https://github.com/mozilla/gecko-dev
Tip revision: 5f72803c6e7aa1b884da20485ccf7f09b885cf7e authored by seabld on 07 November 2011, 20:49:47 UTC
Added tag SEAMONKEY_2_5b4_RELEASE for changeset FIREFOX_8_0b6_BUILD2. CLOSED TREE a=release
Added tag SEAMONKEY_2_5b4_RELEASE for changeset FIREFOX_8_0b6_BUILD2. CLOSED TREE a=release
Tip revision: 5f72803
jsgc.cpp
/* -*- Mode: C++; tab-width: 8; indent-tabs-mode: nil; c-basic-offset: 4 -*-
* vim: set ts=8 sw=4 et tw=78:
*
* ***** BEGIN LICENSE BLOCK *****
* Version: MPL 1.1/GPL 2.0/LGPL 2.1
*
* The contents of this file are subject to the Mozilla Public License Version
* 1.1 (the "License"); you may not use this file except in compliance with
* the License. You may obtain a copy of the License at
* http://www.mozilla.org/MPL/
*
* Software distributed under the License is distributed on an "AS IS" basis,
* WITHOUT WARRANTY OF ANY KIND, either express or implied. See the License
* for the specific language governing rights and limitations under the
* License.
*
* The Original Code is Mozilla Communicator client code, released
* March 31, 1998.
*
* The Initial Developer of the Original Code is
* Netscape Communications Corporation.
* Portions created by the Initial Developer are Copyright (C) 1998
* the Initial Developer. All Rights Reserved.
*
* Contributor(s):
*
* Alternatively, the contents of this file may be used under the terms of
* either of the GNU General Public License Version 2 or later (the "GPL"),
* or the GNU Lesser General Public License Version 2.1 or later (the "LGPL"),
* in which case the provisions of the GPL or the LGPL are applicable instead
* of those above. If you wish to allow use of your version of this file only
* under the terms of either the GPL or the LGPL, and not to allow others to
* use your version of this file under the terms of the MPL, indicate your
* decision by deleting the provisions above and replace them with the notice
* and other provisions required by the GPL or the LGPL. If you do not delete
* the provisions above, a recipient may use your version of this file under
* the terms of any one of the MPL, the GPL or the LGPL.
*
* ***** END LICENSE BLOCK ***** */
/*
* JS Mark-and-Sweep Garbage Collector.
*
* This GC allocates fixed-sized things with sizes up to GC_NBYTES_MAX (see
* jsgc.h). It allocates from a special GC arena pool with each arena allocated
* using malloc. It uses an ideally parallel array of flag bytes to hold the
* mark bit, finalizer type index, etc.
*
* XXX swizzle page to freelist for better locality of reference
*/
#include <math.h>
#include <string.h> /* for memset used when DEBUG */
#include "jstypes.h"
#include "jsstdint.h"
#include "jsutil.h"
#include "jshash.h"
#include "jsbit.h"
#include "jsclist.h"
#include "jsprf.h"
#include "jsapi.h"
#include "jsatom.h"
#include "jscompartment.h"
#include "jscrashreport.h"
#include "jscrashformat.h"
#include "jscntxt.h"
#include "jsversion.h"
#include "jsdbgapi.h"
#include "jsexn.h"
#include "jsfun.h"
#include "jsgc.h"
#include "jsgcchunk.h"
#include "jsgcmark.h"
#include "jshashtable.h"
#include "jsinterp.h"
#include "jsiter.h"
#include "jslock.h"
#include "jsnum.h"
#include "jsobj.h"
#include "jsparse.h"
#include "jsprobes.h"
#include "jsproxy.h"
#include "jsscope.h"
#include "jsscript.h"
#include "jsstaticcheck.h"
#include "jswatchpoint.h"
#include "jsweakmap.h"
#if JS_HAS_XML_SUPPORT
#include "jsxml.h"
#endif
#include "methodjit/MethodJIT.h"
#include "vm/String.h"
#include "vm/Debugger.h"
#include "jsobjinlines.h"
#include "vm/String-inl.h"
#ifdef MOZ_VALGRIND
# define JS_VALGRIND
#endif
#ifdef JS_VALGRIND
# include <valgrind/memcheck.h>
#endif
using namespace js;
using namespace js::gc;
/*
* Check that JSTRACE_XML follows JSTRACE_OBJECT and JSTRACE_STRING.
*/
JS_STATIC_ASSERT(JSTRACE_OBJECT == 0);
JS_STATIC_ASSERT(JSTRACE_STRING == 1);
JS_STATIC_ASSERT(JSTRACE_SHAPE == 2);
JS_STATIC_ASSERT(JSTRACE_XML == 3);
/*
* JS_IS_VALID_TRACE_KIND assumes that JSTRACE_SHAPE is the last non-xml
* trace kind when JS_HAS_XML_SUPPORT is false.
*/
JS_STATIC_ASSERT(JSTRACE_SHAPE + 1 == JSTRACE_XML);
namespace js {
namespace gc {
/* This array should be const, but that doesn't link right under GCC. */
FinalizeKind slotsToThingKind[] = {
/* 0 */ FINALIZE_OBJECT0, FINALIZE_OBJECT2, FINALIZE_OBJECT2, FINALIZE_OBJECT4,
/* 4 */ FINALIZE_OBJECT4, FINALIZE_OBJECT8, FINALIZE_OBJECT8, FINALIZE_OBJECT8,
/* 8 */ FINALIZE_OBJECT8, FINALIZE_OBJECT12, FINALIZE_OBJECT12, FINALIZE_OBJECT12,
/* 12 */ FINALIZE_OBJECT12, FINALIZE_OBJECT16, FINALIZE_OBJECT16, FINALIZE_OBJECT16,
/* 16 */ FINALIZE_OBJECT16
};
JS_STATIC_ASSERT(JS_ARRAY_LENGTH(slotsToThingKind) == SLOTS_TO_THING_KIND_LIMIT);
const uint8 GCThingSizeMap[] = {
sizeof(JSObject), /* FINALIZE_OBJECT0 */
sizeof(JSObject), /* FINALIZE_OBJECT0_BACKGROUND */
sizeof(JSObject_Slots2), /* FINALIZE_OBJECT2 */
sizeof(JSObject_Slots2), /* FINALIZE_OBJECT2_BACKGROUND */
sizeof(JSObject_Slots4), /* FINALIZE_OBJECT4 */
sizeof(JSObject_Slots4), /* FINALIZE_OBJECT4_BACKGROUND */
sizeof(JSObject_Slots8), /* FINALIZE_OBJECT8 */
sizeof(JSObject_Slots8), /* FINALIZE_OBJECT8_BACKGROUND */
sizeof(JSObject_Slots12), /* FINALIZE_OBJECT12 */
sizeof(JSObject_Slots12), /* FINALIZE_OBJECT12_BACKGROUND */
sizeof(JSObject_Slots16), /* FINALIZE_OBJECT16 */
sizeof(JSObject_Slots16), /* FINALIZE_OBJECT16_BACKGROUND */
sizeof(JSFunction), /* FINALIZE_FUNCTION */
sizeof(Shape), /* FINALIZE_SHAPE */
#if JS_HAS_XML_SUPPORT
sizeof(JSXML), /* FINALIZE_XML */
#endif
sizeof(JSShortString), /* FINALIZE_SHORT_STRING */
sizeof(JSString), /* FINALIZE_STRING */
sizeof(JSExternalString), /* FINALIZE_EXTERNAL_STRING */
};
JS_STATIC_ASSERT(JS_ARRAY_LENGTH(GCThingSizeMap) == FINALIZE_LIMIT);
#ifdef DEBUG
void
ArenaHeader::checkSynchronizedWithFreeList() const
{
/*
* Do not allow to access the free list when its real head is still stored
* in FreeLists and is not synchronized with this one.
*/
JS_ASSERT(allocated());
/*
* We can be called from the background finalization thread when the free
* list in the compartment can mutate at any moment. We cannot do any
* checks in this case.
*/
if (!compartment->rt->gcRunning)
return;
FreeSpan firstSpan = FreeSpan::decodeOffsets(arenaAddress(), firstFreeSpanOffsets);
if (firstSpan.isEmpty())
return;
FreeSpan *list = &compartment->freeLists.lists[getThingKind()];
if (list->isEmpty() || firstSpan.arenaAddress() != list->arenaAddress())
return;
/*
* Here this arena has free things, FreeList::lists[thingKind] is not
* empty and also points to this arena. Thus they must the same.
*/
JS_ASSERT(firstSpan.isSameNonEmptySpan(list));
}
#endif
template<typename T>
inline bool
Arena::finalize(JSContext *cx)
{
JS_ASSERT(aheader.allocated());
JS_ASSERT(!aheader.getMarkingDelay()->link);
uintptr_t thing = thingsStart(sizeof(T));
uintptr_t lastByte = thingsEnd() - 1;
FreeSpan nextFree(aheader.getFirstFreeSpan());
nextFree.checkSpan();
FreeSpan newListHead;
FreeSpan *newListTail = &newListHead;
uintptr_t newFreeSpanStart = 0;
bool allClear = true;
#ifdef DEBUG
size_t nmarked = 0;
#endif
for (;; thing += sizeof(T)) {
JS_ASSERT(thing <= lastByte + 1);
if (thing == nextFree.first) {
JS_ASSERT(nextFree.last <= lastByte);
if (nextFree.last == lastByte)
break;
JS_ASSERT(Arena::isAligned(nextFree.last, sizeof(T)));
if (!newFreeSpanStart)
newFreeSpanStart = thing;
thing = nextFree.last;
nextFree = *nextFree.nextSpan();
nextFree.checkSpan();
} else {
T *t = reinterpret_cast<T *>(thing);
if (t->isMarked()) {
allClear = false;
#ifdef DEBUG
nmarked++;
#endif
if (newFreeSpanStart) {
JS_ASSERT(thing >= thingsStart(sizeof(T)) + sizeof(T));
newListTail->first = newFreeSpanStart;
newListTail->last = thing - sizeof(T);
newListTail = newListTail->nextSpanUnchecked(sizeof(T));
newFreeSpanStart = 0;
}
} else {
if (!newFreeSpanStart)
newFreeSpanStart = thing;
t->finalize(cx);
JS_POISON(t, JS_FREE_PATTERN, sizeof(T));
}
}
}
if (allClear) {
JS_ASSERT(newListTail == &newListHead);
JS_ASSERT(newFreeSpanStart == thingsStart(sizeof(T)));
return true;
}
newListTail->first = newFreeSpanStart ? newFreeSpanStart : nextFree.first;
JS_ASSERT(Arena::isAligned(newListTail->first, sizeof(T)));
newListTail->last = lastByte;
#ifdef DEBUG
size_t nfree = 0;
for (const FreeSpan *span = &newListHead; span != newListTail; span = span->nextSpan()) {
span->checkSpan();
JS_ASSERT(Arena::isAligned(span->first, sizeof(T)));
JS_ASSERT(Arena::isAligned(span->last, sizeof(T)));
nfree += (span->last - span->first) / sizeof(T) + 1;
JS_ASSERT(nfree + nmarked <= thingsPerArena(sizeof(T)));
}
nfree += (newListTail->last + 1 - newListTail->first) / sizeof(T);
JS_ASSERT(nfree + nmarked == thingsPerArena(sizeof(T)));
#endif
aheader.setFirstFreeSpan(&newListHead);
return false;
}
/*
* Finalize arenas from the list. On return listHeadp points to the list of
* non-empty arenas.
*/
template<typename T>
static void
FinalizeArenas(JSContext *cx, ArenaHeader **listHeadp)
{
ArenaHeader **ap = listHeadp;
while (ArenaHeader *aheader = *ap) {
bool allClear = aheader->getArena()->finalize<T>(cx);
if (allClear) {
*ap = aheader->next;
aheader->chunk()->releaseArena(aheader);
} else {
ap = &aheader->next;
}
}
}
#ifdef DEBUG
bool
checkArenaListAllUnmarked(JSCompartment *comp)
{
for (unsigned i = 0; i < FINALIZE_LIMIT; i++) {
if (comp->arenas[i].markedThingsInArenaList())
return false;
}
return true;
}
#endif
} /* namespace gc */
} /* namespace js */
void
JSCompartment::finishArenaLists()
{
for (unsigned i = 0; i < FINALIZE_LIMIT; i++)
arenas[i].releaseAll(i);
}
void
Chunk::init(JSRuntime *rt)
{
info.runtime = rt;
info.age = 0;
info.numFree = ArenasPerChunk;
/* Assemble all arenas into a linked list and mark them as not allocated. */
ArenaHeader **prevp = &info.emptyArenaListHead;
Arena *end = &arenas[JS_ARRAY_LENGTH(arenas)];
for (Arena *a = &arenas[0]; a != end; ++a) {
#ifdef DEBUG
memset(a, ArenaSize, JS_FREE_PATTERN);
#endif
*prevp = &a->aheader;
a->aheader.setAsNotAllocated();
prevp = &a->aheader.next;
}
*prevp = NULL;
for (size_t i = 0; i != JS_ARRAY_LENGTH(markingDelay); ++i)
markingDelay[i].init();
/* We clear the bitmap to guard against xpc_IsGrayGCThing being called on
uninitialized data, which would happen before the first GC cycle. */
bitmap.clear();
/* The rest of info fields are initialized in PickChunk. */
}
inline Chunk **
GetAvailableChunkList(JSCompartment *comp)
{
JSRuntime *rt = comp->rt;
return comp->isSystemCompartment
? &rt->gcSystemAvailableChunkListHead
: &rt->gcUserAvailableChunkListHead;
}
inline void
Chunk::addToAvailableList(JSCompartment *comp)
{
Chunk **listHeadp = GetAvailableChunkList(comp);
JS_ASSERT(!info.prevp);
JS_ASSERT(!info.next);
info.prevp = listHeadp;
Chunk *head = *listHeadp;
if (head) {
JS_ASSERT(head->info.prevp == listHeadp);
head->info.prevp = &info.next;
}
info.next = head;
*listHeadp = this;
}
inline void
Chunk::removeFromAvailableList()
{
JS_ASSERT(info.prevp);
*info.prevp = info.next;
if (info.next) {
JS_ASSERT(info.next->info.prevp == &info.next);
info.next->info.prevp = info.prevp;
}
info.prevp = NULL;
info.next = NULL;
}
template <size_t thingSize>
ArenaHeader *
Chunk::allocateArena(JSContext *cx, unsigned thingKind)
{
JSCompartment *comp = cx->compartment;
JS_ASSERT(hasAvailableArenas());
ArenaHeader *aheader = info.emptyArenaListHead;
info.emptyArenaListHead = aheader->next;
aheader->init(comp, thingKind, thingSize);
--info.numFree;
if (!hasAvailableArenas())
removeFromAvailableList();
JSRuntime *rt = info.runtime;
Probes::resizeHeap(comp, rt->gcBytes, rt->gcBytes + ArenaSize);
JS_ATOMIC_ADD(&rt->gcBytes, ArenaSize);
JS_ATOMIC_ADD(&comp->gcBytes, ArenaSize);
if (comp->gcBytes >= comp->gcTriggerBytes)
TriggerCompartmentGC(comp);
return aheader;
}
void
Chunk::releaseArena(ArenaHeader *aheader)
{
JS_ASSERT(aheader->allocated());
JSRuntime *rt = info.runtime;
#ifdef JS_THREADSAFE
Maybe<AutoLockGC> maybeLock;
if (rt->gcHelperThread.sweeping)
maybeLock.construct(info.runtime);
#endif
JSCompartment *comp = aheader->compartment;
Probes::resizeHeap(comp, rt->gcBytes, rt->gcBytes - ArenaSize);
JS_ASSERT(size_t(rt->gcBytes) >= ArenaSize);
JS_ASSERT(size_t(comp->gcBytes) >= ArenaSize);
#ifdef JS_THREADSAFE
if (rt->gcHelperThread.sweeping) {
rt->reduceGCTriggerBytes(GC_HEAP_GROWTH_FACTOR * ArenaSize);
comp->reduceGCTriggerBytes(GC_HEAP_GROWTH_FACTOR * ArenaSize);
}
#endif
JS_ATOMIC_ADD(&rt->gcBytes, -int32(ArenaSize));
JS_ATOMIC_ADD(&comp->gcBytes, -int32(ArenaSize));
aheader->setAsNotAllocated();
aheader->next = info.emptyArenaListHead;
info.emptyArenaListHead = aheader;
++info.numFree;
if (info.numFree == 1) {
JS_ASSERT(!info.prevp);
JS_ASSERT(!info.next);
addToAvailableList(aheader->compartment);
} else if (!unused()) {
JS_ASSERT(info.prevp);
} else {
rt->gcChunkSet.remove(this);
removeFromAvailableList();
/*
* We keep empty chunks until we are done with finalization to allow
* calling IsAboutToBeFinalized/Cell::isMarked for finalized GC things
* in empty chunks. So we add the chunk to the empty set even during
* GC_SHRINK.
*/
info.age = 0;
info.next = rt->gcEmptyChunkListHead;
rt->gcEmptyChunkListHead = this;
rt->gcEmptyChunkCount++;
}
}
inline Chunk *
AllocateGCChunk(JSRuntime *rt)
{
Chunk *p = (Chunk *)rt->gcChunkAllocator->alloc();
#ifdef MOZ_GCTIMER
if (p)
JS_ATOMIC_INCREMENT(&newChunkCount);
#endif
return p;
}
inline void
ReleaseGCChunk(JSRuntime *rt, Chunk *p)
{
JS_ASSERT(p);
#ifdef MOZ_GCTIMER
JS_ATOMIC_INCREMENT(&destroyChunkCount);
#endif
rt->gcChunkAllocator->free_(p);
}
inline Chunk *
PickChunk(JSContext *cx)
{
JSCompartment *comp = cx->compartment;
JSRuntime *rt = comp->rt;
Chunk **listHeadp = GetAvailableChunkList(comp);
Chunk *chunk = *listHeadp;
if (chunk)
return chunk;
/*
* We do not have available chunks, either get one from the empty set or
* allocate one.
*/
chunk = rt->gcEmptyChunkListHead;
if (chunk) {
JS_ASSERT(chunk->unused());
JS_ASSERT(!rt->gcChunkSet.has(chunk));
JS_ASSERT(rt->gcEmptyChunkCount >= 1);
rt->gcEmptyChunkListHead = chunk->info.next;
rt->gcEmptyChunkCount--;
} else {
chunk = AllocateGCChunk(rt);
if (!chunk)
return NULL;
chunk->init(rt);
rt->gcChunkAllocationSinceLastGC = true;
}
/*
* FIXME bug 583732 - chunk is newly allocated and cannot be present in
* the table so using ordinary lookupForAdd is suboptimal here.
*/
GCChunkSet::AddPtr p = rt->gcChunkSet.lookupForAdd(chunk);
JS_ASSERT(!p);
if (!rt->gcChunkSet.add(p, chunk)) {
ReleaseGCChunk(rt, chunk);
return NULL;
}
chunk->info.prevp = NULL;
chunk->info.next = NULL;
chunk->addToAvailableList(comp);
return chunk;
}
static void
ExpireGCChunks(JSRuntime *rt, JSGCInvocationKind gckind)
{
AutoLockGC lock(rt);
/* Return old empty chunks to the system. */
for (Chunk **chunkp = &rt->gcEmptyChunkListHead; *chunkp; ) {
JS_ASSERT(rt->gcEmptyChunkCount);
Chunk *chunk = *chunkp;
JS_ASSERT(chunk->unused());
JS_ASSERT(!rt->gcChunkSet.has(chunk));
JS_ASSERT(chunk->info.age <= MAX_EMPTY_CHUNK_AGE);
if (gckind == GC_SHRINK || chunk->info.age == MAX_EMPTY_CHUNK_AGE) {
*chunkp = chunk->info.next;
--rt->gcEmptyChunkCount;
ReleaseGCChunk(rt, chunk);
} else {
/* Keep the chunk but increase its age. */
++chunk->info.age;
chunkp = &chunk->info.next;
}
}
}
JS_FRIEND_API(bool)
IsAboutToBeFinalized(JSContext *cx, const void *thing)
{
if (JSAtom::isStatic(thing))
return false;
JS_ASSERT(cx);
JSCompartment *thingCompartment = reinterpret_cast<const Cell *>(thing)->compartment();
JSRuntime *rt = cx->runtime;
JS_ASSERT(rt == thingCompartment->rt);
if (rt->gcCurrentCompartment != NULL && rt->gcCurrentCompartment != thingCompartment)
return false;
return !reinterpret_cast<const Cell *>(thing)->isMarked();
}
JS_FRIEND_API(bool)
js_GCThingIsMarked(void *thing, uintN color = BLACK)
{
JS_ASSERT(thing);
AssertValidColor(thing, color);
JS_ASSERT(!JSAtom::isStatic(thing));
return reinterpret_cast<Cell *>(thing)->isMarked(color);
}
/*
* 1/8 life for JIT code. After this number of microseconds have passed, 1/8 of all
* JIT code is discarded in inactive compartments, regardless of how often that
* code runs.
*/
static const int64 JIT_SCRIPT_EIGHTH_LIFETIME = 120 * 1000 * 1000;
JSBool
js_InitGC(JSRuntime *rt, uint32 maxbytes)
{
if (!rt->gcChunkSet.init(INITIAL_CHUNK_CAPACITY))
return false;
if (!rt->gcRootsHash.init(256))
return false;
if (!rt->gcLocksHash.init(256))
return false;
#ifdef JS_THREADSAFE
rt->gcLock = JS_NEW_LOCK();
if (!rt->gcLock)
return false;
rt->gcDone = JS_NEW_CONDVAR(rt->gcLock);
if (!rt->gcDone)
return false;
rt->requestDone = JS_NEW_CONDVAR(rt->gcLock);
if (!rt->requestDone)
return false;
if (!rt->gcHelperThread.init(rt))
return false;
#endif
/*
* Separate gcMaxMallocBytes from gcMaxBytes but initialize to maxbytes
* for default backward API compatibility.
*/
rt->gcMaxBytes = maxbytes;
rt->setGCMaxMallocBytes(maxbytes);
rt->gcEmptyArenaPoolLifespan = 30000;
/*
* The assigned value prevents GC from running when GC memory is too low
* (during JS engine start).
*/
rt->setGCLastBytes(8192, GC_NORMAL);
rt->gcJitReleaseTime = PRMJ_Now() + JIT_SCRIPT_EIGHTH_LIFETIME;
return true;
}
namespace js {
inline bool
InFreeList(ArenaHeader *aheader, uintptr_t addr)
{
if (!aheader->hasFreeThings())
return false;
FreeSpan firstSpan(aheader->getFirstFreeSpan());
for (const FreeSpan *span = &firstSpan;;) {
/* If the thing comes fore the current span, it's not free. */
if (addr < span->first)
return false;
/*
* If we find it inside the span, it's dead. We use here "<=" and not
* "<" even for the last span as we know that thing is inside the
* arena. Thus for the last span thing < span->end.
*/
if (addr <= span->last)
return true;
/*
* The last possible empty span is an the end of the arena. Here
* span->end < thing < thingsEnd and so we must have more spans.
*/
span = span->nextSpan();
}
}
template <typename T>
inline ConservativeGCTest
MarkArenaPtrConservatively(JSTracer *trc, ArenaHeader *aheader, uintptr_t addr)
{
JS_ASSERT(aheader->allocated());
JS_ASSERT(sizeof(T) == aheader->getThingSize());
uintptr_t offset = addr & ArenaMask;
uintptr_t minOffset = Arena::thingsStartOffset(sizeof(T));
if (offset < minOffset)
return CGCT_NOTARENA;
/* addr can point inside the thing so we must align the address. */
uintptr_t shift = (offset - minOffset) % sizeof(T);
addr -= shift;
/*
* Check if the thing is free. We must use the list of free spans as at
* this point we no longer have the mark bits from the previous GC run and
* we must account for newly allocated things.
*/
if (InFreeList(aheader, addr))
return CGCT_NOTLIVE;
T *thing = reinterpret_cast<T *>(addr);
MarkRoot(trc, thing, "machine stack");
#ifdef JS_DUMP_CONSERVATIVE_GC_ROOTS
if (IS_GC_MARKING_TRACER(trc)) {
GCMarker *marker = static_cast<GCMarker *>(trc);
if (marker->conservativeDumpFileName)
marker->conservativeRoots.append(thing);
if (shift)
marker->conservativeStats.unaligned++;
}
#endif
return CGCT_VALID;
}
/*
* Returns CGCT_VALID and mark it if the w can be a live GC thing and sets
* thingKind accordingly. Otherwise returns the reason for rejection.
