ADDV	$-7, R9, R6 // R6 is end pointer-7

	PCALIGN	$16
	SGTU	R6, R4, R8
	BEQ	R8, out
	MOVV	(R5), R7
	ADDV	$8, R5
	MOVV	R7, (R4)
	ADDV	$8, R4
	JMP	-6(PC)

out:
	BEQ	R4, R9, done
	MOVB	(R5), R7
	ADDV	$1, R5
	MOVB	R7, (R4)
	ADDV	$1, R4
	JMP	-5(PC)
done:
	RET

backward:
	ADDV	R6, R5 // from-end pointer
	ADDV	R4, R6, R9 // to-end pointer

	LDADDLD	R5, (R6), ZR                         // df0065f8
	LDADDLW	R5, (R6), ZR                         // df0065b8
	LDADDLH	R5, (R6), ZR                         // df006578
	LDADDLB	R5, (R6), ZR                         // df006538
	LDCLRD	R5, (R6), ZR                         // df1025f8
	LDCLRW	R5, (R6), ZR                         // df1025b8
	LDCLRH	R5, (R6), ZR                         // df102578

	BR     E6

A6:
	MOVD   (R8)(R1*1), R6
	MULHDU R9, R6
	MOVD   (R2)(R1*1), R10
	ADDC   R10, R11        // add to low order bits
	ADDE   R0, R6
	ADDC   R4, R11
	ADDE   R0, R6
	MOVD   R6, R4
	MOVD   R11, (R2)(R1*1)

	MOVD   (8)(R8)(R1*1), R6
	MULHDU R9, R6
	MOVD   (8)(R2)(R1*1), R10
	ADDC   R10, R11           // add to low order bits
	ADDE   R0, R6
	ADDC   R4, R11
	ADDE   R0, R6
	MOVD   R6, R4

	ADD	$32, R8
	ADD	$-32, R6
	CMP	R6, $8
	BLE	small
	// Case for 8 <= n < 32 bytes
	// Process 16 bytes if available
xor16:
	CMP	R6, $16
	BLT	xor8
	LXVD2X	(R4)(R8), VS32
	LXVD2X	(R5)(R8), VS33
	XXLXOR	VS32, VS33, VS32
	STXVD2X	VS32, (R3)(R8)
	ADD	$16, R8
	ADD	$-16, R6
small:
	CMP	R6, $0
	BC	12,2,LR		// BEQLR
xor8:
#ifdef GOPPC64_power10
	SLD	$56,R6,R17
	ADD	R4,R8,R18
	ADD	R5,R8,R19

	MOVD	ptr+0(FP), R4
	MOVW	new+8(FP), R3
	MOVW	(R4), R6
repeat:
	CS	R6, R3, (R4) // if R6==(R4) then (R4)=R3 else R6=(R4)
	BNE	repeat
	MOVW	R6, ret+16(FP)
	RET

// func Xchg64(ptr *uint64, new uint64) uint64
TEXT ·Xchg64(SB), NOSPLIT, $0-24
	MOVD	ptr+0(FP), R4
	MOVD	new+8(FP), R3
	MOVD	(R4), R6
repeat:
	CSG	R6, R3, (R4) // if R6==(R4) then (R4)=R3 else R6=(R4)
	BNE	repeat
	MOVD	R6, ret+16(FP)
	RET

	MOVD.W	-8(R2), R6
	LSR	R4, R6, R5	// return value
	LSL	R3, R6, R8	// x[i] << s
	SUB	$1, R1
one:	TBZ	$0, R1, two
	MOVD.W	-8(R2), R6
	LSR	R4, R6, R7
	ORR	R8, R7
	LSL	R3, R6, R8
	SUB	$1, R1
	MOVD.W	R7, -8(R0)
two:
	TBZ	$1, R1, loop
	LDP.W	-16(R2), (R6, R7)
	LSR	R4, R7, R10
	ORR	R8, R10
	LSL	R3, R7
	LSR	R4, R6, R9
	ORR	R7, R9
	LSL	R3, R6, R8
	SUB	$2, R1

	CMP	$0, R6
	BEQ	exit	// g.m == nil

	// restore libcall
	MOVD	96(R1), R7
	MOVD	R7, (m_libcall+libcall_fn)(R6)
	MOVD	104(R1), R7
	MOVD	R7, (m_libcall+libcall_args)(R6)
	MOVD	112(R1), R7
	MOVD	R7, (m_libcall+libcall_n)(R6)
	MOVD	120(R1), R7
	MOVD	R7, (m_libcall+libcall_r1)(R6)
	MOVD	128(R1), R7
	MOVD	R7, (m_libcall+libcall_r2)(R6)

	// restore errno
	MOVD	(m_mOS+mOS_perrno)(R6), R7

	MOVD	(R1), R5
	EOR	R4, R5
	CBNZ	R5, not_equal
	SUB	$8, R2, R6	// offset of the last 8 bytes
	MOVD	(R0)(R6), R4
	MOVD	(R1)(R6), R5
	EOR	R4, R5
	CBNZ	R5, not_equal
	B	equal
lt_8:
	TBZ	$2, R2, lt_4
	MOVWU	(R0), R4
	MOVWU	(R1), R5
	EOR	R4, R5
	CBNZ	R5, not_equal
	SUB	$4, R2, R6	// offset of the last 4 bytes
	MOVWU	(R0)(R6), R4
	MOVWU	(R1)(R6), R5
	EOR	R4, R5
	CBNZ	R5, not_equal
	B	equal
lt_4:

	// alignment requirements.
	ANDCC	$PAGE_OFFSET, R8, R6	// &sX & PAGE_OFFSET
	ANDCC	$PAGE_OFFSET, R4, R9
	SUBC	R5, $8, R12		// 8-len
	SLD	$3, R12, R14		// (8-len)*8
	CMPU	R6, R12, CR1		// Enough bytes lower in the page to load lower?
	CMPU	R9, R12, CR0
	SUB	R12, R8, R6		// compute lower load address
	SUB	R12, R4, R9
	ISEL	CR1LT, R8, R6, R8	// R8 = R6 < 0 ? R8 (&s1) : R6 (&s1 - (8-len))
	ISEL	CR0LT, R4, R9, R4	// Similar for s2

	RLWNM $3, R4, $29, $31, R6      // 54861f7e
	RLWNM $0, R4, $29, $31, R6      // 5486077e
	RLWNM R0, R4, $29, $31, R6      // 5c86077e
	RLWNM R3, R4, $7, R6            // 5c861f7e
	RLWNM R3, R4, $29, $31, R6      // 5c861f7e
	RLWNMCC $3, R4, $7, R6          // 54861f7f
	RLWNMCC $3, R4, $29, $31, R6    // 54861f7f
	RLWNMCC R3, R4, $7, R6          // 5c861f7f
	RLWNMCC R3, R4, $29, $31, R6    // 5c861f7f
	RLDMI $0, R4, $7, R6            // 7886076c