Skip to content

Commit

Permalink
Remove unused scalar_sqr
Browse files Browse the repository at this point in the history
  • Loading branch information
@sipa
sipa committed Mar 12, 2021
1 parent aa9cc52 commit 5437e7b
Show file tree
Hide file tree
Showing 6 changed files with 1 addition and 287 deletions.
10 changes: 0 additions & 10 deletions src/bench_internal.c
View file
Expand Up @@ -99,15 +99,6 @@ void bench_scalar_negate(void* arg, int iters) {
}
}

void bench_scalar_sqr(void* arg, int iters) {
int i;
bench_inv *data = (bench_inv*)arg;

for (i = 0; i < iters; i++) {
secp256k1_scalar_sqr(&data->scalar[0], &data->scalar[0]);
}
}

void bench_scalar_mul(void* arg, int iters) {
int i;
bench_inv *data = (bench_inv*)arg;
Expand Down Expand Up @@ -393,7 +384,6 @@ int main(int argc, char **argv) {

if (have_flag(argc, argv, "scalar") || have_flag(argc, argv, "add")) run_benchmark("scalar_add", bench_scalar_add, bench_setup, NULL, &data, 10, iters*100);
if (have_flag(argc, argv, "scalar") || have_flag(argc, argv, "negate")) run_benchmark("scalar_negate", bench_scalar_negate, bench_setup, NULL, &data, 10, iters*100);
if (have_flag(argc, argv, "scalar") || have_flag(argc, argv, "sqr")) run_benchmark("scalar_sqr", bench_scalar_sqr, bench_setup, NULL, &data, 10, iters*10);
if (have_flag(argc, argv, "scalar") || have_flag(argc, argv, "mul")) run_benchmark("scalar_mul", bench_scalar_mul, bench_setup, NULL, &data, 10, iters*10);
if (have_flag(argc, argv, "scalar") || have_flag(argc, argv, "split")) run_benchmark("scalar_split", bench_scalar_split, bench_setup, NULL, &data, 10, iters);
if (have_flag(argc, argv, "scalar") || have_flag(argc, argv, "inverse")) run_benchmark("scalar_inverse", bench_scalar_inverse, bench_setup, NULL, &data, 10, 2000);
Expand Down
3 changes: 0 additions & 3 deletions src/scalar.h
View file
Expand Up @@ -63,9 +63,6 @@ static void secp256k1_scalar_mul(secp256k1_scalar *r, const secp256k1_scalar *a,
* the low bits that were shifted off */
static int secp256k1_scalar_shr_int(secp256k1_scalar *r, int n);

/** Compute the square of a scalar (modulo the group order). */
static void secp256k1_scalar_sqr(secp256k1_scalar *r, const secp256k1_scalar *a);

/** Compute the inverse of a scalar (modulo the group order). */
static void secp256k1_scalar_inverse(secp256k1_scalar *r, const secp256k1_scalar *a);

Expand Down
166 changes: 0 additions & 166 deletions src/scalar_4x64_impl.h
View file
Expand Up @@ -214,28 +214,6 @@ static int secp256k1_scalar_cond_negate(secp256k1_scalar *r, int flag) {
VERIFY_CHECK(c1 >= th); \
}

/** Add 2*a*b to the number defined by (c0,c1,c2). c2 must never overflow. */
#define muladd2(a,b) { \
uint64_t tl, th, th2, tl2; \
{ \
uint128_t t = (uint128_t)a * b; \
th = t >> 64; /* at most 0xFFFFFFFFFFFFFFFE */ \
tl = t; \
} \
th2 = th + th; /* at most 0xFFFFFFFFFFFFFFFE (in case th was 0x7FFFFFFFFFFFFFFF) */ \
c2 += (th2 < th); /* never overflows by contract (verified the next line) */ \
VERIFY_CHECK((th2 >= th) || (c2 != 0)); \
tl2 = tl + tl; /* at most 0xFFFFFFFFFFFFFFFE (in case the lowest 63 bits of tl were 0x7FFFFFFFFFFFFFFF) */ \
th2 += (tl2 < tl); /* at most 0xFFFFFFFFFFFFFFFF */ \
c0 += tl2; /* overflow is handled on the next line */ \
th2 += (c0 < tl2); /* second overflow is handled on the next line */ \
c2 += (c0 < tl2) & (th2 == 0); /* never overflows by contract (verified the next line) */ \
VERIFY_CHECK((c0 >= tl2) || (th2 != 0) || (c2 != 0)); \
c1 += th2; /* overflow is handled on the next line */ \
c2 += (c1 < th2); /* never overflows by contract (verified the next line) */ \
VERIFY_CHECK((c1 >= th2) || (c2 != 0)); \
}

