X-Git-Url: https://git.sesse.net/?a=blobdiff_plain;f=libavutil%2Faes.c;h=3ba5e9aef4501960b6489905314402d66ae53805;hb=cde7df25ef74b85b5ce11d9171779f28f0c12d15;hp=1cb139d9e176db90e3b961a361030b86d3f37473;hpb=179dc911607b69cf49073855d3b5ca27d3465fe7;p=ffmpeg diff --git a/libavutil/aes.c b/libavutil/aes.c index 1cb139d9e17..3ba5e9aef45 100644 --- a/libavutil/aes.c +++ b/libavutil/aes.c @@ -1,41 +1,61 @@ /* - * copyright (c) 2007 Michael Niedermayer and Reimar Doeffinger + * copyright (c) 2007 Michael Niedermayer * - * This file is part of FFmpeg. + * some optimization ideas from aes128.c by Reimar Doeffinger * - * FFmpeg is free software; you can redistribute it and/or + * This file is part of Libav. + * + * Libav is free software; you can redistribute it and/or * modify it under the terms of the GNU Lesser General Public * License as published by the Free Software Foundation; either * version 2.1 of the License, or (at your option) any later version. * - * FFmpeg is distributed in the hope that it will be useful, + * Libav is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU * Lesser General Public License for more details. * * You should have received a copy of the GNU Lesser General Public - * License along with FFmpeg; if not, write to the Free Software + * License along with Libav; if not, write to the Free Software * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA */ #include "common.h" -#include "log.h" #include "aes.h" - -typedef struct AVAES{ - uint8_t round_enc_key[15][4][4]; - uint8_t round_dec_key[15][4][4]; - uint8_t state[4][4]; +#include "intreadwrite.h" +#include "timer.h" + +typedef union { + uint64_t u64[2]; + uint32_t u32[4]; + uint8_t u8x4[4][4]; + uint8_t u8[16]; +} av_aes_block; + +typedef struct AVAES { + // Note: round_key[16] is accessed in the init code, but this only + // overwrites state, which does not matter (see also commit ba554c0). + av_aes_block round_key[15]; + av_aes_block state[2]; int rounds; -}AVAES; +} AVAES; -static const uint8_t rcon[11] = { - 0x01, 0x02, 0x04, 0x08, 0x10, 0x20, 0x40, 0x80, 0x1b, 0x36, 0x6c +#if FF_API_CONTEXT_SIZE +const int av_aes_size= sizeof(AVAES); +#endif + +struct AVAES *av_aes_alloc(void) +{ + return av_mallocz(sizeof(struct AVAES)); +} + +static const uint8_t rcon[10] = { + 0x01, 0x02, 0x04, 0x08, 0x10, 0x20, 0x40, 0x80, 0x1b, 0x36 }; static uint8_t sbox[256]; static uint8_t inv_sbox[256]; -#ifdef CONFIG_SMALL +#if CONFIG_SMALL static uint32_t enc_multbl[1][256]; static uint32_t dec_multbl[1][256]; #else @@ -43,207 +63,280 @@ static uint32_t enc_multbl[4][256]; static uint32_t dec_multbl[4][256]; #endif -static inline void addkey(uint64_t state[2], uint64_t