//
// These macros assume a 16 byte aligned stack pointer. This
// is required by ELFv1, ELFv2, and AIX PPC64.

#define SAVE_GPR_SIZE (18*8)
#define SAVE_GPR(offset)               \
	MOVD	R14, (offset+8*0)(R1)  \
	MOVD	R15, (offset+8*1)(R1)  \
	MOVD	R16, (offset+8*2)(R1)  \
	MOVD	R17, (offset+8*3)(R1)  \
	MOVD	R18, (offset+8*4)(R1)  \
	MOVD	R19, (offset+8*5)(R1)  \
	MOVD	R20, (offset+8*6)(R1)  \
	MOVD	R21, (offset+8*7)(R1)  \

	MOVD	p_len+16(FP), R5

	// We assume p_len >= 64
	SRD 	$6, R5
	MOVD	R5, CTR

	MOVWZ	0(R10), R22
	MOVWZ	4(R10), R3
	MOVWZ	8(R10), R4
	MOVWZ	12(R10), R5

loop:
	MOVD	R22, R14
	MOVD	R3, R15
	MOVD	R4, R16
	MOVD	R5, R17

	ENDIAN_MOVE( 0,R6,M00,M15)
	ENDIAN_MOVE( 4,R6,M01,M15)
	ENDIAN_MOVE( 8,R6,M02,M15)
	ENDIAN_MOVE(12,R6,M03,M15)

	ROUND1(R22,R3,R4,R5,M00,0xd76aa478, 7);

// Called on a system stack, and there may be no g yet (during needm).
TEXT ·asmcgocall_no_g(SB),NOSPLIT,$0-16
	MOVD	fn+0(FP), R3
	MOVD	arg+8(FP), R4

	MOVD	R1, R15
	SUB	$(asmcgocallSaveOffset+8), R1
	RLDCR	$0, R1, $~15, R1	// 16-byte alignment for gcc ABI
	MOVD	R15, asmcgocallSaveOffset(R1)

	MOVD	R0, 0(R1)	// clear back chain pointer (TODO can we give it real back trace information?)

func (c *sigctxt) r12() uint64 { return c.regs().__gregs[_REG_R12] }
func (c *sigctxt) r13() uint64 { return c.regs().__gregs[_REG_R13] }
func (c *sigctxt) r14() uint64 { return c.regs().__gregs[_REG_R14] }
func (c *sigctxt) r15() uint64 { return c.regs().__gregs[_REG_R15] }

//go:nosplit
//go:nowritebarrierrec
func (c *sigctxt) rip() uint64 { return c.regs().__gregs[_REG_RIP] }

//	LMOVB rreg ',' addr
//	{
//		outcode(int($1), &$2, 0, &$4);
//	}
	MOVB	R1, foo<>+3(SB)
	MOVB	R5, -18(R4)	// a085ffee
	MOVB	R10, 9(R13)	// a1aa0009
	MOVB	R15, (R13)	// a1af0000
	MOVBU	R5, -18(R4)	// a085ffee
	MOVBU	R10, 9(R13)	// a1aa0009
	MOVBU	R15, (R13)	// a1af0000

//
// store floats
//
//	LMOVW freg ',' addr
//	{
//		outcode(int($1), &$2, 0, &$4);
//	}
	MOVD	F1, foo<>+3(SB)

	MOVD	(13*4+8*6)(R13), F13
	MOVD	(13*4+8*7)(R13), F14
	MOVD	(13*4+8*8)(R13), F15

skipfprest:
	MOVW.P	4(R13), R14
	MOVM.IAW	(R13), [R0, R1, R3, R4, R5, R6, R7, R8, R9, g, R11, R12]
	ADD	$(8*9), R13

func (c *sigctxt) r13() uint64 { return uint64(c.regs().gregs[_REG_R13]) }
func (c *sigctxt) r14() uint64 { return uint64(c.regs().gregs[_REG_R14]) }
func (c *sigctxt) r15() uint64 { return uint64(c.regs().gregs[_REG_R15]) }

//go:nosplit
//go:nowritebarrierrec
func (c *sigctxt) rip() uint64 { return uint64(c.regs().gregs[_REG_RIP]) }

	// To defeat the linker, we make our own stack frame with
	// more space.
	SUB	$144+FIXED_FRAME, R1

	// Save registers
	MOVD	R31, 56(R1)
	MOVD	g, 64(R1)
	MOVD	R29, 72(R1)
	MOVD	R14, 80(R1)
	MOVD	R15, 88(R1)

	BL	runtime·load_g(SB)

	CMP	$0, g
	BEQ	sigtramp // g == nil
	MOVD	g_m(g), R6
	CMP	$0, R6
	BEQ	sigtramp	// g.m == nil

	// Save m->libcall. We need to do this because we

func (c *sigctxt) r12() uint64 { return c.regs().sc_regs[12] }
func (c *sigctxt) r13() uint64 { return c.regs().sc_regs[13] }
func (c *sigctxt) r14() uint64 { return c.regs().sc_regs[14] }
func (c *sigctxt) r15() uint64 { return c.regs().sc_regs[15] }
func (c *sigctxt) r16() uint64 { return c.regs().sc_regs[16] }
func (c *sigctxt) r17() uint64 { return c.regs().sc_regs[17] }
func (c *sigctxt) r18() uint64 { return c.regs().sc_regs[18] }

func (c *sigctxt) r12() uint64 { return c.regs().sc_regs[12] }
func (c *sigctxt) r13() uint64 { return c.regs().sc_regs[13] }
func (c *sigctxt) r14() uint64 { return c.regs().sc_regs[14] }
func (c *sigctxt) r15() uint64 { return c.regs().sc_regs[15] }
func (c *sigctxt) r16() uint64 { return c.regs().sc_regs[16] }
func (c *sigctxt) r17() uint64 { return c.regs().sc_regs[17] }
func (c *sigctxt) r18() uint64 { return c.regs().sc_regs[18] }