Official Go implementation of the Ethereum protocol
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
go-ethereum/crypto/bls12381/arithmetic_x86.s

2151 lines
30 KiB

core/vm, crypto/bls12381, params: add bls12-381 elliptic curve precompiles (#21018) * crypto: add bls12-381 elliptic curve wrapper * params: add bls12-381 precompile gas parameters * core/vm: add bls12-381 precompiles * core/vm: add bls12-381 precompile tests * go.mod, go.sum: use latest bls12381 lib * core/vm: move point encode/decode functions to base library * crypto/bls12381: introduce bls12-381 library init function * crypto/bls12381: import bls12381 elliptic curve implementation * go.mod, go.sum: remove bls12-381 library * remove unsued frobenious coeffs supress warning for inp that used in asm * add mappings tests for zero inputs fix swu g2 minus z inverse constant * crypto/bls12381: fix typo * crypto/bls12381: better comments for bls12381 constants * crypto/bls12381: swu, use single conditional for e2 * crypto/bls12381: utils, delete empty line * crypto/bls12381: utils, use FromHex for string to big * crypto/bls12381: g1, g2, strict length check for FromBytes * crypto/bls12381: field_element, comparision changes * crypto/bls12381: change swu, isogeny constants with hex values * core/vm: fix point multiplication comments * core/vm: fix multiexp gas calculation and lookup for g1 and g2 * core/vm: simpler imput length check for multiexp and pairing precompiles * core/vm: rm empty multiexp result declarations * crypto/bls12381: remove modulus type definition * crypto/bls12381: use proper init function * crypto/bls12381: get rid of new lines at fatal desciprtions * crypto/bls12-381: fix no-adx assembly multiplication * crypto/bls12-381: remove old config function * crypto/bls12381: update multiplication backend this commit changes mul backend to 6limb eip1962 backend mul assign operations are dropped * core/vm/contracts_tests: externalize test vectors for precompiles * core/vm/contracts_test: externalize failure-cases for precompiles * core/vm: linting * go.mod: tiny up sum file * core/vm: fix goimports linter issues * crypto/bls12381: build tags for plain ASM or ADX implementation Co-authored-by: Martin Holst Swende <martin@swende.se> Co-authored-by: Péter Szilágyi <peterke@gmail.com>
4 years ago
// +build amd64,blsasm amd64,blsadx
#include "textflag.h"
// addition w/ modular reduction
// a = (a + b) % p
TEXT ·addAssign(SB), NOSPLIT, $0-16
// |
MOVQ a+0(FP), DI
MOVQ b+8(FP), SI
// |
MOVQ (DI), R8
MOVQ 8(DI), R9
MOVQ 16(DI), R10
MOVQ 24(DI), R11
MOVQ 32(DI), R12
MOVQ 40(DI), R13
// |
ADDQ (SI), R8
ADCQ 8(SI), R9
ADCQ 16(SI), R10
ADCQ 24(SI), R11
ADCQ 32(SI), R12
ADCQ 40(SI), R13
// |
MOVQ R8, R14
MOVQ R9, R15
MOVQ R10, CX
MOVQ R11, DX
MOVQ R12, SI
MOVQ R13, BX
MOVQ $0xb9feffffffffaaab, AX
SUBQ AX, R14
MOVQ $0x1eabfffeb153ffff, AX
SBBQ AX, R15
MOVQ $0x6730d2a0f6b0f624, AX
SBBQ AX, CX
MOVQ $0x64774b84f38512bf, AX
SBBQ AX, DX
MOVQ $0x4b1ba7b6434bacd7, AX
SBBQ AX, SI
MOVQ $0x1a0111ea397fe69a, AX
SBBQ AX, BX
CMOVQCC R14, R8
CMOVQCC R15, R9
CMOVQCC CX, R10
CMOVQCC DX, R11
CMOVQCC SI, R12
CMOVQCC BX, R13
// |
MOVQ R8, (DI)
MOVQ R9, 8(DI)
MOVQ R10, 16(DI)
MOVQ R11, 24(DI)
MOVQ R12, 32(DI)
MOVQ R13, 40(DI)
RET
/* | end */
// addition w/ modular reduction
// c = (a + b) % p
TEXT ·add(SB), NOSPLIT, $0-24
// |
MOVQ a+8(FP), DI
MOVQ b+16(FP), SI
// |
MOVQ (DI), R8
MOVQ 8(DI), R9
MOVQ 16(DI), R10
MOVQ 24(DI), R11
MOVQ 32(DI), R12
MOVQ 40(DI), R13
// |
ADDQ (SI), R8
ADCQ 8(SI), R9
ADCQ 16(SI), R10
ADCQ 24(SI), R11
ADCQ 32(SI), R12
ADCQ 40(SI), R13
// |
MOVQ R8, R14
MOVQ R9, R15
MOVQ R10, CX
MOVQ R11, DX
MOVQ R12, SI
MOVQ R13, BX
MOVQ $0xb9feffffffffaaab, DI
SUBQ DI, R14
MOVQ $0x1eabfffeb153ffff, DI
SBBQ DI, R15
MOVQ $0x6730d2a0f6b0f624, DI
SBBQ DI, CX
MOVQ $0x64774b84f38512bf, DI
SBBQ DI, DX
MOVQ $0x4b1ba7b6434bacd7, DI
SBBQ DI, SI
MOVQ $0x1a0111ea397fe69a, DI
SBBQ DI, BX
CMOVQCC R14, R8
CMOVQCC R15, R9
CMOVQCC CX, R10
CMOVQCC DX, R11
CMOVQCC SI, R12
CMOVQCC BX, R13
// |
MOVQ c+0(FP), DI
MOVQ R8, (DI)
MOVQ R9, 8(DI)
MOVQ R10, 16(DI)
MOVQ R11, 24(DI)
MOVQ R12, 32(DI)
MOVQ R13, 40(DI)
RET
/* | end */
// addition w/o reduction check
// c = (a + b)
TEXT ·ladd(SB), NOSPLIT, $0-24
// |
MOVQ a+8(FP), DI
MOVQ b+16(FP), SI
// |
MOVQ (DI), R8
MOVQ 8(DI), R9
MOVQ 16(DI), R10
MOVQ 24(DI), R11
MOVQ 32(DI), R12
MOVQ 40(DI), R13
// |
ADDQ (SI), R8
ADCQ 8(SI), R9
ADCQ 16(SI), R10
ADCQ 24(SI), R11
ADCQ 32(SI), R12
ADCQ 40(SI), R13
// |
MOVQ c+0(FP), DI
MOVQ R8, (DI)
MOVQ R9, 8(DI)
MOVQ R10, 16(DI)
MOVQ R11, 24(DI)
MOVQ R12, 32(DI)
MOVQ R13, 40(DI)
RET
/* | end */
// addition w/o reduction check
// a = a + b
TEXT ·laddAssign(SB), NOSPLIT, $0-16
// |
MOVQ a+0(FP), DI
MOVQ b+8(FP), SI
// |
MOVQ (DI), R8
MOVQ 8(DI), R9
MOVQ 16(DI), R10
MOVQ 24(DI), R11
MOVQ 32(DI), R12
MOVQ 40(DI), R13
// |
ADDQ (SI), R8
ADCQ 8(SI), R9
ADCQ 16(SI), R10
ADCQ 24(SI), R11
ADCQ 32(SI), R12
ADCQ 40(SI), R13
// |
MOVQ a+0(FP), DI
MOVQ R8, (DI)
MOVQ R9, 8(DI)
MOVQ R10, 16(DI)
MOVQ R11, 24(DI)
MOVQ R12, 32(DI)
MOVQ R13, 40(DI)
RET
/* | end */
// subtraction w/ modular reduction
// c = (a - b) % p
TEXT ·sub(SB), NOSPLIT, $0-24
// |
MOVQ a+8(FP), DI
MOVQ b+16(FP), SI
XORQ AX, AX
// |
MOVQ (DI), R8
MOVQ 8(DI), R9
MOVQ 16(DI), R10
MOVQ 24(DI), R11
MOVQ 32(DI), R12
MOVQ 40(DI), R13
SUBQ (SI), R8
SBBQ 8(SI), R9
SBBQ 16(SI), R10
SBBQ 24(SI), R11
SBBQ 32(SI), R12
SBBQ 40(SI), R13
// |
MOVQ $0xb9feffffffffaaab, R14
MOVQ $0x1eabfffeb153ffff, R15
MOVQ $0x6730d2a0f6b0f624, CX
MOVQ $0x64774b84f38512bf, DX
MOVQ $0x4b1ba7b6434bacd7, SI
MOVQ $0x1a0111ea397fe69a, BX
CMOVQCC AX, R14
CMOVQCC AX, R15
CMOVQCC AX, CX
CMOVQCC AX, DX
CMOVQCC AX, SI
CMOVQCC AX, BX
ADDQ R14, R8
ADCQ R15, R9
ADCQ CX, R10
ADCQ DX, R11
ADCQ SI, R12
ADCQ BX, R13
// |
MOVQ c+0(FP), DI
MOVQ R8, (DI)
MOVQ R9, 8(DI)
MOVQ R10, 16(DI)
MOVQ R11, 24(DI)
MOVQ R12, 32(DI)
MOVQ R13, 40(DI)
RET
/* | end */
// subtraction w/ modular reduction
// a = (a - b) % p
TEXT ·subAssign(SB), NOSPLIT, $0-16
// |
MOVQ a+0(FP), DI
MOVQ b+8(FP), SI
XORQ AX, AX
// |
MOVQ (DI), R8
MOVQ 8(DI), R9
MOVQ 16(DI), R10
MOVQ 24(DI), R11
MOVQ 32(DI), R12
MOVQ 40(DI), R13
SUBQ (SI), R8
SBBQ 8(SI), R9
SBBQ 16(SI), R10
SBBQ 24(SI), R11
SBBQ 32(SI), R12
SBBQ 40(SI), R13
// |
MOVQ $0xb9feffffffffaaab, R14
MOVQ $0x1eabfffeb153ffff, R15
MOVQ $0x6730d2a0f6b0f624, CX
MOVQ $0x64774b84f38512bf, DX
MOVQ $0x4b1ba7b6434bacd7, SI
MOVQ $0x1a0111ea397fe69a, BX
CMOVQCC AX, R14
CMOVQCC AX, R15
CMOVQCC AX, CX
CMOVQCC AX, DX
CMOVQCC AX, SI
CMOVQCC AX, BX
ADDQ R14, R8
ADCQ R15, R9
ADCQ CX, R10
ADCQ DX, R11
ADCQ SI, R12
ADCQ BX, R13
// |
MOVQ a+0(FP), DI
MOVQ R8, (DI)
MOVQ R9, 8(DI)
MOVQ R10, 16(DI)
MOVQ R11, 24(DI)
MOVQ R12, 32(DI)
MOVQ R13, 40(DI)
RET
/* | end */
// subtraction w/o reduction check
// a = (a - b)
TEXT ·lsubAssign(SB), NOSPLIT, $0-16
// |
MOVQ a+0(FP), DI
MOVQ b+8(FP), SI
// |
MOVQ (DI), R8
MOVQ 8(DI), R9
MOVQ 16(DI), R10
MOVQ 24(DI), R11
MOVQ 32(DI), R12
MOVQ 40(DI), R13
SUBQ (SI), R8
SBBQ 8(SI), R9
SBBQ 16(SI), R10
SBBQ 24(SI), R11
SBBQ 32(SI), R12
