|
|
|
// Copyright 2017 The Abseil Authors.
|
|
|
|
//
|
|
|
|
// Licensed under the Apache License, Version 2.0 (the "License");
|
|
|
|
// you may not use this file except in compliance with the License.
|
|
|
|
// You may obtain a copy of the License at
|
|
|
|
//
|
|
|
|
// https://www.apache.org/licenses/LICENSE-2.0
|
|
|
|
//
|
|
|
|
// Unless required by applicable law or agreed to in writing, software
|
|
|
|
// distributed under the License is distributed on an "AS IS" BASIS,
|
|
|
|
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
|
|
// See the License for the specific language governing permissions and
|
|
|
|
// limitations under the License.
|
|
|
|
|
|
|
|
#include "absl/random/internal/randen_slow.h"
|
|
|
|
|
|
|
|
#include <cstddef>
|
|
|
|
#include <cstdint>
|
|
|
|
#include <cstring>
|
|
|
|
|
|
|
|
#include "absl/base/attributes.h"
|
|
|
|
#include "absl/base/internal/endian.h"
|
Export of internal Abseil changes
--
007ce045d5d38a727ededdb5bf06e64785fd73bd by Martijn Vels <mvels@google.com>:
Add `cord_enable_btree` feature flag (default false).
PiperOrigin-RevId: 383729939
--
98e7dc6a0407b0fd7b8713d883cdb3a766e0583d by Benjamin Barenblat <bbaren@google.com>:
Eliminate some byte swapping from randen_slow
Stop swapping bytes when serializing randen_slow’s Vector128 into and
out of memory. Instead, simply index different bytes in the AES round
function. This requires byte swapping the te{0..3} lookup tables, but it
produces an 8% speedup on my Xeon W-2135.
PiperOrigin-RevId: 383689402
--
180b6bf45049188840d439b16a28e6b968669340 by Evan Brown <ezb@google.com>:
Minor simplification in drop_deletes_without_resize() - save probe_offset outside the lambda.
Also, add some consts, avoid an auto, and use lambda capture by value instead of reference.
I realized that the compiler can already optimize this - https://godbolt.org/z/Wxd9c4TfK, but I think this way makes the code a bit clearer.
PiperOrigin-RevId: 383646658
--
781706a974c4dc1c0abbb6b801fca0550229e883 by Martijn Vels <mvels@google.com>:
Change storage to contain 3 bytes.
As per the comments in the code, this allows us to utilize all available space in CordRep that may otherwise be 'lost' in padding in derived clases. For the upcoming CordrepBtree class, we want a strong guarantee on having a 64 bytes aligned implementation.
PiperOrigin-RevId: 383633963
--
8fe22ecf92492fa6649938a2215934ebfe01c714 by Derek Mauro <dmauro@google.com>:
Remove reference to str_format_arg.h, which no longer exists
PiperOrigin-RevId: 383517865
--
79397f3b18f18c1e2d7aea993b687329d626ce64 by Benjamin Barenblat <bbaren@google.com>:
Use absl::uint128 for AES random number generator
Replace randen’s internal 128-bit integer struct, u64x2, with
absl::uint128. This eliminates some code and improves support for
big-endian platforms.
PiperOrigin-RevId: 383475671
GitOrigin-RevId: 007ce045d5d38a727ededdb5bf06e64785fd73bd
Change-Id: Ia9d9c40de557221f1744fb0d6d4d6ca7ac569070
3 years ago
|
|
|
#include "absl/numeric/int128.h"
|
|
|
|
#include "absl/random/internal/platform.h"
|
|
|
|
#include "absl/random/internal/randen_traits.h"
|
|
|
|
|
|
|
|
#if ABSL_HAVE_ATTRIBUTE(always_inline) || \
|
|
|
|
(defined(__GNUC__) && !defined(__clang__))
|
|
|
|
#define ABSL_RANDOM_INTERNAL_ATTRIBUTE_ALWAYS_INLINE \
|
|
|
|
__attribute__((always_inline))
|
|
|
|
#elif defined(_MSC_VER)
|
|
|
|
// We can achieve something similar to attribute((always_inline)) with MSVC by
|
|
|
|
// using the __forceinline keyword, however this is not perfect. MSVC is
|
|
|
|
// much less aggressive about inlining, and even with the __forceinline keyword.
|
|
|
|
#define ABSL_RANDOM_INTERNAL_ATTRIBUTE_ALWAYS_INLINE __forceinline
|
|
|
|
#else
|
|
|
|
#define ABSL_RANDOM_INTERNAL_ATTRIBUTE_ALWAYS_INLINE
|
|
|
|
#endif
|
|
|
|
|
|
|
|
namespace {
|
|
|
|
|
|
|
|
// AES portions based on rijndael-alg-fst.c,
|
Export of internal Abseil changes
--
007ce045d5d38a727ededdb5bf06e64785fd73bd by Martijn Vels <mvels@google.com>:
Add `cord_enable_btree` feature flag (default false).
PiperOrigin-RevId: 383729939
--
98e7dc6a0407b0fd7b8713d883cdb3a766e0583d by Benjamin Barenblat <bbaren@google.com>:
Eliminate some byte swapping from randen_slow
Stop swapping bytes when serializing randen_slow’s Vector128 into and
out of memory. Instead, simply index different bytes in the AES round
function. This requires byte swapping the te{0..3} lookup tables, but it
produces an 8% speedup on my Xeon W-2135.
PiperOrigin-RevId: 383689402
--
180b6bf45049188840d439b16a28e6b968669340 by Evan Brown <ezb@google.com>:
Minor simplification in drop_deletes_without_resize() - save probe_offset outside the lambda.
Also, add some consts, avoid an auto, and use lambda capture by value instead of reference.
I realized that the compiler can already optimize this - https://godbolt.org/z/Wxd9c4TfK, but I think this way makes the code a bit clearer.
PiperOrigin-RevId: 383646658
--
781706a974c4dc1c0abbb6b801fca0550229e883 by Martijn Vels <mvels@google.com>:
Change storage to contain 3 bytes.
As per the comments in the code, this allows us to utilize all available space in CordRep that may otherwise be 'lost' in padding in derived clases. For the upcoming CordrepBtree class, we want a strong guarantee on having a 64 bytes aligned implementation.
PiperOrigin-RevId: 383633963
--
8fe22ecf92492fa6649938a2215934ebfe01c714 by Derek Mauro <dmauro@google.com>:
Remove reference to str_format_arg.h, which no longer exists
PiperOrigin-RevId: 383517865
--
79397f3b18f18c1e2d7aea993b687329d626ce64 by Benjamin Barenblat <bbaren@google.com>:
Use absl::uint128 for AES random number generator
Replace randen’s internal 128-bit integer struct, u64x2, with
absl::uint128. This eliminates some code and improves support for
big-endian platforms.
PiperOrigin-RevId: 383475671
GitOrigin-RevId: 007ce045d5d38a727ededdb5bf06e64785fd73bd
Change-Id: Ia9d9c40de557221f1744fb0d6d4d6ca7ac569070
3 years ago
|
|
|
// https://fastcrypto.org/front/misc/rijndael-alg-fst.c, and modified for
|
|
|
|
// platform-endianness.
|
|
|
|
//
|
|
|
|
// Implementation of
|
|
|
|
// http://www.csrc.nist.gov/publications/fips/fips197/fips-197.pdf
|
|
|
|
constexpr uint32_t te0[256] = {
|
Export of internal Abseil changes
--
007ce045d5d38a727ededdb5bf06e64785fd73bd by Martijn Vels <mvels@google.com>:
Add `cord_enable_btree` feature flag (default false).
PiperOrigin-RevId: 383729939
--
98e7dc6a0407b0fd7b8713d883cdb3a766e0583d by Benjamin Barenblat <bbaren@google.com>:
Eliminate some byte swapping from randen_slow
Stop swapping bytes when serializing randen_slow’s Vector128 into and
out of memory. Instead, simply index different bytes in the AES round
function. This requires byte swapping the te{0..3} lookup tables, but it
produces an 8% speedup on my Xeon W-2135.
PiperOrigin-RevId: 383689402
--
180b6bf45049188840d439b16a28e6b968669340 by Evan Brown <ezb@google.com>:
Minor simplification in drop_deletes_without_resize() - save probe_offset outside the lambda.
Also, add some consts, avoid an auto, and use lambda capture by value instead of reference.
I realized that the compiler can already optimize this - https://godbolt.org/z/Wxd9c4TfK, but I think this way makes the code a bit clearer.
PiperOrigin-RevId: 383646658
--
781706a974c4dc1c0abbb6b801fca0550229e883 by Martijn Vels <mvels@google.com>:
Change storage to contain 3 bytes.
As per the comments in the code, this allows us to utilize all available space in CordRep that may otherwise be 'lost' in padding in derived clases. For the upcoming CordrepBtree class, we want a strong guarantee on having a 64 bytes aligned implementation.
PiperOrigin-RevId: 383633963
--
8fe22ecf92492fa6649938a2215934ebfe01c714 by Derek Mauro <dmauro@google.com>:
Remove reference to str_format_arg.h, which no longer exists
PiperOrigin-RevId: 383517865
--
79397f3b18f18c1e2d7aea993b687329d626ce64 by Benjamin Barenblat <bbaren@google.com>:
Use absl::uint128 for AES random number generator
Replace randen’s internal 128-bit integer struct, u64x2, with
absl::uint128. This eliminates some code and improves support for
big-endian platforms.
