[llvm-branch-commits] [llvm] [Support] Integrate SipHash.cpp into libSupport. (PR #94394)
Ahmed Bougacha via llvm-branch-commits
llvm-branch-commits at lists.llvm.org
Tue Jun 11 16:14:00 PDT 2024
================
@@ -1,185 +1,149 @@
-/*
- SipHash reference C implementation
-
- Copyright (c) 2012-2022 Jean-Philippe Aumasson
- <jeanphilippe.aumasson at gmail.com>
- Copyright (c) 2012-2014 Daniel J. Bernstein <djb at cr.yp.to>
-
- To the extent possible under law, the author(s) have dedicated all copyright
- and related and neighboring rights to this software to the public domain
- worldwide. This software is distributed without any warranty.
-
- You should have received a copy of the CC0 Public Domain Dedication along
- with
- this software. If not, see
- <http://creativecommons.org/publicdomain/zero/1.0/>.
- */
-
-#include "siphash.h"
-#include <assert.h>
-#include <stddef.h>
-#include <stdint.h>
-
-/* default: SipHash-2-4 */
-#ifndef cROUNDS
-#define cROUNDS 2
-#endif
-#ifndef dROUNDS
-#define dROUNDS 4
-#endif
+//===--- SipHash.cpp - An ABI-stable string hash --------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
-#define ROTL(x, b) (uint64_t)(((x) << (b)) | ((x) >> (64 - (b))))
+#include "llvm/Support/Compiler.h"
+#include <cstdint>
-#define U32TO8_LE(p, v) \
- (p)[0] = (uint8_t)((v)); \
- (p)[1] = (uint8_t)((v) >> 8); \
- (p)[2] = (uint8_t)((v) >> 16); \
- (p)[3] = (uint8_t)((v) >> 24);
+// Lightly adapted from the SipHash reference C implementation:
+// https://github.com/veorq/SipHash
+// by Jean-Philippe Aumasson and Daniel J. Bernstein
-#define U64TO8_LE(p, v) \
- U32TO8_LE((p), (uint32_t)((v))); \
- U32TO8_LE((p) + 4, (uint32_t)((v) >> 32));
+#define ROTL(x, b) (uint64_t)(((x) << (b)) | ((x) >> (64 - (b))))
#define U8TO64_LE(p) \
- (((uint64_t)((p)[0])) | ((uint64_t)((p)[1]) << 8) | \
- ((uint64_t)((p)[2]) << 16) | ((uint64_t)((p)[3]) << 24) | \
- ((uint64_t)((p)[4]) << 32) | ((uint64_t)((p)[5]) << 40) | \
- ((uint64_t)((p)[6]) << 48) | ((uint64_t)((p)[7]) << 56))
+ (((uint64_t)((p)[0])) | ((uint64_t)((p)[1]) << 8) | \
+ ((uint64_t)((p)[2]) << 16) | ((uint64_t)((p)[3]) << 24) | \
+ ((uint64_t)((p)[4]) << 32) | ((uint64_t)((p)[5]) << 40) | \
+ ((uint64_t)((p)[6]) << 48) | ((uint64_t)((p)[7]) << 56))
#define SIPROUND \
- do { \
- v0 += v1; \
- v1 = ROTL(v1, 13); \
- v1 ^= v0; \
- v0 = ROTL(v0, 32); \
- v2 += v3; \
- v3 = ROTL(v3, 16); \
- v3 ^= v2; \
- v0 += v3; \
- v3 = ROTL(v3, 21); \
- v3 ^= v0; \
- v2 += v1; \
- v1 = ROTL(v1, 17); \
- v1 ^= v2; \
- v2 = ROTL(v2, 32); \
- } while (0)
-
-#ifdef DEBUG_SIPHASH
-#include <stdio.h>
-
-#define TRACE \
- do { \
- printf("(%3zu) v0 %016" PRIx64 "\n", inlen, v0); \
- printf("(%3zu) v1 %016" PRIx64 "\n", inlen, v1); \