SBBQ 40(SI), R13
// |
MOVQ a+0(FP), DI
MOVQ R8, (DI)
MOVQ R9, 8(DI)
MOVQ R10, 16(DI)
MOVQ R11, 24(DI)
MOVQ R12, 32(DI)
MOVQ R13, 40(DI)
RET
/* | end */
// doubling w/ reduction
// c = (2 * a) % p
TEXT ·double(SB), NOSPLIT, $0-16
// |
MOVQ a+8(FP), DI
MOVQ (DI), R8
MOVQ 8(DI), R9
MOVQ 16(DI), R10
MOVQ 24(DI), R11
MOVQ 32(DI), R12
MOVQ 40(DI), R13
ADDQ R8, R8
ADCQ R9, R9
ADCQ R10, R10
ADCQ R11, R11
ADCQ R12, R12
ADCQ R13, R13
// |
MOVQ R8, R14
MOVQ R9, R15
MOVQ R10, CX
MOVQ R11, DX
MOVQ R12, SI
MOVQ R13, BX
MOVQ $0xb9feffffffffaaab, DI
SUBQ DI, R14
MOVQ $0x1eabfffeb153ffff, DI
SBBQ DI, R15
MOVQ $0x6730d2a0f6b0f624, DI
SBBQ DI, CX
MOVQ $0x64774b84f38512bf, DI
SBBQ DI, DX
MOVQ $0x4b1ba7b6434bacd7, DI
SBBQ DI, SI
MOVQ $0x1a0111ea397fe69a, DI
SBBQ DI, BX
CMOVQCC R14, R8
CMOVQCC R15, R9
CMOVQCC CX, R10
CMOVQCC DX, R11
CMOVQCC SI, R12
CMOVQCC BX, R13
// |
MOVQ c+0(FP), DI
MOVQ R8, (DI)
MOVQ R9, 8(DI)
MOVQ R10, 16(DI)
MOVQ R11, 24(DI)
MOVQ R12, 32(DI)
MOVQ R13, 40(DI)
RET
/* | end */
// doubling w/ reduction
// a = (2 * a) % p
TEXT ·doubleAssign(SB), NOSPLIT, $0-8
// |
MOVQ a+0(FP), DI
MOVQ (DI), R8
MOVQ 8(DI), R9
MOVQ 16(DI), R10
MOVQ 24(DI), R11
MOVQ 32(DI), R12
MOVQ 40(DI), R13
ADDQ R8, R8
ADCQ R9, R9
ADCQ R10, R10
ADCQ R11, R11
ADCQ R12, R12
ADCQ R13, R13
// |
MOVQ R8, R14
MOVQ R9, R15
MOVQ R10, CX
MOVQ R11, DX
MOVQ R12, SI
MOVQ R13, BX
MOVQ $0xb9feffffffffaaab, AX
SUBQ AX, R14
MOVQ $0x1eabfffeb153ffff, AX
SBBQ AX, R15
MOVQ $0x6730d2a0f6b0f624, AX
SBBQ AX, CX
MOVQ $0x64774b84f38512bf, AX
SBBQ AX, DX
MOVQ $0x4b1ba7b6434bacd7, AX
SBBQ AX, SI
MOVQ $0x1a0111ea397fe69a, AX
SBBQ AX, BX
CMOVQCC R14, R8
CMOVQCC R15, R9
CMOVQCC CX, R10
CMOVQCC DX, R11
CMOVQCC SI, R12
CMOVQCC BX, R13
MOVQ R8, (DI)
MOVQ R9, 8(DI)
MOVQ R10, 16(DI)
MOVQ R11, 24(DI)
MOVQ R12, 32(DI)
MOVQ R13, 40(DI)
RET
/* | end */
// doubling w/o reduction
// c = 2 * a
TEXT ·ldouble(SB), NOSPLIT, $0-16
// |
MOVQ a+8(FP), DI
MOVQ (DI), R8
MOVQ 8(DI), R9
MOVQ 16(DI), R10
MOVQ 24(DI), R11
MOVQ 32(DI), R12
MOVQ 40(DI), R13
// |
ADDQ R8, R8
ADCQ R9, R9
ADCQ R10, R10
ADCQ R11, R11
ADCQ R12, R12
ADCQ R13, R13
// |
MOVQ c+0(FP), DI
MOVQ R8, (DI)
MOVQ R9, 8(DI)
MOVQ R10, 16(DI)
MOVQ R11, 24(DI)
MOVQ R12, 32(DI)
MOVQ R13, 40(DI)
RET
/* | end */
TEXT ·_neg(SB), NOSPLIT, $0-16
// |
MOVQ a+8(FP), DI
// |
MOVQ $0xb9feffffffffaaab, R8
MOVQ $0x1eabfffeb153ffff, R9
MOVQ $0x6730d2a0f6b0f624, R10
MOVQ $0x64774b84f38512bf, R11
MOVQ $0x4b1ba7b6434bacd7, R12
MOVQ $0x1a0111ea397fe69a, R13
SUBQ (DI), R8
SBBQ 8(DI), R9
SBBQ 16(DI), R10
SBBQ 24(DI), R11
SBBQ 32(DI), R12
SBBQ 40(DI), R13
// |
MOVQ c+0(FP), DI
MOVQ R8, (DI)
MOVQ R9, 8(DI)
MOVQ R10, 16(DI)
MOVQ R11, 24(DI)
MOVQ R12, 32(DI)
MOVQ R13, 40(DI)
RET
/* | end */
// multiplication without using MULX/ADX
// c = a * b % p
TEXT ·mulNoADX(SB), NOSPLIT, $24-24
// |
/* inputs */
MOVQ a+8(FP), DI
MOVQ b+16(FP), SI
MOVQ $0x00, R9
MOVQ $0x00, R10
MOVQ $0x00, R11
MOVQ $0x00, R12
MOVQ $0x00, R13
MOVQ $0x00, R14
MOVQ $0x00, R15
// |
/* i0 */
// | a0 @ CX
MOVQ (DI), CX
// | a0 * b0
MOVQ (SI), AX
MULQ CX
MOVQ AX, (SP)
MOVQ DX, R8
// | a0 * b1
MOVQ 8(SI), AX
MULQ CX
ADDQ AX, R8
ADCQ DX, R9
// | a0 * b2
MOVQ 16(SI), AX
MULQ CX
ADDQ AX, R9
ADCQ DX, R10
// | a0 * b3
MOVQ 24(SI), AX
MULQ CX
ADDQ AX, R10
ADCQ DX, R11
// | a0 * b4
MOVQ 32(SI), AX
MULQ CX
ADDQ AX, R11
ADCQ DX, R12
// | a0 * b5
MOVQ 40(SI), AX
MULQ CX
ADDQ AX, R12
ADCQ DX, R13
// |
/* i1 */
// | a1 @ CX
MOVQ 8(DI), CX
MOVQ $0x00, BX
// | a1 * b0
MOVQ (SI), AX
MULQ CX
ADDQ AX, R8
ADCQ DX, R9
ADCQ $0x00, R10
ADCQ $0x00, BX
MOVQ R8, 8(SP)
MOVQ $0x00, R8
// | a1 * b1
MOVQ 8(SI), AX
MULQ CX
ADDQ AX, R9
ADCQ DX, R10
ADCQ BX, R11
MOVQ $0x00, BX
ADCQ $0x00, BX
// | a1 * b2
MOVQ 16(SI), AX
MULQ CX
ADDQ AX, R10
ADCQ DX, R11
ADCQ BX, R12
MOVQ $0x00, BX
ADCQ $0x00, BX
// | a1 * b3
MOVQ 24(SI), AX
MULQ CX
ADDQ AX, R11
ADCQ DX, R12
ADCQ BX, R13
MOVQ $0x00, BX
ADCQ $0x00, BX
// | a1 * b4
MOVQ 32(SI), AX
MULQ CX
ADDQ AX, R12
ADCQ DX, R13
ADCQ BX, R14
// | a1 * b5
MOVQ 40(SI), AX
MULQ CX
ADDQ AX, R13
ADCQ DX, R14
// |
/* i2 */
// | a2 @ CX
MOVQ 16(DI), CX
MOVQ $0x00, BX
// | a2 * b0
MOVQ (SI), AX
MULQ CX
ADDQ AX, R9
ADCQ DX, R10
ADCQ $0x00, R11
ADCQ $0x00, BX
MOVQ R9, 16(SP)
MOVQ $0x00, R9
// | a2 * b1
MOVQ 8(SI), AX
MULQ CX
ADDQ AX, R10
ADCQ DX, R11
ADCQ BX, R12
MOVQ $0x00, BX
ADCQ $0x00, BX
// | a2 * b2
MOVQ 16(SI), AX
MULQ CX
ADDQ AX, R11
ADCQ DX, R12
ADCQ BX, R13
MOVQ $0x00, BX
ADCQ $0x00, BX
// | a2 * b3
MOVQ 24(SI), AX
MULQ CX
ADDQ AX, R12
ADCQ DX, R13
ADCQ BX, R14
MOVQ $0x00, BX
ADCQ $0x00, BX
// | a2 * b4
MOVQ 32(SI), AX
MULQ CX
ADDQ AX, R13
ADCQ DX, R14
ADCQ BX, R15
// | a2 * b5
MOVQ 40(SI), AX
MULQ CX
ADDQ AX, R14
ADCQ DX, R15
// |
/* i3 */
// | a3 @ CX
MOVQ 24(DI), CX
MOVQ $0x00, BX
// | a3 * b0
MOVQ (SI), AX
MULQ CX
ADDQ AX, R10
ADCQ DX, R11
ADCQ $0x00, R12
ADCQ $0x00, BX
// | a3 * b1
MOVQ 8(SI), AX
MULQ CX
ADDQ AX, R11
ADCQ DX, R12
ADCQ BX, R13
MOVQ $0x00, BX
ADCQ $0x00, BX
// | a3 * b2
MOVQ 16(SI), AX
MULQ CX
ADDQ AX, R12
ADCQ DX, R13
ADCQ BX, R14
MOVQ $0x00, BX
ADCQ $0x00, BX
// | a3 * b3
MOVQ 24(SI), AX
MULQ CX
ADDQ AX, R13
ADCQ DX, R14
ADCQ BX, R15
MOVQ $0x00, BX
ADCQ $0x00, BX
// | a3 * b4
MOVQ 32(SI), AX
MULQ CX
ADDQ AX, R14
ADCQ DX, R15
ADCQ BX, R8
// | a3 * b5
MOVQ 40(SI), AX
MULQ CX
ADDQ AX, R15
ADCQ DX, R8
// |
/* i4 */
// | a4 @ CX
MOVQ 32(DI), CX
MOVQ $0x00, BX
// | a4 * b0
MOVQ (SI), AX
MULQ CX
ADDQ AX, R11
ADCQ DX, R12
ADCQ $0x00, R13
ADCQ $0x00, BX
// | a4 * b1
MOVQ 8(SI), AX
MULQ CX
ADDQ AX, R12
ADCQ DX, R13
ADCQ BX, R14
MOVQ $0x00, BX
ADCQ $0x00, BX
// | a4 * b2
MOVQ 16(SI), AX
MULQ CX
ADDQ AX, R13
ADCQ DX, R14
ADCQ BX, R15
MOVQ $0x00, BX
ADCQ $0x00, BX
// | a4 * b3
MOVQ 24(SI), AX
MULQ CX
ADDQ AX, R14
ADCQ DX, R15
ADCQ BX, R8
MOVQ $0x00, BX
ADCQ $0x00, BX
// | a4 * b4
MOVQ 32(SI), AX
MULQ CX
ADDQ AX, R15
ADCQ DX, R8
ADCQ BX, R9
// | a4 * b5
MOVQ 40(SI), AX
MULQ CX
ADDQ AX, R8
ADCQ DX, R9
// |
/* i5 */
// | a5 @ CX
MOVQ 40(DI), CX
MOVQ $0x00, BX
// | a5 * b0
MOVQ (SI), AX
MULQ CX
ADDQ AX, R12
ADCQ DX, R13
ADCQ $0x00, R14
ADCQ $0x00, BX
// | a5 * b1
MOVQ 8(SI), AX
MULQ CX
ADDQ AX, R13
ADCQ DX, R14
ADCQ BX, R15
MOVQ $0x00, BX
ADCQ $0x00, BX
// | a5 * b2
MOVQ 16(SI), AX
MULQ CX
ADDQ AX, R14
ADCQ DX, R15
ADCQ BX, R8
MOVQ $0x00, BX
ADCQ $0x00, BX
// | a5 * b3
MOVQ 24(SI), AX
MULQ CX
ADDQ AX, R15
ADCQ DX, R8
ADCQ BX, R9
MOVQ $0x00, BX
ADCQ $0x00, BX
// | a5 * b4
MOVQ 32(SI), AX
MULQ CX
ADDQ AX, R8
ADCQ DX, R9
ADCQ $0x00, BX
// | a5 * b5
MOVQ 40(SI), AX
MULQ CX
ADDQ AX, R9
ADCQ DX, BX
// |
/* */
// |
// | W
// | 0 (SP) | 1 8(SP) | 2 16(SP) | 3 R10 | 4 R11 | 5 R12
// | 6 R13 | 7 R14 | 8 R15 | 9 R8 | 10 R9 | 11 BX
MOVQ (SP), CX
MOVQ 8(SP), DI
MOVQ 16(SP), SI
MOVQ BX, (SP)
MOVQ R9, 8(SP)
// |
/* montgomery reduction */
// |
/* i0 */
// |
// | W
// | 0 CX | 1 DI | 2 SI | 3 R10 | 4 R11 | 5 R12
// | 6 R13 | 7 R14 | 8 R15 | 9 R8 | 10 8(SP) | 11 (SP)
// | | u0 = w0 * inp
MOVQ CX, AX
MULQ ·inp+0(SB)
MOVQ AX, R9
MOVQ $0x00, BX
// |
/* */
// | j0
// | w0 @ CX
MOVQ ·modulus+0(SB), AX
MULQ R9
ADDQ AX, CX
ADCQ DX, BX
// | j1
// | w1 @ DI
MOVQ ·modulus+8(SB), AX