/** Add a to the number defined by (c0,c1,c2). c2 must never overflow. */
#define sumadd(a) { \
unsigned int over; \
Expand Down Expand Up @@ -745,148 +723,10 @@ static void secp256k1_scalar_mul_512(uint64_t l[8], const secp256k1_scalar *a, c
#endif
}

static void secp256k1_scalar_sqr_512(uint64_t l[8], const secp256k1_scalar *a) {
#ifdef USE_ASM_X86_64
__asm__ __volatile__(
/* Preload */
"movq 0(%%rdi), %%r11\n"
"movq 8(%%rdi), %%r12\n"
"movq 16(%%rdi), %%r13\n"
"movq 24(%%rdi), %%r14\n"
/* (rax,rdx) = a0 * a0 */
"movq %%r11, %%rax\n"
"mulq %%r11\n"
/* Extract l0 */
"movq %%rax, 0(%%rsi)\n"
/* (r8,r9,r10) = (rdx,0) */
"movq %%rdx, %%r8\n"
"xorq %%r9, %%r9\n"
"xorq %%r10, %%r10\n"
/* (r8,r9,r10) += 2 * a0 * a1 */
"movq %%r11, %%rax\n"
"mulq %%r12\n"
"addq %%rax, %%r8\n"
"adcq %%rdx, %%r9\n"
"adcq $0, %%r10\n"
"addq %%rax, %%r8\n"
"adcq %%rdx, %%r9\n"
"adcq $0, %%r10\n"
/* Extract l1 */
"movq %%r8, 8(%%rsi)\n"
"xorq %%r8, %%r8\n"
/* (r9,r10,r8) += 2 * a0 * a2 */
"movq %%r11, %%rax\n"
"mulq %%r13\n"
"addq %%rax, %%r9\n"
"adcq %%rdx, %%r10\n"
"adcq $0, %%r8\n"
"addq %%rax, %%r9\n"
"adcq %%rdx, %%r10\n"
"adcq $0, %%r8\n"
/* (r9,r10,r8) += a1 * a1 */
"movq %%r12, %%rax\n"
"mulq %%r12\n"
"addq %%rax, %%r9\n"
"adcq %%rdx, %%r10\n"
"adcq $0, %%r8\n"
/* Extract l2 */
"movq %%r9, 16(%%rsi)\n"
"xorq %%r9, %%r9\n"
/* (r10,r8,r9) += 2 * a0 * a3 */
"movq %%r11, %%rax\n"
"mulq %%r14\n"
"addq %%rax, %%r10\n"
"adcq %%rdx, %%r8\n"
"adcq $0, %%r9\n"
"addq %%rax, %%r10\n"
"adcq %%rdx, %%r8\n"
"adcq $0, %%r9\n"
/* (r10,r8,r9) += 2 * a1 * a2 */
"movq %%r12, %%rax\n"
"mulq %%r13\n"
"addq %%rax, %%r10\n"
"adcq %%rdx, %%r8\n"
"adcq $0, %%r9\n"
"addq %%rax, %%r10\n"
"adcq %%rdx, %%r8\n"
"adcq $0, %%r9\n"
/* Extract l3 */
"movq %%r10, 24(%%rsi)\n"
"xorq %%r10, %%r10\n"
/* (r8,r9,r10) += 2 * a1 * a3 */
"movq %%r12, %%rax\n"
"mulq %%r14\n"
"addq %%rax, %%r8\n"
"adcq %%rdx, %%r9\n"
"adcq $0, %%r10\n"
"addq %%rax, %%r8\n"
"adcq %%rdx, %%r9\n"
"adcq $0, %%r10\n"
/* (r8,r9,r10) += a2 * a2 */
"movq %%r13, %%rax\n"
"mulq %%r13\n"
"addq %%rax, %%r8\n"
"adcq %%rdx, %%r9\n"
"adcq $0, %%r10\n"
/* Extract l4 */
"movq %%r8, 32(%%rsi)\n"
"xorq %%r8, %%r8\n"
/* (r9,r10,r8) += 2 * a2 * a3 */
"movq %%r13, %%rax\n"
"mulq %%r14\n"
"addq %%rax, %%r9\n"
"adcq %%rdx, %%r10\n"
"adcq $0, %%r8\n"
"addq %%rax, %%r9\n"
"adcq %%rdx, %%r10\n"
"adcq $0, %%r8\n"
/* Extract l5 */
"movq %%r9, 40(%%rsi)\n"
/* (r10,r8) += a3 * a3 */
"movq %%r14, %%rax\n"
"mulq %%r14\n"
"addq %%rax, %%r10\n"
"adcq %%rdx, %%r8\n"
/* Extract l6 */
"movq %%r10, 48(%%rsi)\n"
/* Extract l7 */
"movq %%r8, 56(%%rsi)\n"
:
: "S"(l), "D"(a->d)
: "rax", "rdx", "r8", "r9", "r10", "r11", "r12", "r13", "r14", "cc", "memory");
#else
/* 160 bit accumulator. */
uint64_t c0 = 0, c1 = 0;
uint32_t c2 = 0;