round_key[2]){ - state[0] ^= round_key[0]; - state[1] ^= round_key[1]; -} +#if HAVE_BIGENDIAN +# define ROT(x, s) ((x >> s) | (x << (32-s))) +#else +# define ROT(x, s) ((x << s) | (x >> (32-s))) +#endif -#define SUBSHIFT0(s, box) s[0]=box[s[ 0]]; s[ 4]=box[s[ 4]]; s[ 8]=box[s[ 8]]; s[12]=box[s[12]]; -#define SUBSHIFT1(s, box) t=s[0]; s[0]=box[s[ 4]]; s[ 4]=box[s[ 8]]; s[ 8]=box[s[12]]; s[12]=box[t]; -#define SUBSHIFT2(s, box) t=s[0]; s[0]=box[s[ 8]]; s[ 8]=box[ t]; t=s[ 4]; s[ 4]=box[s[12]]; s[12]=box[t]; -#define SUBSHIFT3(s, box) t=s[0]; s[0]=box[s[12]]; s[12]=box[s[ 8]]; s[ 8]=box[s[ 4]]; s[ 4]=box[t]; +static inline void addkey(av_aes_block *dst, const av_aes_block *src, + const av_aes_block *round_key) +{ + dst->u64[0] = src->u64[0] ^ round_key->u64[0]; + dst->u64[1] = src->u64[1] ^ round_key->u64[1]; +} -#define SUBSHIFT1x(s) t=s[0]; s[0]=s[ 4]; s[ 4]=s[ 8]; s[ 8]=s[12]; s[12]=t; -#define SUBSHIFT2x(s) t=s[0]; s[0]=s[ 8]; s[ 8]= t; t=s[ 4]; s[ 4]=s[12]; s[12]=t; -#define SUBSHIFT3x(s) t=s[0]; s[0]=s[12]; s[12]=s[ 8]; s[ 8]=s[ 4]; s[ 4]=t; +static inline void addkey_s(av_aes_block *dst, const uint8_t *src, + const av_aes_block *round_key) +{ + dst->u64[0] = AV_RN64(src) ^ round_key->u64[0]; + dst->u64[1] = AV_RN64(src + 8) ^ round_key->u64[1]; +} -#define ROT(x,s) ((x<>(32-s))) +static inline void addkey_d(uint8_t *dst, const av_aes_block *src, + const av_aes_block *round_key) +{ + AV_WN64(dst, src->u64[0] ^ round_key->u64[0]); + AV_WN64(dst + 8, src->u64[1] ^ round_key->u64[1]); +} -static inline void mix(uint8_t state[4][4], uint32_t multbl[4][256]){ - int i; - for(i=0; i<4; i++) -#ifdef CONFIG_SMALL - ((uint32_t *)(state))[i] = multbl[0][state[i][0]] ^ ROT(multbl[0][state[i][1]], 8) - ^ROT(multbl[0][state[i][2]],16) ^ ROT(multbl[0][state[i][3]],24); +static void subshift(av_aes_block s0[2], int s, const uint8_t *box) +{ + av_aes_block *s1 = (av_aes_block *) (s0[0].u8 - s); + av_aes_block *s3 = (av_aes_block *) (s0[0].u8 + s); + + s0[0].u8[ 0] = box[s0[1].u8[ 0]]; + s0[0].u8[ 4] = box[s0[1].u8[ 4]]; + s0[0].u8[ 8] = box[s0[1].u8[ 8]]; + s0[0].u8[12] = box[s0[1].u8[12]]; + s1[0].u8[ 3] = box[s1[1].u8[ 7]]; + s1[0].u8[ 7] = box[s1[1].u8[11]]; + s1[0].u8[11] = box[s1[1].u8[15]]; + s1[0].u8[15] = box[s1[1].u8[ 3]]; + s0[0].u8[ 2] = box[s0[1].u8[10]]; + s0[0].u8[10] = box[s0[1].u8[ 2]]; + s0[0].u8[ 6] = box[s0[1].u8[14]]; + s0[0].u8[14] = box[s0[1].u8[ 6]]; + s3[0].u8[ 1] = box[s3[1].u8[13]]; + s3[0].u8[13] = box[s3[1].u8[ 9]]; + s3[0].u8[ 9] = box[s3[1].u8[ 5]]; + s3[0].u8[ 5] = box[s3[1].u8[ 1]]; +} +static inline int mix_core(uint32_t multbl[][256], int a, int b, int c, int d){ +#if CONFIG_SMALL + return