MULQ R9
ADDQ AX, DI
ADCQ $0x00, DX
ADDQ BX, DI
MOVQ $0x00, BX
ADCQ DX, BX
// | j2
// | w2 @ SI
MOVQ ·modulus+16(SB), AX
MULQ R9
ADDQ AX, SI
ADCQ $0x00, DX
ADDQ BX, SI
MOVQ $0x00, BX
ADCQ DX, BX
// | j3
// | w3 @ R10
MOVQ ·modulus+24(SB), AX
MULQ R9
ADDQ AX, R10
ADCQ $0x00, DX
ADDQ BX, R10
MOVQ $0x00, BX
ADCQ DX, BX
// | j4
// | w4 @ R11
MOVQ ·modulus+32(SB), AX
MULQ R9
ADDQ AX, R11
ADCQ $0x00, DX
ADDQ BX, R11
MOVQ $0x00, BX
ADCQ DX, BX
// | j5
// | w5 @ R12
MOVQ ·modulus+40(SB), AX
MULQ R9
ADDQ AX, R12
ADCQ $0x00, DX
ADDQ BX, R12
// | w6 @ R13
ADCQ DX, R13
ADCQ $0x00, CX
// |
/* i1 */
// |
// | W
// | 0 - | 1 DI | 2 SI | 3 R10 | 4 R11 | 5 R12
// | 6 R13 | 7 R14 | 8 R15 | 9 R8 | 10 8(SP) | 11 (SP)
// | | u1 = w1 * inp
MOVQ DI, AX
MULQ ·inp+0(SB)
MOVQ AX, R9
MOVQ $0x00, BX
// |
/* */
// | j0
// | w1 @ DI
MOVQ ·modulus+0(SB), AX
MULQ R9
ADDQ AX, DI
ADCQ DX, BX
// | j1
// | w2 @ SI
MOVQ ·modulus+8(SB), AX
MULQ R9
ADDQ AX, SI
ADCQ $0x00, DX
ADDQ BX, SI
MOVQ $0x00, BX
ADCQ DX, BX
// | j2
// | w3 @ R10
MOVQ ·modulus+16(SB), AX
MULQ R9
ADDQ AX, R10
ADCQ $0x00, DX
ADDQ BX, R10
MOVQ $0x00, BX
ADCQ DX, BX
// | j3
// | w4 @ R11
MOVQ ·modulus+24(SB), AX
MULQ R9
ADDQ AX, R11
ADCQ $0x00, DX
ADDQ BX, R11
MOVQ $0x00, BX
ADCQ DX, BX
// | j4
// | w5 @ R12
MOVQ ·modulus+32(SB), AX
MULQ R9
ADDQ AX, R12
ADCQ $0x00, DX
ADDQ BX, R12
MOVQ $0x00, BX
ADCQ DX, BX
// | j5
// | w6 @ R13
MOVQ ·modulus+40(SB), AX
MULQ R9
ADDQ AX, R13
ADCQ DX, CX
ADDQ BX, R13
// | w7 @ R14
ADCQ CX, R14
MOVQ $0x00, CX
ADCQ $0x00, CX
// |
/* i2 */
// |
// | W
// | 0 - | 1 - | 2 SI | 3 R10 | 4 R11 | 5 R12
// | 6 R13 | 7 R14 | 8 R15 | 9 R8 | 10 8(SP) | 11 (SP)
// | | u2 = w2 * inp
MOVQ SI, AX
MULQ ·inp+0(SB)
MOVQ AX, R9
MOVQ $0x00, BX
// |
/* */
// | j0
// | w2 @ SI
MOVQ ·modulus+0(SB), AX
MULQ R9
ADDQ AX, SI
ADCQ DX, BX
// | j1
// | w3 @ R10
MOVQ ·modulus+8(SB), AX
MULQ R9
ADDQ AX, R10
ADCQ $0x00, DX
ADDQ BX, R10
MOVQ $0x00, BX
ADCQ DX, BX
// | j2
// | w4 @ R11
MOVQ ·modulus+16(SB), AX
MULQ R9
ADDQ AX, R11
ADCQ $0x00, DX
ADDQ BX, R11
MOVQ $0x00, BX
ADCQ DX, BX
// | j3
// | w5 @ R12
MOVQ ·modulus+24(SB), AX
MULQ R9
ADDQ AX, R12
ADCQ $0x00, DX
ADDQ BX, R12
MOVQ $0x00, BX
ADCQ DX, BX
// | j4
// | w6 @ R13
MOVQ ·modulus+32(SB), AX
MULQ R9
ADDQ AX, R13
ADCQ $0x00, DX
ADDQ BX, R13
MOVQ $0x00, BX
ADCQ DX, BX
// | j5
// | w7 @ R14
MOVQ ·modulus+40(SB), AX
MULQ R9
ADDQ AX, R14
ADCQ DX, CX
ADDQ BX, R14
// | w8 @ R15
ADCQ CX, R15
MOVQ $0x00, CX
ADCQ $0x00, CX
// |
/* i3 */
// |
// | W
// | 0 - | 1 - | 2 - | 3 R10 | 4 R11 | 5 R12
// | 6 R13 | 7 R14 | 8 R15 | 9 R8 | 10 8(SP) | 11 (SP)
// | | u3 = w3 * inp
MOVQ R10, AX
MULQ ·inp+0(SB)
MOVQ AX, R9
MOVQ $0x00, BX
// |
/* */
// | j0
// | w3 @ R10
MOVQ ·modulus+0(SB), AX
MULQ R9
ADDQ AX, R10
ADCQ DX, BX
// | j1
// | w4 @ R11
MOVQ ·modulus+8(SB), AX
MULQ R9
ADDQ AX, R11
ADCQ $0x00, DX
ADDQ BX, R11
MOVQ $0x00, BX
ADCQ DX, BX
// | j2
// | w5 @ R12
MOVQ ·modulus+16(SB), AX
MULQ R9
ADDQ AX, R12
ADCQ $0x00, DX
ADDQ BX, R12
MOVQ $0x00, BX
ADCQ DX, BX
// | j3
// | w6 @ R13
MOVQ ·modulus+24(SB), AX
MULQ R9
ADDQ AX, R13
ADCQ $0x00, DX
ADDQ BX, R13
MOVQ $0x00, BX
ADCQ DX, BX
// | j4
// | w7 @ R14
MOVQ ·modulus+32(SB), AX
MULQ R9
ADDQ AX, R14
ADCQ $0x00, DX
ADDQ BX, R14
MOVQ $0x00, BX
ADCQ DX, BX
// | j5
// | w8 @ R15
MOVQ ·modulus+40(SB), AX
MULQ R9
ADDQ AX, R15
ADCQ DX, CX
ADDQ BX, R15
// | w9 @ R8
ADCQ CX, R8
MOVQ $0x00, CX
ADCQ $0x00, CX
// |
/* i4 */
// |
// | W
// | 0 - | 1 - | 2 - | 3 - | 4 R11 | 5 R12
// | 6 R13 | 7 R14 | 8 R15 | 9 R8 | 10 8(SP) | 11 (SP)
