Move 5x52 specific code to field_5x52
This commit is contained in:
parent
16fbc0f281
commit
e6d142a8dc
6
Makefile
6
Makefile
|
@ -3,8 +3,8 @@ FLAGS_PROD:=-DNDEBUG -O2 -march=native
|
|||
FLAGS_DEBUG:=-DVERIFY_MAGNITUDE -ggdb3 -O1
|
||||
FLAGS_TEST:=-DVERIFY_MAGNITUDE -ggdb3 -O2 -march=native
|
||||
|
||||
SECP256K1_FILES := num.h field.h group.h ecmult.h ecdsa.h \
|
||||
num.cpp field.cpp group.cpp ecmult.cpp ecdsa.cpp
|
||||
SECP256K1_FILES := num.h field.h field_5x52.h group.h ecmult.h ecdsa.h \
|
||||
num.cpp field.cpp field_5x52.cpp group.cpp ecmult.cpp ecdsa.cpp
|
||||
|
||||
ifndef CONF
|
||||
CONF := gmp
|
||||
|
@ -46,7 +46,7 @@ clean:
|
|||
bench-any: bench-$(CONF)
|
||||
tests-any: tests-$(CONF)
|
||||
|
||||
all-$(CONF): bench-$(CONF) tests-$(CONF)
|
||||
all-$(CONF): bench-$(CONF) tests-$(CONF) obj/secp256k1-$(CONF).o
|
||||
|
||||
clean-$(CONF):
|
||||
rm -f bench-$(CONF) tests-$(CONF) obj/secp256k1-$(CONF).o
|
||||
|
|
409
field.cpp
409
field.cpp
|
@ -1,363 +1,8 @@
|
|||
#include <assert.h>
|
||||
#include <stdint.h>
|
||||
#include <string>
|
||||
#include "num.h"
|
||||
#include "field.h"
|
||||
#include <iostream>
|
||||
|
||||
using namespace std;
|
||||
|
||||
#ifdef INLINE_ASM
|
||||
#include "lin64.h"
|
||||
#endif
|
||||
// just one implementation for now
|
||||
#include "field_5x52.cpp"
|
||||
|
||||
namespace secp256k1 {
|
||||
|
||||
/** Implements arithmetic modulo FFFFFFFF FFFFFFFF FFFFFFFF FFFFFFFF FFFFFFFF FFFFFFFF FFFFFFFE FFFFFC2F,
|
||||
* represented as 5 uint64_t's in base 2^52. The values are allowed to contain >52 each. In particular,
|
||||
* each FieldElem has a 'magnitude' associated with it. Internally, a magnitude M means each element
|
||||
* is at most M*(2^53-1), except the most significant one, which is limited to M*(2^49-1). All operations
|
||||
* accept any input with magnitude at most M, and have different rules for propagating magnitude to their
|
||||
* output.
|
||||
*/
|
||||
|
||||
FieldElem::FieldElem(int x) {
|
||||
n[0] = x;
|
||||
n[1] = n[2] = n[3] = n[4] = 0;
|
||||
#ifdef VERIFY_MAGNITUDE
|
||||
magnitude = 1;
|
||||
normalized = true;
|
||||
#endif
|
||||
}
|
||||
|
||||
FieldElem::FieldElem(const unsigned char *b32) {
|
||||
SetBytes(b32);
|
||||
}
|
||||
|
||||
void FieldElem::Normalize() {
|
||||
uint64_t c;
|
||||
c = n[0];
|
||||
uint64_t t0 = c & 0xFFFFFFFFFFFFFULL;
|
||||
c = (c >> 52) + n[1];
|
||||
uint64_t t1 = c & 0xFFFFFFFFFFFFFULL;
|
||||
c = (c >> 52) + n[2];
|
||||
uint64_t t2 = c & 0xFFFFFFFFFFFFFULL;
|
||||
c = (c >> 52) + n[3];
|
||||
uint64_t t3 = c & 0xFFFFFFFFFFFFFULL;
|
||||
c = (c >> 52) + n[4];
|
||||
uint64_t t4 = c & 0x0FFFFFFFFFFFFULL;
|
||||
c >>= 48;
|
||||
|
||||
// The following code will not modify the t's if c is initially 0.
|
||||
c = c * 0x1000003D1ULL + t0;
|
||||
t0 = c & 0xFFFFFFFFFFFFFULL;
|
||||
c = (c >> 52) + t1;
|
||||
t1 = c & 0xFFFFFFFFFFFFFULL;
|
||||
c = (c >> 52) + t2;
|
||||
t2 = c & 0xFFFFFFFFFFFFFULL;
|
||||
c = (c >> 52) + t3;
|
||||
t3 = c & 0xFFFFFFFFFFFFFULL;
|
||||
c = (c >> 52) + t4;
|
||||
t4 = c & 0x0FFFFFFFFFFFFULL;
|
||||
|
||||
// Replace n's with t's if one of the n's overflows.
|
||||
// If none of the n's overflow to begin with, the t's will just be the n's already and
|
||||
// we effectively ignore the results of the previous computations.