PiperOrigin-RevId: 383475671
GitOrigin-RevId: 007ce045d5d38a727ededdb5bf06e64785fd73bd
Change-Id: Ia9d9c40de557221f1744fb0d6d4d6ca7ac569070
3 years ago
|
|
|
0xa56363c6, 0x847c7cf8, 0x997777ee, 0x8d7b7bf6, 0x0df2f2ff, 0xbd6b6bd6,
|
|
|
|
0xb16f6fde, 0x54c5c591, 0x50303060, 0x03010102, 0xa96767ce, 0x7d2b2b56,
|
|
|
|
0x19fefee7, 0x62d7d7b5, 0xe6abab4d, 0x9a7676ec, 0x45caca8f, 0x9d82821f,
|
|
|
|
0x40c9c989, 0x877d7dfa, 0x15fafaef, 0xeb5959b2, 0xc947478e, 0x0bf0f0fb,
|
|
|
|
0xecadad41, 0x67d4d4b3, 0xfda2a25f, 0xeaafaf45, 0xbf9c9c23, 0xf7a4a453,
|
|
|
|
0x967272e4, 0x5bc0c09b, 0xc2b7b775, 0x1cfdfde1, 0xae93933d, 0x6a26264c,
|
|
|
|
0x5a36366c, 0x413f3f7e, 0x02f7f7f5, 0x4fcccc83, 0x5c343468, 0xf4a5a551,
|
|
|
|
0x34e5e5d1, 0x08f1f1f9, 0x937171e2, 0x73d8d8ab, 0x53313162, 0x3f15152a,
|
|
|
|
0x0c040408, 0x52c7c795, 0x65232346, 0x5ec3c39d, 0x28181830, 0xa1969637,
|
|
|
|
0x0f05050a, 0xb59a9a2f, 0x0907070e, 0x36121224, 0x9b80801b, 0x3de2e2df,
|
|
|
|
0x26ebebcd, 0x6927274e, 0xcdb2b27f, 0x9f7575ea, 0x1b090912, 0x9e83831d,
|
|
|
|
0x742c2c58, 0x2e1a1a34, 0x2d1b1b36, 0xb26e6edc, 0xee5a5ab4, 0xfba0a05b,
|
|
|
|
0xf65252a4, 0x4d3b3b76, 0x61d6d6b7, 0xceb3b37d, 0x7b292952, 0x3ee3e3dd,
|
|
|
|
0x712f2f5e, 0x97848413, 0xf55353a6, 0x68d1d1b9, 0x00000000, 0x2cededc1,
|
|
|
|
0x60202040, 0x1ffcfce3, 0xc8b1b179, 0xed5b5bb6, 0xbe6a6ad4, 0x46cbcb8d,
|
|
|
|
0xd9bebe67, 0x4b393972, 0xde4a4a94, 0xd44c4c98, 0xe85858b0, 0x4acfcf85,
|
|
|
|
0x6bd0d0bb, 0x2aefefc5, 0xe5aaaa4f, 0x16fbfbed, 0xc5434386, 0xd74d4d9a,
|
|
|
|
0x55333366, 0x94858511, 0xcf45458a, 0x10f9f9e9, 0x06020204, 0x817f7ffe,
|
|
|
|
0xf05050a0, 0x443c3c78, 0xba9f9f25, 0xe3a8a84b, 0xf35151a2, 0xfea3a35d,
|
|
|
|
0xc0404080, 0x8a8f8f05, 0xad92923f, 0xbc9d9d21, 0x48383870, 0x04f5f5f1,
|
|
|
|
0xdfbcbc63, 0xc1b6b677, 0x75dadaaf, 0x63212142, 0x30101020, 0x1affffe5,
|
|
|
|
0x0ef3f3fd, 0x6dd2d2bf, 0x4ccdcd81, 0x140c0c18, 0x35131326, 0x2fececc3,
|
|
|
|
0xe15f5fbe, 0xa2979735, 0xcc444488, 0x3917172e, 0x57c4c493, 0xf2a7a755,
|
|
|
|
0x827e7efc, 0x473d3d7a, 0xac6464c8, 0xe75d5dba, 0x2b191932, 0x957373e6,
|
|
|
|
0xa06060c0, 0x98818119, 0xd14f4f9e, 0x7fdcdca3, 0x66222244, 0x7e2a2a54,
|
|
|
|
0xab90903b, 0x8388880b, 0xca46468c, 0x29eeeec7, 0xd3b8b86b, 0x3c141428,
|
|
|
|
0x79dedea7, 0xe25e5ebc, 0x1d0b0b16, 0x76dbdbad, 0x3be0e0db, 0x56323264,
|
|
|
|
0x4e3a3a74, 0x1e0a0a14, 0xdb494992, 0x0a06060c, 0x6c242448, 0xe45c5cb8,
|
|
|
|
0x5dc2c29f, 0x6ed3d3bd, 0xefacac43, 0xa66262c4, 0xa8919139, 0xa4959531,
|
|
|
|
0x37e4e4d3, 0x8b7979f2, 0x32e7e7d5, 0x43c8c88b, 0x5937376e, 0xb76d6dda,
|
|
|
|
0x8c8d8d01, 0x64d5d5b1, 0xd24e4e9c, 0xe0a9a949, 0xb46c6cd8, 0xfa5656ac,
|
|
|
|
0x07f4f4f3, 0x25eaeacf, 0xaf6565ca, 0x8e7a7af4, 0xe9aeae47, 0x18080810,
|
|
|
|
0xd5baba6f, 0x887878f0, 0x6f25254a, 0x722e2e5c, 0x241c1c38, 0xf1a6a657,
|
|
|
|
0xc7b4b473, 0x51c6c697, 0x23e8e8cb, 0x7cdddda1, 0x9c7474e8, 0x211f1f3e,
|
|
|
|
0xdd4b4b96, 0xdcbdbd61, 0x868b8b0d, 0x858a8a0f, 0x907070e0, 0x423e3e7c,
|
|
|
|
0xc4b5b571, 0xaa6666cc, 0xd8484890, 0x05030306, 0x01f6f6f7, 0x120e0e1c,
|
|
|
|
0xa36161c2, 0x5f35356a, 0xf95757ae, 0xd0b9b969, 0x91868617, 0x58c1c199,
|
|
|
|
0x271d1d3a, 0xb99e9e27, 0x38e1e1d9, 0x13f8f8eb, 0xb398982b, 0x33111122,
|
|
|
|
0xbb6969d2, 0x70d9d9a9, 0x898e8e07, 0xa7949433, 0xb69b9b2d, 0x221e1e3c,
|
|
|
|
0x92878715, 0x20e9e9c9, 0x49cece87, 0xff5555aa, 0x78282850, 0x7adfdfa5,
|
|
|
|
0x8f8c8c03, 0xf8a1a159, 0x80898909, 0x170d0d1a, 0xdabfbf65, 0x31e6e6d7,
|
|
|
|
0xc6424284, 0xb86868d0, 0xc3414182, 0xb0999929, 0x772d2d5a, 0x110f0f1e,
|
|
|
|
0xcbb0b07b, 0xfc5454a8, 0xd6bbbb6d, 0x3a16162c,
|
|
|
|
};
|
|
|
|
|
|
|
|
constexpr uint32_t te1[256] = {
|
Export of internal Abseil changes
--
007ce045d5d38a727ededdb5bf06e64785fd73bd by Martijn Vels <mvels@google.com>:
Add `cord_enable_btree` feature flag (default false).
PiperOrigin-RevId: 383729939
--
98e7dc6a0407b0fd7b8713d883cdb3a766e0583d by Benjamin Barenblat <bbaren@google.com>:
Eliminate some byte swapping from randen_slow
Stop swapping bytes when serializing randen_slow’s Vector128 into and
out of memory. Instead, simply index different bytes in the AES round
function. This requires byte swapping the te{0..3} lookup tables, but it
produces an 8% speedup on my Xeon W-2135.
PiperOrigin-RevId: 383689402
--
180b6bf45049188840d439b16a28e6b968669340 by Evan Brown <ezb@google.com>:
Minor simplification in drop_deletes_without_resize() - save probe_offset outside the lambda.
Also, add some consts, avoid an auto, and use lambda capture by value instead of reference.
I realized that the compiler can already optimize this - https://godbolt.org/z/Wxd9c4TfK, but I think this way makes the code a bit clearer.
PiperOrigin-RevId: 383646658
--
781706a974c4dc1c0abbb6b801fca0550229e883 by Martijn Vels <mvels@google.com>:
Change storage to contain 3 bytes.
As per the comments in the code, this allows us to utilize all available space in CordRep that may otherwise be 'lost' in padding in derived clases. For the upcoming CordrepBtree class, we want a strong guarantee on having a 64 bytes aligned implementation.
PiperOrigin-RevId: 383633963
--
8fe22ecf92492fa6649938a2215934ebfe01c714 by Derek Mauro <dmauro@google.com>:
Remove reference to str_format_arg.h, which no longer exists
PiperOrigin-RevId: 383517865
--
79397f3b18f18c1e2d7aea993b687329d626ce64 by Benjamin Barenblat <bbaren@google.com>:
Use absl::uint128 for AES random number generator
Replace randen’s internal 128-bit integer struct, u64x2, with
absl::uint128. This eliminates some code and improves support for
big-endian platforms.
PiperOrigin-RevId: 383475671
GitOrigin-RevId: 007ce045d5d38a727ededdb5bf06e64785fd73bd
Change-Id: Ia9d9c40de557221f1744fb0d6d4d6ca7ac569070
3 years ago
|
|
|
0x6363c6a5, 0x7c7cf884, 0x7777ee99, 0x7b7bf68d, 0xf2f2ff0d, 0x6b6bd6bd,
|
|
|
|
0x6f6fdeb1, 0xc5c59154, 0x30306050, 0x01010203, 0x6767cea9, 0x2b2b567d,
|
|
|
|
0xfefee719, 0xd7d7b562, 0xabab4de6, 0x7676ec9a, 0xcaca8f45, 0x82821f9d,
|
|
|
|
0xc9c98940, 0x7d7dfa87, 0xfafaef15, 0x5959b2eb, 0x47478ec9, 0xf0f0fb0b,
|
|
|
|
0xadad41ec, 0xd4d4b367, 0xa2a25ffd, 0xafaf45ea, 0x9c9c23bf, 0xa4a453f7,
|
|
|
|
0x7272e496, 0xc0c09b5b, 0xb7b775c2, 0xfdfde11c, 0x93933dae, 0x26264c6a,
|
|
|
|
0x36366c5a, 0x3f3f7e41, 0xf7f7f502, 0xcccc834f, 0x3434685c, 0xa5a551f4,
|
|
|
|
0xe5e5d134, 0xf1f1f908, 0x7171e293, 0xd8d8ab73, 0x31316253, 0x15152a3f,
|
|
|
|
0x0404080c, 0xc7c79552, 0x23234665, 0xc3c39d5e, 0x18183028, 0x969637a1,
|
|
|
|
0x05050a0f, 0x9a9a2fb5, 0x07070e09, 0x12122436, 0x80801b9b, 0xe2e2df3d,
|
|
|
|
0xebebcd26, 0x27274e69, 0xb2b27fcd, 0x7575ea9f, 0x0909121b, 0x83831d9e,
|
|
|
|
0x2c2c5874, 0x1a1a342e, 0x1b1b362d, 0x6e6edcb2, 0x5a5ab4ee, 0xa0a05bfb,
|
|
|
|
0x5252a4f6, 0x3b3b764d, 0xd6d6b761, 0xb3b37dce, 0x2929527b, 0xe3e3dd3e,
|
|
|
|
0x2f2f5e71, 0x84841397, 0x5353a6f5, 0xd1d1b968, 0x00000000, 0xededc12c,
|
|
|
|
0x20204060, 0xfcfce31f, 0xb1b179c8, 0x5b5bb6ed, 0x6a6ad4be, 0xcbcb8d46,
|
|
|
|
0xbebe67d9, 0x3939724b, 0x4a4a94de, 0x4c4c98d4, 0x5858b0e8, 0xcfcf854a,
|
|
|
|
0xd0d0bb6b, 0xefefc52a, 0xaaaa4fe5, 0xfbfbed16, 0x434386c5, 0x4d4d9ad7,
|
|
|
|
0x33336655, 0x85851194, 0x45458acf, 0xf9f9e910, 0x02020406, 0x7f7ffe81,
|
|
|
|
0x5050a0f0, 0x3c3c7844, 0x9f9f25ba, 0xa8a84be3, 0x5151a2f3, 0xa3a35dfe,
|
|
|
|
0x404080c0, 0x8f8f058a, 0x92923fad, 0x9d9d21bc, 0x38387048, 0xf5f5f104,
|
|
|
|
0xbcbc63df, 0xb6b677c1, 0xdadaaf75, 0x21214263, 0x10102030, 0xffffe51a,
|
|
|
|
0xf3f3fd0e, 0xd2d2bf6d, 0xcdcd814c, 0x0c0c1814, 0x13132635, 0xececc32f,
|
|
|
|
0x5f5fbee1, 0x979735a2, 0x444488cc, 0x17172e39, 0xc4c49357, 0xa7a755f2,
|
|
|
|
0x7e7efc82, 0x3d3d7a47, 0x6464c8ac, 0x5d5dbae7, 0x1919322b, 0x7373e695,
|
|
|
|
0x6060c0a0, 0x81811998, 0x4f4f9ed1, 0xdcdca37f, 0x22224466, 0x2a2a547e,
|
|
|
|
0x90903bab, 0x88880b83, 0x46468cca, 0xeeeec729, 0xb8b86bd3, 0x1414283c,
|
|
|
|
0xdedea779, 0x5e5ebce2, 0x0b0b161d, 0xdbdbad76, 0xe0e0db3b, 0x32326456,
|
|
|
|
0x3a3a744e, 0x0a0a141e, 0x494992db, 0x06060c0a, 0x2424486c, 0x5c5cb8e4,
|
|
|
|
0xc2c29f5d, 0xd3d3bd6e, 0xacac43ef, 0x6262c4a6, 0x919139a8, 0x959531a4,
|
|
|
|
0xe4e4d337, 0x7979f28b, 0xe7e7d532, 0xc8c88b43, 0x37376e59, 0x6d6ddab7,
|
|
|
|
0x8d8d018c, 0xd5d5b164, 0x4e4e9cd2, 0xa9a949e0, 0x6c6cd8b4, 0x5656acfa,
|
|
|
|
0xf4f4f307, 0xeaeacf25, 0x6565caaf, 0x7a7af48e, 0xaeae47e9, 0x08081018,
|
|
|
|
0xbaba6fd5, 0x7878f088, 0x25254a6f, 0x2e2e5c72, 0x1c1c3824, 0xa6a657f1,
|
|
|
|
0xb4b473c7, 0xc6c69751, 0xe8e8cb23, 0xdddda17c, 0x7474e89c, 0x1f1f3e21,
|
|
|
|
0x4b4b96dd, 0xbdbd61dc, 0x8b8b0d86, 0x8a8a0f85, 0x7070e090, 0x3e3e7c42,
|
|
|
|
0xb5b571c4, 0x6666ccaa, 0x484890d8, 0x03030605, 0xf6f6f701, 0x0e0e1c12,
|
|
|
|
0x6161c2a3, 0x35356a5f, 0x5757aef9, 0xb9b969d0, 0x86861791, 0xc1c19958,
|
|
|
|
0x1d1d3a27, 0x9e9e27b9, 0xe1e1d938, 0xf8f8eb13, 0x98982bb3, 0x11112233,
|
|
|
|
0x6969d2bb, 0xd9d9a970, 0x8e8e0789, 0x949433a7, 0x9b9b2db6, 0x1e1e3c22,
|
|
|
|
0x87871592, 0xe9e9c920, 0xcece8749, 0x5555aaff, 0x28285078, 0xdfdfa57a,
|
|
|
|
0x8c8c038f, 0xa1a159f8, 0x89890980, 0x0d0d1a17, 0xbfbf65da, 0xe6e6d731,
|
|
|
|
0x424284c6, 0x6868d0b8, 0x414182c3, 0x999929b0, 0x2d2d5a77, 0x0f0f1e11,
|
|
|
|
0xb0b07bcb, 0x5454a8fc, 0xbbbb6dd6, 0x16162c3a,
|
|
|
|
};
|
|
|
|
|
|
|
|
constexpr uint32_t te2[256] = {
|
Export of internal Abseil changes
--
007ce045d5d38a727ededdb5bf06e64785fd73bd by Martijn Vels <mvels@google.com>:
Add `cord_enable_btree` feature flag (default false).
