debug/dwarf: don't try to parse addr/rnglists header

In an executable, the debug_addr and debug_rnglists sections are
assembled by concatenating the input sections, and each input section
has a header, and each header may have different attributes. So just
parsing the single header isn't right.  Parsing the header is not
necessary to handle offsets into these sections which is all we do.

Looking at the header is also problematic because GCC with
-gsplit-dwarf when using DWARF versions 2 through 4 emits a
.debug_addr section, but it has no header.  The header was only added
for DWARF 5. So we can't parse the header at all for that case, and we
can't even detect that case in general.

This CL also fixes SeekPC with addrx and strx formats, by not using
the wrong compilation unit to find the address or string base.
To make that work when parsing the compilation unit itself, we add
support for delay the resolution of those values until we know the base.

New test binaries built with

gcc -gdwarf-5 -no-pie debug/dwarf/testdata/line[12].c
(gcc (Debian 10.2.0-15) 10.2.0)

clang -gdwarf-5 -no-pie debug/dwarf/testdata/line[12].c
(clang version 9.0.1-14)

Change-Id: I66783e0eded629bf80c467767f781164d344a54d
Reviewed-on: https://go-review.googlesource.com/c/go/+/277233
Trust: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Than McIntosh <thanm@google.com>

src/debug/dwarf/dwarf5ranges_test.go

@@ -22,7 +22,12 @@ func TestDwarf5Ranges(t *testing.T) {
 	if err := d.AddSection(".debug_rnglists", rngLists); err != nil {
 		t.Fatal(err)
 	}
-	ret, err := d.dwarf5Ranges(nil, 0x5fbd, 0xc, [][2]uint64{})
+	u := &unit{
+		asize: 8,
+		vers:  5,
+		is64:  true,
+	}
+	ret, err := d.dwarf5Ranges(u, nil, 0x5fbd, 0xc, [][2]uint64{})
 	if err != nil {
 		t.Fatalf("could not read rnglist: %v", err)
 	}

src/debug/dwarf/entry.go

@@ -423,6 +423,47 @@ func (b *buf) entry(cu *Entry, atab abbrevTable, ubase Offset, vers int) *Entry
 		Children: a.children,
 		Field:    make([]Field, len(a.field)),
 	}
+
+	// If we are currently parsing the compilation unit,
+	// we can't evaluate Addrx or Strx until we've seen the
+	// relevant base entry.
+	type delayed struct {
+		idx int
+		off uint64
+		fmt format
+	}
+	var delay []delayed
+
+	resolveStrx := func(strBase, off uint64) string {
+		off += strBase
+		if uint64(int(off)) != off {
+			b.error("DW_FORM_strx offset out of range")
+		}
+
+		b1 := makeBuf(b.dwarf, b.format, "str_offsets", 0, b.dwarf.strOffsets)
+		b1.skip(int(off))
+		is64, _ := b.format.dwarf64()
+		if is64 {
+			off = b1.uint64()
+		} else {
+			off = uint64(b1.uint32())
+		}
+		if b1.err != nil {
+			b.err = b1.err
+			return ""
+		}
+		if uint64(int(off)) != off {
+			b.error("DW_FORM_strx indirect offset out of range")
+		}
+		b1 = makeBuf(b.dwarf, b.format, "str", 0, b.dwarf.str)
+		b1.skip(int(off))
+		val := b1.string()
+		if b1.err != nil {
+			b.err = b1.err
+		}
+		return val
+	}
+
 	for i := range e.Field {
 		e.Field[i].Attr = a.field[i].attr
 		e.Field[i].Class = a.field[i].class
@@ -467,10 +508,13 @@ func (b *buf) entry(cu *Entry, atab abbrevTable, ubase Offset, vers int) *Entry
 			var addrBase int64
 			if cu != nil {
 				addrBase, _ = cu.Val(AttrAddrBase).(int64)
+			} else if a.tag == TagCompileUnit {
+				delay = append(delay, delayed{i, off, formAddrx})
+				break
 			}
 