|
||||
n[0] = t0; n[1] = t1; n[2] = t2; n[3] = t3; n[4] = t4;
|
||||
|
||||
// Subtract p if result >= p
|
||||
uint64_t mask = -(int64_t)((n[4] < 0xFFFFFFFFFFFFULL) | (n[3] < 0xFFFFFFFFFFFFFULL) | (n[2] < 0xFFFFFFFFFFFFFULL) | (n[1] < 0xFFFFFFFFFFFFFULL) | (n[0] < 0xFFFFEFFFFFC2FULL));
|
||||
n[4] &= mask;
|
||||
n[3] &= mask;
|
||||
n[2] &= mask;
|
||||
n[1] &= mask;
|
||||
n[0] -= (~mask & 0xFFFFEFFFFFC2FULL);
|
||||
|
||||
#ifdef VERIFY_MAGNITUDE
|
||||
magnitude = 1;
|
||||
normalized = true;
|
||||
#endif
|
||||
}
|
||||
|
||||
bool inline FieldElem::IsZero() const {
|
||||
#ifdef VERIFY_MAGNITUDE
|
||||
assert(normalized);
|
||||
#endif
|
||||
return (n[0] == 0 && n[1] == 0 && n[2] == 0 && n[3] == 0 && n[4] == 0);
|
||||
}
|
||||
|
||||
bool inline operator==(const FieldElem &a, const FieldElem &b) {
|
||||
#ifdef VERIFY_MAGNITUDE
|
||||
assert(a.normalized);
|
||||
assert(b.normalized);
|
||||
#endif
|
||||
return (a.n[0] == b.n[0] && a.n[1] == b.n[1] && a.n[2] == b.n[2] && a.n[3] == b.n[3] && a.n[4] == b.n[4]);
|
||||
}
|
||||
|
||||
void FieldElem::GetBytes(unsigned char *o) {
|
||||
#ifdef VERIFY_MAGNITUDE
|
||||
assert(normalized);
|
||||
#endif
|
||||
for (int i=0; i<32; i++) {
|
||||
int c = 0;
|
||||
for (int j=0; j<2; j++) {
|
||||
int limb = (8*i+4*j)/52;
|
||||
int shift = (8*i+4*j)%52;
|
||||
c |= ((n[limb] >> shift) & 0xF) << (4 * j);
|
||||
}
|
||||
o[31-i] = c;
|
||||
}
|
||||
}
|
||||
|
||||
void FieldElem::SetBytes(const unsigned char *in) {
|
||||
n[0] = n[1] = n[2] = n[3] = n[4] = 0;
|
||||
for (int i=0; i<32; i++) {
|
||||
for (int j=0; j<2; j++) {
|
||||
int limb = (8*i+4*j)/52;
|
||||
int shift = (8*i+4*j)%52;
|
||||
n[limb] |= (uint64_t)((in[31-i] >> (4*j)) & 0xF) << shift;
|
||||
}
|
||||
}
|
||||
#ifdef VERIFY_MAGNITUDE
|
||||
magnitude = 1;
|
||||
normalized = true;
|
||||
#endif
|
||||
}
|
||||
|
||||
void inline FieldElem::SetNeg(const FieldElem &a, int magnitudeIn) {
|
||||
#ifdef VERIFY_MAGNITUDE
|
||||
assert(a.magnitude <= magnitudeIn);
|
||||
magnitude = magnitudeIn + 1;
|
||||
normalized = false;
|
||||
#endif
|
||||
n[0] = 0xFFFFEFFFFFC2FULL * (magnitudeIn + 1) - a.n[0];
|
||||
n[1] = 0xFFFFFFFFFFFFFULL * (magnitudeIn + 1) - a.n[1];
|
||||
n[2] = 0xFFFFFFFFFFFFFULL * (magnitudeIn + 1) - a.n[2];
|
||||
n[3] = 0xFFFFFFFFFFFFFULL * (magnitudeIn + 1) - a.n[3];
|
||||
n[4] = 0x0FFFFFFFFFFFFULL * (magnitudeIn + 1) - a.n[4];
|
||||
}
|
||||
|
||||
void inline FieldElem::operator*=(int v) {
|
||||
#ifdef VERIFY_MAGNITUDE
|
||||
magnitude *= v;
|
||||
normalized = false;
|
||||
#endif
|
||||
n[0] *= v;
|
||||
n[1] *= v;
|
||||
n[2] *= v;
|
||||
n[3] *= v;
|
||||
n[4] *= v;
|
||||
}
|
||||
|
||||
void inline FieldElem::operator+=(const FieldElem &a) {
|
||||
#ifdef VERIFY_MAGNITUDE
|
||||
magnitude += a.magnitude;
|
||||
normalized = false;
|
||||
#endif
|
||||
n[0] += a.n[0];
|
||||
n[1] += a.n[1];
|
||||
n[2] += a.n[2];
|
||||
n[3] += a.n[3];
|
||||
n[4] += a.n[4];
|
||||
}
|
||||
|
||||
void FieldElem::SetMult(const FieldElem &a, const FieldElem &b) {
|
||||
#ifdef VERIFY_MAGNITUDE
|
||||
assert(a.magnitude <= 8);
|
||||
assert(b.magnitude <= 8);
|
||||
#endif
|
||||
|
||||
#ifdef INLINE_ASM
|
||||
ExSetMult((uint64_t *) a.n,(uint64_t *) b.n, (uint64_t *) n);
|
||||
#else
|
||||
unsigned __int128 c = (__int128)a.n[0] * b.n[0];
|
||||
uint64_t t0 = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 0FFFFFFFFFFFFFE0
|
||||
c = c + (__int128)a.n[0] * b.n[1] +
|
||||
(__int128)a.n[1] * b.n[0];
|
||||
uint64_t t1 = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 20000000000000BF
|
||||
c = c + (__int128)a.n[0] * b.n[2] +
|
||||
(__int128)a.n[1] * b.n[1] +
|
||||
(__int128)a.n[2] * b.n[0];
|
||||
uint64_t t2 = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 30000000000001A0
|
||||
c = c + (__int128)a.n[0] * b.n[3] +
|
||||
(__int128)a.n[1] * b.n[2] +
|
||||
(__int128)a.n[2] * b.n[1] +
|
||||
(__int128)a.n[3] * b.n[0];
|
||||
uint64_t t3 = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 4000000000000280
|
||||
c = c + (__int128)a.n[0] * b.n[4] +
|
||||
(__int128)a.n[1] * b.n[3] +
|
||||
(__int128)a.n[2] * b.n[2] +
|
||||
(__int128)a.n[3] * b.n[1] +
|
||||
(__int128)a.n[4] * b.n[0];
|
||||
uint64_t t4 = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 320000000000037E
|
||||
c = c + (__int128)a.n[1] * b.n[4] +
|
||||
(__int128)a.n[2] * b.n[3] +
|
||||
(__int128)a.n[3] * b.n[2] +
|
||||
(__int128)a.n[4] * b.n[1];
|
||||
uint64_t t5 = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 22000000000002BE
|
||||
c = c + (__int128)a.n[2] * b.n[4] +
|
||||
(__int128)a.n[3] * b.n[3] +
|
||||
(__int128)a.n[4] * b.n[2];
|
||||
uint64_t t6 = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 12000000000001DE
|
||||
c = c + (__int128)a.n[3] * b.n[4] +
|
||||
(__int128)a.n[4] * b.n[3];
|
||||
uint64_t t7 = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 02000000000000FE
|
||||
c = c + (__int128)a.n[4] * b.n[4];
|
||||
uint64_t t8 = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 001000000000001E
|
||||
uint64_t t9 = c;
|
||||
|
||||
c = t0 + (__int128)t5 * 0x1000003D10ULL;
|
||||
t0 = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 0000001000003D10
|
||||
c = c + t1 + (__int128)t6 * 0x1000003D10ULL;
|
||||
t1 = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 0000001000003D10
|
||||
c = c + t2 + (__int128)t7 * 0x1000003D10ULL;
|
||||
n[2] = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 0000001000003D10
|
||||
c = c + t3 + (__int128)t8 * 0x1000003D10ULL;
|
||||
n[3] = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 0000001000003D10
|
||||
c = c + t4 + (__int128)t9 * 0x1000003D10ULL;
|
||||
n[4] = c & 0x0FFFFFFFFFFFFULL; c = c >> 48; // c max 000001000003D110
|
||||
c = t0 + (__int128)c * 0x1000003D1ULL;
|
||||
n[0] = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 1000008
|
||||
n[1] = t1 + c;
|
||||
#endif
|
||||
#ifdef VERIFY_MAGNITUDE
|
||||
