pclntab.go

// 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.

/*
 * Line tables
 */

package gosym

import (
	"bytes"
	"encoding/binary"
	"sort"
	"sync"
)

// version of the pclntab
type version int

const (
	verUnknown version = iota
	ver11
	ver12
	ver116
	ver118
	ver120
)

// A LineTable is a data structure mapping program counters to line numbers.
//
// In Go 1.1 and earlier, each function (represented by a Func) had its own LineTable,
// and the line number corresponded to a numbering of all source lines in the
// program, across all files. That absolute line number would then have to be
// converted separately to a file name and line number within the file.
//
// In Go 1.2, the format of the data changed so that there is a single LineTable
// for the entire program, shared by all Funcs, and there are no absolute line
// numbers, just line numbers within specific files.
//
// For the most part, LineTable's methods should be treated as an internal
// detail of the package; callers should use the methods on Table instead.
type LineTable struct {
	Data []byte
	PC   uint64
	Line int

	// This mutex is used to keep parsing of pclntab synchronous.
	mu sync.Mutex

	// Contains the version of the pclntab section.
	version version

	// Go 1.2/1.16/1.18 state
	binary      binary.ByteOrder
	quantum     uint32
	ptrsize     uint32
	textStart   uint64 // address of runtime.text symbol (1.18+)
	funcnametab []byte
	cutab       []byte
	funcdata    []byte
	functab     []byte
	nfunctab    uint32
	filetab     []byte
	pctab       []byte // points to the pctables.
	nfiletab    uint32
	funcNames   map[uint32]string // cache the function names
	strings     map[uint32]string // interned substrings of Data, keyed by offset
	// fileMap varies depending on the version of the object file.
	// For ver12, it maps the name to the index in the file table.
	// For ver116, it maps the name to the offset in filetab.
	fileMap map[string]uint32
}

// NOTE(rsc): This is wrong for GOARCH=arm, which uses a quantum of 4,
// but we have no idea whether we're using arm or not. This only
// matters in the old (pre-Go 1.2) symbol table format, so it's not worth
// fixing.
const oldQuantum = 1

func (t *LineTable) parse(targetPC uint64, targetLine int) (b []byte, pc uint64, line int) {
	// The PC/line table can be thought of as a sequence of
	//  <pc update>* <line update>
	// batches. Each update batch results in a (pc, line) pair,
	// where line applies to every PC from pc up to but not
	// including the pc of the next pair.
	//
	// Here we process each update individually, which simplifies
	// the code, but makes the corner cases more confusing.
	b, pc, line = t.Data, t.PC, t.Line
	for pc <= targetPC && line != targetLine && len(b) > 0 {
		code := b[0]
		b = b[1:]
		switch {
		case code == 0:
			if len(b) < 4 {
				b = b[0:0]
				break
			}
			val := binary.BigEndian.Uint32(b)
			b = b[4:]
			line += int(val)
		case code <= 64:
			line += int(code)
		case code <= 128:
			line -= int(code - 64)
		default:
			pc += oldQuantum * uint64(code-128)
			continue
		}
		pc += oldQuantum
	}
	return b, pc, line
}

func (t *LineTable) slice(pc uint64) *LineTable {
	data, pc, line := t.parse(pc, -1)
	return &LineTable{Data: data, PC: pc, Line: line}
}

// PCToLine returns the line number for the given program counter.
//
// Deprecated: Use Table's PCToLine method instead.
func (t *LineTable) PCToLine(pc uint64) int {
	if t.isGo12() {
		return t.go12PCToLine(pc)
	}
	_, _, line := t.parse(pc, -1)
	return line
}

// LineToPC returns the program counter for the given line number,
// considering only program counters before maxpc.
//
// Deprecated: Use Table's LineToPC method instead.
func (t *LineTable) LineToPC(line int, maxpc uint64) uint64 {
	if t.isGo12() {
		return 0
	}
	_, pc, line1 := t.parse(maxpc, line)
	if line1 != line {
		return 0
	}
	// Subtract quantum from PC to account for post-line increment
	return pc - oldQuantum
}

// NewLineTable returns a new PC/line table
// corresponding to the encoded data.
// Text must be the start address of the
// corresponding text segment.
func NewLineTable(data []byte, text uint64) *LineTable {
	return &LineTable{Data: data, PC: text, Line: 0, funcNames: make(map[uint32]string), strings: make(map[uint32]string)}
}

// Go 1.2 symbol table format.
// See golang.org/s/go12symtab.
//
// A general note about the methods here: rather than try to avoid
// index out of bounds errors, we trust Go to detect them, and then
// we recover from the panics and treat them as indicative of a malformed
// or incomplete table.
//
// The methods called by symtab.go, which begin with "go12" prefixes,
// are expected to have that recovery logic.

// isGo12 reports whether this is a Go 1.2 (or later) symbol table.
func (t *LineTable) isGo12() bool {
	t.parsePclnTab()
	return t.version >= ver12
}

const (
	go12magic  = 0xfffffffb
	go116magic = 0xfffffffa
	go118magic = 0xfffffff0
	go120magic = 0xfffffff1
)

// uintptr returns the pointer-sized value encoded at b.
// The pointer size is dictated by the table being read.
func (t *LineTable) uintptr(b []byte) uint64 {
	if t.ptrsize == 4 {
		return uint64(t.binary.Uint32(b))
	}
	return t.binary.Uint64(b)
}

// parsePclnTab parses the pclntab, setting the version.
func (t *LineTable) parsePclnTab() {
	t.mu.Lock()
	defer t.mu.Unlock()
	if t.version != verUnknown {
		return
	}

	// Note that during this function, setting the version is the last thing we do.
	// If we set the version too early, and parsing failed (likely as a panic on
	// slice lookups), we'd have a mistaken version.
	//
	// Error paths through this code will default the version to 1.1.
	t.version = ver11

	if !disableRecover {
		defer func() {
			// If we panic parsing, assume it's a Go 1.1 pclntab.
			_ = recover()
		}()
	}

	// Check header: 4-byte magic, two zeros, pc quantum, pointer size.
	if len(t.Data) < 16 || t.Data[4] != 0 || t.Data[5] != 0 ||
		(t.Data[6] != 1 && t.Data[6] != 2 && t.Data[6] != 4) || // pc quantum
		(t.Data[7] != 4 && t.Data[7] != 8) { // pointer size
		return
	}

	var possibleVersion version
	leMagic := binary.LittleEndian.Uint32(t.Data)
	beMagic := binary.BigEndian.Uint32(t.Data)
	switch {
	case leMagic == go12magic:
		t.binary, possibleVersion = binary.LittleEndian, ver12
	case beMagic == go12magic:
		t.binary, possibleVersion = binary.BigEndian, ver12
	case leMagic == go116magic:
		t.binary, possibleVersion = binary.LittleEndian, ver116
	case beMagic == go116magic:
		t.binary, possibleVersion = binary.BigEndian, ver116
	case leMagic == go118magic:
		t.binary, possibleVersion = binary.LittleEndian, ver118
	case beMagic == go118magic:
		t.binary, possibleVersion = binary.BigEndian, ver118
	case leMagic == go120magic:
		t.binary, possibleVersion = binary.LittleEndian, ver120
	case beMagic == go120magic:
		t.binary, possibleVersion = binary.BigEndian, ver120
	default:
		return
	}
	t.version = possibleVersion

	// quantum and ptrSize are the same between 1.2, 1.16, and 1.18
	t.quantum = uint32(t.Data[6])
	t.ptrsize = uint32(t.Data[7])

	offset := func(word uint32) uint64 {
		return t.uintptr(t.Data[8+word*t.ptrsize:])
	}
	data := func(word uint32) []byte {
		return t.Data[offset(word):]
	}

	switch possibleVersion {
	case ver118, ver120:
		t.nfunctab = uint32(offset(0))
		t.nfiletab = uint32(offset(1))
		t.textStart = t.PC // use the start PC instead of reading from the table, which may be unrelocated
		t.funcnametab = data(3)
		t.cutab = data(4)
		t.filetab = data(5)
		t.pctab = data(6)
		t.funcdata = data(7)
		t.functab = data(7)
		functabsize := (int(t.nfunctab)*2 + 1) * t.functabFieldSize()
		t.functab = t.functab[:functabsize]
	case ver116:
		t.nfunctab = uint32(offset(0))
		t.nfiletab = uint32(offset(1))
		t.funcnametab = data(2)
		t.cutab = data(3)
		t.filetab = data(4)
		t.pctab = data(5)
		t.funcdata = data(6)
		t.functab = data(6)
		functabsize := (int(t.nfunctab)*2 + 1) * t.functabFieldSize()
		t.functab = t.functab[:functabsize]
	case ver12:
		t.nfunctab = uint32(t.uintptr(t.Data[8:]))
		t.funcdata = t.Data
		t.funcnametab = t.Data
		t.functab = t.Data[8+t.ptrsize:]
		t.pctab = t.Data
		functabsize := (int(t.nfunctab)*2 + 1) * t.functabFieldSize()
		fileoff := t.binary.Uint32(t.functab[functabsize:])
		t.functab = t.functab[:functabsize]
		t.filetab = t.Data[fileoff:]
		t.nfiletab = t.binary.Uint32(t.filetab)
		t.filetab = t.filetab[:t.nfiletab*4]
	default:
		panic("unreachable")
	}
}

