| ... | ... | @@ -1,6 +1,8 @@ |
| 1 | 1 | //! Cross-platform abstraction for this binary's own debug information, with a |
| 2 | 2 | //! goal of minimal code bloat and compilation speed penalty. |
| 3 | 3 | |
| 4 | // MLUGG TODO: audit use of errors in this file. ideally, introduce some concrete error sets |
| 5 | |
| 4 | 6 | const builtin = @import("builtin"); |
| 5 | 7 | const native_os = builtin.os.tag; |
| 6 | 8 | const native_endian = native_arch.endian(); |
| ... | ... | @@ -26,13 +28,7 @@ const regValueNative = Dwarf.abi.regValueNative; |
| 26 | 28 | |
| 27 | 29 | const SelfInfo = @This(); |
| 28 | 30 | |
| 29 | | modules: std.AutoHashMapUnmanaged(usize, struct { |
| 30 | | di: Module.DebugInfo, |
| 31 | | // MLUGG TODO: okay actually these should definitely go on the impl so it can share state. e.g. loading unwind info might require lodaing debug info in some cases |
| 32 | | loaded_locations: bool, |
| 33 | | loaded_unwind: bool, |
| 34 | | const init: @This() = .{ .di = .init, .loaded_locations = false, .loaded_unwind = false }; |
| 35 | | }), |
| 31 | modules: std.AutoHashMapUnmanaged(usize, Module.DebugInfo), |
| 36 | 32 | lookup_cache: Module.LookupCache, |
| 37 | 33 | |
| 38 | 34 | pub const target_supported: bool = switch (native_os) { |
| ... | ... | @@ -77,11 +73,7 @@ pub fn unwindFrame(self: *SelfInfo, gpa: Allocator, context: *UnwindContext) !us |
| 77 | 73 | const module: Module = try .lookup(&self.lookup_cache, gpa, context.pc); |
| 78 | 74 | const gop = try self.modules.getOrPut(gpa, module.load_offset); |
| 79 | 75 | if (!gop.found_existing) gop.value_ptr.* = .init; |
| 80 | | if (!gop.value_ptr.loaded_unwind) { |
| 81 | | try module.loadUnwindInfo(gpa, &gop.value_ptr.di); |
| 82 | | gop.value_ptr.loaded_unwind = true; |
| 83 | | } |
| 84 | | return module.unwindFrame(gpa, &gop.value_ptr.di, context); |
| 76 | return module.unwindFrame(gpa, gop.value_ptr, context); |
| 85 | 77 | } |
| 86 | 78 | |
| 87 | 79 | pub fn getSymbolAtAddress(self: *SelfInfo, gpa: Allocator, address: usize) !std.debug.Symbol { |
| ... | ... | @@ -89,11 +81,7 @@ pub fn getSymbolAtAddress(self: *SelfInfo, gpa: Allocator, address: usize) !std. |
| 89 | 81 | const module: Module = try .lookup(&self.lookup_cache, gpa, address); |
| 90 | 82 | const gop = try self.modules.getOrPut(gpa, module.key()); |
| 91 | 83 | if (!gop.found_existing) gop.value_ptr.* = .init; |
| 92 | | if (!gop.value_ptr.loaded_locations) { |
| 93 | | try module.loadLocationInfo(gpa, &gop.value_ptr.di); |
| 94 | | gop.value_ptr.loaded_locations = true; |
| 95 | | } |
| 96 | | return module.getSymbolAtAddress(gpa, &gop.value_ptr.di, address); |
| 84 | return module.getSymbolAtAddress(gpa, gop.value_ptr, address); |
| 97 | 85 | } |
| 98 | 86 | |
| 99 | 87 | /// Returns the module name for a given address. |
| ... | ... | @@ -168,9 +156,125 @@ const Module = switch (native_os) { |
| 168 | 156 | return error.MissingDebugInfo; |
| 169 | 157 | } |
