kris/pqcrypto
1
1
포크 0
의 미러 https://github.com/henrydcase/pqc.git synced 2024-11-25 17:01:22 +00:00
코드 이슈 1 릴리즈 위키 활동

dilithium: Remove leading underscore from some internal symbols

소스 검색
This commit is contained in:
John M. Schanck 2020-10-08 09:56:15 -04:00 committed by Kris Kwiatkowski
부모 351d17ae70
커밋 3db4fa4876
21개의 변경된 파일과 105개의 추가작업 그리고 105개의 파일을 삭제
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

16
crypto_sign/dilithium2/avx2/invntt.S
파일 보기

@ -45,9 +45,9 @@ vpsrlq $32,%ymm\h3,%ymm\h3
.global cdecl(PQCLEAN_DILITHIUM2_AVX2_invntt_levels0t4_avx)
cdecl(PQCLEAN_DILITHIUM2_AVX2_invntt_levels0t4_avx):
#consts
vmovdqa cdecl(_PQCLEAN_DILITHIUM2_AVX2_8xqinv)(%rip),%ymm0
vmovdqa cdecl(_PQCLEAN_DILITHIUM2_AVX2_8xq)(%rip),%ymm1
vmovdqa cdecl(_PQCLEAN_DILITHIUM2_AVX2_8x256q)(%rip),%ymm2
vmovdqa cdecl(PQCLEAN_DILITHIUM2_AVX2_8xqinv)(%rip),%ymm0
vmovdqa cdecl(PQCLEAN_DILITHIUM2_AVX2_8xq)(%rip),%ymm1
vmovdqa cdecl(PQCLEAN_DILITHIUM2_AVX2_8x256q)(%rip),%ymm2
#load
vmovdqa (%rsi),%ymm6
@ -165,9 +165,9 @@ ret
.global cdecl(PQCLEAN_DILITHIUM2_AVX2_invntt_levels5t7_avx)
cdecl(PQCLEAN_DILITHIUM2_AVX2_invntt_levels5t7_avx):
#consts
vmovdqa cdecl(_PQCLEAN_DILITHIUM2_AVX2_8xqinv)(%rip),%ymm0
vmovdqa cdecl(_PQCLEAN_DILITHIUM2_AVX2_8xq)(%rip),%ymm1
vmovdqa cdecl(_PQCLEAN_DILITHIUM2_AVX2_8x256q)(%rip),%ymm2
vmovdqa cdecl(PQCLEAN_DILITHIUM2_AVX2_8xqinv)(%rip),%ymm0
vmovdqa cdecl(PQCLEAN_DILITHIUM2_AVX2_8xq)(%rip),%ymm1
vmovdqa cdecl(PQCLEAN_DILITHIUM2_AVX2_8x256q)(%rip),%ymm2
#load
vmovdqa (%rsi),%ymm4
@ -237,7 +237,7 @@ vpbroadcastd 24(%rdx),%ymm3
butterfly 4,5,6,7,8,9,10,11,3,3
#consts
vmovdqa cdecl(_PQCLEAN_DILITHIUM2_AVX2_8xdiv)(%rip),%ymm3
vmovdqa cdecl(PQCLEAN_DILITHIUM2_AVX2_8xdiv)(%rip),%ymm3
vpmuludq %ymm3,%ymm4,%ymm4
vpmuludq %ymm3,%ymm5,%ymm5
@ -261,7 +261,7 @@ vpsrlq $32,%ymm6,%ymm6
vpsrlq $32,%ymm7,%ymm7
#store
vmovdqa cdecl(_PQCLEAN_DILITHIUM2_AVX2_mask)(%rip),%ymm3
vmovdqa cdecl(PQCLEAN_DILITHIUM2_AVX2_mask)(%rip),%ymm3
vpermd %ymm4,%ymm3,%ymm4
vpermd %ymm5,%ymm3,%ymm5
vpermd %ymm6,%ymm3,%ymm6

12
crypto_sign/dilithium2/avx2/ntt.S
파일 보기

@ -44,9 +44,9 @@ vpsubd %ymm15,%ymm\rh3,%ymm\rh3
.global cdecl(PQCLEAN_DILITHIUM2_AVX2_ntt_levels0t2_avx)
cdecl(PQCLEAN_DILITHIUM2_AVX2_ntt_levels0t2_avx):
#consts
vmovdqa cdecl(_PQCLEAN_DILITHIUM2_AVX2_8xqinv)(%rip),%ymm0
vmovdqa cdecl(_PQCLEAN_DILITHIUM2_AVX2_8xq)(%rip),%ymm1
vmovdqa cdecl(_PQCLEAN_DILITHIUM2_AVX2_8x2q)(%rip),%ymm2
vmovdqa cdecl(PQCLEAN_DILITHIUM2_AVX2_8xqinv)(%rip),%ymm0
vmovdqa cdecl(PQCLEAN_DILITHIUM2_AVX2_8xq)(%rip),%ymm1
vmovdqa cdecl(PQCLEAN_DILITHIUM2_AVX2_8x2q)(%rip),%ymm2
level0:
#zetas
@ -95,9 +95,9 @@ ret
.global cdecl(PQCLEAN_DILITHIUM2_AVX2_ntt_levels3t8_avx)
cdecl(PQCLEAN_DILITHIUM2_AVX2_ntt_levels3t8_avx):
#consts
vmovdqa cdecl(_PQCLEAN_DILITHIUM2_AVX2_8xqinv)(%rip),%ymm0
vmovdqa cdecl(_PQCLEAN_DILITHIUM2_AVX2_8xq)(%rip),%ymm1
vmovdqa cdecl(_PQCLEAN_DILITHIUM2_AVX2_8x2q)(%rip),%ymm2
vmovdqa cdecl(PQCLEAN_DILITHIUM2_AVX2_8xqinv)(%rip),%ymm0
vmovdqa cdecl(PQCLEAN_DILITHIUM2_AVX2_8xq)(%rip),%ymm1
vmovdqa cdecl(PQCLEAN_DILITHIUM2_AVX2_8x2q)(%rip),%ymm2
#load
vmovdqa (%rsi),%ymm4

