1
0
mirror of https://github.com/golang/go synced 2024-11-24 22:00:09 -07:00

cmd/cgo: bug fixes

* disallow embedding of C type (Fixes issue 2552)
* detect 0-length array (Fixes issue 2806)
* use typedefs when possible, to avoid attribute((unavailable)) (Fixes issue 2888)
* print Go types constructed from C types using original C types (Fixes issue 2612)

This fix changes _cgo_export.h to repeat the preamble from import "C".
Otherwise the fix to issue 2612 is impossible, since it cannot refer to
types that have not been defined.  If people are using //export and
putting non-header information in the preamble, they will need to
refactor their code.

R=golang-dev, r, r
CC=golang-dev
https://golang.org/cl/5672080
This commit is contained in:
Russ Cox 2012-02-19 13:32:55 -05:00
parent 72fb81eeb6
commit 1a0c8fe9bb
14 changed files with 157 additions and 35 deletions

View File

@ -1853,7 +1853,24 @@ Code that uses the old fields will fail to compile and must be updated by hand.
The semantic changes make it difficult for the fix tool to update automatically.
</p>
<h2 id="go_command">The go command</h2>
<h2 id="cmd_go">The go command</h2>
<p>
TODO: Write this.
</p>
<h2 id="cmd_cgo">The cgo command</h2>
<p>
In Go 1, the <a href="/cmd/cgo">cgo command</a>
uses a different <code>_cgo_export.h</code>
file, which is generated for packages containing <code>//export</code> lines.
The <code>_cgo_export.h</code> file now begins with the C preamble comment,
so that exported function definitions can use types defined there.
This has the effect of compiling the preamble multiple times, so a
package using <code>//export</code> must not put function definitions
or variable initializations in the C preamble.
</p
<h2 id="releases">Packaged releases</h2>

View File

@ -1725,7 +1725,24 @@ Code that uses the old fields will fail to compile and must be updated by hand.
The semantic changes make it difficult for the fix tool to update automatically.
</p>
<h2 id="go_command">The go command</h2>
<h2 id="cmd_go">The go command</h2>
<p>
TODO: Write this.
</p>
<h2 id="cmd_cgo">The cgo command</h2>
<p>
In Go 1, the <a href="/cmd/cgo">cgo command</a>
uses a different <code>_cgo_export.h</code>
file, which is generated for packages containing <code>//export</code> lines.
The <code>_cgo_export.h</code> file now begins with the C preamble comment,
so that exported function definitions can use types defined there.
This has the effect of compiling the preamble multiple times, so a
package using <code>//export</code> must not put function definitions
or variable initializations in the C preamble.
</p
<h2 id="releases">Packaged releases</h2>

View File

@ -147,6 +147,9 @@ func (f *File) saveRef(x interface{}, context string) {
if context == "as2" {
context = "expr"
}
if context == "embed-type" {
error_(sel.Pos(), "cannot embed C type")
}
goname := sel.Sel.Name
if goname == "errno" {
error_(sel.Pos(), "cannot refer to errno directly; see documentation")
@ -232,7 +235,11 @@ func (f *File) walk(x interface{}, context string, visit func(*File, interface{}
// These are ordered and grouped to match ../../pkg/go/ast/ast.go
case *ast.Field:
f.walk(&n.Type, "type", visit)
if len(n.Names) == 0 && context == "field" {
f.walk(&n.Type, "embed-type", visit)
} else {
f.walk(&n.Type, "type", visit)
}
case *ast.FieldList:
for _, field := range n.List {
f.walk(field, context, visit)
@ -289,9 +296,9 @@ func (f *File) walk(x interface{}, context string, visit func(*File, interface{}
case *ast.StructType:
f.walk(n.Fields, "field", visit)
case *ast.FuncType:
f.walk(n.Params, "field", visit)
f.walk(n.Params, "param", visit)
if n.Results != nil {
f.walk(n.Results, "field", visit)
f.walk(n.Results, "param", visit)
}
case *ast.InterfaceType:
f.walk(n.Methods, "field", visit)
@ -379,7 +386,7 @@ func (f *File) walk(x interface{}, context string, visit func(*File, interface{}
f.walk(n.Specs, "spec", visit)
case *ast.FuncDecl:
if n.Recv != nil {
f.walk(n.Recv, "field", visit)
f.walk(n.Recv, "param", visit)
}
f.walk(n.Type, "type", visit)
if n.Body != nil {

