implement dwarf extraction

.golangci.yaml

@@ -104,12 +104,12 @@ linters-settings:
 
       - name: cognitive-complexity
         severity: warning
-        arguments: [ 50 ]
+        arguments: [ 300 ]
 
       - name: cyclomatic
-        arguments: [ 30 ]
+        arguments: [ 100 ]
 
-      - name: call-to-gc
+      - name: unchecked-type-assertion
         disabled: true
       - name: struct-tag
         disabled: true

README.md

@@ -178,7 +178,7 @@ Terminal interface options
 ## TODO
 
 - [ ] Add more pattern for disassembling the binary
-- [ ] Extract the information from the DWARF section
+- [x] Extract the information from the DWARF section
 - [x] Count the symbol size itself to package
 - [ ] Add other charts like flame graph, pie chart, etc.
 - [ ] Support C++/Rust symbol demangling in cgo

README_zh-CN.md

@@ -176,7 +176,7 @@ Terminal interface options
 ## TODO
 
 - [ ] 添加更多用于反汇编二进制文件的模式
-- [ ] 从 DWARF 段提取信息
+- [x] 从 DWARF 段提取信息
 - [x] 计算符号本身的大小到包中
 - [ ] 添加其他图表，如火焰图、饼图等
 - [ ] 支持 demangle cgo 中的 C++/Rust 符号

cmd/gsa/command.go

@@ -15,6 +15,7 @@ var Options struct {
 
 	NoDisasm bool `help:"Skip disassembly pass"`
 	NoSymbol bool `help:"Skip symbol pass"`
+	NoDwarf  bool `help:"Skip dwarf pass"`
 
 	HideSections bool `help:"Hide sections" group:"text"`
 	HideMain     bool `help:"Hide main package" group:"text"`

cmd/gsa/entry.go

@@ -40,6 +40,7 @@ func entry() error {
 		internal.Options{
 			SkipSymbol: Options.NoSymbol,
 			SkipDisasm: Options.NoDisasm,
+			SkipDwarf:  Options.NoDwarf,
 		})
 	if err != nil {
 		return err

go.mod

@@ -15,6 +15,7 @@ require (
 	github.com/charmbracelet/x/exp/teatest v0.0.0-20240525152034-77596eb8760e
 	github.com/charmbracelet/x/term v0.1.1
 	github.com/dustin/go-humanize v1.0.1
+	github.com/go-delve/delve v1.22.1
 	github.com/go-json-experiment/json v0.0.0-20240524174822-2d9f40f7385b
 	github.com/jedib0t/go-pretty/v6 v6.5.9
 	github.com/knadh/profiler v0.1.1

go.sum

@@ -39,6 +39,8 @@ github.com/dustin/go-humanize v1.0.1 h1:GzkhY7T5VNhEkwH0PVJgjz+fX1rhBrR7pRT3mDkp
 github.com/dustin/go-humanize v1.0.1/go.mod h1:Mu1zIs6XwVuF/gI1OepvI0qD18qycQx+mFykh5fBlto=
 github.com/erikgeiser/coninput v0.0.0-20211004153227-1c3628e74d0f h1:Y/CXytFA4m6baUTXGLOoWe4PQhGxaX0KpnayAqC48p4=
 github.com/erikgeiser/coninput v0.0.0-20211004153227-1c3628e74d0f/go.mod h1:vw97MGsxSvLiUE2X8qFplwetxpGLQrlU1Q9AUEIzCaM=
+github.com/go-delve/delve v1.22.1 h1:LQSF2sv+lP3mmOzMkadl5HGQGgSS2bFg2tbyALqHu8Y=
+github.com/go-delve/delve v1.22.1/go.mod h1:TfOb+G5H6YYKheZYAmA59ojoHbOimGfs5trbghHdLbM=
 github.com/go-json-experiment/json v0.0.0-20240524174822-2d9f40f7385b h1:IM96IiRXFcd7l+mU8Sys9pcggoBLbH/dEgzOESrS8F8=
 github.com/go-json-experiment/json v0.0.0-20240524174822-2d9f40f7385b/go.mod h1:uDEMZSTQMj7V6Lxdrx4ZwchmHEGdICbjuY+GQd7j9LM=
 github.com/hexops/gotextdiff v1.0.3 h1:gitA9+qJrrTCsiCl7+kh75nPqQt1cx4ZkudSTLoUqJM=

internal/analyze.go

@@ -10,13 +10,15 @@
 	"golang.org/x/exp/maps"
 
 	"github.com/Zxilly/go-size-analyzer/internal/entity"
+	"github.com/Zxilly/go-size-analyzer/internal/knowninfo"
 	"github.com/Zxilly/go-size-analyzer/internal/result"
 	"github.com/Zxilly/go-size-analyzer/internal/wrapper"
 )
 
 type Options struct {
 	SkipSymbol bool
 	SkipDisasm bool
+	SkipDwarf  bool
 }
 
 func Analyze(name string, reader io.ReaderAt, size uint64, options Options) (*result.Result, error) {
@@ -30,15 +32,15 @@
 	slog.Info("Parsing binary done")
 	slog.Info("Finding build info...")
 
-	k := &KnownInfo{
+	k := &knowninfo.KnownInfo{
 		Size:      size,
 		BuildInfo: file.BuildInfo,
 
-		gore:    file,
-		wrapper: wrapper.NewWrapper(file.GetParsedFile()),
+		Gore:    file,
+		Wrapper: wrapper.NewWrapper(file.GetParsedFile()),
 	}
 	k.KnownAddr = entity.NewKnownAddr()
-	k.UpdateVersionFlag()
+	k.VersionFlag = k.UpdateVersionFlag()
 
 	slog.Info("Build info found")
 
@@ -52,21 +54,32 @@
 		return nil, err
 	}
 
-	if !options.SkipSymbol {
-		err = k.AnalyzeSymbol()
-		if err != nil {
-			if !errors.Is(err, wrapper.ErrNoSymbolTable) {
-				return nil, err
+	dwarfOk := false
+	if !options.SkipDwarf {
+		dwarfOk = k.TryLoadDwarf()
+	}
+
+	// DWARF can still add new package
+	k.Deps.FinishLoad()
+
+	if !dwarfOk {
+		// fallback to symbol and disasm
+		if !options.SkipSymbol {
+			err = k.AnalyzeSymbol()
+			if err != nil {
+				if !errors.Is(err, wrapper.ErrNoSymbolTable) {
+					return nil, err
 
+				}
+				slog.Warn("Warning: no symbol table found, this can lead to inaccurate results")
 			}
-			slog.Warn("Warning: no symbol table found, this can lead to inaccurate results")
 		}
-	}
 
-	if !options.SkipDisasm {
-		err = k.Disasm()
-		if err != nil {
-			return nil, err
+		if !options.SkipDisasm {
+			err = k.Disasm()
+			if err != nil {
+				return nil, err
+			}
 		}
 	}
 

internal/disasm/interface_test.go

@@ -1,6 +1,7 @@
 package disasm
 
 import (
+	"debug/dwarf"
 	"errors"
 	"testing"
 
@@ -17,6 +18,10 @@ type TestFileWrapper struct {
 	textErr   error
 }
 
+func (TestFileWrapper) DWARF() (*dwarf.Data, error) {
+	panic("not reachable")
+}
+
 func (t TestFileWrapper) Text() (textStart uint64, text []byte, err error) {
 	return t.textStart, t.text, t.textErr
 }
@@ -26,19 +31,19 @@ func (t TestFileWrapper) GoArch() string {
 }
 
 func (TestFileWrapper) ReadAddr(_, _ uint64) ([]byte, error) {
-	panic("implement me")
+	panic("not reachable")
 }
 
 func (TestFileWrapper) LoadSymbols(_ func(name string, addr uint64, size uint64, typ entity.AddrType) error) error {
-	panic("implement me")
+	panic("not reachable")
 }
 
 func (TestFileWrapper) LoadSections() map[string]*entity.Section {
-	panic("implement me")
+	panic("not reachable")
 }
 
 func (TestFileWrapper) PclntabSections() []string {
-	panic("implement me")
+	panic("not reachable")
 }
 
 func TestNewExtractorNoText(t *testing.T) {

internal/entity/knownaddr.go

@@ -6,21 +6,21 @@
 )
 
 type KnownAddr struct {
-	Pclntab AddrSpace
+	Text AddrSpace
 
 	Symbol         AddrSpace
 	SymbolCoverage AddrCoverage
 }
 
 func NewKnownAddr() *KnownAddr {
 	return &KnownAddr{
-		Pclntab:        make(map[uint64]*Addr),
+		Text:           make(map[uint64]*Addr),
 		Symbol:         make(map[uint64]*Addr),
 		SymbolCoverage: make(AddrCoverage, 0),
 	}
 }
 
-func (f *KnownAddr) InsertPclntab(entry uint64, size uint64, fn *Function, meta GoPclntabMeta) {
+func (f *KnownAddr) InsertTextFromPclnTab(entry uint64, size uint64, fn *Function, meta GoPclntabMeta) {
 	cur := Addr{
 		AddrPos: &AddrPos{
 			Addr: entry,
@@ -33,7 +33,23 @@
 
 		Meta: meta,
 	}
-	f.Pclntab.Insert(&cur)
+	f.Text.Insert(&cur)
+}
+
+func (f *KnownAddr) InsertTextFromDWARF(entry uint64, size uint64, fn *Function, meta DwarfMeta) {
+	cur := Addr{
+		AddrPos: &AddrPos{
+			Addr: entry,
+			Size: size,
+			Type: AddrTypeText,
+		},
+		Pkg:        fn.pkg,
+		Function:   fn,
+		SourceType: AddrSourceDwarf,
+
+		Meta: meta,
+	}
+	f.Text.Insert(&cur)
 }
 
 func (f *KnownAddr) InsertSymbol(entry uint64, size uint64, p *Package, typ AddrType, meta SymbolMeta) *Addr {

internal/entity/meta.go

@@ -16,3 +16,6 @@ type SymbolMeta struct {
 type DisasmMeta struct {
 	Value string
 }
+
+type DwarfMeta struct {
+}

internal/entity/package.go

@@ -1,6 +1,7 @@
 package entity
 
 import (
+	"debug/dwarf"
 	"fmt"
 	"runtime/debug"
 
@@ -22,6 +23,7 @@ const (
 	PackageTypeVendor    PackageType = "vendor"
 	PackageTypeGenerated PackageType = "generated"
 	PackageTypeUnknown   PackageType = "unknown"
+	PackageTypeCGO       PackageType = "cgo"
 )
 
 type Package struct {
@@ -33,20 +35,23 @@ type Package struct {
 
 	Size uint64 `json:"size"` // late filled
 
+	filesCache map[string]*File
+	funcsCache map[string]*Function
+
 	loaded bool // mean it comes from gore
 
 	// should not be used to calculate size,
 	// since linker can create overlapped symbols.
 	// relies on coverage.
-	// currently only data symbol
 	Symbols []*Symbol `json:"symbols"`
 
 	symbolAddrSpace AddrSpace
 	coverage        *utils.ValueOnce[AddrCoverage]
 
-	// should have at least one of them
-	GorePkg  *gore.Package `json:"-"`
-	DebugMod *debug.Module `json:"-"`
+	// should have at least one of them, for cgo pesudo package all nil
+	GorePkg    *gore.Package `json:"-"`
+	DebugMod   *debug.Module `json:"-"`
+	DwarfEntry *dwarf.Entry  `json:"-"`
 }
 
 func NewPackage() *Package {
@@ -56,6 +61,8 @@ func NewPackage() *Package {
 		Symbols:         make([]*Symbol, 0),
 		coverage:        utils.NewOnce[AddrCoverage](),
 		symbolAddrSpace: AddrSpace{},
+		filesCache:      make(map[string]*File),
+		funcsCache:      make(map[string]*Function),
 	}
 }
 
@@ -96,15 +103,34 @@ func NewPackageWithGorePackage(gp *gore.Package, name string, typ PackageType, p
 }
 
 func (p *Package) fileEnsureUnique() {
-	seen := make(map[string]*File)
+	fileSeen := make(map[string]*File)
+
 	for _, f := range p.Files {
-		if old, ok := seen[f.FilePath]; ok {
-			old.Functions = append(old.Functions, f.Functions...)
+		if old, ok := fileSeen[f.FilePath]; ok {
+			funcSeen := make(map[string]*Function)
+			for _, fn := range old.Functions {
+				funcSeen[fn.Name] = fn
+			}
+
+			for _, fn := range f.Functions {
+				if _, ok := funcSeen[fn.Name]; !ok {
+					old.Functions = append(old.Functions, fn)
+				}
+			}
 		} else {
-			seen[f.FilePath] = f
+			fileSeen[f.FilePath] = f
+		}
+	}
+
+	p.Files = maps.Values(fileSeen)
+	p.filesCache = fileSeen
+
+	p.funcsCache = make(map[string]*Function)
+	for _, f := range p.Files {
+		for _, fn := range f.Functions {
+			p.funcsCache[fn.Name] = fn
 		}
 	}
-	p.Files = maps.Values(seen)
 }
 
 func (p *Package) addFunction(path string, fn *Function) {
@@ -115,11 +141,18 @@ func (p *Package) addFunction(path string, fn *Function) {
 	file.Functions = append(file.Functions, fn)
 }
 
+func (p *Package) AddFuncIfNotExists(path string, fn *Function) bool {
+	if _, ok := p.funcsCache[fn.Name]; !ok {
+		p.addFunction(path, fn)
+		p.funcsCache[fn.Name] = fn
+		return true
+	}
+	return false
+}
+
 func (p *Package) getOrInitFile(s string) *File {
-	for _, f := range p.Files {
-		if f.FilePath == s {
-			return f
-		}
+	if f, ok := p.filesCache[s]; ok {
+		return f
 	}
 
 	f := &File{
@@ -129,6 +162,7 @@ func (p *Package) getOrInitFile(s string) *File {
 	}
 
 	p.Files = append(p.Files, f)
+	p.filesCache[f.FilePath] = f
 	return f
 }
 

internal/entity/pcln_symbol.go

@@ -38,3 +38,9 @@ func NewPclnSymbolSize(s *gosym.Func) PclnSymbolSize {
 		PCData:   s.PCDataSize(),
 	}
 }
+
+func NewEmptyPclnSymbolSize() PclnSymbolSize {
+	return PclnSymbolSize{
+		PCData: make(map[string]int),
+	}
+}

internal/entity/type.go

@@ -14,4 +14,5 @@ const (
 	AddrSourceGoPclntab AddrSourceType = "pclntab"
 	AddrSourceSymbol    AddrSourceType = "symbol"
 	AddrSourceDisasm    AddrSourceType = "disasm"
+	AddrSourceDwarf     AddrSourceType = "dwarf"
 )