// | | u4 = w4 * inp
MOVQ R11, AX
MULQ ·inp+0(SB)
MOVQ AX, R9
MOVQ $0x00, BX
// |
/* */
// | j0
// | w4 @ R11
MOVQ ·modulus+0(SB), AX
MULQ R9
ADDQ AX, R11
ADCQ DX, BX
// | j1
// | w5 @ R12
MOVQ ·modulus+8(SB), AX
MULQ R9
ADDQ AX, R12
ADCQ $0x00, DX
ADDQ BX, R12
MOVQ $0x00, BX
ADCQ DX, BX
// | j2
// | w6 @ R13
MOVQ ·modulus+16(SB), AX
MULQ R9
ADDQ AX, R13
ADCQ $0x00, DX
ADDQ BX, R13
MOVQ $0x00, BX
ADCQ DX, BX
// | j3
// | w7 @ R14
MOVQ ·modulus+24(SB), AX
MULQ R9
ADDQ AX, R14
ADCQ $0x00, DX
ADDQ BX, R14
MOVQ $0x00, BX
ADCQ DX, BX
// | j4
// | w8 @ R15
MOVQ ·modulus+32(SB), AX
MULQ R9
ADDQ AX, R15
ADCQ $0x00, DX
ADDQ BX, R15
MOVQ $0x00, BX
ADCQ DX, BX
// | j5
// | w9 @ R8
MOVQ ·modulus+40(SB), AX
MULQ R9
ADDQ AX, R8
ADCQ DX, CX
ADDQ BX, R8
// | move to idle register
MOVQ 8(SP), DI
// | w10 @ DI
ADCQ CX, DI
MOVQ $0x00, CX
ADCQ $0x00, CX
// |
/* i5 */
// |
// | W
// | 0 - | 1 - | 2 - | 3 - | 4 - | 5 R12
// | 6 R13 | 7 R14 | 8 R15 | 9 R8 | 10 DI | 11 (SP)
// | | u5 = w5 * inp
MOVQ R12, AX
MULQ ·inp+0(SB)
MOVQ AX, R9
MOVQ $0x00, BX
// |
/* */
// | j0
// | w5 @ R12
MOVQ ·modulus+0(SB), AX
MULQ R9
ADDQ AX, R12
ADCQ DX, BX
// | j1
// | w6 @ R13
MOVQ ·modulus+8(SB), AX
MULQ R9
ADDQ AX, R13
ADCQ $0x00, DX
ADDQ BX, R13
MOVQ $0x00, BX
ADCQ DX, BX
// | j2
// | w7 @ R14
MOVQ ·modulus+16(SB), AX
MULQ R9
ADDQ AX, R14
ADCQ $0x00, DX
ADDQ BX, R14
MOVQ $0x00, BX
ADCQ DX, BX
// | j3
// | w8 @ R15
MOVQ ·modulus+24(SB), AX
MULQ R9
ADDQ AX, R15
ADCQ $0x00, DX
ADDQ BX, R15
MOVQ $0x00, BX
ADCQ DX, BX
// | j4
// | w9 @ R8
MOVQ ·modulus+32(SB), AX
MULQ R9
ADDQ AX, R8
ADCQ $0x00, DX
ADDQ BX, R8
MOVQ $0x00, BX
ADCQ DX, BX
// | j5
// | w10 @ DI
MOVQ ·modulus+40(SB), AX
MULQ R9
ADDQ AX, DI
ADCQ DX, CX
ADDQ BX, DI
// | w11 @ CX
ADCQ (SP), CX
// |
// | W montgomerry reduction ends
// | 0 - | 1 - | 2 - | 3 - | 4 - | 5 -
// | 6 R13 | 7 R14 | 8 R15 | 9 R8 | 10 DI | 11 CX
// |
/* modular reduction */
MOVQ R13, R10
SUBQ ·modulus+0(SB), R10
MOVQ R14, R11
SBBQ ·modulus+8(SB), R11
MOVQ R15, R12
SBBQ ·modulus+16(SB), R12
MOVQ R8, AX
SBBQ ·modulus+24(SB), AX
MOVQ DI, BX
SBBQ ·modulus+32(SB), BX
MOVQ CX, R9
SBBQ ·modulus+40(SB), R9
// |
/* out */
MOVQ c+0(FP), SI
CMOVQCC R10, R13
MOVQ R13, (SI)
CMOVQCC R11, R14
MOVQ R14, 8(SI)
CMOVQCC R12, R15
MOVQ R15, 16(SI)
CMOVQCC AX, R8
MOVQ R8, 24(SI)
CMOVQCC BX, DI
MOVQ DI, 32(SI)
CMOVQCC R9, CX
MOVQ CX, 40(SI)
RET
// |
/* end */
// multiplication
// c = a * b % p
TEXT ·mulADX(SB), NOSPLIT, $16-24
// |
/* inputs */
MOVQ a+8(FP), DI
MOVQ b+16(FP), SI
XORQ AX, AX
// |
/* i0 */
// | a0 @ DX
MOVQ (DI), DX
// | a0 * b0
MULXQ (SI), AX, CX
MOVQ AX, (SP)
// | a0 * b1
MULXQ 8(SI), AX, R8
ADCXQ AX, CX
// | a0 * b2
MULXQ 16(SI), AX, R9
ADCXQ AX, R8
// | a0 * b3
MULXQ 24(SI), AX, R10
ADCXQ AX, R9
// | a0 * b4
MULXQ 32(SI), AX, R11
ADCXQ AX, R10
// | a0 * b5
MULXQ 40(SI), AX, R12
ADCXQ AX, R11
ADCQ $0x00, R12
// |
/* i1 */
// | a1 @ DX
MOVQ 8(DI), DX
XORQ R13, R13
// | a1 * b0
MULXQ (SI), AX, BX
ADOXQ AX, CX
ADCXQ BX, R8
MOVQ CX, 8(SP)
// | a1 * b1
MULXQ 8(SI), AX, BX
ADOXQ AX, R8
ADCXQ BX, R9
// | a1 * b2
MULXQ 16(SI), AX, BX
ADOXQ AX, R9
ADCXQ BX, R10
// | a1 * b3
MULXQ 24(SI), AX, BX
ADOXQ AX, R10
ADCXQ BX, R11
// | a1 * b4
MULXQ 32(SI), AX, BX
ADOXQ AX, R11
ADCXQ BX, R12
// | a1 * b5
MULXQ 40(SI), AX, BX
ADOXQ AX, R12
ADOXQ R13, R13
ADCXQ BX, R13
// |
/* i2 */
// | a2 @ DX
MOVQ 16(DI), DX
XORQ R14, R14
// | a2 * b0
MULXQ (SI), AX, BX
ADOXQ AX, R8
ADCXQ BX, R9
// | a2 * b1
MULXQ 8(SI), AX, BX
ADOXQ AX, R9
ADCXQ BX, R10
// | a2 * b2
MULXQ 16(SI), AX, BX
ADOXQ AX, R10
