Commit 6de8b4bc3d

William Sengir <william@sengir.com>
2022-03-15 10:07:46
std.dwarf: implement basic DWARF 5 parsing
DWARF 5 moves around some fields and adds a few new ones that can't be parsed or ignored by our current DWARF 4 parser. This isn't a complete implementation of DWARF 5, but this is enough to make stack traces mostly work. Line numbers from C++ don't show up, but I know the info is there. I think the answer is to iterate through .debug_line_str in getLineNumberInfo, but I didn't want to fall into an even deeper rabbit hole tonight.
1 parent 47e004d
lib/std/dwarf/AT.zig
@@ -99,7 +99,35 @@ pub const enum_class = 0x6d;
 pub const linkage_name = 0x6e;
 
 // DWARF 5
+pub const string_length_bit_size = 0x6f;
+pub const string_length_byte_size = 0x70;
+pub const rank = 0x71;
+pub const str_offsets_base = 0x72;
+pub const addr_base = 0x73;
+pub const rnglists_base = 0x74;
+pub const dwo_name = 0x76;
+pub const reference = 0x77;
+pub const rvalue_reference = 0x78;
+pub const macros = 0x79;
+pub const call_all_calls = 0x7a;
+pub const call_all_source_calls = 0x7b;
+pub const call_all_tail_calls = 0x7c;
+pub const call_return_pc = 0x7d;
+pub const call_value = 0x7e;
+pub const call_origin = 0x7f;
+pub const call_parameter = 0x80;
+pub const call_pc = 0x81;
+pub const call_tail_call = 0x82;
+pub const call_target = 0x83;
+pub const call_target_clobbered = 0x84;
+pub const call_data_location = 0x85;
+pub const call_data_value = 0x86;
+pub const @"noreturn" = 0x87;
 pub const alignment = 0x88;
+pub const export_symbols = 0x89;
+pub const deleted = 0x8a;
+pub const defaulted = 0x8b;
+pub const loclists_base = 0x8c;
 
 pub const lo_user = 0x2000; // Implementation-defined range start.
 pub const hi_user = 0x3fff; // Implementation-defined range end.
lib/std/dwarf/ATE.zig
@@ -0,0 +1,46 @@
+pub const @"void" = 0x0;
+pub const address = 0x1;
+pub const boolean = 0x2;
+pub const complex_float = 0x3;
+pub const float = 0x4;
+pub const signed = 0x5;
+pub const signed_char = 0x6;
+pub const unsigned = 0x7;
+pub const unsigned_char = 0x8;
+
+// DWARF 3.
+pub const imaginary_float = 0x9;
+pub const packed_decimal = 0xa;
+pub const numeric_string = 0xb;
+pub const edited = 0xc;
+pub const signed_fixed = 0xd;
+pub const unsigned_fixed = 0xe;
+pub const decimal_float = 0xf;
+
+// DWARF 4.
+pub const UTF = 0x10;
+
+// DWARF 5.
+pub const UCS = 0x11;
+pub const ASCII = 0x12;
+
+pub const lo_user = 0x80;
+pub const hi_user = 0xff;
+
+// HP extensions.
+pub const HP_float80 = 0x80; // Floating-point (80 bit).
+pub const HP_complex_float80 = 0x81; // Complex floating-point (80 bit).
+pub const HP_float128 = 0x82; // Floating-point (128 bit).
+pub const HP_complex_float128 = 0x83; // Complex fp (128 bit).
+pub const HP_floathpintel = 0x84; // Floating-point (82 bit IA64).
+pub const HP_imaginary_float80 = 0x85;
+pub const HP_imaginary_float128 = 0x86;
+pub const HP_VAX_float = 0x88; // F or G floating.
+pub const HP_VAX_float_d = 0x89; // D floating.
+pub const HP_packed_decimal = 0x8a; // Cobol.
+pub const HP_zoned_decimal = 0x8b; // Cobol.
+pub const HP_edited = 0x8c; // Cobol.
+pub const HP_signed_fixed = 0x8d; // Cobol.
+pub const HP_unsigned_fixed = 0x8e; // Cobol.
+pub const HP_VAX_complex_float = 0x8f; // F or G floating complex.
+pub const HP_VAX_complex_float_d = 0x90; // D floating complex.
lib/std/dwarf/FORM.zig
@@ -0,0 +1,52 @@
+pub const addr = 0x01;
+pub const block2 = 0x03;
+pub const block4 = 0x04;
+pub const data2 = 0x05;
+pub const data4 = 0x06;
+pub const data8 = 0x07;
+pub const string = 0x08;
+pub const block = 0x09;
+pub const block1 = 0x0a;
+pub const data1 = 0x0b;
+pub const flag = 0x0c;
+pub const sdata = 0x0d;
+pub const strp = 0x0e;
+pub const udata = 0x0f;
+pub const ref_addr = 0x10;
+pub const ref1 = 0x11;
+pub const ref2 = 0x12;
+pub const ref4 = 0x13;
+pub const ref8 = 0x14;
+pub const ref_udata = 0x15;
+pub const indirect = 0x16;
+pub const sec_offset = 0x17;
+pub const exprloc = 0x18;
+pub const flag_present = 0x19;
+pub const strx = 0x1a;
+pub const addrx = 0x1b;
+pub const ref_sup4 = 0x1c;
+pub const strp_sup = 0x1d;
+pub const data16 = 0x1e;
+pub const line_strp = 0x1f;
+pub const ref_sig8 = 0x20;
+pub const implicit_const = 0x21;
+pub const loclistx = 0x22;
+pub const rnglistx = 0x23;
+pub const ref_sup8 = 0x24;
+pub const strx1 = 0x25;
+pub const strx2 = 0x26;
+pub const strx3 = 0x27;
+pub const strx4 = 0x28;
+pub const addrx1 = 0x29;
+pub const addrx2 = 0x2a;
+pub const addrx3 = 0x2b;
+pub const addrx4 = 0x2c;
+
+// Extensions for Fission.  See http://gcc.gnu.org/wiki/DebugFission.
+pub const GNU_addr_index = 0x1f01;
+pub const GNU_str_index = 0x1f02;
+
+// Extensions for DWZ multifile.
+// See http://www.dwarfstd.org/ShowIssue.php?issue=120604.1&type=open .
+pub const GNU_ref_alt = 0x1f20;
+pub const GNU_strp_alt = 0x1f21;
lib/std/dwarf/LANG.zig
@@ -0,0 +1,48 @@
+pub const C89 = 0x0001;
+pub const C = 0x0002;
+pub const Ada83 = 0x0003;
+pub const C_plus_plus = 0x0004;
+pub const Cobol74 = 0x0005;
+pub const Cobol85 = 0x0006;
+pub const Fortran77 = 0x0007;
+pub const Fortran90 = 0x0008;
+pub const Pascal83 = 0x0009;
+pub const Modula2 = 0x000a;
+pub const Java = 0x000b;
+pub const C99 = 0x000c;
+pub const Ada95 = 0x000d;
+pub const Fortran95 = 0x000e;
+pub const PLI = 0x000f;
+pub const ObjC = 0x0010;
+pub const ObjC_plus_plus = 0x0011;
+pub const UPC = 0x0012;
+pub const D = 0x0013;
+pub const Python = 0x0014;
+pub const OpenCL = 0x0015;
+pub const Go = 0x0016;
+pub const Modula3 = 0x0017;
+pub const Haskell = 0x0018;
+pub const C_plus_plus_03 = 0x0019;
+pub const C_plus_plus_11 = 0x001a;
+pub const OCaml = 0x001b;
+pub const Rust = 0x001c;
+pub const C11 = 0x001d;
+pub const Swift = 0x001e;
+pub const Julia = 0x001f;
+pub const Dylan = 0x0020;
+pub const C_plus_plus_14 = 0x0021;
+pub const Fortran03 = 0x0022;
+pub const Fortran08 = 0x0023;
+pub const RenderScript = 0x0024;
+pub const BLISS = 0x0025;
+
+pub const lo_user = 0x8000;
+pub const hi_user = 0xffff;
+
+pub const Mips_Assembler = 0x8001;
+pub const Upc = 0x8765;
+pub const HP_Bliss = 0x8003;
+pub const HP_Basic91 = 0x8004;
+pub const HP_Pascal91 = 0x8005;
+pub const HP_IMacro = 0x8006;
+pub const HP_Assembler = 0x8007;
lib/std/dwarf/OP.zig
@@ -157,6 +157,18 @@ pub const bit_piece = 0x9d;
 pub const implicit_value = 0x9e;
 pub const stack_value = 0x9f;
 
