Source code
Revision control
Copy as Markdown
Other Tools
/* -*- Mode: C++; tab-width: 8; indent-tabs-mode: nil; c-basic-offset: 2 -*-
* vim: set ts=8 sts=2 et sw=2 tw=80:
* This Source Code Form is subject to the terms of the Mozilla Public
* License, v. 2.0. If a copy of the MPL was not distributed with this
#include "gc/AtomMarking-inl.h"
#include <type_traits>
#include "gc/PublicIterators.h"
#include "gc/GC-inl.h"
#include "gc/Heap-inl.h"
#include "gc/PrivateIterators-inl.h"
namespace js {
namespace gc {
// [SMDOC] GC Atom Marking
//
// Things in the atoms zone (which includes atomized strings and other things,
// all of which we will refer to as 'atoms' here) may be pointed to freely by
// things in other zones. To avoid the need to perform garbage collections of
// the entire runtime to collect atoms, we compute a separate atom mark bitmap
// for each zone that is always an overapproximation of the atoms that zone is
// using. When an atom is not in the mark bitmap for any zone, it can be
// destroyed.
//
// To minimize interference with the rest of the GC, atom marking and sweeping
// is done by manipulating the mark bitmaps in the chunks used for the atoms.
// When the atoms zone is being collected, the mark bitmaps for the chunk(s)
// used by the atoms are updated normally during marking. After marking
// finishes, the chunk mark bitmaps are translated to a more efficient atom mark
// bitmap (see below) that is stored on the zones which the GC collected
// (computeBitmapFromChunkMarkBits). Before sweeping begins, the chunk mark
// bitmaps are updated with any atoms that might be referenced by zones which
// weren't collected (markAtomsUsedByUncollectedZones). The GC sweeping will
// then release all atoms which are not marked by any zone.
//
// The representation of atom mark bitmaps is as follows:
//
// Each arena in the atoms zone has an atomBitmapStart() value indicating the
// word index into the bitmap of the first thing in the arena. Each arena uses
// ArenaBitmapWords of data to store its bitmap, which uses the same
// representation as chunk mark bitmaps: one bit is allocated per Cell, with
// bits for space between things being unused when things are larger than a
// single Cell.
void AtomMarkingRuntime::registerArena(Arena* arena, const AutoLockGC& lock) {
MOZ_ASSERT(arena->getThingSize() != 0);
MOZ_ASSERT(arena->getThingSize() % CellAlignBytes == 0);
MOZ_ASSERT(arena->zone()->isAtomsZone());
// We need to find a range of bits from the atoms bitmap for this arena.
// Look for a free range of bits compatible with this arena.
if (freeArenaIndexes.ref().length()) {
arena->atomBitmapStart() = freeArenaIndexes.ref().popCopy();
return;
}
// Allocate a range of bits from the end for this arena.
arena->atomBitmapStart() = allocatedWords;
allocatedWords += ArenaBitmapWords;
}
void AtomMarkingRuntime::unregisterArena(Arena* arena, const AutoLockGC& lock) {
MOZ_ASSERT(arena->zone()->isAtomsZone());
// Leak these atom bits if we run out of memory.
(void)freeArenaIndexes.ref().emplaceBack(arena->atomBitmapStart());
}
void AtomMarkingRuntime::refineZoneBitmapsForCollectedZones(
GCRuntime* gc, size_t collectedZones) {
// If there is more than one zone to update, copy the chunk mark bits into a
// bitmap and AND that into the atom marking bitmap for each zone.