*/
inline ConservativeGCTest
MarkIfGCThingWord(JSTracer *trc, jsuword w)
{
/*
* We assume that the compiler never uses sub-word alignment to store
* pointers and does not tag pointers on its own. Additionally, the value
* representation for all values and the jsid representation for GC-things
* do not touch the low two bits. Thus any word with the low two bits set
* is not a valid GC-thing.
*/
JS_STATIC_ASSERT(JSID_TYPE_STRING == 0 && JSID_TYPE_OBJECT == 4);
if (w & 0x3)
return CGCT_LOWBITSET;
/*
* An object jsid has its low bits tagged. In the value representation on
* 64-bit, the high bits are tagged.
*/
const jsuword JSID_PAYLOAD_MASK = ~jsuword(JSID_TYPE_MASK);
#if JS_BITS_PER_WORD == 32
jsuword addr = w & JSID_PAYLOAD_MASK;
#elif JS_BITS_PER_WORD == 64
jsuword addr = w & JSID_PAYLOAD_MASK & JSVAL_PAYLOAD_MASK;
#endif
Chunk *chunk = Chunk::fromAddress(addr);
if (!trc->context->runtime->gcChunkSet.has(chunk))
return CGCT_NOTCHUNK;
/*
* We query for pointers outside the arena array after checking for an
* allocated chunk. Such pointers are rare and we want to reject them
* after doing more likely rejections.
*/
if (!Chunk::withinArenasRange(addr))
return CGCT_NOTARENA;
ArenaHeader *aheader = &chunk->arenas[Chunk::arenaIndex(addr)].aheader;
if (!aheader->allocated())
return CGCT_FREEARENA;
ConservativeGCTest test;
unsigned thingKind = aheader->getThingKind();
switch (thingKind) {
case FINALIZE_OBJECT0:
case FINALIZE_OBJECT0_BACKGROUND:
test = MarkArenaPtrConservatively<JSObject>(trc, aheader, addr);
break;
case FINALIZE_OBJECT2:
case FINALIZE_OBJECT2_BACKGROUND:
test = MarkArenaPtrConservatively<JSObject_Slots2>(trc, aheader, addr);
break;
case FINALIZE_OBJECT4:
case FINALIZE_OBJECT4_BACKGROUND:
test = MarkArenaPtrConservatively<JSObject_Slots4>(trc, aheader, addr);
break;
case FINALIZE_OBJECT8:
case FINALIZE_OBJECT8_BACKGROUND:
test = MarkArenaPtrConservatively<JSObject_Slots8>(trc, aheader, addr);
break;
case FINALIZE_OBJECT12:
case FINALIZE_OBJECT12_BACKGROUND:
test = MarkArenaPtrConservatively<JSObject_Slots12>(trc, aheader, addr);
break;
case FINALIZE_OBJECT16:
case FINALIZE_OBJECT16_BACKGROUND:
test = MarkArenaPtrConservatively<JSObject_Slots16>(trc, aheader, addr);
break;
case FINALIZE_STRING:
test = MarkArenaPtrConservatively<JSString>(trc, aheader, addr);
break;
case FINALIZE_EXTERNAL_STRING:
test = MarkArenaPtrConservatively<JSExternalString>(trc, aheader, addr);
break;
case FINALIZE_SHORT_STRING:
test = MarkArenaPtrConservatively<JSShortString>(trc, aheader, addr);
break;
case FINALIZE_FUNCTION:
test = MarkArenaPtrConservatively<JSFunction>(trc, aheader, addr);
break;
case FINALIZE_SHAPE:
test = MarkArenaPtrConservatively<Shape>(trc, aheader, addr);
break;
#if JS_HAS_XML_SUPPORT
case FINALIZE_XML:
test = MarkArenaPtrConservatively<JSXML>(trc, aheader, addr);
break;
#endif
default:
test = CGCT_WRONGTAG;
JS_NOT_REACHED("wrong tag");
}
return test;
}
static void
MarkWordConservatively(JSTracer *trc, jsuword w)
{
/*
* The conservative scanner may access words that valgrind considers as
* undefined. To avoid false positives and not to alter valgrind view of
* the memory we make as memcheck-defined the argument, a copy of the
* original word. See bug 572678.
*/
#ifdef JS_VALGRIND
VALGRIND_MAKE_MEM_DEFINED(&w, sizeof(w));
#endif
MarkIfGCThingWord(trc, w);
}
static void
MarkRangeConservatively(JSTracer *trc, const jsuword *begin, const jsuword *end)
{
JS_ASSERT(begin <= end);
for (const jsuword *i = begin; i != end; ++i)
MarkWordConservatively(trc, *i);
}
static void
MarkThreadDataConservatively(JSTracer *trc, ThreadData *td)
{
ConservativeGCThreadData *ctd = &td->conservativeGC;
JS_ASSERT(ctd->hasStackToScan());
jsuword *stackMin, *stackEnd;
#if JS_STACK_GROWTH_DIRECTION > 0
stackMin = td->nativeStackBase;
stackEnd = ctd->nativeStackTop;
#else
stackMin = ctd->nativeStackTop + 1;
stackEnd = td->nativeStackBase;
#endif
JS_ASSERT(stackMin <= stackEnd);
MarkRangeConservatively(trc, stackMin, stackEnd);
MarkRangeConservatively(trc, ctd->registerSnapshot.words,
JS_ARRAY_END(ctd->registerSnapshot.words));
}
void
MarkStackRangeConservatively(JSTracer *trc, Value *beginv, Value *endv)
{
const jsuword *begin = beginv->payloadWord();
const jsuword *end = endv->payloadWord();;
#ifdef JS_NUNBOX32
/*
* With 64-bit jsvals on 32-bit systems, we can optimize a bit by
* scanning only the payloads.
*/
JS_ASSERT(begin <= end);
for (const jsuword *i = begin; i != end; i += sizeof(Value)/sizeof(jsuword))
MarkWordConservatively(trc, *i);
#else
MarkRangeConservatively(trc, begin, end);
#endif
}
void
MarkConservativeStackRoots(JSTracer *trc)
{
#ifdef JS_THREADSAFE
for (JSThread::Map::Range r = trc->context->runtime->threads.all(); !r.empty(); r.popFront()) {
JSThread *thread = r.front().value;
ConservativeGCThreadData *ctd = &thread->data.conservativeGC;
if (ctd->hasStackToScan()) {
JS_ASSERT_IF(!thread->data.requestDepth, thread->suspendCount);
MarkThreadDataConservatively(trc, &thread->data);
} else {
JS_ASSERT(!thread->suspendCount);
JS_ASSERT(thread->data.requestDepth <= ctd->requestThreshold);
}
}
#else
MarkThreadDataConservatively(trc, &trc->context->runtime->threadData);
#endif
}
JS_NEVER_INLINE void
ConservativeGCThreadData::recordStackTop()
{
/* Update the native stack pointer if it points to a bigger stack. */
jsuword dummy;
nativeStackTop = &dummy;
/*
* To record and update the register snapshot for the conservative
* scanning with the latest values we use setjmp.
*/
#if defined(_MSC_VER)
# pragma warning(push)
# pragma warning(disable: 4611)
#endif
(void) setjmp(registerSnapshot.jmpbuf);
#if defined(_MSC_VER)
# pragma warning(pop)
#endif
}
static inline void
RecordNativeStackTopForGC(JSContext *cx)
{
ConservativeGCThreadData *ctd = &JS_THREAD_DATA(cx)->conservativeGC;
#ifdef JS_THREADSAFE
/* Record the stack top here only if we are called from a request. */
JS_ASSERT(cx->thread()->data.requestDepth >= ctd->requestThreshold);
if (cx->thread()->data.requestDepth == ctd->requestThreshold)
return;
#endif
ctd->recordStackTop();
}
} /* namespace js */
#ifdef DEBUG
static void
CheckLeakedRoots(JSRuntime *rt);
#endif
void
js_FinishGC(JSRuntime *rt)
{
/* Delete all remaining Compartments. */
for (JSCompartment **c = rt->compartments.begin(); c != rt->compartments.end(); ++c) {
JSCompartment *comp = *c;
comp->finishArenaLists();
Foreground::delete_(comp);
}
rt->compartments.clear();
rt->atomsCompartment = NULL;
rt->gcSystemAvailableChunkListHead = NULL;
rt->gcUserAvailableChunkListHead = NULL;
for (GCChunkSet::Range r(rt->gcChunkSet.all()); !r.empty(); r.popFront())
ReleaseGCChunk(rt, r.front());
rt->gcChunkSet.clear();
for (Chunk *chunk = rt->gcEmptyChunkListHead; chunk; ) {
Chunk *next = chunk->info.next;
ReleaseGCChunk(rt, chunk);
chunk = next;
}
rt->gcEmptyChunkListHead = NULL;
rt->gcEmptyChunkCount = 0;
#ifdef JS_THREADSAFE
rt->gcHelperThread.finish(rt);
#endif
#ifdef DEBUG
if (!rt->gcRootsHash.empty())
CheckLeakedRoots(rt);
#endif
rt->gcRootsHash.clear();
rt->gcLocksHash.clear();
}
JSBool
js_AddRoot(JSContext *cx, Value *vp, const char *name)
{
JSBool ok = js_AddRootRT(cx->runtime, Jsvalify(vp), name);
if (!ok)
JS_ReportOutOfMemory(cx);
return ok;
}
JSBool
js_AddGCThingRoot(JSContext *cx, void **rp, const char *name)
{
JSBool ok = js_AddGCThingRootRT(cx->runtime, rp, name);
if (!ok)
JS_ReportOutOfMemory(cx);
return ok;
}
JS_FRIEND_API(JSBool)
js_AddRootRT(JSRuntime *rt, jsval *vp, const char *name)
{
/*
* Due to the long-standing, but now removed, use of rt->gcLock across the
* bulk of js_GC, API users have come to depend on JS_AddRoot etc. locking
* properly with a racing GC, without calling JS_AddRoot from a request.
* We have to preserve API compatibility here, now that we avoid holding
* rt->gcLock across the mark phase (including the root hashtable mark).
*/
AutoLockGC lock(rt);
js_WaitForGC(rt);
return !!rt->gcRootsHash.put((void *)vp,
RootInfo(name, JS_GC_ROOT_VALUE_PTR));
}
JS_FRIEND_API(JSBool)
js_AddGCThingRootRT(JSRuntime *rt, void **rp, const char *name)
{
/*
* Due to the long-standing, but now removed, use of rt->gcLock across the
* bulk of js_GC, API users have come to depend on JS_AddRoot etc. locking
* properly with a racing GC, without calling JS_AddRoot from a request.
* We have to preserve API compatibility here, now that we avoid holding
* rt->gcLock across the mark phase (including the root hashtable mark).
*/
AutoLockGC lock(rt);
js_WaitForGC(rt);
return !!rt->gcRootsHash.put((void *)rp,
RootInfo(name, JS_GC_ROOT_GCTHING_PTR));
}
JS_FRIEND_API(JSBool)
js_RemoveRoot(JSRuntime *rt, void *rp)
{
/*
* Due to the JS_RemoveRootRT API, we may be called outside of a request.
* Same synchronization drill as above in js_AddRoot.
