src/cmd/compile/internal/types/size.go - go - Git at Google

 // Copyright 2009 The Go Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style
 // license that can be found in the LICENSE file.

 package types

 import (
 	"bytes"
 	"fmt"
 	"sort"

 	"cmd/compile/internal/base"
 	"cmd/internal/src"
 )

 var PtrSize int

 var RegSize int

 // Slices in the runtime are represented by three components:
 //
 // type slice struct {
 // 	ptr unsafe.Pointer
 // 	len int
 // 	cap int
 // }
 //
 // Strings in the runtime are represented by two components:
 //
 // type string struct {
 // 	ptr unsafe.Pointer
 // 	len int
 // }
 //
 // These variables are the offsets of fields and sizes of these structs.
 var (
 	SlicePtrOffset int64
 	SliceLenOffset int64
 	SliceCapOffset int64

 	SliceSize  int64
 	StringSize int64
 )

 var SkipSizeForTracing bool

 // typePos returns the position associated with t.
 // This is where t was declared or where it appeared as a type expression.
 func typePos(t *Type) src.XPos {
 	if pos := t.Pos(); pos.IsKnown() {
 		return pos
 	}
 	base.Fatalf("bad type: %v", t)
 	panic("unreachable")
 }

 // MaxWidth is the maximum size of a value on the target architecture.
 var MaxWidth int64

 // CalcSizeDisabled indicates whether it is safe
 // to calculate Types' widths and alignments. See CalcSize.
 var CalcSizeDisabled bool

 // machine size and rounding alignment is dictated around
 // the size of a pointer, set in betypeinit (see ../amd64/galign.go).
 var defercalc int

 func Rnd(o int64, r int64) int64 {
 	if r < 1 || r > 8 || r&(r-1) != 0 {
 		base.Fatalf("rnd %d", r)
 	}
 	return (o + r - 1) &^ (r - 1)
 }

 // expandiface computes the method set for interface type t by
 // expanding embedded interfaces.
 func expandiface(t *Type) {
 	seen := make(map[*Sym]*Field)
 	var methods []*Field

 	addMethod := func(m *Field, explicit bool) {
 		switch prev := seen[m.Sym]; {
 		case prev == nil:
 			seen[m.Sym] = m
 		case AllowsGoVersion(t.Pkg(), 1, 14) && !explicit && Identical(m.Type, prev.Type):
 			return
 		default:
 			base.ErrorfAt(m.Pos, "duplicate method %s", m.Sym.Name)
 		}
 		methods = append(methods, m)
 	}

 	{
 		methods := t.Methods().Slice()
 		sort.SliceStable(methods, func(i, j int) bool {
 			mi, mj := methods[i], methods[j]

 			// Sort embedded types by type name (if any).
 			if mi.Sym == nil && mj.Sym == nil {
 				return mi.Type.Sym().Less(mj.Type.Sym())
 			}

 			// Sort methods before embedded types.
 			if mi.Sym == nil || mj.Sym == nil {
 				return mi.Sym != nil
 			}

 			// Sort methods by symbol name.
 			return mi.Sym.Less(mj.Sym)
 		})
 	}

 	for _, m := range t.Methods().Slice() {
 		if m.Sym == nil {
 			continue
 		}

 		CheckSize(m.Type)
 		addMethod(m, true)
 	}

 	for _, m := range t.Methods().Slice() {
 		if m.Sym != nil || m.Type == nil {
 			continue
 		}

 		if m.Type.IsUnion() {
 			continue
 		}

 		// In 1.18, embedded types can be anything. In Go 1.17, we disallow
 		// embedding anything other than interfaces.
 		if !m.Type.IsInterface() {
 			if AllowsGoVersion(t.Pkg(), 1, 18) {
 				continue
 			}
 			base.ErrorfAt(m.Pos, "interface contains embedded non-interface, non-union %v", m.Type)
 			m.SetBroke(true)
 			t.SetBroke(true)
 			// Add to fields so that error messages
 			// include the broken embedded type when
 			// printing t.
 			// TODO(mdempsky): Revisit this.
 			methods = append(methods, m)
 			continue
 		}

 		// Embedded interface: duplicate all methods
 		// (including broken ones, if any) and add to t's
 		// method set.
 		for _, t1 := range m.Type.AllMethods().Slice() {
 			f := NewField(m.Pos, t1.Sym, t1.Type)
 			addMethod(f, false)

 			// Clear position after typechecking, for consistency with types2.
 			f.Pos = src.NoXPos
 		}

 		// Clear position after typechecking, for consistency with types2.
 		m.Pos = src.NoXPos
 	}

 	sort.Sort(MethodsByName(methods))

 	if int64(len(methods)) >= MaxWidth/int64(PtrSize) {
 		base.ErrorfAt(typePos(t), "interface too large")
 	}
 	for i, m := range methods {
 		m.Offset = int64(i) * int64(PtrSize)
 	}

 	t.SetAllMethods(methods)
 }

 func calcStructOffset(errtype *Type, t *Type, o int64, flag int) int64 {
 	// flag is 0 (receiver), 1 (actual struct), or RegSize (in/out parameters)
 	isStruct := flag == 1
 	starto := o
 	maxalign := int32(flag)
 	if maxalign < 1 {
 		maxalign = 1
 	}
 	lastzero := int64(0)
 	for _, f := range t.Fields().Slice() {
 		if f.Type == nil {
 			// broken field, just skip it so that other valid fields
 			// get a width.
 			continue
 		}

 		CalcSize(f.Type)
 		if int32(f.Type.Align) > maxalign {
 			maxalign = int32(f.Type.Align)
 		}
 		if f.Type.Align > 0 {
 			o = Rnd(o, int64(f.Type.Align))
 		}
 		if isStruct { // For receiver/args/results, do not set, it depends on ABI
 			f.Offset = o
 		}

 		w := f.Type.Width
 		if w < 0 {
 			base.Fatalf("invalid width %d", f.Type.Width)
 		}
 		if w == 0 {
 			lastzero = o
 		}
 		o += w
 		maxwidth := MaxWidth
 		// On 32-bit systems, reflect tables impose an additional constraint
 		// that each field start offset must fit in 31 bits.
 		if maxwidth < 1<<32 {
 			maxwidth = 1<<31 - 1
 		}
 		if o >= maxwidth {
 			base.ErrorfAt(typePos(errtype), "type %L too large", errtype)
 			o = 8 // small but nonzero
 		}
 	}

 	// For nonzero-sized structs which end in a zero-sized thing, we add
 	// an extra byte of padding to the type. This padding ensures that
 	// taking the address of the zero-sized thing can't manufacture a
 	// pointer to the next object in the heap. See issue 9401.
 	if flag == 1 && o > starto && o == lastzero {
 		o++
 	}

 	// final width is rounded
 	if flag != 0 {
 		o = Rnd(o, int64(maxalign))
 	}
 	t.Align = uint8(maxalign)

 	// type width only includes back to first field's offset
 	t.Width = o - starto

 	return o
 }

 // findTypeLoop searches for an invalid type declaration loop involving
 // type t and reports whether one is found. If so, path contains the
 // loop.
 //
 // path points to a slice used for tracking the sequence of types
 // visited. Using a pointer to a slice allows the slice capacity to
 // grow and limit reallocations.
 func findTypeLoop(t *Type, path *[]*Type) bool {
 	// We implement a simple DFS loop-finding algorithm. This
 	// could be faster, but type cycles are rare.

 	if t.Sym() != nil {
 		// Declared type. Check for loops and otherwise
 		// recurse on the type expression used in the type
 		// declaration.

 		// Type imported from package, so it can't be part of
 		// a type loop (otherwise that package should have
 		// failed to compile).
 		if t.Sym().Pkg != LocalPkg {
 			return false
 		}

 		for i, x := range *path {
 			if x == t {
 				*path = (*path)[i:]
 				return true
 			}
 		}

 		*path = append(*path, t)
 		if findTypeLoop(t.Obj().(TypeObject).TypeDefn(), path) {
 			return true
 		}
 		*path = (*path)[:len(*path)-1]
 	} else {
 		// Anonymous type. Recurse on contained types.

 		switch t.Kind() {
 		case TARRAY:
 			if findTypeLoop(t.Elem(), path) {
 				return true
 			}
 		case TSTRUCT:
 			for _, f := range t.Fields().Slice() {
 				if findTypeLoop(f.Type, path) {
 					return true
 				}
 			}
 		case TINTER:
 			for _, m := range t.Methods().Slice() {
 				if m.Type.IsInterface() { // embedded interface
 					if findTypeLoop(m.Type, path) {
 						return true
 					}
 				}
 			}
 		}
 	}

 	return false
 }

 func reportTypeLoop(t *Type) {
 	if t.Broke() {
 		return
 	}

 	var l []*Type
 	if !findTypeLoop(t, &l) {
 		base.Fatalf("failed to find type loop for: %v", t)
 	}

 	// Rotate loop so that the earliest type declaration is first.
 	i := 0
 	for j, t := range l[1:] {
 		if typePos(t).Before(typePos(l[i])) {
 			i = j + 1
 		}
 	}
 	l = append(l[i:], l[:i]...)

 	var msg bytes.Buffer
 	fmt.Fprintf(&msg, "invalid recursive type %v\n", l[0])
 	for _, t := range l {
 		fmt.Fprintf(&msg, "\t%v: %v refers to\n", base.FmtPos(typePos(t)), t)
 		t.SetBroke(true)
 	}
 	fmt.Fprintf(&msg, "\t%v: %v", base.FmtPos(typePos(l[0])), l[0])
 	base.ErrorfAt(typePos(l[0]), msg.String())
 }