// go12Funcs returns a slice of Funcs derived from the Go 1.2+ pcln table.
func (t *LineTable) go12Funcs() []Func {
	// Assume it is malformed and return nil on error.
	if !disableRecover {
		defer func() {
			_ = recover()
		}()
	}

	ft := t.funcTab()
	funcs := make([]Func, ft.Count())
	syms := make([]Sym, len(funcs))
	for i := range funcs {
		f := &funcs[i]
		f.Entry = ft.pc(i)
		f.End = ft.pc(i + 1)
		info := t.funcData(uint32(i))
		f.LineTable = t
		f.FrameSize = int(info.deferreturn())

		// Additions:
		// numFuncField is the number of (32 bit) fields in _func (src/runtime/runtime2.go)
		// Note that the last 4 fields are 32 bits combined. This number is 11 for go1.20,
		// 10 for earlier versions down to go1.16, and 9 before that.
		var numFuncFields uint32 = 11
		if t.version < ver116 {
			numFuncFields = 9
		} else if t.version < ver120 {
			numFuncFields = 10
		}
		f.inlineTreeOffset = info.funcdataOffset(funcdata_InlTree, numFuncFields)
		f.inlineTreeCount = 1 + t.maxInlineTreeIndexValue(info, numFuncFields)

		syms[i] = Sym{
			Value:     f.Entry,
			Type:      'T',
			Name:      t.funcName(info.nameOff()),
			GoType:    0,
			Func:      f,
			goVersion: t.version,
		}
		f.Sym = &syms[i]
	}
	return funcs
}

// findFunc returns the funcData corresponding to the given program counter.
func (t *LineTable) findFunc(pc uint64) funcData {
	ft := t.funcTab()
	if pc < ft.pc(0) || pc >= ft.pc(ft.Count()) {
		return funcData{}
	}
	idx := sort.Search(int(t.nfunctab), func(i int) bool {
		return ft.pc(i) > pc
	})
	idx--
	return t.funcData(uint32(idx))
}

// readvarint reads, removes, and returns a varint from *pp.
func (t *LineTable) readvarint(pp *[]byte) uint32 {
	var v, shift uint32
	p := *pp
	for shift = 0; ; shift += 7 {
		b := p[0]
		p = p[1:]
		v |= (uint32(b) & 0x7F) << shift
		if b&0x80 == 0 {
			break
		}
	}
	*pp = p
	return v
}

// funcName returns the name of the function found at off.
func (t *LineTable) funcName(off uint32) string {
	if s, ok := t.funcNames[off]; ok {
		return s
	}
	i := bytes.IndexByte(t.funcnametab[off:], 0)
	s := string(t.funcnametab[off : off+uint32(i)])
	t.funcNames[off] = s
	return s
}

// stringFrom returns a Go string found at off from a position.
func (t *LineTable) stringFrom(arr []byte, off uint32) string {
	if s, ok := t.strings[off]; ok {
		return s
	}
	i := bytes.IndexByte(arr[off:], 0)
	s := string(arr[off : off+uint32(i)])
	t.strings[off] = s
	return s
}

// string returns a Go string found at off.
func (t *LineTable) string(off uint32) string {
	return t.stringFrom(t.funcdata, off)
}

// functabFieldSize returns the size in bytes of a single functab field.
func (t *LineTable) functabFieldSize() int {
	if t.version >= ver118 {
		return 4
	}
	return int(t.ptrsize)
}

// funcTab returns t's funcTab.
func (t *LineTable) funcTab() funcTab {
	return funcTab{LineTable: t, sz: t.functabFieldSize()}
}

// funcTab is memory corresponding to a slice of functab structs, followed by an invalid PC.
// A functab struct is a PC and a func offset.
type funcTab struct {
	*LineTable
	sz int // cached result of t.functabFieldSize
}

// Count returns the number of func entries in f.
func (f funcTab) Count() int {
	return int(f.nfunctab)
}

