Protecting Against Eavesdroppers with Symmetric Cryptography - Implementing SSL/TLS Using Cryptography and PKI - page 73

Mixing Columns in AES
Armed with this strange multiplication operation, you can implement the matrix
multiplication that performs the column-mixing step in Listing 2-38.
Listing 2-38: “aes.c” mix_columns
static void mix_columns( unsigned char s[ ][ 4 ] )
{
int c;
unsigned char t[ 4 ];
for ( c = 0; c < 4; c++ )
{
t[ 0 ] = dot( 2, s[ 0 ][ c ] ) ^ dot( 3, s[ 1 ][ c ] ) ^
s[ 2 ][ c ] ^ s[ 3 ][ c ];
t[ 1 ] = s[ 0 ][ c ] ^ dot( 2, s[ 1 ][ c ] ) ^
dot( 3, s[ 2 ][ c ] ) ^ s[ 3 ][ c ];
t[ 2 ] = s[ 0 ][ c ] ^ s[ 1 ][ c ] ^ dot( 2, s[ 2 ][ c ] ) ^
dot( 3, s[ 3 ] [ c ] );
t[ 3 ] = dot( 3, s[ 0 ][ c ] ) ^ s[ 1 ][ c ] ^ s[ 2 ][ c ] ^
dot( 2, s[ 3 ][ c ] );
s[ 0 ][ c ] = t[ 0 ];
s[ 1 ][ c ] = t[ 1 ];
s[ 2 ][ c ] = t[ 2 ];
s[ 3 ][ c ] = t[ 3 ];
}
}
Remembering that adding is XORing and mutiplying is dot-ing, this is a
straightforward matrix multiplication. Compare this to Listing 2-35.
And that's it. Each round consists of substituting, shifting, column mixing,
and fi nally adding the round key. Encrypting a block of AES, then, can be done
as shown in Listing 2-39.
Listing 2-39: “aes.c” aes_block_encrypt
static void aes_block_encrypt( const unsigned char *input_block,
unsigned char *output_block,
const unsigned char *key,
int key_size )
{
int r, c;
int round;
int nr;
unsigned char state[ 4 ][ 4 ];
unsigned char w[ 60 ][ 4 ];
for ( r = 0; r < 4; r++ )
{
(Continued)