 // CalcSize calculates and stores the size and alignment for t.
 // If CalcSizeDisabled is set, and the size/alignment
 // have not already been calculated, it calls Fatal.
 // This is used to prevent data races in the back end.
 func CalcSize(t *Type) {
 	// Calling CalcSize when typecheck tracing enabled is not safe.
 	// See issue #33658.
 	if base.EnableTrace && SkipSizeForTracing {
 		return
 	}
 	if PtrSize == 0 {
 		// Assume this is a test.
 		return
 	}

 	if t == nil {
 		return
 	}

 	if t.Width == -2 {
 		reportTypeLoop(t)
 		t.Width = 0
 		t.Align = 1
 		return
 	}

 	if t.WidthCalculated() {
 		return
 	}

 	if CalcSizeDisabled {
 		if t.Broke() {
 			// break infinite recursion from Fatal call below
 			return
 		}
 		t.SetBroke(true)
 		base.Fatalf("width not calculated: %v", t)
 	}

 	// break infinite recursion if the broken recursive type
 	// is referenced again
 	if t.Broke() && t.Width == 0 {
 		return
 	}

 	// defer CheckSize calls until after we're done
 	DeferCheckSize()

 	lno := base.Pos
 	if pos := t.Pos(); pos.IsKnown() {
 		base.Pos = pos
 	}

 	t.Width = -2
 	t.Align = 0 // 0 means use t.Width, below

 	et := t.Kind()
 	switch et {
 	case TFUNC, TCHAN, TMAP, TSTRING:
 		break

 	// SimType == 0 during bootstrap
 	default:
 		if SimType[t.Kind()] != 0 {
 			et = SimType[t.Kind()]
 		}
 	}

 	var w int64
 	switch et {
 	default:
 		base.Fatalf("CalcSize: unknown type: %v", t)

 	// compiler-specific stuff
 	case TINT8, TUINT8, TBOOL:
 		// bool is int8
 		w = 1

 	case TINT16, TUINT16:
 		w = 2

 	case TINT32, TUINT32, TFLOAT32:
 		w = 4

 	case TINT64, TUINT64, TFLOAT64:
 		w = 8
 		t.Align = uint8(RegSize)

 	case TCOMPLEX64:
 		w = 8
 		t.Align = 4

 	case TCOMPLEX128:
 		w = 16
 		t.Align = uint8(RegSize)

 	case TPTR:
 		w = int64(PtrSize)
 		CheckSize(t.Elem())

 	case TUNSAFEPTR:
 		w = int64(PtrSize)

 	case TINTER: // implemented as 2 pointers
 		w = 2 * int64(PtrSize)
 		t.Align = uint8(PtrSize)
 		expandiface(t)

 	case TUNION:
 		// Always part of an interface for now, so size/align don't matter.
 		// Pretend a union is represented like an interface.
 		w = 2 * int64(PtrSize)
 		t.Align = uint8(PtrSize)

 	case TCHAN: // implemented as pointer
 		w = int64(PtrSize)

 		CheckSize(t.Elem())

 		// make fake type to check later to
 		// trigger channel argument check.
 		t1 := NewChanArgs(t)
 		CheckSize(t1)

 	case TCHANARGS:
 		t1 := t.ChanArgs()
 		CalcSize(t1) // just in case
 		if t1.Elem().Width >= 1<<16 {
 			base.ErrorfAt(typePos(t1), "channel element type too large (>64kB)")
 		}
 		w = 1 // anything will do

 	case TMAP: // implemented as pointer
 		w = int64(PtrSize)
 		CheckSize(t.Elem())
 		CheckSize(t.Key())

 	case TFORW: // should have been filled in
 		reportTypeLoop(t)
 		w = 1 // anything will do

 	case TANY:
 		// not a real type; should be replaced before use.
 		base.Fatalf("CalcSize any")

 	case TSTRING:
 		if StringSize == 0 {
 			base.Fatalf("early CalcSize string")
 		}
 		w = StringSize
 		t.Align = uint8(PtrSize)

 	case TARRAY:
 		if t.Elem() == nil {
 			break
 		}

 		CalcSize(t.Elem())
 		if t.Elem().Width != 0 {
 			cap := (uint64(MaxWidth) - 1) / uint64(t.Elem().Width)
 			if uint64(t.NumElem()) > cap {
 				base.ErrorfAt(typePos(t), "type %L larger than address space", t)
 			}
 		}
 		w = t.NumElem() * t.Elem().Width
 		t.Align = t.Elem().Align

 	case TSLICE:
 		if t.Elem() == nil {
 			break
 		}
 		w = SliceSize
 		CheckSize(t.Elem())
 		t.Align = uint8(PtrSize)

 	case TSTRUCT:
 		if t.IsFuncArgStruct() {
 			base.Fatalf("CalcSize fn struct %v", t)
 		}
 		w = calcStructOffset(t, t, 0, 1)

