *tmp, **pp;
- for (g = RtsFlags.GcFlags.generations-1; g > N; g--) {
- /* the act of scavenging the mutable list for this generation
- * might place more objects on the mutable list itself. So we
- * place the current mutable list in a temporary, scavenge it,
- * and then append it to the new list.
- */
- tmp = generations[g].mut_list;
+ for (g = N+1; g < RtsFlags.GcFlags.generations; g++) {
+ generations[g].saved_mut_list = generations[g].mut_list;
generations[g].mut_list = END_MUT_LIST;
- tmp = scavenge_mutable_list(tmp, g);
+ }
+ for (g = RtsFlags.GcFlags.generations-1; g > N; g--) {
+ tmp = scavenge_mutable_list(generations[g].saved_mut_list, g);
pp = &generations[g].mut_list;
while (*pp != END_MUT_LIST) {
pp = &(*pp)->mut_link;
}
*pp = tmp;
}
- }
+ }
+
+ /* follow all the roots that the application knows about.
+ */
+ evac_gen = 0;
+ get_roots();
+
/* And don't forget to mark the TSO if we got here direct from
* Haskell! */
if (CurrentTSO) {
}
}
+ /* Set the maximum blocks for the oldest generation, based on twice
+ * the amount of live data now, adjusted to fit the maximum heap
+ * size if necessary.
+ *
+ * This is an approximation, since in the worst case we'll need
+ * twice the amount of live data plus whatever space the other
+ * generations need.
+ */
+ if (RtsFlags.GcFlags.generations > 1) {
+ if (major_gc) {
+ oldest_gen->max_blocks =
+ stg_max(oldest_gen->steps[0].to_blocks * RtsFlags.GcFlags.oldGenFactor,
+ RtsFlags.GcFlags.minOldGenSize);
+ if (oldest_gen->max_blocks > RtsFlags.GcFlags.maxHeapSize / 2) {
+ oldest_gen->max_blocks = RtsFlags.GcFlags.maxHeapSize / 2;
+ if (((int)oldest_gen->max_blocks -
+ (int)oldest_gen->steps[0].to_blocks) <
+ (RtsFlags.GcFlags.pcFreeHeap *
+ RtsFlags.GcFlags.maxHeapSize / 200)) {
+ heapOverflow();
+ }
+ }
+ }
+ } else {
+ /* For a two-space collector, we need to resize the nursery. */
+
+ /* set up a new nursery. Allocate a nursery size based on a
+ * function of the amount of live data (currently a factor of 2,
+ * should be configurable (ToDo)). Use the blocks from the old
+ * nursery if possible, freeing up any left over blocks.
+ *
+ * If we get near the maximum heap size, then adjust our nursery
+ * size accordingly. If the nursery is the same size as the live
+ * data (L), then we need 3L bytes. We can reduce the size of the
+ * nursery to bring the required memory down near 2L bytes.
+ *
+ * A normal 2-space collector would need 4L bytes to give the same
+ * performance we get from 3L bytes, reducing to the same
+ * performance at 2L bytes.
+ */
+ nat blocks = g0s0->to_blocks;
+
+ if ( blocks * 4 > RtsFlags.GcFlags.maxHeapSize ) {
+ int adjusted_blocks; /* signed on purpose */
+ int pc_free;
+
+ adjusted_blocks = (RtsFlags.GcFlags.maxHeapSize - 2 * blocks);
+ IF_DEBUG(gc, fprintf(stderr, "Near maximum heap size of 0x%x blocks, blocks = %d, adjusted to %d\n", RtsFlags.GcFlags.maxHeapSize, blocks, adjusted_blocks));
+ pc_free = adjusted_blocks * 100 / RtsFlags.GcFlags.maxHeapSize;
+ if (pc_free < RtsFlags.GcFlags.pcFreeHeap) /* might even be < 0 */ {
+ heapOverflow();
+ }
+ blocks = adjusted_blocks;
+
+ } else {
+ blocks *= 2;
+ if (blocks < RtsFlags.GcFlags.minAllocAreaSize) {
+ blocks = RtsFlags.GcFlags.minAllocAreaSize;
+ }
+ }
+
+ if (nursery_blocks < blocks) {
+ IF_DEBUG(gc, fprintf(stderr, "Increasing size of nursery to %d blocks\n",
+ blocks));
+ g0s0->blocks = allocNursery(g0s0->blocks, blocks-nursery_blocks);
+ } else {
+ bdescr *next_bd;
+
+ IF_DEBUG(gc, fprintf(stderr, "Decreasing size of nursery to %d blocks\n",
+ blocks));
+ for (bd = g0s0->blocks; nursery_blocks > blocks; nursery_blocks--) {
+ next_bd = bd->link;
+ freeGroup(bd);
+ bd = next_bd;
+ }
+ g0s0->blocks = bd;
+ }
+
+ g0s0->n_blocks = nursery_blocks = blocks;
+ }
+
/* run through all the generations/steps and tidy up
*/
for (g = 0; g < RtsFlags.GcFlags.generations; g++) {
bdescr *next;
step = &generations[g].steps[s];
- if (!(g == 0 && s == 0)) {
+ if (!(g == 0 && s == 0 && RtsFlags.GcFlags.generations > 1)) {
/* Tidy the end of the to-space chains */
step->hp_bd->free = step->hp;
step->hp_bd->link = NULL;
}
step->large_objects = step->scavenged_large_objects;
- /* Set the maximum blocks for this generation,
- * using an arbitrary factor of the no. of blocks in step 0.
