SLD	$2,R19,R19	// p[5]*4:1
	MOVWZ	(R10)(R18),R22	// tab[1][p[6]]
	ADD	$1024,R10,R10	// tab[2]
	XOR	R21,R22,R21	// xor done R22
	CLRLSLDI $56,R9,$2,R20
	MOVWZ	(R10)(R19),R23	// tab[2][p[5]]
	ADD	$1024,R10,R10	// &tab[3]
	XOR	R21,R23,R21	// xor done R23
	MOVWZ	(R10)(R20),R24	// tab[3][p[4]]
	ADD 	$1024,R10,R10   // &tab[4]
	XOR	R21,R24,R21	// xor done R24
	MOVWZ	(R10)(R8),R25	// tab[4][crc>>24]

	ADD	b, a;


TEXT ·block(SB),NOSPLIT,$0-32
	MOVD	dig+0(FP), R10
	MOVD	p+8(FP), R6
	MOVD	p_len+16(FP), R5

	// We assume p_len >= 64
	SRD 	$6, R5
	MOVD	R5, CTR

	MOVWZ	0(R10), R22
	MOVWZ	4(R10), R3
	MOVWZ	8(R10), R4
	MOVWZ	12(R10), R5

loop:
	MOVD	R22, R14
	MOVD	R3, R15
	MOVD	R4, R16
	MOVD	R5, R17

	ENDIAN_MOVE( 0,R6,M00,M15)
	ENDIAN_MOVE( 4,R6,M01,M15)
	ENDIAN_MOVE( 8,R6,M02,M15)

	XOR	R14, R15, R16   // R16 = a[] ^ b[]
	SUB	$8, R6          // n = n - 8
	MOVD	R16, (R3)(R8)   // Store to dst
	ADD	$8, R8
xor4:
	CMP	R6, $4
	BLT	xor2
	MOVWZ	(R4)(R8), R14
	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

	MOVD  bpx_offset+24(FP), R2 // r2 offset to BPX vector table
	MOVD  R14, R7               // save r14
	MOVD  R15, R8               // save r15
	MOVWZ 16(R0), R9
	MOVWZ 544(R9), R9
	MOVWZ 24(R9), R9            // call vector in r9
	ADD   R2, R9                // add offset to vector table
	MOVWZ (R9), R9              // r9 points to entry point
	BYTE  $0x0D                 // BL R14,R9 --> basr r14,r9

	// ppc64x: "RLWNM\t[$]25, R[0-9]+, [$]27, [$]29, R[0-9]+"
	z[2] = uint32(uint8(a>>7)) & 0x1c
	// ppc64x: -"MOVWZ"
	y[0] = uint64((a >> 6) & 0x1c)
	// ppc64x: -"MOVWZ"
	y[1] = uint64(uint32(b)<<6) + 1
	// ppc64x: -"MOVHZ", -"MOVWZ"
	y[2] = uint64((uint16(a) >> 9) & 0x1F)
	// ppc64x: -"MOVHZ", -"MOVWZ", -"ANDCC"
	y[3] = uint64(((uint16(a) & 0xFF0) >> 9) & 0x1F)
}

// 128 bit shifts

	MOVW (R3)(R4), R5               // 7ca41aaa
	MOVW (R3)(R0), R5               // 7ca01aaa
	MOVW (R3), R5                   // e8a30002
	MOVWZ 4(R3), R4                 // 80830004
	MOVWZ (R3)(R4), R5              // 7ca4182e
	MOVWZ (R3)(R0), R5              // 7ca0182e
	MOVWZ (R3), R5                  // 80a30000
	MOVH 4(R3), R4                  // a8830004
	MOVH (R3)(R4), R5               // 7ca41aae

	NOPH
	MOVWZ 0(R3), R3
	MOVD  R3, err+48(FP)
	MOVD  zosLibVec<>(SB), R8
	ADD   $(__err2ad), R8
	LMG   0(R8), R5, R6
	LE_CALL                   // balr R7, R6 __err2ad (return #2)
	NOPH
	MOVW  (R3), R2            // retrieve errno2
	MOVD  R2, errno2+40(FP)   // store in return area
	XOR   R2, R2
	MOVWZ R2, (R3)            // clear errno2

done:

	// s390x:`MOVHBR\t\(.*\), [R]`,`CMPW\t.*, [$]`
	return s == "xx"
}

func CompareString2(s string) bool {
	// amd64:`CMPL\t\(.*\), [$]`
	// arm64:`MOVWU\t\(.*\), [R]`,`CMPW\t.*, [R]`
	// ppc64le:`MOVWZ\t\(.*\), [R]`,`CMPW\t.*, [R]`
	// s390x:`MOVWBR\t\(.*\), [R]`,`CMPW\t.*, [$]`
	return s == "xxxx"
}

func CompareString3(s string) bool {
	// amd64:`CMPQ\t\(.*\), [A-Z]`
	// arm64:-`CMPW\t`
	// ppc64x:-`CMPW\t`

	io64[2] = (io64[2] >> 4) & 0x0000FFFFFFFFFFFF
	// ppc64x: -"SRD", -"AND", "RLDICL\t[$]36, R[0-9]*, [$]28, R"
	io64[3] = (io64[3] >> 28) & 0x0000FFFFFFFFFFFF

	// ppc64x: "MOVWZ", "RLWNM\t[$]1, R[0-9]*, [$]28, [$]3, R"
	io64[4] = uint64(bits.RotateLeft32(io32[0], 1) & 0xF000000F)

	// ppc64x: "RLWNM\t[$]0, R[0-9]*, [$]4, [$]19, R"
	io32[0] = io32[0] & 0x0FFFF000

#endif						\
	/* copy return values back */		\
	MOVD	regArgs+40(FP), R20;		\
	BL	runtime·spillArgs(SB);			\
	MOVD	stackArgsType+0(FP), R7;		\
	MOVD	stackArgs+16(FP), R3;			\
	MOVWZ	stackArgsSize+24(FP), R4;			\
	MOVWZ	stackRetOffset+28(FP), R6;		\
	ADD	$FIXED_FRAME, R1, R5;		\
	ADD	R6, R5; 			\
	ADD	R6, R3;				\
	SUB	R6, R4;				\
	BL	callRet<>(SB);			\
	RET