 	// make fake type to check later to
 	// trigger function argument computation.
 	case TFUNC:
 		t1 := NewFuncArgs(t)
 		CheckSize(t1)
 		w = int64(PtrSize) // width of func type is pointer

 	// function is 3 cated structures;
 	// compute their widths as side-effect.
 	case TFUNCARGS:
 		t1 := t.FuncArgs()
 		w = calcStructOffset(t1, t1.Recvs(), 0, 0)
 		w = calcStructOffset(t1, t1.Params(), w, RegSize)
 		w = calcStructOffset(t1, t1.Results(), w, RegSize)
 		t1.extra.(*Func).Argwid = w
 		if w%int64(RegSize) != 0 {
 			base.Warn("bad type %v %d\n", t1, w)
 		}
 		t.Align = 1

 	case TTYPEPARAM:
 		// TODO(danscales) - remove when we eliminate the need
 		// to do CalcSize in noder2 (which shouldn't be needed in the noder)
 		w = int64(PtrSize)
 	}

 	if PtrSize == 4 && w != int64(int32(w)) {
 		base.ErrorfAt(typePos(t), "type %v too large", t)
 	}

 	t.Width = w
 	if t.Align == 0 {
 		if w == 0 || w > 8 || w&(w-1) != 0 {
 			base.Fatalf("invalid alignment for %v", t)
 		}
 		t.Align = uint8(w)
 	}

 	base.Pos = lno

 	ResumeCheckSize()
 }

 // CalcStructSize calculates the size of s,
 // filling in s.Width and s.Align,
 // even if size calculation is otherwise disabled.
 func CalcStructSize(s *Type) {
 	s.Width = calcStructOffset(s, s, 0, 1) // sets align
 }

 // RecalcSize is like CalcSize, but recalculates t's size even if it
 // has already been calculated before. It does not recalculate other
 // types.
 func RecalcSize(t *Type) {
 	t.Align = 0
 	CalcSize(t)
 }

 // when a type's width should be known, we call CheckSize
 // to compute it.  during a declaration like
 //
 //	type T *struct { next T }
 //
 // it is necessary to defer the calculation of the struct width
 // until after T has been initialized to be a pointer to that struct.
 // similarly, during import processing structs may be used
 // before their definition.  in those situations, calling
 // DeferCheckSize() stops width calculations until
 // ResumeCheckSize() is called, at which point all the
 // CalcSizes that were deferred are executed.
 // CalcSize should only be called when the type's size
 // is needed immediately.  CheckSize makes sure the
 // size is evaluated eventually.

 var deferredTypeStack []*Type

 func CheckSize(t *Type) {
 	if t == nil {
 		return
 	}

 	// function arg structs should not be checked
 	// outside of the enclosing function.
 	if t.IsFuncArgStruct() {
 		base.Fatalf("CheckSize %v", t)
 	}

 	if defercalc == 0 {
 		CalcSize(t)
 		return
 	}

 	// if type has not yet been pushed on deferredTypeStack yet, do it now
 	if !t.Deferwidth() {
 		t.SetDeferwidth(true)
 		deferredTypeStack = append(deferredTypeStack, t)
 	}
 }

 func DeferCheckSize() {
 	defercalc++
 }

 func ResumeCheckSize() {
 	if defercalc == 1 {
 		for len(deferredTypeStack) > 0 {
 			t := deferredTypeStack[len(deferredTypeStack)-1]
 			deferredTypeStack = deferredTypeStack[:len(deferredTypeStack)-1]
 			t.SetDeferwidth(false)
 			CalcSize(t)
 		}
 	}

 	defercalc--
 }

 // PtrDataSize returns the length in bytes of the prefix of t
 // containing pointer data. Anything after this offset is scalar data.
 func PtrDataSize(t *Type) int64 {
 	if !t.HasPointers() {
 		return 0
 	}

 	switch t.Kind() {
 	case TPTR,
 		TUNSAFEPTR,
 		TFUNC,
 		TCHAN,
 		TMAP:
 		return int64(PtrSize)

 	case TSTRING:
 		// struct { byte *str; intgo len; }
 		return int64(PtrSize)

 	case TINTER:
 		// struct { Itab *tab;	void *data; } or
 		// struct { Type *type; void *data; }
 		// Note: see comment in typebits.Set
 		return 2 * int64(PtrSize)

 	case TSLICE:
 		// struct { byte *array; uintgo len; uintgo cap; }
 		return int64(PtrSize)

 	case TARRAY:
 		// haspointers already eliminated t.NumElem() == 0.
 		return (t.NumElem()-1)*t.Elem().Width + PtrDataSize(t.Elem())

