* The standard x86-64 ABI passes the three arguments m, g, fn
 * in %rdi, %rsi, %rdx.
 */
.globl EXT(crosscall1)
EXT(crosscall1):
	pushq %rbx
	pushq %rbp
	pushq %r12
	pushq %r13
	pushq %r14
	pushq %r15

#if defined(_WIN64)
	movq %r8, %rdi	/* arg of setg_gcc */
	call *%rdx	/* setg_gcc */
	call *%rcx	/* fn */
#else
	movq %rdi, %rbx
	movq %rdx, %rdi	/* arg of setg_gcc */

		return
	}

	// This implementation expects the following function prologue layout:
	// - Stack split code (optional)
	// - PUSHQ	BP
	// - MOVQ	SP,	BP
	//
	// If the prologue layout change, the unwind information should be updated
	// accordingly.

	// Search for the PUSHQ BP instruction inside the prologue.
	var pushbp *obj.Prog
	for p := s.Func().Text; p != nil; p = p.Link {

	MOVQ	g(R12), R14
	MOVQ	g_m(R14), R13
	MOVQ	m_g0(R13), R15
	CMPQ	R13, R15
	JEQ	noswitch	// branch if already on g0
	MOVQ	R15, g(R12)	// g = m->g0
	MOVQ	R15, R14	// set g register
	PUSHQ	RARG1	// func arg
	PUSHQ	RARG0	// func arg
	CALL	runtime·racecallback(SB)
	POPQ	R12
	POPQ	R12
	// All registers are smashed after Go code, reload.
	get_tls(R12)
	MOVQ	g(R12), R13
	MOVQ	g_m(R13), R13

	MOVQ	a1+8(FP), DI
	MOVQ	a2+16(FP), SI
	MOVQ	a3+24(FP), DX
	MOVQ	$0, R10
	MOVQ	$0, R8
	MOVQ	$0, R9
	MOVQ	trap+0(FP), AX	// syscall entry
	POPQ	R12 // preserve return address
	SYSCALL
	PUSHQ	R12
	CMPQ	AX, $0xfffffffffffff001
	JLS	ok2
	MOVQ	$-1, r1+32(FP)
	NEGQ	AX
	MOVQ	AX, err+40(FP)
	RET
ok2:
	MOVQ	AX, r1+32(FP)
	MOVQ	$0, err+40(FP)
	RET

	log.Fatalf("unexpected relocation variant")
	return -1
}

func elfsetupplt(ctxt *ld.Link, ldr *loader.Loader, plt, got *loader.SymbolBuilder, dynamic loader.Sym) {
	if plt.Size() == 0 {
		// pushq got+8(IP)
		plt.AddUint8(0xff)

		plt.AddUint8(0x35)
		plt.AddPCRelPlus(ctxt.Arch, got.Sym(), 8)

		// jmpq got+16(IP)
		plt.AddUint8(0xff)

		plt.AddUint8(0x25)

		}
	}
	lSSE := layout{stack: l.stack, sp: "SP"}
	for _, reg := range regNamesAMD64 {
		if strings.HasPrefix(reg, "X") {
			lSSE.add("MOVUPS", reg, 16)
		}
	}

	// TODO: MXCSR register?

	p("PUSHQ BP")
	p("MOVQ SP, BP")
	p("// Save flags before clobbering them")
	p("PUSHFQ")
	p("// obj doesn't understand ADD/SUB on SP, but does understand ADJSP")
	p("ADJSP $%d", lSSE.stack)

	MOVQ	AX, 8(SP)
	CALL	runtime·args(SB)
	CALL	runtime·osinit(SB)
	CALL	runtime·schedinit(SB)

	// create a new goroutine to start program
	MOVQ	$runtime·mainPC(SB), AX		// entry
	PUSHQ	AX
	CALL	runtime·newproc(SB)
	POPQ	AX

	// start this M
	CALL	runtime·mstart(SB)

	CALL	runtime·abort(SB)	// mstart should never return
	RET