aes-encrypt-internal.c 3.12 KB
Newer Older
1
/* aes-encrypt-internal.c
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

   Encryption function for the aes/rijndael block cipher.

   Copyright (C) 2002, 2013 Niels Möller

   This file is part of GNU Nettle.

   GNU Nettle is free software: you can redistribute it and/or
   modify it under the terms of either:

     * the GNU Lesser General Public License as published by the Free
       Software Foundation; either version 3 of the License, or (at your
       option) any later version.

   or

     * the GNU General Public License as published by the Free
       Software Foundation; either version 2 of the License, or (at your
       option) any later version.

   or both in parallel, as here.

   GNU Nettle 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
   General Public License for more details.

   You should have received copies of the GNU General Public License and
   the GNU Lesser General Public License along with this program.  If
   not, see http://www.gnu.org/licenses/.
*/
33 34 35 36 37 38 39 40 41 42 43

#if HAVE_CONFIG_H
# include "config.h"
#endif

#include <assert.h>

#include "aes-internal.h"
#include "macros.h"

void
44
_nettle_aes_encrypt(unsigned rounds, const uint32_t *keys,
45
		    const struct aes_table *T,
46
		    size_t length, uint8_t *dst,
47 48 49 50 51 52
		    const uint8_t *src)
{
  FOR_BLOCKS(length, dst, src, AES_BLOCK_SIZE)
    {
      uint32_t w0, w1, w2, w3;		/* working ciphertext */
      uint32_t t0, t1, t2, t3;
53
      unsigned i;
54 55 56 57
      
      /* Get clear text, using little-endian byte order.
       * Also XOR with the first subkey. */

58 59 60 61
      w0 = LE_READ_UINT32(src)      ^ keys[0];
      w1 = LE_READ_UINT32(src + 4)  ^ keys[1];
      w2 = LE_READ_UINT32(src + 8)  ^ keys[2];
      w3 = LE_READ_UINT32(src + 12) ^ keys[3];
62

63
      for (i = 1; i < rounds; i++)
64
	{
65 66 67 68
	  t0 = AES_ROUND(T, w0, w1, w2, w3, keys[4*i]);
	  t1 = AES_ROUND(T, w1, w2, w3, w0, keys[4*i + 1]);
	  t2 = AES_ROUND(T, w2, w3, w0, w1, keys[4*i + 2]);
	  t3 = AES_ROUND(T, w3, w0, w1, w2, keys[4*i + 3]);
69

Niels Möller's avatar
Niels Möller committed
70 71 72
	  /* We could unroll the loop twice, to avoid these
	     assignments. If all eight variables fit in registers,
	     that should give a slight speedup. */
73 74 75 76 77 78 79 80
	  w0 = t0;
	  w1 = t1;
	  w2 = t2;
	  w3 = t3;
	}

      /* Final round */

81 82 83 84
      t0 = AES_FINAL_ROUND(T, w0, w1, w2, w3, keys[4*i]);
      t1 = AES_FINAL_ROUND(T, w1, w2, w3, w0, keys[4*i + 1]);
      t2 = AES_FINAL_ROUND(T, w2, w3, w0, w1, keys[4*i + 2]);
      t3 = AES_FINAL_ROUND(T, w3, w0, w1, w2, keys[4*i + 3]);
85 86 87

      LE_WRITE_UINT32(dst, t0);
      LE_WRITE_UINT32(dst + 4, t1);
88
      LE_WRITE_UINT32(dst + 8, t2);
89 90 91 92
      LE_WRITE_UINT32(dst + 12, t3);
    }
}

Niels Möller's avatar
Niels Möller committed
93
/* Some stats, all for AES 128:
94 95 96 97 98 99 100 101 102 103

   A. Table-driven indexing (the approach of the old unified
      _aes_crypt function).
   B. Unrolling the j-loop.

   C. Eliminated the use of IDXk(j) in the main loop.

   D. Put wtxt in four scalar variables.

   E. Also put t in four scalar variables.
Niels Möller's avatar
Niels Möller committed
104 105 106

       P4 2.2 GHz         AMD Duron 1.4GHz
       
107
       MB/s  code size
Niels Möller's avatar
Niels Möller committed
108
   A   35.9  0x202        17 MB/s
109 110 111
   B   37.3  0x334
   C   33.0  0x2a7
   D   40.7  0x3f9
Niels Möller's avatar
Niels Möller committed
112
   E   42.9  0x44a        26 MB/s
113
 */