| 170 | 158 | fn loadLocationInfo(module: *const Module, gpa: Allocator, di: *Module.DebugInfo) !void { |
| 171 | | try loadMachODebugInfo(gpa, module, di); // MLUGG TODO inline |
| 159 | const mapped_mem = mapFileOrSelfExe(module.name) catch |err| switch (err) { |
| 160 | error.FileNotFound => return error.MissingDebugInfo, |
| 161 | error.FileTooBig => return error.InvalidDebugInfo, |
| 162 | else => |e| return e, |
| 163 | }; |
| 164 | errdefer posix.munmap(mapped_mem); |
| 165 | |
| 166 | const hdr: *const macho.mach_header_64 = @ptrCast(@alignCast(mapped_mem.ptr)); |
| 167 | if (hdr.magic != macho.MH_MAGIC_64) |
| 168 | return error.InvalidDebugInfo; |
| 169 | |
| 170 | const symtab: macho.symtab_command = symtab: { |
| 171 | var it: macho.LoadCommandIterator = .{ |
| 172 | .ncmds = hdr.ncmds, |
| 173 | .buffer = mapped_mem[@sizeOf(macho.mach_header_64)..][0..hdr.sizeofcmds], |
| 174 | }; |
| 175 | while (it.next()) |cmd| switch (cmd.cmd()) { |
| 176 | .SYMTAB => break :symtab cmd.cast(macho.symtab_command) orelse return error.InvalidDebugInfo, |
| 177 | else => {}, |
| 178 | }; |
| 179 | return error.MissingDebugInfo; |
| 180 | }; |
| 181 | |
| 182 | const syms_ptr: [*]align(1) const macho.nlist_64 = @ptrCast(mapped_mem[symtab.symoff..]); |
| 183 | const syms = syms_ptr[0..symtab.nsyms]; |
| 184 | const strings = mapped_mem[symtab.stroff..][0 .. symtab.strsize - 1 :0]; |
| 185 | |
| 186 | // MLUGG TODO: does it really make sense to initCapacity here? how many of syms are omitted? |
| 187 | var symbols: std.ArrayList(MachoSymbol) = try .initCapacity(gpa, syms.len); |
| 188 | defer symbols.deinit(gpa); |
| 189 | |
| 190 | var ofile: u32 = undefined; |
| 191 | var last_sym: MachoSymbol = undefined; |
| 192 | var state: enum { |
| 193 | init, |
| 194 | oso_open, |
| 195 | oso_close, |
| 196 | bnsym, |
| 197 | fun_strx, |
| 198 | fun_size, |
| 199 | ensym, |
| 200 | } = .init; |
| 201 | |
| 202 | for (syms) |*sym| { |
| 203 | if (sym.n_type.bits.is_stab == 0) continue; |
| 204 | |
| 205 | // TODO handle globals N_GSYM, and statics N_STSYM |
| 206 | switch (sym.n_type.stab) { |
| 207 | .oso => switch (state) { |
| 208 | .init, .oso_close => { |
| 209 | state = .oso_open; |
| 210 | ofile = sym.n_strx; |
| 211 | }, |
| 212 | else => return error.InvalidDebugInfo, |
| 213 | }, |
| 214 | .bnsym => switch (state) { |
| 215 | .oso_open, .ensym => { |
| 216 | state = .bnsym; |
| 217 | last_sym = .{ |
| 218 | .strx = 0, |
| 219 | .addr = sym.n_value, |
| 220 | .size = 0, |
| 221 | .ofile = ofile, |
| 222 | }; |
| 223 | }, |
| 224 | else => return error.InvalidDebugInfo, |
| 225 | }, |
| 226 | .fun => switch (state) { |
| 227 | .bnsym => { |
| 228 | state = .fun_strx; |
| 229 | last_sym.strx = sym.n_strx; |
| 230 | }, |
| 231 | .fun_strx => { |
| 232 | state = .fun_size; |
| 233 | last_sym.size = @intCast(sym.n_value); |
| 234 | }, |
| 235 | else => return error.InvalidDebugInfo, |
| 236 | }, |
| 237 | .ensym => switch (state) { |
| 238 | .fun_size => { |
| 239 | state = .ensym; |
| 240 | symbols.appendAssumeCapacity(last_sym); |
| 241 | }, |
| 242 | else => return error.InvalidDebugInfo, |
| 243 | }, |
| 244 | .so => switch (state) { |
| 245 | .init, .oso_close => {}, |
| 246 | .oso_open, .ensym => { |
| 247 | state = .oso_close; |
| 248 | }, |
| 249 | else => return error.InvalidDebugInfo, |
| 250 | }, |
| 251 | else => {}, |
| 252 | } |
| 253 | } |
| 254 | |
| 255 | switch (state) { |
| 256 | .init => return error.MissingDebugInfo, |
| 257 | .oso_close => {}, |
| 258 | else => return error.InvalidDebugInfo, |
| 259 | } |
| 260 | |
| 261 | const symbols_slice = try symbols.toOwnedSlice(gpa); |
| 262 | errdefer gpa.free(symbols_slice); |
| 263 | |
| 264 | // Even though lld emits symbols in ascending order, this debug code |
| 265 | // should work for programs linked in any valid way. |
| 266 | // This sort is so that we can binary search later. |
| 267 | mem.sort(MachoSymbol, symbols_slice, {}, MachoSymbol.addressLessThan); |
| 268 | |
| 269 | di.full = .{ |
| 270 | .mapped_memory = mapped_mem, |
| 271 | .symbols = symbols_slice, |
| 272 | .strings = strings, |
| 273 | .ofiles = .empty, |
| 274 | }; |
| 172 | 275 | } |
| 173 | 276 | fn loadUnwindInfo(module: *const Module, gpa: Allocator, di: *Module.DebugInfo) !void { |
| 277 | if (di.unwind != null) return; |
| 174 | 278 | _ = gpa; |
| 175 | 279 | di.unwind = .{ |
| 176 | 280 | .unwind_info = module.unwind_info, |
| ... | ... | @@ -178,27 +282,39 @@ const Module = switch (native_os) { |
| 178 | 282 | }; |
| 179 | 283 | } |
| 180 | 284 | fn getSymbolAtAddress(module: *const Module, gpa: Allocator, di: *DebugInfo, address: usize) !std.debug.Symbol { |
| 285 | if (di.full == null) try module.loadLocationInfo(gpa, di); |
| 181 | 286 | const vaddr = address - module.load_offset; |
| 182 | | const symbol = MachoSymbol.find(di.full.symbols, vaddr) orelse return .{}; // MLUGG TODO null? |
| 287 | const symbol = MachoSymbol.find(di.full.?.symbols, vaddr) orelse return .{ |
| 288 | .name = null, |
| 289 | .compile_unit_name = null, |
| 290 | .source_location = null, |
| 291 | }; |
| 183 | 292 | |
| 184 | 293 | // offset of `address` from start of `symbol` |
| 185 | 294 | const address_symbol_offset = vaddr - symbol.addr; |
| 186 | 295 | |
| 187 | 296 | // Take the symbol name from the N_FUN STAB entry, we're going to |
| 188 | 297 | // use it if we fail to find the DWARF infos |
| 189 | | const stab_symbol = mem.sliceTo(di.full.strings[symbol.strx..], 0); |
| 190 | | const o_file_path = mem.sliceTo(di.full.strings[symbol.ofile..], 0); |
| 298 | const stab_symbol = mem.sliceTo(di.full.?.strings[symbol.strx..], 0); |
| 299 | const o_file_path = mem.sliceTo(di.full.?.strings[symbol.ofile..], 0); |
| 300 | |
| 301 | // If any information is missing, we can at least return this from now on. |
| 302 | const sym_only_result: std.debug.Symbol = .{ |
| 303 | .name = stab_symbol, |
| 304 | .compile_unit_name = null, |
| 305 | .source_location = null, |
| 306 | }; |
| 191 | 307 | |