14
crypto_sign/dilithium2/avx2/nttconsts.c
파일 보기

@ -5,19 +5,19 @@
#define DIV (((MONT*MONT % Q) * (Q-1) % Q) * ((Q-1) >> 8) % Q)
const aligned_uint32x8_t _PQCLEAN_DILITHIUM2_AVX2_8xqinv = {.as_arr = {QINV, QINV, QINV, QINV, QINV, QINV, QINV, QINV}};
const aligned_uint32x8_t _PQCLEAN_DILITHIUM2_AVX2_8xq = {.as_arr = {Q, Q, Q, Q, Q, Q, Q, Q}};
const aligned_uint32x8_t _PQCLEAN_DILITHIUM2_AVX2_8x2q = {.as_arr = {2 * Q, 2 * Q, 2 * Q, 2 * Q, 2 * Q, 2 * Q, 2 * Q, 2 * Q}};
const aligned_uint32x8_t _PQCLEAN_DILITHIUM2_AVX2_8x256q = {.as_arr = {256 * Q, 256 * Q, 256 * Q, 256 * Q, 256 * Q, 256 * Q, 256 * Q,
const aligned_uint32x8_t PQCLEAN_DILITHIUM2_AVX2_8xqinv = {.as_arr = {QINV, QINV, QINV, QINV, QINV, QINV, QINV, QINV}};
const aligned_uint32x8_t PQCLEAN_DILITHIUM2_AVX2_8xq = {.as_arr = {Q, Q, Q, Q, Q, Q, Q, Q}};
const aligned_uint32x8_t PQCLEAN_DILITHIUM2_AVX2_8x2q = {.as_arr = {2 * Q, 2 * Q, 2 * Q, 2 * Q, 2 * Q, 2 * Q, 2 * Q, 2 * Q}};
const aligned_uint32x8_t PQCLEAN_DILITHIUM2_AVX2_8x256q = {.as_arr = {256 * Q, 256 * Q, 256 * Q, 256 * Q, 256 * Q, 256 * Q, 256 * Q,
256 * Q
}
};
const aligned_uint32x8_t _PQCLEAN_DILITHIUM2_AVX2_mask = {.as_arr = {0, 2, 4, 6, 0, 0, 0, 0}};
const aligned_uint32x8_t _PQCLEAN_DILITHIUM2_AVX2_8x23ones = {.as_arr = {0x7FFFFF, 0x7FFFFF, 0x7FFFFF, 0x7FFFFF, 0x7FFFFF, 0x7FFFFF,
const aligned_uint32x8_t PQCLEAN_DILITHIUM2_AVX2_mask = {.as_arr = {0, 2, 4, 6, 0, 0, 0, 0}};
const aligned_uint32x8_t PQCLEAN_DILITHIUM2_AVX2_8x23ones = {.as_arr = {0x7FFFFF, 0x7FFFFF, 0x7FFFFF, 0x7FFFFF, 0x7FFFFF, 0x7FFFFF,
0x7FFFFF, 0x7FFFFF
}
};
const aligned_uint32x8_t _PQCLEAN_DILITHIUM2_AVX2_8xdiv = { .as_arr = {DIV, DIV, DIV, DIV, DIV, DIV, DIV, DIV}};
const aligned_uint32x8_t PQCLEAN_DILITHIUM2_AVX2_8xdiv = { .as_arr = {DIV, DIV, DIV, DIV, DIV, DIV, DIV, DIV}};
#undef QINV
#undef MONT

14
crypto_sign/dilithium2/avx2/nttconsts.h
파일 보기

@ -12,13 +12,13 @@ typedef ALIGNED_UINT32(8) aligned_uint32x8_t;
typedef ALIGNED_UINT32(N) aligned_uint32xN_t;
extern const aligned_uint32x8_t _PQCLEAN_DILITHIUM2_AVX2_8xqinv;
extern const aligned_uint32x8_t _PQCLEAN_DILITHIUM2_AVX2_8xq;
extern const aligned_uint32x8_t _PQCLEAN_DILITHIUM2_AVX2_8x2q;
extern const aligned_uint32x8_t _PQCLEAN_DILITHIUM2_AVX2_8x256q;
extern const aligned_uint32x8_t _PQCLEAN_DILITHIUM2_AVX2_mask;
extern const aligned_uint32x8_t _PQCLEAN_DILITHIUM2_AVX2_8x23ones;
extern const aligned_uint32x8_t _PQCLEAN_DILITHIUM2_AVX2_8xdiv;
extern const aligned_uint32x8_t PQCLEAN_DILITHIUM2_AVX2_8xqinv;
extern const aligned_uint32x8_t PQCLEAN_DILITHIUM2_AVX2_8xq;
extern const aligned_uint32x8_t PQCLEAN_DILITHIUM2_AVX2_8x2q;
extern const aligned_uint32x8_t PQCLEAN_DILITHIUM2_AVX2_8x256q;
extern const aligned_uint32x8_t PQCLEAN_DILITHIUM2_AVX2_mask;
extern const aligned_uint32x8_t PQCLEAN_DILITHIUM2_AVX2_8x23ones;
extern const aligned_uint32x8_t PQCLEAN_DILITHIUM2_AVX2_8xdiv;
extern const aligned_uint32xN_t PQCLEAN_DILITHIUM2_AVX2_zetas;
extern const aligned_uint32xN_t PQCLEAN_DILITHIUM2_AVX2_zetas_inv;

8
crypto_sign/dilithium2/avx2/pointwise.S
파일 보기

@ -4,8 +4,8 @@
.global cdecl(PQCLEAN_DILITHIUM2_AVX2_pointwise_avx)
cdecl(PQCLEAN_DILITHIUM2_AVX2_pointwise_avx):
#consts
vmovdqa cdecl(_PQCLEAN_DILITHIUM2_AVX2_8xqinv)(%rip),%ymm0
vmovdqa cdecl(_PQCLEAN_DILITHIUM2_AVX2_8xq)(%rip),%ymm1
vmovdqa cdecl(PQCLEAN_DILITHIUM2_AVX2_8xqinv)(%rip),%ymm0
vmovdqa cdecl(PQCLEAN_DILITHIUM2_AVX2_8xq)(%rip),%ymm1
xor %eax,%eax
_looptop1:
@ -136,8 +136,8 @@ vpaddq %ymm9,%ymm5,%ymm5
.global cdecl(PQCLEAN_DILITHIUM2_AVX2_pointwise_acc_avx)
cdecl(PQCLEAN_DILITHIUM2_AVX2_pointwise_acc_avx):
#consts
vmovdqa cdecl(_PQCLEAN_DILITHIUM2_AVX2_8xqinv)(%rip),%ymm0
vmovdqa cdecl(_PQCLEAN_DILITHIUM2_AVX2_8xq)(%rip),%ymm1
vmovdqa cdecl(PQCLEAN_DILITHIUM2_AVX2_8xqinv)(%rip),%ymm0
vmovdqa cdecl(PQCLEAN_DILITHIUM2_AVX2_8xq)(%rip),%ymm1
xor %eax,%eax
_looptop2:

2
crypto_sign/dilithium2/avx2/poly.c
파일 보기

@ -82,7 +82,7 @@ void PQCLEAN_DILITHIUM2_AVX2_poly_add(poly *c, const poly *a, const poly *b) {
**************************************************/
void PQCLEAN_DILITHIUM2_AVX2_poly_sub(poly *c, const poly *a, const poly *b) {
__m256i vec0, vec1;
const __m256i twoq = _mm256_load_si256(_PQCLEAN_DILITHIUM2_AVX2_8x2q.as_vec);
const __m256i twoq = _mm256_load_si256(PQCLEAN_DILITHIUM2_AVX2_8x2q.as_vec);
for (size_t i = 0; i < N / 8; i++) {
vec0 = _mm256_load_si256(&a->coeffs_x8[i]);

4
crypto_sign/dilithium2/avx2/reduce.S
파일 보기

@ -3,7 +3,7 @@
.global cdecl(PQCLEAN_DILITHIUM2_AVX2_reduce_avx)
cdecl(PQCLEAN_DILITHIUM2_AVX2_reduce_avx):
#consts
vmovdqa cdecl(_PQCLEAN_DILITHIUM2_AVX2_8x23ones)(%rip),%ymm0
vmovdqa cdecl(PQCLEAN_DILITHIUM2_AVX2_8x23ones)(%rip),%ymm0
xor %eax,%eax
_looptop_rdc32:
@ -51,7 +51,7 @@ ret
.global cdecl(PQCLEAN_DILITHIUM2_AVX2_csubq_avx)
cdecl(PQCLEAN_DILITHIUM2_AVX2_csubq_avx):
#consts
vmovdqa cdecl(_PQCLEAN_DILITHIUM2_AVX2_8xq)(%rip),%ymm0
vmovdqa cdecl(PQCLEAN_DILITHIUM2_AVX2_8xq)(%rip),%ymm0
xor %eax,%eax
_looptop_csubq:

16
crypto_sign/dilithium3/avx2/invntt.S
파일 보기

@ -45,9 +45,9 @@ vpsrlq $32,%ymm\h3,%ymm\h3
.global cdecl(PQCLEAN_DILITHIUM3_AVX2_invntt_levels0t4_avx)
cdecl(PQCLEAN_DILITHIUM3_AVX2_invntt_levels0t4_avx):
#consts
vmovdqa cdecl(_PQCLEAN_DILITHIUM3_AVX2_8xqinv)(%rip),%ymm0
vmovdqa cdecl(_PQCLEAN_DILITHIUM3_AVX2_8xq)(%rip),%ymm1
vmovdqa cdecl(_PQCLEAN_DILITHIUM3_AVX2_8x256q)(%rip),%ymm2
vmovdqa cdecl(PQCLEAN_DILITHIUM3_AVX2_8xqinv)(%rip),%ymm0
vmovdqa cdecl(PQCLEAN_DILITHIUM3_AVX2_8xq)(%rip),%ymm1
vmovdqa cdecl(PQCLEAN_DILITHIUM3_AVX2_8x256q)(%rip),%ymm2
#load
vmovdqa (%rsi),%ymm6
@ -165,9 +165,9 @@ ret
.global cdecl(PQCLEAN_DILITHIUM3_AVX2_invntt_levels5t7_avx)
cdecl(PQCLEAN_DILITHIUM3_AVX2_invntt_levels5t7_avx):
#consts
vmovdqa cdecl(_PQCLEAN_DILITHIUM3_AVX2_8xqinv)(%rip),%ymm0
vmovdqa cdecl(_PQCLEAN_DILITHIUM3_AVX2_8xq)(%rip),%ymm1
vmovdqa cdecl(_PQCLEAN_DILITHIUM3_AVX2_8x256q)(%rip),%ymm2
vmovdqa cdecl(PQCLEAN_DILITHIUM3_AVX2_8xqinv)(%rip),%ymm0
vmovdqa cdecl(PQCLEAN_DILITHIUM3_AVX2_8xq)(%rip),%ymm1
vmovdqa cdecl(PQCLEAN_DILITHIUM3_AVX2_8x256q)(%rip),%ymm2
#load
vmovdqa (%rsi),%ymm4
@ -237,7 +237,7 @@ vpbroadcastd 24(%rdx),%ymm3
butterfly 4,5,6,7,8,9,10,11,3,3
#consts
vmovdqa cdecl(_PQCLEAN_DILITHIUM3_AVX2_8xdiv)(%rip),%ymm3
vmovdqa cdecl(PQCLEAN_DILITHIUM3_AVX2_8xdiv)(%rip),%ymm3
vpmuludq %ymm3,%ymm4,%ymm4
vpmuludq %ymm3,%ymm5,%ymm5
@ -261,7 +261,7 @@ vpsrlq $32,%ymm6,%ymm6
vpsrlq $32,%ymm7,%ymm7
#store
vmovdqa cdecl(_PQCLEAN_DILITHIUM3_AVX2_mask)(%rip),%ymm3
vmovdqa cdecl(PQCLEAN_DILITHIUM3_AVX2_mask)(%rip),%ymm3
vpermd %ymm4,%ymm3,%ymm4
vpermd %ymm5,%ymm3,%ymm5
vpermd %ymm6,%ymm3,%ymm6

12
crypto_sign/dilithium3/avx2/ntt.S
파일 보기

@ -44,9 +44,9 @@ vpsubd %ymm15,%ymm\rh3,%ymm\rh3
.global cdecl(PQCLEAN_DILITHIUM3_AVX2_ntt_levels0t2_avx)
cdecl(PQCLEAN_DILITHIUM3_AVX2_ntt_levels0t2_avx):
#consts
vmovdqa cdecl(_PQCLEAN_DILITHIUM3_AVX2_8xqinv)(%rip),%ymm0
vmovdqa cdecl(_PQCLEAN_DILITHIUM3_AVX2_8xq)(%rip),%ymm1
vmovdqa cdecl(_PQCLEAN_DILITHIUM3_AVX2_8x2q)(%rip),%ymm2
vmovdqa cdecl(PQCLEAN_DILITHIUM3_AVX2_8xqinv)(%rip),%ymm0
vmovdqa cdecl(PQCLEAN_DILITHIUM3_AVX2_8xq)(%rip),%ymm1
vmovdqa cdecl(PQCLEAN_DILITHIUM3_AVX2_8x2q)(%rip),%ymm2
level0:
#zetas
@ -95,9 +95,9 @@ ret
.global cdecl(PQCLEAN_DILITHIUM3_AVX2_ntt_levels3t8_avx)
cdecl(PQCLEAN_DILITHIUM3_AVX2_ntt_levels3t8_avx):
#consts
vmovdqa cdecl(_PQCLEAN_DILITHIUM3_AVX2_8xqinv)(%rip),%ymm0
vmovdqa cdecl(_PQCLEAN_DILITHIUM3_AVX2_8xq)(%rip),%ymm1
vmovdqa cdecl(_PQCLEAN_DILITHIUM3_AVX2_8x2q)(%rip),%ymm2
vmovdqa cdecl(PQCLEAN_DILITHIUM3_AVX2_8xqinv)(%rip),%ymm0
vmovdqa cdecl(PQCLEAN_DILITHIUM3_AVX2_8xq)(%rip),%ymm1
vmovdqa cdecl(PQCLEAN_DILITHIUM3_AVX2_8x2q)(%rip),%ymm2
#load
vmovdqa (%rsi),%ymm4