+// DWARF 5 extensions.
+pub const implicit_pointer = 0xa0;
+pub const addrx = 0xa1;
+pub const constx = 0xa2;
+pub const entry_value = 0xa3;
+pub const const_type = 0xa4;
+pub const regval_type = 0xa5;
+pub const deref_type = 0xa6;
+pub const xderef_type = 0xa7;
+pub const convert = 0xa8;
+pub const reinterpret = 0xa9;
+
 pub const lo_user = 0xe0; // Implementation-defined range start.
 pub const hi_user = 0xff; // Implementation-defined range end.
 
lib/std/dwarf/TAG.zig
@@ -65,6 +65,16 @@ pub const type_unit = 0x41;
 pub const rvalue_reference_type = 0x42;
 pub const template_alias = 0x43;
 
+// DWARF 5
+pub const coarray_type = 0x44;
+pub const generic_subrange = 0x45;
+pub const dynamic_type = 0x46;
+pub const atomic_type = 0x47;
+pub const call_site = 0x48;
+pub const call_site_parameter = 0x49;
+pub const skeleton_unit = 0x4a;
+pub const immutable_type = 0x4b;
+
 pub const lo_user = 0x4080;
 pub const hi_user = 0xffff;
 
lib/std/debug.zig
@@ -797,6 +797,7 @@ fn readCoffDebugInfo(allocator: mem.Allocator, coff_file: File) !ModuleDebugInfo
             const debug_abbrev_data = di.coff.getSectionData(".debug_abbrev", allocator) catch null;
             const debug_str_data = di.coff.getSectionData(".debug_str", allocator) catch null;
             const debug_line_data = di.coff.getSectionData(".debug_line", allocator) catch null;
+            const debug_line_str_data = di.coff.getSectionData(".debug_line_str", allocator) catch null;
             const debug_ranges_data = di.coff.getSectionData(".debug_ranges", allocator) catch null;
 
             var dwarf = DW.DwarfInfo{
@@ -805,6 +806,7 @@ fn readCoffDebugInfo(allocator: mem.Allocator, coff_file: File) !ModuleDebugInfo
                 .debug_abbrev = debug_abbrev_data orelse return error.MissingDebugInfo,
                 .debug_str = debug_str_data orelse return error.MissingDebugInfo,
                 .debug_line = debug_line_data orelse return error.MissingDebugInfo,
+                .debug_line_str = debug_line_str_data,
                 .debug_ranges = debug_ranges_data,
             };
             try DW.openDwarfDebugInfo(&dwarf, allocator);
@@ -871,6 +873,7 @@ pub fn readElfDebugInfo(allocator: mem.Allocator, elf_file: File) !ModuleDebugIn
         var opt_debug_abbrev: ?[]const u8 = null;
         var opt_debug_str: ?[]const u8 = null;
         var opt_debug_line: ?[]const u8 = null;
+        var opt_debug_line_str: ?[]const u8 = null;
         var opt_debug_ranges: ?[]const u8 = null;
 