| 192 | 308 | const o_file: *DebugInfo.OFile = of: { |
| 193 | | const gop = try di.full.ofiles.getOrPut(gpa, o_file_path); |
| 309 | const gop = try di.full.?.ofiles.getOrPut(gpa, o_file_path); |
| 194 | 310 | if (!gop.found_existing) { |
| 195 | 311 | gop.value_ptr.* = DebugInfo.loadOFile(gpa, o_file_path) catch |err| { |
| 196 | | defer _ = di.full.ofiles.pop().?; |
| 312 | defer _ = di.full.?.ofiles.pop().?; |
| 197 | 313 | switch (err) { |
| 198 | 314 | error.FileNotFound, |
| 199 | 315 | error.MissingDebugInfo, |
| 200 | 316 | error.InvalidDebugInfo, |
| 201 | | => return .{ .name = stab_symbol }, |
| 317 | => return sym_only_result, |
| 202 | 318 | else => |e| return e, |
| 203 | 319 | } |
| 204 | 320 | }; |
| ... | ... | @@ -206,10 +322,10 @@ const Module = switch (native_os) { |
| 206 | 322 | break :of gop.value_ptr; |
| 207 | 323 | }; |
| 208 | 324 | |
| 209 | | const symbol_ofile_vaddr = o_file.addr_table.get(stab_symbol) orelse return .{ .name = stab_symbol }; |
| 325 | const symbol_ofile_vaddr = o_file.addr_table.get(stab_symbol) orelse return sym_only_result; |
| 210 | 326 | |
| 211 | 327 | const compile_unit = o_file.dwarf.findCompileUnit(native_endian, symbol_ofile_vaddr) catch |err| switch (err) { |
| 212 | | error.MissingDebugInfo, error.InvalidDebugInfo => return .{ .name = stab_symbol }, |
| 328 | error.MissingDebugInfo, error.InvalidDebugInfo => return sym_only_result, |
| 213 | 329 | else => |e| return e, |
| 214 | 330 | }; |
| 215 | 331 | |
| ... | ... | @@ -222,7 +338,7 @@ const Module = switch (native_os) { |
| 222 | 338 | o_file.dwarf.section(.debug_str), |
| 223 | 339 | compile_unit, |
| 224 | 340 | ) catch |err| switch (err) { |
| 225 | | error.MissingDebugInfo, error.InvalidDebugInfo => "???", |
| 341 | error.MissingDebugInfo, error.InvalidDebugInfo => null, |
| 226 | 342 | }, |
| 227 | 343 | .source_location = o_file.dwarf.getLineNumberInfo( |
| 228 | 344 | gpa, |
| ... | ... | @@ -236,25 +352,27 @@ const Module = switch (native_os) { |
| 236 | 352 | }; |
| 237 | 353 | } |
| 238 | 354 | fn unwindFrame(module: *const Module, gpa: Allocator, di: *DebugInfo, context: *UnwindContext) !usize { |
| 239 | | _ = gpa; |
| 240 | | const unwind_info = di.unwind.unwind_info orelse return error.MissingUnwindInfo; |
| 241 | | // MLUGG TODO: inline |
| 355 | if (di.unwind == null) try module.loadUnwindInfo(gpa, di); |
| 356 | const unwind_info = di.unwind.?.unwind_info orelse return error.MissingUnwindInfo; |
| 357 | // MLUGG TODO: inline? |
| 242 | 358 | return unwindFrameMachO( |
| 243 | 359 | module.text_base, |
| 244 | 360 | module.load_offset, |
| 245 | 361 | context, |
| 246 | 362 | unwind_info, |
| 247 | | di.unwind.eh_frame, |
| 363 | di.unwind.?.eh_frame, |
| 248 | 364 | ); |
| 249 | 365 | } |
| 250 | 366 | const LookupCache = void; |
| 251 | 367 | const DebugInfo = struct { |
| 252 | | unwind: struct { |
| 368 | unwind: ?struct { |
| 369 | // Backed by the in-memory sections mapped by the loader |