multbl[0][a] ^ ROT(multbl[0][b], 8) ^ ROT(multbl[0][c], 16) ^ ROT(multbl[0][d], 24); #else - ((uint32_t *)(state))[i] = multbl[0][state[i][0]] ^ multbl[1][state[i][1]] - ^multbl[2][state[i][2]] ^ multbl[3][state[i][3]]; + return multbl[0][a] ^ multbl[1][b] ^ multbl[2][c] ^ multbl[3][d]; #endif } -void av_aes_decrypt(AVAES *a){ - int t, r; +static inline void mix(av_aes_block state[2], uint32_t multbl[][256], int s1, int s3){ + uint8_t (*src)[4] = state[1].u8x4; + state[0].u32[0] = mix_core(multbl, src[0][0], src[s1 ][1], src[2][2], src[s3 ][3]); + state[0].u32[1] = mix_core(multbl, src[1][0], src[s3-1][1], src[3][2], src[s1-1][3]); + state[0].u32[2] = mix_core(multbl, src[2][0], src[s3 ][1], src[0][2], src[s1 ][3]); + state[0].u32[3] = mix_core(multbl, src[3][0], src[s1-1][1], src[1][2], src[s3-1][3]); +} + +static inline void crypt(AVAES *a, int s, const uint8_t *sbox, + uint32_t multbl[][256]) +{ + int r; - addkey(a->state, a->round_enc_key[a->rounds]); - for(r=a->rounds-2; r>=0; r--){ - SUBSHIFT3x((a->state[0]+1)) - SUBSHIFT2x((a->state[0]+2)) - SUBSHIFT1x((a->state[0]+3)) - mix(a->state, dec_multbl); - addkey(a->state, a->round_dec_key[r+1]); + for (r = a->rounds - 1; r > 0; r--) { + mix(a->state, multbl, 3 - s, 1 + s); + addkey(&a->state[1], &a->state[0], &a->round_key[r]); } - SUBSHIFT0((a->state[0]+0), inv_sbox) - SUBSHIFT3((a->state[0]+1), inv_sbox) - SUBSHIFT2((a->state[0]+2), inv_sbox) - SUBSHIFT1((a->state[0]+3), inv_sbox) - addkey(a->state, a->round_enc_key[0]); -} -void av_aes_encrypt(AVAES *a){ - int r, t; + subshift(&a->state[0], s, sbox); +} - for(r=0; rrounds-1; r++){ - addkey(a->state, a->round_enc_key[r]); - SUBSHIFT1x((a->state[0]+1)) - SUBSHIFT2x((a->state[0]+2)) - SUBSHIFT3x((a->state[0]+3)) - mix(a->state, enc_multbl); //FIXME replace log8 by const / optimze mix as this can be simplified alot +void av_aes_crypt(AVAES *a, uint8_t *dst, const uint8_t *src, + int count, uint8_t *iv, int decrypt) +{ + while (count--) { + addkey_s(&a->state[1], src, &a->round_key[a->rounds]); + if (decrypt) { + crypt(a, 0, inv_sbox, dec_multbl); + if (iv) { + addkey_s(&a->state[0], iv, &a->state[0]); + memcpy(iv, src, 16); + } + addkey_d(dst, &a->state[0], &a->round_key[0]); + } else { + if (iv) + addkey_s(&a->state[1], iv, &a->state[1]); + crypt(a, 2, sbox, enc_multbl); + addkey_d(dst, &a->state[0], &a->round_key[0]); + if (iv) + memcpy(iv, dst, 16); + } + src += 16; + dst += 16; } - addkey(a->state, a->round_enc_key[r]); - SUBSHIFT0((a->state[0]+0), sbox) - SUBSHIFT1((a->state[0]+1), sbox) - SUBSHIFT2((a->state[0]+2), sbox) - SUBSHIFT3((a->state[0]+3), sbox) - addkey(a->state, a->round_enc_key[r+1]); } -static init_multbl2(uint8_t