Add support for LLVM objdump (#534)

Add support for LLVM objdump.

internal/binutils/binutils.go

@@ -19,6 +19,7 @@ import (
 	"debug/elf"
 	"debug/macho"
 	"encoding/binary"
+	"errors"
 	"fmt"
 	"io"
 	"os"
@@ -26,6 +27,7 @@ import (
 	"path/filepath"
 	"regexp"
 	"runtime"
+	"strconv"
 	"strings"
 	"sync"
 
@@ -39,6 +41,8 @@ type Binutils struct {
 	rep *binrep
 }
 
+var objdumpLLVMVerRE = regexp.MustCompile(`LLVM version (?:(\d*)\.(\d*)\.(\d*)|.*(trunk).*)`)
+
 // binrep is an immutable representation for Binutils.  It is atomically
 // replaced on every mutation to provide thread-safe access.
 type binrep struct {
@@ -51,6 +55,7 @@ type binrep struct {
 	nmFound             bool
 	objdump             string
 	objdumpFound        bool
+	isLLVMObjdump       bool
 
 	// if fast, perform symbolization using nm (symbol names only),
 	// instead of file-line detail from the slower addr2line.
@@ -140,7 +145,77 @@ func initTools(b *binrep, config string) {
 		b.addr2line, b.addr2lineFound = findExe("gaddr2line", append(paths["addr2line"], defaultPath...))
 	}
 	b.nm, b.nmFound = findExe("nm", append(paths["nm"], defaultPath...))
-	b.objdump, b.objdumpFound = findExe("objdump", append(paths["objdump"], defaultPath...))
+	b.objdump, b.objdumpFound, b.isLLVMObjdump = findObjdump(append(paths["objdump"], defaultPath...))
+}
+
+// findObjdump finds and returns path to preferred objdump binary.
+// Order of preference is: llvm-objdump, objdump.
+// On MacOS only, also looks for gobjdump with least preference.
+// Accepts a list of paths and returns:
+// a string with path to the preferred objdump binary if found,
+// or an empty string if not found;
+// a boolean if any acceptable objdump was found;
+// a boolen indicating if it is an LLVM objdump.
+func findObjdump(paths []string) (string, bool, bool) {
+	objdumpNames := []string{"llvm-objdump", "objdump"}
+	if runtime.GOOS == "darwin" {
+		objdumpNames = append(objdumpNames, "gobjdump")
+	}
+
+	for _, objdumpName := range objdumpNames {
+		if objdump, objdumpFound := findExe(objdumpName, paths); objdumpFound {
+			cmdOut, err := exec.Command(objdump, "--version").Output()
+			if err != nil {
+				continue
+			}
+			if isLLVMObjdump(string(cmdOut)) {
+				return objdump, true, true
+			}
+			if isBuObjdump(string(cmdOut)) {
+				return objdump, true, false
+			}
+		}
+	}
+	return "", false, false
+}
+
+// isLLVMObjdump accepts a string with path to an objdump binary,
+// and returns a boolean indicating if the given binary is an LLVM
+// objdump binary of an acceptable version.
+func isLLVMObjdump(output string) bool {
+	fields := objdumpLLVMVerRE.FindStringSubmatch(output)
+	if len(fields) != 5 {
+		return false
+	}
+	if fields[4] == "trunk" {
+		return true
+	}
+	verMajor, err := strconv.Atoi(fields[1])
+	if err != nil {
+		return false
+	}
+	verPatch, err := strconv.Atoi(fields[3])
+	if err != nil {
+		return false
+	}
+	if runtime.GOOS == "linux" && verMajor >= 8 {
+		// Ensure LLVM objdump is at least version 8.0 on Linux.
+		// Some flags, like --demangle, and double dashes for options are
+		// not supported by previous versions.
+		return true
+	}
+	if runtime.GOOS == "darwin" {
+		// Ensure LLVM objdump is at least version 10.0.1 on MacOS.
+		return verMajor > 10 || (verMajor == 10 && verPatch >= 1)
+	}
+	return false
+}
+
+// isBuObjdump accepts a string with path to an objdump binary,
+// and returns a boolean indicating if the given binary is a GNU
+// binutils objdump binary. No version check is performed.
+func isBuObjdump(output string) bool {
+	return strings.Contains(output, "GNU objdump") && strings.Contains(output, "Binutils")
 }
 
 // findExe looks for an executable command on a set of paths.
@@ -159,13 +234,16 @@ func findExe(cmd string, paths []string) (string, bool) {
 // of a binary.
 func (bu *Binutils) Disasm(file string, start, end uint64, intelSyntax bool) ([]plugin.Inst, error) {
 	b := bu.get()
-	args := []string{"-d", "-C", "--no-show-raw-insn", "-l",
-		fmt.Sprintf("--start-address=%#x", start),
+	if !b.objdumpFound {
+		return nil, errors.New("cannot disasm: no objdump tool available")
+	}
+	args := []string{"--disassemble-all", "--demangle", "--no-show-raw-insn",
+		"--line-numbers", fmt.Sprintf("--start-address=%#x", start),
 		fmt.Sprintf("--stop-address=%#x", end)}
 
 	if intelSyntax {
-		if runtime.GOOS == "darwin" {
-			args = append(args, "-x86-asm-syntax=intel")
+		if b.isLLVMObjdump {
+			args = append(args, "--x86-asm-syntax=intel")
 		} else {
 			args = append(args, "-M", "intel")
 		}

internal/binutils/binutils_test.go

@@ -189,14 +189,25 @@ func skipUnlessDarwinAmd64(t *testing.T) {
 }
 
 func testDisasm(t *testing.T, intelSyntax bool) {
+	_, llvmObjdump, buObjdump := findObjdump([]string{""})
+	if !(llvmObjdump || buObjdump) {
+		t.Skip("cannot disasm: no objdump tool available")
+	}
+
 	bu := &Binutils{}
-	insts, err := bu.Disasm(filepath.Join("testdata", "exe_linux_64"), 0, math.MaxUint64, intelSyntax)
+	testexe := "exe_linux_64"
+	if runtime.GOOS == "darwin" {
+		testexe = "exe_mac_64"
+	}
+
+	insts, err := bu.Disasm(filepath.Join("testdata", testexe), 0, math.MaxUint64, intelSyntax)
 	if err != nil {
 		t.Fatalf("Disasm: unexpected error %v", err)
 	}
 	mainCount := 0
 	for _, x := range insts {
-		if x.Function == "main" {
+		// Mac symbols have a leading underscore.
+		if x.Function == "main" || x.Function == "_main" {
 			mainCount++
 		}
 	}
@@ -206,7 +217,9 @@ func testDisasm(t *testing.T, intelSyntax bool) {
 }
 
 func TestDisasm(t *testing.T) {
-	skipUnlessLinuxAmd64(t)
+	if runtime.GOOS != "linux" && runtime.GOOS != "darwin" {
+		t.Skip("This test only works on Linux or Mac")
+	}
 	testDisasm(t, true)
 	testDisasm(t, false)
 }
@@ -401,3 +414,92 @@ func TestOpenMalformedMachO(t *testing.T) {
 		t.Errorf("Open: got %v, want error containing 'Mach-O'", err)
 	}
 }
+
+func TestObjdumpVersionChecks(t *testing.T) {
+	// Test that the objdump version strings are parsed properly.
+	type testcase struct {
+		desc string
+		os   string
+		ver  string
+		want bool
+	}
+
+	for _, tc := range []testcase{
+		{
+			desc: "Valid Apple LLVM version string with usable version",
+			os:   "darwin",
+			ver:  "Apple LLVM version 11.0.3 (clang-1103.0.32.62)\nOptimized build.",
+			want: true,
+		},
+		{
+			desc: "Valid Apple LLVM version string with unusable version",
+			os:   "darwin",
+			ver:  "Apple LLVM version 10.0.0 (clang-1000.11.45.5)\nOptimized build.",
+			want: false,
+		},
+		{
+			desc: "Invalid Apple LLVM version string with usable version",
+			os:   "darwin",
+			ver:  "Apple LLVM versions 11.0.3 (clang-1103.0.32.62)\nOptimized build.",
+			want: false,
+		},
+		{
+			desc: "Valid LLVM version string with usable version",
+			os:   "linux",
+			ver:  "LLVM (http://llvm.org/):\nLLVM version 9.0.1\n\nOptimized build.",
+			want: true,
+		},
+		{
+			desc: "Valid LLVM version string with unusable version",
+			os:   "linux",
+			ver:  "LLVM (http://llvm.org/):\nLLVM version 6.0.1\n\nOptimized build.",
+			want: false,
+		},
+		{
+			desc: "Invalid LLVM version string with usable version",
+			os:   "linux",
+			ver:  "LLVM (http://llvm.org/):\nLLVM versions 9.0.1\n\nOptimized build.",
+			want: false,
+		},
+		{
+			desc: "Valid LLVM objdump version string with trunk",
+			os:   runtime.GOOS,
+			ver:  "LLVM (http://llvm.org/):\nLLVM version custom-trunk 124ffeb592a00bfe\nOptimized build.",
+			want: true,
+		},
+		{
+			desc: "Invalid LLVM objdump version string with trunk",
+			os:   runtime.GOOS,
+			ver:  "LLVM (http://llvm.org/):\nLLVM version custom-trank 124ffeb592a00bfe\nOptimized build.",
+			want: false,
+		},
+		{
+			desc: "Invalid LLVM objdump version string with trunk",
+			os:   runtime.GOOS,
+			ver:  "LLVM (http://llvm.org/):\nllvm version custom-trunk 124ffeb592a00bfe\nOptimized build.",
+			want: false,
+		},
+	} {
+		if runtime.GOOS == tc.os {
+			if got := isLLVMObjdump(tc.ver); got != tc.want {
+				t.Errorf("%v: got %v, want %v", tc.desc, got, tc.want)
+			}
+		}
+	}
+	for _, tc := range []testcase{
+		{
+			desc: "Valid GNU objdump version string",
+			ver:  "GNU objdump (GNU Binutils) 2.34\nCopyright (C) 2020 Free Software Foundation, Inc.",
+			want: true,
+		},
+		{
+			desc: "Invalid GNU objdump version string",
+			ver:  "GNU objdump (GNU Banutils) 2.34\nCopyright (C) 2020 Free Software Foundation, Inc.",
+			want: false,
+		},
+	} {
+		if got := isBuObjdump(tc.ver); got != tc.want {
+			t.Errorf("%v: got %v, want %v", tc.desc, got, tc.want)
+		}
+	}
+}

internal/binutils/disasm.go

@@ -25,10 +25,11 @@ import (
 )
 
 var (
-	nmOutputRE            = regexp.MustCompile(`^\s*([[:xdigit:]]+)\s+(.)\s+(.*)`)
-	objdumpAsmOutputRE    = regexp.MustCompile(`^\s*([[:xdigit:]]+):\s+(.*)`)
-	objdumpOutputFileLine = regexp.MustCompile(`^(.*):([0-9]+)`)
-	objdumpOutputFunction = regexp.MustCompile(`^(\S.*)\(\):`)
+	nmOutputRE                = regexp.MustCompile(`^\s*([[:xdigit:]]+)\s+(.)\s+(.*)`)
+	objdumpAsmOutputRE        = regexp.MustCompile(`^\s*([[:xdigit:]]+):\s+(.*)`)
+	objdumpOutputFileLine     = regexp.MustCompile(`^;?\s?(.*):([0-9]+)`)
+	objdumpOutputFunction     = regexp.MustCompile(`^;?\s?(\S.*)\(\):`)
+	objdumpOutputFunctionLLVM = regexp.MustCompile(`^([[:xdigit:]]+)?\s?(.*):`)
 )
 
 func findSymbols(syms []byte, file string, r *regexp.Regexp, address uint64) ([]*plugin.Sym, error) {
@@ -143,6 +144,11 @@ func disassemble(asm []byte) ([]plugin.Inst, error) {
 		if fields := objdumpOutputFunction.FindStringSubmatch(input); len(fields) == 2 {
 			function = fields[1]
 			continue
+		} else {
+			if fields := objdumpOutputFunctionLLVM.FindStringSubmatch(input); len(fields) == 3 {
+				function = fields[2]
+				continue
+			}
 		}
 		// Reset on unrecognized lines.
 		function, file, line = "", "", 0

internal/binutils/disasm_test.go

@@ -110,11 +110,7 @@ func TestFunctionAssembly(t *testing.T) {
 	testcases := []testcase{
 		{
 			plugin.Sym{Name: []string{"symbol1"}, Start: 0x1000, End: 0x1FFF},
-			`  1000: instruction one
-  1001: instruction two
-  1002: instruction three
-  1003: instruction four
-`,
+			"  1000: instruction one\n  1001: instruction two\n  1002: instruction three\n  1003: instruction four",
 			[]plugin.Inst{
 				{Addr: 0x1000, Text: "instruction one"},
 				{Addr: 0x1001, Text: "instruction two"},
@@ -124,14 +120,26 @@ func TestFunctionAssembly(t *testing.T) {
 		},
 		{
 			plugin.Sym{Name: []string{"symbol2"}, Start: 0x2000, End: 0x2FFF},
-			`  2000: instruction one
-  2001: instruction two
-`,
+			"  2000: instruction one\n  2001: instruction two",
 			[]plugin.Inst{
 				{Addr: 0x2000, Text: "instruction one"},
 				{Addr: 0x2001, Text: "instruction two"},
 			},
 		},
+		{
+			plugin.Sym{Name: []string{"_main"}, Start: 0x30000, End: 0x3FFF},
+			"_main:\n; /tmp/hello.c:3\n30001:	push   %rbp",
+			[]plugin.Inst{
+				{Addr: 0x30001, Text: "push   %rbp", Function: "_main", File: "/tmp/hello.c", Line: 3},
+			},
+		},
+		{
+			plugin.Sym{Name: []string{"main"}, Start: 0x4000, End: 0x4FFF},
+			"000000000040052d <main>:\nmain():\n/tmp/hello.c:3\n40001:	push   %rbp",
+			[]plugin.Inst{
+				{Addr: 0x40001, Text: "push   %rbp", Function: "main", File: "/tmp/hello.c", Line: 3},
+			},
+		},
 	}
 
 	for _, tc := range testcases {