Mercurial > libavutil.hg
comparison aes.c @ 955:040bf8859042 libavutil
Chnage AES code to be strict-aliasing-safe.
Makes it give correct results with e.g. gcc 4.4
For unknown reasons the generate asm code also changes
on e.g. gcc 4.3, making the code a bit larger and a bit
slower.
author | reimar |
---|---|
date | Wed, 30 Jun 2010 04:50:35 +0000 |
parents | 2b38811998d1 |
children | 90ee1cca76a2 |
comparison
equal
deleted
inserted
replaced
954:2b38811998d1 | 955:040bf8859042 |
---|---|
21 */ | 21 */ |
22 | 22 |
23 #include "common.h" | 23 #include "common.h" |
24 #include "aes.h" | 24 #include "aes.h" |
25 | 25 |
26 typedef union { | |
27 uint64_t u64[2]; | |
28 uint32_t u32[4]; | |
29 uint8_t u8x4[4][4]; | |
30 uint8_t u8[16]; | |
31 } av_aes_block; | |
32 | |
26 typedef struct AVAES{ | 33 typedef struct AVAES{ |
27 // Note: round_key[16] is accessed in the init code, but this only | 34 // Note: round_key[16] is accessed in the init code, but this only |
28 // overwrites state, which does not matter (see also r7471). | 35 // overwrites state, which does not matter (see also r7471). |
29 uint8_t round_key[15][4][4]; | 36 av_aes_block round_key[15]; |
30 uint8_t state[2][4][4]; | 37 av_aes_block state[2]; |
31 int rounds; | 38 int rounds; |
32 }AVAES; | 39 }AVAES; |
33 | 40 |
34 const int av_aes_size= sizeof(AVAES); | 41 const int av_aes_size= sizeof(AVAES); |
35 | 42 |
45 #else | 52 #else |
46 static uint32_t enc_multbl[4][256]; | 53 static uint32_t enc_multbl[4][256]; |
47 static uint32_t dec_multbl[4][256]; | 54 static uint32_t dec_multbl[4][256]; |
48 #endif | 55 #endif |
49 | 56 |
50 static inline void addkey(uint64_t dst[2], const uint64_t src[2], const uint64_t round_key[2]){ | 57 static inline void addkey(av_aes_block *dst, const av_aes_block *src, const av_aes_block *round_key){ |
51 dst[0] = src[0] ^ round_key[0]; | 58 dst->u64[0] = src->u64[0] ^ round_key->u64[0]; |
52 dst[1] = src[1] ^ round_key[1]; | 59 dst->u64[1] = src->u64[1] ^ round_key->u64[1]; |
53 } | 60 } |
54 | 61 |
55 static void subshift(uint8_t s0[2][16], int s, const uint8_t *box){ | 62 static void subshift(av_aes_block s0[2], int s, const uint8_t *box){ |
56 uint8_t (*s1)[16]= s0[0] - s; | 63 av_aes_block *s1= (av_aes_block *)(s0[0].u8 - s); |
57 uint8_t (*s3)[16]= s0[0] + s; | 64 av_aes_block *s3= (av_aes_block *)(s0[0].u8 + s); |
58 s0[0][0]=box[s0[1][ 0]]; s0[0][ 4]=box[s0[1][ 4]]; s0[0][ 8]=box[s0[1][ 8]]; s0[0][12]=box[s0[1][12]]; | 65 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]]; |
59 s1[0][3]=box[s1[1][ 7]]; s1[0][ 7]=box[s1[1][11]]; s1[0][11]=box[s1[1][15]]; s1[0][15]=box[s1[1][ 3]]; | 66 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]]; |
60 s0[0][2]=box[s0[1][10]]; s0[0][10]=box[s0[1][ 2]]; s0[0][ 6]=box[s0[1][14]]; s0[0][14]=box[s0[1][ 6]]; | 67 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]]; |
61 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]]; | 68 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]]; |
62 } | 69 } |
63 | 70 |
64 static inline int mix_core(uint32_t multbl[][256], int a, int b, int c, int d){ | 71 static inline int mix_core(uint32_t multbl[][256], int a, int b, int c, int d){ |
65 #if CONFIG_SMALL | 72 #if CONFIG_SMALL |
66 #define ROT(x,s) ((x<<s)|(x>>(32-s))) | 73 #define ROT(x,s) ((x<<s)|(x>>(32-s))) |
68 #else | 75 #else |
69 return multbl[0][a] ^ multbl[1][b] ^ multbl[2][c] ^ multbl[3][d]; | 76 return multbl[0][a] ^ multbl[1][b] ^ multbl[2][c] ^ multbl[3][d]; |
70 #endif | 77 #endif |
71 } | 78 } |
72 | 79 |
73 static inline void mix(uint8_t state[2][4][4], uint32_t multbl[][256], int s1, int s3){ | 80 static inline void mix(av_aes_block state[2], uint32_t multbl[][256], int s1, int s3){ |
74 ((uint32_t *)(state))[0] = mix_core(multbl, state[1][0][0], state[1][s1 ][1], state[1][2][2], state[1][s3 ][3]); | 81 state[0].u32[0] = mix_core(multbl, state[1].u8x4[0][0], state[1].u8x4[s1 ][1], state[1].u8x4[2][2], state[1].u8x4[s3 ][3]); |
75 ((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]); | 82 state[0].u32[1] = mix_core(multbl, state[1].u8x4[1][0], state[1].u8x4[s3-1][1], state[1].u8x4[3][2], state[1].u8x4[s1-1][3]); |
76 ((uint32_t *)(state))[2] = mix_core(multbl, state[1][2][0], state[1][s3 ][1], state[1][0][2], state[1][s1 ][3]); | 83 state[0].u32[2] = mix_core(multbl, state[1].u8x4[2][0], state[1].u8x4[s3 ][1], state[1].u8x4[0][2], state[1].u8x4[s1 ][3]); |
77 ((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]); | 84 state[0].u32[3] = mix_core(multbl, state[1].u8x4[3][0], state[1].u8x4[s1-1][1], state[1].u8x4[1][2], state[1].u8x4[s3-1][3]); |
78 } | 85 } |
79 | 86 |
80 static inline void crypt(AVAES *a, int s, const uint8_t *sbox, uint32_t multbl[][256]){ | 87 static inline void crypt(AVAES *a, int s, const uint8_t *sbox, uint32_t multbl[][256]){ |
81 int r; | 88 int r; |
82 | 89 |
83 for(r=a->rounds-1; r>0; r--){ | 90 for(r=a->rounds-1; r>0; r--){ |
84 mix(a->state, multbl, 3-s, 1+s); | 91 mix(a->state, multbl, 3-s, 1+s); |
85 addkey(a->state[1], a->state[0], a->round_key[r]); | 92 addkey(&a->state[1], &a->state[0], &a->round_key[r]); |
86 } | 93 } |
87 subshift(a->state[0][0], s, sbox); | 94 subshift(&a->state[0], s, sbox); |
88 } | 95 } |
89 | 96 |
90 void av_aes_crypt(AVAES *a, uint8_t *dst, const uint8_t *src, int count, uint8_t *iv, int decrypt){ | 97 void av_aes_crypt(AVAES *a, uint8_t *dst_, const uint8_t *src_, int count, uint8_t *iv_, int decrypt){ |
98 av_aes_block *dst = (av_aes_block *)dst_; | |
99 const av_aes_block *src = (const av_aes_block *)src_; | |
100 av_aes_block *iv = (av_aes_block *)iv_; | |
91 while(count--){ | 101 while(count--){ |
92 addkey(a->state[1], src, a->round_key[a->rounds]); | 102 addkey(&a->state[1], src, &a->round_key[a->rounds]); |
93 if(decrypt) { | 103 if(decrypt) { |
94 crypt(a, 0, inv_sbox, dec_multbl); | 104 crypt(a, 0, inv_sbox, dec_multbl); |
95 if(iv){ | 105 if(iv){ |
96 addkey(a->state[0], a->state[0], iv); | 106 addkey(&a->state[0], &a->state[0], iv); |
97 memcpy(iv, src, 16); | 107 memcpy(iv, src, 16); |
98 } | 108 } |
99 addkey(dst, a->state[0], a->round_key[0]); | 109 addkey(dst, &a->state[0], &a->round_key[0]); |
100 }else{ | 110 }else{ |
101 if(iv) addkey(a->state[1], a->state[1], iv); | 111 if(iv) addkey(&a->state[1], &a->state[1], iv); |
102 crypt(a, 2, sbox, enc_multbl); | 112 crypt(a, 2, sbox, enc_multbl); |
103 addkey(dst, a->state[0], a->round_key[0]); | 113 addkey(dst, &a->state[0], &a->round_key[0]); |
104 if(iv) memcpy(iv, dst, 16); | 114 if(iv) memcpy(iv, dst, 16); |
105 } | 115 } |
106 src+=16; | 116 src++; |
107 dst+=16; | 117 dst++; |
108 } | 118 } |
109 } | 119 } |
110 | 120 |
111 static void init_multbl2(uint8_t tbl[1024], const int c[4], const uint8_t *log8, const uint8_t *alog8, const uint8_t *sbox){ | 121 static void init_multbl2(uint8_t tbl[1024], const int c[4], const uint8_t *log8, const uint8_t *alog8, const uint8_t *sbox){ |
112 int i, j; | 122 int i, j; |
156 a->rounds= rounds; | 166 a->rounds= rounds; |
157 | 167 |
158 memcpy(tk, key, KC*4); | 168 memcpy(tk, key, KC*4); |
159 | 169 |
160 for(t= 0; t < (rounds+1)*16;) { | 170 for(t= 0; t < (rounds+1)*16;) { |
161 memcpy(a->round_key[0][0]+t, tk, KC*4); | 171 memcpy(a->round_key[0].u8+t, tk, KC*4); |
162 t+= KC*4; | 172 t+= KC*4; |
163 | 173 |
164 for(i = 0; i < 4; i++) | 174 for(i = 0; i < 4; i++) |
165 tk[0][i] ^= sbox[tk[KC-1][(i+1)&3]]; | 175 tk[0][i] ^= sbox[tk[KC-1][(i+1)&3]]; |
166 tk[0][0] ^= rcon[rconpointer++]; | 176 tk[0][0] ^= rcon[rconpointer++]; |
173 } | 183 } |
174 } | 184 } |
175 | 185 |
176 if(decrypt){ | 186 if(decrypt){ |
177 for(i=1; i<rounds; i++){ | 187 for(i=1; i<rounds; i++){ |
178 uint8_t tmp[3][16]; | 188 av_aes_block tmp[3]; |
179 memcpy(tmp[2], a->round_key[i][0], 16); | 189 memcpy(&tmp[2], &a->round_key[i], 16); |
180 subshift(&tmp[1], 0, sbox); | 190 subshift(&tmp[1], 0, sbox); |
181 mix(tmp, dec_multbl, 1, 3); | 191 mix(tmp, dec_multbl, 1, 3); |
182 memcpy(a->round_key[i][0], tmp[0], 16); | 192 memcpy(&a->round_key[i], &tmp[0], 16); |
183 } | 193 } |
184 }else{ | 194 }else{ |
185 for(i=0; i<(rounds+1)>>1; i++){ | 195 for(i=0; i<(rounds+1)>>1; i++){ |
186 for(j=0; j<16; j++) | 196 for(j=0; j<16; j++) |
187 FFSWAP(int, a->round_key[i][0][j], a->round_key[rounds-i][0][j]); | 197 FFSWAP(int, a->round_key[i].u8[j], a->round_key[rounds-i].u8[j]); |
188 } | 198 } |
189 } | 199 } |
190 | 200 |
191 return 0; | 201 return 0; |
192 } | 202 } |