+ /* Set the maximum blocks for this generation, interpolating
+ * between the maximum size of the oldest and youngest
+ * generations.
+ *
+ * max_blocks = alloc_area_size +
+ * (oldgen_max_blocks - alloc_area_size) * G
+ * -----------------------------------------
+ * oldest_gen
*/
if (g != 0) {
- generation *gen = &generations[g];
- gen->max_blocks =
- stg_max(gen->steps[s].n_blocks * 2,
- RtsFlags.GcFlags.minAllocAreaSize * 4);
- if (gen->max_blocks > RtsFlags.GcFlags.maxHeapSize / 2) {
- gen->max_blocks = RtsFlags.GcFlags.maxHeapSize / 2;
- if (((int)gen->max_blocks - (int)gen->steps[0].n_blocks) <
- (RtsFlags.GcFlags.pcFreeHeap *
- RtsFlags.GcFlags.maxHeapSize / 200)) {
- heapOverflow();
- }
- }
+ generations[g].max_blocks =
+ RtsFlags.GcFlags.minAllocAreaSize +
+ (((oldest_gen->max_blocks - RtsFlags.GcFlags.minAllocAreaSize) * g)
+ / (RtsFlags.GcFlags.generations-1));
}
-
+
/* for older generations... */
} else {
}
}
+ /* Two-space collector:
+ * Free the old to-space, and estimate the amount of live data.
+ */
+ if (RtsFlags.GcFlags.generations == 1) {
+ if (old_to_space != NULL) {
+ freeChain(old_to_space);
+ }
+ for (bd = g0s0->to_space; bd != NULL; bd = bd->link) {
+ bd->evacuated = 0; /* now from-space */
+ }
+ live = g0s0->to_blocks * BLOCK_SIZE_W +
+ ((lnat)g0s0->hp_bd->free - (lnat)g0s0->hp_bd->start) / sizeof(W_);
+
+ /* Generational collector:
+ * estimate the amount of live data.
+ */
+ } else {
+ live = 0;
+ for (g = 0; g < RtsFlags.GcFlags.generations; g++) {
+ for (s = 0; s < generations[g].n_steps; s++) {
+ /* approximate amount of live data (doesn't take into account slop
+ * at end of each block). ToDo: this more accurately.
+ */
+ if (g == 0 && s == 0) { continue; }
+ step = &generations[g].steps[s];
+ live += step->n_blocks * BLOCK_SIZE_W +
+ ((lnat)step->hp_bd->free -(lnat)step->hp_bd->start) / sizeof(W_);
+ }
+ }
+ }
+
/* revert dead CAFs and update enteredCAFs list */
revertDeadCAFs();
}
current_nursery = g0s0->blocks;
- live = 0;
- for (g = 0; g < RtsFlags.GcFlags.generations; g++) {
- for (s = 0; s < generations[g].n_steps; s++) {
- /* approximate amount of live data (doesn't take into account slop
- * at end of each block). ToDo: this more accurately.
- */
- if (g == 0 && s == 0) { continue; }
- step = &generations[g].steps[s];
- live += step->n_blocks * BLOCK_SIZE_W +
- ((lnat)step->hp_bd->free -(lnat)step->hp_bd->start) / sizeof(W_);
- }
- }
-
/* Free the small objects allocated via allocate(), since this will
* all have been copied into G0S1 now.
*/
/* check sanity after GC */
#ifdef DEBUG
- for (g = 0; g <= N; g++) {
- for (s = 0; s < generations[g].n_steps; s++) {
- if (g == 0 && s == 0) { continue; }
- IF_DEBUG(sanity, checkHeap(generations[g].steps[s].blocks, NULL));
- IF_DEBUG(sanity, checkChain(generations[g].steps[s].large_objects));
+ if (RtsFlags.GcFlags.generations == 1) {
+ IF_DEBUG(sanity, checkHeap(g0s0->to_space, NULL));
+ IF_DEBUG(sanity, checkChain(g0s0->large_objects));
+ } else {
+
+ for (g = 0; g <= N; g++) {
+ for (s = 0; s < generations[g].n_steps; s++) {
+ if (g == 0 && s == 0) { continue; }
+ IF_DEBUG(sanity, checkHeap(generations[g].steps[s].blocks, NULL));
+ }
}
- }
- for (g = N+1; g < RtsFlags.GcFlags.generations; g++) {
- for (s = 0; s < generations[g].n_steps; s++) {
- IF_DEBUG(sanity, checkHeap(generations[g].steps[s].old_scan_bd,
- generations[g].steps[s].old_scan));
- IF_DEBUG(sanity, checkChain(generations[g].steps[s].large_objects));
+ for (g = N+1; g < RtsFlags.GcFlags.generations; g++) {
+ for (s = 0; s < generations[g].n_steps; s++) {
+ IF_DEBUG(sanity, checkHeap(generations[g].steps[s].blocks,
+ generations[g].steps[s].blocks->start));
+ IF_DEBUG(sanity, checkChain(generations[g].steps[s].large_objects));
+ }
}
+ IF_DEBUG(sanity, checkFreeListSanity());
}
- IF_DEBUG(sanity, checkFreeListSanity());
#endif
IF_DEBUG(gc, stat_describe_gens());
}
static __inline__ StgClosure *
-copy(StgClosure *src, W_ size, bdescr *bd)
+copy(StgClosure *src, nat size, bdescr *bd)
{
step *step;
P_ to, from, dest;
return (StgClosure *)dest;
}
+/* Special version of copy() for when we only want to copy the info
+ * pointer of an object, but reserve some padding after it. This is
+ * used to optimise evacuation of BLACKHOLEs.
+ */
+
+static __inline__ StgClosure *
+copyPart(StgClosure *src, nat size_to_reserve, nat size_to_copy, bdescr *bd)
+{
+ step *step;
+ P_ dest, to, from;
+
+ step = bd->step->to;
+ if (step->gen->no < evac_gen) {
+ step = &generations[evac_gen].steps[0];
+ }
+
+ if (step->hp + size_to_reserve >= step->hpLim) {
+ addBlock(step);
+ }
+
+ dest = step->hp;
+ step->hp += size_to_reserve;
+ for(to = dest, from = (P_)src; size_to_copy>0; --size_to_copy) {
+ *to++ = *from++;
+ }
+
+ return (StgClosure *)dest;
+}
+
static __inline__ void
upd_evacuee(StgClosure *p, StgClosure *dest)
{
case CAF_BLACKHOLE:
case BLACKHOLE:
- to = copy(q,BLACKHOLE_sizeW(),bd);
+ to = copyPart(q,BLACKHOLE_sizeW(),sizeofW(StgHeader),bd);
upd_evacuee(q,to);
return to;
+ case BLACKHOLE_BQ:
+ to = copy(q,BLACKHOLE_sizeW(),bd);
+ upd_evacuee(q,to);
+ evacuate_mutable((StgMutClosure *)to);
+ return to;
+
case THUNK_SELECTOR:
{
const StgInfoTable* selectee_info;
case CAF_UNENTERED:
case CAF_BLACKHOLE:
case BLACKHOLE:
+ case BLACKHOLE_BQ:
/* not evaluated yet */
break;
/* Large TSOs don't get moved, so no relocation is required.
*/
if (size >= LARGE_OBJECT_THRESHOLD/sizeof(W_)) {
- evacuate_large((P_)q, rtsFalse);
- tso->mut_link = NULL; /* see below */
+ evacuate_large((P_)q, rtsTrue);
return q;
/* To evacuate a small TSO, we need to relocate the update frame
relocate_TSO(tso, new_tso);
upd_evacuee(q,(StgClosure *)new_tso);
- /* don't evac_mutable - these things are marked mutable as
- * required. We *do* need to zero the mut_link field, though:
- * this TSO might have been on the mutable list for this
- * generation, but we're collecting this generation anyway so
- * we didn't follow the mutable list.