         for (shdrs) |*shdr| {
@@ -885,6 +888,8 @@ pub fn readElfDebugInfo(allocator: mem.Allocator, elf_file: File) !ModuleDebugIn
                 opt_debug_str = try chopSlice(mapped_mem, shdr.sh_offset, shdr.sh_size);
             } else if (mem.eql(u8, name, ".debug_line")) {
                 opt_debug_line = try chopSlice(mapped_mem, shdr.sh_offset, shdr.sh_size);
+            } else if (mem.eql(u8, name, ".debug_line_str")) {
+                opt_debug_line_str = try chopSlice(mapped_mem, shdr.sh_offset, shdr.sh_size);
             } else if (mem.eql(u8, name, ".debug_ranges")) {
                 opt_debug_ranges = try chopSlice(mapped_mem, shdr.sh_offset, shdr.sh_size);
             }
@@ -896,6 +901,7 @@ pub fn readElfDebugInfo(allocator: mem.Allocator, elf_file: File) !ModuleDebugIn
             .debug_abbrev = opt_debug_abbrev orelse return error.MissingDebugInfo,
             .debug_str = opt_debug_str orelse return error.MissingDebugInfo,
             .debug_line = opt_debug_line orelse return error.MissingDebugInfo,
+            .debug_line_str = opt_debug_line_str,
             .debug_ranges = opt_debug_ranges,
         };
 
@@ -1434,6 +1440,7 @@ pub const ModuleDebugInfo = switch (native_os) {
             var opt_debug_info: ?*const macho.section_64 = null;
             var opt_debug_abbrev: ?*const macho.section_64 = null;
             var opt_debug_str: ?*const macho.section_64 = null;
+            var opt_debug_line_str: ?*const macho.section_64 = null;
             var opt_debug_ranges: ?*const macho.section_64 = null;
 
             const sections = @ptrCast(
@@ -1456,6 +1463,8 @@ pub const ModuleDebugInfo = switch (native_os) {
                     opt_debug_abbrev = sect;
                 } else if (mem.eql(u8, name, "__debug_str")) {
                     opt_debug_str = sect;
+                } else if (mem.eql(u8, name, "__debug_line_str")) {
+                    opt_debug_line_str = sect;
                 } else if (mem.eql(u8, name, "__debug_ranges")) {
                     opt_debug_ranges = sect;
                 }
@@ -1476,6 +1485,10 @@ pub const ModuleDebugInfo = switch (native_os) {
                 .debug_abbrev = try chopSlice(mapped_mem, debug_abbrev.offset, debug_abbrev.size),
                 .debug_str = try chopSlice(mapped_mem, debug_str.offset, debug_str.size),
                 .debug_line = try chopSlice(mapped_mem, debug_line.offset, debug_line.size),
+                .debug_line_str = if (opt_debug_line_str) |debug_line_str|
+                    try chopSlice(mapped_mem, debug_line_str.offset, debug_line_str.size)
+                else
+                    null,
                 .debug_ranges = if (opt_debug_ranges) |debug_ranges|
                     try chopSlice(mapped_mem, debug_ranges.offset, debug_ranges.size)
                 else
lib/std/dwarf.zig
@@ -11,87 +11,20 @@ const ArrayList = std.ArrayList;
 pub const TAG = @import("dwarf/TAG.zig");
 pub const AT = @import("dwarf/AT.zig");
 pub const OP = @import("dwarf/OP.zig");
-
-pub const FORM = struct {
-    pub const addr = 0x01;
-    pub const block2 = 0x03;
-    pub const block4 = 0x04;
-    pub const data2 = 0x05;
-    pub const data4 = 0x06;
-    pub const data8 = 0x07;
-    pub const string = 0x08;
-    pub const block = 0x09;
-    pub const block1 = 0x0a;
-    pub const data1 = 0x0b;
-    pub const flag = 0x0c;
-    pub const sdata = 0x0d;
-    pub const strp = 0x0e;
-    pub const udata = 0x0f;
-    pub const ref_addr = 0x10;
-    pub const ref1 = 0x11;
-    pub const ref2 = 0x12;
-    pub const ref4 = 0x13;
-    pub const ref8 = 0x14;
-    pub const ref_udata = 0x15;
-    pub const indirect = 0x16;
-    pub const sec_offset = 0x17;
-    pub const exprloc = 0x18;
-    pub const flag_present = 0x19;
-    pub const ref_sig8 = 0x20;
-
-    // Extensions for Fission.  See http://gcc.gnu.org/wiki/DebugFission.
-    pub const GNU_addr_index = 0x1f01;
-    pub const GNU_str_index = 0x1f02;
-
-    // Extensions for DWZ multifile.
-    // See http://www.dwarfstd.org/ShowIssue.php?issue=120604.1&type=open .
-    pub const GNU_ref_alt = 0x1f20;
-    pub const GNU_strp_alt = 0x1f21;
-};
-
-pub const ATE = struct {
-    pub const @"void" = 0x0;
-    pub const address = 0x1;
-    pub const boolean = 0x2;
-    pub const complex_float = 0x3;
-    pub const float = 0x4;
-    pub const signed = 0x5;
-    pub const signed_char = 0x6;
-    pub const unsigned = 0x7;
-    pub const unsigned_char = 0x8;
-
-    // DWARF 3.
-    pub const imaginary_float = 0x9;
-    pub const packed_decimal = 0xa;
-    pub const numeric_string = 0xb;
-    pub const edited = 0xc;
-    pub const signed_fixed = 0xd;
-    pub const unsigned_fixed = 0xe;
-    pub const decimal_float = 0xf;
-
-    // DWARF 4.
-    pub const UTF = 0x10;
-
-    pub const lo_user = 0x80;
-    pub const hi_user = 0xff;
-
-    // HP extensions.
-    pub const HP_float80 = 0x80; // Floating-point (80 bit).
-    pub const HP_complex_float80 = 0x81; // Complex floating-point (80 bit).
-    pub const HP_float128 = 0x82; // Floating-point (128 bit).
-    pub const HP_complex_float128 = 0x83; // Complex fp (128 bit).
-    pub const HP_floathpintel = 0x84; // Floating-point (82 bit IA64).
-    pub const HP_imaginary_float80 = 0x85;
-    pub const HP_imaginary_float128 = 0x86;
-    pub const HP_VAX_float = 0x88; // F or G floating.
-    pub const HP_VAX_float_d = 0x89; // D floating.
-    pub const HP_packed_decimal = 0x8a; // Cobol.
-    pub const HP_zoned_decimal = 0x8b; // Cobol.
-    pub const HP_edited = 0x8c; // Cobol.
-    pub const HP_signed_fixed = 0x8d; // Cobol.
-    pub const HP_unsigned_fixed = 0x8e; // Cobol.
-    pub const HP_VAX_complex_float = 0x8f; // F or G floating complex.
-    pub const HP_VAX_complex_float_d = 0x90; // D floating complex.
+pub const LANG = @import("dwarf/LANG.zig");
+pub const FORM = @import("dwarf/FORM.zig");
+pub const ATE = @import("dwarf/ATE.zig");
+
+pub const LLE = struct {
+    pub const end_of_list = 0x00;
+    pub const base_addressx = 0x01;
+    pub const startx_endx = 0x02;
+    pub const startx_length = 0x03;
+    pub const offset_pair = 0x04;
+    pub const default_location = 0x05;
+    pub const base_address = 0x06;
+    pub const start_end = 0x07;
+    pub const start_length = 0x08;
 };
 
 pub const CFA = struct {
@@ -166,45 +99,6 @@ pub const LNE = struct {
     pub const hi_user = 0xff;
 };
 
-pub const LANG = struct {
-    pub const C89 = 0x0001;
-    pub const C = 0x0002;
-    pub const Ada83 = 0x0003;
-    pub const C_plus_plus = 0x0004;
-    pub const Cobol74 = 0x0005;
-    pub const Cobol85 = 0x0006;
-    pub const Fortran77 = 0x0007;
-    pub const Fortran90 = 0x0008;
-    pub const Pascal83 = 0x0009;
-    pub const Modula2 = 0x000a;
-    pub const Java = 0x000b;
-    pub const C99 = 0x000c;
-    pub const Ada95 = 0x000d;
-    pub const Fortran95 = 0x000e;
-    pub const PLI = 0x000f;
-    pub const ObjC = 0x0010;
-    pub const ObjC_plus_plus = 0x0011;
-    pub const UPC = 0x0012;
-    pub const D = 0x0013;
-    pub const Python = 0x0014;
-    pub const Go = 0x0016;
-    pub const C_plus_plus_11 = 0x001a;
-    pub const Rust = 0x001c;
-    pub const C11 = 0x001d;
-    pub const C_plus_plus_14 = 0x0021;
-    pub const Fortran03 = 0x0022;
-    pub const Fortran08 = 0x0023;
-    pub const lo_user = 0x8000;
-    pub const hi_user = 0xffff;
-    pub const Mips_Assembler = 0x8001;
-    pub const Upc = 0x8765;
-    pub const HP_Bliss = 0x8003;
-    pub const HP_Basic91 = 0x8004;
-    pub const HP_Pascal91 = 0x8005;
-    pub const HP_IMacro = 0x8006;
-    pub const HP_Assembler = 0x8007;
-};
-
 pub const UT = struct {
     pub const compile = 0x01;
     pub const @"type" = 0x02;
@@ -212,6 +106,7 @@ pub const UT = struct {
     pub const skeleton = 0x04;
     pub const split_compile = 0x05;
     pub const split_type = 0x06;
+
     pub const lo_user = 0x80;
     pub const hi_user = 0xff;
 };
@@ -222,10 +117,22 @@ pub const LNCT = struct {
     pub const timestamp = 0x3;
     pub const size = 0x4;
     pub const MD5 = 0x5;
+
     pub const lo_user = 0x2000;
     pub const hi_user = 0x3fff;
 };
 
+pub const RLE = struct {
+    pub const end_of_list = 0x00;
+    pub const base_addressx = 0x01;
+    pub const startx_endx = 0x02;
+    pub const startx_length = 0x03;
+    pub const offset_pair = 0x04;
+    pub const base_address = 0x05;
+    pub const start_end = 0x06;
+    pub const start_length = 0x07;
+};
+
 pub const CC = enum(u8) {
     normal = 0x1,
     program = 0x2,
@@ -276,6 +183,8 @@ const AbbrevTableEntry = struct {
 const AbbrevAttr = struct {
     attr_id: u64,
     form_id: u64,
+    /// Only valid if form_id is .implicit_const
+    payload: i64,
 };
 
 const FormValue = union(enum) {
@@ -289,6 +198,7 @@ const FormValue = union(enum) {
     RefAddr: u64,
     String: []const u8,
     StrPtr: u64,
+    LineStrPtr: u64,
 };
 
 const Constant = struct {
@@ -356,6 +266,7 @@ const Die = struct {
         return switch (form_value.*) {
             FormValue.String => |value| value,
             FormValue.StrPtr => |offset| di.getString(offset),
+            FormValue.LineStrPtr => |offset| di.getLineString(offset),
             else => error.InvalidDebugInfo,
         };
     }
@@ -588,6 +499,7 @@ fn parseFormValue(allocator: mem.Allocator, in_stream: anytype, form_id: u64, en
 
         FORM.string => FormValue{ .String = try in_stream.readUntilDelimiterAlloc(allocator, 0, math.maxInt(usize)) },
         FORM.strp => FormValue{ .StrPtr = try readAddress(in_stream, endian, is_64) },
+        FORM.line_strp => FormValue{ .LineStrPtr = try readAddress(in_stream, endian, is_64) },
         FORM.indirect => {
             const child_form_id = try nosuspend leb.readULEB128(u64, in_stream);
             const F = @TypeOf(async parseFormValue(allocator, in_stream, child_form_id, endian, is_64));
@@ -595,7 +507,12 @@ fn parseFormValue(allocator: mem.Allocator, in_stream: anytype, form_id: u64, en
             defer allocator.destroy(frame);
             return await @asyncCall(frame, {}, parseFormValue, .{ allocator, in_stream, child_form_id, endian, is_64 });
         },
-        else => error.InvalidDebugInfo,
+        FORM.implicit_const => FormValue{ .Const = Constant{ .signed = true, .payload = undefined } },
+
+        else => {
+            std.debug.print("dwarf: unhandled form_id: 0x{x}\n", .{form_id});
+            return error.InvalidDebugInfo;
+        },
     };
 }
 