DenseBitmap marked;
if (collectedZones > 1 && computeBitmapFromChunkMarkBits(gc, marked)) {
for (GCZonesIter zone(gc); !zone.done(); zone.next()) {
refineZoneBitmapForCollectedZone(zone, marked);
}
return;
}
// If there's only one zone (or on OOM), AND the mark bits for each arena into
// the zones' atom marking bitmaps directly.
for (GCZonesIter zone(gc); !zone.done(); zone.next()) {
if (zone->isAtomsZone()) {
continue;
}
for (auto thingKind : AllAllocKinds()) {
for (ArenaIterInGC aiter(gc->atomsZone(), thingKind); !aiter.done();
aiter.next()) {
Arena* arena = aiter.get();
MarkBitmapWord* chunkWords = arena->chunk()->markBits.arenaBits(arena);
zone->markedAtoms().bitwiseAndRangeWith(arena->atomBitmapStart(),
ArenaBitmapWords, chunkWords);
}
}
}
}
bool AtomMarkingRuntime::computeBitmapFromChunkMarkBits(GCRuntime* gc,
DenseBitmap& bitmap) {
MOZ_ASSERT(CurrentThreadIsPerformingGC());
if (!bitmap.ensureSpace(allocatedWords)) {
return false;
}
Zone* atomsZone = gc->atomsZone();
for (auto thingKind : AllAllocKinds()) {
for (ArenaIterInGC aiter(atomsZone, thingKind); !aiter.done();
aiter.next()) {
Arena* arena = aiter.get();
MarkBitmapWord* chunkWords = arena->chunk()->markBits.arenaBits(arena);
bitmap.copyBitsFrom(arena->atomBitmapStart(), ArenaBitmapWords,
chunkWords);
}
}
return true;
}
void AtomMarkingRuntime::refineZoneBitmapForCollectedZone(
Zone* zone, const DenseBitmap& bitmap) {
MOZ_ASSERT(zone->isCollectingFromAnyThread());
if (zone->isAtomsZone()) {
return;
}
// Take the bitwise and between the two mark bitmaps to get the best new
// overapproximation we can. |bitmap| might include bits that are not in
// the zone's mark bitmap, if additional zones were collected by the GC.
zone->markedAtoms().bitwiseAndWith(bitmap);
}
// Set any bits in the chunk mark bitmaps for atoms which are marked in bitmap.
template <typename Bitmap>
static void BitwiseOrIntoChunkMarkBits(Zone* atomsZone, Bitmap& bitmap) {
// Make sure that by copying the mark bits for one arena in word sizes we
// do not affect the mark bits for other arenas.
static_assert(ArenaBitmapBits == ArenaBitmapWords * JS_BITS_PER_WORD,
"ArenaBitmapWords must evenly divide ArenaBitmapBits");
for (auto thingKind : AllAllocKinds()) {
for (ArenaIterInGC aiter(atomsZone, thingKind); !aiter.done();
aiter.next()) {
Arena* arena = aiter.get();
MarkBitmapWord* chunkWords = arena->chunk()->markBits.arenaBits(arena);
bitmap.bitwiseOrRangeInto(arena->atomBitmapStart(), ArenaBitmapWords,
chunkWords);
}
}
}
void AtomMarkingRuntime::markAtomsUsedByUncollectedZones(
GCRuntime* gc, size_t uncollectedZones) {
MOZ_ASSERT(CurrentThreadIsPerformingGC());
// If there are no uncollected non-atom zones then there's no work to do.
if (uncollectedZones == 0) {
return;
}
// If there is more than one zone then try to compute a simple union of the
// zone atom bitmaps before updating the chunk mark bitmaps. If there is only
// one zone or this allocation fails then update the chunk mark bitmaps
// separately for each zone.
DenseBitmap markedUnion;
if (uncollectedZones == 1 || !markedUnion.ensureSpace(allocatedWords)) {
for (ZonesIter zone(gc, SkipAtoms); !zone.done(); zone.next()) {
if (!zone->isCollecting()) {
BitwiseOrIntoChunkMarkBits(gc->atomsZone(), zone->markedAtoms());
}
}
return;
}
for (ZonesIter zone(gc, SkipAtoms); !zone.done(); zone.next()) {
// We only need to update the chunk mark bits for zones which were
// not collected in the current GC. Atoms which are referenced by
// collected zones have already been marked.