- */
- new_tso->mut_link = NULL;
-
+ evacuate_mutable((StgMutClosure *)new_tso);
return (StgClosure *)new_tso;
}
}
case CAF_BLACKHOLE:
case BLACKHOLE:
+ p += BLACKHOLE_sizeW();
+ break;
+
+ case BLACKHOLE_BQ:
{
- StgBlackHole *bh = (StgBlackHole *)p;
+ StgBlockingQueue *bh = (StgBlockingQueue *)p;
(StgClosure *)bh->blocking_queue =
evacuate((StgClosure *)bh->blocking_queue);
+ if (failed_to_evac) {
+ failed_to_evac = rtsFalse;
+ evacuate_mutable((StgMutClosure *)bh);
+ }
p += BLACKHOLE_sizeW();
break;
}
case CAF_BLACKHOLE:
case BLACKHOLE:
- {
- StgBlackHole *bh = (StgBlackHole *)p;
- (StgClosure *)bh->blocking_queue =
- evacuate((StgClosure *)bh->blocking_queue);
break;
- }
case THUNK_SELECTOR:
{
}
continue;
+ case MVAR:
+ {
+ StgMVar *mvar = (StgMVar *)p;
+ (StgClosure *)mvar->head = evacuate((StgClosure *)mvar->head);
+ (StgClosure *)mvar->tail = evacuate((StgClosure *)mvar->tail);
+ (StgClosure *)mvar->value = evacuate((StgClosure *)mvar->value);
+ prev = &p->mut_link;
+ continue;
+ }
+
case TSO:
/* follow ptrs and remove this from the mutable list */
{
}
continue;
+ case BLACKHOLE_BQ:
+ {
+ StgBlockingQueue *bh = (StgBlockingQueue *)p;
+ (StgClosure *)bh->blocking_queue =
+ evacuate((StgClosure *)bh->blocking_queue);
+ prev = &p->mut_link;
+ break;
+ }
+
default:
/* shouldn't have anything else on the mutables list */
barf("scavenge_mutable_object: non-mutable object?");
{
StgInd *ind = (StgInd *)p;
ind->indirectee = evacuate(ind->indirectee);
+
+ /* might fail to evacuate it, in which case we have to pop it
+ * back on the mutable list (and take it off the
+ * scavenged_static list because the static link and mut link
+ * pointers are one and the same).
+ */
+ if (failed_to_evac) {
+ failed_to_evac = rtsFalse;
+ scavenged_static_objects = STATIC_LINK(info,p);
+ ((StgMutClosure *)ind)->mut_link = oldest_gen->mut_list;
+ oldest_gen->mut_list = (StgMutClosure *)ind;
+ }
break;
}
barf("scavenge_static");
}
+ ASSERT(failed_to_evac == rtsFalse);
+
/* get the next static object from the list. Remeber, there might
* be more stuff on this list now that we've done some evacuating!
* (static_objects is a global)
goto follow_srt;
/* Specialised code for update frames, since they're so common.
- * We *know* the updatee points to a BLACKHOLE or CAF_BLACKHOLE,
- * so just inline the code to evacuate it here.
+ * We *know* the updatee points to a BLACKHOLE, CAF_BLACKHOLE,
+ * or BLACKHOLE_BQ, so just inline the code to evacuate it here.