tbl[1024], int c[4], uint8_t *log8, uint8_t *alog8, uint8_t *sbox){ +static void init_multbl2(uint32_t tbl[][256], const int c[4], + const uint8_t *log8, const uint8_t *alog8, + const uint8_t *sbox) +{ int i; - for(i=0; i<1024; i++){ - int x= sbox[i/4]; - if(x) tbl[i]= alog8[ log8[x] + log8[c[i&3]] ]; + + for (i = 0; i < 256; i++) { + int x = sbox[i]; + if (x) { + int k, l, m, n; + x = log8[x]; + k = alog8[x + log8[c[0]]]; + l = alog8[x + log8[c[1]]]; + m = alog8[x + log8[c[2]]]; + n = alog8[x + log8[c[3]]]; + tbl[0][i] = AV_NE(MKBETAG(k,l,m,n), MKTAG(k,l,m,n)); +#if !CONFIG_SMALL + tbl[1][i] = ROT(tbl[0][i], 8); + tbl[2][i] = ROT(tbl[0][i], 16); + tbl[3][i] = ROT(tbl[0][i], 24); +#endif + } } } - // this is based on the reference AES code by Paulo Barreto and Vincent Rijmen -AVAES *av_aes_init(uint8_t *key, int key_bits) { - AVAES *a; +int av_aes_init(AVAES *a, const uint8_t *key, int key_bits, int decrypt) +{ int i, j, t, rconpointer = 0; uint8_t tk[8][4]; - int KC= key_bits/32; - int rounds= KC + 6; - uint8_t log8[256]; + int KC = key_bits >> 5; + int rounds = KC + 6; + uint8_t log8[256]; uint8_t alog8[512]; - if(!sbox[255]){ - j=1; - for(i=0; i<255; i++){ - alog8[i]= - alog8[i+255]= j; - log8[j]= i; - j^= j+j; - if(j>255) j^= 0x11B; + if (!enc_multbl[FF_ARRAY_ELEMS(enc_multbl)-1][FF_ARRAY_ELEMS(enc_multbl[0])-1]) { + j = 1; + for (i = 0; i < 255; i++) { + alog8[i] = alog8[i + 255] = j; + log8[j] = i; + j ^= j + j; + if (j > 255) + j ^= 0x11B; } - log8[0]= 255; - for(i=0; i<256; i++){ - j= i ? alog8[255-log8[i]] : 0; - j ^= (j<<1) ^ (j<<2) ^ (j<<3) ^ (j<<4); - j = (j ^ (j>>8) ^ 99) & 255; - inv_sbox[j]= i; - sbox [i]= j; -// av_log(NULL, AV_LOG_ERROR, "%d, ", log8[i]); + for (i = 0; i < 256; i++) { + j = i ? alog8[255 - log8[i]] : 0; + j ^= (j << 1) ^ (j << 2) ^ (j << 3) ^ (j << 4); + j = (j ^ (j >> 8) ^ 99) & 255; + inv_sbox[j] = i; + sbox[i] = j; } - init_multbl2(dec_multbl[0], (int[4]){0xe, 0x9, 0xd, 0xb}, log8, alog8, inv_sbox); -#ifndef CONFIG_SMALL - init_multbl2(dec_multbl[1], (int[4]){0xb, 0xe, 0x9, 0xd}, log8, alog8, inv_sbox); - init_multbl2(dec_multbl[2], (int[4]){0xd, 0xb, 0xe, 0x9}, log8, alog8, inv_sbox); - init_multbl2(dec_multbl[3], (int[4]){0x9, 0xd, 0xb, 0xe}, log8, alog8, inv_sbox); -#endif - init_multbl2(enc_multbl[0], (int[4]){0x2, 0x1, 0x1, 0x3}, log8, alog8, sbox); -#ifndef CONFIG_SMALL - init_multbl2(enc_multbl[1], (int[4]){0x3, 0x2, 0x1, 0x1}, log8, alog8, sbox); - init_multbl2(enc_multbl[2], (int[4]){0x1, 0x3, 0x2, 0x1}, log8, alog8, sbox); - init_multbl2(enc_multbl[3], (int[4]){0x1, 0x1, 0x3, 0x2}, log8, alog8, sbox); -#endif + init_multbl2(dec_multbl, (const int[4]) { 0xe, 0x9, 0xd, 0xb }, + log8, alog8, inv_sbox); + init_multbl2(enc_multbl, (const int[4]) { 0x2, 0x1, 0x1, 0x3 }, + log8, alog8, sbox); } - if(key_bits!