*/
AutoLockGC lock(rt);
js_WaitForGC(rt);
rt->gcRootsHash.remove(rp);
rt->gcPoke = JS_TRUE;
return JS_TRUE;
}
typedef RootedValueMap::Range RootRange;
typedef RootedValueMap::Entry RootEntry;
typedef RootedValueMap::Enum RootEnum;
#ifdef DEBUG
static void
CheckLeakedRoots(JSRuntime *rt)
{
uint32 leakedroots = 0;
/* Warn (but don't assert) debug builds of any remaining roots. */
for (RootRange r = rt->gcRootsHash.all(); !r.empty(); r.popFront()) {
RootEntry &entry = r.front();
leakedroots++;
fprintf(stderr,
"JS engine warning: leaking GC root \'%s\' at %p\n",
entry.value.name ? entry.value.name : "", entry.key);
}
if (leakedroots > 0) {
if (leakedroots == 1) {
fprintf(stderr,
"JS engine warning: 1 GC root remains after destroying the JSRuntime at %p.\n"
" This root may point to freed memory. Objects reachable\n"
" through it have not been finalized.\n",
(void *) rt);
} else {
fprintf(stderr,
"JS engine warning: %lu GC roots remain after destroying the JSRuntime at %p.\n"
" These roots may point to freed memory. Objects reachable\n"
" through them have not been finalized.\n",
(unsigned long) leakedroots, (void *) rt);
}
}
}
void
js_DumpNamedRoots(JSRuntime *rt,
void (*dump)(const char *name, void *rp, JSGCRootType type, void *data),
void *data)
{
for (RootRange r = rt->gcRootsHash.all(); !r.empty(); r.popFront()) {
RootEntry &entry = r.front();
if (const char *name = entry.value.name)
dump(name, entry.key, entry.value.type, data);
}
}
#endif /* DEBUG */
uint32
js_MapGCRoots(JSRuntime *rt, JSGCRootMapFun map, void *data)
{
AutoLockGC lock(rt);
int ct = 0;
for (RootEnum e(rt->gcRootsHash); !e.empty(); e.popFront()) {
RootEntry &entry = e.front();
ct++;
intN mapflags = map(entry.key, entry.value.type, entry.value.name, data);
if (mapflags & JS_MAP_GCROOT_REMOVE)
e.removeFront();
if (mapflags & JS_MAP_GCROOT_STOP)
break;
}
return ct;
}
void
JSRuntime::setGCLastBytes(size_t lastBytes, JSGCInvocationKind gckind)
{
gcLastBytes = lastBytes;
size_t base = gckind == GC_SHRINK ? lastBytes : Max(lastBytes, GC_ALLOCATION_THRESHOLD);
float trigger = float(base) * GC_HEAP_GROWTH_FACTOR;
gcTriggerBytes = size_t(Min(float(gcMaxBytes), trigger));
}
void
JSRuntime::reduceGCTriggerBytes(uint32 amount) {
JS_ASSERT(amount > 0);
JS_ASSERT(gcTriggerBytes - amount >= 0);
if (gcTriggerBytes - amount < GC_ALLOCATION_THRESHOLD * GC_HEAP_GROWTH_FACTOR)
return;
gcTriggerBytes -= amount;
}
void
JSCompartment::setGCLastBytes(size_t lastBytes, JSGCInvocationKind gckind)
{
gcLastBytes = lastBytes;
size_t base = gckind == GC_SHRINK ? lastBytes : Max(lastBytes, GC_ALLOCATION_THRESHOLD);
float trigger = float(base) * GC_HEAP_GROWTH_FACTOR;
gcTriggerBytes = size_t(Min(float(rt->gcMaxBytes), trigger));
}
void
JSCompartment::reduceGCTriggerBytes(uint32 amount) {
JS_ASSERT(amount > 0);
JS_ASSERT(gcTriggerBytes - amount >= 0);
if (gcTriggerBytes - amount < GC_ALLOCATION_THRESHOLD * GC_HEAP_GROWTH_FACTOR)
return;
gcTriggerBytes -= amount;
}
namespace js {
namespace gc {
inline ArenaHeader *
ArenaList::searchForFreeArena()
{
while (ArenaHeader *aheader = *cursor) {
cursor = &aheader->next;
if (aheader->hasFreeThings())
return aheader;
}
return NULL;
}
template <size_t thingSize>
inline ArenaHeader *
ArenaList::getArenaWithFreeList(JSContext *cx, unsigned thingKind)
{
Chunk *chunk;
#ifdef JS_THREADSAFE
/*
* We cannot search the arena list for free things while the
* background finalization runs and can modify head or cursor at any
* moment.
*/
if (backgroundFinalizeState == BFS_DONE) {
check_arena_list:
if (ArenaHeader *aheader = searchForFreeArena())
return aheader;
}
AutoLockGC lock(cx->runtime);
for (;;) {
if (backgroundFinalizeState == BFS_JUST_FINISHED) {
/*
* Before we took the GC lock or while waiting for the background
* finalization to finish the latter added new arenas to the list.
* Check the list again for free things outside the GC lock.
*/
JS_ASSERT(*cursor);
backgroundFinalizeState = BFS_DONE;
goto check_arena_list;
}
JS_ASSERT(!*cursor);
chunk = PickChunk(cx);
if (chunk || backgroundFinalizeState == BFS_DONE)
break;
/*
* If the background finalization still runs, wait for it to
* finish and retry to check if it populated the arena list or
* added new empty arenas.
*/
JS_ASSERT(backgroundFinalizeState == BFS_RUN);
cx->runtime->gcHelperThread.waitBackgroundSweepEnd(cx->runtime, false);
JS_ASSERT(backgroundFinalizeState == BFS_JUST_FINISHED ||
backgroundFinalizeState == BFS_DONE);
}
#else /* !JS_THREADSAFE */
if (ArenaHeader *aheader = searchForFreeArena())
return aheader;
chunk = PickChunk(cx);
#endif /* !JS_THREADSAFE */
if (!chunk) {
GCREASON(CHUNK);
TriggerGC(cx->runtime);
return NULL;
}
/*
* While we still hold the GC lock get the arena from the chunk and add it
* to the head of the list before the cursor to prevent checking the arena
* for the free things.
*/
ArenaHeader *aheader = chunk->allocateArena<thingSize>(cx, thingKind);
aheader->next = head;
if (cursor == &head)
cursor = &aheader->next;
head = aheader;
return aheader;
}
template<typename T>
void
ArenaList::finalizeNow(JSContext *cx)
{
#ifdef JS_THREADSAFE
JS_ASSERT(backgroundFinalizeState == BFS_DONE);
#endif
FinalizeArenas<T>(cx, &head);
cursor = &head;
}
#ifdef JS_THREADSAFE
template<typename T>
inline void
ArenaList::finalizeLater(JSContext *cx)
{
JS_ASSERT_IF(head,
head->getThingKind() == FINALIZE_OBJECT0_BACKGROUND ||
head->getThingKind() == FINALIZE_OBJECT2_BACKGROUND ||
head->getThingKind() == FINALIZE_OBJECT4_BACKGROUND ||
head->getThingKind() == FINALIZE_OBJECT8_BACKGROUND ||
head->getThingKind() == FINALIZE_OBJECT12_BACKGROUND ||
head->getThingKind() == FINALIZE_OBJECT16_BACKGROUND ||
head->getThingKind() == FINALIZE_SHORT_STRING ||
head->getThingKind() == FINALIZE_STRING);
JS_ASSERT(!cx->runtime->gcHelperThread.sweeping);
/*
* The state can be just-finished if we have not allocated any GC things
* from the arena list after the previous background finalization.
*/
JS_ASSERT(backgroundFinalizeState == BFS_DONE ||
backgroundFinalizeState == BFS_JUST_FINISHED);
if (head && cx->gcBackgroundFree && cx->gcBackgroundFree->finalizeVector.append(head)) {
head = NULL;
cursor = &head;
backgroundFinalizeState = BFS_RUN;
} else {
JS_ASSERT_IF(!head, cursor == &head);
backgroundFinalizeState = BFS_DONE;
finalizeNow<T>(cx);
}
}
/*static*/ void
ArenaList::backgroundFinalize(JSContext *cx, ArenaHeader *listHead)
{
JS_ASSERT(listHead);
unsigned thingKind = listHead->getThingKind();
JSCompartment *comp = listHead->compartment;
ArenaList *al = &comp->arenas[thingKind];
switch (thingKind) {
default:
JS_NOT_REACHED("wrong kind");
break;
case FINALIZE_OBJECT0_BACKGROUND:
FinalizeArenas<JSObject>(cx, &listHead);
break;
case FINALIZE_OBJECT2_BACKGROUND:
FinalizeArenas<JSObject_Slots2>(cx, &listHead);
break;
case FINALIZE_OBJECT4_BACKGROUND:
FinalizeArenas<JSObject_Slots4>(cx, &listHead);
break;
case FINALIZE_OBJECT8_BACKGROUND:
FinalizeArenas<JSObject_Slots8>(cx, &listHead);
break;
case FINALIZE_OBJECT12_BACKGROUND:
FinalizeArenas<JSObject_Slots12>(cx, &listHead);
break;
case FINALIZE_OBJECT16_BACKGROUND:
FinalizeArenas<JSObject_Slots16>(cx, &listHead);
break;
case FINALIZE_STRING:
FinalizeArenas<JSString>(cx, &listHead);
break;
case FINALIZE_SHORT_STRING:
FinalizeArenas<JSShortString>(cx, &listHead);
break;
}
/*
* After we finish the finalization al->cursor must point to the end of
* the head list as we emptied the list before the background finalization
* and the allocation adds new arenas before the cursor.
*/
AutoLockGC lock(cx->runtime);
JS_ASSERT(al->backgroundFinalizeState == BFS_RUN);
JS_ASSERT(!*al->cursor);
if (listHead) {
*al->cursor = listHead;
al->backgroundFinalizeState = BFS_JUST_FINISHED;
} else {
al->backgroundFinalizeState = BFS_DONE;
}
}
#endif /* JS_THREADSAFE */
#ifdef DEBUG
bool
CheckAllocation(JSContext *cx)
{
#ifdef JS_THREADSAFE
JS_ASSERT(cx->thread());
#endif
JS_ASSERT(!cx->runtime->gcRunning);
return true;
}
#endif
inline bool
NeedLastDitchGC(JSContext *cx)
{
JSRuntime *rt = cx->runtime;
return rt->gcIsNeeded;
}
/*
* Return false only if the GC run but could not bring its memory usage under
* JSRuntime::gcMaxBytes.
*/
static bool
RunLastDitchGC(JSContext *cx)
{
JSRuntime *rt = cx->runtime;
#ifdef JS_THREADSAFE
Maybe<AutoUnlockAtomsCompartment> maybeUnlockAtomsCompartment;
if (cx->compartment == rt->atomsCompartment && rt->atomsCompartmentIsLocked)
maybeUnlockAtomsCompartment.construct(cx);
#endif
/* The last ditch GC preserves all atoms. */
AutoKeepAtoms keep(rt);
GCREASON(LASTDITCH);
js_GC(cx, rt->gcTriggerCompartment, GC_NORMAL);
#ifdef JS_THREADSAFE
if (rt->gcBytes >= rt->gcMaxBytes)
cx->runtime->gcHelperThread.waitBackgroundSweepEnd(cx->runtime);
#endif
return rt->gcBytes < rt->gcMaxBytes;
}
static inline bool
IsGCAllowed(JSContext *cx)
{
return !JS_ON_TRACE(cx) && !JS_THREAD_DATA(cx)->waiveGCQuota;
}
template <typename T>
inline void *
RefillTypedFreeList(JSContext *cx, unsigned thingKind)
{
JS_ASSERT(!cx->runtime->gcRunning);
/*
* For compatibility with older code we tolerate calling the allocator
* during the GC in optimized builds.
*/
if (cx->runtime->gcRunning)
return NULL;
JSCompartment *compartment = cx->compartment;
JS_ASSERT(compartment->freeLists.lists[thingKind].isEmpty());
bool canGC = IsGCAllowed(cx);
bool runGC = canGC && JS_UNLIKELY(NeedLastDitchGC(cx));
for (;;) {
if (runGC) {
if (!RunLastDitchGC(cx))
break;
/*
* The JSGC_END callback can legitimately allocate new GC
* things and populate the free list. If that happens, just
* return that list head.
*/
if (void *thing = compartment->freeLists.getNext(thingKind, sizeof(T)))
return thing;
}
ArenaHeader *aheader =
compartment->arenas[thingKind].getArenaWithFreeList<sizeof(T)>(cx, thingKind);
if (aheader) {
JS_ASSERT(sizeof(T) == aheader->getThingSize());
return compartment->freeLists.populate(aheader, thingKind, sizeof(T));
}
/*
* We failed to allocate any arena. Run the GC if we can unless we
* have done it already.