View File

@ -16,8 +16,8 @@ the pseudo-package "C" and then refers to types such as C.size_t,
variables such as C.stdout, or functions such as C.putchar.
If the import of "C" is immediately preceded by a comment, that
comment is used as a header when compiling the C parts of
the package. For example:
comment, called the preamble, is used as a header when compiling
the C parts of the package. For example:
// #include <stdio.h>
// #include <errno.h>
@ -57,6 +57,8 @@ The C type void* is represented by Go's unsafe.Pointer.
To access a struct, union, or enum type directly, prefix it with
struct_, union_, or enum_, as in C.struct_stat.
Go structs cannot embed fields with C types.
Any C function that returns a value may be called in a multiple
assignment context to retrieve both the return value and the
C errno variable as an error. For example:
@ -100,7 +102,8 @@ They will be available in the C code as:
extern int64 MyFunction(int arg1, int arg2, GoString arg3);
extern struct MyFunction2_return MyFunction2(int arg1, int arg2, GoString arg3);
found in _cgo_export.h generated header. Functions with multiple
found in _cgo_export.h generated header, after any preambles
copied from the cgo input files. Functions with multiple
return values are mapped to functions returning a struct.
Not all Go types can be mapped to C types in a useful way.

View File

@ -709,7 +709,7 @@ func (p *Package) rewriteRef(f *File) {
// Substitute definition for mangled type name.
if id, ok := expr.(*ast.Ident); ok {
if t := typedef[id.Name]; t != nil {
expr = t
expr = t.Go
}
if id.Name == r.Name.Mangle && r.Name.Const != "" {
expr = ast.NewIdent(r.Name.Const)
@ -894,7 +894,7 @@ type typeConv struct {
}
var tagGen int
var typedef = make(map[string]ast.Expr)
var typedef = make(map[string]*Type)
var goIdent = make(map[string]*ast.Ident)
func (c *typeConv) Init(ptrSize int64) {
@ -1164,17 +1164,22 @@ func (c *typeConv) Type(dtype dwarf.Type, pos token.Pos) *Type {
goIdent[name.Name] = name
switch dt.Kind {
case "union", "class":
typedef[name.Name] = c.Opaque(t.Size)
if t.C.Empty() {
t.C.Set("typeof(unsigned char[%d])", t.Size)
}
typedef[name.Name] = t
case "struct":
g, csyntax, align := c.Struct(dt, pos)
if t.C.Empty() {
t.C.Set(csyntax)
}
t.Align = align
typedef[name.Name] = g
tt := *t
if tag != "" {
tt.C = &TypeRepr{"struct %s", []interface{}{tag}}
}
tt.Go = g
typedef[name.Name] = &tt
}
case *dwarf.TypedefType:
@ -1203,7 +1208,9 @@ func (c *typeConv) Type(dtype dwarf.Type, pos token.Pos) *Type {
t.Size = sub.Size
t.Align = sub.Align
if _, ok := typedef[name.Name]; !ok {
typedef[name.Name] = sub.Go
tt := *t
tt.Go = sub.Go
typedef[name.Name] = &tt
}
if *godefs || *cdefs {
t.Go = sub.Go
@ -1250,7 +1257,8 @@ func (c *typeConv) Type(dtype dwarf.Type, pos token.Pos) *Type {
}
s = strings.Join(strings.Split(s, " "), "") // strip spaces
name := c.Ident("_Ctype_" + s)
typedef[name.Name] = t.Go
tt := *t
typedef[name.Name] = &tt
if !*godefs && !*cdefs {
t.Go = name
}
@ -1288,9 +1296,18 @@ func (c *typeConv) FuncArg(dtype dwarf.Type, pos token.Pos) *Type {
if ptr, ok := base(dt.Type).(*dwarf.PtrType); ok {
// Unless the typedef happens to point to void* since
// Go has special rules around using unsafe.Pointer.
if _, void := base(ptr.Type).(*dwarf.VoidType); !void {
return c.Type(ptr, pos)
if _, void := base(ptr.Type).(*dwarf.VoidType); void {
break
}
t = c.Type(ptr, pos)
if t == nil {
return nil
}
// Remember the C spelling, in case the struct
// has __attribute__((unavailable)) on it. See issue 2888.
t.Typedef = dt.Name
}
}
return t
@ -1443,7 +1460,7 @@ func (c *typeConv) Struct(dt *dwarf.StructType, pos token.Pos) (expr *ast.Struct
off = dt.ByteSize
}
if off != dt.ByteSize {
fatalf("%s: struct size calculation error", lineno(pos))
fatalf("%s: struct size calculation error off=%d bytesize=%d", lineno(pos), off, dt.ByteSize)
}
buf.WriteString("}")
csyntax = buf.String()

View File

@ -80,7 +80,7 @@ func (p *Package) godefs(f *File, srcfile string) string {
// and xxx is a typedef for yyy, make C.yyy format as T.
for typ, def := range typedef {
if new := override[typ]; new != "" {
if id, ok := def.(*ast.Ident); ok {
if id, ok := def.Go.(*ast.Ident); ok {
override[id.Name] = new
}
}

View File

@ -39,6 +39,7 @@ type Package struct {
Decl []ast.Decl
GoFiles []string // list of Go files
GccFiles []string // list of gcc output files
Preamble string // collected preamble for _cgo_export.h
}
// A File collects information about a single Go input file.
@ -98,6 +99,7 @@ type Type struct {
C *TypeRepr
Go ast.Expr
EnumValues map[string]int64
Typedef string
}
// A FuncType collects information about a function type in both the C and Go worlds.
@ -312,6 +314,9 @@ func (p *Package) Record(f *File) {
}
}
p.ExpFunc = append(p.ExpFunc, f.ExpFunc...)
if f.ExpFunc != nil {
p.ExpFunc = append(p.ExpFunc, f.ExpFunc...)
p.Preamble += "\n" + f.Preamble
}
p.Decl = append(p.Decl, f.AST.Decls...)
}

View File

@ -59,7 +59,7 @@ func (p *Package) writeDefs() {
for name, def := range typedef {
fmt.Fprintf(fgo2, "type %s ", name)
conf.Fprint(fgo2, fset, def)
conf.Fprint(fgo2, fset, def.Go)
fmt.Fprintf(fgo2, "\n\n")
}
fmt.Fprintf(fgo2, "type _Ctype_void [0]byte\n")
@ -196,7 +196,11 @@ func (p *Package) structType(n *Name) (string, int64) {
fmt.Fprintf(&buf, "\t\tchar __pad%d[%d];\n", off, pad)
off += pad
}
fmt.Fprintf(&buf, "\t\t%s p%d;\n", t.C, i)
c := t.Typedef
if c == "" {
c = t.C.String()
}
fmt.Fprintf(&buf, "\t\t%s p%d;\n", c, i)
off += t.Size
}
if off%p.PtrSize != 0 {
@ -428,6 +432,7 @@ func (p *Package) writeExports(fgo2, fc, fm *os.File) {
fgcch := creat(*objDir + "_cgo_export.h")
fmt.Fprintf(fgcch, "/* Created by cgo - DO NOT EDIT. */\n")
fmt.Fprintf(fgcch, "%s\n", p.Preamble)
fmt.Fprintf(fgcch, "%s\n", gccExportHeaderProlog)
fmt.Fprintf(fgcc, "/* Created by cgo - DO NOT EDIT. */\n")
@ -693,10 +698,8 @@ func (p *Package) cgoType(e ast.Expr) *Type {
}
}
}
for name, def := range typedef {
if name == t.Name {
return p.cgoType(def)
}
if def := typedef[t.Name]; def != nil {
return def
}
if t.Name == "uintptr" {
return &Type{Size: p.PtrSize, Align: p.PtrSize, C: c("uintptr")}
@ -721,7 +724,7 @@ func (p *Package) cgoType(e ast.Expr) *Type {
return &Type{Size: p.PtrSize, Align: p.PtrSize, C: c("void*")}
}
}
error_(e.Pos(), "unrecognized Go type %T", e)
error_(e.Pos(), "Go type not supported in export: %s", gofmt(e))
return &Type{Size: 4, Align: 4, C: c("int")}
}

View File

@ -70,7 +70,11 @@ func lineno(pos token.Pos) string {
// Die with an error message.
func fatalf(msg string, args ...interface{}) {
fmt.Fprintf(os.Stderr, msg+"\n", args...)
// If we've already printed other errors, they might have
// caused the fatal condition. Assume they're enough.
if nerrors == 0 {
fmt.Fprintf(os.Stderr, msg+"\n", args...)
}
os.Exit(2)
}

View File

@ -28,8 +28,13 @@ typedef struct my_struct {
volatile int vi;
char x : 1;
int y : 4;
int z[0];
long long array[40];
int zz[0];
} t_my_struct;
typedef struct my_struct1 {
int zz [1];
} t_my_struct1;
typedef union my_union {
volatile int vi;
char x : 1;
@ -65,7 +70,8 @@ t_func_void_of_char *a9;
t_func_void_of_void *a10;
t_func_void_of_ptr_char_dots *a11;
t_my_struct *a12;
t_my_union *a12a;
t_my_struct1 *a12a;
t_my_union *a12b;
t_my_enum *a13;
t_my_list *a14;
t_my_tree *a15;

Binary file not shown.

Binary file not shown.