// pc returns the PC of the i'th func in f.
func (f funcTab) pc(i int) uint64 {
	u := f.uint(f.functab[2*i*f.sz:])
	if f.version >= ver118 {
		u += f.textStart
	}
	return u
}

// funcOff returns the funcdata offset of the i'th func in f.
func (f funcTab) funcOff(i int) uint64 {
	return f.uint(f.functab[(2*i+1)*f.sz:])
}

// uint returns the uint stored at b.
func (f funcTab) uint(b []byte) uint64 {
	if f.sz == 4 {
		return uint64(f.binary.Uint32(b))
	}
	return f.binary.Uint64(b)
}

// funcData is memory corresponding to an _func struct.
type funcData struct {
	t    *LineTable // LineTable this data is a part of
	data []byte     // raw memory for the function
}

// funcData returns the ith funcData in t.functab.
func (t *LineTable) funcData(i uint32) funcData {
	data := t.funcdata[t.funcTab().funcOff(int(i)):]
	return funcData{t: t, data: data}
}

// IsZero reports whether f is the zero value.
func (f funcData) IsZero() bool {
	return f.t == nil && f.data == nil
}

// entryPC returns the func's entry PC.
func (f *funcData) entryPC() uint64 {
	// In Go 1.18, the first field of _func changed
	// from a uintptr entry PC to a uint32 entry offset.
	if f.t.version >= ver118 {
		// TODO: support multiple text sections.
		// See runtime/symtab.go:(*moduledata).textAddr.
		return uint64(f.t.binary.Uint32(f.data)) + f.t.textStart
	}
	return f.t.uintptr(f.data)
}

func (f funcData) nameOff() uint32     { return f.field(1) }
func (f funcData) deferreturn() uint32 { return f.field(3) }
func (f funcData) pcfile() uint32      { return f.field(5) }
func (f funcData) pcln() uint32        { return f.field(6) }
func (f funcData) cuOffset() uint32    { return f.field(8) }

// field returns the nth field of the _func struct.
// It panics if n == 0 or n > 9; for n == 0, call f.entryPC.
// Most callers should use a named field accessor (just above).
func (f funcData) field(n uint32) uint32 {
	if n == 0 || n > 9 {
		panic("bad funcdata field")
	}
	// Addition: some code deleted here to support inlining.
	off := f.fieldOffset(n)
	data := f.data[off:]
	return f.t.binary.Uint32(data)
}

// step advances to the next pc, value pair in the encoded table.
func (t *LineTable) step(p *[]byte, pc *uint64, val *int32, first bool) bool {
	uvdelta := t.readvarint(p)
	if uvdelta == 0 && !first {
		return false
	}
	if uvdelta&1 != 0 {
		uvdelta = ^(uvdelta >> 1)
	} else {
		uvdelta >>= 1
	}
	vdelta := int32(uvdelta)
	pcdelta := t.readvarint(p) * t.quantum
	*pc += uint64(pcdelta)
	*val += vdelta
	return true
}

// pcvalue reports the value associated with the target pc.
// off is the offset to the beginning of the pc-value table,
// and entry is the start PC for the corresponding function.
func (t *LineTable) pcvalue(off uint32, entry, targetpc uint64) int32 {
	p := t.pctab[off:]

	val := int32(-1)
	pc := entry
	for t.step(&p, &pc, &val, pc == entry) {
		if targetpc < pc {
			return val
		}
	}
	return -1
}

// findFileLine scans one function in the binary looking for a
// program counter in the given file on the given line.
// It does so by running the pc-value tables mapping program counter
// to file number. Since most functions come from a single file, these
// are usually short and quick to scan. If a file match is found, then the
// code goes to the expense of looking for a simultaneous line number match.
func (t *LineTable) findFileLine(entry uint64, filetab, linetab uint32, filenum, line int32, cutab []byte) uint64 {
	if filetab == 0 || linetab == 0 {
		return 0
	}

	fp := t.pctab[filetab:]
	fl := t.pctab[linetab:]
	fileVal := int32(-1)
	filePC := entry
	lineVal := int32(-1)
	linePC := entry
	fileStartPC := filePC
	for t.step(&fp, &filePC, &fileVal, filePC == entry) {
		fileIndex := fileVal
		if t.version == ver116 || t.version == ver118 || t.version == ver120 {
			fileIndex = int32(t.binary.Uint32(cutab[fileVal*4:]))
		}
		if fileIndex == filenum && fileStartPC < filePC {
			// fileIndex is in effect starting at fileStartPC up to
			// but not including filePC, and it's the file we want.
			// Run the PC table looking for a matching line number
			// or until we reach filePC.
			lineStartPC := linePC
			for linePC < filePC && t.step(&fl, &linePC, &lineVal, linePC == entry) {
				// lineVal is in effect until linePC, and lineStartPC < filePC.
				if lineVal == line {
					if fileStartPC <= lineStartPC {
						return lineStartPC
					}
					if fileStartPC < linePC {
						return fileStartPC
					}
				}
				lineStartPC = linePC
			}
		}
		fileStartPC = filePC
	}
	return 0
}