| 370 | // MLUGG TODO: these are duplicated state. i actually reckon they should be removed from Module, and loadLocationInfo should be the one discovering them! |
| 253 | 371 | unwind_info: ?[]const u8, |
| 254 | 372 | eh_frame: ?[]const u8, |
| 255 | 373 | }, |
| 256 | 374 | // MLUGG TODO: awful field name |
| 257 | | full: struct { |
| 375 | full: ?struct { |
| 258 | 376 | mapped_memory: []align(std.heap.page_size_min) const u8, |
| 259 | 377 | symbols: []const MachoSymbol, |
| 260 | 378 | strings: [:0]const u8, |
| ... | ... | @@ -262,11 +380,10 @@ const Module = switch (native_os) { |
| 262 | 380 | ofiles: std.StringArrayHashMapUnmanaged(OFile), |
| 263 | 381 | }, |
| 264 | 382 | |
| 265 | | // Backed by the in-memory sections mapped by the loader |
| 266 | | // MLUGG TODO: these are duplicated state. i actually reckon they should be removed from Module, and loadMachODebugInfo should be the one discovering them! |
| 267 | | |
| 268 | | // MLUGG TODO HACKHACK: this is awful |
| 269 | | const init: DebugInfo = undefined; |
| 383 | const init: DebugInfo = .{ |
| 384 | .unwind = null, |
| 385 | .full = null, |
| 386 | }; |
| 270 | 387 | |
| 271 | 388 | const OFile = struct { |
| 272 | 389 | dwarf: Dwarf, |
| ... | ... | @@ -388,18 +505,6 @@ const Module = switch (native_os) { |
| 388 | 505 | _ = address; |
| 389 | 506 | unreachable; |
| 390 | 507 | } |
| 391 | | fn loadLocationInfo(module: *const Module, gpa: Allocator, di: *DebugInfo) !void { |
| 392 | | _ = module; |
| 393 | | _ = gpa; |
| 394 | | _ = di; |
| 395 | | unreachable; |
| 396 | | } |
| 397 | | fn loadUnwindInfo(module: *const Module, gpa: Allocator, di: *DebugInfo) !void { |
| 398 | | _ = module; |
| 399 | | _ = gpa; |
| 400 | | _ = di; |
| 401 | | unreachable; |
| 402 | | } |
| 403 | 508 | }, |
| 404 | 509 | .linux, .netbsd, .freebsd, .dragonfly, .openbsd, .haiku, .solaris, .illumos => struct { |
| 405 | 510 | load_offset: usize, |
| ... | ... | @@ -408,9 +513,12 @@ const Module = switch (native_os) { |
| 408 | 513 | gnu_eh_frame: ?[]const u8, |
| 409 | 514 | const LookupCache = void; |
| 410 | 515 | const DebugInfo = struct { |
| 411 | | const init: DebugInfo = undefined; // MLUGG TODO: this makes me sad |
| 412 | | em: Dwarf.ElfModule, // MLUGG TODO: bad field name (and, frankly, type) |
| 413 | | unwind: Dwarf.Unwind, |
| 516 | em: ?Dwarf.ElfModule, // MLUGG TODO: bad field name (and, frankly, type) |
| 517 | unwind: ?Dwarf.Unwind, |
| 518 | const init: DebugInfo = .{ |
| 519 | .em = null, |
| 520 | .unwind = null, |
| 521 | }; |
| 414 | 522 | }; |
| 415 | 523 | fn key(m: Module) usize { |
| 416 | 524 | return m.load_offset; // MLUGG TODO: is this technically valid? idk |
| ... | ... | @@ -496,19 +604,20 @@ const Module = switch (native_os) { |
| 496 | 604 | errdefer posix.munmap(mapped_mem); |
| 497 | 605 | di.em = try .load(gpa, mapped_mem, module.build_id, null, null, null, filename); |
| 498 | 606 | } |
| 607 | fn getSymbolAtAddress(module: *const Module, gpa: Allocator, di: *DebugInfo, address: usize) !std.debug.Symbol { |
| 608 | if (di.em == null) try module.loadLocationInfo(gpa, di); |
| 609 | return di.em.?.getSymbolAtAddress(gpa, native_endian, module.load_offset, address); |
| 610 | } |
| 499 | 611 | fn loadUnwindInfo(module: *const Module, gpa: Allocator, di: *Module.DebugInfo) !void { |
| 500 | 612 | const section_bytes = module.gnu_eh_frame orelse return error.MissingUnwindInfo; // MLUGG TODO: load from file |
| 501 | 613 | const section_vaddr: u64 = @intFromPtr(section_bytes.ptr) - module.load_offset; |
| 502 | 614 | const header: Dwarf.Unwind.EhFrameHeader = try .parse(section_vaddr, section_bytes, @sizeOf(usize), native_endian); |
| 503 | 615 | di.unwind = .initEhFrameHdr(header, section_vaddr, @ptrFromInt(module.load_offset + header.eh_frame_vaddr)); |
| 504 | | try di.unwind.prepareLookup(gpa, @sizeOf(usize), native_endian); |
| 505 | | } |
| 506 | | fn getSymbolAtAddress(module: *const Module, gpa: Allocator, di: *DebugInfo, address: usize) !std.debug.Symbol { |
| 507 | | return di.em.getSymbolAtAddress(gpa, native_endian, module.load_offset, address); |
| 616 | try di.unwind.?.prepareLookup(gpa, @sizeOf(usize), native_endian); |
| 508 | 617 | } |
| 509 | 618 | fn unwindFrame(module: *const Module, gpa: Allocator, di: *DebugInfo, context: *UnwindContext) !usize { |
| 510 | | _ = gpa; |
| 511 | | return unwindFrameDwarf(&di.unwind, module.load_offset, context, null); |
| 619 | if (di.unwind == null) try module.loadUnwindInfo(gpa, di); |
| 620 | return unwindFrameDwarf(&di.unwind.?, module.load_offset, context, null); |
| 512 | 621 | } |
| 513 | 622 | }, |
| 514 | 623 | .uefi, .windows => struct { |
| ... | ... | @@ -660,12 +769,16 @@ const Module = switch (native_os) { |
| 660 | 769 | |
| 661 | 770 | di.coff_section_headers = try coff_obj.getSectionHeadersAlloc(gpa); |
| 662 | 771 | } |
| 772 | |
| 773 | di.loaded = true; |
| 663 | 774 | } |
| 664 | 775 | const LookupCache = struct { |
| 665 | 776 | modules: std.ArrayListUnmanaged(windows.MODULEENTRY32), |
| 666 | 777 | const init: LookupCache = .{ .modules = .empty }; |
| 667 | 778 | }; |
| 668 | 779 | const DebugInfo = struct { |
| 780 | loaded: bool, |
| 781 | |
| 669 | 782 | coff_image_base: u64, |
| 670 | 783 | mapped_file: ?struct { |
| 671 | 784 | file: File, |
| ... | ... | @@ -686,6 +799,7 @@ const Module = switch (native_os) { |
| 686 | 799 | coff_section_headers: []coff.SectionHeader, |
| 687 | 800 | |
| 688 | 801 | const init: DebugInfo = .{ |
| 802 | .loaded = false, |
| 689 | 803 | .coff_image_base = undefined, |
| 690 | 804 | .mapped_file = null, |
| 691 | 805 | .dwarf = null, |
| ... | ... | @@ -717,28 +831,24 @@ const Module = switch (native_os) { |
| 717 | 831 | return null; |
| 718 | 832 | }; |
| 719 | 833 | |
| 720 | | const module = (try di.pdb.?.getModule(mod_index)) orelse |
| 721 | | return error.InvalidDebugInfo; |
| 722 | | const obj_basename = fs.path.basename(module.obj_file_name); |
| 723 | | |
| 724 | | const symbol_name = di.pdb.?.getSymbolName( |
| 725 | | module, |
| 726 | | relocated_address - coff_section.virtual_address, |
| 727 | | ) orelse "???"; |
| 728 | | const opt_line_info = try di.pdb.?.getLineNumberInfo( |
| 729 | | module, |
| 730 | | relocated_address - coff_section.virtual_address, |
| 731 | | ); |
| 834 | const module = try di.pdb.?.getModule(mod_index) orelse return error.InvalidDebugInfo; |
| 732 | 835 | |
| 733 | 836 | return .{ |
| 734 | | .name = symbol_name, |
| 735 | | .compile_unit_name = obj_basename, |
| 736 | | .source_location = opt_line_info, |
| 837 | .name = di.pdb.?.getSymbolName( |
| 838 | module, |
| 839 | relocated_address - coff_section.virtual_address, |
| 840 | ), |
| 841 | .compile_unit_name = fs.path.basename(module.obj_file_name), |
| 842 | .source_location = try di.pdb.?.getLineNumberInfo( |
| 843 | module, |
| 844 | relocated_address - coff_section.virtual_address, |
| 845 | ), |
| 737 | 846 | }; |
| 738 | 847 | } |
| 739 | 848 | }; |
| 740 | 849 | |
| 741 | 850 | fn getSymbolAtAddress(module: *const Module, gpa: Allocator, di: *DebugInfo, address: usize) !std.debug.Symbol { |
| 851 | if (!di.loaded) try module.loadLocationInfo(gpa, di); |
| 742 | 852 | // Translate the runtime address into a virtual address into the module |
| 743 | 853 | const vaddr = address - module.base_address; |
| 744 | 854 | |
| ... | ... | @@ -756,125 +866,6 @@ const Module = switch (native_os) { |
| 756 | 866 | }, |
| 757 | 867 | }; |
| 758 | 868 | |
| 759 | | fn loadMachODebugInfo(gpa: Allocator, module: *const Module, di: *Module.DebugInfo) !void { |
| 760 | | const mapped_mem = mapFileOrSelfExe(module.name) catch |err| switch (err) { |
| 761 | | error.FileNotFound => return error.MissingDebugInfo, |
| 762 | | error.FileTooBig => return error.InvalidDebugInfo, |
| 763 | | else => |e| return e, |
| 764 | | }; |
| 765 | | errdefer posix.munmap(mapped_mem); |
| 766 | | |
| 767 | | const hdr: *const macho.mach_header_64 = @ptrCast(@alignCast(mapped_mem.ptr)); |
| 768 | | if (hdr.magic != macho.MH_MAGIC_64) |
| 769 | | return error.InvalidDebugInfo; |
| 770 | | |
| 771 | | const symtab: macho.symtab_command = symtab: { |
| 772 | | var it: macho.LoadCommandIterator = .{ |
| 773 | | .ncmds = hdr.ncmds, |
| 774 | | .buffer = mapped_mem[@sizeOf(macho.mach_header_64)..][0..hdr.sizeofcmds], |
| 775 | | }; |
| 776 | | while (it.next()) |cmd| switch (cmd.cmd()) { |
| 777 | | .SYMTAB => break :symtab cmd.cast(macho.symtab_command) orelse return error.InvalidDebugInfo, |
| 778 | | else => {}, |
| 779 | | }; |
| 780 | | return error.MissingDebugInfo; |
| 781 | | }; |
| 782 | | |
| 783 | | const syms_ptr: [*]align(1) const macho.nlist_64 = @ptrCast(mapped_mem[symtab.symoff..]); |
| 784 | | const syms = syms_ptr[0..symtab.nsyms]; |
| 785 | | const strings = mapped_mem[symtab.stroff..][0 .. symtab.strsize - 1 :0]; |
| 786 | | |
| 787 | | // MLUGG TODO: does it really make sense to initCapacity here? how many of syms are omitted? |
| 788 | | var symbols: std.ArrayList(MachoSymbol) = try .initCapacity(gpa, syms.len); |