*/
case UPDATE_FRAME:
{
continue;
} else {
bdescr *bd = Bdescr((P_)frame->updatee);
- ASSERT(type == BLACKHOLE || type == CAF_BLACKHOLE);
- if (bd->gen->no >= evac_gen && bd->gen->no > N) { continue; }
- to = copy(frame->updatee, BLACKHOLE_sizeW(), bd);
- upd_evacuee(frame->updatee,to);
- frame->updatee = to;
- continue;
+ if (bd->gen->no > N) {
+ if (bd->gen->no < evac_gen) {
+ failed_to_evac = rtsTrue;
+ }
+ continue;
+ }
+ switch (type) {
+ case BLACKHOLE:
+ case CAF_BLACKHOLE:
+ to = copyPart(frame->updatee, BLACKHOLE_sizeW(),
+ sizeofW(StgHeader), bd);
+ upd_evacuee(frame->updatee,to);
+ frame->updatee = to;
+ continue;
+ case BLACKHOLE_BQ:
+ to = copy(frame->updatee, BLACKHOLE_sizeW(), bd);
+ upd_evacuee(frame->updatee,to);
+ frame->updatee = to;
+ evacuate_mutable((StgMutClosure *)to);
+ continue;
+ default:
+ barf("scavenge_stack: UPDATE_FRAME updatee");
+ }
}
}
threadLazyBlackHole(StgTSO *tso)
{
StgUpdateFrame *update_frame;
- StgBlackHole *bh;
+ StgBlockingQueue *bh;
StgPtr stack_end;
stack_end = &tso->stack[tso->stack_size];
break;
case UPDATE_FRAME:
- bh = stgCast(StgBlackHole*,update_frame->updatee);
+ bh = (StgBlockingQueue *)update_frame->updatee;
/* if the thunk is already blackholed, it means we've also
* already blackholed the rest of the thunks on this stack,
* so we can stop early.
+ *
+ * The blackhole made for a CAF is a CAF_BLACKHOLE, so they
+ * don't interfere with this optimisation.
*/
+ if (bh->header.info == &BLACKHOLE_info) {
+ return;
+ }
- /* Don't for now: when we enter a CAF, we create a black hole on
- * the heap and make the update frame point to it. Thus the
- * above optimisation doesn't apply.
- */
- if (bh->header.info != &BLACKHOLE_info
- && bh->header.info != &CAF_BLACKHOLE_info) {
+ if (bh->header.info != &BLACKHOLE_BQ_info &&
+ bh->header.info != &CAF_BLACKHOLE_info) {
SET_INFO(bh,&BLACKHOLE_info);
- bh->blocking_queue = END_TSO_QUEUE;
}
update_frame = update_frame->link;
* added to the stack, rather than the way we see them in this
* walk. (It makes the next loop less confusing.)
*
- * Could stop if we find an update frame pointing to a black hole,
- * but see comment in threadLazyBlackHole().
+ * Stop if we find an update frame pointing to a black hole
+ * (see comment in threadLazyBlackHole()).
*/
next_frame = NULL;
frame->link = next_frame;
next_frame = frame;
frame = prev_frame;
+ if (get_itbl(frame)->type == UPDATE_FRAME
+ && frame->updatee->header.info == &BLACKHOLE_info) {
+ break;
+ }
}
/* Now, we're at the bottom. Frame points to the lowest update
* slower --SDM
*/
#if 0 /* do it properly... */
- if (GET_INFO(updatee_bypass) == BLACKHOLE_info
- || GET_INFO(updatee_bypass) == CAF_BLACKHOLE_info
- ) {
+ if (GET_INFO(updatee_bypass) == BLACKHOLE_BQ_info) {
/* Sigh. It has one. Don't lose those threads! */
- if (GET_INFO(updatee_keep) == BLACKHOLE_info
- || GET_INFO(updatee_keep) == CAF_BLACKHOLE_info
- ) {
+ if (GET_INFO(updatee_keep) == BLACKHOLE_BQ_info) {
/* Urgh. Two queues. Merge them. */
- P_ keep_tso = ((StgBlackHole *)updatee_keep)->blocking_queue;
+ P_ keep_tso = ((StgBlockingQueue *)updatee_keep)->blocking_queue;
while (keep_tso->link != END_TSO_QUEUE) {
keep_tso = keep_tso->link;
}
- keep_tso->link = ((StgBlackHole *)updatee_bypass)->blocking_queue;
+ keep_tso->link = ((StgBlockingQueue *)updatee_bypass)->blocking_queue;
} else {
/* For simplicity, just swap the BQ for the BH */
/* Do lazy black-holing.
*/
if (is_update_frame) {
- StgBlackHole *bh = (StgBlackHole *)frame->updatee;
- if (bh->header.info != &BLACKHOLE_info
- && bh->header.info != &CAF_BLACKHOLE_info
- ) {
+ StgBlockingQueue *bh = (StgBlockingQueue *)frame->updatee;
+ if (bh->header.info != &BLACKHOLE_BQ_info &&
+ bh->header.info != &CAF_BLACKHOLE_info) {
SET_INFO(bh,&BLACKHOLE_info);
- bh->blocking_queue = END_TSO_QUEUE;
}
}
projects / ghc-hetmet.git / blobdiff