ADCXQ BX, R11
// | a2 * b3
MULXQ 24(SI), AX, BX
ADOXQ AX, R11
ADCXQ BX, R12
// | a2 * b4
MULXQ 32(SI), AX, BX
ADOXQ AX, R12
ADCXQ BX, R13
// | a2 * b5
MULXQ 40(SI), AX, BX
ADOXQ AX, R13
ADOXQ R14, R14
ADCXQ BX, R14
// |
/* i3 */
// | a3 @ DX
MOVQ 24(DI), DX
XORQ R15, R15
// | a3 * b0
MULXQ (SI), AX, BX
ADOXQ AX, R9
ADCXQ BX, R10
// | a3 * b1
MULXQ 8(SI), AX, BX
ADOXQ AX, R10
ADCXQ BX, R11
// | a3 * b2
MULXQ 16(SI), AX, BX
ADOXQ AX, R11
ADCXQ BX, R12
// | a3 * b3
MULXQ 24(SI), AX, BX
ADOXQ AX, R12
ADCXQ BX, R13
// | a3 * b4
MULXQ 32(SI), AX, BX
ADOXQ AX, R13
ADCXQ BX, R14
// | a3 * b5
MULXQ 40(SI), AX, BX
ADOXQ AX, R14
ADOXQ R15, R15
ADCXQ BX, R15
// |
/* i4 */
// | a4 @ DX
MOVQ 32(DI), DX
XORQ CX, CX
// | a4 * b0
MULXQ (SI), AX, BX
ADOXQ AX, R10
ADCXQ BX, R11
// | a4 * b1
MULXQ 8(SI), AX, BX
ADOXQ AX, R11
ADCXQ BX, R12
// | a4 * b2
MULXQ 16(SI), AX, BX
ADOXQ AX, R12
ADCXQ BX, R13
// | a4 * b3
MULXQ 24(SI), AX, BX
ADOXQ AX, R13
ADCXQ BX, R14
// | a4 * b4
MULXQ 32(SI), AX, BX
ADOXQ AX, R14
ADCXQ BX, R15
// | a4 * b5
MULXQ 40(SI), AX, BX
ADOXQ AX, R15
ADOXQ CX, CX
ADCXQ BX, CX
// |
/* i5 */
// | a5 @ DX
MOVQ 40(DI), DX
XORQ DI, DI
// | a5 * b0
MULXQ (SI), AX, BX
ADOXQ AX, R11
ADCXQ BX, R12
// | a5 * b1
MULXQ 8(SI), AX, BX
ADOXQ AX, R12
ADCXQ BX, R13
// | a5 * b2
MULXQ 16(SI), AX, BX
ADOXQ AX, R13
ADCXQ BX, R14
// | a5 * b3
MULXQ 24(SI), AX, BX
ADOXQ AX, R14
ADCXQ BX, R15
// | a5 * b4
MULXQ 32(SI), AX, BX
ADOXQ AX, R15
ADCXQ BX, CX
// | a5 * b5
MULXQ 40(SI), AX, BX
ADOXQ AX, CX
ADOXQ BX, DI
ADCQ $0x00, DI
// |
/* */
// |
// | W
// | 0 (SP) | 1 8(SP) | 2 R8 | 3 R9 | 4 R10 | 5 R11
// | 6 R12 | 7 R13 | 8 R14 | 9 R15 | 10 CX | 11 DI
MOVQ (SP), BX
MOVQ 8(SP), SI
MOVQ DI, (SP)
// |
// | W ready to mont
// | 0 BX | 1 SI | 2 R8 | 3 R9 | 4 R10 | 5 R11
// | 6 R12 | 7 R13 | 8 R14 | 9 R15 | 10 CX | 11 (SP)
// |
/* montgomery reduction */
// | clear flags
XORQ AX, AX
// |
/* i0 */
// |
// | W
// | 0 BX | 1 SI | 2 R8 | 3 R9 | 4 R10 | 5 R11
// | 6 R12 | 7 R13 | 8 R14 | 9 R15 | 10 CX | 11 (SP)
// | | u0 = w0 * inp
MOVQ BX, DX
MULXQ ·inp+0(SB), DX, DI
// |
/* */
// | j0
// | w0 @ BX
MULXQ ·modulus+0(SB), AX, DI
ADOXQ AX, BX
ADCXQ DI, SI
// | j1
// | w1 @ SI
MULXQ ·modulus+8(SB), AX, DI
ADOXQ AX, SI
ADCXQ DI, R8
// | j2
// | w2 @ R8
MULXQ ·modulus+16(SB), AX, DI
ADOXQ AX, R8
ADCXQ DI, R9
// | j3
// | w3 @ R9
MULXQ ·modulus+24(SB), AX, DI
ADOXQ AX, R9
ADCXQ DI, R10
// | j4
// | w4 @ R10
MULXQ ·modulus+32(SB), AX, DI
ADOXQ AX, R10
ADCXQ DI, R11
// | j5
// | w5 @ R11
MULXQ ·modulus+40(SB), AX, DI
ADOXQ AX, R11
ADCXQ DI, R12
ADOXQ BX, R12
ADCXQ BX, BX
MOVQ $0x00, AX
ADOXQ AX, BX
// | clear flags
XORQ AX, AX
// |
/* i1 */
// |
// | W
// | 0 - | 1 SI | 2 R8 | 3 R9 | 4 R10 | 5 R11
// | 6 R12 | 7 R13 | 8 R14 | 9 R15 | 10 CX | 11 (SP)
// | | u1 = w1 * inp
MOVQ SI, DX
MULXQ ·inp+0(SB), DX, DI
// |
/* */
// | j0
// | w1 @ SI
MULXQ ·modulus+0(SB), AX, DI
ADOXQ AX, SI
ADCXQ DI, R8
// | j1
// | w2 @ R8
MULXQ ·modulus+8(SB), AX, DI
ADOXQ AX, R8
ADCXQ DI, R9
// | j2
// | w3 @ R9
MULXQ ·modulus+16(SB), AX, DI
ADOXQ AX, R9
ADCXQ DI, R10
// | j3
// | w4 @ R10
MULXQ ·modulus+24(SB), AX, DI
ADOXQ AX, R10
ADCXQ DI, R11
// | j4
// | w5 @ R11
MULXQ ·modulus+32(SB), AX, DI
ADOXQ AX, R11
ADCXQ DI, R12
// | j5
// | w6 @ R12
MULXQ ·modulus+40(SB), AX, DI
ADOXQ AX, R12
ADCXQ DI, R13
ADOXQ BX, R13
ADCXQ SI, SI
MOVQ $0x00, AX
ADOXQ AX, SI
// | clear flags
XORQ AX, AX
// |
/* i2 */
// |
// | W
// | 0 - | 1 - | 2 R8 | 3 R9 | 4 R10 | 5 R11
// | 6 R12 | 7 R13 | 8 R14 | 9 R15 | 10 CX | 11 (SP)
// | | u2 = w2 * inp
MOVQ R8, DX
MULXQ ·inp+0(SB), DX, DI
// |
/* */
// | j0
// | w2 @ R8
MULXQ ·modulus+0(SB), AX, DI
ADOXQ AX, R8
ADCXQ DI, R9
// | j1
// | w3 @ R9
MULXQ ·modulus+8(SB), AX, DI
ADOXQ AX, R9
ADCXQ DI, R10
// | j2
// | w4 @ R10
MULXQ ·modulus+16(SB), AX, DI
ADOXQ AX, R10
ADCXQ DI, R11
// | j3
// | w5 @ R11
MULXQ ·modulus+24(SB), AX, DI
ADOXQ AX, R11
ADCXQ DI, R12
// | j4
// | w6 @ R12
MULXQ ·modulus+32(SB), AX, DI
ADOXQ AX, R12
ADCXQ DI, R13
// | j5
// | w7 @ R13
MULXQ ·modulus+40(SB), AX, DI
ADOXQ AX, R13
ADCXQ DI, R14
ADOXQ SI, R14
ADCXQ R8, R8
MOVQ $0x00, AX
ADOXQ AX, R8
// | clear flags
XORQ AX, AX
// |
/* i3 */
// |
// | W
// | 0 - | 1 - | 2 - | 3 R9 | 4 R10 | 5 R11
// | 6 R12 | 7 R13 | 8 R14 | 9 R15 | 10 CX | 11 (SP)
// | | u3 = w3 * inp
MOVQ R9, DX
MULXQ ·inp+0(SB), DX, DI
// |
/* */
// | j0
// | w3 @ R9
MULXQ ·modulus+0(SB), AX, DI
ADOXQ AX, R9
ADCXQ DI, R10
// | j1
// | w4 @ R10
MULXQ ·modulus+8(SB), AX, DI
ADOXQ AX, R10
ADCXQ DI, R11
// | j2
// | w5 @ R11
MULXQ ·modulus+16(SB), AX, DI
ADOXQ AX, R11
ADCXQ DI, R12
// | j3
// | w6 @ R12
MULXQ ·modulus+24(SB), AX, DI
ADOXQ AX, R12
ADCXQ DI, R13
// | j4
// | w7 @ R13
MULXQ ·modulus+32(SB), AX, DI
ADOXQ AX, R13
ADCXQ DI, R14
// | j5
// | w8 @ R14
MULXQ ·modulus+40(SB), AX, DI
ADOXQ AX, R14
ADCXQ DI, R15
ADOXQ R8, R15
ADCXQ R9, R9
MOVQ $0x00, AX
ADOXQ AX, R9
// | clear flags
XORQ AX, AX
// |
/* i4 */
// |
// | W
// | 0 - | 1 - | 2 - | 3 - | 4 R10 | 5 R11
// | 6 R12 | 7 R13 | 8 R14 | 9 R15 | 10 CX | 11 (SP)
// | | u4 = w4 * inp
MOVQ R10, DX
MULXQ ·inp+0(SB), DX, DI
// |
/* */
// | j0
// | w4 @ R10
MULXQ ·modulus+0(SB), AX, DI
ADOXQ AX, R10
ADCXQ DI, R11
// | j1
// | w5 @ R11
MULXQ ·modulus+8(SB), AX, DI
ADOXQ AX, R11
ADCXQ DI, R12
// | j2
// | w6 @ R12
MULXQ ·modulus+16(SB), AX, DI
ADOXQ AX, R12
ADCXQ DI, R13
// | j3
// | w7 @ R13
MULXQ ·modulus+24(SB), AX, DI
ADOXQ AX, R13
ADCXQ DI, R14
// | j4
// | w8 @ R14
MULXQ ·modulus+32(SB), AX, DI
ADOXQ AX, R14
ADCXQ DI, R15
// | j5
// | w9 @ R15
MULXQ ·modulus+40(SB), AX, DI
ADOXQ AX, R15
ADCXQ DI, CX
ADOXQ R9, CX
ADCXQ R10, R10
MOVQ $0x00, AX
ADOXQ AX, R10
// | clear flags
XORQ AX, AX
// |
/* i5 */
// |
// | W
// | 0 - | 1 - | 2 - | 3 - | 4 - | 5 R11
// | 6 R12 | 7 R13 | 8 R14 | 9 R15 | 10 CX | 11 (SP)
// | | u5 = w5 * inp
MOVQ R11, DX
MULXQ ·inp+0(SB), DX, DI
// |
/* */
// | j0
// | w5 @ R11
MULXQ ·modulus+0(SB), AX, DI
ADOXQ AX, R11
ADCXQ DI, R12
// | j1
// | w6 @ R12
MULXQ ·modulus+8(SB), AX, DI
ADOXQ AX, R12
ADCXQ DI, R13
// | j2
// | w7 @ R13
MULXQ ·modulus+16(SB), AX, DI
ADOXQ AX, R13
ADCXQ DI, R14
// | j3
// | w8 @ R14
MULXQ ·modulus+24(SB), AX, DI
ADOXQ AX, R14
ADCXQ DI, R15
// | j4
// | w9 @ R15
MULXQ ·modulus+32(SB), AX, DI
ADOXQ AX, R15
ADCXQ DI, CX
// | j5
// | w10 @ CX
MULXQ ·modulus+40(SB), AX, DI
ADOXQ AX, CX
// | w11 @ (SP)
// | move to an idle register
MOVQ (SP), BX
ADCXQ DI, BX
ADOXQ R10, BX
// |
// | W montgomery reduction ends
// | 0 - | 1 - | 2 - | 3 - | 4 - | 5 -
// | 6 R12 | 7 R13 | 8 R14 | 9 R15 | 10 CX | 11 BX
// |
/* modular reduction */
MOVQ R12, AX
SUBQ ·modulus+0(SB), AX
MOVQ R13, DI
SBBQ ·modulus+8(SB), DI
MOVQ R14, SI
SBBQ ·modulus+16(SB), SI
MOVQ R15, R8
SBBQ ·modulus+24(SB), R8
MOVQ CX, R9
SBBQ ·modulus+32(SB), R9
MOVQ BX, R10
SBBQ ·modulus+40(SB), R10
// |
/* out */
MOVQ c+0(FP), R11
CMOVQCC AX, R12
MOVQ R12, (R11)
CMOVQCC DI, R13
MOVQ R13, 8(R11)
CMOVQCC SI, R14
MOVQ R14, 16(R11)
CMOVQCC R8, R15
MOVQ R15, 24(R11)
CMOVQCC R9, CX
MOVQ CX, 32(R11)
CMOVQCC R10, BX
MOVQ BX, 40(R11)
RET
// |
/* end */