if (!zone->isCollecting()) {
zone->markedAtoms().bitwiseOrInto(markedUnion);
}
}
BitwiseOrIntoChunkMarkBits(gc->atomsZone(), markedUnion);
}
template <typename T>
void AtomMarkingRuntime::markAtom(JSContext* cx, T* thing) {
return inlinedMarkAtom(cx, thing);
}
template void AtomMarkingRuntime::markAtom(JSContext* cx, JSAtom* thing);
template void AtomMarkingRuntime::markAtom(JSContext* cx, JS::Symbol* thing);
void AtomMarkingRuntime::markId(JSContext* cx, jsid id) {
if (id.isAtom()) {
markAtom(cx, id.toAtom());
return;
}
if (id.isSymbol()) {
markAtom(cx, id.toSymbol());
return;
}
MOZ_ASSERT(!id.isGCThing());
}
void AtomMarkingRuntime::markAtomValue(JSContext* cx, const Value& value) {
if (value.isString()) {
if (value.toString()->isAtom()) {
markAtom(cx, &value.toString()->asAtom());
}
return;
}
if (value.isSymbol()) {
markAtom(cx, value.toSymbol());
return;
}
MOZ_ASSERT_IF(value.isGCThing(), value.isObject() ||
value.isPrivateGCThing() ||
value.isBigInt());
}
#ifdef DEBUG
template <typename T>
bool AtomMarkingRuntime::atomIsMarked(Zone* zone, T* thing) {
static_assert(std::is_same_v<T, JSAtom> || std::is_same_v<T, JS::Symbol>,
"Should only be called with JSAtom* or JS::Symbol* argument");
MOZ_ASSERT(thing);
MOZ_ASSERT(!IsInsideNursery(thing));
MOZ_ASSERT(thing->zoneFromAnyThread()->isAtomsZone());
if (!zone->runtimeFromAnyThread()->permanentAtomsPopulated()) {
return true;
}
if (thing->isPermanentAndMayBeShared()) {
return true;
}
if constexpr (std::is_same_v<T, JSAtom>) {
if (thing->isPinned()) {
return true;
}
}
size_t bit = GetAtomBit(&thing->asTenured());
return zone->markedAtoms().readonlyThreadsafeGetBit(bit);
}
template bool AtomMarkingRuntime::atomIsMarked(Zone* zone, JSAtom* thing);
template bool AtomMarkingRuntime::atomIsMarked(Zone* zone, JS::Symbol* thing);
template <>
bool AtomMarkingRuntime::atomIsMarked(Zone* zone, TenuredCell* thing) {
if (!thing) {
return true;
}
if (thing->is<JSString>()) {
JSString* str = thing->as<JSString>();
if (!str->isAtom()) {
return true;
}
return atomIsMarked(zone, &str->asAtom());
}
if (thing->is<JS::Symbol>()) {
return atomIsMarked(zone, thing->as<JS::Symbol>());
}
return true;
}
bool AtomMarkingRuntime::idIsMarked(Zone* zone, jsid id) {
if (id.isAtom()) {
return atomIsMarked(zone, id.toAtom());
}
if (id.isSymbol()) {
return atomIsMarked(zone, id.toSymbol());
}
MOZ_ASSERT(!id.isGCThing());
return true;
}
bool AtomMarkingRuntime::valueIsMarked(Zone* zone, const Value& value) {
if (value.isString()) {
if (value.toString()->isAtom()) {
return atomIsMarked(zone, &value.toString()->asAtom());
}
return true;
}
if (value.isSymbol()) {
return atomIsMarked(zone, value.toSymbol());
}
MOZ_ASSERT_IF(value.isGCThing(), value.hasObjectPayload() ||
value.isPrivateGCThing() ||
value.isBigInt());
return true;
}
#endif // DEBUG
} // namespace gc
#ifdef DEBUG
bool AtomIsMarked(Zone* zone, JSAtom* atom) {
return zone->runtimeFromAnyThread()->gc.atomMarking.atomIsMarked(zone, atom);
}
bool AtomIsMarked(Zone* zone, jsid id) {
return zone->runtimeFromAnyThread()->gc.atomMarking.idIsMarked(zone, id);
}
bool AtomIsMarked(Zone* zone, const Value& value) {
return zone->runtimeFromAnyThread()->gc.atomMarking.valueIsMarked(zone,
value);
}
#endif // DEBUG
} // namespace js