// index 0 in the heap arena map.
	//
	// On amd64, the address space is 48 bits, sign extended to 64
	// bits. This offset lets us handle "negative" addresses (or
	// high addresses if viewed as unsigned).
	//
	// On aix/ppc64, this offset allows to keep the heapAddrBits to
	// 48. Otherwise, it would be 60 in order to handle mmap addresses
	// (in range 0x0a00000000000000 - 0x0afffffffffffff). But in this

	fFunc := runtime.FuncForPC(uintptr(abi.FuncPCABIInternal(f)))
	if fFunc == nil || fFunc.Entry() == 0 {
		panic("failed to locate function entry")
	}

	for offset := 0; offset < maxBytes; offset++ {
		innerPC := fFunc.Entry() + uintptr(offset)
		inner := runtime.FuncForPC(innerPC)
		if inner == nil {
			// No function known for this PC value.
			// It might simply be misaligned, so keep searching.
			continue
		}

	needsUnsafe := false
	// Walk backward so that in C.f1(C.f2()) we rewrite C.f2 first.
	for _, call := range f.Calls {
		if call.Done {
			continue
		}
		start := f.offset(call.Call.Pos())
		end := f.offset(call.Call.End())
		str, nu := p.rewriteCall(f, call)
		if str != "" {
			f.Edit.Replace(start, end, str)
			if nu {
				needsUnsafe = true
			}
		}
	}
	return needsUnsafe
}

	}
	if o := reflect.TypeOf(&signed).Elem().Field(1).Offset; o != 8 {
		t.Fatalf("Int64 offset = %d, want 8", o)
	}
	if p := reflect.ValueOf(&signed).Elem().Field(1).Addr().Pointer(); p&7 != 0 {
		t.Fatalf("Int64 pointer = %#x, want 8-aligned", p)
	}

	var unsigned struct {
		_ uint32
		i Uint64
	}
	if o := reflect.TypeOf(&unsigned).Elem().Field(1).Offset; o != 8 {
		t.Fatalf("Uint64 offset = %d, want 8", o)

	if p == nil {
		return
	}

	liveInfo := funcdata(f, abi.FUNCDATA_ArgLiveInfo)
	liveIdx := pcdatavalue(f, abi.PCDATA_ArgLiveIndex, pc)
	startOffset := uint8(0xff) // smallest offset that needs liveness info (slots with a lower offset is always live)
	if liveInfo != nil {
		startOffset = *(*uint8)(liveInfo)
	}

	isLive := func(off, slotIdx uint8) bool {
		if liveInfo == nil || liveIdx <= 0 {

build:rbe_linux_cpu --copt=-Wno-error=unused-command-line-argument
# This was added in clang-16 by https://reviews.llvm.org/D133574.
# Can be removed once upb is updated, since a type definition is used within
# offset of in the current version of ubp.
# See https://github.com/protocolbuffers/upb/blob/9effcbcb27f0a665f9f345030188c0b291e32482/upb/upb.c#L183.
build:rbe_linux_cpu --copt=-Wno-gnu-offsetof-extensions

// Indexed ops generate indexed load or store instructions for all GOPPC64 values.
// Non-indexed ops generate DS-form loads and stores when the offset fits in 16 bits,
// and on power8 and power9, a multiple of 4 is required for MOVW and MOVD ops.
// On power10, prefixed loads and stores can be used for offsets > 16 bits and <= 32 bits.
// and support for PC relative addressing must be available if relocation is needed.

//              mean * scale * rsqrt(variance + epsilon)
// CHECK:  %[[MUL3:.*]] = "tf.Mul"(%[[ARG3:.*]], %[[MUL1]])
//              offset - mean * scale * rsqrt(variance + epsilon)
// CHECK:  %[[SUB:.*]] = "tf.Sub"(%[[ARG2:.*]], %[[MUL3]])
//              x * scale * rsqrt(variance + epsilon) +
//              offset - mean * scale * rsqrt(variance + epsilon)
// CHECK:  %[[ADD2:.*]] = "tf.Add"(%[[MUL2]], %[[SUB]])

		// duffzero
		// arg0 = address of memory to zero
		// arg1 = mem
		// auxint = offset into duffzero code to start executing
		// returns mem
		// R20 changed as side effect
		// R16 and R17 may be clobbered by linker trampoline.
		{
			name:      "DUFFZERO",
			aux:       "Int64",

	if length < 0 {
		length = fi.Size - startOffset
	}

	// Reply back invalid range if the input offset and length fall out of range.
	if startOffset > fi.Size || startOffset+length > fi.Size {
		return InvalidRange{startOffset, length, fi.Size}
	}

	// Get start part index and offset.
	partIndex, partOffset, err := fi.ObjectToPartOffset(ctx, startOffset)
	if err != nil {