*/
if (!canGC || runGC)
break;
runGC = true;
}
js_ReportOutOfMemory(cx);
return NULL;
}
void *
RefillFinalizableFreeList(JSContext *cx, unsigned thingKind)
{
switch (thingKind) {
case FINALIZE_OBJECT0:
case FINALIZE_OBJECT0_BACKGROUND:
return RefillTypedFreeList<JSObject>(cx, thingKind);
case FINALIZE_OBJECT2:
case FINALIZE_OBJECT2_BACKGROUND:
return RefillTypedFreeList<JSObject_Slots2>(cx, thingKind);
case FINALIZE_OBJECT4:
case FINALIZE_OBJECT4_BACKGROUND:
return RefillTypedFreeList<JSObject_Slots4>(cx, thingKind);
case FINALIZE_OBJECT8:
case FINALIZE_OBJECT8_BACKGROUND:
return RefillTypedFreeList<JSObject_Slots8>(cx, thingKind);
case FINALIZE_OBJECT12:
case FINALIZE_OBJECT12_BACKGROUND:
return RefillTypedFreeList<JSObject_Slots12>(cx, thingKind);
case FINALIZE_OBJECT16:
case FINALIZE_OBJECT16_BACKGROUND:
return RefillTypedFreeList<JSObject_Slots16>(cx, thingKind);
case FINALIZE_STRING:
return RefillTypedFreeList<JSString>(cx, thingKind);
case FINALIZE_EXTERNAL_STRING:
return RefillTypedFreeList<JSExternalString>(cx, thingKind);
case FINALIZE_SHORT_STRING:
return RefillTypedFreeList<JSShortString>(cx, thingKind);
case FINALIZE_FUNCTION:
return RefillTypedFreeList<JSFunction>(cx, thingKind);
case FINALIZE_SHAPE:
return RefillTypedFreeList<Shape>(cx, thingKind);
#if JS_HAS_XML_SUPPORT
case FINALIZE_XML:
return RefillTypedFreeList<JSXML>(cx, thingKind);
#endif
default:
JS_NOT_REACHED("bad finalize kind");
return 0;
}
}
} /* namespace gc */
} /* namespace js */
uint32
js_GetGCThingTraceKind(void *thing)
{
return GetGCThingTraceKind(thing);
}
JSBool
js_LockGCThingRT(JSRuntime *rt, void *thing)
{
if (!thing)
return true;
AutoLockGC lock(rt);
if (GCLocks::Ptr p = rt->gcLocksHash.lookupWithDefault(thing, 0)) {
p->value++;
return true;
}
return false;
}
void
js_UnlockGCThingRT(JSRuntime *rt, void *thing)
{
if (!thing)
return;
AutoLockGC lock(rt);
GCLocks::Ptr p = rt->gcLocksHash.lookup(thing);
if (p) {
rt->gcPoke = true;
if (--p->value == 0)
rt->gcLocksHash.remove(p);
}
}
namespace js {
/*
* When the native stack is low, the GC does not call JS_TraceChildren to mark
* the reachable "children" of the thing. Rather the thing is put aside and
* JS_TraceChildren is called later with more space on the C stack.
*
* To implement such delayed marking of the children with minimal overhead for
* the normal case of sufficient native stack, the code adds a field per
* arena. The field markingDelay->link links all arenas with delayed things
* into a stack list with the pointer to stack top in
* GCMarker::unmarkedArenaStackTop. delayMarkingChildren adds
* arenas to the stack as necessary while markDelayedChildren pops the arenas
* from the stack until it empties.
*/
GCMarker::GCMarker(JSContext *cx)
: color(0),
unmarkedArenaStackTop(MarkingDelay::stackBottom()),
objStack(cx->runtime->gcMarkStackObjs, sizeof(cx->runtime->gcMarkStackObjs)),
ropeStack(cx->runtime->gcMarkStackRopes, sizeof(cx->runtime->gcMarkStackRopes)),
xmlStack(cx->runtime->gcMarkStackXMLs, sizeof(cx->runtime->gcMarkStackXMLs)),
largeStack(cx->runtime->gcMarkStackLarges, sizeof(cx->runtime->gcMarkStackLarges))
{
JS_TRACER_INIT(this, cx, NULL);
#ifdef DEBUG
markLaterArenas = 0;
#endif
#ifdef JS_DUMP_CONSERVATIVE_GC_ROOTS
conservativeDumpFileName = getenv("JS_DUMP_CONSERVATIVE_GC_ROOTS");
memset(&conservativeStats, 0, sizeof(conservativeStats));
#endif
}
GCMarker::~GCMarker()
{
#ifdef JS_DUMP_CONSERVATIVE_GC_ROOTS
dumpConservativeRoots();
#endif
}
void
GCMarker::delayMarkingChildren(const void *thing)
{
const Cell *cell = reinterpret_cast<const Cell *>(thing);
ArenaHeader *aheader = cell->arenaHeader();
if (aheader->getMarkingDelay()->link) {
/* Arena already scheduled to be marked later */
return;
}
aheader->getMarkingDelay()->link = unmarkedArenaStackTop;
unmarkedArenaStackTop = aheader;
#ifdef DEBUG
markLaterArenas++;
#endif
}
static void
MarkDelayedChildren(JSTracer *trc, ArenaHeader *aheader)
{
unsigned traceKind = GetFinalizableTraceKind(aheader->getThingKind());
size_t thingSize = aheader->getThingSize();
Arena *a = aheader->getArena();
uintptr_t end = a->thingsEnd();
for (uintptr_t thing = a->thingsStart(thingSize); thing != end; thing += thingSize) {
Cell *t = reinterpret_cast<Cell *>(thing);
if (t->isMarked())
JS_TraceChildren(trc, t, traceKind);
}
}
void
GCMarker::markDelayedChildren()
{
while (unmarkedArenaStackTop != MarkingDelay::stackBottom()) {
/*
* If marking gets delayed at the same arena again, we must repeat
* marking of its things. For that we pop arena from the stack and
* clear its nextDelayedMarking before we begin the marking.
*/
ArenaHeader *aheader = unmarkedArenaStackTop;
unmarkedArenaStackTop = aheader->getMarkingDelay()->link;
JS_ASSERT(unmarkedArenaStackTop);
aheader->getMarkingDelay()->link = NULL;
#ifdef DEBUG
JS_ASSERT(markLaterArenas);
markLaterArenas--;
#endif
MarkDelayedChildren(this, aheader);
}
JS_ASSERT(!markLaterArenas);
}
} /* namespace js */
#ifdef DEBUG
static void
EmptyMarkCallback(JSTracer *trc, void *thing, uint32 kind)
{
}
#endif
static void
gc_root_traversal(JSTracer *trc, const RootEntry &entry)
{
#ifdef DEBUG
void *ptr;
if (entry.value.type == JS_GC_ROOT_GCTHING_PTR) {
ptr = *reinterpret_cast<void **>(entry.key);
} else {
Value *vp = reinterpret_cast<Value *>(entry.key);
ptr = vp->isGCThing() ? vp->toGCThing() : NULL;
}
if (ptr) {
if (!JSAtom::isStatic(ptr)) {
/* Use conservative machinery to find if ptr is a valid GC thing. */
JSTracer checker;
JS_TRACER_INIT(&checker, trc->context, EmptyMarkCallback);
ConservativeGCTest test = MarkIfGCThingWord(&checker, reinterpret_cast<jsuword>(ptr));
if (test != CGCT_VALID && entry.value.name) {
fprintf(stderr,
"JS API usage error: the address passed to JS_AddNamedRoot currently holds an\n"
"invalid gcthing. This is usually caused by a missing call to JS_RemoveRoot.\n"
"The root's name is \"%s\".\n",
entry.value.name);
}
JS_ASSERT(test == CGCT_VALID);
}
}
#endif
JS_SET_TRACING_NAME(trc, entry.value.name ? entry.value.name : "root");
if (entry.value.type == JS_GC_ROOT_GCTHING_PTR)
MarkGCThing(trc, *reinterpret_cast<void **>(entry.key));
else
MarkValueRaw(trc, *reinterpret_cast<Value *>(entry.key));
}
static void
gc_lock_traversal(const GCLocks::Entry &entry, JSTracer *trc)
{
JS_ASSERT(entry.value >= 1);
MarkGCThing(trc, entry.key, "locked object");
}
void
js_TraceStackFrame(JSTracer *trc, StackFrame *fp)
{
MarkObject(trc, fp->scopeChain(), "scope chain");
if (fp->isDummyFrame())
return;
if (fp->hasArgsObj())
MarkObject(trc, fp->argsObj(), "arguments");
js_TraceScript(trc, fp->script(), NULL);
fp->script()->compartment->active = true;
MarkValue(trc, fp->returnValue(), "rval");
}
void
AutoIdArray::trace(JSTracer *trc)
{
JS_ASSERT(tag == IDARRAY);
gc::MarkIdRange(trc, idArray->length, idArray->vector, "JSAutoIdArray.idArray");
}
void
AutoEnumStateRooter::trace(JSTracer *trc)
{
gc::MarkObject(trc, *obj, "js::AutoEnumStateRooter.obj");
}
inline void
AutoGCRooter::trace(JSTracer *trc)
{
switch (tag) {
case JSVAL:
MarkValue(trc, static_cast<AutoValueRooter *>(this)->val, "js::AutoValueRooter.val");
return;
case SHAPE:
MarkShape(trc, static_cast<AutoShapeRooter *>(this)->shape, "js::AutoShapeRooter.val");
return;
case PARSER:
static_cast<Parser *>(this)->trace(trc);
return;
case SCRIPT:
if (JSScript *script = static_cast<AutoScriptRooter *>(this)->script)
js_TraceScript(trc, script, NULL);
return;
case ENUMERATOR:
static_cast<AutoEnumStateRooter *>(this)->trace(trc);
return;
case IDARRAY: {
JSIdArray *ida = static_cast<AutoIdArray *>(this)->idArray;
MarkIdRange(trc, ida->length, ida->vector, "js::AutoIdArray.idArray");
return;
}
case DESCRIPTORS: {
PropDescArray &descriptors =
static_cast<AutoPropDescArrayRooter *>(this)->descriptors;
for (size_t i = 0, len = descriptors.length(); i < len; i++) {
PropDesc &desc = descriptors[i];
MarkValue(trc, desc.pd, "PropDesc::pd");
MarkValue(trc, desc.value, "PropDesc::value");
MarkValue(trc, desc.get, "PropDesc::get");
MarkValue(trc, desc.set, "PropDesc::set");
}
return;
}
case DESCRIPTOR : {
PropertyDescriptor &desc = *static_cast<AutoPropertyDescriptorRooter *>(this);
if (desc.obj)
MarkObject(trc, *desc.obj, "Descriptor::obj");
MarkValue(trc, desc.value, "Descriptor::value");
if ((desc.attrs & JSPROP_GETTER) && desc.getter)
MarkObject(trc, *CastAsObject(desc.getter), "Descriptor::get");
if (desc.attrs & JSPROP_SETTER && desc.setter)
MarkObject(trc, *CastAsObject(desc.setter), "Descriptor::set");
return;
}
case NAMESPACES: {
JSXMLArray &array = static_cast<AutoNamespaceArray *>(this)->array;
MarkObjectRange(trc, array.length, reinterpret_cast<JSObject **>(array.vector),
"JSXMLArray.vector");
array.cursors->trace(trc);
return;
}
case XML:
js_TraceXML(trc, static_cast<AutoXMLRooter *>(this)->xml);
return;
case OBJECT:
if (JSObject *obj = static_cast<AutoObjectRooter *>(this)->obj)
MarkObject(trc, *obj, "js::AutoObjectRooter.obj");
return;
case ID:
MarkId(trc, static_cast<AutoIdRooter *>(this)->id_, "js::AutoIdRooter.val");
return;
case VALVECTOR: {
AutoValueVector::VectorImpl &vector = static_cast<AutoValueVector *>(this)->vector;
MarkValueRange(trc, vector.length(), vector.begin(), "js::AutoValueVector.vector");
return;
}
case STRING:
if (JSString *str = static_cast<AutoStringRooter *>(this)->str)
MarkString(trc, str, "js::AutoStringRooter.str");
return;
case IDVECTOR: {
AutoIdVector::VectorImpl &vector = static_cast<AutoIdVector *>(this)->vector;
MarkIdRange(trc, vector.length(), vector.begin(), "js::AutoIdVector.vector");
return;
}
case SHAPEVECTOR: {
AutoShapeVector::VectorImpl &vector = static_cast<js::AutoShapeVector *>(this)->vector;
MarkShapeRange(trc, vector.length(), vector.begin(), "js::AutoShapeVector.vector");
return;
}
case BINDINGS: {
static_cast<js::AutoBindingsRooter *>(this)->bindings.trace(trc);
return;
}
}
JS_ASSERT(tag >= 0);
MarkValueRange(trc, tag, static_cast<AutoArrayRooter *>(this)->array, "js::AutoArrayRooter.array");
}
namespace js {
JS_FRIEND_API(void)
MarkContext(JSTracer *trc, JSContext *acx)
{
/* Stack frames and slots are traced by StackSpace::mark. */
/* Mark other roots-by-definition in acx. */
if (acx->globalObject && !acx->hasRunOption(JSOPTION_UNROOTED_GLOBAL))
MarkObject(trc, *acx->globalObject, "global object");
if (acx->isExceptionPending())
MarkValue(trc, acx->getPendingException(), "exception");
for (js::AutoGCRooter *gcr = acx->autoGCRooters; gcr; gcr = gcr->down)
gcr->trace(trc);
if (acx->sharpObjectMap.depth > 0)
js_TraceSharpMap(trc, &acx->sharpObjectMap);
MarkValue(trc, acx->iterValue, "iterValue");
}
#define PER_COMPARTMENT_OP(rt, op) \
if ((rt)->gcCurrentCompartment) { \
JSCompartment *c = (rt)->gcCurrentCompartment; \
op; \
} else { \
for (JSCompartment **i = rt->compartments.begin(); i != rt->compartments.end(); ++i) { \
JSCompartment *c = *i; \
op; \
} \
}
JS_REQUIRES_STACK void
MarkRuntime(JSTracer *trc)
{
JSRuntime *rt = trc->context->runtime;
if (rt->state != JSRTS_LANDING)
MarkConservativeStackRoots(trc);
for (RootRange r = rt->gcRootsHash.all(); !r.empty(); r.popFront())
gc_root_traversal(trc, r.front());
for (GCLocks::Range r = rt->gcLocksHash.all(); !r.empty(); r.popFront())
gc_lock_traversal(r.front(), trc);
js_TraceAtomState(trc);
JSContext *iter = NULL;
while (JSContext *acx = js_ContextIterator(rt, JS_TRUE, &iter))
MarkContext(trc, acx);
#ifdef JS_TRACER
PER_COMPARTMENT_OP(rt, if (c->hasTraceMonitor()) c->traceMonitor()->mark(trc));
#endif
for (ThreadDataIter i(rt); !i.empty(); i.popFront())
i.threadData()->mark(trc);
/*
* We mark extra roots at the end so that the hook can use additional
* colors to implement cycle collection.
