yading@11: /*
yading@11:  * copyright (c) 2007 Michael Niedermayer <michaelni@gmx.at>
yading@11:  *
yading@11:  * some optimization ideas from aes128.c by Reimar Doeffinger
yading@11:  *
yading@11:  * This file is part of FFmpeg.
yading@11:  *
yading@11:  * FFmpeg is free software; you can redistribute it and/or
yading@11:  * modify it under the terms of the GNU Lesser General Public
yading@11:  * License as published by the Free Software Foundation; either
yading@11:  * version 2.1 of the License, or (at your option) any later version.
yading@11:  *
yading@11:  * FFmpeg is distributed in the hope that it will be useful,
yading@11:  * but WITHOUT ANY WARRANTY; without even the implied warranty of
yading@11:  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
yading@11:  * Lesser General Public License for more details.
yading@11:  *
yading@11:  * You should have received a copy of the GNU Lesser General Public
yading@11:  * License along with FFmpeg; if not, write to the Free Software
yading@11:  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
yading@11:  */
yading@11: 
yading@11: #include "common.h"
yading@11: #include "aes.h"
yading@11: #include "intreadwrite.h"
yading@11: 
yading@11: typedef union {
yading@11:     uint64_t u64[2];
yading@11:     uint32_t u32[4];
yading@11:     uint8_t u8x4[4][4];
yading@11:     uint8_t u8[16];
yading@11: } av_aes_block;
yading@11: 
yading@11: typedef struct AVAES {
yading@11:     // Note: round_key[16] is accessed in the init code, but this only
yading@11:     // overwrites state, which does not matter (see also commit ba554c0).
yading@11:     av_aes_block round_key[15];
yading@11:     av_aes_block state[2];
yading@11:     int rounds;
yading@11: } AVAES;
yading@11: 
yading@11: const int av_aes_size= sizeof(AVAES);
yading@11: 
yading@11: struct AVAES *av_aes_alloc(void)
yading@11: {
yading@11:     return av_mallocz(sizeof(struct AVAES));
yading@11: }
yading@11: 
yading@11: static const uint8_t rcon[10] = {
yading@11:   0x01, 0x02, 0x04, 0x08, 0x10, 0x20, 0x40, 0x80, 0x1b, 0x36
yading@11: };
yading@11: 
yading@11: static uint8_t     sbox[256];
yading@11: static uint8_t inv_sbox[256];
yading@11: #if CONFIG_SMALL
yading@11: static uint32_t enc_multbl[1][256];
yading@11: static uint32_t dec_multbl[1][256];
yading@11: #else
yading@11: static uint32_t enc_multbl[4][256];
yading@11: static uint32_t dec_multbl[4][256];
yading@11: #endif
yading@11: 
yading@11: #if HAVE_BIGENDIAN
yading@11: #   define ROT(x, s) ((x >> s) | (x << (32-s)))
yading@11: #else
yading@11: #   define ROT(x, s) ((x << s) | (x >> (32-s)))
yading@11: #endif
yading@11: 
yading@11: static inline void addkey(av_aes_block *dst, const av_aes_block *src,
yading@11:                           const av_aes_block *round_key)
yading@11: {
yading@11:     dst->u64[0] = src->u64[0] ^ round_key->u64[0];
yading@11:     dst->u64[1] = src->u64[1] ^ round_key->u64[1];
yading@11: }
yading@11: 
yading@11: static inline void addkey_s(av_aes_block *dst, const uint8_t *src,
yading@11:                             const av_aes_block *round_key)
yading@11: {
yading@11:     dst->u64[0] = AV_RN64(src)     ^ round_key->u64[0];
yading@11:     dst->u64[1] = AV_RN64(src + 8) ^ round_key->u64[1];
yading@11: }
yading@11: 
yading@11: static inline void addkey_d(uint8_t *dst, const av_aes_block *src,
yading@11:                             const av_aes_block *round_key)
yading@11: {
yading@11:     AV_WN64(dst,     src->u64[0] ^ round_key->u64[0]);
yading@11:     AV_WN64(dst + 8, src->u64[1] ^ round_key->u64[1]);
yading@11: }
yading@11: 
yading@11: static void subshift(av_aes_block s0[2], int s, const uint8_t *box)
yading@11: {
yading@11:     av_aes_block *s1 = (av_aes_block *) (s0[0].u8 - s);
yading@11:     av_aes_block *s3 = (av_aes_block *) (s0[0].u8 + s);
yading@11: 
yading@11:     s0[0].u8[ 0] = box[s0[1].u8[ 0]];
yading@11:     s0[0].u8[ 4] = box[s0[1].u8[ 4]];
yading@11:     s0[0].u8[ 8] = box[s0[1].u8[ 8]];
yading@11:     s0[0].u8[12] = box[s0[1].u8[12]];
yading@11:     s1[0].u8[ 3] = box[s1[1].u8[ 7]];
yading@11:     s1[0].u8[ 7] = box[s1[1].u8[11]];
yading@11:     s1[0].u8[11] = box[s1[1].u8[15]];
yading@11:     s1[0].u8[15] = box[s1[1].u8[ 3]];
yading@11:     s0[0].u8[ 2] = box[s0[1].u8[10]];
yading@11:     s0[0].u8[10] = box[s0[1].u8[ 2]];
yading@11:     s0[0].u8[ 6] = box[s0[1].u8[14]];
yading@11:     s0[0].u8[14] = box[s0[1].u8[ 6]];
yading@11:     s3[0].u8[ 1] = box[s3[1].u8[13]];
yading@11:     s3[0].u8[13] = box[s3[1].u8[ 9]];
yading@11:     s3[0].u8[ 9] = box[s3[1].u8[ 5]];
yading@11:     s3[0].u8[ 5] = box[s3[1].u8[ 1]];
yading@11: }
yading@11: 
yading@11: static inline int mix_core(uint32_t multbl[][256], int a, int b, int c, int d){
yading@11: #if CONFIG_SMALL
yading@11:     return multbl[0][a] ^ ROT(multbl[0][b], 8) ^ ROT(multbl[0][c], 16) ^ ROT(multbl[0][d], 24);
yading@11: #else
yading@11:     return multbl[0][a] ^ multbl[1][b] ^ multbl[2][c] ^ multbl[3][d];
yading@11: #endif
yading@11: }
yading@11: 
yading@11: static inline void mix(av_aes_block state[2], uint32_t multbl[][256], int s1, int s3){
yading@11:     uint8_t (*src)[4] = state[1].u8x4;
yading@11:     state[0].u32[0] = mix_core(multbl, src[0][0], src[s1  ][1], src[2][2], src[s3  ][3]);
yading@11:     state[0].u32[1] = mix_core(multbl, src[1][0], src[s3-1][1], src[3][2], src[s1-1][3]);
yading@11:     state[0].u32[2] = mix_core(multbl, src[2][0], src[s3  ][1], src[0][2], src[s1  ][3]);
yading@11:     state[0].u32[3] = mix_core(multbl, src[3][0], src[s1-1][1], src[1][2], src[s3-1][3]);
yading@11: }
yading@11: 
yading@11: static inline void crypt(AVAES *a, int s, const uint8_t *sbox,
yading@11:                          uint32_t multbl[][256])
yading@11: {
yading@11:     int r;
yading@11: 
yading@11:     for (r = a->rounds - 1; r > 0; r--) {
yading@11:         mix(a->state, multbl, 3 - s, 1 + s);
yading@11:         addkey(&a->state[1], &a->state[0], &a->round_key[r]);
yading@11:     }
yading@11: 
yading@11:     subshift(&a->state[0], s, sbox);
yading@11: }
yading@11: 
yading@11: void av_aes_crypt(AVAES *a, uint8_t *dst, const uint8_t *src,
yading@11:                   int count, uint8_t *iv, int decrypt)
yading@11: {
yading@11:     while (count--) {
yading@11:         addkey_s(&a->state[1], src, &a->round_key[a->rounds]);
yading@11:         if (decrypt) {
yading@11:             crypt(a, 0, inv_sbox, dec_multbl);
yading@11:             if (iv) {
yading@11:                 addkey_s(&a->state[0], iv, &a->state[0]);
yading@11:                 memcpy(iv, src, 16);
yading@11:             }
yading@11:             addkey_d(dst, &a->state[0], &a->round_key[0]);
yading@11:         } else {
yading@11:             if (iv)
yading@11:                 addkey_s(&a->state[1], iv, &a->state[1]);
yading@11:             crypt(a, 2, sbox, enc_multbl);
yading@11:             addkey_d(dst, &a->state[0], &a->round_key[0]);
yading@11:             if (iv)
yading@11:                 memcpy(iv, dst, 16);
yading@11:         }
yading@11:         src += 16;
yading@11:         dst += 16;
yading@11:     }
yading@11: }
yading@11: 
yading@11: static void init_multbl2(uint32_t tbl[][256], const int c[4],
yading@11:                          const uint8_t *log8, const uint8_t *alog8,
yading@11:                          const uint8_t *sbox)
yading@11: {
yading@11:     int i;
yading@11: 
yading@11:     for (i = 0; i < 256; i++) {
yading@11:         int x = sbox[i];
yading@11:         if (x) {
yading@11:             int k, l, m, n;
yading@11:             x = log8[x];
yading@11:             k = alog8[x + log8[c[0]]];
yading@11:             l = alog8[x + log8[c[1]]];
yading@11:             m = alog8[x + log8[c[2]]];
yading@11:             n = alog8[x + log8[c[3]]];
yading@11:             tbl[0][i] = AV_NE(MKBETAG(k,l,m,n), MKTAG(k,l,m,n));
yading@11: #if !CONFIG_SMALL
yading@11:             tbl[1][i] = ROT(tbl[0][i], 8);
yading@11:             tbl[2][i] = ROT(tbl[0][i], 16);
yading@11:             tbl[3][i] = ROT(tbl[0][i], 24);
yading@11: #endif
yading@11:         }
yading@11:     }
yading@11: }
yading@11: 
yading@11: // this is based on the reference AES code by Paulo Barreto and Vincent Rijmen
yading@11: int av_aes_init(AVAES *a, const uint8_t *key, int key_bits, int decrypt)
yading@11: {
yading@11:     int i, j, t, rconpointer = 0;
yading@11:     uint8_t tk[8][4];
yading@11:     int KC = key_bits >> 5;
yading@11:     int rounds = KC + 6;
yading@11:     uint8_t log8[256];
yading@11:     uint8_t alog8[512];
yading@11: 
yading@11:     if (!enc_multbl[FF_ARRAY_ELEMS(enc_multbl)-1][FF_ARRAY_ELEMS(enc_multbl[0])-1]) {
yading@11:         j = 1;
yading@11:         for (i = 0; i < 255; i++) {
yading@11:             alog8[i] = alog8[i + 255] = j;
yading@11:             log8[j] = i;
yading@11:             j ^= j + j;
yading@11:             if (j > 255)
yading@11:                 j ^= 0x11B;
yading@11:         }
yading@11:         for (i = 0; i < 256; i++) {
yading@11:             j = i ? alog8[255 - log8[i]] : 0;
yading@11:             j ^= (j << 1) ^ (j << 2) ^ (j << 3) ^ (j << 4);
yading@11:             j = (j ^ (j >> 8) ^ 99) & 255;
yading@11:             inv_sbox[j] = i;
yading@11:             sbox[i] = j;
yading@11:         }
yading@11:         init_multbl2(dec_multbl, (const int[4]) { 0xe, 0x9, 0xd, 0xb },
yading@11:                      log8, alog8, inv_sbox);
yading@11:         init_multbl2(enc_multbl, (const int[4]) { 0x2, 0x1, 0x1, 0x3 },
yading@11:                      log8, alog8, sbox);
yading@11:     }
yading@11: 
yading@11:     if (key_bits != 128 && key_bits != 192 && key_bits != 256)
yading@11:         return -1;
yading@11: 
yading@11:     a->rounds = rounds;
yading@11: 
yading@11:     memcpy(tk, key, KC * 4);
yading@11:     memcpy(a->round_key[0].u8, key, KC * 4);
yading@11: 
yading@11:     for (t = KC * 4; t < (rounds + 1) * 16; t += KC * 4) {
yading@11:         for (i = 0; i < 4; i++)
yading@11:             tk[0][i] ^= sbox[tk[KC - 1][(i + 1) & 3]];
yading@11:         tk[0][0] ^= rcon[rconpointer++];
yading@11: 
yading@11:         for (j = 1; j < KC; j++) {
yading@11:             if (KC != 8 || j != KC >> 1)
yading@11:                 for (i = 0; i < 4; i++)
yading@11:                     tk[j][i] ^= tk[j - 1][i];
yading@11:             else
yading@11:                 for (i = 0; i < 4; i++)
yading@11:                     tk[j][i] ^= sbox[tk[j - 1][i]];
yading@11:         }
yading@11: 
yading@11:         memcpy(a->round_key[0].u8 + t, tk, KC * 4);
yading@11:     }
yading@11: 
yading@11:     if (decrypt) {
yading@11:         for (i = 1; i < rounds; i++) {
yading@11:             av_aes_block tmp[3];
yading@11:             tmp[2] = a->round_key[i];
yading@11:             subshift(&tmp[1], 0, sbox);
yading@11:             mix(tmp, dec_multbl, 1, 3);
yading@11:             a->round_key[i] = tmp[0];
yading@11:         }
yading@11:     } else {
yading@11:         for (i = 0; i < (rounds + 1) >> 1; i++) {
yading@11:             FFSWAP(av_aes_block, a->round_key[i], a->round_key[rounds-i]);
yading@11:         }
yading@11:     }
yading@11: 
yading@11:     return 0;
yading@11: }
yading@11: 
yading@11: #ifdef TEST
yading@11: // LCOV_EXCL_START
yading@11: #include <string.h>
yading@11: #include "lfg.h"
yading@11: #include "log.h"
yading@11: 
yading@11: int main(int argc, char **argv)
yading@11: {
yading@11:     int i, j;
yading@11:     AVAES b;
yading@11:     uint8_t rkey[2][16] = {
yading@11:         { 0 },
yading@11:         { 0x10, 0xa5, 0x88, 0x69, 0xd7, 0x4b, 0xe5, 0xa3,
yading@11:           0x74, 0xcf, 0x86, 0x7c, 0xfb, 0x47, 0x38, 0x59 }
yading@11:     };
yading@11:     uint8_t pt[16], rpt[2][16]= {
yading@11:         { 0x6a, 0x84, 0x86, 0x7c, 0xd7, 0x7e, 0x12, 0xad,
yading@11:           0x07, 0xea, 0x1b, 0xe8, 0x95, 0xc5, 0x3f, 0xa3 },
yading@11:         { 0 }
yading@11:     };
yading@11:     uint8_t rct[2][16]= {
yading@11:         { 0x73, 0x22, 0x81, 0xc0, 0xa0, 0xaa, 0xb8, 0xf7,
yading@11:           0xa5, 0x4a, 0x0c, 0x67, 0xa0, 0xc4, 0x5e, 0xcf },
yading@11:         { 0x6d, 0x25, 0x1e, 0x69, 0x44, 0xb0, 0x51, 0xe0,
yading@11:           0x4e, 0xaa, 0x6f, 0xb4, 0xdb, 0xf7, 0x84, 0x65 }
yading@11:     };
yading@11:     uint8_t temp[16];
yading@11:     int err = 0;
yading@11: 
yading@11:     av_log_set_level(AV_LOG_DEBUG);
yading@11: 
yading@11:     for (i = 0; i < 2; i++) {
yading@11:         av_aes_init(&b, rkey[i], 128, 1);
yading@11:         av_aes_crypt(&b, temp, rct[i], 1, NULL, 1);
yading@11:         for (j = 0; j < 16; j++) {
yading@11:             if (rpt[i][j] != temp[j]) {
yading@11:                 av_log(NULL, AV_LOG_ERROR, "%d %02X %02X\n",
yading@11:                        j, rpt[i][j], temp[j]);
yading@11:                 err = 1;
yading@11:             }
yading@11:         }
yading@11:     }
yading@11: 
yading@11:     if (argc > 1 && !strcmp(argv[1], "-t")) {
yading@11:         AVAES ae, ad;
yading@11:         AVLFG prng;
yading@11: 
yading@11:         av_aes_init(&ae, "PI=3.141592654..", 128, 0);
yading@11:         av_aes_init(&ad, "PI=3.141592654..", 128, 1);
yading@11:         av_lfg_init(&prng, 1);
yading@11: 
yading@11:         for (i = 0; i < 10000; i++) {
yading@11:             for (j = 0; j < 16; j++) {
yading@11:                 pt[j] = av_lfg_get(&prng);
yading@11:             }
yading@11:             {
yading@11:                 START_TIMER;
yading@11:                 av_aes_crypt(&ae, temp, pt, 1, NULL, 0);
yading@11:                 if (!(i & (i - 1)))
yading@11:                     av_log(NULL, AV_LOG_ERROR, "%02X %02X %02X %02X\n",
yading@11:                            temp[0], temp[5], temp[10], temp[15]);
yading@11:                 av_aes_crypt(&ad, temp, temp, 1, NULL, 1);
yading@11:                 STOP_TIMER("aes");
yading@11:             }
yading@11:             for (j = 0; j < 16; j++) {
yading@11:                 if (pt[j] != temp[j]) {
yading@11:                     av_log(NULL, AV_LOG_ERROR, "%d %d %02X %02X\n",
yading@11:                            i, j, pt[j], temp[j]);
yading@11:                 }
yading@11:             }
yading@11:         }
yading@11:     }
yading@11:     return err;
yading@11: }
yading@11: // LCOV_EXCL_STOP
yading@11: #endif