ADCXQ BX, R8
	ADOXQ 32(AX), R8
	MOVQ  R8, 32(AX)

	// Iteration 5
	MULXQ 40(CX), R8, BX
	ADCXQ DI, R8
	ADOXQ 40(AX), R8
	MOVQ  R8, 40(AX)

	// Iteration 6
	MULXQ 48(CX), R8, DI
	ADCXQ BX, R8
	ADOXQ 48(AX), R8
	MOVQ  R8, 48(AX)

	// Iteration 7
	MULXQ 56(CX), R8, BX
	ADCXQ DI, R8
	ADOXQ 56(AX), R8
	MOVQ  R8, 56(AX)

	// Iteration 8
	MULXQ 64(CX), R8, DI

	BEQ	R3, R4, samebytes
	SGTU	R1, R2, R7
	MOVW	R1, R8
	CMOVN	R7, R2, R8	// R8 is min(R1, R2)

	ADDU	R3, R8	// R3 is current byte in a, R8 is last byte in a to compare
loop:
	BEQ	R3, R8, samebytes

	MOVBU	(R3), R6
	ADDU	$1, R3
	MOVBU	(R4), R7
	ADDU	$1, R4
	BEQ	R6, R7 , loop

	SGTU	R6, R7, R8
	MOVW	$-1, R6
	CMOVZ	R8, R6, R8
	JMP	cmp_ret
samebytes:
	SGTU	R1, R2, R6
	SGTU	R2, R1, R7

	JA   _5_or_more
	MOVL (R8), R8
	LEAQ -3(DI)(DX*1), DX
loop4:
	MOVL (DI), SI
	CMPL SI,R8
	JZ   success
	ADDQ $1,DI
	CMPQ DI,DX
	JB loop4
	JMP fail
_5_or_more:
	CMPQ AX, $7
	JA   _8_or_more
	LEAQ 1(DI)(DX*1), DX
	SUBQ AX, DX
	MOVL -4(R8)(AX*1), BX
	MOVL (R8), R8
loop5to7:
	MOVL (DI), SI
	CMPL SI,R8
	JZ   partial_success5to7
	ADDQ $1,DI
	CMPQ DI,DX
	JB loop5to7
	JMP fail

	CMPBNE	R5, $16, movemt16
	MOVD	0(R4), R7
	MOVD	8(R4), R8
	MOVD	R7, 0(R6)
	MOVD	R8, 8(R6)
	RET

movemt16:
	CMPBGT	R4, R6, forwards
	ADD	R5, R4, R7
	CMPBLE	R7, R6, forwards
	ADD	R5, R6, R8
backwards:
	MOVD	-8(R7), R3
	MOVD	R3, -8(R8)
	MOVD	-16(R7), R3
	MOVD	R3, -16(R8)
	ADD	$-16, R5
	ADD	$-16, R7
	ADD	$-16, R8
	CMP	R5, $16
	BGE	backwards
	BR	start

forwards:

DATA zosLibVec<>(SB)/8, $0
GLOBL zosLibVec<>(SB), NOPTR, $8

TEXT ·initZosLibVec(SB), NOSPLIT|NOFRAME, $0-0
	MOVW PSALAA, R8
	MOVD LCA64(R8), R8
	MOVD CAA(R8), R8
	MOVD EDCHPXV(R8), R8
	MOVD R8, zosLibVec<>(SB)
	RET

TEXT ·GetZosLibVec(SB), NOSPLIT|NOFRAME, $0-0
	MOVD zosLibVec<>(SB), R8
	MOVD R8, ret+0(FP)
	RET

TEXT ·clearErrno(SB), NOSPLIT, $0-0
	BL   addrerrno<>(SB)
	MOVD $0, 0(R3)
	RET

largeLoop:  // Copying 256 bytes at a time.
	MVC    $256, 0(R6), 0(R8)
	MOVD   $256(R6), R6
	MOVD   $256(R8), R8
	MOVD   $-32(R5), R5
	CMPBGE R5, $32, largeLoop
	BR     mediumLoop

mediumLoopBody:  // Copying 32 bytes at a time
	MVC    $32, 0(R6), 0(R8)
	MOVD   $32(R6), R6
	MOVD   $32(R8), R8
	MOVD   $-4(R5), R5
	CMPBGE R5, $4, mediumLoopBody
	BR     smallLoop

returnC:

chunk16_loop:
	BEQ	R0, R14, byte_loop
	MOVV	(R4), R8
	MOVV	(R6), R9
	BNE	R8, R9, byte_loop
	MOVV	8(R4), R16
	MOVV	8(R6), R17
	ADDV	$16, R4
	ADDV	$16, R6
	SUBVU	$1, R14
	BEQ	R16, R17, chunk16_loop
	SUBV	$8, R4
	SUBV	$8, R6

byte_loop:
	BEQ	R4, R12, samebytes
	MOVBU	(R4), R8
	ADDVU	$1, R4
	MOVBU	(R6), R9
	ADDVU	$1, R6
	BEQ	R8, R9, byte_loop

byte_cmp:
	SGTU	R8, R9, R4 // R12 = 1 if (R8 > R9)

	BEQ	R1, R6, f_tail_ua
	MOVWHI	0(R2), R8
	ADDU	$4, R1
	MOVWLO	3(R2), R8
	ADDU	$4, R2
	MOVW	R8, -4(R1)
	JMP	f_words_ua

f_tail_ua:
	BEQ	R1, R5, ret
	MOVWHI	-4(R4), R8
	MOVWLO	-1(R4), R8
	MOVWLO	R8, -1(R5)
	JMP	ret

f_small_copy:
	BEQ	R1, R5, ret
	ADDU	$1, R1
	MOVB	0(R2), R6
	ADDU	$1, R2
	MOVB	R6, -1(R1)
	JMP	f_small_copy

backward:
	SGTU	$4, R3, R6
	BNE	R6, b_small_copy

	// 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))

	MOVWZ	(R5)(R8), R15
	XOR	R14, R15, R16
	MOVW	R16, (R3)(R8)
	ADD	$4,R8
	ADD	$-4,R6
xor2:
	CMP	R6, $2
	BLT	xor1
	MOVHZ	(R4)(R8), R14
	MOVHZ	(R5)(R8), R15
	XOR	R14, R15, R16
	MOVH	R16, (R3)(R8)
	ADD	$2,R8
	ADD	$-2,R6
xor1:
	CMP	R6, $0
	BC	12,2,LR		// BEQLR
	MOVBZ	(R4)(R8), R14	// R14 = a[i]
	MOVBZ	(R5)(R8), R15	// R15 = b[i]
	XOR	R14, R15, R16	// R16 = a[i] ^ b[i]