/* AES implementation by Tom St Denis 
 *
 * Derived from the Public Domain source code by
 
---  
  * rijndael-alg-fst.c
  *
  * @version 3.0 (December 2000)
  *
  * Optimised ANSI C code for the Rijndael cipher (now AES)
  *
  * @author Vincent Rijmen <vincent.rijmen@esat.kuleuven.ac.be>
  * @author Antoon Bosselaers <antoon.bosselaers@esat.kuleuven.ac.be>
  * @author Paulo Barreto <paulo.barreto@terra.com.br>
---
 */

#include "mycrypt.h"

#ifdef RIJNDAEL

const struct _cipher_descriptor rijndael_desc =
{
    "rijndael",
    6,
    16, 32, 16, 10,
    &rijndael_setup,
    &rijndael_ecb_encrypt,
    &rijndael_ecb_decrypt,
    &rijndael_test,
    &rijndael_keysize
};

const struct _cipher_descriptor aes_desc =
{
    "aes",
    6,
    16, 32, 16, 10,
    &rijndael_setup,
    &rijndael_ecb_encrypt,
    &rijndael_ecb_decrypt,
    &rijndael_test,
    &rijndael_keysize
};

#include "aes_tab.c"

int rijndael_setup(const unsigned char *key, int keylen, int rounds, symmetric_key *skey)
{
    int i = 0, j;
    unsigned long temp, *rk, *rrk;
    
    _ARGCHK(key != NULL);
    _ARGCHK(skey != NULL);
    
    if (keylen != 16 && keylen != 24 && keylen != 32) {
       return CRYPT_INVALID_KEYSIZE;
    }
    
    if (rounds != 0 && rounds != (10 + ((keylen/8)-2)*2)) {
       return CRYPT_INVALID_ROUNDS;
    }
    
    skey->rijndael.Nr = 10 + ((keylen/8)-2)*2;
        
    /* setup the forward key */
    rk                = skey->rijndael.eK;
    LOAD32H(rk[0], key     );
    LOAD32H(rk[1], key +  4);
    LOAD32H(rk[2], key +  8);
    LOAD32H(rk[3], key + 12);
    if (keylen == 16) {
        j = 44;
        for (;;) {
            temp  = rk[3];
            rk[4] = rk[0] ^
                (Te4_3[byte(temp, 2)]) ^
                (Te4_2[byte(temp, 1)]) ^
                (Te4_1[byte(temp, 0)]) ^
                (Te4_0[byte(temp, 3)]) ^
                rcon[i];
            rk[5] = rk[1] ^ rk[4];
            rk[6] = rk[2] ^ rk[5];
            rk[7] = rk[3] ^ rk[6];
            if (++i == 10) {
                break;
            }
            rk += 4;
        }
    } else if (keylen == 24) {
        j = 52;   
        LOAD32H(rk[4], key + 16);
        LOAD32H(rk[5], key + 20);
        for (;;) {
        #ifdef _MSC_VER
            temp = skey->rijndael.eK[rk - skey->rijndael.eK + 5]; 
        #else
            temp = rk[5];
        #endif
            rk[ 6] = rk[ 0] ^
                (Te4_3[byte(temp, 2)]) ^
                (Te4_2[byte(temp, 1)]) ^
                (Te4_1[byte(temp, 0)]) ^
                (Te4_0[byte(temp, 3)]) ^
                rcon[i];
            rk[ 7] = rk[ 1] ^ rk[ 6];
            rk[ 8] = rk[ 2] ^ rk[ 7];
            rk[ 9] = rk[ 3] ^ rk[ 8];
            if (++i == 8) {
                break;
            }
            rk[10] = rk[ 4] ^ rk[ 9];
            rk[11] = rk[ 5] ^ rk[10];
            rk += 6;
        }
    } else if (keylen == 32) {
        j = 60;
        LOAD32H(rk[4], key + 16);
        LOAD32H(rk[5], key + 20);
        LOAD32H(rk[6], key + 24);
        LOAD32H(rk[7], key + 28);
        for (;;) {
        #ifdef _MSC_VER
            temp = skey->rijndael.eK[rk - skey->rijndael.eK + 7]; 
        #else
            temp = rk[7];
        #endif
            rk[ 8] = rk[ 0] ^
                (Te4_3[byte(temp, 2)]) ^
                (Te4_2[byte(temp, 1)]) ^
                (Te4_1[byte(temp, 0)]) ^
                (Te4_0[byte(temp, 3)]) ^
                rcon[i];
            rk[ 9] = rk[ 1] ^ rk[ 8];
            rk[10] = rk[ 2] ^ rk[ 9];
            rk[11] = rk[ 3] ^ rk[10];
            if (++i == 7) {
                break;
            }
            temp = rk[11];
            rk[12] = rk[ 4] ^
                (Te4_3[byte(temp, 3)]) ^
                (Te4_2[byte(temp, 2)]) ^
                (Te4_1[byte(temp, 1)]) ^
                (Te4_0[byte(temp, 0)]);
            rk[13] = rk[ 5] ^ rk[12];
            rk[14] = rk[ 6] ^ rk[13];
            rk[15] = rk[ 7] ^ rk[14];
            rk += 8;
        }
    } else {
       /* this can't happen */
       j = 4;
    }
    