@@ -613,6 +530,7 @@ pub const DwarfInfo = struct {
     debug_abbrev: []const u8,
     debug_str: []const u8,
     debug_line: []const u8,
+    debug_line_str: ?[]const u8,
     debug_ranges: ?[]const u8,
     // Filled later by the initializer
     abbrev_table_list: ArrayList(AbbrevTableHeader) = undefined,
@@ -652,9 +570,20 @@ pub const DwarfInfo = struct {
             const version = try in.readInt(u16, di.endian);
             if (version < 2 or version > 5) return error.InvalidDebugInfo;
 
-            const debug_abbrev_offset = if (is_64) try in.readInt(u64, di.endian) else try in.readInt(u32, di.endian);
-
-            const address_size = try in.readByte();
+            var address_size: u8 = undefined;
+            var debug_abbrev_offset: u64 = undefined;
+            switch (version) {
+                5 => {
+                    const unit_type = try in.readInt(u8, di.endian);
+                    if (unit_type != UT.compile) return error.InvalidDebugInfo;
+                    address_size = try in.readByte();
+                    debug_abbrev_offset = if (is_64) try in.readInt(u64, di.endian) else try in.readInt(u32, di.endian);
+                },
+                else => {
+                    debug_abbrev_offset = if (is_64) try in.readInt(u64, di.endian) else try in.readInt(u32, di.endian);
+                    address_size = try in.readByte();
+                },
+            }
             if (address_size != @sizeOf(usize)) return error.InvalidDebugInfo;
 
             const compile_unit_pos = try seekable.getPos();
@@ -756,9 +685,20 @@ pub const DwarfInfo = struct {
             const version = try in.readInt(u16, di.endian);
             if (version < 2 or version > 5) return error.InvalidDebugInfo;
 
-            const debug_abbrev_offset = if (is_64) try in.readInt(u64, di.endian) else try in.readInt(u32, di.endian);
-
-            const address_size = try in.readByte();
+            var address_size: u8 = undefined;
+            var debug_abbrev_offset: u64 = undefined;
+            switch (version) {
+                5 => {
+                    const unit_type = try in.readInt(u8, di.endian);
+                    if (unit_type != UT.compile) return error.InvalidDebugInfo;
+                    address_size = try in.readByte();
+                    debug_abbrev_offset = if (is_64) try in.readInt(u64, di.endian) else try in.readInt(u32, di.endian);
+                },
+                else => {
+                    debug_abbrev_offset = if (is_64) try in.readInt(u64, di.endian) else try in.readInt(u32, di.endian);
+                    address_size = try in.readByte();
+                },
+            }
             if (address_size != @sizeOf(usize)) return error.InvalidDebugInfo;
 
             const compile_unit_pos = try seekable.getPos();
@@ -890,9 +830,12 @@ pub const DwarfInfo = struct {
                 const attr_id = try leb.readULEB128(u64, in);
                 const form_id = try leb.readULEB128(u64, in);
                 if (attr_id == 0 and form_id == 0) break;
+                // DW_FORM_implicit_const stores its value immediately after the attribute pair :(
+                const payload = if (form_id == FORM.implicit_const) try leb.readILEB128(i64, in) else undefined;
                 try attrs.append(AbbrevAttr{
                     .attr_id = attr_id,
                     .form_id = form_id,
+                    .payload = payload,
                 });
             }
         }
@@ -914,6 +857,9 @@ pub const DwarfInfo = struct {
                 .id = attr.attr_id,
                 .value = try parseFormValue(di.allocator(), in_stream, attr.form_id, di.endian, is_64),
             };
+            if (attr.form_id == FORM.implicit_const) {
+                result.attrs.items[i].value.Const.payload = @bitCast(u64, attr.payload);
+            }
         }
         return result;
     }
@@ -1101,6 +1047,21 @@ pub const DwarfInfo = struct {
 
