//go:build !purego

#include "textflag.h"

// func addMulVVW1024(z, x *uint, y uint) (c uint)
TEXT ·addMulVVW1024(SB), $0-16
	MOVL	$32, BX
	JMP		addMulVVWx(SB)

// func addMulVVW1536(z, x *uint, y uint) (c uint)
TEXT ·addMulVVW1536(SB), $0-16
	MOVL	$48, BX
	JMP		addMulVVWx(SB)

// func addMulVVW2048(z, x *uint, y uint) (c uint)
TEXT ·addMulVVW2048(SB), $0-16
	MOVL	$64, BX
	JMP		addMulVVWx(SB)

#include "textflag.h"

// func addMulVVW1024(z, x *uint, y uint) (c uint)
TEXT ·addMulVVW1024(SB), $0-32
	MOVD	$4, R6 // R6 = z_len/4
	JMP		addMulVVWx<>(SB)

// func addMulVVW1536(z, x *uint, y uint) (c uint)
TEXT ·addMulVVW1536(SB), $0-32
	MOVD	$6, R6 // R6 = z_len/4
	JMP		addMulVVWx<>(SB)

// func addMulVVW2048(z, x *uint, y uint) (c uint)
TEXT ·addMulVVW2048(SB), $0-32
	MOVD	$8, R6 // R6 = z_len/4

//go:build !purego

#include "textflag.h"

// func addMulVVW1024(z, x *uint, y uint) (c uint)
TEXT ·addMulVVW1024(SB), $0-32
	MOVD	$16, R0
	JMP		addMulVVWx(SB)

// func addMulVVW1536(z, x *uint, y uint) (c uint)
TEXT ·addMulVVW1536(SB), $0-32
	MOVD	$24, R0
	JMP		addMulVVWx(SB)

// func addMulVVW2048(z, x *uint, y uint) (c uint)
TEXT ·addMulVVW2048(SB), $0-32
	MOVD	$32, R0
	JMP		addMulVVWx(SB)

//go:build !purego

#include "textflag.h"

// func addMulVVW1024(z, x *uint, y uint) (c uint)
TEXT ·addMulVVW1024(SB), $0-16
	MOVW	$32, R5
	JMP		addMulVVWx(SB)

// func addMulVVW1536(z, x *uint, y uint) (c uint)
TEXT ·addMulVVW1536(SB), $0-16
	MOVW	$48, R5
	JMP		addMulVVWx(SB)

// func addMulVVW2048(z, x *uint, y uint) (c uint)
TEXT ·addMulVVW2048(SB), $0-16
	MOVW	$64, R5
	JMP		addMulVVWx(SB)

//go:build !purego

#include "textflag.h"

// func addMulVVW1024(z, x *uint, y uint) (c uint)
TEXT ·addMulVVW1024(SB), $0-32
	MOVD	$16, R5
	JMP		addMulVVWx(SB)

// func addMulVVW1536(z, x *uint, y uint) (c uint)
TEXT ·addMulVVW1536(SB), $0-32
	MOVD	$24, R5
	JMP		addMulVVWx(SB)

// func addMulVVW2048(z, x *uint, y uint) (c uint)
TEXT ·addMulVVW2048(SB), $0-32
	MOVD	$32, R5
	JMP		addMulVVWx(SB)

package bigmod

import "unsafe"

func addMulVVW1024(z, x *uint, y uint) (c uint) {
	return addMulVVW(unsafe.Slice(z, 1024/_W), unsafe.Slice(x, 1024/_W), y)
}

func addMulVVW1536(z, x *uint, y uint) (c uint) {
	return addMulVVW(unsafe.Slice(z, 1536/_W), unsafe.Slice(x, 1536/_W), y)
}

func addMulVVW2048(z, x *uint, y uint) (c uint) {

// TODO(filippo): file cmd/compile performance issue.

var supportADX = cpu.X86.HasADX && cpu.X86.HasBMI2

//go:noescape
func addMulVVW1024(z, x *uint, y uint) (c uint)

//go:noescape
func addMulVVW1536(z, x *uint, y uint) (c uint)

//go:noescape

//go:build !purego

#include "textflag.h"

// func addMulVVW1024(z, x *uint, y uint) (c uint)
TEXT ·addMulVVW1024(SB),$0-32
	MOV	$16, X30
	JMP	addMulVVWx(SB)

// func addMulVVW1536(z, x *uint, y uint) (c uint)
TEXT ·addMulVVW1536(SB),$0-32
	MOV	$24, X30
	JMP	addMulVVWx(SB)

// func addMulVVW2048(z, x *uint, y uint) (c uint)
TEXT ·addMulVVW2048(SB),$0-32
	MOV	$32, X30
	JMP	addMulVVWx(SB)

	ADOXQ 120(AX), R8
	MOVQ  R8, 120(AX)

	// Add back carry flags and return
	ADCXQ SI, BX
	ADOXQ SI, BX
	MOVQ  BX, c+24(FP)
	RET

// func addMulVVW1536(z *uint, x *uint, y uint) (c uint)
// Requires: ADX, BMI2
TEXT ·addMulVVW1536(SB), $0-32
	CMPB ·supportADX+0(SB), $0x01
	JEQ  adx
	MOVQ z+0(FP), CX
	MOVQ x+8(FP), BX
	MOVQ y+16(FP), SI
	XORQ DI, DI

	// Iteration 0
	MOVQ (BX), AX

	case 1536 / _W:
		const n = 1536 / _W // compiler hint
		T := make([]uint, n*2)
		var c uint
		for i := 0; i < n; i++ {
			d := bLimbs[i]
			c1 := addMulVVW1536(&T[i], &aLimbs[0], d)
			Y := T[i] * m.m0inv
			c2 := addMulVVW1536(&T[i], &mLimbs[0], Y)
			T[n+i], c = bits.Add(c1, c2, c)
		}
		copy(x.reset(n).limbs, T[n:])
		x.maybeSubtractModulus(choice(c), m)

	case 2048 / _W: