| /* |
| * Copyright (C) 2004-2019 Apple Inc. All rights reserved. |
| * |
| * This library is free software; you can redistribute it and/or |
| * modify it under the terms of the GNU Library General Public |
| * License as published by the Free Software Foundation; either |
| * version 2 of the License, or (at your option) any later version. |
| * |
| * This library is distributed in the hope that it will be useful, |
| * but WITHOUT ANY WARRANTY; without even the implied warranty of |
| * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
| * Library General Public License for more details. |
| * |
| * You should have received a copy of the GNU Library General Public License |
| * along with this library; see the file COPYING.LIB. If not, write to |
| * the Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor, |
| * Boston, MA 02110-1301, USA. |
| * |
| */ |
| |
| #pragma once |
| |
| #include "JSExportMacros.h" |
| #include "PropertyOffset.h" |
| #include "Structure.h" |
| #include "WriteBarrier.h" |
| #include <wtf/HashTable.h> |
| #include <wtf/MathExtras.h> |
| #include <wtf/Vector.h> |
| #include <wtf/text/AtomStringImpl.h> |
| |
| |
| #define DUMP_PROPERTYMAP_STATS 0 |
| #define DUMP_PROPERTYMAP_COLLISIONS 0 |
| |
| #define PROPERTY_MAP_DELETED_ENTRY_KEY ((UniquedStringImpl*)1) |
| |
| namespace JSC { |
| |
| DECLARE_ALLOCATOR_WITH_HEAP_IDENTIFIER(PropertyTable); |
| |
| #if DUMP_PROPERTYMAP_STATS |
| |
| struct PropertyMapHashTableStats { |
| std::atomic<unsigned> numFinds; |
| std::atomic<unsigned> numCollisions; |
| std::atomic<unsigned> numLookups; |
| std::atomic<unsigned> numLookupProbing; |
| std::atomic<unsigned> numAdds; |
| std::atomic<unsigned> numRemoves; |
| std::atomic<unsigned> numRehashes; |
| std::atomic<unsigned> numReinserts; |
| }; |
| |
| JS_EXPORT_PRIVATE extern PropertyMapHashTableStats* propertyMapHashTableStats; |
| |
| #endif |
| |
| inline bool isPowerOf2(unsigned v) |
| { |
| return hasOneBitSet(v); |
| } |
| |
| inline unsigned nextPowerOf2(unsigned v) |
| { |
| // Taken from http://www.cs.utk.edu/~vose/c-stuff/bithacks.html |
| // Devised by Sean Anderson, Sepember 14, 2001 |
| |
| v--; |
| v |= v >> 1; |
| v |= v >> 2; |
| v |= v >> 4; |
| v |= v >> 8; |
| v |= v >> 16; |
| v++; |
| |
| return v; |
| } |
| |
| class PropertyTable final : public JSCell { |
| |
| // This is the implementation for 'iterator' and 'const_iterator', |
| // used for iterating over the table in insertion order. |
| template<typename T> |
| class ordered_iterator { |
| public: |
| ordered_iterator<T>& operator++() |
| { |
| m_valuePtr = skipDeletedEntries(m_valuePtr + 1, m_endValuePtr); |
| return *this; |
| } |
| |
| bool operator==(const ordered_iterator<T>& other) |
| { |
| return m_valuePtr == other.m_valuePtr; |
| } |
| |
| bool operator!=(const ordered_iterator<T>& other) |
| { |
| return m_valuePtr != other.m_valuePtr; |
| } |
| |
| T& operator*() |
| { |
| return *m_valuePtr; |
| } |
| |
| T* operator->() |
| { |
| return m_valuePtr; |
| } |
| |
| ordered_iterator(T* valuePtr, T* endValuePtr) |
| : m_valuePtr(valuePtr) |
| , m_endValuePtr(endValuePtr) |
| { |
| } |
| |
| private: |
| T* m_valuePtr; |
| T* m_endValuePtr; |
| }; |
| |
| public: |
| typedef JSCell Base; |
| static constexpr unsigned StructureFlags = Base::StructureFlags | StructureIsImmortal; |
| |
| template<typename CellType, SubspaceAccess> |
| static IsoSubspace* subspaceFor(VM& vm) |
| { |
| return &vm.propertyTableSpace; |
| } |
| |
| static constexpr bool needsDestruction = true; |
| static void destroy(JSCell*); |
| |
| DECLARE_EXPORT_INFO; |
| |
| static Structure* createStructure(VM& vm, JSGlobalObject* globalObject, JSValue prototype) |
| { |
| return Structure::create(vm, globalObject, prototype, TypeInfo(CellType, StructureFlags), info()); |
| } |
| |
| typedef UniquedStringImpl* KeyType; |
| typedef PropertyMapEntry ValueType; |
| |
| // The in order iterator provides overloaded * and -> to access the Value at the current position. |
| typedef ordered_iterator<ValueType> iterator; |
| typedef ordered_iterator<const ValueType> const_iterator; |
| |
| // The find_iterator is a pair of a pointer to a Value* an the entry in the index. |
| // If 'find' does not find an entry then iter.first will be 0, and iter.second will |
| // give the point in m_index where an entry should be inserted. |
| typedef std::pair<ValueType*, unsigned> find_iterator; |
| |
| // Constructor is passed an initial capacity, a PropertyTable to copy, or both. |
| static PropertyTable* create(VM&, unsigned initialCapacity); |
| static PropertyTable* clone(VM&, const PropertyTable&); |
| static PropertyTable* clone(VM&, unsigned initialCapacity, const PropertyTable&); |
| ~PropertyTable(); |
| |
| // Ordered iteration methods. |
| iterator begin(); |
| iterator end(); |
| const_iterator begin() const; |
| const_iterator end() const; |
| |
| // Find a value in the table. |
| find_iterator find(const KeyType&); |
| ValueType* get(const KeyType&); |
| // Add a value to the table |
| std::pair<find_iterator, bool> WARN_UNUSED_RETURN add(const ValueType& entry); |
| // Remove a value from the table. |
| void remove(const find_iterator& iter); |
| void remove(const KeyType& key); |
| |
| // Returns the number of values in the hashtable. |
| unsigned size() const; |
| |
| // Checks if there are any values in the hashtable. |
| bool isEmpty() const; |
| |
| // Number of slots in the property storage array in use, included deletedOffsets. |
| unsigned propertyStorageSize() const; |
| |
| // Used to maintain a list of unused entries in the property storage. |
| void clearDeletedOffsets(); |
| bool hasDeletedOffset(); |
| PropertyOffset getDeletedOffset(); |
| void addDeletedOffset(PropertyOffset); |
| |
| PropertyOffset nextOffset(PropertyOffset inlineCapacity); |
| |
| // Copy this PropertyTable, ensuring the copy has at least the capacity provided. |
| PropertyTable* copy(VM&, unsigned newCapacity); |
| |
| #ifndef NDEBUG |
| size_t sizeInMemory(); |
| void checkConsistency(); |
| #endif |
| |
| static ptrdiff_t offsetOfIndexSize() { return OBJECT_OFFSETOF(PropertyTable, m_indexSize); } |
| static ptrdiff_t offsetOfIndexMask() { return OBJECT_OFFSETOF(PropertyTable, m_indexMask); } |
| static ptrdiff_t offsetOfIndex() { return OBJECT_OFFSETOF(PropertyTable, m_index); } |
| |
| static constexpr unsigned EmptyEntryIndex = 0; |
| |
| private: |
| PropertyTable(VM&, unsigned initialCapacity); |
| PropertyTable(VM&, const PropertyTable&); |
| PropertyTable(VM&, unsigned initialCapacity, const PropertyTable&); |
| |
| PropertyTable(const PropertyTable&); |
| // Used to insert a value known not to be in the table, and where we know capacity to be available. |
| void reinsert(const ValueType& entry); |
| |
| // Rehash the table. Used to grow, or to recover deleted slots. |
| void rehash(unsigned newCapacity); |
| |
| // The capacity of the table of values is half of the size of the index. |
| unsigned tableCapacity() const; |
| |
| // We keep an extra deleted slot after the array to make iteration work, |
| // and to use for deleted values. Index values into the array are 1-based, |
| // so this is tableCapacity() + 1. |
| // For example, if m_tableSize is 16, then tableCapacity() is 8 - but the |
| // values array is actually 9 long (the 9th used for the deleted value/ |
| // iteration guard). The 8 valid entries are numbered 1..8, so the |
| // deleted index is 9 (0 being reserved for empty). |
| unsigned deletedEntryIndex() const; |
| |
| // Used in iterator creation/progression. |
| template<typename T> |
| static T* skipDeletedEntries(T* valuePtr, T* endValuePtr); |
| |
| // The table of values lies after the hash index. |
| ValueType* table(); |
| const ValueType* table() const; |
| |
| ValueType* tableEnd() { return table() + usedCount(); } |
| const ValueType* tableEnd() const { return table() + usedCount(); } |
| |
| // total number of used entries in the values array - by either valid entries, or deleted ones. |
| unsigned usedCount() const; |
| |
| // The size in bytes of data needed for by the table. |
| size_t dataSize(); |
| |
| // Calculates the appropriate table size (rounds up to a power of two). |
| static unsigned sizeForCapacity(unsigned capacity); |
| |
| // Check if capacity is available. |
| bool canInsert(); |
| |
| unsigned m_indexSize; |
| unsigned m_indexMask; |
| unsigned* m_index; |
| unsigned m_keyCount; |
| unsigned m_deletedCount; |
| std::unique_ptr<Vector<PropertyOffset>> m_deletedOffsets; |
| |
| static constexpr unsigned MinimumTableSize = 16; |
| }; |
| |
| inline PropertyTable::iterator PropertyTable::begin() |
| { |
| auto* tableEnd = this->tableEnd(); |
| return iterator(skipDeletedEntries(table(), tableEnd), tableEnd); |
| } |
| |
| inline PropertyTable::iterator PropertyTable::end() |
| { |
| auto* tableEnd = this->tableEnd(); |
| return iterator(tableEnd, tableEnd); |
| } |
| |
| inline PropertyTable::const_iterator PropertyTable::begin() const |
| { |
| auto* tableEnd = this->tableEnd(); |
| return const_iterator(skipDeletedEntries(table(), tableEnd), tableEnd); |
| } |
| |
| inline PropertyTable::const_iterator PropertyTable::end() const |
| { |
| auto* tableEnd = this->tableEnd(); |
| return const_iterator(tableEnd, tableEnd); |
| } |
| |
| inline PropertyTable::find_iterator PropertyTable::find(const KeyType& key) |
| { |
| ASSERT(key); |
| ASSERT(key->isAtom() || key->isSymbol()); |
| unsigned hash = IdentifierRepHash::hash(key); |
| |
| #if DUMP_PROPERTYMAP_STATS |
| ++propertyMapHashTableStats->numFinds; |
| #endif |
| |
| while (true) { |
| unsigned entryIndex = m_index[hash & m_indexMask]; |
| if (entryIndex == EmptyEntryIndex) |
| return std::make_pair((ValueType*)0, hash & m_indexMask); |
| if (key == table()[entryIndex - 1].key) |
| return std::make_pair(&table()[entryIndex - 1], hash & m_indexMask); |
| |
| #if DUMP_PROPERTYMAP_STATS |
| ++propertyMapHashTableStats->numCollisions; |
| #endif |
| |
| #if DUMP_PROPERTYMAP_COLLISIONS |
| dataLog("PropertyTable collision for ", key, " (", hash, ")\n"); |
| dataLog("Collided with ", table()[entryIndex - 1].key, "(", IdentifierRepHash::hash(table()[entryIndex - 1].key), ")\n"); |
| #endif |
| |
| hash++; |
| } |
| } |
| |
| inline PropertyTable::ValueType* PropertyTable::get(const KeyType& key) |
| { |
| ASSERT(key); |
| ASSERT(key->isAtom() || key->isSymbol()); |
| ASSERT(key != PROPERTY_MAP_DELETED_ENTRY_KEY); |
| |
| if (!m_keyCount) |
| return nullptr; |
| |