PiperOrigin-RevId: 383729939
--
98e7dc6a0407b0fd7b8713d883cdb3a766e0583d by Benjamin Barenblat <bbaren@google.com>:
Eliminate some byte swapping from randen_slow
Stop swapping bytes when serializing randen_slow’s Vector128 into and
out of memory. Instead, simply index different bytes in the AES round
function. This requires byte swapping the te{0..3} lookup tables, but it
produces an 8% speedup on my Xeon W-2135.
PiperOrigin-RevId: 383689402
--
180b6bf45049188840d439b16a28e6b968669340 by Evan Brown <ezb@google.com>:
Minor simplification in drop_deletes_without_resize() - save probe_offset outside the lambda.
Also, add some consts, avoid an auto, and use lambda capture by value instead of reference.
I realized that the compiler can already optimize this - https://godbolt.org/z/Wxd9c4TfK, but I think this way makes the code a bit clearer.
PiperOrigin-RevId: 383646658
--
781706a974c4dc1c0abbb6b801fca0550229e883 by Martijn Vels <mvels@google.com>:
Change storage to contain 3 bytes.
As per the comments in the code, this allows us to utilize all available space in CordRep that may otherwise be 'lost' in padding in derived clases. For the upcoming CordrepBtree class, we want a strong guarantee on having a 64 bytes aligned implementation.
PiperOrigin-RevId: 383633963
--
8fe22ecf92492fa6649938a2215934ebfe01c714 by Derek Mauro <dmauro@google.com>:
Remove reference to str_format_arg.h, which no longer exists
PiperOrigin-RevId: 383517865
--
79397f3b18f18c1e2d7aea993b687329d626ce64 by Benjamin Barenblat <bbaren@google.com>:
Use absl::uint128 for AES random number generator
Replace randen’s internal 128-bit integer struct, u64x2, with
absl::uint128. This eliminates some code and improves support for
big-endian platforms.
PiperOrigin-RevId: 383475671
GitOrigin-RevId: 007ce045d5d38a727ededdb5bf06e64785fd73bd
Change-Id: Ia9d9c40de557221f1744fb0d6d4d6ca7ac569070
3 years ago
|
|
|
0x63c6a563, 0x7cf8847c, 0x77ee9977, 0x7bf68d7b, 0xf2ff0df2, 0x6bd6bd6b,
|
|
|
|
0x6fdeb16f, 0xc59154c5, 0x30605030, 0x01020301, 0x67cea967, 0x2b567d2b,
|
|
|
|
0xfee719fe, 0xd7b562d7, 0xab4de6ab, 0x76ec9a76, 0xca8f45ca, 0x821f9d82,
|
|
|
|
0xc98940c9, 0x7dfa877d, 0xfaef15fa, 0x59b2eb59, 0x478ec947, 0xf0fb0bf0,
|
|
|
|
0xad41ecad, 0xd4b367d4, 0xa25ffda2, 0xaf45eaaf, 0x9c23bf9c, 0xa453f7a4,
|
|
|
|
0x72e49672, 0xc09b5bc0, 0xb775c2b7, 0xfde11cfd, 0x933dae93, 0x264c6a26,
|
|
|
|
0x366c5a36, 0x3f7e413f, 0xf7f502f7, 0xcc834fcc, 0x34685c34, 0xa551f4a5,
|
|
|
|
0xe5d134e5, 0xf1f908f1, 0x71e29371, 0xd8ab73d8, 0x31625331, 0x152a3f15,
|
|
|
|
0x04080c04, 0xc79552c7, 0x23466523, 0xc39d5ec3, 0x18302818, 0x9637a196,
|
|
|
|
0x050a0f05, 0x9a2fb59a, 0x070e0907, 0x12243612, 0x801b9b80, 0xe2df3de2,
|
|
|
|
0xebcd26eb, 0x274e6927, 0xb27fcdb2, 0x75ea9f75, 0x09121b09, 0x831d9e83,
|
|
|
|
0x2c58742c, 0x1a342e1a, 0x1b362d1b, 0x6edcb26e, 0x5ab4ee5a, 0xa05bfba0,
|
|
|
|
0x52a4f652, 0x3b764d3b, 0xd6b761d6, 0xb37dceb3, 0x29527b29, 0xe3dd3ee3,
|
|
|
|
0x2f5e712f, 0x84139784, 0x53a6f553, 0xd1b968d1, 0x00000000, 0xedc12ced,
|
|
|
|
0x20406020, 0xfce31ffc, 0xb179c8b1, 0x5bb6ed5b, 0x6ad4be6a, 0xcb8d46cb,
|
|
|
|
0xbe67d9be, 0x39724b39, 0x4a94de4a, 0x4c98d44c, 0x58b0e858, 0xcf854acf,
|
|
|
|
0xd0bb6bd0, 0xefc52aef, 0xaa4fe5aa, 0xfbed16fb, 0x4386c543, 0x4d9ad74d,
|
|
|
|
0x33665533, 0x85119485, 0x458acf45, 0xf9e910f9, 0x02040602, 0x7ffe817f,
|
|
|
|
0x50a0f050, 0x3c78443c, 0x9f25ba9f, 0xa84be3a8, 0x51a2f351, 0xa35dfea3,
|
|
|
|
0x4080c040, 0x8f058a8f, 0x923fad92, 0x9d21bc9d, 0x38704838, 0xf5f104f5,
|
|
|
|
0xbc63dfbc, 0xb677c1b6, 0xdaaf75da, 0x21426321, 0x10203010, 0xffe51aff,
|
|
|
|
0xf3fd0ef3, 0xd2bf6dd2, 0xcd814ccd, 0x0c18140c, 0x13263513, 0xecc32fec,
|
|
|
|
0x5fbee15f, 0x9735a297, 0x4488cc44, 0x172e3917, 0xc49357c4, 0xa755f2a7,
|
|
|
|
0x7efc827e, 0x3d7a473d, 0x64c8ac64, 0x5dbae75d, 0x19322b19, 0x73e69573,
|
|
|
|
0x60c0a060, 0x81199881, 0x4f9ed14f, 0xdca37fdc, 0x22446622, 0x2a547e2a,
|
|
|
|
0x903bab90, 0x880b8388, 0x468cca46, 0xeec729ee, 0xb86bd3b8, 0x14283c14,
|
|
|
|
0xdea779de, 0x5ebce25e, 0x0b161d0b, 0xdbad76db, 0xe0db3be0, 0x32645632,
|
|
|
|
0x3a744e3a, 0x0a141e0a, 0x4992db49, 0x060c0a06, 0x24486c24, 0x5cb8e45c,
|
|
|
|
0xc29f5dc2, 0xd3bd6ed3, 0xac43efac, 0x62c4a662, 0x9139a891, 0x9531a495,
|
|
|
|
0xe4d337e4, 0x79f28b79, 0xe7d532e7, 0xc88b43c8, 0x376e5937, 0x6ddab76d,
|
|
|
|
0x8d018c8d, 0xd5b164d5, 0x4e9cd24e, 0xa949e0a9, 0x6cd8b46c, 0x56acfa56,
|
|
|
|
0xf4f307f4, 0xeacf25ea, 0x65caaf65, 0x7af48e7a, 0xae47e9ae, 0x08101808,
|
|
|
|
0xba6fd5ba, 0x78f08878, 0x254a6f25, 0x2e5c722e, 0x1c38241c, 0xa657f1a6,
|
|
|
|
0xb473c7b4, 0xc69751c6, 0xe8cb23e8, 0xdda17cdd, 0x74e89c74, 0x1f3e211f,
|
|
|
|
0x4b96dd4b, 0xbd61dcbd, 0x8b0d868b, 0x8a0f858a, 0x70e09070, 0x3e7c423e,
|
|
|
|
0xb571c4b5, 0x66ccaa66, 0x4890d848, 0x03060503, 0xf6f701f6, 0x0e1c120e,
|
|
|
|
0x61c2a361, 0x356a5f35, 0x57aef957, 0xb969d0b9, 0x86179186, 0xc19958c1,
|
|
|
|
0x1d3a271d, 0x9e27b99e, 0xe1d938e1, 0xf8eb13f8, 0x982bb398, 0x11223311,
|
|
|
|
0x69d2bb69, 0xd9a970d9, 0x8e07898e, 0x9433a794, 0x9b2db69b, 0x1e3c221e,
|
|
|
|
0x87159287, 0xe9c920e9, 0xce8749ce, 0x55aaff55, 0x28507828, 0xdfa57adf,
|
|
|
|
0x8c038f8c, 0xa159f8a1, 0x89098089, 0x0d1a170d, 0xbf65dabf, 0xe6d731e6,
|
|
|
|
0x4284c642, 0x68d0b868, 0x4182c341, 0x9929b099, 0x2d5a772d, 0x0f1e110f,
|
|
|
|
0xb07bcbb0, 0x54a8fc54, 0xbb6dd6bb, 0x162c3a16,
|
|
|
|
};
|
|
|
|
|
|
|
|
constexpr uint32_t te3[256] = {
|
Export of internal Abseil changes
--
007ce045d5d38a727ededdb5bf06e64785fd73bd by Martijn Vels <mvels@google.com>:
Add `cord_enable_btree` feature flag (default false).
PiperOrigin-RevId: 383729939
--
98e7dc6a0407b0fd7b8713d883cdb3a766e0583d by Benjamin Barenblat <bbaren@google.com>:
Eliminate some byte swapping from randen_slow
Stop swapping bytes when serializing randen_slow’s Vector128 into and
out of memory. Instead, simply index different bytes in the AES round
function. This requires byte swapping the te{0..3} lookup tables, but it
produces an 8% speedup on my Xeon W-2135.
PiperOrigin-RevId: 383689402
--
180b6bf45049188840d439b16a28e6b968669340 by Evan Brown <ezb@google.com>:
Minor simplification in drop_deletes_without_resize() - save probe_offset outside the lambda.