=128 && key_bits!=192 && key_bits!=256) - return NULL; + if (key_bits != 128 && key_bits != 192 && key_bits != 256) + return -1; - a= av_malloc(sizeof(AVAES)); - a->rounds= rounds; + a->rounds = rounds; - memcpy(tk, key, KC*4); + memcpy(tk, key, KC * 4); + memcpy(a->round_key[0].u8, key, KC * 4); - for(t= 0; t < (rounds+1)*4;) { - memcpy(a->round_enc_key[0][t], tk, KC*4); - t+= KC; - - for(i = 0; i < 4; i++) - tk[0][i] ^= sbox[tk[KC-1][(i+1)&3]]; + for (t = KC * 4; t < (rounds + 1) * 16; t += KC * 4) { + for (i = 0; i < 4; i++) + tk[0][i] ^= sbox[tk[KC - 1][(i + 1) & 3]]; tk[0][0] ^= rcon[rconpointer++]; - for(j = 1; j < KC; j++){ - if(KC != 8 || j != KC/2) - for(i = 0; i < 4; i++) tk[j][i] ^= tk[j-1][i]; + for (j = 1; j < KC; j++) { + if (KC != 8 || j != KC >> 1) + for (i = 0; i < 4; i++) + tk[j][i] ^= tk[j - 1][i]; else - for(i = 0; i < 4; i++) tk[j][i] ^= sbox[tk[j-1][i]]; + for (i = 0; i < 4; i++) + tk[j][i] ^= sbox[tk[j - 1][i]]; } + + memcpy(a->round_key[0].u8 + t, tk, KC * 4); } - for(i=0; iround_enc_key); i++) - a->round_dec_key[0][0][i]= sbox[a->round_enc_key[0][0][i]]; - for(i=1; iround_dec_key[i], dec_multbl); + if (decrypt) { + for (i = 1; i < rounds; i++) { + av_aes_block tmp[3]; + tmp[2] = a->round_key[i]; + subshift(&tmp[1], 0, sbox); + mix(tmp, dec_multbl, 1, 3); + a->round_key[i] = tmp[0]; + } + } else { + for (i = 0; i < (rounds + 1) >> 1; i++) { + FFSWAP(av_aes_block, a->round_key[i], a->round_key[rounds-i]); + } + } - return a; + return 0; } #ifdef TEST +#include +#include "lfg.h" +#include "log.h" -int main(){ - int i,j,k; - AVAES *a= av_aes_init("PI=3.141592654..", 128); - uint8_t zero[16]= {0}; - uint8_t pt[16]= {0x6a, 0x84, 0x86, 0x7c, 0xd7, 0x7e, 0x12, 0xad, 0x07, 0xea, 0x1b, 0xe8, 0x95, 0xc5, 0x3f, 0xa3}; - uint8_t ct[16]= {0x73, 0x22, 0x81, 0xc0, 0xa0, 0xaa, 0xb8, 0xf7, 0xa5, 0x4a, 0x0c, 0x67, 0xa0, 0xc4, 0x5e, 0xcf}; - AVAES *b= av_aes_init(zero, 128); - -/* uint8_t key[16]= {0x42, 0x78, 0xb8, 0x40, 0xfb, 0x44, 0xaa, 0xa7, 0x57, 0xc1, 0xbf, 0x04, 0xac, 0xbe, 0x1a, 0x3e}; - uint8_t IV[16] = {0x57, 0xf0, 0x2a, 0x5c, 0x53, 0x39, 0xda, 0xeb, 0x0a, 0x29, 0x08, 0xa0, 0x6a, 0xc6, 0x39, 0x3f}; - uint8_t pt[16] = {0x3c, 0x88, 0x8b, 0xbb, 0xb1, 0xa8, 0xeb, 0x9f, 0x3e, 0x9b, 0x87, 0xac, 0xaa, 0xd9, 0x86, 0xc4}; -// 66e2f7071c83083b8a557971918850e5 - uint8_t ct[16] = {0x47, 0x9c, 0x89, 0xec, 0x14, 0xbc, 0x98, 0x99, 0x4e, 0x62, 0xb2, 0xc7, 0x05, 0xb5, 0x0, 0x14e}; -// 175bd7832e7e60a1e92aac568a861eb7*/ - uint8_t ckey[16]= {0x10, 0xa5, 0x88, 0x69, 0xd7, 0x4b, 0xe5, 0xa3, 0x74, 0xcf, 0x86, 0x7c, 0xfb, 0x47, 0x38, 0x59}; - uint8_t cct[16] = {0x6d, 0x25, 0x1e, 0x69, 0x44, 0xb0, 0x51, 0xe0, 0x4e, 0xaa, 0x6f, 0xb4, 0xdb, 0xf7, 0x84, 0x65}; - AVAES *c= av_aes_init(ckey, 128); - - av_log_level= AV_LOG_DEBUG; - - memcpy(b->state, ct, 16); - av_aes_decrypt(b); - for(j=0; j<16; j++) - if(pt[j] != b->state[0][j]){ - av_log(NULL, AV_LOG_ERROR, "%d %02X %02X\n", j, pt[j], b->state[0][j]); +int main(int argc, char **argv) +{ + int i, j; + AVAES b; + uint8_t rkey[2][16] = { + { 0 }, + { 0x10, 0xa5, 0x88, 0x69, 0xd7, 0x4b, 0xe5, 0xa3, + 0x74, 0xcf, 0x86, 0x7c, 0xfb, 0x47, 0x38, 0x59 } + }; + uint8_t pt[16], rpt[2][16]= { + { 0x6a, 0x84, 0x86, 0x7c, 0xd7, 0x7e, 0x12, 0xad, + 0x07, 0xea, 0x1b, 0xe8, 0x95, 0xc5, 0x3f, 0xa3 }, + { 0 } + }; + uint8_t rct[2][16]= { + { 0x73, 0x22, 0x81, 0xc0, 0xa0, 0xaa, 0xb8, 0xf7, + 0xa5, 0x4a, 0x0c, 0x67, 0xa0, 0xc4, 0x5e, 0xcf }, + { 0x6d, 0x25, 0x1e, 0x69, 0x44, 0xb0, 0x51, 0xe0, + 0x4e, 0xaa, 0x6f, 0xb4, 0xdb, 0xf7, 0x84, 0x65 } + }; + uint8_t temp[16]; + int err = 0; + + av_log_set_level(AV_LOG_DEBUG); + + for (i = 0; i < 2; i++) { + av_aes_init(&b, rkey[i], 128, 1); + av_aes_crypt(&b, temp, rct[i], 1, NULL, 1); + for (j = 0; j < 16; j++) { + if (rpt[i][j] != temp[j]) { + av_log(NULL, AV_LOG_ERROR, "%d %02X %02X\n", + j, rpt[i][j], temp[j]); + err = 1; + } } + } - memcpy(c->state, cct, 16); - av_aes_decrypt(c); - for(j=0; j<16; j++) - if(zero[j] != c->state[0][j]){ - av_log(NULL, AV_LOG_ERROR, "%d %02X %02X\n", j, zero[j], c->state[0][j]); - } + if (argc > 1 && !strcmp(argv[1], "-t")) { + AVAES ae, ad; + AVLFG prng; - for(i=0; i<10000; i++){ - for(j=0; j<16; j++){ - pt[j]= random(); - } - memcpy(a->state, pt, 16); -{START_TIMER - av_aes_encrypt(a); - if(!(i&(i-1))) - av_log(NULL, AV_LOG_ERROR, "%02X %02X %02X %02X\n", a->state[0][0], a->state[1][1], a->state[2][2], a->state[3][3]); - av_aes_decrypt(a); -STOP_TIMER("aes")} - for(j=0; j<16; j++){ - if(pt[j] != a->state[0][j]){ - av_log(NULL, AV_LOG_ERROR, "%d %d %02X %02X\n", i,j, pt[j], a->state[0][j]); + av_aes_init(&ae, "PI=3.141592654..", 128, 0); + av_aes_init(&ad, "PI=3.141592654..", 128, 1); + av_lfg_init(&prng, 1); + + for (i = 0; i < 10000; i++) { + for (j = 0; j < 16; j++) { + pt[j] = av_lfg_get(&prng); + } + { + START_TIMER; + av_aes_crypt(&ae, temp, pt, 1, NULL, 0); + if (!(i & (i - 1))) + av_log(NULL, AV_LOG_ERROR, "%02X %02X %02X %02X\n", + temp[0], temp[5], temp[10], temp[15]); + av_aes_crypt(&ad, temp, temp, 1, NULL, 1); + STOP_TIMER("aes"); + } + for (j = 0; j < 16; j++) { + if (pt[j] != temp[j]) { + av_log(NULL, AV_LOG_ERROR, "%d %d %02X %02X\n", + i, j, pt[j], temp[j]); + } } } } - return 0; + return err; } #endif