*/
if (rt->gcExtraRootsTraceOp)
rt->gcExtraRootsTraceOp(trc, rt->gcExtraRootsData);
}
void
TriggerGC(JSRuntime *rt)
{
JS_ASSERT(!rt->gcRunning);
if (rt->gcIsNeeded)
return;
/*
* Trigger the GC when it is safe to call an operation callback on any
* thread.
*/
rt->gcIsNeeded = true;
rt->gcTriggerCompartment = NULL;
TriggerAllOperationCallbacks(rt);
}
void
TriggerCompartmentGC(JSCompartment *comp)
{
JSRuntime *rt = comp->rt;
JS_ASSERT(!rt->gcRunning);
GCREASON(COMPARTMENT);
if (rt->gcZeal()) {
TriggerGC(rt);
return;
}
if (rt->gcMode != JSGC_MODE_COMPARTMENT || comp == rt->atomsCompartment) {
/* We can't do a compartmental GC of the default compartment. */
TriggerGC(rt);
return;
}
if (rt->gcIsNeeded) {
/* If we need to GC more than one compartment, run a full GC. */
if (rt->gcTriggerCompartment != comp)
rt->gcTriggerCompartment = NULL;
return;
}
if (rt->gcBytes > 8192 && rt->gcBytes >= 3 * (rt->gcTriggerBytes / 2)) {
/* If we're using significantly more than our quota, do a full GC. */
TriggerGC(rt);
return;
}
/*
* Trigger the GC when it is safe to call an operation callback on any
* thread.
*/
rt->gcIsNeeded = true;
rt->gcTriggerCompartment = comp;
TriggerAllOperationCallbacks(comp->rt);
}
void
MaybeGC(JSContext *cx)
{
JSRuntime *rt = cx->runtime;
if (rt->gcZeal()) {
GCREASON(MAYBEGC);
js_GC(cx, NULL, GC_NORMAL);
return;
}
JSCompartment *comp = cx->compartment;
if (rt->gcIsNeeded) {
GCREASON(MAYBEGC);
js_GC(cx, (comp == rt->gcTriggerCompartment) ? comp : NULL, GC_NORMAL);
return;
}
if (comp->gcBytes > 8192 && comp->gcBytes >= 3 * (comp->gcTriggerBytes / 4)) {
GCREASON(MAYBEGC);
js_GC(cx, (rt->gcMode == JSGC_MODE_COMPARTMENT) ? comp : NULL, GC_NORMAL);
return;
}
/*
* On 32 bit setting gcNextFullGCTime below is not atomic and a race condition
* could trigger an GC. We tolerate this.
*/
int64 now = PRMJ_Now();
if (rt->gcNextFullGCTime && rt->gcNextFullGCTime <= now) {
if (rt->gcChunkAllocationSinceLastGC || rt->gcEmptyChunkListHead) {
GCREASON(MAYBEGC);
js_GC(cx, NULL, GC_SHRINK);
} else {
rt->gcNextFullGCTime = now + GC_IDLE_FULL_SPAN;
}
}
}
} /* namespace js */
void
js_DestroyScriptsToGC(JSContext *cx, JSCompartment *comp)
{
JSScript **listp, *script;
for (size_t i = 0; i != JS_ARRAY_LENGTH(comp->scriptsToGC); ++i) {
listp = &comp->scriptsToGC[i];
while ((script = *listp) != NULL) {
*listp = script->u.nextToGC;
script->u.nextToGC = NULL;
js_DestroyCachedScript(cx, script);
}
}
}
void
JSCompartment::finalizeObjectArenaLists(JSContext *cx)
{
arenas[FINALIZE_OBJECT0]. finalizeNow<JSObject>(cx);
arenas[FINALIZE_OBJECT2]. finalizeNow<JSObject_Slots2>(cx);
arenas[FINALIZE_OBJECT4]. finalizeNow<JSObject_Slots4>(cx);
arenas[FINALIZE_OBJECT8]. finalizeNow<JSObject_Slots8>(cx);
arenas[FINALIZE_OBJECT12].finalizeNow<JSObject_Slots12>(cx);
arenas[FINALIZE_OBJECT16].finalizeNow<JSObject_Slots16>(cx);
arenas[FINALIZE_FUNCTION].finalizeNow<JSFunction>(cx);
#ifdef JS_THREADSAFE
arenas[FINALIZE_OBJECT0_BACKGROUND]. finalizeLater<JSObject>(cx);
arenas[FINALIZE_OBJECT2_BACKGROUND]. finalizeLater<JSObject_Slots2>(cx);
arenas[FINALIZE_OBJECT4_BACKGROUND]. finalizeLater<JSObject_Slots4>(cx);
arenas[FINALIZE_OBJECT8_BACKGROUND]. finalizeLater<JSObject_Slots8>(cx);
arenas[FINALIZE_OBJECT12_BACKGROUND].finalizeLater<JSObject_Slots12>(cx);
arenas[FINALIZE_OBJECT16_BACKGROUND].finalizeLater<JSObject_Slots16>(cx);
#endif
#if JS_HAS_XML_SUPPORT
arenas[FINALIZE_XML].finalizeNow<JSXML>(cx);
#endif
}
void
JSCompartment::finalizeStringArenaLists(JSContext *cx)
{
#ifdef JS_THREADSAFE
arenas[FINALIZE_SHORT_STRING].finalizeLater<JSShortString>(cx);
arenas[FINALIZE_STRING].finalizeLater<JSString>(cx);
#else
arenas[FINALIZE_SHORT_STRING].finalizeNow<JSShortString>(cx);
arenas[FINALIZE_STRING].finalizeNow<JSString>(cx);
#endif
arenas[FINALIZE_EXTERNAL_STRING].finalizeNow<JSExternalString>(cx);
}
void
JSCompartment::finalizeShapeArenaLists(JSContext *cx)
{
arenas[FINALIZE_SHAPE].finalizeNow<Shape>(cx);
}
#ifdef JS_THREADSAFE
namespace js {
bool
GCHelperThread::init(JSRuntime *rt)
{
if (!(wakeup = PR_NewCondVar(rt->gcLock)))
return false;
if (!(sweepingDone = PR_NewCondVar(rt->gcLock)))
return false;
thread = PR_CreateThread(PR_USER_THREAD, threadMain, rt, PR_PRIORITY_NORMAL,
PR_LOCAL_THREAD, PR_JOINABLE_THREAD, 0);
return !!thread;
}
void
GCHelperThread::finish(JSRuntime *rt)
{
PRThread *join = NULL;
{
AutoLockGC lock(rt);
if (thread && !shutdown) {
shutdown = true;
PR_NotifyCondVar(wakeup);
join = thread;
}
}
if (join) {
/* PR_DestroyThread is not necessary. */
PR_JoinThread(join);
}
if (wakeup)
PR_DestroyCondVar(wakeup);
if (sweepingDone)
PR_DestroyCondVar(sweepingDone);
}
/* static */
void
GCHelperThread::threadMain(void *arg)
{
JSRuntime *rt = static_cast<JSRuntime *>(arg);
rt->gcHelperThread.threadLoop(rt);
}
void
GCHelperThread::threadLoop(JSRuntime *rt)
{
AutoLockGC lock(rt);
while (!shutdown) {
/*
* Sweeping can be true here on the first iteration if a GC and the
* corresponding startBackgroundSweep call happen before this thread
* has a chance to run.
*/
if (!sweeping)
PR_WaitCondVar(wakeup, PR_INTERVAL_NO_TIMEOUT);
if (sweeping) {
AutoUnlockGC unlock(rt);
doSweep();
}
sweeping = false;
PR_NotifyAllCondVar(sweepingDone);
}
}
void
GCHelperThread::startBackgroundSweep(JSRuntime *rt, JSGCInvocationKind gckind)
{
/* The caller takes the GC lock. */
JS_ASSERT(!sweeping);
lastGCKind = gckind;
sweeping = true;
PR_NotifyCondVar(wakeup);
}
void
GCHelperThread::waitBackgroundSweepEnd(JSRuntime *rt, bool gcUnlocked)
{
Maybe<AutoLockGC> lock;
if (gcUnlocked)
lock.construct(rt);
while (sweeping)
PR_WaitCondVar(sweepingDone, PR_INTERVAL_NO_TIMEOUT);
}
JS_FRIEND_API(void)
GCHelperThread::replenishAndFreeLater(void *ptr)
{
JS_ASSERT(freeCursor == freeCursorEnd);
do {
if (freeCursor && !freeVector.append(freeCursorEnd - FREE_ARRAY_LENGTH))
break;
freeCursor = (void **) OffTheBooks::malloc_(FREE_ARRAY_SIZE);
if (!freeCursor) {
freeCursorEnd = NULL;
break;
}
freeCursorEnd = freeCursor + FREE_ARRAY_LENGTH;
*freeCursor++ = ptr;
return;
} while (false);
Foreground::free_(ptr);
}
void
GCHelperThread::doSweep()
{
JS_ASSERT(cx);
for (ArenaHeader **i = finalizeVector.begin(); i != finalizeVector.end(); ++i)
ArenaList::backgroundFinalize(cx, *i);
finalizeVector.resize(0);
ExpireGCChunks(cx->runtime, lastGCKind);
cx = NULL;
if (freeCursor) {
void **array = freeCursorEnd - FREE_ARRAY_LENGTH;
freeElementsAndArray(array, freeCursor);
freeCursor = freeCursorEnd = NULL;
} else {
JS_ASSERT(!freeCursorEnd);
}
for (void ***iter = freeVector.begin(); iter != freeVector.end(); ++iter) {
void **array = *iter;
freeElementsAndArray(array, array + FREE_ARRAY_LENGTH);
}
freeVector.resize(0);
}
}
#endif /* JS_THREADSAFE */
static void
SweepCrossCompartmentWrappers(JSContext *cx)
{
JSRuntime *rt = cx->runtime;
/*
* Figure out how much JIT code should be released from inactive compartments.
* If multiple eighth-lives have passed, compound the release interval linearly;
* if enough time has passed, all inactive JIT code will be released.
