aboutsummaryrefslogtreecommitdiff
path: root/src/scalar_8x32_impl.h
diff options
context:
space:
mode:
Diffstat (limited to 'src/scalar_8x32_impl.h')
-rw-r--r--src/scalar_8x32_impl.h126
1 files changed, 111 insertions, 15 deletions
diff --git a/src/scalar_8x32_impl.h b/src/scalar_8x32_impl.h
index e58be1365f..915cbcddbe 100644
--- a/src/scalar_8x32_impl.h
+++ b/src/scalar_8x32_impl.h
@@ -45,9 +45,31 @@ SECP256K1_INLINE static void secp256k1_scalar_clear(secp256k1_scalar_t *r) {
r->d[7] = 0;
}
-SECP256K1_INLINE static int secp256k1_scalar_get_bits(const secp256k1_scalar_t *a, int offset, int count) {
- VERIFY_CHECK((offset + count - 1) / 32 == offset / 32);
- return (a->d[offset / 32] >> (offset % 32)) & ((1 << count) - 1);
+SECP256K1_INLINE static void secp256k1_scalar_set_int(secp256k1_scalar_t *r, unsigned int v) {
+ r->d[0] = v;
+ r->d[1] = 0;
+ r->d[2] = 0;
+ r->d[3] = 0;
+ r->d[4] = 0;
+ r->d[5] = 0;
+ r->d[6] = 0;
+ r->d[7] = 0;
+}
+
+SECP256K1_INLINE static unsigned int secp256k1_scalar_get_bits(const secp256k1_scalar_t *a, unsigned int offset, unsigned int count) {
+ VERIFY_CHECK((offset + count - 1) >> 5 == offset >> 5);
+ return (a->d[offset >> 5] >> (offset & 0x1F)) & ((1 << count) - 1);
+}
+
+SECP256K1_INLINE static unsigned int secp256k1_scalar_get_bits_var(const secp256k1_scalar_t *a, unsigned int offset, unsigned int count) {
+ VERIFY_CHECK(count < 32);
+ VERIFY_CHECK(offset + count <= 256);
+ if ((offset + count - 1) >> 5 == offset >> 5) {
+ return secp256k1_scalar_get_bits(a, offset, count);
+ } else {
+ VERIFY_CHECK((offset >> 5) + 1 < 8);
+ return ((a->d[offset >> 5] >> (offset & 0x1F)) | (a->d[(offset >> 5) + 1] << (32 - (offset & 0x1F)))) & ((((uint32_t)1) << count) - 1);
+ }
}
SECP256K1_INLINE static int secp256k1_scalar_check_overflow(const secp256k1_scalar_t *a) {
@@ -89,7 +111,7 @@ SECP256K1_INLINE static int secp256k1_scalar_reduce(secp256k1_scalar_t *r, uint3
return overflow;
}
-static void secp256k1_scalar_add(secp256k1_scalar_t *r, const secp256k1_scalar_t *a, const secp256k1_scalar_t *b) {
+static int secp256k1_scalar_add(secp256k1_scalar_t *r, const secp256k1_scalar_t *a, const secp256k1_scalar_t *b) {
uint64_t t = (uint64_t)a->d[0] + b->d[0];
r->d[0] = t & 0xFFFFFFFFULL; t >>= 32;
t += (uint64_t)a->d[1] + b->d[1];
@@ -106,7 +128,34 @@ static void secp256k1_scalar_add(secp256k1_scalar_t *r, const secp256k1_scalar_t
r->d[6] = t & 0xFFFFFFFFULL; t >>= 32;
t += (uint64_t)a->d[7] + b->d[7];
r->d[7] = t & 0xFFFFFFFFULL; t >>= 32;
- secp256k1_scalar_reduce(r, t + secp256k1_scalar_check_overflow(r));
+ int overflow = t + secp256k1_scalar_check_overflow(r);
+ VERIFY_CHECK(overflow == 0 || overflow == 1);
+ secp256k1_scalar_reduce(r, overflow);
