debug/plan9obj: implement parsing of Plan 9 a.out executables

It implements parsing of the header and symbol table for both
32-bit and 64-bit Plan 9 binaries. The nm tool was updated to
use this package.

R=rsc, aram
CC=golang-codereviews
https://golang.org/cl/49970044

src/cmd/nm/nm.go

@@ -105,6 +105,10 @@ var parsers = []struct {
 	{[]byte("\xCE\xFA\xED\xFE"), machoSymbols},
 	{[]byte("\xCF\xFA\xED\xFE"), machoSymbols},
 	{[]byte("MZ"), peSymbols},
+	{[]byte("\x00\x00\x01\xEB"), plan9Symbols}, // 386
+	{[]byte("\x00\x00\x04\x07"), plan9Symbols}, // mips
+	{[]byte("\x00\x00\x06\x47"), plan9Symbols}, // arm
+	{[]byte("\x00\x00\x8A\x97"), plan9Symbols}, // amd64
 }
 
 func nm(file string) {

src/cmd/nm/plan9obj.go

+// Copyright 2014 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.
+
+// Parsing of Plan 9 a.out executables.
+
+package main
+
+import (
+	"debug/plan9obj"
+	"os"
+	"sort"
+)
+
+func plan9Symbols(f *os.File) []Sym {
+	p, err := plan9obj.NewFile(f)
+	if err != nil {
+		errorf("parsing %s: %v", f.Name(), err)
+		return nil
+	}
+
+	plan9Syms, err := p.Symbols()
+	if err != nil {
+		errorf("parsing %s: %v", f.Name(), err)
+		return nil
+	}
+
+	// Build sorted list of addresses of all symbols.
+	// We infer the size of a symbol by looking at where the next symbol begins.
+	var addrs []uint64
+	for _, s := range plan9Syms {
+		addrs = append(addrs, s.Value)
+	}
+	sort.Sort(uint64s(addrs))
+
+	var syms []Sym
+
+	for _, s := range plan9Syms {
+		sym := Sym{Addr: s.Value, Name: s.Name, Code: rune(s.Type)}
+		i := sort.Search(len(addrs), func(x int) bool { return addrs[x] > s.Value })
+		if i < len(addrs) {
+			sym.Size = int64(addrs[i] - s.Value)
+		}
+		syms = append(syms, sym)
+	}
+
+	return syms
+}

src/pkg/debug/plan9obj/file.go

+// Copyright 2014 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 plan9obj implements access to Plan 9 a.out object files.
+package plan9obj
+
+import (
+	"encoding/binary"
+	"errors"
+	"fmt"
+	"io"
+	"os"
+)
+
+// A FileHeader represents an Plan 9 a.out file header.
+type FileHeader struct {
+	Ptrsz int
+}
+
+// A File represents an open Plan 9 a.out file.
+type File struct {
+	FileHeader
+	Sections []*Section
+	closer   io.Closer
+}
+
+type SectionHeader struct {
+	Name   string
+	Size   uint32
+	Offset uint32
+}
+
+// A Section represents a single section in an Plan 9 a.out file.
+type Section struct {
+	SectionHeader
+
+	// Embed ReaderAt for ReadAt method.
+	// Do not embed SectionReader directly
+	// to avoid having Read and Seek.
+	// If a client wants Read and Seek it must use
+	// Open() to avoid fighting over the seek offset
+	// with other clients.
+	io.ReaderAt
+	sr *io.SectionReader
+}
+
+// Data reads and returns the contents of the Plan 9 a.out section.
+func (s *Section) Data() ([]byte, error) {
+	dat := make([]byte, s.sr.Size())
+	n, err := s.sr.ReadAt(dat, 0)
+	return dat[0:n], err
+}
+
+// Open returns a new ReadSeeker reading the Plan 9 a.out section.
+func (s *Section) Open() io.ReadSeeker { return io.NewSectionReader(s.sr, 0, 1<<63-1) }
+
+// A ProgHeader represents a single Plan 9 a.out program header.
+type ProgHeader struct {
+	Magic uint32
+	Text  uint32
+	Data  uint32
+	Bss   uint32
+	Syms  uint32
+	Entry uint64
+	Spsz  uint32
+	Pcsz  uint32
+}
+
+// A Prog represents the program header in an Plan 9 a.out binary.
+type Prog struct {
+	ProgHeader
+
+	// Embed ReaderAt for ReadAt method.
+	// Do not embed SectionReader directly
+	// to avoid having Read and Seek.
+	// If a client wants Read and Seek it must use
+	// Open() to avoid fighting over the seek offset
+	// with other clients.
+	io.ReaderAt
+	sr *io.SectionReader
+}
+
+// Open returns a new ReadSeeker reading the Plan 9 a.out program body.
+func (p *Prog) Open() io.ReadSeeker { return io.NewSectionReader(p.sr, 0, 1<<63-1) }
+
+// A Symbol represents an entry in a Plan 9 a.out symbol table section.
+type Sym struct {
+	Value uint64
+	Type  rune
+	Name  string
+}
+
+/*
+ * Plan 9 a.out reader
+ */
+
+type FormatError struct {
+	off int
+	msg string
+	val interface{}
+}
+
+func (e *FormatError) Error() string {
+	msg := e.msg
+	if e.val != nil {
+		msg += fmt.Sprintf(" '%v'", e.val)
+	}
+	msg += fmt.Sprintf(" in record at byte %#x", e.off)
+	return msg
+}
+
+// Open opens the named file using os.Open and prepares it for use as an Plan 9 a.out binary.
+func Open(name string) (*File, error) {
+	f, err := os.Open(name)
+	if err != nil {
+		return nil, err
+	}
+	ff, err := NewFile(f)
+	if err != nil {
+		f.Close()
+		return nil, err
+	}
+	ff.closer = f
+	return ff, nil
+}
+
+// Close closes the File.
+// If the File was created using NewFile directly instead of Open,
+// Close has no effect.
+func (f *File) Close() error {
+	var err error
+	if f.closer != nil {
+		err = f.closer.Close()
+		f.closer = nil
+	}
+	return err
+}
+
+func parseMagic(magic [4]byte) (*ExecTable, error) {
+	for _, e := range exectab {
+		if string(magic[:]) == e.Magic {
+			return &e, nil
+		}
+	}
+	return nil, &FormatError{0, "bad magic number", magic[:]}
+}
+
+// NewFile creates a new File for accessing an Plan 9 binary in an underlying reader.
+// The Plan 9 binary is expected to start at position 0 in the ReaderAt.
+func NewFile(r io.ReaderAt) (*File, error) {
+	sr := io.NewSectionReader(r, 0, 1<<63-1)
+	// Read and decode Plan 9 magic
+	var magic [4]byte
+	if _, err := r.ReadAt(magic[:], 0); err != nil {
+		return nil, err
+	}
+	mp, err := parseMagic(magic)
+	if err != nil {
+		return nil, err
+	}
+
+	f := &File{FileHeader{mp.Ptrsz}, nil, nil}
+
+	ph := new(prog)
+	if err := binary.Read(sr, binary.BigEndian, ph); err != nil {
+		return nil, err
+	}
+
+	p := new(Prog)
+	p.ProgHeader = ProgHeader{
+		Magic: ph.Magic,
+		Text:  ph.Text,
+		Data:  ph.Data,
+		Bss:   ph.Bss,
+		Syms:  ph.Syms,
+		Entry: uint64(ph.Entry),
+		Spsz:  ph.Spsz,
+		Pcsz:  ph.Pcsz,
+	}
+
+	if mp.Ptrsz == 8 {
+		if err := binary.Read(sr, binary.BigEndian, &p.Entry); err != nil {
+			return nil, err
+		}
+	}
+
+	var sects = []struct {
+		name string
+		size uint32
+	}{
+		{"text", ph.Text},
+		{"data", ph.Data},
+		{"syms", ph.Syms},
+		{"spsz", ph.Spsz},
+		{"pcsz", ph.Pcsz},
+	}
+
+	f.Sections = make([]*Section, 5)
+
+	off := mp.Hsize
+
+	for i, sect := range sects {
+		s := new(Section)
+		s.SectionHeader = SectionHeader{
+			Name:   sect.name,
+			Size:   sect.size,
+			Offset: off,
+		}
+		off += sect.size
+		s.sr = io.NewSectionReader(r, int64(s.SectionHeader.Offset), int64(s.SectionHeader.Size))
+		s.ReaderAt = s.sr
+		f.Sections[i] = s
+	}
+
+	return f, nil
+}
+
+func walksymtab(data []byte, ptrsz int, fn func(sym) error) error {
+	var order binary.ByteOrder = binary.BigEndian
+	var s sym
+	p := data
+	for len(p) >= 4 {
+		// Symbol type, value.
+		if len(p) < ptrsz {
+			return &FormatError{len(data), "unexpected EOF", nil}
+		}
+		// fixed-width value
+		if ptrsz == 8 {
+			s.value = order.Uint64(p[0:8])
+			p = p[8:]
+		} else {
+			s.value = uint64(order.Uint32(p[0:4]))
+			p = p[4:]
+		}
+
+		var typ byte
+		typ = p[0] & 0x7F
+		s.typ = typ
+		p = p[1:]
+
+		// Name.
+		var i int
+		var nnul int
+		for i = 0; i < len(p); i++ {
+			if p[i] == 0 {
+				nnul = 1
+				break
+			}
+		}
+		switch typ {
+		case 'z', 'Z':
+			p = p[i+nnul:]
+			for i = 0; i+2 <= len(p); i += 2 {
+				if p[i] == 0 && p[i+1] == 0 {
+					nnul = 2
+					break
+				}
+			}
+		}
+		if len(p) < i+nnul {
+			return &FormatError{len(data), "unexpected EOF", nil}
+		}
+		s.name = p[0:i]
+		i += nnul
+		p = p[i:]
+
+		fn(s)
+	}
+	return nil
+}
+
+// NewTable decodes the Go symbol table in data,
+// returning an in-memory representation.
+func newTable(symtab []byte, ptrsz int) ([]Sym, error) {
+	var n int
+	err := walksymtab(symtab, ptrsz, func(s sym) error {
+		n++
+		return nil
+	})
+	if err != nil {
+		return nil, err
+	}
+
+	fname := make(map[uint16]string)
+	syms := make([]Sym, 0, n)
+	err = walksymtab(symtab, ptrsz, func(s sym) error {
+		n := len(syms)
+		syms = syms[0 : n+1]
+		ts := &syms[n]
+		ts.Type = rune(s.typ)
+		ts.Value = s.value
+		switch s.typ {
+		default:
+			ts.Name = string(s.name[:])
+		case 'z', 'Z':
+			for i := 0; i < len(s.name); i += 2 {
+				eltIdx := binary.BigEndian.Uint16(s.name[i : i+2])
+				elt, ok := fname[eltIdx]
+				if !ok {
+					return &FormatError{-1, "bad filename code", eltIdx}
+				}
+				if n := len(ts.Name); n > 0 && ts.Name[n-1] != '/' {
+					ts.Name += "/"
+				}
+				ts.Name += elt
+			}
+		}
+		switch s.typ {
+		case 'f':
+			fname[uint16(s.value)] = ts.Name
+		}
+		return nil
+	})
+	if err != nil {
+		return nil, err
+	}
+
+	return syms, nil
+}
+
+// Symbols returns the symbol table for f.
+func (f *File) Symbols() ([]Sym, error) {
+	symtabSection := f.Section("syms")
+	if symtabSection == nil {
+		return nil, errors.New("no symbol section")
+	}
+
+	symtab, err := symtabSection.Data()
+	if err != nil {
+		return nil, errors.New("cannot load symbol section")
+	}
+
+	return newTable(symtab, f.Ptrsz)
+}
+
+// Section returns a section with the given name, or nil if no such
+// section exists.
+func (f *File) Section(name string) *Section {
+	for _, s := range f.Sections {
+		if s.Name == name {
+			return s
+		}
+	}
+	return nil
+}

src/pkg/debug/plan9obj/file_test.go

+// Copyright 2014 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 plan9obj
+
+import (
+	"reflect"
+	"testing"
+)
+
+type fileTest struct {
+	file     string
+	hdr      FileHeader
+	sections []*SectionHeader
+}
+
+var fileTests = []fileTest{
+	{
+		"testdata/386-plan9-exec",
+		FileHeader{4},
+		[]*SectionHeader{
+			{"text", 0x4c5f, 0x20},
+			{"data", 0x94c, 0x4c7f},
+			{"syms", 0x2c2b, 0x55cb},
+			{"spsz", 0x0, 0x81f6},
+			{"pcsz", 0xf7a, 0x81f6},
+		},
+	},
+	{
+		"testdata/amd64-plan9-exec",
+		FileHeader{8},
+		[]*SectionHeader{
+			{"text", 0x4213, 0x28},
+			{"data", 0xa80, 0x423b},
+			{"syms", 0x2c8c, 0x4cbb},
+			{"spsz", 0x0, 0x7947},
+			{"pcsz", 0xca0, 0x7947},
+		},
+	},
+}
+
+func TestOpen(t *testing.T) {
+	for i := range fileTests {
+		tt := &fileTests[i]
+
+		f, err := Open(tt.file)
+		if err != nil {
+			t.Error(err)
+			continue
+		}
+		if !reflect.DeepEqual(f.FileHeader, tt.hdr) {
+			t.Errorf("open %s:\n\thave %#v\n\twant %#v\n", tt.file, f.FileHeader, tt.hdr)
+			continue
+		}
+
+		for i, sh := range f.Sections {
+			if i >= len(tt.sections) {
+				break
+			}
+			have := &sh.SectionHeader
+			want := tt.sections[i]
+			if !reflect.DeepEqual(have, want) {
+				t.Errorf("open %s, section %d:\n\thave %#v\n\twant %#v\n", tt.file, i, have, want)
+			}
+		}
+		tn := len(tt.sections)
+		fn := len(f.Sections)
+		if tn != fn {
+			t.Errorf("open %s: len(Sections) = %d, want %d", tt.file, fn, tn)
+		}
+	}
+}
+
+func TestOpenFailure(t *testing.T) {
+	filename := "file.go"    // not a Plan 9 a.out file
+	_, err := Open(filename) // don't crash
+	if err == nil {
+		t.Errorf("open %s: succeeded unexpectedly", filename)
+	}
+}