14
crypto_sign/dilithium3/avx2/nttconsts.c
파일 보기

@ -5,19 +5,19 @@
#define DIV (((MONT*MONT % Q) * (Q-1) % Q) * ((Q-1) >> 8) % Q)
const aligned_uint32x8_t _PQCLEAN_DILITHIUM3_AVX2_8xqinv = {.as_arr = {QINV, QINV, QINV, QINV, QINV, QINV, QINV, QINV}};
const aligned_uint32x8_t _PQCLEAN_DILITHIUM3_AVX2_8xq = {.as_arr = {Q, Q, Q, Q, Q, Q, Q, Q}};
const aligned_uint32x8_t _PQCLEAN_DILITHIUM3_AVX2_8x2q = {.as_arr = {2 * Q, 2 * Q, 2 * Q, 2 * Q, 2 * Q, 2 * Q, 2 * Q, 2 * Q}};
const aligned_uint32x8_t _PQCLEAN_DILITHIUM3_AVX2_8x256q = {.as_arr = {256 * Q, 256 * Q, 256 * Q, 256 * Q, 256 * Q, 256 * Q, 256 * Q,
const aligned_uint32x8_t PQCLEAN_DILITHIUM3_AVX2_8xqinv = {.as_arr = {QINV, QINV, QINV, QINV, QINV, QINV, QINV, QINV}};
const aligned_uint32x8_t PQCLEAN_DILITHIUM3_AVX2_8xq = {.as_arr = {Q, Q, Q, Q, Q, Q, Q, Q}};
const aligned_uint32x8_t PQCLEAN_DILITHIUM3_AVX2_8x2q = {.as_arr = {2 * Q, 2 * Q, 2 * Q, 2 * Q, 2 * Q, 2 * Q, 2 * Q, 2 * Q}};
const aligned_uint32x8_t PQCLEAN_DILITHIUM3_AVX2_8x256q = {.as_arr = {256 * Q, 256 * Q, 256 * Q, 256 * Q, 256 * Q, 256 * Q, 256 * Q,
256 * Q
}
};
const aligned_uint32x8_t _PQCLEAN_DILITHIUM3_AVX2_mask = {.as_arr = {0, 2, 4, 6, 0, 0, 0, 0}};
const aligned_uint32x8_t _PQCLEAN_DILITHIUM3_AVX2_8x23ones = {.as_arr = {0x7FFFFF, 0x7FFFFF, 0x7FFFFF, 0x7FFFFF, 0x7FFFFF, 0x7FFFFF,
const aligned_uint32x8_t PQCLEAN_DILITHIUM3_AVX2_mask = {.as_arr = {0, 2, 4, 6, 0, 0, 0, 0}};
const aligned_uint32x8_t PQCLEAN_DILITHIUM3_AVX2_8x23ones = {.as_arr = {0x7FFFFF, 0x7FFFFF, 0x7FFFFF, 0x7FFFFF, 0x7FFFFF, 0x7FFFFF,
0x7FFFFF, 0x7FFFFF
}
};
const aligned_uint32x8_t _PQCLEAN_DILITHIUM3_AVX2_8xdiv = { .as_arr = {DIV, DIV, DIV, DIV, DIV, DIV, DIV, DIV}};
const aligned_uint32x8_t PQCLEAN_DILITHIUM3_AVX2_8xdiv = { .as_arr = {DIV, DIV, DIV, DIV, DIV, DIV, DIV, DIV}};
#undef QINV
#undef MONT

14
crypto_sign/dilithium3/avx2/nttconsts.h
파일 보기

@ -12,13 +12,13 @@ typedef ALIGNED_UINT32(8) aligned_uint32x8_t;
typedef ALIGNED_UINT32(N) aligned_uint32xN_t;
extern const aligned_uint32x8_t _PQCLEAN_DILITHIUM3_AVX2_8xqinv;
extern const aligned_uint32x8_t _PQCLEAN_DILITHIUM3_AVX2_8xq;
extern const aligned_uint32x8_t _PQCLEAN_DILITHIUM3_AVX2_8x2q;
extern const aligned_uint32x8_t _PQCLEAN_DILITHIUM3_AVX2_8x256q;
extern const aligned_uint32x8_t _PQCLEAN_DILITHIUM3_AVX2_mask;
extern const aligned_uint32x8_t _PQCLEAN_DILITHIUM3_AVX2_8x23ones;
extern const aligned_uint32x8_t _PQCLEAN_DILITHIUM3_AVX2_8xdiv;
extern const aligned_uint32x8_t PQCLEAN_DILITHIUM3_AVX2_8xqinv;
extern const aligned_uint32x8_t PQCLEAN_DILITHIUM3_AVX2_8xq;
extern const aligned_uint32x8_t PQCLEAN_DILITHIUM3_AVX2_8x2q;
extern const aligned_uint32x8_t PQCLEAN_DILITHIUM3_AVX2_8x256q;
extern const aligned_uint32x8_t PQCLEAN_DILITHIUM3_AVX2_mask;
extern const aligned_uint32x8_t PQCLEAN_DILITHIUM3_AVX2_8x23ones;
extern const aligned_uint32x8_t PQCLEAN_DILITHIUM3_AVX2_8xdiv;
extern const aligned_uint32xN_t PQCLEAN_DILITHIUM3_AVX2_zetas;
extern const aligned_uint32xN_t PQCLEAN_DILITHIUM3_AVX2_zetas_inv;

8
crypto_sign/dilithium3/avx2/pointwise.S
파일 보기

@ -4,8 +4,8 @@
.global cdecl(PQCLEAN_DILITHIUM3_AVX2_pointwise_avx)
cdecl(PQCLEAN_DILITHIUM3_AVX2_pointwise_avx):
#consts
vmovdqa cdecl(_PQCLEAN_DILITHIUM3_AVX2_8xqinv)(%rip),%ymm0
vmovdqa cdecl(_PQCLEAN_DILITHIUM3_AVX2_8xq)(%rip),%ymm1
vmovdqa cdecl(PQCLEAN_DILITHIUM3_AVX2_8xqinv)(%rip),%ymm0
vmovdqa cdecl(PQCLEAN_DILITHIUM3_AVX2_8xq)(%rip),%ymm1
xor %eax,%eax
_looptop1:
@ -136,8 +136,8 @@ vpaddq %ymm9,%ymm5,%ymm5
.global cdecl(PQCLEAN_DILITHIUM3_AVX2_pointwise_acc_avx)
cdecl(PQCLEAN_DILITHIUM3_AVX2_pointwise_acc_avx):
#consts
vmovdqa cdecl(_PQCLEAN_DILITHIUM3_AVX2_8xqinv)(%rip),%ymm0
vmovdqa cdecl(_PQCLEAN_DILITHIUM3_AVX2_8xq)(%rip),%ymm1
vmovdqa cdecl(PQCLEAN_DILITHIUM3_AVX2_8xqinv)(%rip),%ymm0
vmovdqa cdecl(PQCLEAN_DILITHIUM3_AVX2_8xq)(%rip),%ymm1
xor %eax,%eax
_looptop2:

2
crypto_sign/dilithium3/avx2/poly.c
파일 보기

@ -82,7 +82,7 @@ void PQCLEAN_DILITHIUM3_AVX2_poly_add(poly *c, const poly *a, const poly *b) {
**************************************************/
void PQCLEAN_DILITHIUM3_AVX2_poly_sub(poly *c, const poly *a, const poly *b) {
__m256i vec0, vec1;
const __m256i twoq = _mm256_load_si256(_PQCLEAN_DILITHIUM3_AVX2_8x2q.as_vec);
const __m256i twoq = _mm256_load_si256(PQCLEAN_DILITHIUM3_AVX2_8x2q.as_vec);
for (size_t i = 0; i < N / 8; i++) {
vec0 = _mm256_load_si256(&a->coeffs_x8[i]);

4
crypto_sign/dilithium3/avx2/reduce.S
파일 보기

@ -3,7 +3,7 @@
.global cdecl(PQCLEAN_DILITHIUM3_AVX2_reduce_avx)
cdecl(PQCLEAN_DILITHIUM3_AVX2_reduce_avx):
#consts
vmovdqa cdecl(_PQCLEAN_DILITHIUM3_AVX2_8x23ones)(%rip),%ymm0
vmovdqa cdecl(PQCLEAN_DILITHIUM3_AVX2_8x23ones)(%rip),%ymm0
xor %eax,%eax
_looptop_rdc32:
@ -51,7 +51,7 @@ ret
.global cdecl(PQCLEAN_DILITHIUM3_AVX2_csubq_avx)
cdecl(PQCLEAN_DILITHIUM3_AVX2_csubq_avx):
#consts
vmovdqa cdecl(_PQCLEAN_DILITHIUM3_AVX2_8xq)(%rip),%ymm0
vmovdqa cdecl(PQCLEAN_DILITHIUM3_AVX2_8xq)(%rip),%ymm0
xor %eax,%eax
_looptop_csubq:

16
crypto_sign/dilithium4/avx2/invntt.S
파일 보기

@ -45,9 +45,9 @@ vpsrlq $32,%ymm\h3,%ymm\h3
.global cdecl(PQCLEAN_DILITHIUM4_AVX2_invntt_levels0t4_avx)
cdecl(PQCLEAN_DILITHIUM4_AVX2_invntt_levels0t4_avx):
#consts
vmovdqa cdecl(_PQCLEAN_DILITHIUM4_AVX2_8xqinv)(%rip),%ymm0
vmovdqa cdecl(_PQCLEAN_DILITHIUM4_AVX2_8xq)(%rip),%ymm1
vmovdqa cdecl(_PQCLEAN_DILITHIUM4_AVX2_8x256q)(%rip),%ymm2
vmovdqa cdecl(PQCLEAN_DILITHIUM4_AVX2_8xqinv)(%rip),%ymm0
vmovdqa cdecl(PQCLEAN_DILITHIUM4_AVX2_8xq)(%rip),%ymm1
vmovdqa cdecl(PQCLEAN_DILITHIUM4_AVX2_8x256q)(%rip),%ymm2
#load
vmovdqa (%rsi),%ymm6
@ -165,9 +165,9 @@ ret
.global cdecl(PQCLEAN_DILITHIUM4_AVX2_invntt_levels5t7_avx)
cdecl(PQCLEAN_DILITHIUM4_AVX2_invntt_levels5t7_avx):
#consts
vmovdqa cdecl(_PQCLEAN_DILITHIUM4_AVX2_8xqinv)(%rip),%ymm0
vmovdqa cdecl(_PQCLEAN_DILITHIUM4_AVX2_8xq)(%rip),%ymm1
vmovdqa cdecl(_PQCLEAN_DILITHIUM4_AVX2_8x256q)(%rip),%ymm2
vmovdqa cdecl(PQCLEAN_DILITHIUM4_AVX2_8xqinv)(%rip),%ymm0
vmovdqa cdecl(PQCLEAN_DILITHIUM4_AVX2_8xq)(%rip),%ymm1
vmovdqa cdecl(PQCLEAN_DILITHIUM4_AVX2_8x256q)(%rip),%ymm2
#load
vmovdqa (%rsi),%ymm4
@ -237,7 +237,7 @@ vpbroadcastd 24(%rdx),%ymm3
butterfly 4,5,6,7,8,9,10,11,3,3
#consts
vmovdqa cdecl(_PQCLEAN_DILITHIUM4_AVX2_8xdiv)(%rip),%ymm3
vmovdqa cdecl(PQCLEAN_DILITHIUM4_AVX2_8xdiv)(%rip),%ymm3
vpmuludq %ymm3,%ymm4,%ymm4
vpmuludq %ymm3,%ymm5,%ymm5
@ -261,7 +261,7 @@ vpsrlq $32,%ymm6,%ymm6
vpsrlq $32,%ymm7,%ymm7
#store
vmovdqa cdecl(_PQCLEAN_DILITHIUM4_AVX2_mask)(%rip),%ymm3
vmovdqa cdecl(PQCLEAN_DILITHIUM4_AVX2_mask)(%rip),%ymm3
vpermd %ymm4,%ymm3,%ymm4
vpermd %ymm5,%ymm3,%ymm5
vpermd %ymm6,%ymm3,%ymm6