+ return overflow;
+}
+
+static void secp256k1_scalar_add_bit(secp256k1_scalar_t *r, unsigned int bit) {
+ VERIFY_CHECK(bit < 256);
+ uint64_t t = (uint64_t)r->d[0] + (((uint32_t)((bit >> 5) == 0)) << (bit & 0x1F));
+ r->d[0] = t & 0xFFFFFFFFULL; t >>= 32;
+ t += (uint64_t)r->d[1] + (((uint32_t)((bit >> 5) == 1)) << (bit & 0x1F));
+ r->d[1] = t & 0xFFFFFFFFULL; t >>= 32;
+ t += (uint64_t)r->d[2] + (((uint32_t)((bit >> 5) == 2)) << (bit & 0x1F));
+ r->d[2] = t & 0xFFFFFFFFULL; t >>= 32;
+ t += (uint64_t)r->d[3] + (((uint32_t)((bit >> 5) == 3)) << (bit & 0x1F));
+ r->d[3] = t & 0xFFFFFFFFULL; t >>= 32;
+ t += (uint64_t)r->d[4] + (((uint32_t)((bit >> 5) == 4)) << (bit & 0x1F));
+ r->d[4] = t & 0xFFFFFFFFULL; t >>= 32;
+ t += (uint64_t)r->d[5] + (((uint32_t)((bit >> 5) == 5)) << (bit & 0x1F));
+ r->d[5] = t & 0xFFFFFFFFULL; t >>= 32;
+ t += (uint64_t)r->d[6] + (((uint32_t)((bit >> 5) == 6)) << (bit & 0x1F));
+ r->d[6] = t & 0xFFFFFFFFULL; t >>= 32;
+ t += (uint64_t)r->d[7] + (((uint32_t)((bit >> 5) == 7)) << (bit & 0x1F));
+ r->d[7] = t & 0xFFFFFFFFULL;
+#ifdef VERIFY
+ VERIFY_CHECK((t >> 32) == 0);
+ VERIFY_CHECK(secp256k1_scalar_check_overflow(r) == 0);
+#endif
}
static void secp256k1_scalar_set_b32(secp256k1_scalar_t *r, const unsigned char *b32, int *overflow) {
@@ -405,12 +454,10 @@ static void secp256k1_scalar_reduce_512(secp256k1_scalar_t *r, const uint32_t *l
secp256k1_scalar_reduce(r, c + secp256k1_scalar_check_overflow(r));
}
-static void secp256k1_scalar_mul(secp256k1_scalar_t *r, const secp256k1_scalar_t *a, const secp256k1_scalar_t *b) {
+static void secp256k1_scalar_mul_512(uint32_t l[16], const secp256k1_scalar_t *a, const secp256k1_scalar_t *b) {
/* 96 bit accumulator. */
uint32_t c0 = 0, c1 = 0, c2 = 0;
- uint32_t l[16];
-
/* l[0..15] = a[0..7] * b[0..7]. */
muladd_fast(a->d[0], b->d[0]);
extract_fast(l[0]);
@@ -493,16 +540,12 @@ static void secp256k1_scalar_mul(secp256k1_scalar_t *r, const secp256k1_scalar_t
extract_fast(l[14]);
VERIFY_CHECK(c1 == 0);
l[15] = c0;
-
- secp256k1_scalar_reduce_512(r, l);
}
-static void secp256k1_scalar_sqr(secp256k1_scalar_t *r, const secp256k1_scalar_t *a) {
+static void secp256k1_scalar_sqr_512(uint32_t l[16], const secp256k1_scalar_t *a) {
/* 96 bit accumulator. */
uint32_t c0 = 0, c1 = 0, c2 = 0;
- uint32_t l[16];
-
/* l[0..15] = a[0..7]^2. */
muladd_fast(a->d[0], a->d[0]);
extract_fast(l[0]);
@@ -557,8 +600,6 @@ static void secp256k1_scalar_sqr(secp256k1_scalar_t *r, const secp256k1_scalar_t
extract_fast(l[14]);
VERIFY_CHECK(c1 == 0);
l[15] = c0;
-
- secp256k1_scalar_reduce_512(r, l);
}
#undef sumadd
@@ -569,4 +610,59 @@ static void secp256k1_scalar_sqr(secp256k1_scalar_t *r, const secp256k1_scalar_t