View File

@ -426,6 +426,8 @@ func (d *Data) Type(off Offset) (Type, error) {
t.StructName, _ = e.Val(AttrName).(string)
t.Incomplete = e.Val(AttrDeclaration) != nil
t.Field = make([]*StructField, 0, 8)
var lastFieldType Type
var lastFieldBitOffset int64
for kid := next(); kid != nil; kid = next() {
if kid.Tag == TagMember {
f := new(StructField)
@ -444,11 +446,32 @@ func (d *Data) Type(off Offset) (Type, error) {
goto Error
}
}
haveBitOffset := false
f.Name, _ = kid.Val(AttrName).(string)
f.ByteSize, _ = kid.Val(AttrByteSize).(int64)
f.BitOffset, _ = kid.Val(AttrBitOffset).(int64)
f.BitOffset, haveBitOffset = kid.Val(AttrBitOffset).(int64)
f.BitSize, _ = kid.Val(AttrBitSize).(int64)
t.Field = append(t.Field, f)
bito := f.BitOffset
if !haveBitOffset {
bito = f.ByteOffset * 8
}
if bito == lastFieldBitOffset && t.Kind != "union" {
// Last field was zero width. Fix array length.
// (DWARF writes out 0-length arrays as if they were 1-length arrays.)
zeroArray(lastFieldType)
}
lastFieldType = f.Type
lastFieldBitOffset = bito
}
}
if t.Kind != "union" {
b, ok := e.Val(AttrByteSize).(int64)
if ok && b*8 == lastFieldBitOffset {
// Final field must be zero width. Fix array length.
zeroArray(lastFieldType)
}
}
@ -579,3 +602,14 @@ Error:
delete(d.typeCache, off)
return nil, err
}
func zeroArray(t Type) {
for {
at, ok := t.(*ArrayType)
if !ok {
break
}
at.Count = 0
t = at.Type
}
}

View File

@ -25,13 +25,22 @@ var typedefTests = map[string]string{
"t_func_void_of_char": "func(char) void",
"t_func_void_of_void": "func() void",
"t_func_void_of_ptr_char_dots": "func(*char, ...) void",
"t_my_struct": "struct my_struct {vi volatile int@0; x char@4 : 1@7; y int@4 : 4@27; array [40]long long int@8}",
"t_my_struct": "struct my_struct {vi volatile int@0; x char@4 : 1@7; y int@4 : 4@27; z [0]int@8; array [40]long long int@8; zz [0]int@328}",
"t_my_struct1": "struct my_struct1 {zz [1]int@0}",
"t_my_union": "union my_union {vi volatile int@0; x char@0 : 1@7; y int@0 : 4@28; array [40]long long int@0}",
"t_my_enum": "enum my_enum {e1=1; e2=2; e3=-5; e4=1000000000000000}",
"t_my_list": "struct list {val short int@0; next *t_my_list@8}",
"t_my_tree": "struct tree {left *struct tree@0; right *struct tree@8; val long long unsigned int@16}",
}
// As Apple converts gcc to a clang-based front end
// they keep breaking the DWARF output. This map lists the
// conversion from real answer to Apple answer.
var machoBug = map[string]string{
"func(*char, ...) void": "func(*char) void",
"enum my_enum {e1=1; e2=2; e3=-5; e4=1000000000000000}": "enum my_enum {e1=1; e2=2; e3=-5; e4=-1530494976}",
}
func elfData(t *testing.T, name string) *Data {
f, err := elf.Open(name)
if err != nil {
@ -58,13 +67,13 @@ func machoData(t *testing.T, name string) *Data {
return d
}
func TestTypedefsELF(t *testing.T) { testTypedefs(t, elfData(t, "testdata/typedef.elf")) }
func TestTypedefsELF(t *testing.T) { testTypedefs(t, elfData(t, "testdata/typedef.elf"), "elf") }
func TestTypedefsMachO(t *testing.T) {
testTypedefs(t, machoData(t, "testdata/typedef.macho"))
testTypedefs(t, machoData(t, "testdata/typedef.macho"), "macho")
}
func testTypedefs(t *testing.T, d *Data) {
func testTypedefs(t *testing.T, d *Data, kind string) {
r := d.Reader()
seen := make(map[string]bool)
for {
@ -93,7 +102,7 @@ func testTypedefs(t *testing.T, d *Data) {
t.Errorf("multiple definitions for %s", t1.Name)
}
seen[t1.Name] = true
if typstr != want {
if typstr != want && (kind != "macho" || typstr != machoBug[want]) {
t.Errorf("%s:\n\thave %s\n\twant %s", t1.Name, typstr, want)
}
}