/* l[0..7] = a[0..3] * b[0..3]. */
muladd_fast(a->d[0], a->d[0]);
extract_fast(l[0]);
muladd2(a->d[0], a->d[1]);
extract(l[1]);
muladd2(a->d[0], a->d[2]);
muladd(a->d[1], a->d[1]);
extract(l[2]);
muladd2(a->d[0], a->d[3]);
muladd2(a->d[1], a->d[2]);
extract(l[3]);
muladd2(a->d[1], a->d[3]);
muladd(a->d[2], a->d[2]);
extract(l[4]);
muladd2(a->d[2], a->d[3]);
extract(l[5]);
muladd_fast(a->d[3], a->d[3]);
extract_fast(l[6]);
VERIFY_CHECK(c1 == 0);
l[7] = c0;
#endif
}

#undef sumadd
#undef sumadd_fast
#undef muladd
#undef muladd_fast
#undef muladd2
#undef extract
#undef extract_fast

Expand All @@ -908,12 +748,6 @@ static int secp256k1_scalar_shr_int(secp256k1_scalar *r, int n) {
return ret;
}

static void secp256k1_scalar_sqr(secp256k1_scalar *r, const secp256k1_scalar *a) {
uint64_t l[8];
secp256k1_scalar_sqr_512(l, a);
secp256k1_scalar_reduce_512(r, l);
}

static void secp256k1_scalar_split_128(secp256k1_scalar *r1, secp256k1_scalar *r2, const secp256k1_scalar *k) {
r1->d[0] = k->d[0];
r1->d[1] = k->d[1];
Expand Down
89 changes: 0 additions & 89 deletions src/scalar_8x32_impl.h
View file
Expand Up @@ -293,28 +293,6 @@ static int secp256k1_scalar_cond_negate(secp256k1_scalar *r, int flag) {
VERIFY_CHECK(c1 >= th); \
}

/** Add 2*a*b to the number defined by (c0,c1,c2). c2 must never overflow. */
#define muladd2(a,b) { \
uint32_t tl, th, th2, tl2; \
{ \
uint64_t t = (uint64_t)a * b; \
th = t >> 32; /* at most 0xFFFFFFFE */ \
tl = t; \
} \
th2 = th + th; /* at most 0xFFFFFFFE (in case th was 0x7FFFFFFF) */ \
c2 += (th2 < th); /* never overflows by contract (verified the next line) */ \
VERIFY_CHECK((th2 >= th) || (c2 != 0)); \
tl2 = tl + tl; /* at most 0xFFFFFFFE (in case the lowest 63 bits of tl were 0x7FFFFFFF) */ \
th2 += (tl2 < tl); /* at most 0xFFFFFFFF */ \
c0 += tl2; /* overflow is handled on the next line */ \
th2 += (c0 < tl2); /* second overflow is handled on the next line */ \
c2 += (c0 < tl2) & (th2 == 0); /* never overflows by contract (verified the next line) */ \
VERIFY_CHECK((c0 >= tl2) || (th2 != 0) || (c2 != 0)); \
c1 += th2; /* overflow is handled on the next line */ \
c2 += (c1 < th2); /* never overflows by contract (verified the next line) */ \
VERIFY_CHECK((c1 >= th2) || (c2 != 0)); \
}

/** Add a to the number defined by (c0,c1,c2). c2 must never overflow. */
#define sumadd(a) { \
unsigned int over; \
Expand Down Expand Up @@ -578,71 +556,10 @@ static void secp256k1_scalar_mul_512(uint32_t *l, const secp256k1_scalar *a, con
l[15] = c0;
}

static void secp256k1_scalar_sqr_512(uint32_t *l, const secp256k1_scalar *a) {
/* 96 bit accumulator. */
uint32_t c0 = 0, c1 = 0, c2 = 0;

