From 791ed32c133174adb975715f01b90c9b846da283 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Mikk=20Margus=20M=C3=B6ll?= <mikk@you.die.ifyou.work>
Date: Sat, 10 Feb 2024 15:08:15 +0200
Subject: [PATCH] use DWARF info, if available, to get goroot and goversion

---
 elf.go       | 156 +++++++++++++++++++++++++++++++++++++++++++++++++++
 goroot.go    |   9 +++
 goversion.go |  10 ++++
 3 files changed, 175 insertions(+)

diff --git a/elf.go b/elf.go
index b5fb75e..ad6d3ed 100644
--- a/elf.go
+++ b/elf.go
@@ -18,13 +18,169 @@
 package gore
 
 import (
+	"bytes"
+	"debug/dwarf"
 	"debug/elf"
 	"debug/gosym"
+	"encoding/binary"
 	"errors"
 	"fmt"
 	"os"
 )
 
+const (
+	// official DWARF language ID for Go
+	// https://dwarfstd.org/languages.html
+	dwLangGo int64 = 0x0016
+
+	// DWARF operation; used to encode type offsets
+	dwOpAddr = 0x03
+)
+
+func (e *elfFile) getGoRootFromDwarf() (string, bool) {
+	return getStringFromDwarf(e, getDwarfStringCheck("runtime.defaultGOROOT"))
+}
+
+func (e *elfFile) getBuildVersionFromDwarf() (string, bool) {
+	return getStringFromDwarf(e, getDwarfStringCheck("runtime.buildVersion"))
+}
+
+// DWARF entry plus any associated children
+type dwarfEntryPlus struct {
+	entry    *dwarf.Entry
+	children []*dwarfEntryPlus
+}
+
+type dwarfwalkStatus uint8
+
+const (
+	dwStop dwarfwalkStatus = iota + 1
+	dwContinue
+	dwFound
+)
+
+func getStringFromDwarf(e *elfFile, check func(e *elfFile, entry *dwarfEntryPlus) (string, dwarfwalkStatus)) (string, bool) {
+	dw, err := e.file.DWARF()
+	if err != nil {
+		return "", false
+	}
+	r := dw.Reader()
+	// walk through compilation units
+getValOuter:
+	for cu := dwarfReadEntry(r); cu != nil; cu = dwarfReadEntry(r) {
+		if langField := cu.entry.AttrField(dwarf.AttrLanguage); langField == nil || langField.Val != dwLangGo {
+			continue
+		}
+	getValInner:
+		for _, entry := range cu.children {
+			ret, status := check(e, entry)
+			switch status {
+			case dwStop:
+				break getValOuter
+			case dwFound:
+				return ret, true
+			case dwContinue:
+				continue getValInner
+			}
+		}
+	}
+	return "", false
+}
+
+// get, by name, a DWARF entry corresponding to a string constant
+func getDwarfStringCheck(name string) func(e *elfFile, entry *dwarfEntryPlus) (string, dwarfwalkStatus) {
+	return func(e *elfFile, d *dwarfEntryPlus) (string, dwarfwalkStatus) {
+		entry := d.entry
+		nameField := entry.AttrField(dwarf.AttrName)
+		if nameField == nil {
+			return "", dwContinue
+		}
+
+		if fieldName := nameField.Val.(string); fieldName != name {
+			return "", dwContinue
+		}
+
+		return commonStringCheck(e, entry)
+	}
+}
+
+func commonStringCheck(e *elfFile, entry *dwarf.Entry) (string, dwarfwalkStatus) {
+	locationField := entry.AttrField(dwarf.AttrLocation)
+	if locationField == nil {
+		// unexpected failure
+		return "", dwStop
+	}
+	location := locationField.Val.([]byte)
+	// DWARF address operation followed by the machine byte order encoded address
+	if location[0] != dwOpAddr {
+		return "", dwStop
+	}
+	var addr uint64
+	if e.file.Class == elf.ELFCLASS32 {
+		addr = uint64(e.file.ByteOrder.Uint32(location[1:]))
+	} else {
+		addr = e.file.ByteOrder.Uint64(location[1:])
+	}
+
+	sectionBase, data, err := e.getSectionDataFromAddress(addr)
+	if err != nil {
+		return "", dwStop
+	}
+	off := addr - sectionBase
+	r := bytes.NewReader(data[off:])
+	var stringData [2]uint64
+	if e.file.Class == elf.ELFCLASS32 {
+		var stringData32 [2]uint32
+		err = binary.Read(r, e.file.ByteOrder, &stringData32)
+		if err != nil {
+			return "", dwStop
+		}
+		stringData[0] = uint64(stringData32[0])
+		stringData[1] = uint64(stringData32[1])
+	} else {
+		err = binary.Read(r, e.file.ByteOrder, &stringData)
+		if err != nil {
+			return "", dwStop
+		}
+	}
+	addr = stringData[0]
+	stringLen := stringData[1]
+	sectionBase, data, err = e.getSectionDataFromAddress(addr)
+	if err != nil {
+		return "", dwStop
+	}
+	off = addr - sectionBase
+	raw := data[off : off+stringLen]
+	return string(raw), dwFound
+}
+
+func dwarfReadEntry(r *dwarf.Reader) *dwarfEntryPlus {
+	entry, _ := r.Next()
+	if entry == nil {
+		return nil
+	}
+	var children []*dwarfEntryPlus
+	if entry.Children {
+		children = dwarfReadChildren(r)
+	}
+	return &dwarfEntryPlus{
+		entry:    entry,
+		children: children,
+	}
+}
+
+func dwarfReadChildren(r *dwarf.Reader) []*dwarfEntryPlus {
+	var ret []*dwarfEntryPlus
+
+	for {
+		e := dwarfReadEntry(r)
+		if e.entry.Tag == 0 {
+			return ret
+		}
+		ret = append(ret, e)
+	}
+}
+
 func openELF(fp string) (*elfFile, error) {
 	osFile, err := os.Open(fp)
 	if err != nil {
diff --git a/goroot.go b/goroot.go
index 6d5efde..b42272c 100644
--- a/goroot.go
+++ b/goroot.go
@@ -327,6 +327,15 @@ pkgLoop:
 
 func findGoRootPath(f *GoFile) (string, error) {
 	var goroot string
+
+	// if DWARF debug info exists, then this can simply be obtained from there
+	if e, ok := f.fh.(*elfFile); ok {
+		goroot, ok = e.getGoRootFromDwarf()
+		if ok {
+			return goroot, nil
+		}
+	}
+
 	// There is no GOROOT function may be inlined (after go1.16)
 	// at this time GOROOT is obtained through time_init function
 	goroot, err := tryFromGOROOT(f)
diff --git a/goversion.go b/goversion.go
index eeccfd0..94e4110 100644
--- a/goversion.go
+++ b/goversion.go
@@ -79,6 +79,16 @@ func buildSemVerString(v string) string {
 }
 
 func findGoCompilerVersion(f *GoFile) (*GoVersion, error) {
+	// if DWARF debug info exists, then this can simply be obtained from there
+	if e, ok := f.fh.(*elfFile); ok {
+		s, ok := e.getBuildVersionFromDwarf()
+		if ok {
+			if ver := ResolveGoVersion(s); ver != nil {
+				return ver, nil
+			}
+		}
+	}
+
 	// Try to determine the version based on the schedinit function.
 	if v := tryFromSchedInit(f); v != nil {
 		return v, nil