12
crypto_sign/dilithium4/avx2/ntt.S
파일 보기

@ -44,9 +44,9 @@ vpsubd %ymm15,%ymm\rh3,%ymm\rh3
.global cdecl(PQCLEAN_DILITHIUM4_AVX2_ntt_levels0t2_avx)
cdecl(PQCLEAN_DILITHIUM4_AVX2_ntt_levels0t2_avx):
#consts
vmovdqa cdecl(_PQCLEAN_DILITHIUM4_AVX2_8xqinv)(%rip),%ymm0
vmovdqa cdecl(_PQCLEAN_DILITHIUM4_AVX2_8xq)(%rip),%ymm1
vmovdqa cdecl(_PQCLEAN_DILITHIUM4_AVX2_8x2q)(%rip),%ymm2
vmovdqa cdecl(PQCLEAN_DILITHIUM4_AVX2_8xqinv)(%rip),%ymm0
vmovdqa cdecl(PQCLEAN_DILITHIUM4_AVX2_8xq)(%rip),%ymm1
vmovdqa cdecl(PQCLEAN_DILITHIUM4_AVX2_8x2q)(%rip),%ymm2
level0:
#zetas
@ -95,9 +95,9 @@ ret
.global cdecl(PQCLEAN_DILITHIUM4_AVX2_ntt_levels3t8_avx)
cdecl(PQCLEAN_DILITHIUM4_AVX2_ntt_levels3t8_avx):
#consts
vmovdqa cdecl(_PQCLEAN_DILITHIUM4_AVX2_8xqinv)(%rip),%ymm0
vmovdqa cdecl(_PQCLEAN_DILITHIUM4_AVX2_8xq)(%rip),%ymm1
vmovdqa cdecl(_PQCLEAN_DILITHIUM4_AVX2_8x2q)(%rip),%ymm2
vmovdqa cdecl(PQCLEAN_DILITHIUM4_AVX2_8xqinv)(%rip),%ymm0
vmovdqa cdecl(PQCLEAN_DILITHIUM4_AVX2_8xq)(%rip),%ymm1
vmovdqa cdecl(PQCLEAN_DILITHIUM4_AVX2_8x2q)(%rip),%ymm2
#load
vmovdqa (%rsi),%ymm4

14
crypto_sign/dilithium4/avx2/nttconsts.c
파일 보기

@ -5,19 +5,19 @@
#define DIV (((MONT*MONT % Q) * (Q-1) % Q) * ((Q-1) >> 8) % Q)
const aligned_uint32x8_t _PQCLEAN_DILITHIUM4_AVX2_8xqinv = {.as_arr = {QINV, QINV, QINV, QINV, QINV, QINV, QINV, QINV}};
const aligned_uint32x8_t _PQCLEAN_DILITHIUM4_AVX2_8xq = {.as_arr = {Q, Q, Q, Q, Q, Q, Q, Q}};
const aligned_uint32x8_t _PQCLEAN_DILITHIUM4_AVX2_8x2q = {.as_arr = {2 * Q, 2 * Q, 2 * Q, 2 * Q, 2 * Q, 2 * Q, 2 * Q, 2 * Q}};
const aligned_uint32x8_t _PQCLEAN_DILITHIUM4_AVX2_8x256q = {.as_arr = {256 * Q, 256 * Q, 256 * Q, 256 * Q, 256 * Q, 256 * Q, 256 * Q,
const aligned_uint32x8_t PQCLEAN_DILITHIUM4_AVX2_8xqinv = {.as_arr = {QINV, QINV, QINV, QINV, QINV, QINV, QINV, QINV}};
const aligned_uint32x8_t PQCLEAN_DILITHIUM4_AVX2_8xq = {.as_arr = {Q, Q, Q, Q, Q, Q, Q, Q}};
const aligned_uint32x8_t PQCLEAN_DILITHIUM4_AVX2_8x2q = {.as_arr = {2 * Q, 2 * Q, 2 * Q, 2 * Q, 2 * Q, 2 * Q, 2 * Q, 2 * Q}};
const aligned_uint32x8_t PQCLEAN_DILITHIUM4_AVX2_8x256q = {.as_arr = {256 * Q, 256 * Q, 256 * Q, 256 * Q, 256 * Q, 256 * Q, 256 * Q,
256 * Q
}
};
const aligned_uint32x8_t _PQCLEAN_DILITHIUM4_AVX2_mask = {.as_arr = {0, 2, 4, 6, 0, 0, 0, 0}};
const aligned_uint32x8_t _PQCLEAN_DILITHIUM4_AVX2_8x23ones = {.as_arr = {0x7FFFFF, 0x7FFFFF, 0x7FFFFF, 0x7FFFFF, 0x7FFFFF, 0x7FFFFF,
const aligned_uint32x8_t PQCLEAN_DILITHIUM4_AVX2_mask = {.as_arr = {0, 2, 4, 6, 0, 0, 0, 0}};
const aligned_uint32x8_t PQCLEAN_DILITHIUM4_AVX2_8x23ones = {.as_arr = {0x7FFFFF, 0x7FFFFF, 0x7FFFFF, 0x7FFFFF, 0x7FFFFF, 0x7FFFFF,
0x7FFFFF, 0x7FFFFF
}
};
const aligned_uint32x8_t _PQCLEAN_DILITHIUM4_AVX2_8xdiv = { .as_arr = {DIV, DIV, DIV, DIV, DIV, DIV, DIV, DIV}};
const aligned_uint32x8_t PQCLEAN_DILITHIUM4_AVX2_8xdiv = { .as_arr = {DIV, DIV, DIV, DIV, DIV, DIV, DIV, DIV}};
#undef QINV
#undef MONT

