ecp: clarify contract of internal function

The "shortcut" implementation was using the fact that this function is
called in a single place where we know that the top limb of A is zero.

But the function's documentation was not stating that as a requirement.

Align the documentation and "non-shortcut" implementation on the
"shortcut" implementation. This should be a tiny bit better for
performance: using the fact that A's top limb is zero means we have one
limb less to load from memory. (Copying is load + store, while setting
to 0 is only a store.)

Signed-off-by: Manuel Pégourié-Gonnard <[email protected]>
This commit is contained in:
Manuel Pégourié-Gonnard
2026-05-18 10:45:11 +02:00
parent 422dcad498
commit 8f682e8f8e
+7 -5
View File
@@ -5492,23 +5492,25 @@ cleanup:
}
/* Copy and shift right: X = A >> 224
* Both X and A must be P448_WIDTH + 1 limbs */
* X must be P448_WIDTH + 1, and A must be P448_WIDTH limbs */
static void ecp_copy_shift_r_224(mbedtls_mpi_uint *X,
const mbedtls_mpi_uint *A)
{
#if defined(MBEDTLS_HAVE_INT64) && defined(MBEDTLS_IS_BIG_ENDIAN)
/* No need to copy lower limbs, they'll be shifted away.
* P448_WIDTH + 1 = P224_WIDTH_MIN + (P224_WIDTH_MAX + 1) */
memcpy(X + P224_WIDTH_MIN, A + P224_WIDTH_MIN, (P224_WIDTH_MAX + 1) * ciL);
* P448_WIDTH = P224_WIDTH_MIN + P224_WIDTH_MAX */
memcpy(X + P224_WIDTH_MIN, A + P224_WIDTH_MIN, P224_WIDTH_MAX * ciL);
X[P448_WIDTH] = 0;
mbedtls_mpi_core_shift_r(X, P448_WIDTH + 1, 224);
#else
/* Shortcut for 32-bit and little-endian 64-bit. */
/* Shortcut for 32-bit and little-endian 64-bit.
* P448_WITDH * ciL = 2 * P224_SIZE */
memcpy(X, (char *) A + P224_SIZE, P224_SIZE);
memset((char *) X + P224_SIZE, 0, P224_SIZE + ciL);
#endif
}
/* Shift left: X <<= 224
/* In-place shift left: X <<= 224
* X must be P448_WIDTH + 1 limbs */
static void ecp_shift_l_224(mbedtls_mpi_uint *X)
{