*/
uint32 releaseInterval = 0;
int64 now = PRMJ_Now();
if (now >= rt->gcJitReleaseTime) {
releaseInterval = 8;
while (now >= rt->gcJitReleaseTime) {
if (--releaseInterval == 1)
rt->gcJitReleaseTime = now;
rt->gcJitReleaseTime += JIT_SCRIPT_EIGHTH_LIFETIME;
}
}
/*
* Sweep the compartment:
* (1) Remove dead wrappers from the compartment map.
* (2) Finalize any unused empty shapes.
* (3) Sweep the trace JIT of unused code.
* (4) Sweep the method JIT ICs and release infrequently used JIT code.
*/
for (JSCompartment **c = rt->compartments.begin(); c != rt->compartments.end(); ++c)
(*c)->sweep(cx, releaseInterval);
}
static void
SweepCompartments(JSContext *cx, JSGCInvocationKind gckind)
{
JSRuntime *rt = cx->runtime;
JSCompartmentCallback callback = rt->compartmentCallback;
/* Skip the atomsCompartment. */
JSCompartment **read = rt->compartments.begin() + 1;
JSCompartment **end = rt->compartments.end();
JSCompartment **write = read;
JS_ASSERT(rt->compartments.length() >= 1);
JS_ASSERT(*rt->compartments.begin() == rt->atomsCompartment);
while (read < end) {
JSCompartment *compartment = *read++;
if (!compartment->hold &&
(compartment->arenaListsAreEmpty() || gckind == GC_LAST_CONTEXT))
{
compartment->freeLists.checkEmpty();
if (callback)
JS_ALWAYS_TRUE(callback(cx, compartment, JSCOMPARTMENT_DESTROY));
if (compartment->principals)
JSPRINCIPALS_DROP(cx, compartment->principals);
cx->delete_(compartment);
continue;
}
*write++ = compartment;
}
rt->compartments.resize(write - rt->compartments.begin());
}
/*
* Perform mark-and-sweep GC.
*
* In a JS_THREADSAFE build, the calling thread must be rt->gcThread and each
* other thread must be either outside all requests or blocked waiting for GC
* to finish. Note that the caller does not hold rt->gcLock.
* If comp is set, we perform a single-compartment GC.
*/
static void
MarkAndSweep(JSContext *cx, JSCompartment *comp, JSGCInvocationKind gckind GCTIMER_PARAM)
{
JS_ASSERT_IF(comp, gckind != GC_LAST_CONTEXT);
JS_ASSERT_IF(comp, comp != comp->rt->atomsCompartment);
JS_ASSERT_IF(comp, comp->rt->gcMode == JSGC_MODE_COMPARTMENT);
JSRuntime *rt = cx->runtime;
rt->gcNumber++;
/* Clear gcIsNeeded now, when we are about to start a normal GC cycle. */
rt->gcIsNeeded = false;
rt->gcTriggerCompartment = NULL;
/* Reset malloc counter. */
rt->resetGCMallocBytes();
/*
* Reset the property cache's type id generator so we can compress ids.
* Same for the protoHazardShape proxy-shape standing in for all object
* prototypes having readonly or setter properties.
*/
if (rt->shapeGen & SHAPE_OVERFLOW_BIT || (rt->gcZeal() && !rt->gcCurrentCompartment)) {
rt->gcRegenShapes = true;
rt->shapeGen = 0;
rt->protoHazardShape = 0;
}
PER_COMPARTMENT_OP(rt, c->purge(cx));
js_PurgeThreads(cx);
{
JSContext *iter = NULL;
while (JSContext *acx = js_ContextIterator(rt, JS_TRUE, &iter))
acx->purge();
}
JS_ASSERT_IF(comp, !rt->gcRegenShapes);
/*
* Mark phase.
*/
GCTIMESTAMP(startMark);
GCMarker gcmarker(cx);
JS_ASSERT(IS_GC_MARKING_TRACER(&gcmarker));
JS_ASSERT(gcmarker.getMarkColor() == BLACK);
rt->gcMarkingTracer = &gcmarker;
for (GCChunkSet::Range r(rt->gcChunkSet.all()); !r.empty(); r.popFront())
r.front()->bitmap.clear();
if (comp) {
for (JSCompartment **c = rt->compartments.begin(); c != rt->compartments.end(); ++c)
(*c)->markCrossCompartmentWrappers(&gcmarker);
Debugger::markCrossCompartmentDebuggerObjectReferents(&gcmarker);
}
MarkRuntime(&gcmarker);
gcmarker.drainMarkStack();
/*
* Mark weak roots.
*/
while (WatchpointMap::markAllIteratively(&gcmarker) ||
WeakMapBase::markAllIteratively(&gcmarker) ||
Debugger::markAllIteratively(&gcmarker, gckind))
{
gcmarker.drainMarkStack();
}
rt->gcMarkingTracer = NULL;
if (rt->gcCallback)
(void) rt->gcCallback(cx, JSGC_MARK_END);
#ifdef DEBUG
/* Make sure that we didn't mark an object in another compartment */
if (comp) {
for (JSCompartment **c = rt->compartments.begin(); c != rt->compartments.end(); ++c)
JS_ASSERT_IF(*c != comp && *c != rt->atomsCompartment, checkArenaListAllUnmarked(*c));
}
#endif
/*
* Sweep phase.
*
* Finalize as we sweep, outside of rt->gcLock but with rt->gcRunning set
* so that any attempt to allocate a GC-thing from a finalizer will fail,
* rather than nest badly and leave the unmarked newborn to be swept.
*
* We first sweep atom state so we can use IsAboutToBeFinalized on
* JSString held in a hashtable to check if the hashtable entry can be
* freed. Note that even after the entry is freed, JSObject finalizers can
* continue to access the corresponding JSString* assuming that they are
* unique. This works since the atomization API must not be called during
* the GC.
*/
GCTIMESTAMP(startSweep);
/* Finalize unreachable (key,value) pairs in all weak maps. */
WeakMapBase::sweepAll(&gcmarker);
js_SweepAtomState(cx);
/* Collect watch points associated with unreachable objects. */
WatchpointMap::sweepAll(cx);
/*
* We finalize objects before other GC things to ensure that object's finalizer
* can access them even if they will be freed. Sweep the runtime's property trees
* after finalizing objects, in case any had watchpoints referencing tree nodes.
* Do this before sweeping compartments, so that we sweep all shapes in
* unreachable compartments.
*/
if (comp) {
Probes::GCStartSweepPhase(comp);
comp->sweep(cx, 0);
comp->finalizeObjectArenaLists(cx);
GCTIMESTAMP(sweepObjectEnd);
comp->finalizeStringArenaLists(cx);
GCTIMESTAMP(sweepStringEnd);
comp->finalizeShapeArenaLists(cx);
GCTIMESTAMP(sweepShapeEnd);
Probes::GCEndSweepPhase(comp);
} else {
/*
* Some sweeping is not compartment-specific. Start a NULL-compartment
* phase to demarcate all of that. (The compartment sweeps will nest
* within.)
*/
Probes::GCStartSweepPhase(NULL);
Debugger::sweepAll(cx);
SweepCrossCompartmentWrappers(cx);
for (JSCompartment **c = rt->compartments.begin(); c != rt->compartments.end(); c++) {
Probes::GCStartSweepPhase(*c);
(*c)->finalizeObjectArenaLists(cx);
}
GCTIMESTAMP(sweepObjectEnd);
for (JSCompartment **c = rt->compartments.begin(); c != rt->compartments.end(); c++)
(*c)->finalizeStringArenaLists(cx);
GCTIMESTAMP(sweepStringEnd);
for (JSCompartment **c = rt->compartments.begin(); c != rt->compartments.end(); c++) {
(*c)->finalizeShapeArenaLists(cx);
Probes::GCEndSweepPhase(*c);
}
GCTIMESTAMP(sweepShapeEnd);
}
#ifdef DEBUG
PropertyTree::dumpShapes(cx);
#endif
/*
* Sweep script filenames after sweeping functions in the generic loop
* above. In this way when a scripted function's finalizer destroys the
* script and calls rt->destroyScriptHook, the hook can still access the
* script's filename. See bug 323267.
*/
PER_COMPARTMENT_OP(rt, js_SweepScriptFilenames(c));
if (!comp) {
SweepCompartments(cx, gckind);
/* non-compartmental sweep pieces */
Probes::GCEndSweepPhase(NULL);
}
#ifndef JS_THREADSAFE
/*
* Destroy arenas after we finished the sweeping so finalizers can safely
* use IsAboutToBeFinalized().
* This is done on the GCHelperThread if JS_THREADSAFE is defined.
*/
ExpireGCChunks(rt, gckind);
#endif
GCTIMESTAMP(sweepDestroyEnd);
if (rt->gcCallback)
(void) rt->gcCallback(cx, JSGC_FINALIZE_END);
#ifdef DEBUG_srcnotesize
{ extern void DumpSrcNoteSizeHist();
DumpSrcNoteSizeHist();
printf("GC HEAP SIZE %lu\n", (unsigned long)rt->gcBytes);
}
#endif
}
#ifdef JS_THREADSAFE
/*
* If the GC is running and we're called on another thread, wait for this GC
* activation to finish. We can safely wait here without fear of deadlock (in
* the case where we are called within a request on another thread's context)
* because the GC doesn't set rt->gcRunning until after it has waited for all
* active requests to end.
*
* We call here js_CurrentThreadId() after checking for rt->gcState to avoid
* an expensive call when the GC is not running.
*/
void
js_WaitForGC(JSRuntime *rt)
{
if (rt->gcRunning && rt->gcThread->id != js_CurrentThreadId()) {
do {
JS_AWAIT_GC_DONE(rt);
} while (rt->gcRunning);
}
}
/*
* GC is running on another thread. Temporarily suspend all requests running
* on the current thread and wait until the GC is done.
*/
static void
LetOtherGCFinish(JSContext *cx)
{
JSRuntime *rt = cx->runtime;
JS_ASSERT(rt->gcThread);
JS_ASSERT(cx->thread() != rt->gcThread);
size_t requestDebit = cx->thread()->data.requestDepth ? 1 : 0;
JS_ASSERT(requestDebit <= rt->requestCount);
#ifdef JS_TRACER
JS_ASSERT_IF(requestDebit == 0, !JS_ON_TRACE(cx));
#endif
if (requestDebit != 0) {
#ifdef JS_TRACER
if (JS_ON_TRACE(cx)) {
/*
* Leave trace before we decrease rt->requestCount and notify the
* GC. Otherwise the GC may start immediately after we unlock while
* this thread is still on trace.
*/
AutoUnlockGC unlock(rt);
LeaveTrace(cx);
}
#endif
rt->requestCount -= requestDebit;
if (rt->requestCount == 0)
JS_NOTIFY_REQUEST_DONE(rt);
/*
* Update the native stack before we wait so the GC thread see the
* correct stack bounds.
*/
RecordNativeStackTopForGC(cx);
}
/*
* Check that we did not release the GC lock above and let the GC to
* finish before we wait.
*/
JS_ASSERT(rt->gcThread);
/*
* Wait for GC to finish on the other thread, even if requestDebit is 0
* and even if GC has not started yet because the gcThread is waiting in
* AutoGCSession. This ensures that js_GC never returns without a full GC
* cycle happening.
*/
do {
JS_AWAIT_GC_DONE(rt);
} while (rt->gcThread);
rt->requestCount += requestDebit;
}
#endif
class AutoGCSession {
public:
explicit AutoGCSession(JSContext *cx);
~AutoGCSession();
private:
JSContext *context;
/* Disable copy constructor or assignments */
AutoGCSession(const AutoGCSession&);
void operator=(const AutoGCSession&);
};
/*
* Start a new GC session. Together with LetOtherGCFinish this function
* contains the rendezvous algorithm by which we stop the world for GC.
*
* This thread becomes the GC thread. Wait for all other threads to quiesce.
* Then set rt->gcRunning and return.
*/
AutoGCSession::AutoGCSession(JSContext *cx)
: context(cx)
{
JSRuntime *rt = cx->runtime;
#ifdef JS_THREADSAFE
if (rt->gcThread && rt->gcThread != cx->thread())
LetOtherGCFinish(cx);
#endif
JS_ASSERT(!rt->gcRunning);
#ifdef JS_THREADSAFE
/* No other thread is in GC, so indicate that we're now in GC. */
JS_ASSERT(!rt->gcThread);
rt->gcThread = cx->thread();
/*
* Notify operation callbacks on other threads, which will give them a
* chance to yield their requests. Threads without requests perform their
* callback at some later point, which then will be unnecessary, but
* harmless.
*/
for (JSThread::Map::Range r = rt->threads.all(); !r.empty(); r.popFront()) {
JSThread *thread = r.front().value;
if (thread != cx->thread())
thread->data.triggerOperationCallback(rt);
}
/*
* Discount the request on the current thread from contributing to
* rt->requestCount before we wait for all other requests to finish.
* JS_NOTIFY_REQUEST_DONE, which will wake us up, is only called on
* rt->requestCount transitions to 0.
*/
size_t requestDebit = cx->thread()->data.requestDepth ? 1 : 0;
JS_ASSERT(requestDebit <= rt->requestCount);
if (requestDebit != rt->requestCount) {
rt->requestCount -= requestDebit;
do {
JS_AWAIT_REQUEST_DONE(rt);
} while (rt->requestCount > 0);
rt->requestCount += requestDebit;
}
#endif /* JS_THREADSAFE */
/*
* Set rt->gcRunning here within the GC lock, and after waiting for any
* active requests to end. This way js_WaitForGC called outside a request
* would not block on the GC that is waiting for other requests to finish
* with rt->gcThread set while JS_BeginRequest would do such wait.
*/
rt->gcRunning = true;
}
/* End the current GC session and allow other threads to proceed. */
AutoGCSession::~AutoGCSession()
{
JSRuntime *rt = context->runtime;
rt->gcRunning = false;
#ifdef JS_THREADSAFE
JS_ASSERT(rt->gcThread == context->thread());
rt->gcThread = NULL;
JS_NOTIFY_GC_DONE(rt);
#endif
}
/*
* GC, repeatedly if necessary, until we think we have not created any new
* garbage and no other threads are demanding more GC. We disable inlining
* to ensure that the bottom of the stack with possible GC roots recorded in
* js_GC excludes any pointers we use during the marking implementation.
*/
static JS_NEVER_INLINE void
GCCycle(JSContext *cx, JSCompartment *comp, JSGCInvocationKind gckind GCTIMER_PARAM)
{
JSRuntime *rt = cx->runtime;
/*
* Recursive GC is no-op and a call from another thread waits the started
* GC cycle to finish.
*/
if (rt->gcMarkAndSweep) {
#ifdef JS_THREADSAFE
JS_ASSERT(rt->gcThread);
if (rt->gcThread != cx->thread()) {
/* We do not return until another GC finishes. */
LetOtherGCFinish(cx);
}
#endif
return;
}
AutoGCSession gcsession(cx);
/*
* Don't GC if any thread is reporting an OOM. We check the flag after we
* have set up the GC session and know that the thread that reported OOM
* is either the current thread or waits for the GC to complete on this
* thread.
*/
if (rt->inOOMReport) {
JS_ASSERT(gckind != GC_LAST_CONTEXT);
return;
}
/*
* We should not be depending on cx->compartment in the GC, so set it to
* NULL to look for violations.
*/
SwitchToCompartment sc(cx, (JSCompartment *)NULL);
JS_ASSERT(!rt->gcCurrentCompartment);
rt->gcCurrentCompartment = comp;
rt->gcMarkAndSweep = true;
{
AutoUnlockGC unlock(rt);
#ifdef JS_THREADSAFE
/*
* As we about to purge caches and clear the mark bits we must wait
* for any background finalization to finish.
*/
JS_ASSERT(!cx->gcBackgroundFree);
rt->gcHelperThread.waitBackgroundSweepEnd(rt);
if (gckind != GC_LAST_CONTEXT && rt->state != JSRTS_LANDING) {
cx->gcBackgroundFree = &rt->gcHelperThread;
cx->gcBackgroundFree->setContext(cx);
}
#endif
MarkAndSweep(cx, comp, gckind GCTIMER_ARG);
}
#ifdef JS_THREADSAFE
if (gckind != GC_LAST_CONTEXT && rt->state != JSRTS_LANDING) {
JS_ASSERT(cx->gcBackgroundFree == &rt->gcHelperThread);
cx->gcBackgroundFree = NULL;
rt->gcHelperThread.startBackgroundSweep(rt, gckind);
} else {
JS_ASSERT(!cx->gcBackgroundFree);
}
#endif
rt->gcMarkAndSweep = false;
rt->gcRegenShapes = false;
rt->setGCLastBytes(rt->gcBytes, gckind);
rt->gcCurrentCompartment = NULL;
rt->gcWeakMapList = NULL;
for (JSCompartment **c = rt->compartments.begin(); c != rt->compartments.end(); ++c)
(*c)->setGCLastBytes((*c)->gcBytes, gckind);
}
struct GCCrashData
{
int isRegen;
int isCompartment;
};
void
js_GC(JSContext *cx, JSCompartment *comp, JSGCInvocationKind gckind)
{
JSRuntime *rt = cx->runtime;
/*
* Don't collect garbage if the runtime isn't up, and cx is not the last
* context in the runtime. The last context must force a GC, and nothing
* should suppress that final collection or there may be shutdown leaks,
* or runtime bloat until the next context is created.
*/
if (rt->state != JSRTS_UP && gckind != GC_LAST_CONTEXT)
return;
if (JS_ON_TRACE(cx)) {
JS_ASSERT(gckind != GC_LAST_CONTEXT);
return;
}
RecordNativeStackTopForGC(cx);
GCCrashData crashData;
crashData.isRegen = rt->shapeGen & SHAPE_OVERFLOW_BIT;
crashData.isCompartment = !!comp;
crash::SaveCrashData(crash::JS_CRASH_TAG_GC, &crashData, sizeof(crashData));
GCTIMER_BEGIN(rt, comp);
struct AutoGCProbe {
JSCompartment *comp;
AutoGCProbe(JSCompartment *comp) : comp(comp) {
Probes::GCStart(comp);
}
~AutoGCProbe() {
Probes::GCEnd(comp); /* background thread may still be sweeping */
}
} autoGCProbe(comp);
do {
/*
* Let the API user decide to defer a GC if it wants to (unless this
* is the last context). Invoke the callback regardless. Sample the
* callback in case we are freely racing with a JS_SetGCCallback{,RT}
* on another thread.
*/
if (JSGCCallback callback = rt->gcCallback) {
if (!callback(cx, JSGC_BEGIN) && gckind != GC_LAST_CONTEXT)
return;
}
{
#ifdef JS_THREADSAFE
rt->gcHelperThread.waitBackgroundSweepEnd(rt);
#endif
/* Lock out other GC allocator and collector invocations. */
AutoLockGC lock(rt);
rt->gcPoke = false;
GCCycle(cx, comp, gckind GCTIMER_ARG);
}
/* We re-sample the callback again as the finalizers can change it. */
if (JSGCCallback callback = rt->gcCallback)
(void) callback(cx, JSGC_END);
/*
* On shutdown, iterate until finalizers or the JSGC_END callback
* stop creating garbage.
*/
} while (gckind == GC_LAST_CONTEXT && rt->gcPoke);
rt->gcNextFullGCTime = PRMJ_Now() + GC_IDLE_FULL_SPAN;
rt->gcChunkAllocationSinceLastGC = false;
GCTIMER_END(gckind == GC_LAST_CONTEXT);
crash::SnapshotGCStack();
}
namespace js {
class AutoCopyFreeListToArenas {
JSRuntime *rt;
public:
AutoCopyFreeListToArenas(JSRuntime *rt)
: rt(rt) {
for (JSCompartment **c = rt->compartments.begin(); c != rt->compartments.end(); ++c)
(*c)->freeLists.copyToArenas();
}
~AutoCopyFreeListToArenas() {
for (JSCompartment **c = rt->compartments.begin(); c != rt->compartments.end(); ++c)
(*c)->freeLists.clearInArenas();
}
};
void
TraceRuntime(JSTracer *trc)
{
JS_ASSERT(!IS_GC_MARKING_TRACER(trc));
LeaveTrace(trc->context);
#ifdef JS_THREADSAFE
{
JSContext *cx = trc->context;
JSRuntime *rt = cx->runtime;
if (rt->gcThread != cx->thread()) {
AutoLockGC lock(rt);
AutoGCSession gcsession(cx);
rt->gcHelperThread.waitBackgroundSweepEnd(rt, false);
AutoUnlockGC unlock(rt);
AutoCopyFreeListToArenas copy(rt);
RecordNativeStackTopForGC(trc->context);
MarkRuntime(trc);
return;
}
}
#else
AutoCopyFreeListToArenas copy(trc->context->runtime);
RecordNativeStackTopForGC(trc->context);
#endif
/*
* Calls from inside a normal GC or a recursive calls are OK and do not
* require session setup.
*/
MarkRuntime(trc);
}
void
IterateCompartmentsArenasCells(JSContext *cx, void *data,
IterateCompartmentCallback compartmentCallback,
IterateArenaCallback arenaCallback,
IterateCellCallback cellCallback)
{
CHECK_REQUEST(cx);
LeaveTrace(cx);
JSRuntime *rt = cx->runtime;
JS_ASSERT(!rt->gcRunning);
AutoLockGC lock(rt);
AutoGCSession gcsession(cx);
#ifdef JS_THREADSAFE
rt->gcHelperThread.waitBackgroundSweepEnd(rt, false);
#endif
AutoUnlockGC unlock(rt);
AutoCopyFreeListToArenas copy(rt);
for (JSCompartment **c = rt->compartments.begin(); c != rt->compartments.end(); ++c) {
JSCompartment *compartment = *c;
(*compartmentCallback)(cx, data, compartment);
for (unsigned thingKind = 0; thingKind < FINALIZE_LIMIT; thingKind++) {
size_t traceKind = GetFinalizableTraceKind(thingKind);
size_t thingSize = GCThingSizeMap[thingKind];
ArenaHeader *aheader = compartment->arenas[thingKind].getHead();
for (; aheader; aheader = aheader->next) {
Arena *arena = aheader->getArena();
(*arenaCallback)(cx, data, arena, traceKind, thingSize);
FreeSpan firstSpan(aheader->getFirstFreeSpan());
const FreeSpan *span = &firstSpan;
for (uintptr_t thing = arena->thingsStart(thingSize); ; thing += thingSize) {
JS_ASSERT(thing <= arena->thingsEnd());
if (thing == span->first) {
if (!span->hasNext())
break;
thing = span->last;
span = span->nextSpan();
} else {
void *t = reinterpret_cast<void *>(thing);
(*cellCallback)(cx, data, t, traceKind, thingSize);
}
}
}
}
}
}
namespace gc {
JSCompartment *
NewCompartment(JSContext *cx, JSPrincipals *principals)
{
JSRuntime *rt = cx->runtime;
JSCompartment *compartment = cx->new_<JSCompartment>(rt);
if (compartment && compartment->init()) {
// Any compartment with the trusted principals -- and there can be
// multiple -- is a system compartment.
compartment->isSystemCompartment = principals && rt->trustedPrincipals() == principals;
if (principals) {
compartment->principals = principals;
JSPRINCIPALS_HOLD(cx, principals);
}
compartment->setGCLastBytes(8192, GC_NORMAL);
/*
* Before reporting the OOM condition, |lock| needs to be cleaned up,
* hence the scoping.
*/
{
AutoLockGC lock(rt);
if (rt->compartments.append(compartment))
return compartment;
}
js_ReportOutOfMemory(cx);
}
Foreground::delete_(compartment);
return NULL;
}
void
RunDebugGC(JSContext *cx)
{
#ifdef JS_GC_ZEAL
if (IsGCAllowed(cx)) {
JSRuntime *rt = cx->runtime;
/*
* If rt->gcDebugCompartmentGC is true, only GC the current
* compartment. But don't GC the atoms compartment.
*/
rt->gcTriggerCompartment = rt->gcDebugCompartmentGC ? cx->compartment : NULL;
if (rt->gcTriggerCompartment == rt->atomsCompartment)
rt->gcTriggerCompartment = NULL;
RunLastDitchGC(cx);
}
#endif
}
} /* namespace gc */
} /* namespace js */
#if JS_HAS_XML_SUPPORT
extern size_t sE4XObjectsCreated;
JSXML *
js_NewGCXML(JSContext *cx)
{
if (!cx->runningWithTrustedPrincipals())
++sE4XObjectsCreated;
return NewGCThing<JSXML>(cx, js::gc::FINALIZE_XML, sizeof(JSXML));
}
#endif