Also, add some consts, avoid an auto, and use lambda capture by value instead of reference.
I realized that the compiler can already optimize this - https://godbolt.org/z/Wxd9c4TfK, but I think this way makes the code a bit clearer.
PiperOrigin-RevId: 383646658
--
781706a974c4dc1c0abbb6b801fca0550229e883 by Martijn Vels <mvels@google.com>:
Change storage to contain 3 bytes.
As per the comments in the code, this allows us to utilize all available space in CordRep that may otherwise be 'lost' in padding in derived clases. For the upcoming CordrepBtree class, we want a strong guarantee on having a 64 bytes aligned implementation.
PiperOrigin-RevId: 383633963
--
8fe22ecf92492fa6649938a2215934ebfe01c714 by Derek Mauro <dmauro@google.com>:
Remove reference to str_format_arg.h, which no longer exists
PiperOrigin-RevId: 383517865
--
79397f3b18f18c1e2d7aea993b687329d626ce64 by Benjamin Barenblat <bbaren@google.com>:
Use absl::uint128 for AES random number generator
Replace randen’s internal 128-bit integer struct, u64x2, with
absl::uint128. This eliminates some code and improves support for
big-endian platforms.
PiperOrigin-RevId: 383475671
GitOrigin-RevId: 007ce045d5d38a727ededdb5bf06e64785fd73bd
Change-Id: Ia9d9c40de557221f1744fb0d6d4d6ca7ac569070
3 years ago
|
|
|
0xc6a56363, 0xf8847c7c, 0xee997777, 0xf68d7b7b, 0xff0df2f2, 0xd6bd6b6b,
|
|
|
|
0xdeb16f6f, 0x9154c5c5, 0x60503030, 0x02030101, 0xcea96767, 0x567d2b2b,
|
|
|
|
0xe719fefe, 0xb562d7d7, 0x4de6abab, 0xec9a7676, 0x8f45caca, 0x1f9d8282,
|
|
|
|
0x8940c9c9, 0xfa877d7d, 0xef15fafa, 0xb2eb5959, 0x8ec94747, 0xfb0bf0f0,
|
|
|
|
0x41ecadad, 0xb367d4d4, 0x5ffda2a2, 0x45eaafaf, 0x23bf9c9c, 0x53f7a4a4,
|
|
|
|
0xe4967272, 0x9b5bc0c0, 0x75c2b7b7, 0xe11cfdfd, 0x3dae9393, 0x4c6a2626,
|
|
|
|
0x6c5a3636, 0x7e413f3f, 0xf502f7f7, 0x834fcccc, 0x685c3434, 0x51f4a5a5,
|
|
|
|
0xd134e5e5, 0xf908f1f1, 0xe2937171, 0xab73d8d8, 0x62533131, 0x2a3f1515,
|
|
|
|
0x080c0404, 0x9552c7c7, 0x46652323, 0x9d5ec3c3, 0x30281818, 0x37a19696,
|
|
|
|
0x0a0f0505, 0x2fb59a9a, 0x0e090707, 0x24361212, 0x1b9b8080, 0xdf3de2e2,
|
|
|
|
0xcd26ebeb, 0x4e692727, 0x7fcdb2b2, 0xea9f7575, 0x121b0909, 0x1d9e8383,
|
|
|
|
0x58742c2c, 0x342e1a1a, 0x362d1b1b, 0xdcb26e6e, 0xb4ee5a5a, 0x5bfba0a0,
|
|
|
|
0xa4f65252, 0x764d3b3b, 0xb761d6d6, 0x7dceb3b3, 0x527b2929, 0xdd3ee3e3,
|
|
|
|
0x5e712f2f, 0x13978484, 0xa6f55353, 0xb968d1d1, 0x00000000, 0xc12ceded,
|
|
|
|
0x40602020, 0xe31ffcfc, 0x79c8b1b1, 0xb6ed5b5b, 0xd4be6a6a, 0x8d46cbcb,
|
|
|
|
0x67d9bebe, 0x724b3939, 0x94de4a4a, 0x98d44c4c, 0xb0e85858, 0x854acfcf,
|
|
|
|
0xbb6bd0d0, 0xc52aefef, 0x4fe5aaaa, 0xed16fbfb, 0x86c54343, 0x9ad74d4d,
|
|
|
|
0x66553333, 0x11948585, 0x8acf4545, 0xe910f9f9, 0x04060202, 0xfe817f7f,
|
|
|
|
0xa0f05050, 0x78443c3c, 0x25ba9f9f, 0x4be3a8a8, 0xa2f35151, 0x5dfea3a3,
|
|
|
|
0x80c04040, 0x058a8f8f, 0x3fad9292, 0x21bc9d9d, 0x70483838, 0xf104f5f5,
|
|
|
|
0x63dfbcbc, 0x77c1b6b6, 0xaf75dada, 0x42632121, 0x20301010, 0xe51affff,
|
|
|
|
0xfd0ef3f3, 0xbf6dd2d2, 0x814ccdcd, 0x18140c0c, 0x26351313, 0xc32fecec,
|
|
|
|
0xbee15f5f, 0x35a29797, 0x88cc4444, 0x2e391717, 0x9357c4c4, 0x55f2a7a7,
|
|
|
|
0xfc827e7e, 0x7a473d3d, 0xc8ac6464, 0xbae75d5d, 0x322b1919, 0xe6957373,
|
|
|
|
0xc0a06060, 0x19988181, 0x9ed14f4f, 0xa37fdcdc, 0x44662222, 0x547e2a2a,
|
|
|
|
0x3bab9090, 0x0b838888, 0x8cca4646, 0xc729eeee, 0x6bd3b8b8, 0x283c1414,
|
|
|
|
0xa779dede, 0xbce25e5e, 0x161d0b0b, 0xad76dbdb, 0xdb3be0e0, 0x64563232,
|
|
|
|
0x744e3a3a, 0x141e0a0a, 0x92db4949, 0x0c0a0606, 0x486c2424, 0xb8e45c5c,
|
|
|
|
0x9f5dc2c2, 0xbd6ed3d3, 0x43efacac, 0xc4a66262, 0x39a89191, 0x31a49595,
|
|
|
|
0xd337e4e4, 0xf28b7979, 0xd532e7e7, 0x8b43c8c8, 0x6e593737, 0xdab76d6d,
|
|
|
|
0x018c8d8d, 0xb164d5d5, 0x9cd24e4e, 0x49e0a9a9, 0xd8b46c6c, 0xacfa5656,
|
|
|
|
0xf307f4f4, 0xcf25eaea, 0xcaaf6565, 0xf48e7a7a, 0x47e9aeae, 0x10180808,
|
|
|
|
0x6fd5baba, 0xf0887878, 0x4a6f2525, 0x5c722e2e, 0x38241c1c, 0x57f1a6a6,
|
|
|
|
0x73c7b4b4, 0x9751c6c6, 0xcb23e8e8, 0xa17cdddd, 0xe89c7474, 0x3e211f1f,
|
|
|
|
0x96dd4b4b, 0x61dcbdbd, 0x0d868b8b, 0x0f858a8a, 0xe0907070, 0x7c423e3e,
|
|
|
|
0x71c4b5b5, 0xccaa6666, 0x90d84848, 0x06050303, 0xf701f6f6, 0x1c120e0e,
|
|
|
|
0xc2a36161, 0x6a5f3535, 0xaef95757, 0x69d0b9b9, 0x17918686, 0x9958c1c1,
|
|
|
|
0x3a271d1d, 0x27b99e9e, 0xd938e1e1, 0xeb13f8f8, 0x2bb39898, 0x22331111,
|
|
|
|
0xd2bb6969, 0xa970d9d9, 0x07898e8e, 0x33a79494, 0x2db69b9b, 0x3c221e1e,
|
|
|
|
0x15928787, 0xc920e9e9, 0x8749cece, 0xaaff5555, 0x50782828, 0xa57adfdf,
|
|
|
|
0x038f8c8c, 0x59f8a1a1, 0x09808989, 0x1a170d0d, 0x65dabfbf, 0xd731e6e6,
|
|
|
|
0x84c64242, 0xd0b86868, 0x82c34141, 0x29b09999, 0x5a772d2d, 0x1e110f0f,
|
|
|
|
0x7bcbb0b0, 0xa8fc5454, 0x6dd6bbbb, 0x2c3a1616,
|
|
|
|
};
|
|
|
|
|
|
|
|
// Software implementation of the Vector128 class, using uint32_t
|
|
|
|
// as an underlying vector register.
|
|
|
|
struct alignas(16) Vector128 {
|
|
|
|
uint32_t s[4];
|
|
|
|
};
|
|
|
|
|
|
|
|
inline ABSL_RANDOM_INTERNAL_ATTRIBUTE_ALWAYS_INLINE Vector128
|
|
|
|
Vector128Load(const void* from) {
|
|
|
|
Vector128 result;
|
Export of internal Abseil changes
--
007ce045d5d38a727ededdb5bf06e64785fd73bd by Martijn Vels <mvels@google.com>:
Add `cord_enable_btree` feature flag (default false).
PiperOrigin-RevId: 383729939
--
98e7dc6a0407b0fd7b8713d883cdb3a766e0583d by Benjamin Barenblat <bbaren@google.com>:
Eliminate some byte swapping from randen_slow
Stop swapping bytes when serializing randen_slow’s Vector128 into and
out of memory. Instead, simply index different bytes in the AES round
function. This requires byte swapping the te{0..3} lookup tables, but it
produces an 8% speedup on my Xeon W-2135.
PiperOrigin-RevId: 383689402
--
180b6bf45049188840d439b16a28e6b968669340 by Evan Brown <ezb@google.com>:
Minor simplification in drop_deletes_without_resize() - save probe_offset outside the lambda.
Also, add some consts, avoid an auto, and use lambda capture by value instead of reference.
I realized that the compiler can already optimize this - https://godbolt.org/z/Wxd9c4TfK, but I think this way makes the code a bit clearer.
PiperOrigin-RevId: 383646658
--
781706a974c4dc1c0abbb6b801fca0550229e883 by Martijn Vels <mvels@google.com>:
Change storage to contain 3 bytes.
As per the comments in the code, this allows us to utilize all available space in CordRep that may otherwise be 'lost' in padding in derived clases. For the upcoming CordrepBtree class, we want a strong guarantee on having a 64 bytes aligned implementation.
PiperOrigin-RevId: 383633963
--
8fe22ecf92492fa6649938a2215934ebfe01c714 by Derek Mauro <dmauro@google.com>:
Remove reference to str_format_arg.h, which no longer exists
PiperOrigin-RevId: 383517865
--
79397f3b18f18c1e2d7aea993b687329d626ce64 by Benjamin Barenblat <bbaren@google.com>:
Use absl::uint128 for AES random number generator
Replace randen’s internal 128-bit integer struct, u64x2, with
absl::uint128. This eliminates some code and improves support for
big-endian platforms.
PiperOrigin-RevId: 383475671
GitOrigin-RevId: 007ce045d5d38a727ededdb5bf06e64785fd73bd
Change-Id: Ia9d9c40de557221f1744fb0d6d4d6ca7ac569070
3 years ago
|
|
|
std::memcpy(result.s, from, sizeof(Vector128));
|
|
|
|
return result;
|
|
|
|
}
|
|
|
|
|
|
|
|
inline ABSL_RANDOM_INTERNAL_ATTRIBUTE_ALWAYS_INLINE void Vector128Store(
|
|
|
|
const Vector128& v, void* to) {
|
Export of internal Abseil changes
--
007ce045d5d38a727ededdb5bf06e64785fd73bd by Martijn Vels <mvels@google.com>:
Add `cord_enable_btree` feature flag (default false).
PiperOrigin-RevId: 383729939
--
98e7dc6a0407b0fd7b8713d883cdb3a766e0583d by Benjamin Barenblat <bbaren@google.com>:
Eliminate some byte swapping from randen_slow
Stop swapping bytes when serializing randen_slow’s Vector128 into and
out of memory. Instead, simply index different bytes in the AES round
function. This requires byte swapping the te{0..3} lookup tables, but it
produces an 8% speedup on my Xeon W-2135.