- printf("(%3zu) v2 %016" PRIx64 "\n", inlen, v2); \
- printf("(%3zu) v3 %016" PRIx64 "\n", inlen, v3); \
- } while (0)
-#else
-#define TRACE
-#endif
-
-/*
- Computes a SipHash value
- *in: pointer to input data (read-only)
- inlen: input data length in bytes (any size_t value)
- *k: pointer to the key data (read-only), must be 16 bytes
- *out: pointer to output data (write-only), outlen bytes must be allocated
- outlen: length of the output in bytes, must be 8 or 16
-*/
-int siphash(const void *in, const size_t inlen, const void *k, uint8_t *out,
- const size_t outlen) {
-
- const unsigned char *ni = (const unsigned char *)in;
- const unsigned char *kk = (const unsigned char *)k;
-
- assert((outlen == 8) || (outlen == 16));
- uint64_t v0 = UINT64_C(0x736f6d6570736575);
- uint64_t v1 = UINT64_C(0x646f72616e646f6d);
- uint64_t v2 = UINT64_C(0x6c7967656e657261);
- uint64_t v3 = UINT64_C(0x7465646279746573);
- uint64_t k0 = U8TO64_LE(kk);
- uint64_t k1 = U8TO64_LE(kk + 8);
- uint64_t m;
- int i;
- const unsigned char *end = ni + inlen - (inlen % sizeof(uint64_t));
- const int left = inlen & 7;
- uint64_t b = ((uint64_t)inlen) << 56;
- v3 ^= k1;
- v2 ^= k0;
- v1 ^= k1;
- v0 ^= k0;
-
- if (outlen == 16)
- v1 ^= 0xee;
-
- for (; ni != end; ni += 8) {
- m = U8TO64_LE(ni);
- v3 ^= m;
-
- TRACE;
- for (i = 0; i < cROUNDS; ++i)
- SIPROUND;
-
- v0 ^= m;
- }
-
- switch (left) {
- case 7:
- b |= ((uint64_t)ni[6]) << 48;
- /* FALLTHRU */
- case 6:
- b |= ((uint64_t)ni[5]) << 40;
- /* FALLTHRU */
- case 5:
- b |= ((uint64_t)ni[4]) << 32;
- /* FALLTHRU */
- case 4:
- b |= ((uint64_t)ni[3]) << 24;
- /* FALLTHRU */
- case 3:
- b |= ((uint64_t)ni[2]) << 16;
- /* FALLTHRU */
- case 2:
- b |= ((uint64_t)ni[1]) << 8;
- /* FALLTHRU */
- case 1:
- b |= ((uint64_t)ni[0]);
- break;
- case 0:
- break;
- }
-
- v3 ^= b;
-
- TRACE;
- for (i = 0; i < cROUNDS; ++i)
- SIPROUND;
-
- v0 ^= b;
-
- if (outlen == 16)
- v2 ^= 0xee;
- else
- v2 ^= 0xff;
-
- TRACE;
- for (i = 0; i < dROUNDS; ++i)
- SIPROUND;
-
- b = v0 ^ v1 ^ v2 ^ v3;
- U64TO8_LE(out, b);
+ do { \
+ v0 += v1; \
+ v1 = ROTL(v1, 13); \
+ v1 ^= v0; \
+ v0 = ROTL(v0, 32); \
+ v2 += v3; \
+ v3 = ROTL(v3, 16); \
+ v3 ^= v2; \
+ v0 += v3; \
+ v3 = ROTL(v3, 21); \
+ v3 ^= v0; \
+ v2 += v1; \
+ v1 = ROTL(v1, 17); \
+ v1 ^= v2; \
+ v2 = ROTL(v2, 32); \
+ } while (0)
+
+namespace {
+
+/// Computes a SipHash value
+///
+/// \param in: pointer to input data (read-only)
+/// \param inlen: input data length in bytes (any size_t value)
+/// \param k: reference to the key data 16-byte array (read-only)
+/// \returns output data, must be 8 or 16 bytes
+///
+template <int cROUNDS, int dROUNDS, class ResultTy>
+ResultTy siphash(const unsigned char *in, uint64_t inlen,