magnitude = 1;
|
||||
normalized = false;
|
||||
#endif
|
||||
}
|
||||
|
||||
void FieldElem::SetSquare(const FieldElem &a) {
|
||||
#ifdef VERIFY_MAGNITUDE
|
||||
assert(a.magnitude <= 8);
|
||||
#endif
|
||||
|
||||
#ifdef INLINE_ASM
|
||||
ExSetSquare((uint64_t *)a.n,(uint64_t *)n);
|
||||
#else
|
||||
__int128 c = (__int128)a.n[0] * a.n[0];
|
||||
uint64_t t0 = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 0FFFFFFFFFFFFFE0
|
||||
c = c + (__int128)(a.n[0]*2) * a.n[1];
|
||||
uint64_t t1 = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 20000000000000BF
|
||||
c = c + (__int128)(a.n[0]*2) * a.n[2] +
|
||||
(__int128)a.n[1] * a.n[1];
|
||||
uint64_t t2 = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 30000000000001A0
|
||||
c = c + (__int128)(a.n[0]*2) * a.n[3] +
|
||||
(__int128)(a.n[1]*2) * a.n[2];
|
||||
uint64_t t3 = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 4000000000000280
|
||||
c = c + (__int128)(a.n[0]*2) * a.n[4] +
|
||||
(__int128)(a.n[1]*2) * a.n[3] +
|
||||
(__int128)a.n[2] * a.n[2];
|
||||
uint64_t t4 = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 320000000000037E
|
||||
c = c + (__int128)(a.n[1]*2) * a.n[4] +
|
||||
(__int128)(a.n[2]*2) * a.n[3];
|
||||
uint64_t t5 = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 22000000000002BE
|
||||
c = c + (__int128)(a.n[2]*2) * a.n[4] +
|
||||
(__int128)a.n[3] * a.n[3];
|
||||
uint64_t t6 = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 12000000000001DE
|
||||
c = c + (__int128)(a.n[3]*2) * a.n[4];
|
||||
uint64_t t7 = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 02000000000000FE
|
||||
c = c + (__int128)a.n[4] * a.n[4];
|
||||
uint64_t t8 = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 001000000000001E
|
||||
uint64_t t9 = c;
|
||||
c = t0 + (__int128)t5 * 0x1000003D10ULL;
|
||||
t0 = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 0000001000003D10
|
||||
c = c + t1 + (__int128)t6 * 0x1000003D10ULL;
|
||||
t1 = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 0000001000003D10
|
||||
c = c + t2 + (__int128)t7 * 0x1000003D10ULL;
|
||||
n[2] = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 0000001000003D10
|
||||
c = c + t3 + (__int128)t8 * 0x1000003D10ULL;
|
||||
n[3] = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 0000001000003D10
|
||||
c = c + t4 + (__int128)t9 * 0x1000003D10ULL;
|
||||
n[4] = c & 0x0FFFFFFFFFFFFULL; c = c >> 48; // c max 000001000003D110
|
||||
c = t0 + (__int128)c * 0x1000003D1ULL;
|
||||
n[0] = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 1000008
|
||||
n[1] = t1 + c;
|
||||
#endif
|
||||
|
||||
#ifdef VERIFY_MAGNITUDE
|
||||
assert(a.magnitude <= 8);
|
||||
normalized = false;
|
||||
#endif
|
||||
}
|
||||
|
||||
void FieldElem::SetSquareRoot(const FieldElem &a) {
|
||||
// calculate a^p, with p={15,780,1022,1023}
|
||||
FieldElem a2; a2.SetSquare(a);
|
||||
FieldElem a3; a3.SetMult(a2,a);
|
||||
FieldElem a6; a6.SetSquare(a3);
|
||||
FieldElem a12; a12.SetSquare(a6);
|
||||
FieldElem a15; a15.SetMult(a12,a3);
|
||||
FieldElem a30; a30.SetSquare(a15);
|
||||
FieldElem a60; a60.SetSquare(a30);
|
||||
FieldElem a120; a120.SetSquare(a60);
|
||||
FieldElem a240; a240.SetSquare(a120);
|
||||
FieldElem a255; a255.SetMult(a240,a15);
|
||||
FieldElem a510; a510.SetSquare(a255);
|
||||
FieldElem a750; a750.SetMult(a510,a240);
|
||||
FieldElem a780; a780.SetMult(a750,a30);
|
||||
FieldElem a1020; a1020.SetSquare(a510);
|
||||
FieldElem a1022; a1022.SetMult(a1020,a2);
|
||||
FieldElem a1023; a1023.SetMult(a1022,a);
|
||||
FieldElem x = a15;
|
||||
for (int i=0; i<21; i++) {
|
||||
for (int j=0; j<10; j++) x.SetSquare(x);
|
||||
x.SetMult(x,a1023);
|
||||
}
|
||||
for (int j=0; j<10; j++) x.SetSquare(x);
|
||||
x.SetMult(x,a1022);
|
||||
for (int i=0; i<2; i++) {
|
||||
for (int j=0; j<10; j++) x.SetSquare(x);
|
||||
x.SetMult(x,a1023);
|
||||
}
|
||||
for (int j=0; j<10; j++) x.SetSquare(x);
|
||||
SetMult(x,a780);
|
||||
}
|
||||
|
||||
bool FieldElem::IsOdd() const {
|
||||
#ifdef VERIFY_MAGNITUDE
|
||||
assert(normalized);
|
||||
#endif
|
||||
return n[0] & 1;
|
||||
}
|
||||
|
||||
std::string FieldElem::ToString() {
|
||||
unsigned char tmp[32];
|
||||
Normalize();
|
||||
GetBytes(tmp);
|
||||
std::string ret;
|
||||
for (int i=0; i<32; i++) {
|
||||
static const char *c = "0123456789ABCDEF";
|
||||
ret += c[(tmp[i] >> 4) & 0xF];
|
||||
ret += c[(tmp[i]) & 0xF];
|
||||
}
|
||||
return ret;
|
||||
}
|
||||
|
||||
void FieldElem::SetHex(const std::string &str) {
|
||||
unsigned char tmp[32] = {};
|
||||
static const int cvt[256] = {0, 0, 0, 0, 0, 0, 0,0,0,0,0,0,0,0,0,0,
|
||||
0, 0, 0, 0, 0, 0, 0,0,0,0,0,0,0,0,0,0,
|
||||
0, 0, 0, 0, 0, 0, 0,0,0,0,0,0,0,0,0,0,
|
||||
0, 1, 2, 3, 4, 5, 6,7,8,9,0,0,0,0,0,0,
|
||||
0,10,11,12,13,14,15,0,0,0,0,0,0,0,0,0,
|
||||
0, 0, 0, 0, 0, 0, 0,0,0,0,0,0,0,0,0,0,
|
||||
0,10,11,12,13,14,15,0,0,0,0,0,0,0,0,0,
|
||||
0, 0, 0, 0, 0, 0, 0,0,0,0,0,0,0,0,0,0,
|
||||
0, 0, 0, 0, 0, 0, 0,0,0,0,0,0,0,0,0,0,
|
||||
0, 0, 0, 0, 0, 0, 0,0,0,0,0,0,0,0,0,0,
|
||||
0, 0, 0, 0, 0, 0, 0,0,0,0,0,0,0,0,0,0,
|
||||
0, 0, 0, 0, 0, 0, 0,0,0,0,0,0,0,0,0,0,
|
||||
0, 0, 0, 0, 0, 0, 0,0,0,0,0,0,0,0,0,0,
|
||||
0, 0, 0, 0, 0, 0, 0,0,0,0,0,0,0,0,0,0,
|
||||
0, 0, 0, 0, 0, 0, 0,0,0,0,0,0,0,0,0,0,
|
||||
0, 0, 0, 0, 0, 0, 0,0,0,0,0,0,0,0,0,0};
|
||||
for (unsigned int i=0; i<32; i++) {
|
||||
if (str.length() > i*2)
|
||||
tmp[32 - str.length()/2 + i] = (cvt[(unsigned char)str[2*i]] << 4) + cvt[(unsigned char)str[2*i+1]];
|
||||
}
|
||||
SetBytes(tmp);
|
||||
}
|
||||
|
||||
static const unsigned char field_p_[] = {0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,
|
||||
0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,
|
||||
0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,
|
||||
|
@ -377,54 +22,4 @@ const FieldConstants &GetFieldConst() {
|
|||
return field_const;
|
||||
}
|
||||
|
||||
// Nonbuiltin Field Inverse is not constant time.
|
||||
void FieldElem::SetInverse(FieldElem &a) {
|
||||
#if defined(USE_FIELDINVERSE_BUILTIN)
|
||||
// calculate a^p, with p={45,63,1019,1023}
|
||||
FieldElem a2; a2.SetSquare(a);
|
||||
FieldElem a3; a3.SetMult(a2,a);
|
||||
FieldElem a4; a4.SetSquare(a2);
|
||||
FieldElem a5; a5.SetMult(a4,a);
|
||||
FieldElem a10; a10.SetSquare(a5);
|
||||
FieldElem a11; a11.SetMult(a10,a);
|
||||
FieldElem a21; a21.SetMult(a11,a10);
|
||||
FieldElem a42; a42.SetSquare(a21);
|
||||
FieldElem a45; a45.SetMult(a42,a3);
|
||||
FieldElem a63; a63.SetMult(a42,a21);
|
||||
FieldElem a126; a126.SetSquare(a63);
|
||||
FieldElem a252; a252.SetSquare(a126);
|
||||
FieldElem a504; a504.SetSquare(a252);
|
||||
FieldElem a1008; a1008.SetSquare(a504);
|
||||
FieldElem a1019; a1019.SetMult(a1008,a11);
|
||||
FieldElem a1023; a1023.SetMult(a1019,a4);
|
||||
FieldElem x = a63;
|
||||
for (int i=0; i<21; i++) {
|
||||
for (int j=0; j<10; j++) x.SetSquare(x);
|
||||
x.SetMult(x,a1023);
|
||||
}
|
||||
for (int j=0; j<10; j++) x.SetSquare(x);
|
||||
x.SetMult(x,a1019);
|
||||
for (int i=0; i<2; i++) {
|
||||
for (int j=0; j<10; j++) x.SetSquare(x);
|
||||
x.SetMult(x,a1023);
|
||||
}
|
||||
for (int j=0; j<10; j++) x.SetSquare(x);
|
||||
SetMult(x,a45);
|
||||
#else
|
||||
unsigned char b[32];
|
||||
a.Normalize();
|
||||
a.GetBytes(b);
|
||||
{
|
||||
const secp256k1_num_t &p = GetFieldConst().field_p;
|
||||
secp256k1_num_t n;
|
||||
secp256k1_num_init(&n);
|
||||
secp256k1_num_set_bin(&n, b, 32);
|
||||
secp256k1_num_mod_inverse(&n, &n, &p);
|
||||
secp256k1_num_get_bin(b, 32, &n);
|
||||
secp256k1_num_free(&n);
|
||||
}
|
||||
SetBytes(b);
|
||||
#endif
|
||||
}
|
||||
|
||||
}
|
||||
|
|
73
field.h
73
field.h
|
@ -1,80 +1,11 @@
|
|||
#ifndef _SECP256K1_FIELD_
|
||||
#define _SECP256K1_FIELD_
|
||||
|
||||
using namespace std;
|
||||
|
||||
#include <stdint.h>
|
||||
#include <string>
|
||||
|
||||
#include "num.h"
|
||||
|
||||
// #define VERIFY_MAGNITUDE 1
|
||||
// just one implementation for now
|
||||
#include "field_5x52.h"
|
||||
|
||||
namespace secp256k1 {
|
||||
|
||||
/** Implements arithmetic modulo FFFFFFFF FFFFFFFF FFFFFFFF FFFFFFFF FFFFFFFF FFFFFFFF FFFFFFFE FFFFFC2F,
|
||||
* represented as 5 uint64_t's in base 2^52. he values are allowed to contain >52 each. In particular,
|
||||
* each FieldElem has a 'magnitude' associated with it. Internally, a magnitude M means each element
|
||||
* is at most M*(2^53-1), except the most significant one, which is limited to M*(2^49-1). All operations
|
||||
* accept any input with magnitude at most M, and have different rules for propagating magnitude to their
|
||||
* output.
|
||||
*/
|
||||
class FieldElem {
|
||||
private:
|
||||
// X = sum(i=0..4, elem[i]*2^52) mod n
|
||||
uint64_t n[5];
|
||||
#ifdef VERIFY_MAGNITUDE
|
||||
int magnitude;
|
||||
bool normalized;
|
||||
#endif
|
||||
|
||||
public:
|
||||
|
||||
/** Creates a constant field element. Magnitude=1 */
|
||||
FieldElem(int x = 0);
|
||||
|
||||
FieldElem(const unsigned char *b32);
|
||||
|
||||
/** Normalizes the internal representation entries. Magnitude=1 */
|
||||
void Normalize();
|
||||
|
||||
bool IsZero() const;
|
||||
|
||||
bool friend operator==(const FieldElem &a, const FieldElem &b);
|
||||
|
||||
/** extract as 32-byte big endian array */
|
||||
void GetBytes(unsigned char *o);
|
||||
|
||||
/** set value of 32-byte big endian array */
|
||||
void SetBytes(const unsigned char *in);
|
||||
|
||||
/** Set a FieldElem to be the negative of another. Increases magnitude by one. */
|
||||
void SetNeg(const FieldElem &a, int magnitudeIn);
|
||||
|
||||
/** Multiplies this FieldElem with an integer constant. Magnitude is multiplied by v */
|
||||
void operator*=(int v);
|
||||
|
||||
void operator+=(const FieldElem &a);
|
||||
|
||||
/** Set this FieldElem to be the multiplication of two others. Magnitude=1 (variable time) */
|
||||
void SetMult(const FieldElem &a, const FieldElem &b);
|
||||
|
||||
/** Set this FieldElem to be the square of another. Magnitude=1 (variable time) */
|
||||
void SetSquare(const FieldElem &a);
|
||||
|
||||
/** Set this to be the (modular) square root of another FieldElem. Magnitude=1 */
|
||||
void SetSquareRoot(const FieldElem &a);
|
||||
|
||||
bool IsOdd() const;
|
||||
|
||||
/** Set this to be the (modular) inverse of another FieldElem. Magnitude=1 (variable time) */
|
||||
void SetInverse(FieldElem &a);
|
||||
|
||||
std::string ToString();
|
||||
|
||||
void SetHex(const std::string &str);
|
||||
};
|
||||
|
||||
class FieldConstants {
|
||||
public:
|
||||
secp256k1_num_t field_p;
|
||||
|
|
|
@ -0,0 +1,412 @@
|
|||
#include <assert.h>
|
||||
#include <stdint.h>
|
||||
#include <string>
|
||||
#include "num.h"
|
||||
#include "field.h"
|
||||
#include <iostream>
|
||||
|
||||
using namespace std;
|
||||
|
||||
#ifdef INLINE_ASM
|
||||
#include "lin64.h"
|
||||
#endif
|
||||
|
||||
namespace secp256k1 {
|
||||
|
||||
/** Implements arithmetic modulo FFFFFFFF FFFFFFFF FFFFFFFF FFFFFFFF FFFFFFFF FFFFFFFF FFFFFFFE FFFFFC2F,
|
||||
* represented as 5 uint64_t's in base 2^52. The values are allowed to contain >52 each. In particular,
|
||||
* each FieldElem has a 'magnitude' associated with it. Internally, a magnitude M means each element
|
||||
* is at most M*(2^53-1), except the most significant one, which is limited to M*(2^49-1). All operations
|
||||
* accept any input with magnitude at most M, and have different rules for propagating magnitude to their
|
||||
* output.
|
||||
*/
|
||||
|
||||
FieldElem::FieldElem(int x) {
|
||||
n[0] = x;
|
||||
n[1] = n[2] = n[3] = n[4] = 0;
|
||||
#ifdef VERIFY_MAGNITUDE
|
||||
magnitude = 1;
|
||||
normalized = true;
|
||||
#endif
|
||||
}
|
||||
|
||||
FieldElem::FieldElem(const unsigned char *b32) {
|
||||
SetBytes(b32);
|
||||
}
|
||||
|
||||
void FieldElem::Normalize() {
|
||||
uint64_t c;
|
||||
c = n[0];
|
||||
uint64_t t0 = c & 0xFFFFFFFFFFFFFULL;
|
||||
c = (c >> 52) + n[1];
|
||||
uint64_t t1 = c & 0xFFFFFFFFFFFFFULL;
|
||||
c = (c >> 52) + n[2];
|
||||
uint64_t t2 = c & 0xFFFFFFFFFFFFFULL;
|
||||
c = (c >> 52) + n[3];
|
||||
uint64_t t3 = c & 0xFFFFFFFFFFFFFULL;
|
||||
c = (c >> 52) + n[4];
|
||||
uint64_t t4 = c & 0x0FFFFFFFFFFFFULL;
|
||||
c >>= 48;
|
||||
|
||||
// The following code will not modify the t's if c is initially 0.
|
||||
c = c * 0x1000003D1ULL + t0;
|
||||
t0 = c & 0xFFFFFFFFFFFFFULL;
|
||||
c = (c >> 52) + t1;
|
||||
t1 = c & 0xFFFFFFFFFFFFFULL;
|
||||
c = (c >> 52) + t2;
|
||||
t2 = c & 0xFFFFFFFFFFFFFULL;
|
||||
c = (c >> 52) + t3;
|
||||
t3 = c & 0xFFFFFFFFFFFFFULL;
|
||||
c = (c >> 52) + t4;
|
||||
t4 = c & 0x0FFFFFFFFFFFFULL;
|
||||
|
||||
// Replace n's with t's if one of the n's overflows.
|
||||
// If none of the n's overflow to begin with, the t's will just be the n's already and
|
||||
// we effectively ignore the results of the previous computations.
|
||||
n[0] = t0; n[1] = t1; n[2] = t2; n[3] = t3; n[4] = t4;
|
||||
|
||||
// Subtract p if result >= p
|
||||
uint64_t mask = -(int64_t)((n[4] < 0xFFFFFFFFFFFFULL) | (n[3] < 0xFFFFFFFFFFFFFULL) | (n[2] < 0xFFFFFFFFFFFFFULL) | (n[1] < 0xFFFFFFFFFFFFFULL) | (n[0] < 0xFFFFEFFFFFC2FULL));
|
||||
n[4] &= mask;
|
||||
n[3] &= mask;
|
||||
n[2] &= mask;
|
||||
n[1] &= mask;
|
||||
n[0] -= (~mask & 0xFFFFEFFFFFC2FULL);
|
||||
|
||||
#ifdef VERIFY_MAGNITUDE
|
||||
magnitude = 1;
|
||||
normalized = true;
|
||||
#endif
|
||||
}
|
||||
|
||||
bool inline FieldElem::IsZero() const {
|
||||
#ifdef VERIFY_MAGNITUDE
|
||||
assert(normalized);
|
||||
#endif
|
||||
return (n[0] == 0 && n[1] == 0 && n[2] == 0 && n[3] == 0 && n[4] == 0);
|
||||
}
|
||||
|
||||
bool inline operator==(const FieldElem &a, const FieldElem &b) {
|
||||
#ifdef VERIFY_MAGNITUDE
|
||||
assert(a.normalized);
|
||||
assert(b.normalized);
|
||||
#endif
|
||||
return (a.n[0] == b.n[0] && a.n[1] == b.n[1] && a.n[2] == b.n[2] && a.n[3] == b.n[3] && a.n[4] == b.n[4]);
