static uint32_t dec_multbl[4][256];
#endif
-static inline void addkey(uint64_t state[2], uint64_t round_key[2]){
- state[2] = state[0] ^ round_key[0];
- state[3] = state[1] ^ round_key[1];
+static inline void addkey(uint64_t dst[2], uint64_t src[2], uint64_t round_key[2]){
+ dst[0] = src[0] ^ round_key[0];
+ dst[1] = src[1] ^ round_key[1];
}
static void subshift(uint8_t s0[2][16], int s, uint8_t *box){
s3[0][1]=box[s3[1][13]]; s3[0][13]=box[s3[1][ 9]]; s3[0][ 9]=box[s3[1][ 5]]; s3[0][ 5]=box[s3[1][ 1]];
}
-#define ROT(x,s) ((x<<s)|(x>>(32-s)))
-#if 0
-static inline void mix(uint8_t state[4][4], uint32_t multbl[4][256]){
- int i;
- for(i=0; i<4; i++)
+static inline int mix_core(uint32_t multbl[4][256], int a, int b, int c, int d){
#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);
+#define ROT(x,s) ((x<<s)|(x>>(32-s)))
+ 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
}
-#endif
static inline void mix(uint8_t state[2][4][4], uint32_t multbl[4][256], int s1, int s3){
- ((uint32_t *)(state))[0] = multbl[0][state[1][0][0]] ^ multbl[1][state[1][s1 ][1]]
- ^multbl[2][state[1][2][2]] ^ multbl[3][state[1][s3 ][3]];
- ((uint32_t *)(state))[1] = multbl[0][state[1][1][0]] ^ multbl[1][state[1][s3-1][1]]
- ^multbl[2][state[1][3][2]] ^ multbl[3][state[1][s1-1][3]];
- ((uint32_t *)(state))[2] = multbl[0][state[1][2][0]] ^ multbl[1][state[1][s3 ][1]]
- ^multbl[2][state[1][0][2]] ^ multbl[3][state[1][s1 ][3]];
- ((uint32_t *)(state))[3] = multbl[0][state[1][3][0]] ^ multbl[1][state[1][s1-1][1]]
- ^multbl[2][state[1][1][2]] ^ multbl[3][state[1][s3-1][3]];
+ ((uint32_t *)(state))[0] = mix_core(multbl, state[1][0][0], state[1][s1 ][1], state[1][2][2], state[1][s3 ][3]);
+ ((uint32_t *)(state))[1] = mix_core(multbl, state[1][1][0], state[1][s3-1][1], state[1][3][2], state[1][s1-1][3]);
+ ((uint32_t *)(state))[2] = mix_core(multbl, state[1][2][0], state[1][s3 ][1], state[1][0][2], state[1][s1 ][3]);
+ ((uint32_t *)(state))[3] = mix_core(multbl, state[1][3][0], state[1][s1-1][1], state[1][1][2], state[1][s3-1][3]);
}
static inline void crypt(AVAES *a, int s, uint8_t *sbox, uint32_t *multbl){
int r;
- for(r=a->rounds; r>1; r--){
- addkey(a->state, a->round_key[r]);
+ 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]);
}
- addkey(a->state, a->round_key[1]);
subshift(a->state[0][0], s, sbox);
- addkey(a->state, a->round_key[0]);
-}
-
-static void aes_decrypt(AVAES *a){
- crypt(a, 0, inv_sbox, dec_multbl);
}
-static void aes_encrypt(AVAES *a){
- crypt(a, 2, sbox, enc_multbl);
+void av_aes_crypt(AVAES *a, uint8_t *dst, uint8_t *src, int count, uint8_t *iv, int decrypt){
+ while(count--){
+ addkey(a->state[1], src, a->round_key[a->rounds]);
+ if(decrypt) {
+ crypt(a, 0, inv_sbox, dec_multbl);
+ if(iv){
+ addkey(a->state[0], a->state[0], iv);
+ memcpy(iv, src, 16);
+ }
+ addkey(dst, a->state[0], a->round_key[0]);
+ }else{
+ if(iv) addkey(a->state[1], a->state[1], iv);
+ crypt(a, 2, sbox, enc_multbl);
+ addkey(dst, a->state[0], a->round_key[0]);
+ if(iv) memcpy(iv, dst, 16);
+ }
+ src+=16;
+ dst+=16;
+ }
}
static void init_multbl2(uint8_t tbl[1024], int c[4], uint8_t *log8, uint8_t *alog8, uint8_t *sbox){
}
// this is based on the reference AES code by Paulo Barreto and Vincent Rijmen
-int av_aes_init(AVAES *a, uint8_t *key, int key_bits, int decrypt) {
+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>>5;
memcpy(tk, key, KC*4);
- for(t= 0; t < (rounds+1)*4;) {
- memcpy(a->round_key[0][t], tk, KC*4);
- t+= KC;
+ for(t= 0; t < (rounds+1)*16;) {
+ memcpy(a->round_key[0][0]+t, tk, KC*4);
+ t+= KC*4;
for(i = 0; i < 4; i++)
tk[0][i] ^= sbox[tk[KC-1][(i+1)&3]];
if(decrypt){
for(i=1; i<rounds; i++){
- uint8_t tmp[2][16];
- memcpy(tmp[1], a->round_key[i][0], 16);
- subshift(tmp[0], 0, sbox);
- memcpy(tmp[1], tmp[0], 16);
+ uint8_t tmp[3][16];
+ memcpy(tmp[2], a->round_key[i][0], 16);
+ subshift(tmp[1], 0, sbox);
mix(tmp, dec_multbl, 1, 3);
memcpy(a->round_key[i][0], tmp[0], 16);
}
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];
av_aes_init(&ae, "PI=3.141592654..", 128, 0);
av_aes_init(&ad, "PI=3.141592654..", 128, 1);
for(i=0; i<2; i++){
av_aes_init(&b, rkey[i], 128, 1);
- memcpy(b.state, rct[i], 16);
- aes_decrypt(&b);
+ av_aes_crypt(&b, temp, rct[i], 1, NULL, 1);
for(j=0; j<16; j++)
- if(rpt[i][j] != b.state[1][0][j])
- av_log(NULL, AV_LOG_ERROR, "%d %02X %02X\n", j, rpt[i][j], b.state[1][0][j]);
+ if(rpt[i][j] != temp[j])
+ av_log(NULL, AV_LOG_ERROR, "%d %02X %02X\n", j, rpt[i][j], temp[j]);
}
for(i=0; i<10000; i++){
for(j=0; j<16; j++){
pt[j]= random();
}
- memcpy(ae.state, pt, 16);
{START_TIMER
- aes_encrypt(&ae);
+ av_aes_crypt(&ae, temp, pt, 1, NULL, 0);
if(!(i&(i-1)))
- av_log(NULL, AV_LOG_ERROR, "%02X %02X %02X %02X\n", ae.state[1][0][0], ae.state[1][1][1], ae.state[1][2][2], ae.state[1][3][3]);
- memcpy(ad.state[0], ae.state[1], 16);
- aes_decrypt(&ad);
+ 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] != ad.state[1][0][j]){
- av_log(NULL, AV_LOG_ERROR, "%d %d %02X %02X\n", i,j, pt[j], ad.state[1][0][j]);
+ if(pt[j] != temp[j]){
+ av_log(NULL, AV_LOG_ERROR, "%d %d %02X %02X\n", i,j, pt[j], temp[j]);
}
}
}