sharpetronics/liboqs
1
0
Fork 0
mirror of https://github.com/open-quantum-safe/liboqs.git synced 2025-10-08 00:03:38 -04:00
Code Issues Projects Releases Wiki Activity

Fix undefined behaviour in keccak4x (#919)

Browse Source 
* keccak4x: Fix undefined load from improperly aligned static constant

* keccak4x: Backport some minor changes

* astyle
This commit is contained in:
John Schanck 2021-02-17 20:38:19 -05:00 committed by GitHub
parent 78a190ffa2
commit 952c628051
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 14 additions and 9 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

23
src/common/sha3/keccak4x/KeccakP-1600-times4-SIMD256.c
View File

@ -42,25 +42,30 @@ typedef __m256i V256;
#if defined(KeccakP1600times4_useAVX2)
#define ANDnu256(a, b) _mm256_andnot_si256(a, b)
#define CONST256(a) _mm256_load_si256((const V256 *) &(a))
#define CONST256_64(a) _mm256_castpd_si256(_mm256_broadcast_sd((const double *) (&a)))
#define CONST256_64(a) _mm256_set1_epi64x((long long) (a))
#define LOAD256(a) _mm256_load_si256((const V256 *) &(a))
#define LOAD256u(a) _mm256_loadu_si256((const V256 *) &(a))
#define LOAD4_64(a, b, c, d) _mm256_set_epi64x((UINT64)(a), (UINT64)(b), (UINT64)(c), (UINT64)(d))
#define ROL64in256(d, a, o) d = _mm256_or_si256(_mm256_slli_epi64(a, o), _mm256_srli_epi64(a, 64 - (o)))
#define ROL64in256_8(d, a) d = _mm256_shuffle_epi8(a, CONST256(rho8))
#define ROL64in256_56(d, a) d = _mm256_shuffle_epi8(a, CONST256(rho56))
static const UINT64 rho8[4] = {0x0605040302010007, 0x0E0D0C0B0A09080F, 0x1615141312111017, 0x1E1D1C1B1A19181F};
static const UINT64 rho56[4] = {0x0007060504030201, 0x080F0E0D0C0B0A09, 0x1017161514131211, 0x181F1E1D1C1B1A19};
#define ROL64in256_8(d, a) d = _mm256_shuffle_epi8(a, rho8.v)
#define ROL64in256_56(d, a) d = _mm256_shuffle_epi8(a, rho56.v)
static const union {
UINT64 i[4];
V256 v;
} rho8 = {{0x0605040302010007, 0x0E0D0C0B0A09080F, 0x1615141312111017, 0x1E1D1C1B1A19181F}};
static const union {
UINT64 i[4];
V256 v;
} rho56 = {{0x0007060504030201, 0x080F0E0D0C0B0A09, 0x1017161514131211, 0x181F1E1D1C1B1A19}};
#define STORE256(a, b) _mm256_store_si256((V256 *) &(a), b)
#define STORE256u(a, b) _mm256_storeu_si256((V256 *) &(a), b)
#define STORE2_128(ah, al, v) _mm256_storeu2_m128d((V128 *) &(ah), (V128 *) &(al), v)
#define STORE2_128(ah, al, v) _mm256_storeu2_m128i((V128 *) &(ah), (V128 *) &(al), v)
#define XOR256(a, b) _mm256_xor_si256(a, b)
#define XOReq256(a, b) a = _mm256_xor_si256(a, b)
#define UNPACKL(a, b) _mm256_unpacklo_epi64((a), (b))
#define UNPACKH(a, b) _mm256_unpackhi_epi64((a), (b))
#define PERM128(a, b, c) (V256) _mm256_permute2f128_ps((__m256)(a), (__m256)(b), c)
#define SHUFFLE64(a, b, c) (V256) _mm256_shuffle_pd((__m256d)(a), (__m256d)(b), c)
#define PERM128(a, b, c) _mm256_permute2f128_si256((a), (b), c)
#define SHUFFLE64(a, b, c) _mm256_castpd_si256(_mm256_shuffle_pd(_mm256_castsi256_pd(a), _mm256_castsi256_pd(b), c))
#define UNINTLEAVE() lanesL01 = UNPACKL(lanes0, lanes1), \
lanesH01 = UNPACKH(lanes0, lanes1), \