|
||||
}
|
||||
|
||||
void FieldElem::GetBytes(unsigned char *o) {
|
||||
#ifdef VERIFY_MAGNITUDE
|
||||
assert(normalized);
|
||||
#endif
|
||||
for (int i=0; i<32; i++) {
|
||||
int c = 0;
|
||||
for (int j=0; j<2; j++) {
|
||||
int limb = (8*i+4*j)/52;
|
||||
int shift = (8*i+4*j)%52;
|
||||
c |= ((n[limb] >> shift) & 0xF) << (4 * j);
|
||||
}
|
||||
o[31-i] = c;
|
||||
}
|
||||
}
|
||||
|
||||
void FieldElem::SetBytes(const unsigned char *in) {
|
||||
n[0] = n[1] = n[2] = n[3] = n[4] = 0;
|
||||
for (int i=0; i<32; i++) {
|
||||
for (int j=0; j<2; j++) {
|
||||
int limb = (8*i+4*j)/52;
|
||||
int shift = (8*i+4*j)%52;
|
||||
n[limb] |= (uint64_t)((in[31-i] >> (4*j)) & 0xF) << shift;
|
||||
}
|
||||
}
|
||||
#ifdef VERIFY_MAGNITUDE
|
||||
magnitude = 1;
|
||||
normalized = true;
|
||||
#endif
|
||||
}
|
||||
|
||||
void inline FieldElem::SetNeg(const FieldElem &a, int magnitudeIn) {
|
||||
#ifdef VERIFY_MAGNITUDE
|
||||
assert(a.magnitude <= magnitudeIn);
|
||||
magnitude = magnitudeIn + 1;
|
||||
normalized = false;
|
||||
#endif
|
||||
n[0] = 0xFFFFEFFFFFC2FULL * (magnitudeIn + 1) - a.n[0];
|
||||
n[1] = 0xFFFFFFFFFFFFFULL * (magnitudeIn + 1) - a.n[1];
|
||||
n[2] = 0xFFFFFFFFFFFFFULL * (magnitudeIn + 1) - a.n[2];
|
||||
n[3] = 0xFFFFFFFFFFFFFULL * (magnitudeIn + 1) - a.n[3];
|
||||
n[4] = 0x0FFFFFFFFFFFFULL * (magnitudeIn + 1) - a.n[4];
|
||||
}
|
||||
|
||||
void inline FieldElem::operator*=(int v) {
|
||||
#ifdef VERIFY_MAGNITUDE
|
||||
magnitude *= v;
|
||||
normalized = false;
|
||||
#endif
|
||||
n[0] *= v;
|
||||
n[1] *= v;
|
||||
n[2] *= v;
|
||||
n[3] *= v;
|
||||
n[4] *= v;
|
||||
}
|
||||
|
||||
void inline FieldElem::operator+=(const FieldElem &a) {
|
||||
#ifdef VERIFY_MAGNITUDE
|
||||
magnitude += a.magnitude;
|
||||
normalized = false;
|
||||
#endif
|
||||
n[0] += a.n[0];
|
||||
n[1] += a.n[1];
|
||||
n[2] += a.n[2];
|
||||
n[3] += a.n[3];
|
||||
n[4] += a.n[4];
|
||||
}
|
||||
|
||||
void FieldElem::SetMult(const FieldElem &a, const FieldElem &b) {
|
||||
#ifdef VERIFY_MAGNITUDE
|
||||
assert(a.magnitude <= 8);
|
||||
assert(b.magnitude <= 8);
|
||||
#endif
|
||||
|
||||
#ifdef INLINE_ASM
|
||||
ExSetMult((uint64_t *) a.n,(uint64_t *) b.n, (uint64_t *) n);
|
||||
#else
|
||||
unsigned __int128 c = (__int128)a.n[0] * b.n[0];
|
||||
uint64_t t0 = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 0FFFFFFFFFFFFFE0
|
||||
c = c + (__int128)a.n[0] * b.n[1] +
|
||||
(__int128)a.n[1] * b.n[0];
|
||||
uint64_t t1 = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 20000000000000BF
|
||||
c = c + (__int128)a.n[0] * b.n[2] +
|
||||
(__int128)a.n[1] * b.n[1] +
|
||||
(__int128)a.n[2] * b.n[0];
|
||||
uint64_t t2 = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 30000000000001A0
|
||||
c = c + (__int128)a.n[0] * b.n[3] +
|
||||
(__int128)a.n[1] * b.n[2] +
|
||||
(__int128)a.n[2] * b.n[1] +
|
||||
(__int128)a.n[3] * b.n[0];
|
||||
uint64_t t3 = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 4000000000000280
|
||||
c = c + (__int128)a.n[0] * b.n[4] +
|
||||
(__int128)a.n[1] * b.n[3] +
|
||||
(__int128)a.n[2] * b.n[2] +
|
||||
(__int128)a.n[3] * b.n[1] +
|
||||
(__int128)a.n[4] * b.n[0];
|
||||
uint64_t t4 = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 320000000000037E
|
||||
c = c + (__int128)a.n[1] * b.n[4] +
|
||||
(__int128)a.n[2] * b.n[3] +
|
||||
(__int128)a.n[3] * b.n[2] +
|
||||
(__int128)a.n[4] * b.n[1];
|
||||
uint64_t t5 = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 22000000000002BE
|
||||
c = c + (__int128)a.n[2] * b.n[4] +
|
||||
(__int128)a.n[3] * b.n[3] +
|
||||
(__int128)a.n[4] * b.n[2];
|
||||
uint64_t t6 = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 12000000000001DE
|
||||
c = c + (__int128)a.n[3] * b.n[4] +
|
||||
(__int128)a.n[4] * b.n[3];
|
||||
uint64_t t7 = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 02000000000000FE
|
||||
c = c + (__int128)a.n[4] * b.n[4];
|
||||
uint64_t t8 = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 001000000000001E
|
||||
uint64_t t9 = c;
|
||||
|
||||
c = t0 + (__int128)t5 * 0x1000003D10ULL;
|
||||
t0 = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 0000001000003D10
|
||||
c = c + t1 + (__int128)t6 * 0x1000003D10ULL;
|
||||
t1 = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 0000001000003D10
|
||||
c = c + t2 + (__int128)t7 * 0x1000003D10ULL;
|
||||
n[2] = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 0000001000003D10
|
||||
c = c + t3 + (__int128)t8 * 0x1000003D10ULL;
|
||||
n[3] = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 0000001000003D10
|
||||
c = c + t4 + (__int128)t9 * 0x1000003D10ULL;
|
||||
n[4] = c & 0x0FFFFFFFFFFFFULL; c = c >> 48; // c max 000001000003D110
|
||||
c = t0 + (__int128)c * 0x1000003D1ULL;
|
||||
n[0] = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 1000008
|
||||
n[1] = t1 + c;
|
||||
#endif
|
||||
#ifdef VERIFY_MAGNITUDE
|
||||
magnitude = 1;
|
||||
normalized = false;
|
||||
#endif
|
||||
}
|
||||
|
||||
void FieldElem::SetSquare(const FieldElem &a) {
|
||||
#ifdef VERIFY_MAGNITUDE
|
||||
assert(a.magnitude <= 8);
|
||||
#endif
|
||||
|
||||
#ifdef INLINE_ASM
|
||||
ExSetSquare((uint64_t *)a.n,(uint64_t *)n);
|
||||
#else
|
||||
__int128 c = (__int128)a.n[0] * a.n[0];
|