 			var err error
-			val, err = b.dwarf.debugAddr(uint64(addrBase), off)
+			val, err = b.dwarf.debugAddr(b.format, uint64(addrBase), off)
 			if err != nil {
 				if b.err == nil {
 					b.err = err
@@ -611,38 +655,16 @@ func (b *buf) entry(cu *Entry, atab abbrevTable, ubase Offset, vers int) *Entry
 			// compilation unit. This won't work if the
 			// program uses Reader.Seek to skip over the
 			// unit. Not much we can do about that.
+			var strBase int64
 			if cu != nil {
-				cuOff, ok := cu.Val(AttrStrOffsetsBase).(int64)
-				if ok {
-					off += uint64(cuOff)
-				}
+				strBase, _ = cu.Val(AttrStrOffsetsBase).(int64)
+			} else if a.tag == TagCompileUnit {
+				delay = append(delay, delayed{i, off, formStrx})
+				break
 			}
 
-			if uint64(int(off)) != off {
-				b.error("DW_FORM_strx offset out of range")
-			}
+			val = resolveStrx(uint64(strBase), off)
 
-			b1 := makeBuf(b.dwarf, b.format, "str_offsets", 0, b.dwarf.strOffsets)
-			b1.skip(int(off))
-			if is64 {
-				off = b1.uint64()
-			} else {
-				off = uint64(b1.uint32())
-			}
-			if b1.err != nil {
-				b.err = b1.err
-				return nil
-			}
-			if uint64(int(off)) != off {
-				b.error("DW_FORM_strx indirect offset out of range")
-			}
-			b1 = makeBuf(b.dwarf, b.format, "str", 0, b.dwarf.str)
-			b1.skip(int(off))
-			val = b1.string()
-			if b1.err != nil {
-				b.err = b1.err
-				return nil
-			}
 		case formStrpSup:
 			is64, known := b.format.dwarf64()
 			if !known {
@@ -689,11 +711,32 @@ func (b *buf) entry(cu *Entry, atab abbrevTable, ubase Offset, vers int) *Entry
 		case formRnglistx:
 			val = b.uint()
 		}
+
 		e.Field[i].Val = val
 	}
 	if b.err != nil {
 		return nil
 	}
+
+	for _, del := range delay {
+		switch del.fmt {
+		case formAddrx:
+			addrBase, _ := e.Val(AttrAddrBase).(int64)
+			val, err := b.dwarf.debugAddr(b.format, uint64(addrBase), del.off)
+			if err != nil {
+				b.err = err
+				return nil
+			}
+			e.Field[del.idx].Val = val
+		case formStrx:
+			strBase, _ := e.Val(AttrStrOffsetsBase).(int64)
+			e.Field[del.idx].Val = resolveStrx(uint64(strBase), del.off)
+			if b.err != nil {
+				return nil
+			}
+		}
+	}
+
 	return e
 }
 
@@ -877,6 +920,7 @@ func (r *Reader) SeekPC(pc uint64) (*Entry, error) {
 		r.err = nil
 		r.lastChildren = false
 		r.unit = unit
+		r.cu = nil
 		u := &r.d.unit[unit]
 		r.b = makeBuf(r.d, u, "info", u.off, u.data)
 		e, err := r.Next()
@@ -946,7 +990,7 @@ func (d *Data) Ranges(e *Entry) ([][2]uint64, error) {
 			if err != nil {
 				return nil, err
 			}
-			return d.dwarf5Ranges(cu, base, ranges, ret)
+			return d.dwarf5Ranges(u, cu, base, ranges, ret)
 
 		case ClassRngList:
 			// TODO: support DW_FORM_rnglistx
@@ -1023,13 +1067,13 @@ func (d *Data) dwarf2Ranges(u *unit, base uint64, ranges int64, ret [][2]uint64)
 
 // dwarf5Ranges interpets a debug_rnglists sequence, see DWARFv5 section
 // 2.17.3 (page 53).
-func (d *Data) dwarf5Ranges(cu *Entry, base uint64, ranges int64, ret [][2]uint64) ([][2]uint64, error) {
+func (d *Data) dwarf5Ranges(u *unit, cu *Entry, base uint64, ranges int64, ret [][2]uint64) ([][2]uint64, error) {
 	var addrBase int64
 	if cu != nil {
 		addrBase, _ = cu.Val(AttrAddrBase).(int64)
 	}
 
-	buf := makeBuf(d, d.rngLists, "rnglists", 0, d.rngLists.data)
+	buf := makeBuf(d, u, "rnglists", 0, d.rngLists)
 	buf.skip(int(ranges))
 	for {
 		opcode := buf.uint8()
@@ -1043,7 +1087,7 @@ func (d *Data) dwarf5Ranges(cu *Entry, base uint64, ranges int64, ret [][2]uint6
 		case rleBaseAddressx:
 			baseIdx := buf.uint()
 			var err error
-			base, err = d.debugAddr(uint64(addrBase), baseIdx)
+			base, err = d.debugAddr(u, uint64(addrBase), baseIdx)
 			if err != nil {
 				return nil, err
 			}
@@ -1052,11 +1096,11 @@ func (d *Data) dwarf5Ranges(cu *Entry, base uint64, ranges int64, ret [][2]uint6
 			startIdx := buf.uint()
 			endIdx := buf.uint()
 
-			start, err := d.debugAddr(uint64(addrBase), startIdx)
+			start, err := d.debugAddr(u, uint64(addrBase), startIdx)
 			if err != nil {
 				return nil, err
 			}
-			end, err := d.debugAddr(uint64(addrBase), endIdx)
+			end, err := d.debugAddr(u, uint64(addrBase), endIdx)
 			if err != nil {
 				return nil, err
 			}
@@ -1065,7 +1109,7 @@ func (d *Data) dwarf5Ranges(cu *Entry, base uint64, ranges int64, ret [][2]uint6
 		case rleStartxLength:
 			startIdx := buf.uint()
 			len := buf.uint()
-			start, err := d.debugAddr(uint64(addrBase), startIdx)
+			start, err := d.debugAddr(u, uint64(addrBase), startIdx)
 			if err != nil {
 				return nil, err
 			}
@@ -1093,19 +1137,18 @@ func (d *Data) dwarf5Ranges(cu *Entry, base uint64, ranges int64, ret [][2]uint6
 }
 
 // debugAddr returns the address at idx in debug_addr
-func (d *Data) debugAddr(addrBase, idx uint64) (uint64, error) {
-	off := idx*uint64(d.addr.addrsize()) + addrBase
+func (d *Data) debugAddr(format dataFormat, addrBase, idx uint64) (uint64, error) {
+	off := idx*uint64(format.addrsize()) + addrBase
 
 	if uint64(int(off)) != off {
 		return 0, errors.New("offset out of range")
 	}
 
-	b := makeBuf(d, d.addr, "addr", 0, d.addr.data)
+	b := makeBuf(d, format, "addr", 0, d.addr)
 	b.skip(int(off))
 	val := b.addr()
 	if b.err != nil {
 		return 0, b.err
 	}
-
 	return val, nil
 }

src/debug/dwarf/entry_test.go

@@ -55,6 +55,20 @@ func TestReaderSeek(t *testing.T) {
 		{0x400611, nil},
 	}
 	testRanges(t, "testdata/line-gcc.elf", want)
+
+	want = []wantRange{
+		{0x401122, [][2]uint64{{0x401122, 0x401166}}},
+		{0x401165, [][2]uint64{{0x401122, 0x401166}}},
+		{0x401166, [][2]uint64{{0x401166, 0x401179}}},
+	}
+	testRanges(t, "testdata/line-gcc-dwarf5.elf", want)
+
+	want = []wantRange{
+		{0x401130, [][2]uint64{{0x401130, 0x40117e}}},
+		{0x40117d, [][2]uint64{{0x401130, 0x40117e}}},
+		{0x40117e, nil},
+	}
+	testRanges(t, "testdata/line-clang-dwarf5.elf", want)
 }
 
 func TestRangesSection(t *testing.T) {
@@ -97,44 +111,72 @@ func testRanges(t *testing.T, name string, want []wantRange) {
 }
 
 func TestReaderRanges(t *testing.T) {
-	d := elfData(t, "testdata/line-gcc.elf")
-
-	subprograms := []struct {
+	type subprograms []struct {
 		name   string
 		ranges [][2]uint64
+	}
+	tests := []struct {
+		filename    string
+		subprograms subprograms
 	}{
-		{"f1", [][2]uint64{{0x40059d, 0x4005e7}}},
-		{"main", [][2]uint64{{0x4005e7, 0x400601}}},
-		{"f2", [][2]uint64{{0x400601, 0x400611}}},
+		{
+			"testdata/line-gcc.elf",
+			subprograms{
+				{"f1", [][2]uint64{{0x40059d, 0x4005e7}}},
+				{"main", [][2]uint64{{0x4005e7, 0x400601}}},
+				{"f2", [][2]uint64{{0x400601, 0x400611}}},
+			},
+		},
+		{
+			"testdata/line-gcc-dwarf5.elf",
+			subprograms{
+				{"main", [][2]uint64{{0x401147, 0x401166}}},
+				{"f1", [][2]uint64{{0x401122, 0x401147}}},
+				{"f2", [][2]uint64{{0x401166, 0x401179}}},
+			},
+		},
+		{
+			"testdata/line-clang-dwarf5.elf",
+			subprograms{
+				{"main", [][2]uint64{{0x401130, 0x401144}}},
+				{"f1", [][2]uint64{{0x401150, 0x40117e}}},
+				{"f2", [][2]uint64{{0x401180, 0x401197}}},
+			},
+		},
 	}
 
-	r := d.Reader()
-	i := 0
-	for entry, err := r.Next(); entry != nil && err == nil; entry, err = r.Next() {
-		if entry.Tag != TagSubprogram {
-			continue
+	for _, test := range tests {
+		d := elfData(t, test.filename)
+		subprograms := test.subprograms
+
+		r := d.Reader()
+		i := 0
+		for entry, err := r.Next(); entry != nil && err == nil; entry, err = r.Next() {
+			if entry.Tag != TagSubprogram {
+				continue
+			}
+
+			if i > len(subprograms) {
+				t.Fatalf("%s: too many subprograms (expected at most %d)", test.filename, i)
+			}
+
+			if got := entry.Val(AttrName).(string); got != subprograms[i].name {
+				t.Errorf("%s: subprogram %d name is %s, expected %s", test.filename, i, got, subprograms[i].name)
+			}
+			ranges, err := d.Ranges(entry)
+			if err != nil {
+				t.Errorf("%s: subprogram %d: %v", test.filename, i, err)
+				continue
+			}
+			if !reflect.DeepEqual(ranges, subprograms[i].ranges) {
+				t.Errorf("%s: subprogram %d ranges are %x, expected %x", test.filename, i, ranges, subprograms[i].ranges)
+			}
+			i++
 		}
 
-		if i > len(subprograms) {
-			t.Fatalf("too many subprograms (expected at most %d)", i)
+		if i < len(subprograms) {
+			t.Errorf("%s: saw only %d subprograms, expected %d", test.filename, i, len(subprograms))
 		}
-
-		if got := entry.Val(AttrName).(string); got != subprograms[i].name {
-			t.Errorf("subprogram %d name is %s, expected %s", i, got, subprograms[i].name)
-		}
-		ranges, err := d.Ranges(entry)
-		if err != nil {
-			t.Errorf("subprogram %d: %v", i, err)
-			continue
-		}
-		if !reflect.DeepEqual(ranges, subprograms[i].ranges) {
-			t.Errorf("subprogram %d ranges are %x, expected %x", i, ranges, subprograms[i].ranges)
-		}
-		i++
-	}
-
-	if i < len(subprograms) {
-		t.Errorf("saw only %d subprograms, expected %d", i, len(subprograms))
 	}
 }
 

src/debug/dwarf/open.go

@@ -26,10 +26,10 @@ type Data struct {
 	str      []byte
 
 	// New sections added in DWARF 5.
-	addr       *debugAddr
+	addr       []byte
 	lineStr    []byte
 	strOffsets []byte
-	rngLists   *rngLists
+	rngLists   []byte
 
 	// parsed data
 	abbrevCache map[uint64]abbrevTable
@@ -40,21 +40,6 @@ type Data struct {
 	unit        []unit
 }
 
-// rngLists represents the contents of a debug_rnglists section (DWARFv5).
-type rngLists struct {
-	is64  bool
-	asize uint8
-	data  []byte
-	ver   uint16
-}
-
-// debugAddr represents the contents of a debug_addr section (DWARFv5).
-type debugAddr struct {
-	is64  bool
-	asize uint8
-	data  []byte
-}
-
 var errSegmentSelector = errors.New("non-zero segment_selector size not supported")
 
 // New returns a new Data object initialized from the given parameters.
@@ -132,76 +117,14 @@ func (d *Data) AddSection(name string, contents []byte) error {
 	var err error
 	switch name {
 	case ".debug_addr":
-		d.addr, err = d.parseAddrHeader(contents)
+		d.addr = contents
 	case ".debug_line_str":
 		d.lineStr = contents
 	case ".debug_str_offsets":
 		d.strOffsets = contents
 	case ".debug_rnglists":
-		d.rngLists, err = d.parseRngListsHeader(contents)
+		d.rngLists = contents
 	}
 	// Just ignore names that we don't yet support.
 	return err
 }
-
-// parseRngListsHeader reads the header of a debug_rnglists section, see
-// DWARFv5 section 7.28 (page 242).
-func (d *Data) parseRngListsHeader(bytes []byte) (*rngLists, error) {
-	rngLists := &rngLists{data: bytes}
-
-	buf := makeBuf(d, unknownFormat{}, "rnglists", 0, bytes)
-	_, rngLists.is64 = buf.unitLength()
-
-	rngLists.ver = buf.uint16() // version
-
-	rngLists.asize = buf.uint8()
-	segsize := buf.uint8()
-	if segsize != 0 {
-		return nil, errSegmentSelector
-	}
-
-	// Header fields not read: offset_entry_count, offset table
-
-	return rngLists, nil
-}
-
-func (rngLists *rngLists) version() int {
-	return int(rngLists.ver)
-}
-
-func (rngLists *rngLists) dwarf64() (bool, bool) {
-	return rngLists.is64, true
-}
-
-func (rngLists *rngLists) addrsize() int {
-	return int(rngLists.asize)
-}
-
-// parseAddrHeader reads the header of a debug_addr section, see DWARFv5
-// section 7.27 (page 241).
-func (d *Data) parseAddrHeader(bytes []byte) (*debugAddr, error) {
-	addr := &debugAddr{data: bytes}
-
-	buf := makeBuf(d, unknownFormat{}, "addr", 0, bytes)
-	_, addr.is64 = buf.unitLength()
-
-	addr.asize = buf.uint8()
-	segsize := buf.uint8()
-	if segsize != 0 {
-		return nil, errSegmentSelector
-	}
-
-	return addr, nil
-}
-
-func (addr *debugAddr) version() int {
-	return 5
-}
-
-func (addr *debugAddr) dwarf64() (bool, bool) {
-	return addr.is64, true
-}
-
-func (addr *debugAddr) addrsize() int {
-	return int(addr.asize)
-}