PiperOrigin-RevId: 383689402
--
180b6bf45049188840d439b16a28e6b968669340 by Evan Brown <ezb@google.com>:
Minor simplification in drop_deletes_without_resize() - save probe_offset outside the lambda.
Also, add some consts, avoid an auto, and use lambda capture by value instead of reference.
I realized that the compiler can already optimize this - https://godbolt.org/z/Wxd9c4TfK, but I think this way makes the code a bit clearer.
PiperOrigin-RevId: 383646658
--
781706a974c4dc1c0abbb6b801fca0550229e883 by Martijn Vels <mvels@google.com>:
Change storage to contain 3 bytes.
As per the comments in the code, this allows us to utilize all available space in CordRep that may otherwise be 'lost' in padding in derived clases. For the upcoming CordrepBtree class, we want a strong guarantee on having a 64 bytes aligned implementation.
PiperOrigin-RevId: 383633963
--
8fe22ecf92492fa6649938a2215934ebfe01c714 by Derek Mauro <dmauro@google.com>:
Remove reference to str_format_arg.h, which no longer exists
PiperOrigin-RevId: 383517865
--
79397f3b18f18c1e2d7aea993b687329d626ce64 by Benjamin Barenblat <bbaren@google.com>:
Use absl::uint128 for AES random number generator
Replace randen’s internal 128-bit integer struct, u64x2, with
absl::uint128. This eliminates some code and improves support for
big-endian platforms.
PiperOrigin-RevId: 383475671
GitOrigin-RevId: 007ce045d5d38a727ededdb5bf06e64785fd73bd
Change-Id: Ia9d9c40de557221f1744fb0d6d4d6ca7ac569070
3 years ago
|
|
|
std::memcpy(to, v.s, sizeof(Vector128));
|
|
|
|
}
|
|
|
|
|
|
|
|
// One round of AES. "round_key" is a public constant for breaking the
|
|
|
|
// symmetry of AES (ensures previously equal columns differ afterwards).
|
|
|
|
inline ABSL_RANDOM_INTERNAL_ATTRIBUTE_ALWAYS_INLINE Vector128
|
|
|
|
AesRound(const Vector128& state, const Vector128& round_key) {
|
|
|
|
Vector128 result;
|
|
|
|
#ifdef ABSL_IS_LITTLE_ENDIAN
|
|
|
|
result.s[0] = round_key.s[0] ^ //
|
Export of internal Abseil changes
--
007ce045d5d38a727ededdb5bf06e64785fd73bd by Martijn Vels <mvels@google.com>:
Add `cord_enable_btree` feature flag (default false).
PiperOrigin-RevId: 383729939
--
98e7dc6a0407b0fd7b8713d883cdb3a766e0583d by Benjamin Barenblat <bbaren@google.com>:
Eliminate some byte swapping from randen_slow
Stop swapping bytes when serializing randen_slow’s Vector128 into and
out of memory. Instead, simply index different bytes in the AES round
function. This requires byte swapping the te{0..3} lookup tables, but it
produces an 8% speedup on my Xeon W-2135.
PiperOrigin-RevId: 383689402
--
180b6bf45049188840d439b16a28e6b968669340 by Evan Brown <ezb@google.com>:
Minor simplification in drop_deletes_without_resize() - save probe_offset outside the lambda.
Also, add some consts, avoid an auto, and use lambda capture by value instead of reference.
I realized that the compiler can already optimize this - https://godbolt.org/z/Wxd9c4TfK, but I think this way makes the code a bit clearer.
PiperOrigin-RevId: 383646658
--
781706a974c4dc1c0abbb6b801fca0550229e883 by Martijn Vels <mvels@google.com>:
Change storage to contain 3 bytes.
As per the comments in the code, this allows us to utilize all available space in CordRep that may otherwise be 'lost' in padding in derived clases. For the upcoming CordrepBtree class, we want a strong guarantee on having a 64 bytes aligned implementation.
PiperOrigin-RevId: 383633963
--
8fe22ecf92492fa6649938a2215934ebfe01c714 by Derek Mauro <dmauro@google.com>:
Remove reference to str_format_arg.h, which no longer exists
PiperOrigin-RevId: 383517865
--
79397f3b18f18c1e2d7aea993b687329d626ce64 by Benjamin Barenblat <bbaren@google.com>:
Use absl::uint128 for AES random number generator
Replace randen’s internal 128-bit integer struct, u64x2, with
absl::uint128. This eliminates some code and improves support for
big-endian platforms.
PiperOrigin-RevId: 383475671
GitOrigin-RevId: 007ce045d5d38a727ededdb5bf06e64785fd73bd
Change-Id: Ia9d9c40de557221f1744fb0d6d4d6ca7ac569070
3 years ago
|
|
|
te0[uint8_t(state.s[0])] ^ //
|
|
|
|
te1[uint8_t(state.s[1] >> 8)] ^ //
|
|
|
|
te2[uint8_t(state.s[2] >> 16)] ^ //
|
|
|
|
te3[uint8_t(state.s[3] >> 24)];
|
|
|
|
result.s[1] = round_key.s[1] ^ //
|
Export of internal Abseil changes
--
007ce045d5d38a727ededdb5bf06e64785fd73bd by Martijn Vels <mvels@google.com>:
Add `cord_enable_btree` feature flag (default false).
PiperOrigin-RevId: 383729939
--
98e7dc6a0407b0fd7b8713d883cdb3a766e0583d by Benjamin Barenblat <bbaren@google.com>:
Eliminate some byte swapping from randen_slow
Stop swapping bytes when serializing randen_slow’s Vector128 into and
out of memory. Instead, simply index different bytes in the AES round
function. This requires byte swapping the te{0..3} lookup tables, but it
produces an 8% speedup on my Xeon W-2135.
PiperOrigin-RevId: 383689402
--
180b6bf45049188840d439b16a28e6b968669340 by Evan Brown <ezb@google.com>:
Minor simplification in drop_deletes_without_resize() - save probe_offset outside the lambda.
Also, add some consts, avoid an auto, and use lambda capture by value instead of reference.
I realized that the compiler can already optimize this - https://godbolt.org/z/Wxd9c4TfK, but I think this way makes the code a bit clearer.
PiperOrigin-RevId: 383646658
--
781706a974c4dc1c0abbb6b801fca0550229e883 by Martijn Vels <mvels@google.com>:
Change storage to contain 3 bytes.
As per the comments in the code, this allows us to utilize all available space in CordRep that may otherwise be 'lost' in padding in derived clases. For the upcoming CordrepBtree class, we want a strong guarantee on having a 64 bytes aligned implementation.
PiperOrigin-RevId: 383633963
--
8fe22ecf92492fa6649938a2215934ebfe01c714 by Derek Mauro <dmauro@google.com>:
Remove reference to str_format_arg.h, which no longer exists
PiperOrigin-RevId: 383517865
--
79397f3b18f18c1e2d7aea993b687329d626ce64 by Benjamin Barenblat <bbaren@google.com>:
Use absl::uint128 for AES random number generator
Replace randen’s internal 128-bit integer struct, u64x2, with
absl::uint128. This eliminates some code and improves support for
big-endian platforms.
PiperOrigin-RevId: 383475671
GitOrigin-RevId: 007ce045d5d38a727ededdb5bf06e64785fd73bd
Change-Id: Ia9d9c40de557221f1744fb0d6d4d6ca7ac569070
3 years ago
|
|
|
te0[uint8_t(state.s[1])] ^ //
|
|
|
|
te1[uint8_t(state.s[2] >> 8)] ^ //
|
|
|
|
te2[uint8_t(state.s[3] >> 16)] ^ //
|
|
|
|
te3[uint8_t(state.s[0] >> 24)];
|
|
|
|
result.s[2] = round_key.s[2] ^ //
|
Export of internal Abseil changes
--
007ce045d5d38a727ededdb5bf06e64785fd73bd by Martijn Vels <mvels@google.com>:
Add `cord_enable_btree` feature flag (default false).
PiperOrigin-RevId: 383729939
--
98e7dc6a0407b0fd7b8713d883cdb3a766e0583d by Benjamin Barenblat <bbaren@google.com>:
Eliminate some byte swapping from randen_slow
Stop swapping bytes when serializing randen_slow’s Vector128 into and
out of memory. Instead, simply index different bytes in the AES round
function. This requires byte swapping the te{0..3} lookup tables, but it
produces an 8% speedup on my Xeon W-2135.
PiperOrigin-RevId: 383689402
--
180b6bf45049188840d439b16a28e6b968669340 by Evan Brown <ezb@google.com>:
Minor simplification in drop_deletes_without_resize() - save probe_offset outside the lambda.
Also, add some consts, avoid an auto, and use lambda capture by value instead of reference.
I realized that the compiler can already optimize this - https://godbolt.org/z/Wxd9c4TfK, but I think this way makes the code a bit clearer.
PiperOrigin-RevId: 383646658
--
781706a974c4dc1c0abbb6b801fca0550229e883 by Martijn Vels <mvels@google.com>:
Change storage to contain 3 bytes.
As per the comments in the code, this allows us to utilize all available space in CordRep that may otherwise be 'lost' in padding in derived clases. For the upcoming CordrepBtree class, we want a strong guarantee on having a 64 bytes aligned implementation.
PiperOrigin-RevId: 383633963
--
8fe22ecf92492fa6649938a2215934ebfe01c714 by Derek Mauro <dmauro@google.com>:
Remove reference to str_format_arg.h, which no longer exists
PiperOrigin-RevId: 383517865
--
79397f3b18f18c1e2d7aea993b687329d626ce64 by Benjamin Barenblat <bbaren@google.com>:
Use absl::uint128 for AES random number generator
Replace randen’s internal 128-bit integer struct, u64x2, with
absl::uint128. This eliminates some code and improves support for
big-endian platforms.
PiperOrigin-RevId: 383475671
GitOrigin-RevId: 007ce045d5d38a727ededdb5bf06e64785fd73bd
Change-Id: Ia9d9c40de557221f1744fb0d6d4d6ca7ac569070
3 years ago
|
|
|
te0[uint8_t(state.s[2])] ^ //
|
|
|
|
te1[uint8_t(state.s[3] >> 8)] ^ //
|
|
|
|
te2[uint8_t(state.s[0] >> 16)] ^ //
|
|
|
|
te3[uint8_t(state.s[1] >> 24)];
|
|
|
|
result.s[3] = round_key.s[3] ^ //
|
Export of internal Abseil changes
--
007ce045d5d38a727ededdb5bf06e64785fd73bd by Martijn Vels <mvels@google.com>:
Add `cord_enable_btree` feature flag (default false).
PiperOrigin-RevId: 383729939
--
98e7dc6a0407b0fd7b8713d883cdb3a766e0583d by Benjamin Barenblat <bbaren@google.com>:
Eliminate some byte swapping from randen_slow
Stop swapping bytes when serializing randen_slow’s Vector128 into and
out of memory. Instead, simply index different bytes in the AES round
function. This requires byte swapping the te{0..3} lookup tables, but it
produces an 8% speedup on my Xeon W-2135.
PiperOrigin-RevId: 383689402
--
180b6bf45049188840d439b16a28e6b968669340 by Evan Brown <ezb@google.com>:
Minor simplification in drop_deletes_without_resize() - save probe_offset outside the lambda.
Also, add some consts, avoid an auto, and use lambda capture by value instead of reference.
I realized that the compiler can already optimize this - https://godbolt.org/z/Wxd9c4TfK, but I think this way makes the code a bit clearer.
PiperOrigin-RevId: 383646658
--
781706a974c4dc1c0abbb6b801fca0550229e883 by Martijn Vels <mvels@google.com>:
Change storage to contain 3 bytes.
As per the comments in the code, this allows us to utilize all available space in CordRep that may otherwise be 'lost' in padding in derived clases. For the upcoming CordrepBtree class, we want a strong guarantee on having a 64 bytes aligned implementation.