||||
uint64_t t0 = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 0FFFFFFFFFFFFFE0
|
||||
c = c + (__int128)(a.n[0]*2) * a.n[1];
|
||||
uint64_t t1 = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 20000000000000BF
|
||||
c = c + (__int128)(a.n[0]*2) * a.n[2] +
|
||||
(__int128)a.n[1] * a.n[1];
|
||||
uint64_t t2 = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 30000000000001A0
|
||||
c = c + (__int128)(a.n[0]*2) * a.n[3] +
|
||||
(__int128)(a.n[1]*2) * a.n[2];
|
||||
uint64_t t3 = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 4000000000000280
|
||||
c = c + (__int128)(a.n[0]*2) * a.n[4] +
|
||||
(__int128)(a.n[1]*2) * a.n[3] +
|
||||
(__int128)a.n[2] * a.n[2];
|
||||
uint64_t t4 = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 320000000000037E
|
||||
c = c + (__int128)(a.n[1]*2) * a.n[4] +
|
||||
(__int128)(a.n[2]*2) * a.n[3];
|
||||
uint64_t t5 = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 22000000000002BE
|
||||
c = c + (__int128)(a.n[2]*2) * a.n[4] +
|
||||
(__int128)a.n[3] * a.n[3];
|
||||
uint64_t t6 = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 12000000000001DE
|
||||
c = c + (__int128)(a.n[3]*2) * a.n[4];
|
||||
uint64_t t7 = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 02000000000000FE
|
||||
c = c + (__int128)a.n[4] * a.n[4];
|
||||
uint64_t t8 = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 001000000000001E
|
||||
uint64_t t9 = c;
|
||||
c = t0 + (__int128)t5 * 0x1000003D10ULL;
|
||||
t0 = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 0000001000003D10
|
||||
c = c + t1 + (__int128)t6 * 0x1000003D10ULL;
|
||||
t1 = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 0000001000003D10
|
||||
c = c + t2 + (__int128)t7 * 0x1000003D10ULL;
|
||||
n[2] = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 0000001000003D10
|
||||
c = c + t3 + (__int128)t8 * 0x1000003D10ULL;
|
||||
n[3] = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 0000001000003D10
|
||||
c = c + t4 + (__int128)t9 * 0x1000003D10ULL;
|
||||
n[4] = c & 0x0FFFFFFFFFFFFULL; c = c >> 48; // c max 000001000003D110
|
||||
c = t0 + (__int128)c * 0x1000003D1ULL;
|
||||
n[0] = c & 0xFFFFFFFFFFFFFULL; c = c >> 52; // c max 1000008
|
||||
n[1] = t1 + c;
|
||||
#endif
|
||||
|
||||
#ifdef VERIFY_MAGNITUDE
|
||||
assert(a.magnitude <= 8);
|
||||
normalized = false;
|
||||
#endif
|
||||
}
|
||||
|
||||
void FieldElem::SetSquareRoot(const FieldElem &a) {
|
||||
// calculate a^p, with p={15,780,1022,1023}
|
||||
FieldElem a2; a2.SetSquare(a);
|
||||
FieldElem a3; a3.SetMult(a2,a);
|
||||
FieldElem a6; a6.SetSquare(a3);
|
||||
FieldElem a12; a12.SetSquare(a6);
|
||||
FieldElem a15; a15.SetMult(a12,a3);
|
||||
FieldElem a30; a30.SetSquare(a15);
|
||||
FieldElem a60; a60.SetSquare(a30);
|
||||
FieldElem a120; a120.SetSquare(a60);
|
||||
FieldElem a240; a240.SetSquare(a120);
|
||||
FieldElem a255; a255.SetMult(a240,a15);
|
||||
FieldElem a510; a510.SetSquare(a255);
|
||||
FieldElem a750; a750.SetMult(a510,a240);
|
||||
FieldElem a780; a780.SetMult(a750,a30);
|
||||
FieldElem a1020; a1020.SetSquare(a510);
|
||||
FieldElem a1022; a1022.SetMult(a1020,a2);
|
||||
FieldElem a1023; a1023.SetMult(a1022,a);
|
||||
FieldElem x = a15;
|
||||
for (int i=0; i<21; i++) {
|
||||
for (int j=0; j<10; j++) x.SetSquare(x);
|
||||
x.SetMult(x,a1023);
|
||||
}
|
||||
for (int j=0; j<10; j++) x.SetSquare(x);
|
||||
x.SetMult(x,a1022);
|
||||
for (int i=0; i<2; i++) {
|
||||
for (int j=0; j<10; j++) x.SetSquare(x);
|
||||
x.SetMult(x,a1023);
|
||||
}
|
||||
for (int j=0; j<10; j++) x.SetSquare(x);
|
||||
SetMult(x,a780);
|
||||
}
|
||||
|
||||
bool FieldElem::IsOdd() const {
|
||||
#ifdef VERIFY_MAGNITUDE
|
||||
assert(normalized);
|
||||
#endif
|
||||
return n[0] & 1;
|
||||
}
|
||||
|
||||
std::string FieldElem::ToString() {
|
||||
unsigned char tmp[32];
|
||||
Normalize();
|
||||
GetBytes(tmp);
|
||||
std::string ret;
|
||||
for (int i=0; i<32; i++) {
|
||||
static const char *c = "0123456789ABCDEF";
|
||||
ret += c[(tmp[i] >> 4) & 0xF];
|
||||
ret += c[(tmp[i]) & 0xF];
|
||||
}
|
||||
return ret;
|
||||
}
|
||||
|
||||
void FieldElem::SetHex(const std::string &str) {
|
||||
unsigned char tmp[32] = {};
|
||||
static const int cvt[256] = {0, 0, 0, 0, 0, 0, 0,0,0,0,0,0,0,0,0,0,
|
||||
0, 0, 0, 0, 0, 0, 0,0,0,0,0,0,0,0,0,0,
|
||||
0, 0, 0, 0, 0, 0, 0,0,0,0,0,0,0,0,0,0,
|
||||
0, 1, 2, 3, 4, 5, 6,7,8,9,0,0,0,0,0,0,
|
||||
0,10,11,12,13,14,15,0,0,0,0,0,0,0,0,0,
|
||||
0, 0, 0, 0, 0, 0, 0,0,0,0,0,0,0,0,0,0,
|
||||
0,10,11,12,13,14,15,0,0,0,0,0,0,0,0,0,
|
||||
0, 0, 0, 0, 0, 0, 0,0,0,0,0,0,0,0,0,0,
|
||||
0, 0, 0, 0, 0, 0, 0,0,0,0,0,0,0,0,0,0,
|
||||
0, 0, 0, 0, 0, 0, 0,0,0,0,0,0,0,0,0,0,
|
||||
0, 0, 0, 0, 0, 0, 0,0,0,0,0,0,0,0,0,0,
|
||||
0, 0, 0, 0, 0, 0, 0,0,0,0,0,0,0,0,0,0,
|
||||
0, 0, 0, 0, 0, 0, 0,0,0,0,0,0,0,0,0,0,
|
||||
0, 0, 0, 0, 0, 0, 0,0,0,0,0,0,0,0,0,0,
|
||||
0, 0, 0, 0, 0, 0, 0,0,0,0,0,0,0,0,0,0,
|
||||
0, 0, 0, 0, 0, 0, 0,0,0,0,0,0,0,0,0,0};
|
||||
for (unsigned int i=0; i<32; i++) {
|
||||
if (str.length() > i*2)
|
||||
tmp[32 - str.length()/2 + i] = (cvt[(unsigned char)str[2*i]] << 4) + cvt[(unsigned char)str[2*i+1]];
|
||||
}
|
||||
SetBytes(tmp);
|
||||
}
|
||||
|
||||
|
||||
// Nonbuiltin Field Inverse is not constant time.
