MOVQ CR2, (AX)                  // ERROR "invalid instruction"
	MOVQ CR3, (AX)                  // ERROR "invalid instruction"
	MOVQ CR4, (AX)                  // ERROR "invalid instruction"
	MOVQ CR8, (AX)                  // ERROR "invalid instruction"
	MOVQ (AX), CR0                  // ERROR "invalid instruction"
	MOVQ (AX), CR2                  // ERROR "invalid instruction"
	MOVQ (AX), CR3                  // ERROR "invalid instruction"

	DCBT	(R6)		// cache hint

	SRD	$6,R9,R14	// There is at least one iteration.
	MOVD	R14,CTR
	ANDCC   $63,R9,R9
	CMP	R9,$16,CR1	// Do setup for tail check early on.
	CMP	R9,$32,CR2
	CMP	R9,$48,CR3
	ADD	$-16,R9,R9

	MOVD	$32,R11		// set offsets to load into vector
	MOVD	$48,R12		// set offsets to load into vector

	PCALIGN	$16
cmp64_loop:
	LXVD2X	(R5)(R0),V3	// load bytes of A at offset 0 into vector

	DS:   "DS",
	ES:   "ES",
	FS:   "FS",
	GS:   "GS",
	GDTR: "GDTR",
	IDTR: "IDTR",
	LDTR: "LDTR",
	MSW:  "MSW",
	TASK: "TASK",
	CR0:  "CR0",
	CR1:  "CR1",
	CR2:  "CR2",
	CR3:  "CR3",
	CR4:  "CR4",
	CR5:  "CR5",
	CR6:  "CR6",
	CR7:  "CR7",
	CR8:  "CR8",
	CR9:  "CR9",
	CR10: "CR10",
	CR11: "CR11",
	CR12: "CR12",
	CR13: "CR13",
	CR14: "CR14",
	CR15: "CR15",

	CMP	LEN, $16
	BGT	mcopy
	SLD	$56, LEN, TMP
	LXVL	SRC, TMP, V0
	STXVL	V0, TGT, TMP
	RET
#endif
mcopy:
	ANDCC	$7, LEN, BYTES	// R7: bytes to copy
	SRD	$3, LEN, DWORDS	// R6: double words to copy
	MOVFL	CR0, CR3	// save CR from ANDCC
	CMP	DWORDS, $0, CR1	// CR1[EQ] set if no double words to copy

	// Determine overlap by subtracting dest - src and comparing against the