CMP     R11, $0
	BEQ     done

	MOVD    0(R8), R20
	ADD     $-1, R11
	MULLD   R9, R20, R6       // R6 = z0 = Low-order(x[i]*y)
	MULHDU  R9, R20, R7       // R7 = z1 = High-order(x[i]*y)
	ADDC    R4, R6            // R6 = z0 + r
	ADDZE   R7, R4            // R4 = z1 + CA
	CMP     R11, $0
	MOVD    R6, 0(R10)        // z[i]
	BEQ     done

	// We will read 4 elements per iteration

	RET

// memequal_varlen(a, b unsafe.Pointer) bool
TEXT runtime·memequal_varlen<ABIInternal>(SB),NOSPLIT,$40-17
	BEQ	R4, R5, eq
	MOVV	8(REGCTXT), R6    // compiler stores size at offset 8 in the closure
	MOVV	R4, 8(R3)
	MOVV	R5, 16(R3)
	MOVV	R6, 24(R3)
	JAL	runtime·memequal(SB)
	MOVBU	32(R3), R4
	RET
eq:
	MOVV	$1, R4

TEXT runtime·memclrNoHeapPointers<ABIInternal>(SB),NOSPLIT,$0-16
	ADDV	R4, R5, R6

	// if less than 8 bytes, do one byte at a time
	SGTU	$8, R5, R8
	BNE	R8, out

	// do one byte at a time until 8-aligned
	AND	$7, R4, R8
	BEQ	R8, words
	MOVB	R0, (R4)
	ADDV	$1, R4
	JMP	-4(PC)

words:
	// do 8 bytes at a time if there is room
	ADDV	$-7, R6, R5

	PCALIGN	$16
	SGTU	R5, R4, R8
	BEQ	R8, out
	MOVV	R0, (R4)

		ANDSW	$0x100, R13, ZR
		CSELW	NE, R14, R13, R13 // Fake modulo
		SUBS	$1, R8
		VMOV	V2.S[0], R0
		EORW	R0, R4
		EORW	R4, R5
		EORW	R5, R6
		EORW	R6, R7
		STPW.P	(R4, R5), 8(R10)
		STPW.P	(R6, R7), 8(R10)
	BNE	ks128Loop
	CBZ	R11, ksDone       // If dec is nil we are done
	SUB	$176, R10
	// Decryption keys are encryption keys with InverseMixColumns applied

	MOVD	callpc+8(FP), R5
	MOVD	pc+16(FP), R6
	// void __tsan_write_pc(ThreadState *thr, void *addr, void *callpc, void *pc);
	MOVD	$__tsan_write_pc(SB), R8
	BR	racecalladdr<>(SB)

// func runtime·RaceReadRange(addr, size uintptr)
// Called from instrumented Go code.
TEXT	runtime·racereadrange<ABIInternal>(SB), NOSPLIT, $0-16
	MOVD	R4, R5 // size
	MOVD	R3, R4 // addr
	MOVD	LR, R6

	LVX	(R3),byteswap
#endif

	CMPU	R5,$256		// length of bytes
	BLT	short

	RLDICR	$0,R5,$56,R6 // chunk to process

	// First step for larger sizes
l1:	MOVD	$32768,R7
	MOVD	R7,R9
	CMP	R6,R7   // compare R6, R7 (MAX SIZE)
	BGT	top	// less than MAX, just do remainder
	MOVD	R6,R7
top:
	SUB	R7,R6,R6

	// mainloop does 128 bytes at a time
	SRD	$7,R7

	genSize1 := (1 << 16) + 16
	genSize2 := (1 << 21) + 16

	fmt.Fprintln(buf, "TEXT f(SB),0,$0-0")
	fmt.Fprintln(buf, "BEQ R5, R6, label18")
	fmt.Fprintln(buf, "BNE R5, R6, label18")
	fmt.Fprintln(buf, "BGE R5, R6, label18")

	fmt.Fprintln(buf, "BGEU R5, R6, label18")
	fmt.Fprintln(buf, "BLTU R5, R6, label18")

	fmt.Fprintln(buf, "BLEZ R5, label18")
	fmt.Fprintln(buf, "BGEZ R5, label18")
	fmt.Fprintln(buf, "BLTZ R5, label18")

	MOVD	savedm-8(SP), R6
	CMP	R6, $0
	BNE	droppedm

	// Skip dropm to reuse it in the next call, when a pthread key has been created.
	MOVD	_cgo_pthread_key_created(SB), R6
	// It means cgo is disabled when _cgo_pthread_key_created is a nil pointer, need dropm.
	CMP	R6, $0
	BEQ	dropm
	MOVD	(R6), R6
	CMP	R6, $0
	BNE	droppedm

dropm:
	MOVD	$runtime·dropm(SB), R12

	MOVD	fn+0(FP), R0        // function code (encryption)
	MOVD	key_base+8(FP), R1  // key
	MOVD	buf_base+80(FP), R2 // counter values
	MOVD	dst_base+32(FP), R4 // dst
	MOVD	src_base+56(FP), R6 // src
	MOVD	src_len+64(FP), R7  // len
loop:
	KMCTR	R4, R2, R6          // cipher message with counter (KMCTR)
	BVS	loop                // branch back if interrupted
	RET
crash:
	MOVD	$0, (R0)
	RET

	// We get called to create new OS threads too, and those
	// come in on the m->g0 stack already.
	MOVV	g_m(g), R5
	MOVV	m_gsignal(R5), R6
	BEQ	R6, g, g0
	MOVV	m_g0(R5), R6
	BEQ	R6, g, g0

	JAL	gosave_systemstack_switch<>(SB)
	MOVV	R6, g
	JAL	runtime·save_g(SB)
	MOVV	(g_sched+gobuf_sp)(g), R3

	// Now on a scheduling stack (a pthread-created stack).
g0: