// 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:

// It's such a pointer chain: _crosscall2_ptr -> x_crosscall2_ptr -> crosscall2
TEXT ·set_crosscall2(SB),NOSPLIT,$0-0
	MOVD	_crosscall2_ptr(SB), R5
	MOVD	CROSSCALL2_FPTR, R6
	MOVD	R6, (R5)
	RET

TEXT crosscall2_trampoline<>(SB),NOSPLIT,$0-0
	JMP	crosscall2(SB)

// Called by C code generated by cmd/cgo.
// func crosscall2(fn, a unsafe.Pointer, n int32, ctxt uintptr)

// Use a local trampoline, to avoid taking the address of a dynamically exported
// function.
TEXT ·set_crosscall2(SB),NOSPLIT,$0-0
	MOVV	_crosscall2_ptr(SB), R5
	MOVV	$crosscall2_trampoline<>(SB), R6
	MOVV	R6, (R5)
	RET

TEXT crosscall2_trampoline<>(SB),NOSPLIT,$0-0
	JMP	crosscall2(SB)

// Called by C code generated by cmd/cgo.
// func crosscall2(fn, a unsafe.Pointer, n int32, ctxt uintptr)

	{"R2", "R2"},
	{"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)"},

//

// func rawVforkSyscall(trap, a1, a2, a3 uintptr) (r1, err uintptr)
TEXT ·rawVforkSyscall(SB),NOSPLIT|NOFRAME,$0-48
	MOVD	a1+8(FP), R3
	MOVD	a2+16(FP), R4
	MOVD	a3+24(FP), R5
	MOVD	R0, R6
	MOVD	R0, R7
	MOVD	R0, R8
	MOVD	trap+0(FP), R9	// syscall entry
	SYSCALL R9
	BVC	ok
	MOVD	$-1, R4
	MOVD	R4, r1+32(FP)	// r1
	MOVD	R3, err+40(FP)	// errno
	RET
ok:
	MOVD	R3, r1+32(FP)	// r1

	MOVD	nr+0(FP), R6   // Round count/Key size
	MOVD	xk+8(FP), R5   // Key pointer
	MOVD	dst+16(FP), R3 // Dest pointer
	MOVD	src+24(FP), R4 // Src pointer
#ifdef NEEDS_ESPERM
	MOVD	$·rcon(SB), R7
	LVX	(R7), ESPERM   // Permute value for P8_ macros.
#endif

	// Set CR{1,2,3}EQ to hold the key size information.
	CMPU	R6, $10, CR1
	CMPU	R6, $12, CR2
	CMPU	R6, $14, CR3

	MOVD	$16, R6

	VPERM  vreg, vreg, SWAP, vreg

GLOBL byteswap<>+0(SB), RODATA, $16

TEXT ·Index<ABIInternal>(SB),NOSPLIT|NOFRAME,$0-56
	// R3 = byte array pointer
	// R4 = length
	MOVD R6, R5             // R5 = separator pointer
	MOVD R7, R6             // R6 = separator length

#ifdef GOARCH_ppc64le
	MOVBZ internal∕cpu·PPC64+const_offsetPPC64HasPOWER9(SB), R7
	CMP   R7, $1
	BNE   power8
	BR    indexbodyp9<>(SB)
#endif
power8:

	FCMPU   F0, F1
	BEQ     atanIsZero

	MOVD	$·atanrodataL8<>+0(SB), R5
	MOVH	$0x3FE0, R3
	LGDR	F0, R1
	RISBGNZ	$32, $63, $32, R1, R1
	RLL	$16, R1, R2
	ANDW	$0x7FF0, R2
	MOVW	R2, R6
	MOVW	R3, R7
	CMPUBLE	R6, R7, L6
	MOVD	$·atanxmone<>+0(SB), R3
	FMOVD	0(R3), F2
	WFDDB	V0, V2, V0
	RISBGZ	$63, $63, $33, R1, R1
	MOVD	$·atanxpi2h<>+0(SB), R3
	MOVWZ	R1, R1
	SLD	$3, R1, R1

	// R3 = ptr
	// R4 = n

	// Determine if there are doublewords to clear
check:
	ANDCC $7, R4, R5  // R5: leftover bytes to clear
	SRD   $3, R4, R6  // R6: double words to clear
	CMP   R6, $0, CR1 // CR1[EQ] set if no double words

	BC    12, 6, nozerolarge // only single bytes
	CMP   R4, $512
	BLT   under512           // special case for < 512

	MOVW	num+0(FP), R7	// syscall entry
	MOVW	a1+4(FP), R0
	MOVW	a2+8(FP), R1
	MOVW	a3+12(FP), R2
	MOVW	a4+16(FP), R3
	MOVW	a5+20(FP), R4
	MOVW	a6+24(FP), R5
	SWI	$0
	MOVW	$0xfffff001, R6
	CMP	R6, R0
	BLS	ok
	MOVW	$-1, R1
	MOVW	R1, r1+28(FP)
	MOVW	$0, R2
	MOVW	R2, r2+32(FP)
	RSB	$0, R0, R0
	MOVW	R0, errno+36(FP)
	RET
ok:
	MOVW	R0, r1+28(FP)
	MOVW	R1, r2+32(FP)
	MOVW	$0, R0