/* Note: This function requires LENGTH > 0. */
static void
salsa20_do_encrypt_stream (SALSA20_context_t *ctx,
byte *outbuf, const byte *inbuf,
unsigned int length)
{
if (ctx->unused)
{
unsigned char *p = (void*)ctx->pad;
unsigned int n;
gcry_assert (ctx->unused < SALSA20_BLOCK_SIZE);
n = ctx->unused;
if (n > length)
n = length;
buf_xor (outbuf, inbuf, p + SALSA20_BLOCK_SIZE - ctx->unused, n);
length -= n;
outbuf += n;
inbuf += n;
ctx->unused -= n;
if (!length)
return;
gcry_assert (!ctx->unused);
}
for (;;)
{
/* Create the next pad and bump the block counter. Note that it
is the user's duty to change to another nonce not later than
after 2^70 processed bytes. */
salsa20_core (ctx->pad, ctx->input);
if (!++ctx->input[8])
ctx->input[9]++;
if (length <= SALSA20_BLOCK_SIZE)
{
buf_xor (outbuf, inbuf, ctx->pad, length);
ctx->unused = SALSA20_BLOCK_SIZE - length;
return;
}
buf_xor (outbuf, inbuf, ctx->pad, SALSA20_BLOCK_SIZE);
length -= SALSA20_BLOCK_SIZE;
outbuf += SALSA20_BLOCK_SIZE;
inbuf += SALSA20_BLOCK_SIZE;
}
}
static void blockmix(uint32_t *b, uint32_t *t, const size_t r)
{
uint32_t x[16];
size_t i;
memcpy(x, b + (r << 5) - 16, 64);
for(i = 0; i < (r << 5);)
{
x[0] ^= b[i++];
x[1] ^= b[i++];
x[2] ^= b[i++];
x[3] ^= b[i++];
x[4] ^= b[i++];
x[5] ^= b[i++];
x[6] ^= b[i++];
x[7] ^= b[i++];
x[8] ^= b[i++];
x[9] ^= b[i++];
x[10] ^= b[i++];
x[11] ^= b[i++];
x[12] ^= b[i++];
x[13] ^= b[i++];
x[14] ^= b[i++];
x[15] ^= b[i++];
salsa20_core(x);
memcpy(t + i - 16, x, 64);
}
for(i = 0; i < r; i++)
{
memcpy(b + (i << 4), t + (i << 5), 64);
memcpy(b + ((i + r) << 4), t + (i << 5) + 16, 64);
}
}
static void
scrypt_block_mix (u32 r, unsigned char *B, unsigned char *tmp2)
{
u64 i;
unsigned char *X = tmp2;
unsigned char *Y = tmp2 + 64;
#if 0
if (r == 1)
{
for (i = 0; i < 2 * r; i++)
{
size_t j;
printf ("B[%d] = ", (int)i);
for (j = 0; j < 64; j++)
{
if (j && !(j % 16))
printf ("\n ");
printf (" %02x", B[i * 64 + j]);
}
putchar ('\n');
}
}
#endif
/* X = B[2 * r - 1] */
memcpy (X, &B[(2 * r - 1) * 64], 64);
/* for i = 0 to 2 * r - 1 do */
for (i = 0; i <= 2 * r - 1; i++)
{
/* T = X xor B[i] */
buf_xor(X, X, &B[i * 64], 64);
/* X = Salsa (T) */
salsa20_core ((u32*)(void*)X, (u32*)(void*)X, 8);
/* Y[i] = X */
memcpy (&Y[i * 64], X, 64);
}
for (i = 0; i < r; i++)
{
memcpy (&B[i * 64], &Y[2 * i * 64], 64);
memcpy (&B[(r + i) * 64], &Y[(2 * i + 1) * 64], 64);
}
#if 0
if (r==1)
{
for (i = 0; i < 2 * r; i++)
{
size_t j;
printf ("B'[%d] =", (int)i);
for (j = 0; j < 64; j++)
{
if (j && !(j % 16))
printf ("\n ");
printf (" %02x", B[i * 64 + j]);
}
putchar ('\n');
}
}
#endif
}