14
crypto_sign/dilithium4/avx2/nttconsts.h
파일 보기

@ -12,13 +12,13 @@ typedef ALIGNED_UINT32(8) aligned_uint32x8_t;
typedef ALIGNED_UINT32(N) aligned_uint32xN_t;
extern const aligned_uint32x8_t _PQCLEAN_DILITHIUM4_AVX2_8xqinv;
extern const aligned_uint32x8_t _PQCLEAN_DILITHIUM4_AVX2_8xq;
extern const aligned_uint32x8_t _PQCLEAN_DILITHIUM4_AVX2_8x2q;
extern const aligned_uint32x8_t _PQCLEAN_DILITHIUM4_AVX2_8x256q;
extern const aligned_uint32x8_t _PQCLEAN_DILITHIUM4_AVX2_mask;
extern const aligned_uint32x8_t _PQCLEAN_DILITHIUM4_AVX2_8x23ones;
extern const aligned_uint32x8_t _PQCLEAN_DILITHIUM4_AVX2_8xdiv;
extern const aligned_uint32x8_t PQCLEAN_DILITHIUM4_AVX2_8xqinv;
extern const aligned_uint32x8_t PQCLEAN_DILITHIUM4_AVX2_8xq;
extern const aligned_uint32x8_t PQCLEAN_DILITHIUM4_AVX2_8x2q;
extern const aligned_uint32x8_t PQCLEAN_DILITHIUM4_AVX2_8x256q;
extern const aligned_uint32x8_t PQCLEAN_DILITHIUM4_AVX2_mask;
extern const aligned_uint32x8_t PQCLEAN_DILITHIUM4_AVX2_8x23ones;
extern const aligned_uint32x8_t PQCLEAN_DILITHIUM4_AVX2_8xdiv;
extern const aligned_uint32xN_t PQCLEAN_DILITHIUM4_AVX2_zetas;
extern const aligned_uint32xN_t PQCLEAN_DILITHIUM4_AVX2_zetas_inv;

8
crypto_sign/dilithium4/avx2/pointwise.S
파일 보기

@ -4,8 +4,8 @@
.global cdecl(PQCLEAN_DILITHIUM4_AVX2_pointwise_avx)
cdecl(PQCLEAN_DILITHIUM4_AVX2_pointwise_avx):
#consts
vmovdqa cdecl(_PQCLEAN_DILITHIUM4_AVX2_8xqinv)(%rip),%ymm0
vmovdqa cdecl(_PQCLEAN_DILITHIUM4_AVX2_8xq)(%rip),%ymm1
vmovdqa cdecl(PQCLEAN_DILITHIUM4_AVX2_8xqinv)(%rip),%ymm0
vmovdqa cdecl(PQCLEAN_DILITHIUM4_AVX2_8xq)(%rip),%ymm1
xor %eax,%eax
_looptop1:
@ -136,8 +136,8 @@ vpaddq %ymm9,%ymm5,%ymm5
.global cdecl(PQCLEAN_DILITHIUM4_AVX2_pointwise_acc_avx)
cdecl(PQCLEAN_DILITHIUM4_AVX2_pointwise_acc_avx):
#consts
vmovdqa cdecl(_PQCLEAN_DILITHIUM4_AVX2_8xqinv)(%rip),%ymm0
vmovdqa cdecl(_PQCLEAN_DILITHIUM4_AVX2_8xq)(%rip),%ymm1
vmovdqa cdecl(PQCLEAN_DILITHIUM4_AVX2_8xqinv)(%rip),%ymm0
vmovdqa cdecl(PQCLEAN_DILITHIUM4_AVX2_8xq)(%rip),%ymm1
xor %eax,%eax
_looptop2:

2
crypto_sign/dilithium4/avx2/poly.c
파일 보기

@ -82,7 +82,7 @@ void PQCLEAN_DILITHIUM4_AVX2_poly_add(poly *c, const poly *a, const poly *b) {
**************************************************/
void PQCLEAN_DILITHIUM4_AVX2_poly_sub(poly *c, const poly *a, const poly *b) {
__m256i vec0, vec1;
const __m256i twoq = _mm256_load_si256(_PQCLEAN_DILITHIUM4_AVX2_8x2q.as_vec);
const __m256i twoq = _mm256_load_si256(PQCLEAN_DILITHIUM4_AVX2_8x2q.as_vec);
for (size_t i = 0; i < N / 8; i++) {
vec0 = _mm256_load_si256(&a->coeffs_x8[i]);

4
crypto_sign/dilithium4/avx2/reduce.S
파일 보기

@ -3,7 +3,7 @@
.global cdecl(PQCLEAN_DILITHIUM4_AVX2_reduce_avx)
cdecl(PQCLEAN_DILITHIUM4_AVX2_reduce_avx):
#consts
vmovdqa cdecl(_PQCLEAN_DILITHIUM4_AVX2_8x23ones)(%rip),%ymm0
vmovdqa cdecl(PQCLEAN_DILITHIUM4_AVX2_8x23ones)(%rip),%ymm0
xor %eax,%eax
_looptop_rdc32:
@ -51,7 +51,7 @@ ret
.global cdecl(PQCLEAN_DILITHIUM4_AVX2_csubq_avx)
cdecl(PQCLEAN_DILITHIUM4_AVX2_csubq_avx):
#consts
vmovdqa cdecl(_PQCLEAN_DILITHIUM4_AVX2_8xq)(%rip),%ymm0
vmovdqa cdecl(PQCLEAN_DILITHIUM4_AVX2_8xq)(%rip),%ymm0
xor %eax,%eax
_looptop_csubq: