void
_salsa20_core(uint32_t *dst, const uint32_t *src, unsigned rounds)
{
uint32_t x[_SALSA20_INPUT_LENGTH];
unsigned i;
assert ( (rounds & 1) == 0);
memcpy (x, src, sizeof(x));
for (i = 0; i < rounds;i += 2)
{
DEBUG (i);
QROUND(x[0], x[4], x[8], x[12]);
QROUND(x[5], x[9], x[13], x[1]);
QROUND(x[10], x[14], x[2], x[6]);
QROUND(x[15], x[3], x[7], x[11]);
DEBUG (i+1);
QROUND(x[0], x[1], x[2], x[3]);
QROUND(x[5], x[6], x[7], x[4]);
QROUND(x[10], x[11], x[8], x[9]);
QROUND(x[15], x[12], x[13], x[14]);
}
DEBUG (i);
for (i = 0; i < _SALSA20_INPUT_LENGTH; i++)
{
uint32_t t = x[i] + src[i];
dst[i] = LE_SWAP32 (t);
}
}
static void
salsa20_core (u32 *dst, const u32 *src, unsigned int rounds)
{
u32 x[SALSA20_INPUT_LENGTH];
unsigned i;
assert ( (rounds & 1) == 0);
for (i = 0; i < SALSA20_INPUT_LENGTH; i++)
x[i] = LE_SWAP32(src[i]);
for (i = 0; i < rounds;i += 2)
{
QROUND(x[0], x[4], x[8], x[12]);
QROUND(x[5], x[9], x[13], x[1]);
QROUND(x[10], x[14], x[2], x[6]);
QROUND(x[15], x[3], x[7], x[11]);
QROUND(x[0], x[1], x[2], x[3]);
QROUND(x[5], x[6], x[7], x[4]);
QROUND(x[10], x[11], x[8], x[9]);
QROUND(x[15], x[12], x[13], x[14]);
}
for (i = 0; i < SALSA20_INPUT_LENGTH; i++)
{
u32 t = x[i] + LE_SWAP32(src[i]);
dst[i] = LE_SWAP32(t);
}
}
static void
salsa20_core (u32 *dst, const u32 *src)
{
u32 pad[SALSA20_INPUT_LENGTH];
unsigned int i;
memcpy (pad, src, sizeof(pad));
for (i = 0; i < SALSA20_ROUNDS; i += 2)
{
SALSA20_CORE_DEBUG (i);
QROUND (pad[0], pad[4], pad[8], pad[12]);
QROUND (pad[5], pad[9], pad[13], pad[1] );
QROUND (pad[10], pad[14], pad[2], pad[6] );
QROUND (pad[15], pad[3], pad[7], pad[11]);
SALSA20_CORE_DEBUG (i+1);
QROUND (pad[0], pad[1], pad[2], pad[3] );
QROUND (pad[5], pad[6], pad[7], pad[4] );
QROUND (pad[10], pad[11], pad[8], pad[9] );
QROUND (pad[15], pad[12], pad[13], pad[14]);
}
SALSA20_CORE_DEBUG (i);
for (i = 0; i < SALSA20_INPUT_LENGTH; i++)
{
u32 t = pad[i] + src[i];
dst[i] = LE_SWAP32 (t);
}
}
static unsigned int
salsa20_core (u32 *dst, SALSA20_context_t *ctx, unsigned rounds)
{
u32 pad[SALSA20_INPUT_LENGTH], *src = ctx->input;
unsigned int i;
memcpy (pad, src, sizeof(pad));
for (i = 0; i < rounds; i += 2)
{
SALSA20_CORE_DEBUG (i);
QROUND (pad[0], pad[4], pad[8], pad[12]);
QROUND (pad[5], pad[9], pad[13], pad[1] );
QROUND (pad[10], pad[14], pad[2], pad[6] );
QROUND (pad[15], pad[3], pad[7], pad[11]);
SALSA20_CORE_DEBUG (i+1);
QROUND (pad[0], pad[1], pad[2], pad[3] );
QROUND (pad[5], pad[6], pad[7], pad[4] );
QROUND (pad[10], pad[11], pad[8], pad[9] );
QROUND (pad[15], pad[12], pad[13], pad[14]);
}
SALSA20_CORE_DEBUG (i);
for (i = 0; i < SALSA20_INPUT_LENGTH; i++)
{
u32 t = pad[i] + src[i];
dst[i] = LE_SWAP32 (t);
}
/* Update counter. */
if (!++src[8])
src[9]++;
/* burn_stack */
return ( 3*sizeof (void*) \
+ 2*sizeof (void*) \
+ 64 \
+ sizeof (unsigned int) \
+ sizeof (u32) );
}
void akmos_salsa_stream(akmos_salsa_t *ctx, uint8_t *out_blk)
{
size_t i;
uint32_t s[16];
for(i = 0; i < 16; i++)
s[i] = ctx->s[i];
for(i = 0; i < AKMOS_SALSA_ROUNDS / 2; i++) {
QROUND( 4, 0, 12, 7); QROUND( 9, 5, 1, 7);
QROUND(14, 10, 6, 7); QROUND( 3, 15, 11, 7);
QROUND( 8, 4, 0, 9); QROUND(13, 9, 5, 9);
QROUND( 2, 14, 10, 9); QROUND( 7, 3, 15, 9);
QROUND(12, 8, 4, 13); QROUND( 1, 13, 9, 13);
QROUND( 6, 2, 14, 13); QROUND(11, 7, 3, 13);
QROUND( 0, 12, 8, 18); QROUND( 5, 1, 13, 18);
QROUND(10, 6, 2, 18); QROUND(15, 11, 7, 18);
QROUND( 1, 0, 3, 7); QROUND( 6, 5, 4, 7);
QROUND(11, 10, 9, 7); QROUND(12, 15, 14, 7);
QROUND( 2, 1, 0, 9); QROUND( 7, 6, 5, 9);
QROUND( 8, 11, 10, 9); QROUND(13, 12, 15, 9);
QROUND( 3, 2, 1, 13); QROUND( 4, 7, 6, 13);
QROUND( 9, 8, 11, 13); QROUND(14, 13, 12, 13);
QROUND( 0, 3, 2, 18); QROUND( 5, 4, 7, 18);
QROUND(10, 9, 8, 18); QROUND(15, 14, 13, 18);
}
for(i = 0; i < 16; i++)
s[i] += ctx->s[i];
ctx->s[8]++;
if(!ctx->s[8])
ctx->s[9]++;
for(i = 0; i < AKMOS_SALSA_BLKLEN / 4; i++, out_blk += 4)
UNPACK32BE(out_blk, s[i]);
}