    /* setup the inverse key now */
    rk   = skey->rijndael.dK;
    rrk  = skey->rijndael.eK + j - 4; 
    
    /* apply the inverse MixColumn transform to all round keys but the first and the last: */
    /* copy first */
    *rk++ = *rrk++;
    *rk++ = *rrk++;
    *rk++ = *rrk++;
    *rk   = *rrk;
    rk -= 3; rrk -= 3;
    
    for (i = 1; i < skey->rijndael.Nr; i++) {
        rrk -= 4;
        rk  += 4;
    #ifdef SMALL_CODE        
        temp = rrk[0];
        rk[0] =
            Td0[255 & Te4[byte(temp, 3)]] ^
            Td1[255 & Te4[byte(temp, 2)]] ^
            Td2[255 & Te4[byte(temp, 1)]] ^
            Td3[255 & Te4[byte(temp, 0)]];
        temp = rrk[1];
        rk[1] =
            Td0[255 & Te4[byte(temp, 3)]] ^
            Td1[255 & Te4[byte(temp, 2)]] ^
            Td2[255 & Te4[byte(temp, 1)]] ^
            Td3[255 & Te4[byte(temp, 0)]];
        temp = rrk[2];
        rk[2] =
            Td0[255 & Te4[byte(temp, 3)]] ^
            Td1[255 & Te4[byte(temp, 2)]] ^
            Td2[255 & Te4[byte(temp, 1)]] ^
            Td3[255 & Te4[byte(temp, 0)]];
        temp = rrk[3];
        rk[3] =
            Td0[255 & Te4[byte(temp, 3)]] ^
            Td1[255 & Te4[byte(temp, 2)]] ^
            Td2[255 & Te4[byte(temp, 1)]] ^
            Td3[255 & Te4[byte(temp, 0)]];
     #else
        temp = rrk[0];
        rk[0] =
            Tks0[byte(temp, 3)] ^
            Tks1[byte(temp, 2)] ^
            Tks2[byte(temp, 1)] ^
            Tks3[byte(temp, 0)];
        temp = rrk[1];
        rk[1] =
            Tks0[byte(temp, 3)] ^
            Tks1[byte(temp, 2)] ^
            Tks2[byte(temp, 1)] ^
            Tks3[byte(temp, 0)];
        temp = rrk[2];
        rk[2] =
            Tks0[byte(temp, 3)] ^
            Tks1[byte(temp, 2)] ^
            Tks2[byte(temp, 1)] ^
            Tks3[byte(temp, 0)];
        temp = rrk[3];
        rk[3] =
            Tks0[byte(temp, 3)] ^
            Tks1[byte(temp, 2)] ^
            Tks2[byte(temp, 1)] ^
            Tks3[byte(temp, 0)];
      #endif            
     
    }

    /* copy last */
    rrk -= 4;
    rk  += 4;
    *rk++ = *rrk++;
    *rk++ = *rrk++;
    *rk++ = *rrk++;
    *rk   = *rrk;

    return CRYPT_OK;   
}

void rijndael_ecb_encrypt(const unsigned char *pt, unsigned char *ct, symmetric_key *skey) 
{
    unsigned long s0, s1, s2, s3, t0, t1, t2, t3, *rk;
    int Nr, r;
   
    _ARGCHK(pt != NULL);
    _ARGCHK(ct != NULL);
    _ARGCHK(skey != NULL);
    
    Nr = skey->rijndael.Nr;
    rk = skey->rijndael.eK;
    
    /*
     * map byte array block to cipher state
     * and add initial round key:
     */
    LOAD32H(s0, pt      ); s0 ^= rk[0];
    LOAD32H(s1, pt  +  4); s1 ^= rk[1];
    LOAD32H(s2, pt  +  8); s2 ^= rk[2];
    LOAD32H(s3, pt  + 12); s3 ^= rk[3];