|
||||
void FieldElem::SetInverse(FieldElem &a) {
|
||||
#if defined(USE_FIELDINVERSE_BUILTIN)
|
||||
// calculate a^p, with p={45,63,1019,1023}
|
||||
FieldElem a2; a2.SetSquare(a);
|
||||
FieldElem a3; a3.SetMult(a2,a);
|
||||
FieldElem a4; a4.SetSquare(a2);
|
||||
FieldElem a5; a5.SetMult(a4,a);
|
||||
FieldElem a10; a10.SetSquare(a5);
|
||||
FieldElem a11; a11.SetMult(a10,a);
|
||||
FieldElem a21; a21.SetMult(a11,a10);
|
||||
FieldElem a42; a42.SetSquare(a21);
|
||||
FieldElem a45; a45.SetMult(a42,a3);
|
||||
FieldElem a63; a63.SetMult(a42,a21);
|
||||
FieldElem a126; a126.SetSquare(a63);
|
||||
FieldElem a252; a252.SetSquare(a126);
|
||||
FieldElem a504; a504.SetSquare(a252);
|
||||
FieldElem a1008; a1008.SetSquare(a504);
|
||||
FieldElem a1019; a1019.SetMult(a1008,a11);
|
||||
FieldElem a1023; a1023.SetMult(a1019,a4);
|
||||
FieldElem x = a63;
|
||||
for (int i=0; i<21; i++) {
|
||||
for (int j=0; j<10; j++) x.SetSquare(x);
|
||||
x.SetMult(x,a1023);
|
||||
}
|
||||
for (int j=0; j<10; j++) x.SetSquare(x);
|
||||
x.SetMult(x,a1019);
|
||||
for (int i=0; i<2; i++) {
|
||||
for (int j=0; j<10; j++) x.SetSquare(x);
|
||||
x.SetMult(x,a1023);
|
||||
}
|
||||
for (int j=0; j<10; j++) x.SetSquare(x);
|
||||
SetMult(x,a45);
|
||||
#else
|
||||
unsigned char b[32];
|
||||
a.Normalize();
|
||||
a.GetBytes(b);
|
||||
{
|
||||
const secp256k1_num_t &p = GetFieldConst().field_p;
|
||||
secp256k1_num_t n;
|
||||
secp256k1_num_init(&n);
|
||||
secp256k1_num_set_bin(&n, b, 32);
|
||||
secp256k1_num_mod_inverse(&n, &n, &p);
|
||||
secp256k1_num_get_bin(b, 32, &n);
|
||||
secp256k1_num_free(&n);
|
||||
}
|
||||
SetBytes(b);
|
||||
#endif
|
||||
}
|
||||
|
||||
}
|
|
@ -0,0 +1,80 @@
|
|||
#ifndef _SECP256K1_FIELD_5x52_
|
||||
#define _SECP256K1_FIELD_5x52_
|
||||
|
||||
using namespace std;
|
||||
|
||||
#include <stdint.h>
|
||||
#include <string>
|
||||
|
||||
#include "num.h"
|
||||
|
||||
// #define VERIFY_MAGNITUDE 1
|
||||
|
||||
namespace secp256k1 {
|
||||
|
||||
/** Implements arithmetic modulo FFFFFFFF FFFFFFFF FFFFFFFF FFFFFFFF FFFFFFFF FFFFFFFF FFFFFFFE FFFFFC2F,
|
||||
* represented as 5 uint64_t's in base 2^52. he values are allowed to contain >52 each. In particular,
|
||||
* each FieldElem has a 'magnitude' associated with it. Internally, a magnitude M means each element
|
||||
* is at most M*(2^53-1), except the most significant one, which is limited to M*(2^49-1). All operations
|
||||
* accept any input with magnitude at most M, and have different rules for propagating magnitude to their
|
||||
* output.
|
||||
*/
|
||||
class FieldElem {
|
||||
private:
|
||||
// X = sum(i=0..4, elem[i]*2^52) mod n
|
||||
uint64_t n[5];
|
||||
#ifdef VERIFY_MAGNITUDE
|
||||
int magnitude;
|
||||
bool normalized;
|
||||
#endif
|
||||
|
||||
public:
|
||||
|
||||
/** Creates a constant field element. Magnitude=1 */
|
||||
FieldElem(int x = 0);
|
||||
|
||||
FieldElem(const unsigned char *b32);
|
||||
|
||||
/** Normalizes the internal representation entries. Magnitude=1 */
|
||||
void Normalize();
|
||||
|
||||
bool IsZero() const;
|
||||
|
||||
bool friend operator==(const FieldElem &a, const FieldElem &b);
|
||||
|
||||
/** extract as 32-byte big endian array */
|
||||
void GetBytes(unsigned char *o);
|
||||
|
||||
/** set value of 32-byte big endian array */
|
||||
void SetBytes(const unsigned char *in);
|
||||
|
||||
/** Set a FieldElem to be the negative of another. Increases magnitude by one. */
|
||||
void SetNeg(const FieldElem &a, int magnitudeIn);
|
||||
|
||||
/** Multiplies this FieldElem with an integer constant. Magnitude is multiplied by v */
|
||||
void operator*=(int v);
|
||||
|
||||
void operator+=(const FieldElem &a);
|
||||
|
||||
/** Set this FieldElem to be the multiplication of two others. Magnitude=1 (variable time) */
|
||||
void SetMult(const FieldElem &a, const FieldElem &b);
|
||||
|
||||
/** Set this FieldElem to be the square of another. Magnitude=1 (variable time) */
|
||||
void SetSquare(const FieldElem &a);
|
||||
|
||||
/** Set this to be the (modular) square root of another FieldElem. Magnitude=1 */
|
||||
void SetSquareRoot(const FieldElem &a);
|
||||
|
||||
bool IsOdd() const;
|
||||
|
||||
/** Set this to be the (modular) inverse of another FieldElem. Magnitude=1 (variable time) */
|
||||
void SetInverse(FieldElem &a);
|
||||
|
||||
std::string ToString();
|
||||
|
||||
void SetHex(const std::string &str);
|
||||
};
|
||||
|
||||
}
|
||||
|
||||
#endif
|
Loading…
Reference in New Issue