PiperOrigin-RevId: 383633963
--
8fe22ecf92492fa6649938a2215934ebfe01c714 by Derek Mauro <dmauro@google.com>:
Remove reference to str_format_arg.h, which no longer exists
PiperOrigin-RevId: 383517865
--
79397f3b18f18c1e2d7aea993b687329d626ce64 by Benjamin Barenblat <bbaren@google.com>:
Use absl::uint128 for AES random number generator
Replace randen’s internal 128-bit integer struct, u64x2, with
absl::uint128. This eliminates some code and improves support for
big-endian platforms.
PiperOrigin-RevId: 383475671
GitOrigin-RevId: 007ce045d5d38a727ededdb5bf06e64785fd73bd
Change-Id: Ia9d9c40de557221f1744fb0d6d4d6ca7ac569070
3 years ago
|
|
|
te0[uint8_t(state.s[3])] ^ //
|
|
|
|
te1[uint8_t(state.s[0] >> 8)] ^ //
|
|
|
|
te2[uint8_t(state.s[1] >> 16)] ^ //
|
|
|
|
te3[uint8_t(state.s[2] >> 24)];
|
|
|
|
#else
|
|
|
|
result.s[0] = round_key.s[0] ^ //
|
|
|
|
te0[uint8_t(state.s[0])] ^ //
|
|
|
|
te1[uint8_t(state.s[3] >> 8)] ^ //
|
|
|
|
te2[uint8_t(state.s[2] >> 16)] ^ //
|
|
|
|
te3[uint8_t(state.s[1] >> 24)];
|
|
|
|
result.s[1] = round_key.s[1] ^ //
|
|
|
|
te0[uint8_t(state.s[1])] ^ //
|
|
|
|
te1[uint8_t(state.s[0] >> 8)] ^ //
|
|
|
|
te2[uint8_t(state.s[3] >> 16)] ^ //
|
|
|
|
te3[uint8_t(state.s[2] >> 24)];
|
|
|
|
result.s[2] = round_key.s[2] ^ //
|
|
|
|
te0[uint8_t(state.s[2])] ^ //
|
|
|
|
te1[uint8_t(state.s[1] >> 8)] ^ //
|
|
|
|
te2[uint8_t(state.s[0] >> 16)] ^ //
|
|
|
|
te3[uint8_t(state.s[3] >> 24)];
|
|
|
|
result.s[3] = round_key.s[3] ^ //
|
|
|
|
te0[uint8_t(state.s[3])] ^ //
|
|
|
|
te1[uint8_t(state.s[2] >> 8)] ^ //
|
|
|
|
te2[uint8_t(state.s[1] >> 16)] ^ //
|
|
|
|
te3[uint8_t(state.s[0] >> 24)];
|
|
|
|
#endif
|
|
|
|
return result;
|
|
|
|
}
|
|
|
|
|
|
|
|
using ::absl::random_internal::RandenTraits;
|
|
|
|
|
|
|
|
// The improved Feistel block shuffle function for 16 blocks.
|
|
|
|
inline ABSL_RANDOM_INTERNAL_ATTRIBUTE_ALWAYS_INLINE void BlockShuffle(
|
Export of internal Abseil changes
--
007ce045d5d38a727ededdb5bf06e64785fd73bd by Martijn Vels <mvels@google.com>:
Add `cord_enable_btree` feature flag (default false).
PiperOrigin-RevId: 383729939
--
98e7dc6a0407b0fd7b8713d883cdb3a766e0583d by Benjamin Barenblat <bbaren@google.com>:
Eliminate some byte swapping from randen_slow
Stop swapping bytes when serializing randen_slow’s Vector128 into and
out of memory. Instead, simply index different bytes in the AES round
function. This requires byte swapping the te{0..3} lookup tables, but it
produces an 8% speedup on my Xeon W-2135.
PiperOrigin-RevId: 383689402
--
180b6bf45049188840d439b16a28e6b968669340 by Evan Brown <ezb@google.com>:
Minor simplification in drop_deletes_without_resize() - save probe_offset outside the lambda.
Also, add some consts, avoid an auto, and use lambda capture by value instead of reference.
I realized that the compiler can already optimize this - https://godbolt.org/z/Wxd9c4TfK, but I think this way makes the code a bit clearer.
PiperOrigin-RevId: 383646658
--
781706a974c4dc1c0abbb6b801fca0550229e883 by Martijn Vels <mvels@google.com>:
Change storage to contain 3 bytes.
As per the comments in the code, this allows us to utilize all available space in CordRep that may otherwise be 'lost' in padding in derived clases. For the upcoming CordrepBtree class, we want a strong guarantee on having a 64 bytes aligned implementation.
PiperOrigin-RevId: 383633963
--
8fe22ecf92492fa6649938a2215934ebfe01c714 by Derek Mauro <dmauro@google.com>:
Remove reference to str_format_arg.h, which no longer exists
PiperOrigin-RevId: 383517865
--
79397f3b18f18c1e2d7aea993b687329d626ce64 by Benjamin Barenblat <bbaren@google.com>:
Use absl::uint128 for AES random number generator
Replace randen’s internal 128-bit integer struct, u64x2, with
absl::uint128. This eliminates some code and improves support for
big-endian platforms.
PiperOrigin-RevId: 383475671
GitOrigin-RevId: 007ce045d5d38a727ededdb5bf06e64785fd73bd
Change-Id: Ia9d9c40de557221f1744fb0d6d4d6ca7ac569070
3 years ago
|
|
|
absl::uint128* state) {
|
|
|
|
static_assert(RandenTraits::kFeistelBlocks == 16,
|
|
|
|
"Feistel block shuffle only works for 16 blocks.");
|
|
|
|
|
|
|
|
constexpr size_t shuffle[RandenTraits::kFeistelBlocks] = {
|
|
|
|
7, 2, 13, 4, 11, 8, 3, 6, 15, 0, 9, 10, 1, 14, 5, 12};
|
|
|
|
|
|
|
|
// The fully unrolled loop without the memcpy improves the speed by about
|
|
|
|
// 30% over the equivalent:
|
|
|
|
#if 0
|
Export of internal Abseil changes
--
007ce045d5d38a727ededdb5bf06e64785fd73bd by Martijn Vels <mvels@google.com>:
Add `cord_enable_btree` feature flag (default false).
PiperOrigin-RevId: 383729939
--
98e7dc6a0407b0fd7b8713d883cdb3a766e0583d by Benjamin Barenblat <bbaren@google.com>:
Eliminate some byte swapping from randen_slow
Stop swapping bytes when serializing randen_slow’s Vector128 into and
out of memory. Instead, simply index different bytes in the AES round
function. This requires byte swapping the te{0..3} lookup tables, but it
produces an 8% speedup on my Xeon W-2135.
PiperOrigin-RevId: 383689402
--
180b6bf45049188840d439b16a28e6b968669340 by Evan Brown <ezb@google.com>:
Minor simplification in drop_deletes_without_resize() - save probe_offset outside the lambda.
Also, add some consts, avoid an auto, and use lambda capture by value instead of reference.
I realized that the compiler can already optimize this - https://godbolt.org/z/Wxd9c4TfK, but I think this way makes the code a bit clearer.
PiperOrigin-RevId: 383646658
--
781706a974c4dc1c0abbb6b801fca0550229e883 by Martijn Vels <mvels@google.com>:
Change storage to contain 3 bytes.
As per the comments in the code, this allows us to utilize all available space in CordRep that may otherwise be 'lost' in padding in derived clases. For the upcoming CordrepBtree class, we want a strong guarantee on having a 64 bytes aligned implementation.
PiperOrigin-RevId: 383633963
--
8fe22ecf92492fa6649938a2215934ebfe01c714 by Derek Mauro <dmauro@google.com>:
Remove reference to str_format_arg.h, which no longer exists
PiperOrigin-RevId: 383517865
--
79397f3b18f18c1e2d7aea993b687329d626ce64 by Benjamin Barenblat <bbaren@google.com>:
Use absl::uint128 for AES random number generator
Replace randen’s internal 128-bit integer struct, u64x2, with
absl::uint128. This eliminates some code and improves support for
big-endian platforms.
PiperOrigin-RevId: 383475671
GitOrigin-RevId: 007ce045d5d38a727ededdb5bf06e64785fd73bd
Change-Id: Ia9d9c40de557221f1744fb0d6d4d6ca7ac569070
3 years ago
|
|
|
absl::uint128 source[RandenTraits::kFeistelBlocks];
|
|
|
|
std::memcpy(source, state, sizeof(source));
|
|
|
|
for (size_t i = 0; i < RandenTraits::kFeistelBlocks; i++) {
|
Export of internal Abseil changes
--
007ce045d5d38a727ededdb5bf06e64785fd73bd by Martijn Vels <mvels@google.com>:
Add `cord_enable_btree` feature flag (default false).
PiperOrigin-RevId: 383729939
--
98e7dc6a0407b0fd7b8713d883cdb3a766e0583d by Benjamin Barenblat <bbaren@google.com>:
Eliminate some byte swapping from randen_slow
Stop swapping bytes when serializing randen_slow’s Vector128 into and
out of memory. Instead, simply index different bytes in the AES round
function. This requires byte swapping the te{0..3} lookup tables, but it
produces an 8% speedup on my Xeon W-2135.
PiperOrigin-RevId: 383689402
--
180b6bf45049188840d439b16a28e6b968669340 by Evan Brown <ezb@google.com>:
Minor simplification in drop_deletes_without_resize() - save probe_offset outside the lambda.
Also, add some consts, avoid an auto, and use lambda capture by value instead of reference.
I realized that the compiler can already optimize this - https://godbolt.org/z/Wxd9c4TfK, but I think this way makes the code a bit clearer.
PiperOrigin-RevId: 383646658
--
781706a974c4dc1c0abbb6b801fca0550229e883 by Martijn Vels <mvels@google.com>:
Change storage to contain 3 bytes.
As per the comments in the code, this allows us to utilize all available space in CordRep that may otherwise be 'lost' in padding in derived clases. For the upcoming CordrepBtree class, we want a strong guarantee on having a 64 bytes aligned implementation.
PiperOrigin-RevId: 383633963
--
8fe22ecf92492fa6649938a2215934ebfe01c714 by Derek Mauro <dmauro@google.com>:
Remove reference to str_format_arg.h, which no longer exists
PiperOrigin-RevId: 383517865
--
79397f3b18f18c1e2d7aea993b687329d626ce64 by Benjamin Barenblat <bbaren@google.com>:
Use absl::uint128 for AES random number generator
Replace randen’s internal 128-bit integer struct, u64x2, with
absl::uint128. This eliminates some code and improves support for
big-endian platforms.
PiperOrigin-RevId: 383475671
GitOrigin-RevId: 007ce045d5d38a727ededdb5bf06e64785fd73bd
Change-Id: Ia9d9c40de557221f1744fb0d6d4d6ca7ac569070
3 years ago
|
|
|
const absl::uint128 v0 = source[shuffle[i]];
|
|
|
|
state[i] = v0;
|
|
|
|
}
|
|
|
|
return;
|
|
|
|
#endif
|
|
|
|
|
Export of internal Abseil changes
--
007ce045d5d38a727ededdb5bf06e64785fd73bd by Martijn Vels <mvels@google.com>:
Add `cord_enable_btree` feature flag (default false).
PiperOrigin-RevId: 383729939
--
98e7dc6a0407b0fd7b8713d883cdb3a766e0583d by Benjamin Barenblat <bbaren@google.com>:
Eliminate some byte swapping from randen_slow
Stop swapping bytes when serializing randen_slow’s Vector128 into and
out of memory. Instead, simply index different bytes in the AES round
function. This requires byte swapping the te{0..3} lookup tables, but it
produces an 8% speedup on my Xeon W-2135.
PiperOrigin-RevId: 383689402
--
180b6bf45049188840d439b16a28e6b968669340 by Evan Brown <ezb@google.com>:
Minor simplification in drop_deletes_without_resize() - save probe_offset outside the lambda.