| unsigned hash = IdentifierRepHash::hash(key); |
| |
| #if DUMP_PROPERTYMAP_STATS |
| ++propertyMapHashTableStats->numLookups; |
| #endif |
| |
| while (true) { |
| unsigned entryIndex = m_index[hash & m_indexMask]; |
| if (entryIndex == EmptyEntryIndex) |
| return nullptr; |
| if (key == table()[entryIndex - 1].key) { |
| ASSERT(!m_deletedOffsets || !m_deletedOffsets->contains(table()[entryIndex - 1].offset)); |
| return &table()[entryIndex - 1]; |
| } |
| |
| #if DUMP_PROPERTYMAP_STATS |
| ++propertyMapHashTableStats->numLookupProbing; |
| #endif |
| |
| hash++; |
| } |
| } |
| |
| inline std::pair<PropertyTable::find_iterator, bool> WARN_UNUSED_RETURN PropertyTable::add(const ValueType& entry) |
| { |
| ASSERT(!m_deletedOffsets || !m_deletedOffsets->contains(entry.offset)); |
| |
| // Look for a value with a matching key already in the array. |
| find_iterator iter = find(entry.key); |
| if (iter.first) |
| return std::make_pair(iter, false); |
| |
| #if DUMP_PROPERTYMAP_STATS |
| ++propertyMapHashTableStats->numAdds; |
| #endif |
| |
| // Ref the key |
| entry.key->ref(); |
| |
| // ensure capacity is available. |
| if (!canInsert()) { |
| rehash(m_keyCount + 1); |
| iter = find(entry.key); |
| ASSERT(!iter.first); |
| } |
| |
| // Allocate a slot in the hashtable, and set the index to reference this. |
| unsigned entryIndex = usedCount() + 1; |
| m_index[iter.second] = entryIndex; |
| iter.first = &table()[entryIndex - 1]; |
| *iter.first = entry; |
| |
| ++m_keyCount; |
| |
| return std::make_pair(iter, true); |
| } |
| |
| inline void PropertyTable::remove(const find_iterator& iter) |
| { |
| // Removing a key that doesn't exist does nothing! |
| if (!iter.first) |
| return; |
| |
| #if DUMP_PROPERTYMAP_STATS |
| ++propertyMapHashTableStats->numRemoves; |
| #endif |
| |
| // Replace this one element with the deleted sentinel. Also clear out |
| // the entry so we can iterate all the entries as needed. |
| m_index[iter.second] = deletedEntryIndex(); |
| iter.first->key->deref(); |
| iter.first->key = PROPERTY_MAP_DELETED_ENTRY_KEY; |
| |
| ASSERT(m_keyCount >= 1); |
| --m_keyCount; |
| ++m_deletedCount; |
| |
| if (m_deletedCount * 4 >= m_indexSize) |
| rehash(m_keyCount); |
| } |
| |
| inline void PropertyTable::remove(const KeyType& key) |
| { |
| remove(find(key)); |
| } |
| |
| // returns the number of values in the hashtable. |
| inline unsigned PropertyTable::size() const |
| { |
| return m_keyCount; |
| } |
| |
| inline bool PropertyTable::isEmpty() const |
| { |
| return !m_keyCount; |
| } |
| |
| inline unsigned PropertyTable::propertyStorageSize() const |
| { |
| return size() + (m_deletedOffsets ? m_deletedOffsets->size() : 0); |
| } |
| |
| inline void PropertyTable::clearDeletedOffsets() |
| { |
| m_deletedOffsets = nullptr; |
| } |
| |
| inline bool PropertyTable::hasDeletedOffset() |
| { |
| return m_deletedOffsets && !m_deletedOffsets->isEmpty(); |
| } |
| |
| inline PropertyOffset PropertyTable::getDeletedOffset() |
| { |
| PropertyOffset offset = m_deletedOffsets->last(); |
| m_deletedOffsets->removeLast(); |
| return offset; |
| } |
| |
| inline void PropertyTable::addDeletedOffset(PropertyOffset offset) |
| { |
| if (!m_deletedOffsets) |
| m_deletedOffsets = makeUnique<Vector<PropertyOffset>>(); |
| ASSERT(!m_deletedOffsets->contains(offset)); |
| m_deletedOffsets->append(offset); |
| } |
| |
| inline PropertyOffset PropertyTable::nextOffset(PropertyOffset inlineCapacity) |
| { |
| if (hasDeletedOffset()) |
| return getDeletedOffset(); |
| |
| return offsetForPropertyNumber(size(), inlineCapacity); |
| } |
| |
| inline PropertyTable* PropertyTable::copy(VM& vm, unsigned newCapacity) |
| { |
| ASSERT(newCapacity >= m_keyCount); |
| |
| // Fast case; if the new table will be the same m_indexSize as this one, we can memcpy it, |
| // save rehashing all keys. |
| if (sizeForCapacity(newCapacity) == m_indexSize) |
| return PropertyTable::clone(vm, *this); |
| return PropertyTable::clone(vm, newCapacity, *this); |
| } |
| |
| #ifndef NDEBUG |
| inline size_t PropertyTable::sizeInMemory() |
| { |
| size_t result = sizeof(PropertyTable) + dataSize(); |
| if (m_deletedOffsets) |
| result += (m_deletedOffsets->capacity() * sizeof(PropertyOffset)); |
| return result; |
| } |
| #endif |
| |
| inline void PropertyTable::reinsert(const ValueType& entry) |
| { |
| #if DUMP_PROPERTYMAP_STATS |
| ++propertyMapHashTableStats->numReinserts; |
| #endif |
| |
| // Used to insert a value known not to be in the table, and where |
| // we know capacity to be available. |
| ASSERT(canInsert()); |
| find_iterator iter = find(entry.key); |
| ASSERT(!iter.first); |
| |
| unsigned entryIndex = usedCount() + 1; |
| m_index[iter.second] = entryIndex; |
| table()[entryIndex - 1] = entry; |
| |
| ++m_keyCount; |
| } |
| |
| inline void PropertyTable::rehash(unsigned newCapacity) |
| { |
| #if DUMP_PROPERTYMAP_STATS |
| ++propertyMapHashTableStats->numRehashes; |
| #endif |
| |
| unsigned* oldEntryIndices = m_index; |
| iterator iter = this->begin(); |
| iterator end = this->end(); |
| |
| m_indexSize = sizeForCapacity(newCapacity); |
| m_indexMask = m_indexSize - 1; |
| m_keyCount = 0; |
| m_deletedCount = 0; |
| |
| m_index = static_cast<unsigned*>(PropertyTableMalloc::zeroedMalloc(dataSize())); |
| |
| for (; iter != end; ++iter) { |
| ASSERT(canInsert()); |
| reinsert(*iter); |
| } |
| |
| PropertyTableMalloc::free(oldEntryIndices); |
| } |
| |
| inline unsigned PropertyTable::tableCapacity() const { return m_indexSize >> 1; } |
| |
| inline unsigned PropertyTable::deletedEntryIndex() const { return tableCapacity() + 1; } |
| |
| template<typename T> |
| inline T* PropertyTable::skipDeletedEntries(T* valuePtr, T* endValuePtr) |
| { |
| while (valuePtr < endValuePtr && valuePtr->key == PROPERTY_MAP_DELETED_ENTRY_KEY) |
| ++valuePtr; |
| return valuePtr; |
| } |
| |
| inline PropertyTable::ValueType* PropertyTable::table() |
| { |
| // The table of values lies after the hash index. |
| return reinterpret_cast<ValueType*>(m_index + m_indexSize); |
| } |
| |
| inline const PropertyTable::ValueType* PropertyTable::table() const |
| { |
| // The table of values lies after the hash index. |
| return reinterpret_cast<const ValueType*>(m_index + m_indexSize); |
| } |
| |
| inline unsigned PropertyTable::usedCount() const |
| { |
| // Total number of used entries in the values array - by either valid entries, or deleted ones. |
| return m_keyCount + m_deletedCount; |
| } |
| |
| inline size_t PropertyTable::dataSize() |
| { |
| // The size in bytes of data needed for by the table. |
| return m_indexSize * sizeof(unsigned) + ((tableCapacity()) + 1) * sizeof(ValueType); |
| } |
| |
| inline unsigned PropertyTable::sizeForCapacity(unsigned capacity) |
| { |
| if (capacity < MinimumTableSize / 2) |
| return MinimumTableSize; |
| return nextPowerOf2(capacity + 1) * 2; |
| } |
| |
| inline bool PropertyTable::canInsert() |
| { |
| return usedCount() < tableCapacity(); |
| } |
| |
| } // namespace JSC |