[Cryptech-Commits] [sw/libhal] 02/04: Fold the coef* functions into a single function, because that really was premature optimization.
git at cryptech.is
git at cryptech.is
Sat Mar 10 15:00:42 UTC 2018
This is an automated email from the git hooks/post-receive script.
paul at psgd.org pushed a commit to branch hashsig
in repository sw/libhal.
commit 6dfdbd46031799504fe062725ae1621b18b541d1
Author: Paul Selkirk <paul at psgd.org>
AuthorDate: Sat Mar 10 08:31:51 2018 -0500
Fold the coef* functions into a single function, because that really was premature optimization.
---
hashsig.c | 63 +++++++++++++++++++++++----------------------------------------
1 file changed, 23 insertions(+), 40 deletions(-)
diff --git a/hashsig.c b/hashsig.c
index e32358d..e2acfc5 100644
--- a/hashsig.c
+++ b/hashsig.c
@@ -137,21 +137,15 @@ static inline hal_error_t hal_xdr_decode_bytestring16(const uint8_t ** const inb
* LM-OTS
*/
-static uint8_t coef1(const uint8_t * const S, const size_t i);
-static uint8_t coef2(const uint8_t * const S, const size_t i);
-static uint8_t coef4(const uint8_t * const S, const size_t i);
-static uint8_t coef8(const uint8_t * const S, const size_t i);
-
typedef const struct lmots_parameter_set {
lmots_algorithm_t type;
size_t n, w, w2, p, ls;
- uint8_t (*coef)(const uint8_t * const S, const size_t i);
} lmots_parameter_t;
static lmots_parameter_t lmots_parameters[] = {
- { lmots_sha256_n32_w1, 32, 1, 2, 265, 7, coef1 },
- { lmots_sha256_n32_w2, 32, 2, 4, 133, 6, coef2 },
- { lmots_sha256_n32_w4, 32, 4, 16, 67, 4, coef4 },
- { lmots_sha256_n32_w8, 32, 8, 256, 34, 0, coef8 },
+ { lmots_sha256_n32_w1, 32, 1, 2, 265, 7 },
+ { lmots_sha256_n32_w2, 32, 2, 4, 133, 6 },
+ { lmots_sha256_n32_w4, 32, 4, 16, 67, 4 },
+ { lmots_sha256_n32_w8, 32, 8, 256, 34, 0 },
};
typedef struct lmots_key {
@@ -265,32 +259,21 @@ static hal_error_t lmots_generate(lmots_key_t * const key)
}
#endif
-/* coef() functions for the supported values of w.
- * This is a bit of premature optimization, because coef() gets called a lot.
- */
-
-/* w = 1 */
-static uint8_t coef1(const uint8_t * const S, const size_t i)
-{
- return (S[i/8] >> (7 - (i % 8))) & 0x01;
-}
-
-/* w = 2 */
-static uint8_t coef2(const uint8_t * const S, const size_t i)
-{
- return (S[i/4] >> (6 - (2 * (i % 4)))) & 0x03;
-}
-
-/* w = 4 */
-static uint8_t coef4(const uint8_t * const S, const size_t i)
+/* strings of w-bit elements */
+static uint8_t coef(const uint8_t * const S, const size_t i, size_t w)
{
- return (S[i/2] >> (4 - (4 * (i % 2)))) & 0x0f;
-}
-
-/* w = 8 */
-static uint8_t coef8(const uint8_t * const S, const size_t i)
-{
- return S[i];
+ switch (w) {
+ case 1:
+ return (S[i/8] >> (7 - (i % 8))) & 0x01;
+ case 2:
+ return (S[i/4] >> (6 - (2 * (i % 4)))) & 0x03;
+ case 4:
+ return (S[i/2] >> (4 - (4 * (i % 2)))) & 0x0f;
+ case 8:
+ return S[i];
+ default:
+ return 0;
+ }
}
/* checksum */
@@ -299,7 +282,7 @@ static uint16_t Cksm(const uint8_t * const S, lmots_parameter_t *lmots)
uint16_t sum = 0;
for (size_t i = 0; i < (lmots->n * 8 / lmots->w); ++i)
- sum += (lmots->w2 - 1) - lmots->coef(S, i);
+ sum += (lmots->w2 - 1) - coef(S, i, lmots->w);
return (sum << lmots->ls);
}
@@ -321,7 +304,7 @@ static hal_error_t lmots_sign(lmots_key_t *key,
size_t n = key->lmots->n;
size_t p = key->lmots->p;
- uint8_t (*coef)() = key->lmots->coef;
+ size_t w = key->lmots->w;
if (sig_max < lmots_signature_len(key->lmots))
return HAL_ERROR_BAD_ARGUMENTS;
@@ -360,7 +343,7 @@ static hal_error_t lmots_sign(lmots_key_t *key,
for (size_t i = 0; i < p; ++i) {
// a = coef(Q || Cksm(Q), i, w)
- uint8_t a = coef(Q, i);
+ uint8_t a = coef(Q, i, w);
// tmp = x[i]
bytestring32 tmp;
@@ -431,8 +414,8 @@ static hal_error_t lmots_public_key_candidate(const lmots_key_t * const key,
size_t n = key->lmots->n;
size_t p = key->lmots->p;
+ size_t w = key->lmots->w;
size_t w2 = key->lmots->w2;
- uint8_t (*coef)() = key->lmots->coef;
// d. C = next n bytes of signature
@@ -475,7 +458,7 @@ static hal_error_t lmots_public_key_candidate(const lmots_key_t * const key,
for (size_t i = 0; i < p; ++i) {
// a = coef(Q || Cksm(Q), i, w)
- uint8_t a = coef(Q, i);
+ uint8_t a = coef(Q, i, w);
// tmp = y[i]
bytestring32 tmp;
More information about the Commits
mailing list