#undef extract
#undef extract_fast
+static void secp256k1_scalar_mul(secp256k1_scalar_t *r, const secp256k1_scalar_t *a, const secp256k1_scalar_t *b) {
+ uint32_t l[16];
+ secp256k1_scalar_mul_512(l, a, b);
+ secp256k1_scalar_reduce_512(r, l);
+}
+
+static void secp256k1_scalar_sqr(secp256k1_scalar_t *r, const secp256k1_scalar_t *a) {
+ uint32_t l[16];
+ secp256k1_scalar_sqr_512(l, a);
+ secp256k1_scalar_reduce_512(r, l);
+}
+
+static void secp256k1_scalar_split_128(secp256k1_scalar_t *r1, secp256k1_scalar_t *r2, const secp256k1_scalar_t *a) {
+ r1->d[0] = a->d[0];
+ r1->d[1] = a->d[1];
+ r1->d[2] = a->d[2];
+ r1->d[3] = a->d[3];
+ r1->d[4] = 0;
+ r1->d[5] = 0;
+ r1->d[6] = 0;
+ r1->d[7] = 0;
+ r2->d[0] = a->d[4];
+ r2->d[1] = a->d[5];
+ r2->d[2] = a->d[6];
+ r2->d[3] = a->d[7];
+ r2->d[4] = 0;
+ r2->d[5] = 0;
+ r2->d[6] = 0;
+ r2->d[7] = 0;
+}
+
+SECP256K1_INLINE static int secp256k1_scalar_eq(const secp256k1_scalar_t *a, const secp256k1_scalar_t *b) {
+ return ((a->d[0] ^ b->d[0]) | (a->d[1] ^ b->d[1]) | (a->d[2] ^ b->d[2]) | (a->d[3] ^ b->d[3]) | (a->d[4] ^ b->d[4]) | (a->d[5] ^ b->d[5]) | (a->d[6] ^ b->d[6]) | (a->d[7] ^ b->d[7])) == 0;
+}
+
+SECP256K1_INLINE static void secp256k1_scalar_mul_shift_var(secp256k1_scalar_t *r, const secp256k1_scalar_t *a, const secp256k1_scalar_t *b, unsigned int shift) {
+ VERIFY_CHECK(shift >= 256);
+ uint32_t l[16];
+ secp256k1_scalar_mul_512(l, a, b);
+ unsigned int shiftlimbs = shift >> 5;
+ unsigned int shiftlow = shift & 0x1F;
+ unsigned int shifthigh = 32 - shiftlow;
+ r->d[0] = shift < 512 ? (l[0 + shiftlimbs] >> shiftlow | (shift < 480 && shiftlow ? (l[1 + shiftlimbs] << shifthigh) : 0)) : 0;
+ r->d[1] = shift < 480 ? (l[1 + shiftlimbs] >> shiftlow | (shift < 448 && shiftlow ? (l[2 + shiftlimbs] << shifthigh) : 0)) : 0;
+ r->d[2] = shift < 448 ? (l[2 + shiftlimbs] >> shiftlow | (shift < 416 && shiftlow ? (l[3 + shiftlimbs] << shifthigh) : 0)) : 0;
+ r->d[3] = shift < 416 ? (l[3 + shiftlimbs] >> shiftlow | (shift < 384 && shiftlow ? (l[4 + shiftlimbs] << shifthigh) : 0)) : 0;
+ r->d[4] = shift < 384 ? (l[4 + shiftlimbs] >> shiftlow | (shift < 352 && shiftlow ? (l[5 + shiftlimbs] << shifthigh) : 0)) : 0;
+ r->d[5] = shift < 352 ? (l[5 + shiftlimbs] >> shiftlow | (shift < 320 && shiftlow ? (l[6 + shiftlimbs] << shifthigh) : 0)) : 0;
+ r->d[6] = shift < 320 ? (l[6 + shiftlimbs] >> shiftlow | (shift < 288 && shiftlow ? (l[7 + shiftlimbs] << shifthigh) : 0)) : 0;
+ r->d[7] = shift < 288 ? (l[7 + shiftlimbs] >> shiftlow) : 0;
+ if ((l[(shift - 1) >> 5] >> ((shift - 1) & 0x1f)) & 1) {
+ secp256k1_scalar_add_bit(r, 0);
+ }
+}
+
#endif
generated by cgit v1.2.3 (git 2.26.2) at 2024-11-14 22:16:30 +0000