    /*
     * Nr - 1 full rounds:
     */
    r = Nr >> 1;
    for (;;) {

/* Both of these blocks are equivalent except the top is more friendlier for x86 processors */
#if defined(__GNUC__)
        t0 = rk[4]; t1 = rk[5]; t2 = rk[6]; t3 = rk[7];
        t1 ^= Te3[byte(s0, 0)]; t2 ^= Te2[byte(s0, 1)]; t3 ^= Te1[byte(s0, 2)]; t0 ^= Te0[byte(s0, 3)];
        t2 ^= Te3[byte(s1, 0)]; t3 ^= Te2[byte(s1, 1)]; t0 ^= Te1[byte(s1, 2)]; t1 ^= Te0[byte(s1, 3)];
        t3 ^= Te3[byte(s2, 0)]; t0 ^= Te2[byte(s2, 1)]; t1 ^= Te1[byte(s2, 2)]; t2 ^= Te0[byte(s2, 3)];
        t0 ^= Te3[byte(s3, 0)]; t1 ^= Te2[byte(s3, 1)]; t2 ^= Te1[byte(s3, 2)]; t3 ^= Te0[byte(s3, 3)];
#else
        t0 =
            Te0[byte(s0, 3)] ^
            Te1[byte(s1, 2)] ^
            Te2[byte(s2, 1)] ^
            Te3[byte(s3, 0)] ^
            rk[4];
        t1 =
            Te0[byte(s1, 3)] ^
            Te1[byte(s2, 2)] ^
            Te2[byte(s3, 1)] ^
            Te3[byte(s0, 0)] ^
            rk[5];
        t2 =
            Te0[byte(s2, 3)] ^
            Te1[byte(s3, 2)] ^
            Te2[byte(s0, 1)] ^
            Te3[byte(s1, 0)] ^
            rk[6];
        t3 =
            Te0[byte(s3, 3)] ^
            Te1[byte(s0, 2)] ^
            Te2[byte(s1, 1)] ^
            Te3[byte(s2, 0)] ^
            rk[7];
#endif
       
        rk += 8;
        if (--r == 0) {
            break;
        }
        
/* this second half optimization actually makes it slower on the Athlon, use with caution. */
#if 0
        s1 = rk[1]; s2 = rk[2]; s3 = rk[3]; s0 = rk[0]; 
        s1 ^= Te3[byte(t0, 0)]; s2 ^= Te2[byte(t0, 1)]; s3 ^= Te1[byte(t0, 2)]; s0 ^= Te0[byte(t0, 3)];
        s2 ^= Te3[byte(t1, 0)]; s3 ^= Te2[byte(t1, 1)]; s0 ^= Te1[byte(t1, 2)]; s1 ^= Te0[byte(t1, 3)];
        s3 ^= Te3[byte(t2, 0)]; s0 ^= Te2[byte(t2, 1)]; s1 ^= Te1[byte(t2, 2)]; s2 ^= Te0[byte(t2, 3)];
        s0 ^= Te3[byte(t3, 0)]; s1 ^= Te2[byte(t3, 1)]; s2 ^= Te1[byte(t3, 2)]; s3 ^= Te0[byte(t3, 3)];
#else
        s0 =
            Te0[byte(t0, 3)] ^
            Te1[byte(t1, 2)] ^
            Te2[byte(t2, 1)] ^
            Te3[byte(t3, 0)] ^
            rk[0];
        s1 =
            Te0[byte(t1, 3)] ^
            Te1[byte(t2, 2)] ^
            Te2[byte(t3, 1)] ^
            Te3[byte(t0, 0)] ^
            rk[1];
        s2 =
            Te0[byte(t2, 3)] ^
            Te1[byte(t3, 2)] ^
            Te2[byte(t0, 1)] ^
            Te3[byte(t1, 0)] ^
            rk[2];
        s3 =
            Te0[byte(t3, 3)] ^
            Te1[byte(t0, 2)] ^
            Te2[byte(t1, 1)] ^
            Te3[byte(t2, 0)] ^
            rk[3];
#endif            
    }
    /*
     * apply last round and
     * map cipher state to byte array block:
     */
    s0 =
        (Te4_3[(t0 >> 24)       ]) ^
        (Te4_2[(t1 >> 16) & 0xff]) ^
        (Te4_1[(t2 >>  8) & 0xff]) ^
        (Te4_0[(t3      ) & 0xff]) ^
        rk[0];
    STORE32H(s0, ct);
    s1 =
        (Te4_3[(t1 >> 24)       ]) ^
        (Te4_2[(t2 >> 16) & 0xff]) ^
        (Te4_1[(t3 >>  8) & 0xff]) ^
        (Te4_0[(t0      ) & 0xff]) ^
        rk[1];
    STORE32H(s1, ct+4);
    s2 =
        (Te4_3[(t2 >> 24)       ]) ^
        (Te4_2[(t3 >> 16) & 0xff]) ^
        (Te4_1[(t0 >>  8) & 0xff]) ^
        (Te4_0[(t1      ) & 0xff]) ^
        rk[2];
    STORE32H(s2, ct+8);
    s3 =
        (Te4_3[(t3 >> 24)       ]) ^
        (Te4_2[(t0 >> 16) & 0xff]) ^
        (Te4_1[(t1 >>  8) & 0xff]) ^
        (Te4_0[(t2      ) & 0xff]) ^ 
        rk[3];
    STORE32H(s3, ct+12);
}