/* l[0..15] = a[0..7]^2. */
muladd_fast(a->d[0], a->d[0]);
extract_fast(l[0]);
muladd2(a->d[0], a->d[1]);
extract(l[1]);
muladd2(a->d[0], a->d[2]);
muladd(a->d[1], a->d[1]);
extract(l[2]);
muladd2(a->d[0], a->d[3]);
muladd2(a->d[1], a->d[2]);
extract(l[3]);
muladd2(a->d[0], a->d[4]);
muladd2(a->d[1], a->d[3]);
muladd(a->d[2], a->d[2]);
extract(l[4]);
muladd2(a->d[0], a->d[5]);
muladd2(a->d[1], a->d[4]);
muladd2(a->d[2], a->d[3]);
extract(l[5]);
muladd2(a->d[0], a->d[6]);
muladd2(a->d[1], a->d[5]);
muladd2(a->d[2], a->d[4]);
muladd(a->d[3], a->d[3]);
extract(l[6]);
muladd2(a->d[0], a->d[7]);
muladd2(a->d[1], a->d[6]);
muladd2(a->d[2], a->d[5]);
muladd2(a->d[3], a->d[4]);
extract(l[7]);
muladd2(a->d[1], a->d[7]);
muladd2(a->d[2], a->d[6]);
muladd2(a->d[3], a->d[5]);
muladd(a->d[4], a->d[4]);
extract(l[8]);
muladd2(a->d[2], a->d[7]);
muladd2(a->d[3], a->d[6]);
muladd2(a->d[4], a->d[5]);
extract(l[9]);
muladd2(a->d[3], a->d[7]);
muladd2(a->d[4], a->d[6]);
muladd(a->d[5], a->d[5]);
extract(l[10]);
muladd2(a->d[4], a->d[7]);
muladd2(a->d[5], a->d[6]);
extract(l[11]);
muladd2(a->d[5], a->d[7]);
muladd(a->d[6], a->d[6]);
extract(l[12]);
muladd2(a->d[6], a->d[7]);
extract(l[13]);
muladd_fast(a->d[7], a->d[7]);
extract_fast(l[14]);
VERIFY_CHECK(c1 == 0);
l[15] = c0;
}

#undef sumadd
#undef sumadd_fast
#undef muladd
#undef muladd_fast
#undef muladd2
#undef extract
#undef extract_fast

Expand All @@ -668,12 +585,6 @@ static int secp256k1_scalar_shr_int(secp256k1_scalar *r, int n) {
return ret;
}

static void secp256k1_scalar_sqr(secp256k1_scalar *r, const secp256k1_scalar *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 *r1, secp256k1_scalar *r2, const secp256k1_scalar *k) {
r1->d[0] = k->d[0];
r1->d[1] = k->d[1];
Expand Down
4 changes: 0 additions & 4 deletions src/scalar_low_impl.h
View file
Expand Up @@ -104,10 +104,6 @@ static int secp256k1_scalar_shr_int(secp256k1_scalar *r, int n) {
return ret;
}

static void secp256k1_scalar_sqr(secp256k1_scalar *r, const secp256k1_scalar *a) {
*r = (*a * *a) % EXHAUSTIVE_TEST_ORDER;
}

static void secp256k1_scalar_split_128(secp256k1_scalar *r1, secp256k1_scalar *r2, const secp256k1_scalar *a) {
*r1 = *a;
*r2 = 0;
Expand Down
16 changes: 1 addition & 15 deletions src/tests.c
View file
Expand Up @@ -793,7 +793,7 @@ void test_num_jacobi(void) {
/** test with secp group order as order */
secp256k1_scalar_order_get_num(&order);
random_scalar_order_test(&sqr);
secp256k1_scalar_sqr(&sqr, &sqr);
secp256k1_scalar_mul(&sqr, &sqr, &sqr);
/* test residue */
secp256k1_scalar_get_num(&n, &sqr);
CHECK(secp256k1_num_jacobi(&n, &order) == 1);
Expand Down Expand Up @@ -1488,14 +1488,6 @@ void scalar_test(void) {
CHECK(secp256k1_scalar_eq(&r1, &r2));
}

{
/* Test square. */
secp256k1_scalar r1, r2;
secp256k1_scalar_sqr(&r1, &s1);
secp256k1_scalar_mul(&r2, &s1, &s1);
CHECK(secp256k1_scalar_eq(&r1, &r2));
}

{
/* Test multiplicative identity. */
secp256k1_scalar r1, v1;
Expand Down Expand Up @@ -2187,12 +2179,6 @@ void run_scalar_tests(void) {
CHECK(!secp256k1_scalar_check_overflow(&zz));
CHECK(secp256k1_scalar_eq(&one, &zz));
}
secp256k1_scalar_mul(&z, &x, &x);
CHECK(!secp256k1_scalar_check_overflow(&z));
secp256k1_scalar_sqr(&zz, &x);
CHECK(!secp256k1_scalar_check_overflow(&zz));
CHECK(secp256k1_scalar_eq(&zz, &z));
CHECK(secp256k1_scalar_eq(&r2, &zz));
}
}
}
Expand Down

0 comments on commit 5437e7b

Please sign in to comment.