         return error.InvalidDebugInfo;
     }
+
+    fn getLineString(di: *DwarfInfo, offset: u64) ![]const u8 {
+        const debug_line_str = di.debug_line_str orelse return error.InvalidDebugInfo;
+        if (offset > debug_line_str.len)
+            return error.InvalidDebugInfo;
+        const casted_offset = math.cast(usize, offset) catch
+            return error.InvalidDebugInfo;
+
+        // Valid strings always have a terminating zero byte
+        if (mem.indexOfScalarPos(u8, debug_line_str, casted_offset, 0)) |last| {
+            return debug_line_str[casted_offset..last];
+        }
+
+        return error.InvalidDebugInfo;
+    }
 };
 
 /// Initialize DWARF info. The caller has the responsibility to initialize most
src/link/MachO/Object.zig
@@ -38,6 +38,7 @@ dwarf_debug_info_index: ?u16 = null,
 dwarf_debug_abbrev_index: ?u16 = null,
 dwarf_debug_str_index: ?u16 = null,
 dwarf_debug_line_index: ?u16 = null,
+dwarf_debug_line_str_index: ?u16 = null,
 dwarf_debug_ranges_index: ?u16 = null,
 
 symtab: std.ArrayListUnmanaged(macho.nlist_64) = .{},
@@ -68,6 +69,7 @@ const DebugInfo = struct {
     debug_abbrev: []u8,
     debug_str: []u8,
     debug_line: []u8,
+    debug_line_str: []u8,
     debug_ranges: []u8,
 
     pub fn parseFromObject(allocator: Allocator, object: *const Object) !?DebugInfo {
@@ -87,6 +89,12 @@ const DebugInfo = struct {
             const index = object.dwarf_debug_line_index orelse return null;
             break :blk try object.readSection(allocator, index);
         };
+        var debug_line_str = blk: {
+            if (object.dwarf_debug_line_str_index) |ind| {
+                break :blk try object.readSection(allocator, ind);
+            }
+            break :blk try allocator.alloc(u8, 0);
+        };
         var debug_ranges = blk: {
             if (object.dwarf_debug_ranges_index) |ind| {
                 break :blk try object.readSection(allocator, ind);
@@ -100,6 +108,7 @@ const DebugInfo = struct {
             .debug_abbrev = debug_abbrev,
             .debug_str = debug_str,
             .debug_line = debug_line,
+            .debug_line_str = debug_line_str,
             .debug_ranges = debug_ranges,
         };
         try dwarf.openDwarfDebugInfo(&inner, allocator);
@@ -110,6 +119,7 @@ const DebugInfo = struct {
             .debug_abbrev = debug_abbrev,
             .debug_str = debug_str,
             .debug_line = debug_line,
+            .debug_line_str = debug_line_str,
             .debug_ranges = debug_ranges,
         };
     }
@@ -119,6 +129,7 @@ const DebugInfo = struct {
         allocator.free(self.debug_abbrev);
         allocator.free(self.debug_str);
         allocator.free(self.debug_line);
+        allocator.free(self.debug_line_str);
         allocator.free(self.debug_ranges);
         self.inner.abbrev_table_list.deinit();
         self.inner.compile_unit_list.deinit();
@@ -285,6 +296,8 @@ pub fn readLoadCommands(self: *Object, allocator: Allocator, reader: anytype) !v
                             self.dwarf_debug_str_index = index;
                         } else if (mem.eql(u8, sectname, "__debug_line")) {
                             self.dwarf_debug_line_index = index;
+                        } else if (mem.eql(u8, sectname, "__debug_line_str")) {
+                            self.dwarf_debug_line_str_index = index;
                         } else if (mem.eql(u8, sectname, "__debug_ranges")) {
                             self.dwarf_debug_ranges_index = index;
                         }