// go12PCToLine maps program counter to line number for the Go 1.2+ pcln table.
func (t *LineTable) go12PCToLine(pc uint64) (line int) {
	defer func() {
		if !disableRecover && recover() != nil {
			line = -1
		}
	}()

	f := t.findFunc(pc)
	if f.IsZero() {
		return -1
	}
	entry := f.entryPC()
	linetab := f.pcln()
	return int(t.pcvalue(linetab, entry, pc))
}

// go12PCToFile maps program counter to file name for the Go 1.2+ pcln table.
func (t *LineTable) go12PCToFile(pc uint64) (file string) {
	defer func() {
		if !disableRecover && recover() != nil {
			file = ""
		}
	}()

	f := t.findFunc(pc)
	if f.IsZero() {
		return ""
	}
	entry := f.entryPC()
	filetab := f.pcfile()
	fno := t.pcvalue(filetab, entry, pc)
	if t.version == ver12 {
		if fno <= 0 {
			return ""
		}
		return t.string(t.binary.Uint32(t.filetab[4*fno:]))
	}
	// Go ≥ 1.16
	if fno < 0 { // 0 is valid for ≥ 1.16
		return ""
	}
	cuoff := f.cuOffset()
	if fnoff := t.binary.Uint32(t.cutab[(cuoff+uint32(fno))*4:]); fnoff != ^uint32(0) {
		return t.stringFrom(t.filetab, fnoff)
	}
	return ""
}

// go12LineToPC maps a (file, line) pair to a program counter for the Go 1.2+ pcln table.
func (t *LineTable) go12LineToPC(file string, line int) (pc uint64) {
	defer func() {
		if !disableRecover && recover() != nil {
			pc = 0
		}
	}()

	t.initFileMap()
	filenum, ok := t.fileMap[file]
	if !ok {
		return 0
	}

	// Scan all functions.
	// If this turns out to be a bottleneck, we could build a map[int32][]int32
	// mapping file number to a list of functions with code from that file.
	var cutab []byte
	for i := uint32(0); i < t.nfunctab; i++ {
		f := t.funcData(i)
		entry := f.entryPC()
		filetab := f.pcfile()
		linetab := f.pcln()
		if t.version == ver116 || t.version == ver118 || t.version == ver120 {
			if f.cuOffset() == ^uint32(0) {
				// skip functions without compilation unit (not real function, or linker generated)
				continue
			}
			cutab = t.cutab[f.cuOffset()*4:]
		}
		pc := t.findFileLine(entry, filetab, linetab, int32(filenum), int32(line), cutab)
		if pc != 0 {
			return pc
		}
	}
	return 0
}

// initFileMap initializes the map from file name to file number.
func (t *LineTable) initFileMap() {
	t.mu.Lock()
	defer t.mu.Unlock()

	if t.fileMap != nil {
		return
	}
	m := make(map[string]uint32)

	if t.version == ver12 {
		for i := uint32(1); i < t.nfiletab; i++ {
			s := t.string(t.binary.Uint32(t.filetab[4*i:]))
			m[s] = i
		}
	} else {
		var pos uint32
		for i := uint32(0); i < t.nfiletab; i++ {
			s := t.stringFrom(t.filetab, pos)
			m[s] = pos
			pos += uint32(len(s) + 1)
		}
	}
	t.fileMap = m
}

// go12MapFiles adds to m a key for every file in the Go 1.2 LineTable.
// Every key maps to obj. That's not a very interesting map, but it provides
// a way for callers to obtain the list of files in the program.
func (t *LineTable) go12MapFiles(m map[string]*Obj, obj *Obj) {
	if !disableRecover {
		defer func() {
			_ = recover()
		}()
	}

	t.initFileMap()
	for file := range t.fileMap {
		m[file] = obj
	}
}

// disableRecover causes this package not to swallow panics.
// This is useful when making changes.
const disableRecover = true