ADCW ZR, R8, R10                           // 0a011f1a
	ADC R0, R2, R12                            // 4c00009a
	ADCSW R9, R21, R6                          // a602093a
	ADCS R23, R22, R22                         // d60217ba
	ADDW R5.UXTH, R8, R9                       // 0921250b
	ADD R8.SXTB<<3, R23, R14                   // ee8e288b

	VPCONFLICTD 15(R8), K6, X6                         // 62d27d0ec4b00f000000
	VPCONFLICTD (BP), K6, X6                           // 62f27d0ec47500
	VPCONFLICTD X6, K6, X17                            // 62e27d0ec4ce
	VPCONFLICTD X1, K6, X17                            // 62e27d0ec4c9
	VPCONFLICTD X8, K6, X17                            // 62c27d0ec4c8
	VPCONFLICTD 15(R8), K6, X17                        // 62c27d0ec4880f000000

	RET

gcm_ghash_p8_4x:
	LVSL     (R8)(R0), T0      // 0x0001..0e0f
	MOVD     $0x70, R8
	LXVD2X   (HTBL)(R9), VH2
	MOVD     $0x80, R9
	VSPLTISB $8, T1            // 0x0808..0808
	MOVD     $0x90, R10
	LXVD2X   (HTBL)(R8), VH3L  // load H^3
	MOVD     $0xa0, R8
	LXVD2X   (HTBL)(R9), VH3
	MOVD     $0xb0, R9
	LXVD2X   (HTBL)(R10), VH3H
	MOVD     $0xc0, R10
	LXVD2X   (HTBL)(R8), VH4L  // load H^4
	MOVD     $0x10, R8

	MOVW	12(R13), R0
	MOVW	$1234, R1
	CMP	R0, R1
	BEQ	2(PC)
	BL	runtime·abort(SB)

	MOVW	0(R13), R8    // m
	MOVW	4(R13), R0    // g

	CMP	$0, R8
	BEQ	nog
	CMP	$0, R0
	BEQ	nog

	MOVW	R0, g
	MOVW	R8, g_m(g)

	// paranoia; check they are not nil
	MOVW	0(R8), R0
	MOVW	0(g), R0

	BL	runtime·emptyfunc(SB)	// fault if stack check is wrong

	{"R3", "R3"},
	{"R4", "R4"},
	{"R(4)", "R4"},
	{"R5", "R5"},
	{"R6", "R6"},
	{"R7", "R7"},
	{"R8", "R8"},
	{"[R0,R1,g,R15]", "[R0,R1,g,R15]"},
	{"[R0-R7]", "[R0,R1,R2,R3,R4,R5,R6,R7]"},
	{"[R(0)-R(7)]", "[R0,R1,R2,R3,R4,R5,R6,R7]"},
	{"[R0]", "[R0]"},
	{"[R1-R12]", "[R1,R2,R3,R4,R5,R6,R7,R8,R9,g,R11,R12]"},
	{"armCAS64(SB)", "armCAS64(SB)"},
	{"asmcgocall<>(SB)", "asmcgocall<>(SB)"},
	{"c+28(FP)", "c+28(FP)"},

	VPDPWSSD 15(R8)(R14*4), Y2, K3, Y14                // 62126d2b52b4b00f000000
	VPDPWSSD -7(CX)(DX*4), Y2, K3, Y14                 // 62726d2b52b491f9ffffff
	VPDPWSSD Y8, Y22, K3, Y14                          // 62524d2352f0
	VPDPWSSD Y9, Y22, K3, Y14                          // 62524d2352f1
	VPDPWSSD Y22, Y22, K3, Y14                         // 62324d2352f6

	MOVW	m_gsignal(R8), R4
	CMP	g, R4
	BNE	3(PC)
	BL	runtime·badmorestackgsignal(SB)
	B	runtime·abort(SB)

	// Called from f.
	// Set m->morebuf to f's caller.
	MOVW	R3, (m_morebuf+gobuf_pc)(R8)	// f's caller's PC
	MOVW	R13, (m_morebuf+gobuf_sp)(R8)	// f's caller's SP
	MOVW	g, (m_morebuf+gobuf_g)(R8)

	// Call newstack on m->g0's stack.
	MOVW	m_g0(R8), R0
	BL	setg<>(SB)

	SYSCALL $SYS_gettid

	MOVD	-24(R1), R12       // fn
	MOVD	-16(R1), R8        // g
	MOVD	-8(R1), R7         // m

	CMP	R7, $0
	BEQ	nog
	CMP	R8, $0
	BEQ	nog

	MOVD	R3, m_procid(R7)

	// TODO: setup TLS.

	// In child, set up new stack
	MOVD	R7, g_m(R8)
	MOVD	R8, g
	//CALL	runtime·stackcheck(SB)

nog:
	// Call fn
	MOVD	R12, CTR
	BL	(CTR)

	MOVD	R15, (g_sched+gobuf_sp)(g)
	MOVD	LR, R8
	MOVD	R8, (g_sched+gobuf_pc)(g)
	MOVD	R5, (g_sched+gobuf_lr)(g)
	MOVD	R12, (g_sched+gobuf_ctxt)(g)

	// Cannot grow scheduler stack (m->g0).
	MOVD	g_m(g), R7
	MOVD	m_g0(R7), R8
	CMPBNE	g, R8, 3(PC)
	BL	runtime·badmorestackg0(SB)
	BL	runtime·abort(SB)

	// Cannot grow signal stack (m->gsignal).
	MOVD	m_gsignal(R7), R8
	CMP	g, R8
	BNE	3(PC)

	SYSCALL
	JCC	2(PC)
	NEGQ	AX
	MOVL	AX, ret+24(FP)
	RET

TEXT runtime·lwp_tramp(SB),NOSPLIT,$0

	// Set FS to point at m->tls.
	LEAQ	m_tls(R8), DI
	CALL	runtime·settls(SB)

	// Set up new stack.
	get_tls(CX)
	MOVQ	R8, g_m(R9)
	MOVQ	R9, g(CX)
	CALL	runtime·stackcheck(SB)

	// Call fn. This is an ABI0 PC.
	CALL	R12

	// It shouldn't return. If it does, exit.
	MOVL	$SYS__lwp_exit, AX