diff options
author | Daniel Baumann <daniel.baumann@progress-linux.org> | 2024-04-07 17:32:43 +0000 |
---|---|---|
committer | Daniel Baumann <daniel.baumann@progress-linux.org> | 2024-04-07 17:32:43 +0000 |
commit | 6bf0a5cb5034a7e684dcc3500e841785237ce2dd (patch) | |
tree | a68f146d7fa01f0134297619fbe7e33db084e0aa /security/sandbox/chromium/base/hash | |
parent | Initial commit. (diff) | |
download | thunderbird-upstream.tar.xz thunderbird-upstream.zip |
Adding upstream version 1:115.7.0.upstream/1%115.7.0upstream
Signed-off-by: Daniel Baumann <daniel.baumann@progress-linux.org>
Diffstat (limited to 'security/sandbox/chromium/base/hash')
-rw-r--r-- | security/sandbox/chromium/base/hash/hash.cc | 167 | ||||
-rw-r--r-- | security/sandbox/chromium/base/hash/hash.h | 86 |
2 files changed, 253 insertions, 0 deletions
diff --git a/security/sandbox/chromium/base/hash/hash.cc b/security/sandbox/chromium/base/hash/hash.cc new file mode 100644 index 0000000000..c96f8bc843 --- /dev/null +++ b/security/sandbox/chromium/base/hash/hash.cc @@ -0,0 +1,167 @@ +// Copyright 2014 The Chromium Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. + +#include "base/hash/hash.h" + +#include "base/rand_util.h" +#include "base/third_party/cityhash/city.h" +#include "build/build_config.h" + +// Definition in base/third_party/superfasthash/superfasthash.c. (Third-party +// code did not come with its own header file, so declaring the function here.) +// Note: This algorithm is also in Blink under Source/wtf/StringHasher.h. +extern "C" uint32_t SuperFastHash(const char* data, int len); + +namespace base { + +namespace { + +size_t FastHashImpl(base::span<const uint8_t> data) { + // We use the updated CityHash within our namespace (not the deprecated + // version from third_party/smhasher). +#if defined(ARCH_CPU_64_BITS) + return base::internal::cityhash_v111::CityHash64( + reinterpret_cast<const char*>(data.data()), data.size()); +#else + return base::internal::cityhash_v111::CityHash32( + reinterpret_cast<const char*>(data.data()), data.size()); +#endif +} + +// Implement hashing for pairs of at-most 32 bit integer values. +// When size_t is 32 bits, we turn the 64-bit hash code into 32 bits by using +// multiply-add hashing. This algorithm, as described in +// Theorem 4.3.3 of the thesis "Über die Komplexität der Multiplikation in +// eingeschränkten Branchingprogrammmodellen" by Woelfel, is: +// +// h32(x32, y32) = (h64(x32, y32) * rand_odd64 + rand16 * 2^16) % 2^64 / 2^32 +// +// Contact danakj@chromium.org for any questions. +size_t HashInts32Impl(uint32_t value1, uint32_t value2) { + uint64_t value1_64 = value1; + uint64_t hash64 = (value1_64 << 32) | value2; + + if (sizeof(size_t) >= sizeof(uint64_t)) + return static_cast<size_t>(hash64); + + uint64_t odd_random = 481046412LL << 32 | 1025306955LL; + uint32_t shift_random = 10121U << 16; + + hash64 = hash64 * odd_random + shift_random; + size_t high_bits = + static_cast<size_t>(hash64 >> (8 * (sizeof(uint64_t) - sizeof(size_t)))); + return high_bits; +} + +// Implement hashing for pairs of up-to 64-bit integer values. +// We use the compound integer hash method to produce a 64-bit hash code, by +// breaking the two 64-bit inputs into 4 32-bit values: +// http://opendatastructures.org/versions/edition-0.1d/ods-java/node33.html#SECTION00832000000000000000 +// Then we reduce our result to 32 bits if required, similar to above. +size_t HashInts64Impl(uint64_t value1, uint64_t value2) { + uint32_t short_random1 = 842304669U; + uint32_t short_random2 = 619063811U; + uint32_t short_random3 = 937041849U; + uint32_t short_random4 = 3309708029U; + + uint32_t value1a = static_cast<uint32_t>(value1 & 0xffffffff); + uint32_t value1b = static_cast<uint32_t>((value1 >> 32) & 0xffffffff); + uint32_t value2a = static_cast<uint32_t>(value2 & 0xffffffff); + uint32_t value2b = static_cast<uint32_t>((value2 >> 32) & 0xffffffff); + + uint64_t product1 = static_cast<uint64_t>(value1a) * short_random1; + uint64_t product2 = static_cast<uint64_t>(value1b) * short_random2; + uint64_t product3 = static_cast<uint64_t>(value2a) * short_random3; + uint64_t product4 = static_cast<uint64_t>(value2b) * short_random4; + + uint64_t hash64 = product1 + product2 + product3 + product4; + + if (sizeof(size_t) >= sizeof(uint64_t)) + return static_cast<size_t>(hash64); + + uint64_t odd_random = 1578233944LL << 32 | 194370989LL; + uint32_t shift_random = 20591U << 16; + + hash64 = hash64 * odd_random + shift_random; + size_t high_bits = + static_cast<size_t>(hash64 >> (8 * (sizeof(uint64_t) - sizeof(size_t)))); + return high_bits; +} + +// The random seed is used to perturb the output of base::FastHash() and +// base::HashInts() so that it is only deterministic within the lifetime of a +// process. This prevents inadvertent dependencies on the underlying +// implementation, e.g. anything that persists the hash value and expects it to +// be unchanging will break. +// +// Note: this is the same trick absl uses to generate a random seed. This is +// more robust than using base::RandBytes(), which can fail inside a sandboxed +// environment. Note that without ASLR, the seed won't be quite as random... +#if DCHECK_IS_ON() +constexpr const void* kSeed = &kSeed; +#endif + +template <typename T> +T Scramble(T input) { +#if DCHECK_IS_ON() + return HashInts64Impl(input, reinterpret_cast<uintptr_t>(kSeed)); +#else + return input; +#endif +} + +} // namespace + +size_t FastHash(base::span<const uint8_t> data) { + return Scramble(FastHashImpl(data)); +} + +uint32_t Hash(const void* data, size_t length) { + // Currently our in-memory hash is the same as the persistent hash. The + // split between in-memory and persistent hash functions is maintained to + // allow the in-memory hash function to be updated in the future. + return PersistentHash(data, length); +} + +uint32_t Hash(const std::string& str) { + return PersistentHash(as_bytes(make_span(str))); +} + +uint32_t Hash(const string16& str) { + return PersistentHash(as_bytes(make_span(str))); +} + +uint32_t PersistentHash(span<const uint8_t> data) { + // This hash function must not change, since it is designed to be persistable + // to disk. + if (data.size() > static_cast<size_t>(std::numeric_limits<int>::max())) { + NOTREACHED(); + return 0; + } + return ::SuperFastHash(reinterpret_cast<const char*>(data.data()), + static_cast<int>(data.size())); +} + +uint32_t PersistentHash(const void* data, size_t length) { + return PersistentHash(make_span(static_cast<const uint8_t*>(data), length)); +} + +uint32_t PersistentHash(const std::string& str) { + return PersistentHash(str.data(), str.size()); +} + +size_t HashInts32(uint32_t value1, uint32_t value2) { + return Scramble(HashInts32Impl(value1, value2)); +} + +// Implement hashing for pairs of up-to 64-bit integer values. +// We use the compound integer hash method to produce a 64-bit hash code, by +// breaking the two 64-bit inputs into 4 32-bit values: +// http://opendatastructures.org/versions/edition-0.1d/ods-java/node33.html#SECTION00832000000000000000 +// Then we reduce our result to 32 bits if required, similar to above. +size_t HashInts64(uint64_t value1, uint64_t value2) { + return Scramble(HashInts64Impl(value1, value2)); +} + +} // namespace base diff --git a/security/sandbox/chromium/base/hash/hash.h b/security/sandbox/chromium/base/hash/hash.h new file mode 100644 index 0000000000..f55ae45b38 --- /dev/null +++ b/security/sandbox/chromium/base/hash/hash.h @@ -0,0 +1,86 @@ +// Copyright (c) 2011 The Chromium Authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. + +#ifndef BASE_HASH_HASH_H_ +#define BASE_HASH_HASH_H_ + +#include <stddef.h> +#include <stdint.h> + +#include <limits> +#include <string> +#include <utility> + +#include "base/base_export.h" +#include "base/containers/span.h" +#include "base/logging.h" +#include "base/strings/string16.h" +#include "base/strings/string_piece.h" + +namespace base { + +// WARNING: This hash functions should not be used for any cryptographic +// purpose. + +// Deprecated: Computes a hash of a memory buffer, use FastHash() instead. +// If you need to persist a change on disk or between computers, use +// PersistentHash(). +// TODO(https://crbug.com/1025358): Migrate client code to new hash function. +BASE_EXPORT uint32_t Hash(const void* data, size_t length); +BASE_EXPORT uint32_t Hash(const std::string& str); +BASE_EXPORT uint32_t Hash(const string16& str); + +// Really *fast* and high quality hash. +// Recommended hash function for general use, we pick the best performant +// hash for each build target. +// It is prone to be updated whenever a newer/faster hash function is +// publicly available. +// May changed without warning, do not expect stability of outputs. +BASE_EXPORT size_t FastHash(base::span<const uint8_t> data); +inline size_t FastHash(StringPiece str) { + return FastHash(as_bytes(make_span(str))); +} + +// Computes a hash of a memory buffer. This hash function must not change so +// that code can use the hashed values for persistent storage purposes or +// sending across the network. If a new persistent hash function is desired, a +// new version will have to be added in addition. +// +// WARNING: This hash function should not be used for any cryptographic purpose. +BASE_EXPORT uint32_t PersistentHash(base::span<const uint8_t> data); +BASE_EXPORT uint32_t PersistentHash(const void* data, size_t length); +BASE_EXPORT uint32_t PersistentHash(const std::string& str); + +// Hash pairs of 32-bit or 64-bit numbers. +BASE_EXPORT size_t HashInts32(uint32_t value1, uint32_t value2); +BASE_EXPORT size_t HashInts64(uint64_t value1, uint64_t value2); + +template <typename T1, typename T2> +inline size_t HashInts(T1 value1, T2 value2) { + // This condition is expected to be compile-time evaluated and optimised away + // in release builds. + if (sizeof(T1) > sizeof(uint32_t) || (sizeof(T2) > sizeof(uint32_t))) + return HashInts64(value1, value2); + + return HashInts32(static_cast<uint32_t>(value1), + static_cast<uint32_t>(value2)); +} + +// A templated hasher for pairs of integer types. Example: +// +// using MyPair = std::pair<int32_t, int32_t>; +// std::unordered_set<MyPair, base::IntPairHash<MyPair>> set; +template <typename T> +struct IntPairHash; + +template <typename Type1, typename Type2> +struct IntPairHash<std::pair<Type1, Type2>> { + size_t operator()(std::pair<Type1, Type2> value) const { + return HashInts(value.first, value.second); + } +}; + +} // namespace base + +#endif // BASE_HASH_HASH_H_ |