| 1 |
// 20250302 djb: port to -Wc++-compat |
| 2 |
#include <immintrin.h> |
| 3 |
#include "crypto_decode.h" |
| 4 |
#include "crypto_uint8.h" |
| 5 |
#define uint8 crypto_uint8 |
| 6 |
|
| 7 |
#define p 761 |
| 8 |
#define loops 6 |
| 9 |
#define overshoot 2 |
| 10 |
|
| 11 |
void crypto_decode(void *v,const unsigned char *s) |
| 12 |
{ |
| 13 |
uint8 *f = (uint8 *) v; |
| 14 |
int loop; |
| 15 |
uint8 *nextf = f+128-4*overshoot; |
| 16 |
const unsigned char *nexts = s+32-overshoot; |
| 17 |
|
| 18 |
for (loop = loops;loop > 0;--loop) { |
| 19 |
__m256i s0 = _mm256_loadu_si256((const __m256i *) s); |
| 20 |
s = nexts; |
| 21 |
nexts += 32; |
| 22 |
|
| 23 |
__m256i s1 = _mm256_srli_epi16(s0&_mm256_set1_epi8(-16),4); |
| 24 |
s0 &= _mm256_set1_epi8(15); |
| 25 |
|
| 26 |
__m256i a0 = _mm256_unpacklo_epi8(s0,s1); |
| 27 |
/* 0 0>>4 1 1>>4 2 2>>4 3 3>>4 4 4>>4 5 5>>4 6 6>>4 7 7>>4 */ |
| 28 |
/* 16 16>>4 ... */ |
| 29 |
__m256i a1 = _mm256_unpackhi_epi8(s0,s1); |
| 30 |
/* 8 8>>4 9 9>>4 10 10>>4 ... */ |
| 31 |
/* 24 24>>4 ... */ |
| 32 |
|
| 33 |
__m256i a2 = _mm256_srli_epi16(a0&_mm256_set1_epi8(12),2); |
| 34 |
__m256i a3 = _mm256_srli_epi16(a1&_mm256_set1_epi8(12),2); |
| 35 |
a0 &= _mm256_set1_epi8(3); |
| 36 |
a1 &= _mm256_set1_epi8(3); |
| 37 |
|
| 38 |
__m256i b0 = _mm256_unpacklo_epi8(a0,a2); |
| 39 |
/* 0 0>>2 0>>4 0>>6 1 1>>2 1>>4 1>>6 */ |
| 40 |
/* 2 2>>2 2>>4 2>>6 3 3>>2 3>>4 3>.6 */ |
| 41 |
/* 16 16>>2 16>>4 16>>6 ... */ |
| 42 |
__m256i b2 = _mm256_unpackhi_epi8(a0,a2); |
| 43 |
/* 4 4>>2 ... */ |
| 44 |
__m256i b1 = _mm256_unpacklo_epi8(a1,a3); |
| 45 |
/* 8 8>>2 ... */ |
| 46 |
__m256i b3 = _mm256_unpackhi_epi8(a1,a3); |
| 47 |
/* 12 12>>2 ... */ |
| 48 |
|
| 49 |
__m256i f0 = _mm256_permute2x128_si256(b0,b2,0x20); |
| 50 |
__m256i f2 = _mm256_permute2x128_si256(b0,b2,0x31); |
| 51 |
__m256i f1 = _mm256_permute2x128_si256(b1,b3,0x20); |
| 52 |
__m256i f3 = _mm256_permute2x128_si256(b1,b3,0x31); |
| 53 |
|
| 54 |
f0 = _mm256_add_epi8(f0,_mm256_set1_epi8(-1)); |
| 55 |
f1 = _mm256_add_epi8(f1,_mm256_set1_epi8(-1)); |
| 56 |
f2 = _mm256_add_epi8(f2,_mm256_set1_epi8(-1)); |
| 57 |
f3 = _mm256_add_epi8(f3,_mm256_set1_epi8(-1)); |
| 58 |
|
| 59 |
_mm256_storeu_si256((__m256i *) (f+0),f0); |
| 60 |
_mm256_storeu_si256((__m256i *) (f+32),f1); |
| 61 |
_mm256_storeu_si256((__m256i *) (f+64),f2); |
| 62 |
_mm256_storeu_si256((__m256i *) (f+96),f3); |
| 63 |
f = nextf; |
| 64 |
nextf += 128; |
| 65 |
} |
| 66 |
|
| 67 |
*f = ((uint8)(*s&3))-1; |
| 68 |
} |