void rijndael_ecb_decrypt(const unsigned char *ct, unsigned char *pt, symmetric_key *skey) {
    unsigned long s0, s1, s2, s3, t0, t1, t2, t3, *rk;
    int Nr, r;

    _ARGCHK(pt != NULL);
    _ARGCHK(ct != NULL);
    _ARGCHK(skey != NULL);
    
    Nr = skey->rijndael.Nr;
    rk = skey->rijndael.dK;

    /*
     * map byte array block to cipher state
     * and add initial round key:
     */
    LOAD32H(s0, ct      ); s0 ^= rk[0];
    LOAD32H(s1, ct  +  4); s1 ^= rk[1];
    LOAD32H(s2, ct  +  8); s2 ^= rk[2];
    LOAD32H(s3, ct  + 12); s3 ^= rk[3];

    /*
     * Nr - 1 full rounds:
     */
    r = Nr >> 1;
    for (;;) {

        t0 =
            Td0[byte(s0, 3)] ^
            Td1[byte(s3, 2)] ^
            Td2[byte(s2, 1)] ^
            Td3[byte(s1, 0)] ^
            rk[4];
        t1 =
            Td0[byte(s1, 3)] ^
            Td1[byte(s0, 2)] ^
            Td2[byte(s3, 1)] ^
            Td3[byte(s2, 0)] ^
            rk[5];
        t2 =
            Td0[byte(s2, 3)] ^
            Td1[byte(s1, 2)] ^
            Td2[byte(s0, 1)] ^
            Td3[byte(s3, 0)] ^
            rk[6];
        t3 =
            Td0[byte(s3, 3)] ^
            Td1[byte(s2, 2)] ^
            Td2[byte(s1, 1)] ^
            Td3[byte(s0, 0)] ^
            rk[7];

        rk += 8;
        if (--r == 0) {
            break;
        }


        s0 =
            Td0[byte(t0, 3)] ^
            Td1[byte(t3, 2)] ^
            Td2[byte(t2, 1)] ^
            Td3[byte(t1, 0)] ^
            rk[0];
        s1 =
            Td0[byte(t1, 3)] ^
            Td1[byte(t0, 2)] ^
            Td2[byte(t3, 1)] ^
            Td3[byte(t2, 0)] ^
            rk[1];
        s2 =
            Td0[byte(t2, 3)] ^
            Td1[byte(t1, 2)] ^
            Td2[byte(t0, 1)] ^
            Td3[byte(t3, 0)] ^
            rk[2];
        s3 =
            Td0[byte(t3, 3)] ^
            Td1[byte(t2, 2)] ^
            Td2[byte(t1, 1)] ^
            Td3[byte(t0, 0)] ^
            rk[3];
    }

    /*
     * apply last round and
     * map cipher state to byte array block:
     */
    s0 =
        (Td4[(t0 >> 24)       ] & 0xff000000) ^
        (Td4[(t3 >> 16) & 0xff] & 0x00ff0000) ^
        (Td4[(t2 >>  8) & 0xff] & 0x0000ff00) ^
        (Td4[(t1      ) & 0xff] & 0x000000ff) ^
        rk[0];
    STORE32H(s0, pt);
    s1 =
        (Td4[(t1 >> 24)       ] & 0xff000000) ^
        (Td4[(t0 >> 16) & 0xff] & 0x00ff0000) ^
        (Td4[(t3 >>  8) & 0xff] & 0x0000ff00) ^
        (Td4[(t2      ) & 0xff] & 0x000000ff) ^
        rk[1];
    STORE32H(s1, pt+4);
    s2 =
        (Td4[(t2 >> 24)       ] & 0xff000000) ^
        (Td4[(t1 >> 16) & 0xff] & 0x00ff0000) ^
        (Td4[(t0 >>  8) & 0xff] & 0x0000ff00) ^
        (Td4[(t3      ) & 0xff] & 0x000000ff) ^
        rk[2];
    STORE32H(s2, pt+8);
    s3 =
        (Td4[(t3 >> 24)       ] & 0xff000000) ^
        (Td4[(t2 >> 16) & 0xff] & 0x00ff0000) ^
        (Td4[(t1 >>  8) & 0xff] & 0x0000ff00) ^
        (Td4[(t0      ) & 0xff] & 0x000000ff) ^
        rk[3];
    STORE32H(s3, pt+12);
}

