ADDV	$8, R21

	MOVV	(R20), R30
	ADDV	$8, R20
	MOVV	R30, (R21)
	ADDV	$8, R21

	MOVV	(R20), R30
	ADDV	$8, R20
	MOVV	R30, (R21)
	ADDV	$8, R21

	MOVV	(R20), R30
	ADDV	$8, R20
	MOVV	R30, (R21)
	ADDV	$8, R21

	MOVV	(R20), R30
	ADDV	$8, R20
	MOVV	R30, (R21)
	ADDV	$8, R21

	MOVV	(R20), R30
	ADDV	$8, R20
	MOVV	R30, (R21)
	ADDV	$8, R21

	MOVV	(R20), R30

	ORRW $16252928, ZR, R21                    // f5130d32
	MOVD $-4260607558625, R11                  // eb6b16b2
	MOVD R30, R7                               // e7031eaa
	MOVKW $(3905<<0), R21                      // MOVKW $3905, R21              // 35e88172
	MOVKW $(3905<<16), R21                     // MOVKW $255918080, R21         // 35e8a172

	MOVD	g_m(g), R21	// R21 = m

	MOVD	$0, R3		// CLOCK_REALTIME

	MOVD	runtime·vdsoClockgettimeSym(SB), R12	// Check for VDSO availability
	CMP	R12, $0
	BEQ	fallback

	// Set vdsoPC and vdsoSP for SIGPROF traceback.
	// Save the old values on stack and restore them on exit,
	// so this function is reentrant.
	MOVD	m_vdsoPC(R21), R4
	MOVD	m_vdsoSP(R21), R5
	MOVD	R4, 32(R1)
	MOVD	R5, 40(R1)

	MOVD	RSP, R1

	MOVD	g_m(g), R21	// R21 = m

	// Set vdsoPC and vdsoSP for SIGPROF traceback.
	// Save the old values on stack and restore them on exit,
	// so this function is reentrant.
	MOVD	m_vdsoPC(R21), R2
	MOVD	m_vdsoSP(R21), R3
	MOVD	R2, 8(RSP)
	MOVD	R3, 16(RSP)

	MOVD	$ret-8(FP), R2 // caller's SP
	MOVD	LR, m_vdsoPC(R21)
	MOVD	R2, m_vdsoSP(R21)

	MOVD	m_curg(R21), R0
	CMP	g, R0

	XOR	R21,R24,R21	// xor done R24
	MOVWZ	(R10)(R8),R25	// tab[4][crc>>24]
	RLDICL	$48,R7,$56,R24	// crc>>16&0xFF
	XOR	R21,R25,R21	// xor done R25
	ADD	$1024,R10,R10	// &tab[5]
	SLD	$2,R24,R24	// crc>>16&0xFF*4
	MOVWZ	(R10)(R24),R26	// tab[5][crc>>16&0xFF]
	XOR	R21,R26,R21	// xor done R26
	RLDICL	$56,R7,$56,R25	// crc>>8
	ADD	$1024,R10,R10	// &tab[6]
	SLD	$2,R25,R25	// crc>>8&FF*2
	MOVBZ   R7,R26          // crc&0xFF

loop:
	MOVD    8(R8), R20        // R20 = x[i]
	MOVD    16(R8), R21       // R21 = x[i+1]
	MOVD    24(R8), R22       // R22 = x[i+2]
	MOVDU   32(R8), R23       // R23 = x[i+3]
	MULLD   R9, R20, R24      // R24 = z0[i]
	MULHDU  R9, R20, R20      // R20 = z1[i]
	ADDC    R4, R24           // R24 = z0[i] + c
	MULLD   R9, R21, R25
	MULHDU  R9, R21, R21
	ADDE    R20, R25
	MULLD   R9, R22, R26
	MULHDU  R9, R22, R22

	MOVD $2, R15      // Set up index

	// Set up masks for use with VSEL
	MOVD   $0xff, R21        // Set up mask 0xff000000ff000000...
	SLD    $24, R21
	MTVSRD R21, V10
	VSPLTW $1, V10, V29
	VSLDOI $2, V29, V29, V30 // Mask 0x0000ff000000ff00...
	MOVD   $0xffff, R21
	SLD    $16, R21
	MTVSRD R21, V10
	VSPLTW $1, V10, V31      // Mask 0xffff0000ffff0000...
	VSPLTW $0, V0, V1        // Splat 1st word of separator

//	}
	SLL	$19, R22, R21	// 0016acc0
	SLLV	$19, R22, R21	// 0016acf8
	SRL	$31, R6, R17	// 00068fc2
	SRLV	$31, R6, R17	// 00068ffa
	SRA	$8, R8, R19	// 00089a03
	SRAV	$19, R8, R7	// 00083cfb
	ROTR	$12, R8, R3	// 00281b02
	ROTRV	$8, R22, R22	// 0036b23a

//	LSHW imm ',' rreg
//	{
//		outcode(int($1), &$2, 0, &$4);
//	}
	SLL	$19, R21	// 0015acc0
	SLLV	$19, R21	// 0015acf8

	LSL	R4, R6, R20
	LSR	R3, R6
	ORR	R8, R20
	LSL	R4, R7, R21
	LSR	R3, R7, R8
	ORR	R6, R21
	STP.P	(R20, R21), 16(R0)
	SUB	$2, R1
loop:
	CBZ	R1, done
	LDP.P	32(R2), (R10, R11)
	LDP	-16(R2), (R12, R13)
	LSL	R4, R10, R20
	LSR	R3, R10
	ORR	R8, R20		// z[i] = (x[i] >> s) | (x[i+1] << (64 - s))
	LSL	R4, R11, R21
	LSR	R3, R11
	ORR	R10, R21
	LSL	R4, R12, R22
	LSR	R3, R12
	ORR	R11, R22

		// large or unaligned move
		// arg0 = address of dst memory (in R21, changed as side effect)
		// arg1 = address of src memory (in R20, changed as side effect)
		// arg2 = address of the last element of src
		// arg3 = mem
		// auxint = alignment
		// returns mem
		//	MOVx	(R20), Rtmp
		//	MOVx	Rtmp, (R21)
		//	ADDV	$sz, R20
		//	ADDV	$sz, R21
		//	BGEU	Rarg2, R20, -4(PC)
		{
			name:      "LoweredMove",