| 789 | | defer symbols.deinit(gpa); |
| 790 | | |
| 791 | | var ofile: u32 = undefined; |
| 792 | | var last_sym: MachoSymbol = undefined; |
| 793 | | var state: enum { |
| 794 | | init, |
| 795 | | oso_open, |
| 796 | | oso_close, |
| 797 | | bnsym, |
| 798 | | fun_strx, |
| 799 | | fun_size, |
| 800 | | ensym, |
| 801 | | } = .init; |
| 802 | | |
| 803 | | for (syms) |*sym| { |
| 804 | | if (sym.n_type.bits.is_stab == 0) continue; |
| 805 | | |
| 806 | | // TODO handle globals N_GSYM, and statics N_STSYM |
| 807 | | switch (sym.n_type.stab) { |
| 808 | | .oso => switch (state) { |
| 809 | | .init, .oso_close => { |
| 810 | | state = .oso_open; |
| 811 | | ofile = sym.n_strx; |
| 812 | | }, |
| 813 | | else => return error.InvalidDebugInfo, |
| 814 | | }, |
| 815 | | .bnsym => switch (state) { |
| 816 | | .oso_open, .ensym => { |
| 817 | | state = .bnsym; |
| 818 | | last_sym = .{ |
| 819 | | .strx = 0, |
| 820 | | .addr = sym.n_value, |
| 821 | | .size = 0, |
| 822 | | .ofile = ofile, |
| 823 | | }; |
| 824 | | }, |
| 825 | | else => return error.InvalidDebugInfo, |
| 826 | | }, |
| 827 | | .fun => switch (state) { |
| 828 | | .bnsym => { |
| 829 | | state = .fun_strx; |
| 830 | | last_sym.strx = sym.n_strx; |
| 831 | | }, |
| 832 | | .fun_strx => { |
| 833 | | state = .fun_size; |
| 834 | | last_sym.size = @intCast(sym.n_value); |
| 835 | | }, |
| 836 | | else => return error.InvalidDebugInfo, |
| 837 | | }, |
| 838 | | .ensym => switch (state) { |
| 839 | | .fun_size => { |
| 840 | | state = .ensym; |
| 841 | | symbols.appendAssumeCapacity(last_sym); |
| 842 | | }, |
| 843 | | else => return error.InvalidDebugInfo, |
| 844 | | }, |
| 845 | | .so => switch (state) { |
| 846 | | .init, .oso_close => {}, |
| 847 | | .oso_open, .ensym => { |
| 848 | | state = .oso_close; |
| 849 | | }, |
| 850 | | else => return error.InvalidDebugInfo, |
| 851 | | }, |
| 852 | | else => {}, |
| 853 | | } |
| 854 | | } |
| 855 | | |
| 856 | | switch (state) { |
| 857 | | .init => return error.MissingDebugInfo, |
| 858 | | .oso_close => {}, |
| 859 | | else => return error.InvalidDebugInfo, |
| 860 | | } |
| 861 | | |
| 862 | | const symbols_slice = try symbols.toOwnedSlice(gpa); |
| 863 | | errdefer gpa.free(symbols_slice); |
| 864 | | |
| 865 | | // Even though lld emits symbols in ascending order, this debug code |
| 866 | | // should work for programs linked in any valid way. |
| 867 | | // This sort is so that we can binary search later. |
| 868 | | mem.sort(MachoSymbol, symbols_slice, {}, MachoSymbol.addressLessThan); |
| 869 | | |
| 870 | | di.full = .{ |
| 871 | | .mapped_memory = mapped_mem, |
| 872 | | .symbols = symbols_slice, |
| 873 | | .strings = strings, |
| 874 | | .ofiles = .empty, |
| 875 | | }; |
| 876 | | } |
| 877 | | |
| 878 | 869 | const MachoSymbol = struct { |
| 879 | 870 | strx: u32, |
| 880 | 871 | addr: u64, |
Matthew Lugg <mlugg@mlugg.co.uk> 2025-09-02 14:44:15+01:00
Commit is signed but in an unrecognized format.
lib