Also, add some consts, avoid an auto, and use lambda capture by value instead of reference.
I realized that the compiler can already optimize this - https://godbolt.org/z/Wxd9c4TfK, but I think this way makes the code a bit clearer.
PiperOrigin-RevId: 383646658
--
781706a974c4dc1c0abbb6b801fca0550229e883 by Martijn Vels <mvels@google.com>:
Change storage to contain 3 bytes.
As per the comments in the code, this allows us to utilize all available space in CordRep that may otherwise be 'lost' in padding in derived clases. For the upcoming CordrepBtree class, we want a strong guarantee on having a 64 bytes aligned implementation.
PiperOrigin-RevId: 383633963
--
8fe22ecf92492fa6649938a2215934ebfe01c714 by Derek Mauro <dmauro@google.com>:
Remove reference to str_format_arg.h, which no longer exists
PiperOrigin-RevId: 383517865
--
79397f3b18f18c1e2d7aea993b687329d626ce64 by Benjamin Barenblat <bbaren@google.com>:
Use absl::uint128 for AES random number generator
Replace randen’s internal 128-bit integer struct, u64x2, with
absl::uint128. This eliminates some code and improves support for
big-endian platforms.
PiperOrigin-RevId: 383475671
GitOrigin-RevId: 007ce045d5d38a727ededdb5bf06e64785fd73bd
Change-Id: Ia9d9c40de557221f1744fb0d6d4d6ca7ac569070
3 years ago
|
|
|
const absl::uint128 v0 = state[shuffle[0]];
|
|
|
|
const absl::uint128 v1 = state[shuffle[1]];
|
|
|
|
const absl::uint128 v2 = state[shuffle[2]];
|
|
|
|
const absl::uint128 v3 = state[shuffle[3]];
|
|
|
|
const absl::uint128 v4 = state[shuffle[4]];
|
|
|
|
const absl::uint128 v5 = state[shuffle[5]];
|
|
|
|
const absl::uint128 v6 = state[shuffle[6]];
|
|
|
|
const absl::uint128 v7 = state[shuffle[7]];
|
|
|
|
const absl::uint128 w0 = state[shuffle[8]];
|
|
|
|
const absl::uint128 w1 = state[shuffle[9]];
|
|
|
|
const absl::uint128 w2 = state[shuffle[10]];
|
|
|
|
const absl::uint128 w3 = state[shuffle[11]];
|
|
|
|
const absl::uint128 w4 = state[shuffle[12]];
|
|
|
|
const absl::uint128 w5 = state[shuffle[13]];
|
|
|
|
const absl::uint128 w6 = state[shuffle[14]];
|
|
|
|
const absl::uint128 w7 = state[shuffle[15]];
|
|
|
|
state[0] = v0;
|
|
|
|
state[1] = v1;
|
|
|
|
state[2] = v2;
|
|
|
|
state[3] = v3;
|
|
|
|
state[4] = v4;
|
|
|
|
state[5] = v5;
|
|
|
|
state[6] = v6;
|
|
|
|
state[7] = v7;
|
|
|
|
state[8] = w0;
|
|
|
|
state[9] = w1;
|
|
|
|
state[10] = w2;
|
|
|
|
state[11] = w3;
|
|
|
|
state[12] = w4;
|
|
|
|
state[13] = w5;
|
|
|
|
state[14] = w6;
|
|
|
|
state[15] = w7;
|
|
|
|
}
|
|
|
|
|
|
|
|
// Feistel round function using two AES subrounds. Very similar to F()
|
|
|
|
// from Simpira v2, but with independent subround keys. Uses 17 AES rounds
|
|
|
|
// per 16 bytes (vs. 10 for AES-CTR). Computing eight round functions in
|
|
|
|
// parallel hides the 7-cycle AESNI latency on HSW. Note that the Feistel
|
|
|
|
// XORs are 'free' (included in the second AES instruction).
|
Export of internal Abseil changes
--
007ce045d5d38a727ededdb5bf06e64785fd73bd by Martijn Vels <mvels@google.com>:
Add `cord_enable_btree` feature flag (default false).
PiperOrigin-RevId: 383729939
--
98e7dc6a0407b0fd7b8713d883cdb3a766e0583d by Benjamin Barenblat <bbaren@google.com>:
Eliminate some byte swapping from randen_slow
Stop swapping bytes when serializing randen_slow’s Vector128 into and
out of memory. Instead, simply index different bytes in the AES round
function. This requires byte swapping the te{0..3} lookup tables, but it
produces an 8% speedup on my Xeon W-2135.
PiperOrigin-RevId: 383689402
--
180b6bf45049188840d439b16a28e6b968669340 by Evan Brown <ezb@google.com>:
Minor simplification in drop_deletes_without_resize() - save probe_offset outside the lambda.
Also, add some consts, avoid an auto, and use lambda capture by value instead of reference.
I realized that the compiler can already optimize this - https://godbolt.org/z/Wxd9c4TfK, but I think this way makes the code a bit clearer.
PiperOrigin-RevId: 383646658
--
781706a974c4dc1c0abbb6b801fca0550229e883 by Martijn Vels <mvels@google.com>:
Change storage to contain 3 bytes.
As per the comments in the code, this allows us to utilize all available space in CordRep that may otherwise be 'lost' in padding in derived clases. For the upcoming CordrepBtree class, we want a strong guarantee on having a 64 bytes aligned implementation.
PiperOrigin-RevId: 383633963
--
8fe22ecf92492fa6649938a2215934ebfe01c714 by Derek Mauro <dmauro@google.com>:
Remove reference to str_format_arg.h, which no longer exists
PiperOrigin-RevId: 383517865
--
79397f3b18f18c1e2d7aea993b687329d626ce64 by Benjamin Barenblat <bbaren@google.com>:
Use absl::uint128 for AES random number generator
Replace randen’s internal 128-bit integer struct, u64x2, with
absl::uint128. This eliminates some code and improves support for
big-endian platforms.
PiperOrigin-RevId: 383475671
GitOrigin-RevId: 007ce045d5d38a727ededdb5bf06e64785fd73bd
Change-Id: Ia9d9c40de557221f1744fb0d6d4d6ca7ac569070
3 years ago
|
|
|
inline ABSL_RANDOM_INTERNAL_ATTRIBUTE_ALWAYS_INLINE const absl::uint128*
|
|
|
|
FeistelRound(absl::uint128* ABSL_RANDOM_INTERNAL_RESTRICT state,
|
|
|
|
const absl::uint128* ABSL_RANDOM_INTERNAL_RESTRICT keys) {
|
|
|
|
for (size_t branch = 0; branch < RandenTraits::kFeistelBlocks; branch += 4) {
|
|
|
|
const Vector128 s0 = Vector128Load(state + branch);
|
|
|
|
const Vector128 s1 = Vector128Load(state + branch + 1);
|
|
|
|
const Vector128 f0 = AesRound(s0, Vector128Load(keys));
|
|
|
|
keys++;
|
|
|
|
const Vector128 o1 = AesRound(f0, s1);
|
|
|
|
Vector128Store(o1, state + branch + 1);
|
|
|
|
|
|
|
|
// Manually unroll this loop once. about 10% better than not unrolled.
|
|
|
|
const Vector128 s2 = Vector128Load(state + branch + 2);
|
|
|
|
const Vector128 s3 = Vector128Load(state + branch + 3);
|
|
|
|
const Vector128 f2 = AesRound(s2, Vector128Load(keys));
|
|
|
|
keys++;
|
|
|
|
const Vector128 o3 = AesRound(f2, s3);
|
|
|
|
Vector128Store(o3, state + branch + 3);
|
|
|
|
}
|
|
|
|
return keys;
|
|
|
|
}
|
|
|
|
|
|
|
|
// Cryptographic permutation based via type-2 Generalized Feistel Network.
|
|
|
|
// Indistinguishable from ideal by chosen-ciphertext adversaries using less than
|
|
|
|
// 2^64 queries if the round function is a PRF. This is similar to the b=8 case
|
|
|
|
// of Simpira v2, but more efficient than its generic construction for b=16.
|
|
|
|
inline ABSL_RANDOM_INTERNAL_ATTRIBUTE_ALWAYS_INLINE void Permute(
|
Export of internal Abseil changes
--
007ce045d5d38a727ededdb5bf06e64785fd73bd by Martijn Vels <mvels@google.com>:
Add `cord_enable_btree` feature flag (default false).
PiperOrigin-RevId: 383729939
--
98e7dc6a0407b0fd7b8713d883cdb3a766e0583d by Benjamin Barenblat <bbaren@google.com>:
Eliminate some byte swapping from randen_slow
Stop swapping bytes when serializing randen_slow’s Vector128 into and
out of memory. Instead, simply index different bytes in the AES round
function. This requires byte swapping the te{0..3} lookup tables, but it
produces an 8% speedup on my Xeon W-2135.
PiperOrigin-RevId: 383689402
--
180b6bf45049188840d439b16a28e6b968669340 by Evan Brown <ezb@google.com>:
Minor simplification in drop_deletes_without_resize() - save probe_offset outside the lambda.
Also, add some consts, avoid an auto, and use lambda capture by value instead of reference.
I realized that the compiler can already optimize this - https://godbolt.org/z/Wxd9c4TfK, but I think this way makes the code a bit clearer.
PiperOrigin-RevId: 383646658
--
781706a974c4dc1c0abbb6b801fca0550229e883 by Martijn Vels <mvels@google.com>:
Change storage to contain 3 bytes.
As per the comments in the code, this allows us to utilize all available space in CordRep that may otherwise be 'lost' in padding in derived clases. For the upcoming CordrepBtree class, we want a strong guarantee on having a 64 bytes aligned implementation.
PiperOrigin-RevId: 383633963
--
8fe22ecf92492fa6649938a2215934ebfe01c714 by Derek Mauro <dmauro@google.com>:
Remove reference to str_format_arg.h, which no longer exists
PiperOrigin-RevId: 383517865
--
79397f3b18f18c1e2d7aea993b687329d626ce64 by Benjamin Barenblat <bbaren@google.com>:
Use absl::uint128 for AES random number generator
Replace randen’s internal 128-bit integer struct, u64x2, with
absl::uint128. This eliminates some code and improves support for
big-endian platforms.
