abc/GmSSL
1
0
Fork 0
mirror of https://github.com/guanzhi/GmSSL.git synced 2026-05-07 08:56:17 +08:00
Code Issues Packages Projects Releases Wiki Activity

Update gf128_avx.c

Browse Source
This commit is contained in:
Zhi Guan
2022-12-22 10:27:46 +08:00
parent 868f197f09
commit fe0712090d
1 changed files with 50 additions and 59 deletions
Download Patch File Download Diff File Expand all files Collapse all files

109
src/gf128_avx.c
View File

@@ -115,103 +115,96 @@ void gf128_to_bytes(gf128_t a, uint8_t p[16])
gf128_t gf128_add(gf128_t ga, gf128_t gb)
{
uint8_t r[16], a[16], b[16];
gf128_to_bytes(ga, a);
gf128_to_bytes(gb, b);
__m128i a1 = _mm_loadu_si128((const __m128i*)a);
__m128i b1 = _mm_loadu_si128((const __m128i*)b);
__m128i a1 = _mm_loadu_si128((const __m128i*)a);
__m128i b1 = _mm_loadu_si128((const __m128i*)b);
__m128i T0 = _mm_xor_si128(a1, b1);
_mm_storeu_si128((__m128i*)r, T0);
return gf128_from_bytes(r);
}
gf128_t gf128_mul(gf128_t ga, gf128_t gb)
{
uint8_t r[16], a[16], b[16];
{
const __m128i MASK = _mm_set_epi8(0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15);
__m128i a1, b1;
__m128i T0, T1, T2, T3, T4, T5;
uint8_t r[16], a[16], b[16];
gf128_to_bytes(ga, a);
gf128_to_bytes(gb, b);
__m128i a1 = _mm_loadu_si128((const __m128i*)a);
__m128i b1 = _mm_loadu_si128((const __m128i*)b);
a1 = _mm_loadu_si128((const __m128i*)a);
b1 = _mm_loadu_si128((const __m128i*)b);
a1 = _mm_shuffle_epi8(a1, MASK);
b1 = _mm_shuffle_epi8(b1, MASK);
a1 = _mm_shuffle_epi8(a1, MASK);
b1 = _mm_shuffle_epi8(b1, MASK);
__m128i T0, T1, T2, T3, T4, T5;
T0 = _mm_clmulepi64_si128(a1, b1, 0x00);
T1 = _mm_clmulepi64_si128(a1, b1, 0x01);
T2 = _mm_clmulepi64_si128(a1, b1, 0x10);
T3 = _mm_clmulepi64_si128(a1, b1, 0x11);
T0 = _mm_clmulepi64_si128(a1, b1, 0x00);
T1 = _mm_clmulepi64_si128(a1, b1, 0x01);
T2 = _mm_clmulepi64_si128(a1, b1, 0x10);
T3 = _mm_clmulepi64_si128(a1, b1, 0x11);
T1 = _mm_xor_si128(T1, T2);
T2 = _mm_slli_si128(T1, 8);
T1 = _mm_srli_si128(T1, 8);
T0 = _mm_xor_si128(T0, T2);
T3 = _mm_xor_si128(T3, T1);
T1 = _mm_xor_si128(T1, T2);
T2 = _mm_slli_si128(T1, 8);
T1 = _mm_srli_si128(T1, 8);
T0 = _mm_xor_si128(T0, T2);
T3 = _mm_xor_si128(T3, T1);
T4 = _mm_srli_epi32(T0, 31);
T0 = _mm_slli_epi32(T0, 1);
T4 = _mm_srli_epi32(T0, 31);
T0 = _mm_slli_epi32(T0, 1);
T5 = _mm_srli_epi32(T3, 31);
T3 = _mm_slli_epi32(T3, 1);
T5 = _mm_srli_epi32(T3, 31);
T3 = _mm_slli_epi32(T3, 1);
T2 = _mm_srli_si128(T4, 12);
T5 = _mm_slli_si128(T5, 4);
T4 = _mm_slli_si128(T4, 4);
T0 = _mm_or_si128(T0, T4);
T3 = _mm_or_si128(T3, T5);
T3 = _mm_or_si128(T3, T2);
T2 = _mm_srli_si128(T4, 12);
T5 = _mm_slli_si128(T5, 4);
T4 = _mm_slli_si128(T4, 4);
T0 = _mm_or_si128(T0, T4);
T3 = _mm_or_si128(T3, T5);
T3 = _mm_or_si128(T3, T2);
T4 = _mm_slli_epi32(T0, 31);
T5 = _mm_slli_epi32(T0, 30);
T2 = _mm_slli_epi32(T0, 25);
T4 = _mm_slli_epi32(T0, 31);
T5 = _mm_slli_epi32(T0, 30);
T2 = _mm_slli_epi32(T0, 25);
T4 = _mm_xor_si128(T4, T5);
T4 = _mm_xor_si128(T4, T2);
T5 = _mm_srli_si128(T4, 4);
T3 = _mm_xor_si128(T3, T5);
T4 = _mm_slli_si128(T4, 12);
T0 = _mm_xor_si128(T0, T4);
T3 = _mm_xor_si128(T3, T0);
T4 = _mm_xor_si128(T4, T5);
T4 = _mm_xor_si128(T4, T2);
T5 = _mm_srli_si128(T4, 4);
T3 = _mm_xor_si128(T3, T5);
T4 = _mm_slli_si128(T4, 12);
T0 = _mm_xor_si128(T0, T4);
T3 = _mm_xor_si128(T3, T0);
T4 = _mm_srli_epi32(T0, 1);
T1 = _mm_srli_epi32(T0, 2);
T2 = _mm_srli_epi32(T0, 7);
T3 = _mm_xor_si128(T3, T1);
T3 = _mm_xor_si128(T3, T2);
T3 = _mm_xor_si128(T3, T4);
T3 = _mm_shuffle_epi8(T3, MASK);
_mm_storeu_si128((__m128i*)r, T3);
T4 = _mm_srli_epi32(T0, 1);
T1 = _mm_srli_epi32(T0, 2);
T2 = _mm_srli_epi32(T0, 7);
T3 = _mm_xor_si128(T3, T1);
T3 = _mm_xor_si128(T3, T2);
T3 = _mm_xor_si128(T3, T4);
T3 = _mm_shuffle_epi8(T3, MASK);
_mm_storeu_si128((__m128i*)r, T3);
return gf128_from_bytes(r);
}
gf128_t gf128_mul2(gf128_t ga)
{
uint8_t r[16], a[16];
const __m128i MASK = _mm_set_epi8(0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15);
__m128i MASK1 = _mm_set_epi8(0xe1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0);
__m128i MASK2 = _mm_set_epi8(0x80,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0);
__m128i a1;
__m128i T0, T1, T2, T3, T4, T5;
uint8_t r[16], a[16];
gf128_to_bytes(ga, a);
__m128i a1 = _mm_loadu_si128((const __m128i*)a);
a1 = _mm_loadu_si128((const __m128i*)a);
a1 = _mm_shuffle_epi8(a1, MASK);
T0 = _mm_srli_epi64(a1,1);
@@ -227,8 +220,6 @@ gf128_t gf128_mul2(gf128_t ga)
T5 = _mm_xor_si128(T5,T3);
T5 = _mm_shuffle_epi8(T5, MASK);
_mm_storeu_si128((__m128i*)r, T5);
return gf128_from_bytes(r);
}