MOVW	b+4(FP), R2
	BEQ	R1, R2, eq
	MOVW	size+8(FP), R3
	ADDU	R1, R3, R4
loop:
	BNE	R1, R4, test
	MOVW	$1, R1
	MOVB	R1, ret+12(FP)
	RET
test:
	MOVBU	(R1), R6
	ADDU	$1, R1
	MOVBU	(R2), R7
	ADDU	$1, R2
	BEQ	R6, R7, loop

	MOVB	R0, ret+12(FP)
	RET
eq:
	MOVW	$1, R1
	MOVB	R1, ret+12(FP)
	RET

// memequal_varlen(a, b unsafe.Pointer) bool
TEXT runtime·memequal_varlen(SB),NOSPLIT,$0-9

	MOVD.P -8(R10), R8         <=>      ldr x8, [x10],#-8
	MOVB.W 16(R16), R10        <=>      ldrsb x10, [x16,#16]!
	MOVBU.W 16(R16), R10       <=>      ldrb x10, [x16,#16]!

3. Go uses a series of MOV instructions as load and store.

64-bit variant ldr, str, stur => MOVD;
32-bit variant str, stur, ldrsw => MOVW;
32-bit variant ldr => MOVWU;
ldrb => MOVBU; ldrh => MOVHU;
ldrsb, sturb, strb => MOVB;
ldrsh, sturh, strh =>  MOVH.

	MOVHZ (R3), R5                  // a0a30000
	MOVB 1(R3), R4                  // 888300017c840774
	MOVB (R3)(R4), R5               // 7ca418ae7ca50774
	MOVB (R3)(R0), R5               // 7ca018ae7ca50774
	MOVB (R3), R5                   // 88a300007ca50774
	MOVBZ 1(R3), R4                 // 88830001
	MOVBZ (R3)(R4), R5              // 7ca418ae
	MOVBZ (R3)(R0), R5              // 7ca018ae
	MOVBZ (R3), R5                  // 88a30000

	MOVV	R4, result+65540(FP)		// 1e020014de8f1000c433c029
	MOVV	R4, result+4097(FP)   		// 3e000014de8f1000c427c029
	MOVB	R4, result+65540(FP)		// 1e020014de8f1000c4330029
	MOVB	R4, result+4097(FP)   		// 3e000014de8f1000c4270029
	MOVBU	R4, result+65540(FP)		// 1e020014de8f1000c4330029
	MOVBU	R4, result+4097(FP)		// 3e000014de8f1000c4270029
	MOVW	R4, 65536(R5)			// 1e020014de971000c4038029

	MOVHU	(R1), R6
	MOVHU	(R2), R7
	BNE	R6, R7, not_eq
	ADDV	$2, R1
	ADDV	$2, R2

chunk_tail_1:
	AND	$1, R3, R14
	BEQ	R0, R14, eq
	MOVBU	(R1), R6
	MOVBU	(R2), R7
	BEQ	R6, R7, eq

not_eq:
	MOVB	R0, ret+24(FP)
	RET
eq:
	MOVV	$1, R1
	MOVB	R1, ret+24(FP)
	RET

// memequal_varlen(a, b unsafe.Pointer) bool
TEXT runtime·memequal_varlen(SB),NOSPLIT,$40-17
	MOVV	a+0(FP), R1

TEXT ·Cas(SB), NOSPLIT, $0-17
	MOVD	ptr+0(FP), R0
	MOVW	old+8(FP), R1
	MOVW	new+12(FP), R2
#ifndef GOARM64_LSE
	MOVBU	internal∕cpu·ARM64+const_offsetARM64HasATOMICS(SB), R4
	CBZ 	R4, load_store_loop
#endif
	MOVD	R1, R3
	CASALW	R3, (R0), R2
	CMP 	R1, R3
	CSET	EQ, R0
	MOVB	R0, ret+16(FP)
	RET
#ifndef GOARM64_LSE
load_store_loop:
	LDAXRW	(R0), R3
	CMPW	R1, R3
	BNE	ok
	STLXRW	R2, (R0), R3

	case LDR_EQ, MSR_EQ, MRS_EQ:
		op = "MOVW" + op[3:] + suffix
	case VMRS_EQ, VMSR_EQ:
		op = "MOVW" + op[4:] + suffix
	case LDRB_EQ, UXTB_EQ:
		op = "MOVBU" + op[4:] + suffix
	case LDRSB_EQ:
		op = "MOVBS" + op[5:] + suffix
	case SXTB_EQ:
		op = "MOVBS" + op[4:] + suffix
	case LDRH_EQ, UXTH_EQ:
		op = "MOVHU" + op[4:] + suffix
	case LDRSH_EQ:
		op = "MOVHS" + op[5:] + suffix
	case SXTH_EQ:

		// conversions
		{name: "MOVBreg", argLength: 1, reg: gp11, asm: "MOVB"},   // move from arg0, sign-extended from byte
		{name: "MOVBUreg", argLength: 1, reg: gp11, asm: "MOVBU"}, // move from arg0, unsign-extended from byte
		{name: "MOVHreg", argLength: 1, reg: gp11, asm: "MOVH"},   // move from arg0, sign-extended from half

	BLT	X5, X6, check4_unaligned
compare8_unaligned:
	MOVBU	0(X10), X8
	MOVBU	1(X10), X15
	MOVBU	2(X10), X17
	MOVBU	3(X10), X19
	MOVBU	4(X10), X21
	MOVBU	5(X10), X23
	MOVBU	6(X10), X25
	MOVBU	7(X10), X29
	MOVBU	0(X12), X9
	MOVBU	1(X12), X16
	MOVBU	2(X12), X18
	MOVBU	3(X12), X20
	MOVBU	4(X12), X22
	MOVBU	5(X12), X24
	MOVBU	6(X12), X28
	MOVBU	7(X12), X30
	BNE	X8, X9, cmp1a
	BNE	X15, X16, cmp1b

TEXT runtime·memequal<ABIInternal>(SB),NOSPLIT|NOFRAME,$0-25
	BEQ	R4, R5, eq
	ADDV	R4, R6, R7
	PCALIGN	$16
loop:
	BNE	R4, R7, test
	MOVV	$1, R4
	RET
test:
	MOVBU	(R4), R9
	ADDV	$1, R4
	MOVBU	(R5), R10
	ADDV	$1, R5
	BEQ	R9, R10, loop

	MOVB    R0, R4
	RET
eq:
	MOVV	$1, R4
	RET

// memequal_varlen(a, b unsafe.Pointer) bool
TEXT runtime·memequal_varlen<ABIInternal>(SB),NOSPLIT,$40-17
	BEQ	R4, R5, eq