ASUBMEVCC
	ASUBMEV
	ASUBV
	ASUBE
	ASUBECC
	ASUBEV
	ASUBEVCC
	ASUBZE
	ASUBZECC
	ASUBZEVCC
	ASUBZEV
	ASYNC
	AXOR
	AXORCC
	AXORIS

	ADCBF
	ADCBI
	ADCBST
	ADCBT
	ADCBTST
	ADCBZ
	AEIEIO
	AICBI
	AISYNC
	APTESYNC
	ATLBIE
	ATLBIEL
	ATLBSYNC
	ATW

	ASYSCALL
	AWORD

	ARFCI

	AFCPSGN
	AFCPSGNCC

	ADCS	acc2, acc2
	ADCS	acc3, acc3
	ADCS	acc4, acc4
	ADCS	acc5, acc5
	ADCS	acc6, acc6
	ADC	$0, acc7
	// Missing products
	MUL	x0, x0, acc0
	UMULH	x0, x0, t0
	ADDS	t0, acc1, acc1

	MUL	x1, x1, t0
	ADCS	t0, acc2, acc2
	UMULH	x1, x1, t1
	ADCS	t1, acc3, acc3

	MUL	x2, x2, t0
	ADCS	t0, acc4, acc4
	UMULH	x2, x2, t1
	ADCS	t1, acc5, acc5

	MUL	x3, x3, t0
	ADCS	t0, acc6, acc6

	ADDQ t1, acc3
	ADCQ $0, DX
	ADDQ AX, acc3
	ADCQ $0, DX
	MOVQ DX, t1

	MOVQ (8*3)(x_ptr), AX
	MULQ t0
	ADDQ t1, acc4
	ADCQ $0, DX
	ADDQ AX, acc4
	ADCQ DX, acc5
	ADCQ $0, acc0
	// Second reduction step
	MOVQ acc1, AX
	MOVQ acc1, t1
	SHLQ $32, acc1
	MULQ p256const1<>(SB)
	SHRQ $32, t1
	ADDQ acc1, acc2
	ADCQ t1, acc3
	ADCQ AX, acc4
	ADCQ DX, acc5
	ADCQ $0, acc0

	testenv.ParallelOn64Bit(t)
	h := newHashSet()
	var b [3]byte
	for i := 0; i < 256; i++ {
		b[0] = byte(i)
		h.addB(b[:1])
		for j := 0; j < 256; j++ {
			b[1] = byte(j)
			h.addB(b[:2])
			if !testing.Short() {
				for k := 0; k < 256; k++ {
					b[2] = byte(k)
					h.addB(b[:3])
				}
			}
		}
	}
	h.check(t)
}

// Different length strings of all zeros have distinct hashes.

	}
	testenv.ParallelOn64Bit(t)
	h := newHashSet()
	var b [3]byte
	for i := 0; i < 256; i++ {
		b[0] = byte(i)
		h.addB(b[:1])
		for j := 0; j < 256; j++ {
			b[1] = byte(j)
			h.addB(b[:2])
			if !testing.Short() {
				for k := 0; k < 256; k++ {
					b[2] = byte(k)
					h.addB(b[:3])
				}
			}
		}
	}
	h.check(t)
}

// Different length strings of all zeros have distinct hashes.

	i := (addr - span.base()) / goarch.PtrSize / ptrBits
	j := (addr - span.base()) / goarch.PtrSize % ptrBits
	bits := span.elemsize / goarch.PtrSize
	word0 := (*uintptr)(unsafe.Pointer(addb(hbits, goarch.PtrSize*(i+0))))
	word1 := (*uintptr)(unsafe.Pointer(addb(hbits, goarch.PtrSize*(i+1))))

	var read uintptr
	if j+bits > ptrBits {
		// Two reads.
		bits0 := ptrBits - j
		bits1 := bits - bits0
		read = *word0 >> j

			ASTBCCC,
			ASTWCCC,
			AEIEIO,
			AICBI,
			AISYNC,
			ATLBIE,
			ATLBIEL,
			ASLBIA,
			ASLBIE,
			ASLBMFEE,
			ASLBMFEV,
			ASLBMTE,
			ADCBF,
			ADCBI,
			ADCBST,
			ADCBT,
			ADCBTST,
			ADCBZ,
			ASYNC,
			ATLBSYNC,
			APTESYNC,
			ALWSYNC,
			ATW,
			AWORD,
			ARFI,
			ARFCI,
			ARFID,
			AHRFID:
			q = p
			p.Mark |= LABEL | SYNC

	skipMask := off / goarch.PtrSize / 8
	skipBytes := skipMask * goarch.PtrSize * 8
	ptrmask := addb(gcbits, skipMask)
	src = add(src, skipBytes)
	off -= skipBytes
	size += off
	var bits uint32
	for i := uintptr(0); i < size; i += goarch.PtrSize {
		if i&(goarch.PtrSize*8-1) == 0 {
			bits = uint32(*ptrmask)
			ptrmask = addb(ptrmask, 1)
		} else {
			bits >>= 1
		}
		if off > 0 {
			off -= goarch.PtrSize

	{as: ATW, a1: C_32CON, a2: C_REG, a6: C_S16CON, type_: 61, size: 4},
	{as: ADCBF, a1: C_SOREG, type_: 43, size: 4},
	{as: ADCBF, a1: C_XOREG, type_: 43, size: 4},
	{as: ADCBF, a1: C_XOREG, a2: C_REG, a6: C_U15CON, type_: 43, size: 4},
	{as: ADCBF, a1: C_SOREG, a6: C_U15CON, type_: 43, size: 4},
	{as: ADCBF, a1: C_XOREG, a6: C_U15CON, type_: 43, size: 4},

	MOVL	delta_lo+4(FP), SI
	MOVL	delta_hi+8(FP), DI
	// DX:AX = *addr
	MOVL	0(BP), AX
	MOVL	4(BP), DX
addloop:
	// CX:BX = DX:AX (*addr) + DI:SI (delta)
	MOVL	AX, BX
	MOVL	DX, CX
	ADDL	SI, BX
	ADCL	DI, CX

	// if *addr == DX:AX {
	//	*addr = CX:BX
	// } else {
	//	DX:AX = *addr
	// }
	// all in one instruction
	LOCK
	CMPXCHG8B	0(BP)

	JNZ	addloop

	// success