src/pkg/debug/plan9obj/plan9obj.go

+// Copyright 2014 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.
+
+/*
+ * Plan 9 a.out constants and data structures
+ */
+
+package plan9obj
+
+import (
+	"bytes"
+	"encoding/binary"
+)
+
+// Plan 9 Program header.
+type prog struct {
+	Magic uint32 /* magic number */
+	Text  uint32 /* size of text segment */
+	Data  uint32 /* size of initialized data */
+	Bss   uint32 /* size of uninitialized data */
+	Syms  uint32 /* size of symbol table */
+	Entry uint32 /* entry point */
+	Spsz  uint32 /* size of pc/sp offset table */
+	Pcsz  uint32 /* size of pc/line number table */
+}
+
+// Plan 9 symbol table entries.
+type sym struct {
+	value uint64
+	typ   byte
+	name  []byte
+}
+
+const (
+	hsize      = 4 * 8
+	_HDR_MAGIC = 0x00008000 /* header expansion */
+)
+
+func magic(f, b int) string {
+	buf := new(bytes.Buffer)
+	var i uint32 = uint32((f) | ((((4 * (b)) + 0) * (b)) + 7))
+	binary.Write(buf, binary.BigEndian, i)
+	return string(buf.Bytes())
+}
+
+var (
+	_A_MAGIC = magic(0, 8)           /* 68020 (retired) */
+	_I_MAGIC = magic(0, 11)          /* intel 386 */
+	_J_MAGIC = magic(0, 12)          /* intel 960 (retired) */
+	_K_MAGIC = magic(0, 13)          /* sparc */
+	_V_MAGIC = magic(0, 16)          /* mips 3000 BE */
+	_X_MAGIC = magic(0, 17)          /* att dsp 3210 (retired) */
+	_M_MAGIC = magic(0, 18)          /* mips 4000 BE */
+	_D_MAGIC = magic(0, 19)          /* amd 29000 (retired) */
+	_E_MAGIC = magic(0, 20)          /* arm */
+	_Q_MAGIC = magic(0, 21)          /* powerpc */
+	_N_MAGIC = magic(0, 22)          /* mips 4000 LE */
+	_L_MAGIC = magic(0, 23)          /* dec alpha (retired) */
+	_P_MAGIC = magic(0, 24)          /* mips 3000 LE */
+	_U_MAGIC = magic(0, 25)          /* sparc64 (retired) */
+	_S_MAGIC = magic(_HDR_MAGIC, 26) /* amd64 */
+	_T_MAGIC = magic(_HDR_MAGIC, 27) /* powerpc64 */
+	_R_MAGIC = magic(_HDR_MAGIC, 28) /* arm64 */
+)
+
+type ExecTable struct {
+	Magic string
+	Ptrsz int
+	Hsize uint32
+}
+
+var exectab = []ExecTable{
+	{_A_MAGIC, 4, hsize},
+	{_I_MAGIC, 4, hsize},
+	{_J_MAGIC, 4, hsize},
+	{_K_MAGIC, 4, hsize},
+	{_V_MAGIC, 4, hsize},
+	{_X_MAGIC, 4, hsize},
+	{_M_MAGIC, 4, hsize},
+	{_D_MAGIC, 4, hsize},
+	{_E_MAGIC, 4, hsize},
+	{_Q_MAGIC, 4, hsize},
+	{_N_MAGIC, 4, hsize},
+	{_L_MAGIC, 4, hsize},
+	{_P_MAGIC, 4, hsize},
+	{_U_MAGIC, 4, hsize},
+	{_S_MAGIC, 8, hsize + 8},
+	{_T_MAGIC, 8, hsize + 8},
+	{_R_MAGIC, 8, hsize + 8},
+}

src/pkg/debug/plan9obj/testdata/hello.c

+#include <u.h>
+#include <libc.h>
+
+void
+main(void)
+{
+	print("hello, world\n");
+}