PiperOrigin-RevId: 383475671
GitOrigin-RevId: 007ce045d5d38a727ededdb5bf06e64785fd73bd
Change-Id: Ia9d9c40de557221f1744fb0d6d4d6ca7ac569070
3 years ago
|
|
|
absl::uint128* state,
|
|
|
|
const absl::uint128* ABSL_RANDOM_INTERNAL_RESTRICT keys) {
|
|
|
|
for (size_t round = 0; round < RandenTraits::kFeistelRounds; ++round) {
|
|
|
|
keys = FeistelRound(state, keys);
|
|
|
|
BlockShuffle(state);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
// Enables native loads in the round loop by pre-swapping.
|
|
|
|
inline ABSL_RANDOM_INTERNAL_ATTRIBUTE_ALWAYS_INLINE void SwapEndian(
|
|
|
|
absl::uint128* state) {
|
|
|
|
#ifdef ABSL_IS_BIG_ENDIAN
|
|
|
|
for (uint32_t block = 0; block < RandenTraits::kFeistelBlocks; ++block) {
|
|
|
|
uint64_t new_lo = absl::little_endian::ToHost64(
|
|
|
|
static_cast<uint64_t>(state[block] >> 64));
|
|
|
|
uint64_t new_hi = absl::little_endian::ToHost64(
|
|
|
|
static_cast<uint64_t>((state[block] << 64) >> 64));
|
|
|
|
state[block] = (static_cast<absl::uint128>(new_hi) << 64) | new_lo;
|
|
|
|
}
|
|
|
|
#else
|
|
|
|
// Avoid warning about unused variable.
|
|
|
|
(void)state;
|
|
|
|
#endif
|
|
|
|
}
|
|
|
|
|
|
|
|
} // namespace
|
|
|
|
|
|
|
|
namespace absl {
|
|
|
|
ABSL_NAMESPACE_BEGIN
|
|
|
|
namespace random_internal {
|
|
|
|
|
|
|
|
const void* RandenSlow::GetKeys() {
|
|
|
|
// Round keys for one AES per Feistel round and branch.
|
|
|
|
// The canonical implementation uses first digits of Pi.
|
|
|
|
#ifdef ABSL_IS_LITTLE_ENDIAN
|
|
|
|
return kRandenRoundKeys;
|
|
|
|
#else
|
|
|
|
return kRandenRoundKeysBE;
|
|
|
|
#endif
|
|
|
|
}
|
|
|
|
|
|
|
|
void RandenSlow::Absorb(const void* seed_void, void* state_void) {
|
|
|
|
auto* state =
|
|
|
|
reinterpret_cast<uint64_t * ABSL_RANDOM_INTERNAL_RESTRICT>(state_void);
|
|
|
|
const auto* seed =
|
|
|
|
reinterpret_cast<const uint64_t * ABSL_RANDOM_INTERNAL_RESTRICT>(
|
|
|
|
seed_void);
|
|
|
|
|
|
|
|
constexpr size_t kCapacityBlocks =
|
|
|
|
RandenTraits::kCapacityBytes / sizeof(uint64_t);
|
|
|
|
static_assert(
|
|
|
|
kCapacityBlocks * sizeof(uint64_t) == RandenTraits::kCapacityBytes,
|
|
|
|
"Not i*V");
|
|
|
|
|
|
|
|
for (size_t i = kCapacityBlocks;
|
|
|
|
i < RandenTraits::kStateBytes / sizeof(uint64_t); ++i) {
|
|
|
|
state[i] ^= seed[i - kCapacityBlocks];
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
void RandenSlow::Generate(const void* keys_void, void* state_void) {
|
Export of internal Abseil changes
--
007ce045d5d38a727ededdb5bf06e64785fd73bd by Martijn Vels <mvels@google.com>:
Add `cord_enable_btree` feature flag (default false).
PiperOrigin-RevId: 383729939
--
98e7dc6a0407b0fd7b8713d883cdb3a766e0583d by Benjamin Barenblat <bbaren@google.com>:
Eliminate some byte swapping from randen_slow
Stop swapping bytes when serializing randen_slow’s Vector128 into and
out of memory. Instead, simply index different bytes in the AES round
function. This requires byte swapping the te{0..3} lookup tables, but it
produces an 8% speedup on my Xeon W-2135.
PiperOrigin-RevId: 383689402
--
180b6bf45049188840d439b16a28e6b968669340 by Evan Brown <ezb@google.com>:
Minor simplification in drop_deletes_without_resize() - save probe_offset outside the lambda.
Also, add some consts, avoid an auto, and use lambda capture by value instead of reference.
I realized that the compiler can already optimize this - https://godbolt.org/z/Wxd9c4TfK, but I think this way makes the code a bit clearer.
PiperOrigin-RevId: 383646658
--
781706a974c4dc1c0abbb6b801fca0550229e883 by Martijn Vels <mvels@google.com>:
Change storage to contain 3 bytes.
As per the comments in the code, this allows us to utilize all available space in CordRep that may otherwise be 'lost' in padding in derived clases. For the upcoming CordrepBtree class, we want a strong guarantee on having a 64 bytes aligned implementation.
PiperOrigin-RevId: 383633963
--
8fe22ecf92492fa6649938a2215934ebfe01c714 by Derek Mauro <dmauro@google.com>:
Remove reference to str_format_arg.h, which no longer exists
PiperOrigin-RevId: 383517865
--
79397f3b18f18c1e2d7aea993b687329d626ce64 by Benjamin Barenblat <bbaren@google.com>:
Use absl::uint128 for AES random number generator
Replace randen’s internal 128-bit integer struct, u64x2, with
absl::uint128. This eliminates some code and improves support for
big-endian platforms.
PiperOrigin-RevId: 383475671
GitOrigin-RevId: 007ce045d5d38a727ededdb5bf06e64785fd73bd
Change-Id: Ia9d9c40de557221f1744fb0d6d4d6ca7ac569070
3 years ago
|
|
|
static_assert(RandenTraits::kCapacityBytes == sizeof(absl::uint128),
|
|
|
|
"Capacity mismatch");
|
|
|
|
|
Export of internal Abseil changes
--
007ce045d5d38a727ededdb5bf06e64785fd73bd by Martijn Vels <mvels@google.com>:
Add `cord_enable_btree` feature flag (default false).
PiperOrigin-RevId: 383729939
--
98e7dc6a0407b0fd7b8713d883cdb3a766e0583d by Benjamin Barenblat <bbaren@google.com>:
Eliminate some byte swapping from randen_slow
Stop swapping bytes when serializing randen_slow’s Vector128 into and
out of memory. Instead, simply index different bytes in the AES round
function. This requires byte swapping the te{0..3} lookup tables, but it
produces an 8% speedup on my Xeon W-2135.
PiperOrigin-RevId: 383689402
--
180b6bf45049188840d439b16a28e6b968669340 by Evan Brown <ezb@google.com>:
Minor simplification in drop_deletes_without_resize() - save probe_offset outside the lambda.
Also, add some consts, avoid an auto, and use lambda capture by value instead of reference.
I realized that the compiler can already optimize this - https://godbolt.org/z/Wxd9c4TfK, but I think this way makes the code a bit clearer.
PiperOrigin-RevId: 383646658
--
781706a974c4dc1c0abbb6b801fca0550229e883 by Martijn Vels <mvels@google.com>:
Change storage to contain 3 bytes.
As per the comments in the code, this allows us to utilize all available space in CordRep that may otherwise be 'lost' in padding in derived clases. For the upcoming CordrepBtree class, we want a strong guarantee on having a 64 bytes aligned implementation.
PiperOrigin-RevId: 383633963
--
8fe22ecf92492fa6649938a2215934ebfe01c714 by Derek Mauro <dmauro@google.com>:
Remove reference to str_format_arg.h, which no longer exists
PiperOrigin-RevId: 383517865
--
79397f3b18f18c1e2d7aea993b687329d626ce64 by Benjamin Barenblat <bbaren@google.com>:
Use absl::uint128 for AES random number generator
Replace randen’s internal 128-bit integer struct, u64x2, with
absl::uint128. This eliminates some code and improves support for
big-endian platforms.
PiperOrigin-RevId: 383475671
GitOrigin-RevId: 007ce045d5d38a727ededdb5bf06e64785fd73bd
Change-Id: Ia9d9c40de557221f1744fb0d6d4d6ca7ac569070
3 years ago
|
|
|
auto* state = reinterpret_cast<absl::uint128*>(state_void);
|
|
|
|
const auto* keys = reinterpret_cast<const absl::uint128*>(keys_void);
|
|
|
|
|
Export of internal Abseil changes
--
007ce045d5d38a727ededdb5bf06e64785fd73bd by Martijn Vels <mvels@google.com>:
Add `cord_enable_btree` feature flag (default false).
PiperOrigin-RevId: 383729939
--
98e7dc6a0407b0fd7b8713d883cdb3a766e0583d by Benjamin Barenblat <bbaren@google.com>:
Eliminate some byte swapping from randen_slow
Stop swapping bytes when serializing randen_slow’s Vector128 into and
out of memory. Instead, simply index different bytes in the AES round
function. This requires byte swapping the te{0..3} lookup tables, but it
produces an 8% speedup on my Xeon W-2135.
PiperOrigin-RevId: 383689402
--
180b6bf45049188840d439b16a28e6b968669340 by Evan Brown <ezb@google.com>:
Minor simplification in drop_deletes_without_resize() - save probe_offset outside the lambda.
Also, add some consts, avoid an auto, and use lambda capture by value instead of reference.
I realized that the compiler can already optimize this - https://godbolt.org/z/Wxd9c4TfK, but I think this way makes the code a bit clearer.
PiperOrigin-RevId: 383646658
--
781706a974c4dc1c0abbb6b801fca0550229e883 by Martijn Vels <mvels@google.com>:
Change storage to contain 3 bytes.
As per the comments in the code, this allows us to utilize all available space in CordRep that may otherwise be 'lost' in padding in derived clases. For the upcoming CordrepBtree class, we want a strong guarantee on having a 64 bytes aligned implementation.
PiperOrigin-RevId: 383633963
--
8fe22ecf92492fa6649938a2215934ebfe01c714 by Derek Mauro <dmauro@google.com>:
Remove reference to str_format_arg.h, which no longer exists
PiperOrigin-RevId: 383517865
--
79397f3b18f18c1e2d7aea993b687329d626ce64 by Benjamin Barenblat <bbaren@google.com>:
Use absl::uint128 for AES random number generator
Replace randen’s internal 128-bit integer struct, u64x2, with
absl::uint128. This eliminates some code and improves support for
big-endian platforms.
PiperOrigin-RevId: 383475671
GitOrigin-RevId: 007ce045d5d38a727ededdb5bf06e64785fd73bd
Change-Id: Ia9d9c40de557221f1744fb0d6d4d6ca7ac569070
3 years ago
|
|
|
const absl::uint128 prev_inner = state[0];
|
|
|
|
|
|
|
|
SwapEndian(state);
|
|
|
|
|
|
|
|
Permute(state, keys);
|
|
|
|
|
|
|
|
SwapEndian(state);
|
|
|
|
|
|
|
|
// Ensure backtracking resistance.
|
Export of internal Abseil changes
--
007ce045d5d38a727ededdb5bf06e64785fd73bd by Martijn Vels <mvels@google.com>:
Add `cord_enable_btree` feature flag (default false).
PiperOrigin-RevId: 383729939
--
98e7dc6a0407b0fd7b8713d883cdb3a766e0583d by Benjamin Barenblat <bbaren@google.com>:
Eliminate some byte swapping from randen_slow
Stop swapping bytes when serializing randen_slow’s Vector128 into and
out of memory. Instead, simply index different bytes in the AES round
function. This requires byte swapping the te{0..3} lookup tables, but it
produces an 8% speedup on my Xeon W-2135.
PiperOrigin-RevId: 383689402
--
180b6bf45049188840d439b16a28e6b968669340 by Evan Brown <ezb@google.com>:
Minor simplification in drop_deletes_without_resize() - save probe_offset outside the lambda.
Also, add some consts, avoid an auto, and use lambda capture by value instead of reference.
I realized that the compiler can already optimize this - https://godbolt.org/z/Wxd9c4TfK, but I think this way makes the code a bit clearer.
PiperOrigin-RevId: 383646658
--
781706a974c4dc1c0abbb6b801fca0550229e883 by Martijn Vels <mvels@google.com>:
Change storage to contain 3 bytes.
As per the comments in the code, this allows us to utilize all available space in CordRep that may otherwise be 'lost' in padding in derived clases. For the upcoming CordrepBtree class, we want a strong guarantee on having a 64 bytes aligned implementation.
PiperOrigin-RevId: 383633963
--
8fe22ecf92492fa6649938a2215934ebfe01c714 by Derek Mauro <dmauro@google.com>:
Remove reference to str_format_arg.h, which no longer exists
PiperOrigin-RevId: 383517865
--
79397f3b18f18c1e2d7aea993b687329d626ce64 by Benjamin Barenblat <bbaren@google.com>:
Use absl::uint128 for AES random number generator
Replace randen’s internal 128-bit integer struct, u64x2, with
absl::uint128. This eliminates some code and improves support for
big-endian platforms.
PiperOrigin-RevId: 383475671
GitOrigin-RevId: 007ce045d5d38a727ededdb5bf06e64785fd73bd
Change-Id: Ia9d9c40de557221f1744fb0d6d4d6ca7ac569070
3 years ago
|
|
|
*state ^= prev_inner;
|
|
|
|
}
|
|
|
|
|
|
|
|
} // namespace random_internal
|
|
|
|
ABSL_NAMESPACE_END
|
|
|
|
} // namespace absl
|