// Push the LR to stack, as we'll clobber it in order to
	// push the call. The function being pushed is responsible
	// for restoring the LR and setting the SP back.
	// This extra slot is known to gentraceback.
	sp := c.sp() - 4
	c.set_sp(sp)
	*(*uint32)(unsafe.Pointer(uintptr(sp))) = c.lr()
	// Set up PC and LR to pretend the function being signaled
	// calls targetPC at resumePC.
	c.set_lr(uint32(resumePC))
	c.set_pc(uint32(targetPC))

	MOV	$8(SP), (X5)			// ERROR "address load must target register"
	MOVB	$8(SP), X5			// ERROR "unsupported address load"
	MOVH	$8(SP), X5			// ERROR "unsupported address load"
	MOVW	$8(SP), X5			// ERROR "unsupported address load"
	MOVF	$8(SP), X5			// ERROR "unsupported address load"
	MOV	$1234, 0(SP)			// ERROR "constant load must target register"
	MOV	$1234, 8(SP)			// ERROR "constant load must target register"

	increment(5)
	MOVOU T0, (8*16 + 6*16)(SP)
	increment(6)
	MOVOU T0, (8*16 + 7*16)(SP)
	increment(7)

	MOVOU (8*16 + 0*16)(SP), B0
	MOVOU (8*16 + 1*16)(SP), B1
	MOVOU (8*16 + 2*16)(SP), B2
	MOVOU (8*16 + 3*16)(SP), B3
	MOVOU (8*16 + 4*16)(SP), B4
	MOVOU (8*16 + 5*16)(SP), B5
	MOVOU (8*16 + 6*16)(SP), B6
	MOVOU (8*16 + 7*16)(SP), B7

	aesRound(1)
	increment(0)
	aesRound(2)
	increment(1)
	aesRound(3)

//
// The stack holds the intermediate word array - 16 uint32s - at 0(SP) up to 64(SP).
// The saved a, b, c, d, e (R11 through R15 on amd64) are at 64(SP) up to 84(SP).
// The saved limit pointer (DI on amd64) is at 84(SP).
// The saved data pointer (SI on amd64) is at 88(SP).

#define LOAD(index, e) \
	MOVL	88(SP), SI; \
	MOVL	(index*4)(SI), DI; \
	BSWAPL	DI; \
	MOVL	DI, (index*4)(SP); \
	ADDL	DI, e

#define SHUFFLE(index, e) \

TEXT runtime·morestack(SB), NOSPLIT, $0-0
	// R1 = g.m
	MOVD g_m(g), R1

	// R2 = g0
	MOVD m_g0(R1), R2

	// Set g->sched to context in f.
	NOP	SP	// tell vet SP changed - stop checking offsets
	MOVD 0(SP), g_sched+gobuf_pc(g)
	MOVD $8(SP), g_sched+gobuf_sp(g) // f's SP
	MOVD CTXT, g_sched+gobuf_ctxt(g)

	// Cannot grow scheduler stack (m->g0).
	Get g
	Get R2
	I64Eq
	If

// adjust Gobuf as if it executed a call to fn with context ctxt
// and then stopped before the first instruction in fn.
func gostartcall(buf *gobuf, fn, ctxt unsafe.Pointer) {
	sp := buf.sp
	sp -= goarch.PtrSize
	*(*uintptr)(unsafe.Pointer(sp)) = buf.pc
	buf.sp = sp
	buf.pc = uintptr(fn)
	buf.ctxt = ctxt

	var l = layout{sp: "SP"}
	for _, reg := range regNames386 {
		if reg == "SP" || strings.HasPrefix(reg, "X") {
			continue
		}
		l.add("MOVL", reg, 4)
	}

	softfloat := "GO386_softfloat"

	// Save SSE state only if supported.
	lSSE := layout{stack: l.stack, sp: "SP"}
	for i := 0; i < 8; i++ {
		lSSE.add("MOVUPS", fmt.Sprintf("X%d", i), 16)
	}

	p("ADJSP $%d", lSSE.stack)
	p("NOP SP")
	l.save()

// func runtime·doasanread(addr unsafe.Pointer, sz, sp, pc uintptr)
TEXT	runtime·doasanread(SB), NOSPLIT, $0-32
	MOVV	addr+0(FP), RARG0
	MOVV	sz+8(FP), RARG1
	MOVV	sp+16(FP), RARG2
	MOVV	pc+24(FP), RARG3
	// void __asan_read_go(void *addr, uintptr_t sz, void *sp, void *pc);
	MOVV	$__asan_read_go(SB), FARG
	JMP	asancall<>(SB)

// func runtime·doasanwrite(addr unsafe.Pointer, sz, sp, pc uintptr)
TEXT	runtime·doasanwrite(SB), NOSPLIT, $0-32

// func runtime·doasanread(addr unsafe.Pointer, sz, sp, pc uintptr)
TEXT	runtime·doasanread(SB),NOSPLIT|NOFRAME,$0-32
	MOVD	addr+0(FP), RARG0
	MOVD	sz+8(FP), RARG1
	MOVD	sp+16(FP), RARG2
	MOVD	pc+24(FP), RARG3
	// void __asan_read_go(void *addr, uintptr_t sz, void *sp, void *pc);
	MOVD	$__asan_read_go(SB), FARG
	BR	asancall<>(SB)

// func runtime·doasanwrite(addr unsafe.Pointer, sz, sp, pc uintptr)

	return dst == src &&
		r.Overlaps(r.Add(sp.Sub(r.Min))) &&
		(sp.Y < r.Min.Y || (sp.Y == r.Min.Y && sp.X < r.Min.X))
}

// Draw calls [DrawMask] with a nil mask.
func Draw(dst Image, r image.Rectangle, src image.Image, sp image.Point, op Op) {
	DrawMask(dst, r, src, sp, nil, image.Point{}, op)
}