 	case TSTRUCT:
 		// Find the last field that has pointers.
 		var lastPtrField *Field
 		fs := t.Fields().Slice()
 		for i := len(fs) - 1; i >= 0; i-- {
 			if fs[i].Type.HasPointers() {
 				lastPtrField = fs[i]
 				break
 			}
 		}
 		return lastPtrField.Offset + PtrDataSize(lastPtrField.Type)

 	default:
 		base.Fatalf("PtrDataSize: unexpected type, %v", t)
 		return 0
 	}
 }
	// Copyright 2009 The Go Authors. All rights reserved.
	// Use of this source code is governed by a BSD-style
	// license that can be found in the LICENSE file.

	package types

	import (
	"bytes"
	"fmt"
	"sort"

	"cmd/compile/internal/base"
	"cmd/internal/src"
	)

	var PtrSize int

	var RegSize int

	// Slices in the runtime are represented by three components:
	//
	// type slice struct {
	// ptr unsafe.Pointer
	// len int
	// cap int
	// }
	//
	// Strings in the runtime are represented by two components:
	//
	// type string struct {
	// ptr unsafe.Pointer
	// len int
	// }
	//
	// These variables are the offsets of fields and sizes of these structs.
	var (
	SlicePtrOffset int64
	SliceLenOffset int64
	SliceCapOffset int64

	SliceSize int64
	StringSize int64
	)

	var SkipSizeForTracing bool

	// typePos returns the position associated with t.
	// This is where t was declared or where it appeared as a type expression.
	func typePos(t *Type) src.XPos {
	if pos := t.Pos(); pos.IsKnown() {
	return pos
	}
	base.Fatalf("bad type: %v", t)
	panic("unreachable")
	}

	// MaxWidth is the maximum size of a value on the target architecture.
	var MaxWidth int64

	// CalcSizeDisabled indicates whether it is safe
	// to calculate Types' widths and alignments. See CalcSize.
	var CalcSizeDisabled bool

	// machine size and rounding alignment is dictated around
	// the size of a pointer, set in betypeinit (see ../amd64/galign.go).
	var defercalc int

	func Rnd(o int64, r int64) int64 {
	if r < 1 \|\| r > 8 \|\| r&(r-1) != 0 {
	base.Fatalf("rnd %d", r)
	}
	return (o + r - 1) &^ (r - 1)
	}

	// expandiface computes the method set for interface type t by
	// expanding embedded interfaces.
	func expandiface(t *Type) {
	seen := make(map[Sym]Field)
	var methods []*Field

	addMethod := func(m *Field, explicit bool) {
	switch prev := seen[m.Sym]; {
	case prev == nil:
	seen[m.Sym] = m
	case AllowsGoVersion(t.Pkg(), 1, 14) && !explicit && Identical(m.Type, prev.Type):
	return
	default:
	base.ErrorfAt(m.Pos, "duplicate method %s", m.Sym.Name)
	}
	methods = append(methods, m)
	}

	{
	methods := t.Methods().Slice()
	sort.SliceStable(methods, func(i, j int) bool {
	mi, mj := methods[i], methods[j]

	// Sort embedded types by type name (if any).
	if mi.Sym == nil && mj.Sym == nil {
	return mi.Type.Sym().Less(mj.Type.Sym())
	}

	// Sort methods before embedded types.
	if mi.Sym == nil \|\| mj.Sym == nil {
	return mi.Sym != nil
	}

	// Sort methods by symbol name.
	return mi.Sym.Less(mj.Sym)
	})
	}

	for _, m := range t.Methods().Slice() {
	if m.Sym == nil {
	continue
	}

	CheckSize(m.Type)
	addMethod(m, true)
	}

	for _, m := range t.Methods().Slice() {
	if m.Sym != nil \|\| m.Type == nil {
	continue
	}

	if m.Type.IsUnion() {
	continue
	}

	// In 1.18, embedded types can be anything. In Go 1.17, we disallow
	// embedding anything other than interfaces.
	if !m.Type.IsInterface() {
	if AllowsGoVersion(t.Pkg(), 1, 18) {
	continue
	}
	base.ErrorfAt(m.Pos, "interface contains embedded non-interface, non-union %v", m.Type)
	m.SetBroke(true)
	t.SetBroke(true)
	// Add to fields so that error messages
	// include the broken embedded type when
	// printing t.
	// TODO(mdempsky): Revisit this.
	methods = append(methods, m)
	continue
	}

	// Embedded interface: duplicate all methods
	// (including broken ones, if any) and add to t's
	// method set.
	for _, t1 := range m.Type.AllMethods().Slice() {
	f := NewField(m.Pos, t1.Sym, t1.Type)
	addMethod(f, false)

	// Clear position after typechecking, for consistency with types2.
	f.Pos = src.NoXPos
	}

	// Clear position after typechecking, for consistency with types2.
	m.Pos = src.NoXPos
	}

	sort.Sort(MethodsByName(methods))

	if int64(len(methods)) >= MaxWidth/int64(PtrSize) {
	base.ErrorfAt(typePos(t), "interface too large")
	}
	for i, m := range methods {
	m.Offset = int64(i) * int64(PtrSize)
	}

	t.SetAllMethods(methods)
	}

	func calcStructOffset(errtype Type, t Type, o int64, flag int) int64 {
	// flag is 0 (receiver), 1 (actual struct), or RegSize (in/out parameters)
	isStruct := flag == 1
	starto := o
	maxalign := int32(flag)
	if maxalign < 1 {
	maxalign = 1
	}
	lastzero := int64(0)
	for _, f := range t.Fields().Slice() {
	if f.Type == nil {
	// broken field, just skip it so that other valid fields
	// get a width.
	continue
	}

	CalcSize(f.Type)
	if int32(f.Type.Align) > maxalign {
	maxalign = int32(f.Type.Align)
	}
	if f.Type.Align > 0 {
	o = Rnd(o, int64(f.Type.Align))
	}
	if isStruct { // For receiver/args/results, do not set, it depends on ABI
	f.Offset = o
	}

	w := f.Type.Width
	if w < 0 {
	base.Fatalf("invalid width %d", f.Type.Width)
	}
	if w == 0 {
	lastzero = o
	}
	o += w
	maxwidth := MaxWidth
	// On 32-bit systems, reflect tables impose an additional constraint
	// that each field start offset must fit in 31 bits.
	if maxwidth < 1<<32 {
	maxwidth = 1<<31 - 1
	}
	if o >= maxwidth {
	base.ErrorfAt(typePos(errtype), "type %L too large", errtype)
	o = 8 // small but nonzero
	}
	}

	// For nonzero-sized structs which end in a zero-sized thing, we add
	// an extra byte of padding to the type. This padding ensures that
	// taking the address of the zero-sized thing can't manufacture a
	// pointer to the next object in the heap. See issue 9401.
	if flag == 1 && o > starto && o == lastzero {
	o++
	}

	// final width is rounded
	if flag != 0 {
	o = Rnd(o, int64(maxalign))
	}
	t.Align = uint8(maxalign)

	// type width only includes back to first field's offset
	t.Width = o - starto

	return o
	}

	// findTypeLoop searches for an invalid type declaration loop involving
	// type t and reports whether one is found. If so, path contains the
	// loop.
	//
	// path points to a slice used for tracking the sequence of types
	// visited. Using a pointer to a slice allows the slice capacity to
	// grow and limit reallocations.
	func findTypeLoop(t Type, path []*Type) bool {
	// We implement a simple DFS loop-finding algorithm. This
	// could be faster, but type cycles are rare.

	if t.Sym() != nil {
	// Declared type. Check for loops and otherwise
	// recurse on the type expression used in the type
	// declaration.

	// Type imported from package, so it can't be part of
	// a type loop (otherwise that package should have
	// failed to compile).
	if t.Sym().Pkg != LocalPkg {
	return false
	}

	for i, x := range *path {
	if x == t {
	path = (path)[i:]
	return true
	}
	}

	path = append(path, t)
	if findTypeLoop(t.Obj().(TypeObject).TypeDefn(), path) {
	return true
	}
	path = (path)[:len(*path)-1]
	} else {
	// Anonymous type. Recurse on contained types.

	switch t.Kind() {
	case TARRAY:
	if findTypeLoop(t.Elem(), path) {
	return true
	}
	case TSTRUCT:
	for _, f := range t.Fields().Slice() {
	if findTypeLoop(f.Type, path) {
	return true
	}
	}
	case TINTER:
	for _, m := range t.Methods().Slice() {
	if m.Type.IsInterface() { // embedded interface
	if findTypeLoop(m.Type, path) {
	return true
	}
	}
	}
	}
	}

	return false
	}

	func reportTypeLoop(t *Type) {
	if t.Broke() {
	return
	}

	var l []*Type
	if !findTypeLoop(t, &l) {
	base.Fatalf("failed to find type loop for: %v", t)
	}

	// Rotate loop so that the earliest type declaration is first.
	i := 0
	for j, t := range l[1:] {
	if typePos(t).Before(typePos(l[i])) {
	i = j + 1
	}
	}
	l = append(l[i:], l[:i]...)

	var msg bytes.Buffer
	fmt.Fprintf(&msg, "invalid recursive type %v\n", l[0])
	for _, t := range l {
	fmt.Fprintf(&msg, "\t%v: %v refers to\n", base.FmtPos(typePos(t)), t)
	t.SetBroke(true)
	}
	fmt.Fprintf(&msg, "\t%v: %v", base.FmtPos(typePos(l[0])), l[0])
	base.ErrorfAt(typePos(l[0]), msg.String())
	}

	// CalcSize calculates and stores the size and alignment for t.
	// If CalcSizeDisabled is set, and the size/alignment
	// have not already been calculated, it calls Fatal.
	// This is used to prevent data races in the back end.
	func CalcSize(t *Type) {
	// Calling CalcSize when typecheck tracing enabled is not safe.
	// See issue #33658.
	if base.EnableTrace && SkipSizeForTracing {
	return
	}
	if PtrSize == 0 {
	// Assume this is a test.
	return
	}

	if t == nil {
	return
	}

	if t.Width == -2 {
	reportTypeLoop(t)
	t.Width = 0
	t.Align = 1
	return
	}

	if t.WidthCalculated() {
	return
	}

	if CalcSizeDisabled {
	if t.Broke() {
	// break infinite recursion from Fatal call below
	return
	}
	t.SetBroke(true)
	base.Fatalf("width not calculated: %v", t)
	}

	// break infinite recursion if the broken recursive type
	// is referenced again
	if t.Broke() && t.Width == 0 {
	return
	}

	// defer CheckSize calls until after we're done
	DeferCheckSize()

	lno := base.Pos
	if pos := t.Pos(); pos.IsKnown() {
	base.Pos = pos
	}

	t.Width = -2
	t.Align = 0 // 0 means use t.Width, below

	et := t.Kind()
	switch et {
	case TFUNC, TCHAN, TMAP, TSTRING:
	break

	// SimType == 0 during bootstrap
	default:
	if SimType[t.Kind()] != 0 {
	et = SimType[t.Kind()]
	}
	}

	var w int64
	switch et {
	default:
	base.Fatalf("CalcSize: unknown type: %v", t)

	// compiler-specific stuff
	case TINT8, TUINT8, TBOOL:
	// bool is int8
	w = 1

	case TINT16, TUINT16:
	w = 2

	case TINT32, TUINT32, TFLOAT32:
	w = 4

	case TINT64, TUINT64, TFLOAT64:
	w = 8
	t.Align = uint8(RegSize)

	case TCOMPLEX64:
	w = 8
	t.Align = 4

	case TCOMPLEX128:
	w = 16
	t.Align = uint8(RegSize)

	case TPTR:
	w = int64(PtrSize)
	CheckSize(t.Elem())

	case TUNSAFEPTR:
	w = int64(PtrSize)

	case TINTER: // implemented as 2 pointers
	w = 2 * int64(PtrSize)
	t.Align = uint8(PtrSize)
	expandiface(t)

	case TUNION:
	// Always part of an interface for now, so size/align don't matter.
	// Pretend a union is represented like an interface.
	w = 2 * int64(PtrSize)
	t.Align = uint8(PtrSize)

	case TCHAN: // implemented as pointer
	w = int64(PtrSize)

	CheckSize(t.Elem())

	// make fake type to check later to
	// trigger channel argument check.
	t1 := NewChanArgs(t)
	CheckSize(t1)

	case TCHANARGS:
	t1 := t.ChanArgs()
	CalcSize(t1) // just in case
	if t1.Elem().Width >= 1<<16 {
	base.ErrorfAt(typePos(t1), "channel element type too large (>64kB)")
	}
	w = 1 // anything will do

	case TMAP: // implemented as pointer
	w = int64(PtrSize)
	CheckSize(t.Elem())
	CheckSize(t.Key())

	case TFORW: // should have been filled in
	reportTypeLoop(t)
	w = 1 // anything will do

	case TANY:
	// not a real type; should be replaced before use.
	base.Fatalf("CalcSize any")

	case TSTRING:
	if StringSize == 0 {
	base.Fatalf("early CalcSize string")
	}
	w = StringSize
	t.Align = uint8(PtrSize)

	case TARRAY:
	if t.Elem() == nil {
	break
	}

	CalcSize(t.Elem())
	if t.Elem().Width != 0 {
	cap := (uint64(MaxWidth) - 1) / uint64(t.Elem().Width)
	if uint64(t.NumElem()) > cap {
	base.ErrorfAt(typePos(t), "type %L larger than address space", t)
	}
	}
	w = t.NumElem() * t.Elem().Width
	t.Align = t.Elem().Align

	case TSLICE:
	if t.Elem() == nil {
	break
	}
	w = SliceSize
	CheckSize(t.Elem())
	t.Align = uint8(PtrSize)

	case TSTRUCT:
	if t.IsFuncArgStruct() {
	base.Fatalf("CalcSize fn struct %v", t)
	}
	w = calcStructOffset(t, t, 0, 1)

	// make fake type to check later to
	// trigger function argument computation.
	case TFUNC:
	t1 := NewFuncArgs(t)
	CheckSize(t1)
	w = int64(PtrSize) // width of func type is pointer

	// function is 3 cated structures;
	// compute their widths as side-effect.
	case TFUNCARGS:
	t1 := t.FuncArgs()
	w = calcStructOffset(t1, t1.Recvs(), 0, 0)
	w = calcStructOffset(t1, t1.Params(), w, RegSize)
	w = calcStructOffset(t1, t1.Results(), w, RegSize)
	t1.extra.(*Func).Argwid = w
	if w%int64(RegSize) != 0 {
	base.Warn("bad type %v %d\n", t1, w)
	}
	t.Align = 1

	case TTYPEPARAM:
	// TODO(danscales) - remove when we eliminate the need
	// to do CalcSize in noder2 (which shouldn't be needed in the noder)
	w = int64(PtrSize)
	}

	if PtrSize == 4 && w != int64(int32(w)) {
	base.ErrorfAt(typePos(t), "type %v too large", t)
	}

	t.Width = w
	if t.Align == 0 {
	if w == 0 \|\| w > 8 \|\| w&(w-1) != 0 {
	base.Fatalf("invalid alignment for %v", t)
	}
	t.Align = uint8(w)
	}

	base.Pos = lno

	ResumeCheckSize()
	}

	// CalcStructSize calculates the size of s,
	// filling in s.Width and s.Align,
	// even if size calculation is otherwise disabled.
	func CalcStructSize(s *Type) {
	s.Width = calcStructOffset(s, s, 0, 1) // sets align
	}

	// RecalcSize is like CalcSize, but recalculates t's size even if it
	// has already been calculated before. It does not recalculate other
	// types.
	func RecalcSize(t *Type) {
	t.Align = 0
	CalcSize(t)
	}

	// when a type's width should be known, we call CheckSize
	// to compute it. during a declaration like
	//
	// type T *struct { next T }
	//
	// it is necessary to defer the calculation of the struct width
	// until after T has been initialized to be a pointer to that struct.
	// similarly, during import processing structs may be used
	// before their definition. in those situations, calling
	// DeferCheckSize() stops width calculations until
	// ResumeCheckSize() is called, at which point all the
	// CalcSizes that were deferred are executed.
	// CalcSize should only be called when the type's size
	// is needed immediately. CheckSize makes sure the
	// size is evaluated eventually.

	var deferredTypeStack []*Type

	func CheckSize(t *Type) {
	if t == nil {
	return
	}

	// function arg structs should not be checked
	// outside of the enclosing function.
	if t.IsFuncArgStruct() {
	base.Fatalf("CheckSize %v", t)
	}

	if defercalc == 0 {
	CalcSize(t)
	return
	}

	// if type has not yet been pushed on deferredTypeStack yet, do it now
	if !t.Deferwidth() {
	t.SetDeferwidth(true)
	deferredTypeStack = append(deferredTypeStack, t)
	}
	}

	func DeferCheckSize() {
	defercalc++
	}

	func ResumeCheckSize() {
	if defercalc == 1 {
	for len(deferredTypeStack) > 0 {
	t := deferredTypeStack[len(deferredTypeStack)-1]
	deferredTypeStack = deferredTypeStack[:len(deferredTypeStack)-1]
	t.SetDeferwidth(false)
	CalcSize(t)
	}
	}

	defercalc--
	}

	// PtrDataSize returns the length in bytes of the prefix of t
	// containing pointer data. Anything after this offset is scalar data.
	func PtrDataSize(t *Type) int64 {
	if !t.HasPointers() {
	return 0
	}

	switch t.Kind() {
	case TPTR,
	TUNSAFEPTR,
	TFUNC,
	TCHAN,
	TMAP:
	return int64(PtrSize)

	case TSTRING:
	// struct { byte *str; intgo len; }
	return int64(PtrSize)

	case TINTER:
	// struct { Itab tab; void data; } or
	// struct { Type type; void data; }
	// Note: see comment in typebits.Set
	return 2 * int64(PtrSize)

	case TSLICE:
	// struct { byte *array; uintgo len; uintgo cap; }
	return int64(PtrSize)

	case TARRAY:
	// haspointers already eliminated t.NumElem() == 0.
	return (t.NumElem()-1)*t.Elem().Width + PtrDataSize(t.Elem())

	case TSTRUCT:
	// Find the last field that has pointers.
	var lastPtrField *Field
	fs := t.Fields().Slice()
	for i := len(fs) - 1; i >= 0; i-- {
	if fs[i].Type.HasPointers() {
	lastPtrField = fs[i]
	break
	}
	}
	return lastPtrField.Offset + PtrDataSize(lastPtrField.Type)

	default:
	base.Fatalf("PtrDataSize: unexpected type, %v", t)
	return 0
	}
	}