+ const unsigned char (&k)[16]) {
+
+ const unsigned char *ni = (const unsigned char *)in;
+ const unsigned char *kk = (const unsigned char *)k;
+
+ constexpr size_t outlen = sizeof(ResultTy);
+ static_assert(outlen == 8 || outlen == 16,
+ "result type should be uint64_t or uint128_t");
+
+ uint64_t v0 = UINT64_C(0x736f6d6570736575);
+ uint64_t v1 = UINT64_C(0x646f72616e646f6d);
+ uint64_t v2 = UINT64_C(0x6c7967656e657261);
+ uint64_t v3 = UINT64_C(0x7465646279746573);
+ uint64_t k0 = U8TO64_LE(kk);
+ uint64_t k1 = U8TO64_LE(kk + 8);
+ uint64_t m;
+ int i;
+ const unsigned char *end = ni + inlen - (inlen % sizeof(uint64_t));
+ const int left = inlen & 7;
+ uint64_t b = ((uint64_t)inlen) << 56;
+ v3 ^= k1;
+ v2 ^= k0;
+ v1 ^= k1;
+ v0 ^= k0;
+
+ if (outlen == 16)
+ v1 ^= 0xee;
+
+ for (; ni != end; ni += 8) {
+ m = U8TO64_LE(ni);
+ v3 ^= m;
- if (outlen == 8)
- return 0;
-
- v1 ^= 0xdd;
-
- TRACE;
- for (i = 0; i < dROUNDS; ++i)
- SIPROUND;
-
- b = v0 ^ v1 ^ v2 ^ v3;
- U64TO8_LE(out + 8, b);
-
- return 0;
+ for (i = 0; i < cROUNDS; ++i)
+ SIPROUND;
+
+ v0 ^= m;
+ }
+
+ switch (left) {
+ case 7:
+ b |= ((uint64_t)ni[6]) << 48;
+ LLVM_FALLTHROUGH;
+ case 6:
+ b |= ((uint64_t)ni[5]) << 40;
+ LLVM_FALLTHROUGH;
+ case 5:
+ b |= ((uint64_t)ni[4]) << 32;
+ LLVM_FALLTHROUGH;
+ case 4:
+ b |= ((uint64_t)ni[3]) << 24;
+ LLVM_FALLTHROUGH;
+ case 3:
+ b |= ((uint64_t)ni[2]) << 16;
+ LLVM_FALLTHROUGH;
+ case 2:
+ b |= ((uint64_t)ni[1]) << 8;
+ LLVM_FALLTHROUGH;
+ case 1:
+ b |= ((uint64_t)ni[0]);
+ break;
+ case 0:
+ break;
+ }
+
+ v3 ^= b;
+
+ for (i = 0; i < cROUNDS; ++i)
+ SIPROUND;
+
+ v0 ^= b;
+
+ if (outlen == 16)
+ v2 ^= 0xee;
+ else
+ v2 ^= 0xff;
+
+ for (i = 0; i < dROUNDS; ++i)
+ SIPROUND;
+
+ b = v0 ^ v1 ^ v2 ^ v3;
+ uint64_t firstHalf = b;
+
+ if (outlen == 8)
+ return firstHalf;
+
+ v1 ^= 0xdd;
+
+ for (i = 0; i < dROUNDS; ++i)
+ SIPROUND;
+
+ b = v0 ^ v1 ^ v2 ^ v3;
+ uint64_t secondHalf = b;
+
+ return firstHalf | (ResultTy(secondHalf) << 64);
----------------
ahmedbougacha wrote:
In the 128-bit result case we'd be exposed to the big-endian host ABI's treatment of uint128_t results, and from what I can find that seems to be quite messy! [644b473](https://github.com/llvm/llvm-project/pull/94394/commits/644b473fa7d0382b035d4dfd2ea4ed88fcd97233) describes it a little bit more, and brings back the original's returning via an i8 array. The caller can figure out what it wants to do with the result bytes.
For ptrauth in #93902 I made it `read64le` the result again (and we do define the truncation as based on the little-endian interpretation of the raw hash bytes already.)
https://github.com/llvm/llvm-project/pull/94394
More information about the llvm-branch-commits
mailing list