int rijndael_test(void)
{
 #ifndef LTC_TEST
    return CRYPT_NOP;
 #else    
 int err;
 static const struct {
     int keylen;
     unsigned char key[32], pt[16], ct[16];
 } tests[] = {
    { 16,
      { 0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, 
        0x08, 0x09, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f }, 
      { 0x00, 0x11, 0x22, 0x33, 0x44, 0x55, 0x66, 0x77,
        0x88, 0x99, 0xaa, 0xbb, 0xcc, 0xdd, 0xee, 0xff },
      { 0x69, 0xc4, 0xe0, 0xd8, 0x6a, 0x7b, 0x04, 0x30, 
        0xd8, 0xcd, 0xb7, 0x80, 0x70, 0xb4, 0xc5, 0x5a }
    }, { 
      24,
      { 0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, 
        0x08, 0x09, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f,
        0x10, 0x11, 0x12, 0x13, 0x14, 0x15, 0x16, 0x17 },
      { 0x00, 0x11, 0x22, 0x33, 0x44, 0x55, 0x66, 0x77,
        0x88, 0x99, 0xaa, 0xbb, 0xcc, 0xdd, 0xee, 0xff },
      { 0xdd, 0xa9, 0x7c, 0xa4, 0x86, 0x4c, 0xdf, 0xe0, 
        0x6e, 0xaf, 0x70, 0xa0, 0xec, 0x0d, 0x71, 0x91 }
    }, {
      32,
      { 0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, 
        0x08, 0x09, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f,
        0x10, 0x11, 0x12, 0x13, 0x14, 0x15, 0x16, 0x17, 
        0x18, 0x19, 0x1a, 0x1b, 0x1c, 0x1d, 0x1e, 0x1f },
      { 0x00, 0x11, 0x22, 0x33, 0x44, 0x55, 0x66, 0x77,
        0x88, 0x99, 0xaa, 0xbb, 0xcc, 0xdd, 0xee, 0xff },
      { 0x8e, 0xa2, 0xb7, 0xca, 0x51, 0x67, 0x45, 0xbf, 
        0xea, 0xfc, 0x49, 0x90, 0x4b, 0x49, 0x60, 0x89 }
    }
 };
 
 symmetric_key key;
 unsigned char tmp[2][16];
 int i;
 
 for (i = 0; i < (int)(sizeof(tests)/sizeof(tests[0])); i++) {
    zeromem(&key, sizeof(key));
    if ((err = rijndael_setup(tests[i].key, tests[i].keylen, 0, &key)) != CRYPT_OK) { 
       return err;
    }
  
    rijndael_ecb_encrypt(tests[i].pt, tmp[0], &key);
    rijndael_ecb_decrypt(tmp[0], tmp[1], &key);
    if (memcmp(tmp[0], tests[i].ct, 16) || memcmp(tmp[1], tests[i].pt, 16)) { 
#if 0
       printf("\n\nTest %d failed\n", i);
       if (memcmp(tmp[0], tests[i].ct, 16)) {
          printf("CT: ");
          for (i = 0; i < 16; i++) {
             printf("%02x ", tmp[0][i]);
          }
          printf("\n");
       } else {
          printf("PT: ");
          for (i = 0; i < 16; i++) {
             printf("%02x ", tmp[1][i]);
          }
          printf("\n");
       }
#endif       
        return CRYPT_FAIL_TESTVECTOR;
    }
 }       
 return CRYPT_OK;
 #endif
}

int rijndael_keysize(int *desired_keysize)
{
   _ARGCHK(desired_keysize != NULL);

   if (*desired_keysize < 16)
      return CRYPT_INVALID_KEYSIZE;
   if (*desired_keysize < 24) {
      *desired_keysize = 16;
      return CRYPT_OK;
   } else if (*desired_keysize < 32) {
      *desired_keysize = 24;
      return CRYPT_OK;
   } else {
      *desired_keysize = 32;
      return CRYPT_OK;
   }
}

#endif