}

	// ±0 - y
	// x - ±Inf
	return z.Neg(y)
}

// Mul sets z to the rounded product x*y and returns z.
// Precision, rounding, and accuracy reporting are as for [Float.Add].
// Mul panics with [ErrNaN] if one operand is zero and the other
// operand an infinity. The value of z is undefined in that case.
func (z *Float) Mul(x, y *Float) *Float {
	if debugFloat {
		x.validate()
		y.validate()
	}

	// FMUL <Sd>, <Sn>, <Sm>
	{0xffe0fc00, 0x1e200800, FMUL, instArgs{arg_Sd, arg_Sn, arg_Sm}, nil},
	// FMUL <Dd>, <Dn>, <Dm>
	{0xffe0fc00, 0x1e600800, FMUL, instArgs{arg_Dd, arg_Dn, arg_Dm}, nil},
	// FMUL <Vd>.<t>, <Vn>.<t>, <Vm>.<t>

	"UBFIZ",
	"UBFIZW",
	"UBFM",
	"UBFMW",
	"UBFX",
	"UBFXW",
	"UCVTFD",
	"UCVTFS",
	"UCVTFWD",
	"UCVTFWS",
	"UDIV",
	"UDIVW",
	"UMADDL",
	"UMNEGL",
	"UMSUBL",
	"UMULH",
	"UMULL",
	"UREM",
	"UREMW",
	"UXTB",
	"UXTBW",
	"UXTH",
	"UXTHW",
	"UXTW",
	"VADD",
	"VADDP",
	"VADDV",
	"VAND",
	"VBCAX",
	"VBIF",
	"VBIT",
	"VBSL",
	"VCMEQ",

		index := aa.IfIndex
		if index == 0 { // ipv6IfIndex is a substitute for ifIndex
			index = aa.Ipv6IfIndex
		}
		if ifi == nil || ifi.Index == int(index) {
			for pmul := aa.FirstMulticastAddress; pmul != nil; pmul = pmul.Next {
				sa, err := pmul.Address.Sockaddr.Sockaddr()
				if err != nil {
					return nil, os.NewSyscallError("sockaddr", err)
				}
				switch sa := sa.(type) {
				case *syscall.SockaddrInet4:

loop:
	CBZ	R0, done
	LDP.P	32(R2), (R5, R6)
	LDP	-16(R2), (R7, R8)

	MUL	R3, R5, R10
	UMULH	R3, R5, R11
	ADDS	R4, R10
	MUL	R3, R6, R12
	UMULH	R3, R6, R13
	ADCS	R11, R12

	MUL	R3, R7, R14
	UMULH	R3, R7, R15
	ADCS	R13, R14
	MUL	R3, R8, R16
	UMULH	R3, R8, R17
	ADCS	R15, R16
	ADC	$0, R17, R4

	STP.P	(R10, R12), 32(R1)
	STP	(R14, R16), -16(R1)
	SUB	$4, R0
	B	loop

	LDP	(R1), (R9, R10)
	ADDS	R4, R9
	MUL	R6, R3, R14
	ADCS	R14, R10
	MUL	R7, R3, R15
	LDP	16(R1), (R11, R12)
	ADCS	R15, R11
	MUL	R8, R3, R16
	ADCS	R16, R12
	UMULH	R8, R3, R20
	ADC	$0, R20

	MUL	R5, R3, R13
	ADDS	R13, R9
	UMULH	R5, R3, R17
	ADCS	R17, R10
	UMULH	R6, R3, R21
	STP.P	(R9, R10), 16(R1)
	ADCS	R21, R11
	UMULH	R7, R3, R19
	ADCS	R19, R12
	STP.P	(R11, R12), 16(R1)

	// arm/7:"ADDD",-"MULD"
	// arm64:"FADDD",-"FMULD"
	// ppc64x:"FADD",-"FMUL"
	// riscv64:"FADDD",-"FMULD"
	return f * 2.0
}

func DivPow2(f1, f2, f3 float64) (float64, float64, float64) {
	// 386/sse2:"MULSD",-"DIVSD"
	// amd64:"MULSD",-"DIVSD"
	// arm/7:"MULD",-"DIVD"
	// arm64:"FMULD",-"FDIVD"
	// ppc64x:"FMUL",-"FDIV"
	// riscv64:"FMULD",-"FDIVD"
	x := f1 / 16.0

	// 386/sse2:"MULSD",-"DIVSD"

//	then UV = u*v + x*(u*VV+v*UU) + x*x*UU*VV

func Mul(U, V PS) PS {
	Z := mkPS()
	go func() {
		<-Z.req
		uv := get2(U, V)
		if end(uv[0]) != 0 || end(uv[1]) != 0 {
			Z.dat <- finis
		} else {
			Z.dat <- mul(uv[0], uv[1])
			UU := Split(U)
			VV := Split(V)
			W := Add(Cmul(uv[0], VV[0]), Cmul(uv[1], UU[0]))
			<-Z.req
			Z.dat <- get(W)
			copy(Add(W, Mul(UU[1], VV[1])), Z)
		}
	}()
	return Z

// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.

package main

import (
	"fmt"
)

// When using soft-float, OMUL might be rewritten to function
// call so we should ensure it was evaluated first. Stack frame
// setup for "test" function call should happen after call to runtime.fmul32

var x int32 = 1

func main() {

	LDGR	R2, F0
	FMADD	F0, F4, F0
	MOVD	$·expx4ff<>+0(SB), R3
	FMOVD	0(R3), F2
	FMUL	F2, F0
	FMOVD	F0, ret+8(FP)
	RET
L13:
	FMOVD	$0, F0
	FMOVD	F0, ret+8(FP)
	RET
L21:
	ADDW	$0x1000, R1
	RISBGN	$0, $15, $48, R1, R2
	LDGR	R2, F0
	FMADD	F0, F4, F0
	MOVD	$·expx2ff<>+0(SB), R3
	FMOVD	0(R3), F2
	FMUL	F2, F0
	FMOVD	F0, ret+8(FP)
	RET
LEXITTAGexp:
	FMOVD	F0, ret+8(FP)