| 1 | //===-- lib/DebugInfo/Symbolize/MarkupFilter.cpp -------------------------===// |
| 2 | // |
| 3 | // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions. |
| 4 | // See https://llvm.org/LICENSE.txt for license information. |
| 5 | // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception |
| 6 | // |
| 7 | //===----------------------------------------------------------------------===// |
| 8 | /// |
| 9 | /// \file |
| 10 | /// This file defines the implementation of a filter that replaces symbolizer |
| 11 | /// markup with human-readable expressions. |
| 12 | /// |
| 13 | /// See https://llvm.org/docs/SymbolizerMarkupFormat.html |
| 14 | /// |
| 15 | //===----------------------------------------------------------------------===// |
| 16 | |
| 17 | #include "llvm/DebugInfo/Symbolize/MarkupFilter.h" |
| 18 | |
| 19 | #include "llvm/ADT/STLExtras.h" |
| 20 | #include "llvm/ADT/StringExtras.h" |
| 21 | #include "llvm/ADT/StringSwitch.h" |
| 22 | #include "llvm/DebugInfo/DIContext.h" |
| 23 | #include "llvm/DebugInfo/Symbolize/Markup.h" |
| 24 | #include "llvm/DebugInfo/Symbolize/Symbolize.h" |
| 25 | #include "llvm/Demangle/Demangle.h" |
| 26 | #include "llvm/Support/Error.h" |
| 27 | #include "llvm/Support/FormatVariadic.h" |
| 28 | #include "llvm/Support/WithColor.h" |
| 29 | #include "llvm/Support/raw_ostream.h" |
| 30 | #include <optional> |
| 31 | |
| 32 | using namespace llvm; |
| 33 | using namespace llvm::symbolize; |
| 34 | |
| 35 | MarkupFilter::MarkupFilter(raw_ostream &OS, LLVMSymbolizer &Symbolizer, |
| 36 | std::optional<bool> ColorsEnabled) |
| 37 | : OS(OS), Symbolizer(Symbolizer), |
| 38 | ColorsEnabled( |
| 39 | ColorsEnabled.value_or(u: WithColor::defaultAutoDetectFunction()(OS))) {} |
| 40 | |
| 41 | void MarkupFilter::filter(std::string &&InputLine) { |
| 42 | Line = std::move(InputLine); |
| 43 | resetColor(); |
| 44 | |
| 45 | Parser.parseLine(Line); |
| 46 | SmallVector<MarkupNode> DeferredNodes; |
| 47 | // See if the line is a contextual (i.e. contains a contextual element). |
| 48 | // In this case, anything after the contextual element is elided, or the whole |
| 49 | // line may be elided. |
| 50 | while (std::optional<MarkupNode> Node = Parser.nextNode()) { |
| 51 | // If this was a contextual line, then summarily stop processing. |
| 52 | if (tryContextualElement(Node: *Node, DeferredNodes)) |
| 53 | return; |
| 54 | // This node may yet be part of an elided contextual line. |
| 55 | DeferredNodes.push_back(Elt: *Node); |
| 56 | } |
| 57 | |
| 58 | // This was not a contextual line, so nothing in it should be elided. |
| 59 | endAnyModuleInfoLine(); |
| 60 | for (const MarkupNode &Node : DeferredNodes) |
| 61 | filterNode(Node); |
| 62 | } |
| 63 | |
| 64 | void MarkupFilter::finish() { |
| 65 | Parser.flush(); |
| 66 | while (std::optional<MarkupNode> Node = Parser.nextNode()) |
| 67 | filterNode(Node: *Node); |
| 68 | endAnyModuleInfoLine(); |
| 69 | resetColor(); |
| 70 | Modules.clear(); |
| 71 | MMaps.clear(); |
| 72 | } |
| 73 | |
| 74 | // See if the given node is a contextual element and handle it if so. This may |
| 75 | // either output or defer the element; in the former case, it will first emit |
| 76 | // any DeferredNodes. |
| 77 | // |
| 78 | // Returns true if the given element was a contextual element. In this case, |
| 79 | // DeferredNodes should be considered handled and should not be emitted. The |
| 80 | // rest of the containing line must also be ignored in case the element was |
| 81 | // deferred to a following line. |
| 82 | bool MarkupFilter::tryContextualElement( |
| 83 | const MarkupNode &Node, const SmallVector<MarkupNode> &DeferredNodes) { |
| 84 | if (tryMMap(Element: Node, DeferredNodes)) |
| 85 | return true; |
| 86 | if (tryReset(Element: Node, DeferredNodes)) |
| 87 | return true; |
| 88 | return tryModule(Element: Node, DeferredNodes); |
| 89 | } |
| 90 | |
| 91 | bool MarkupFilter::tryMMap(const MarkupNode &Node, |
| 92 | const SmallVector<MarkupNode> &DeferredNodes) { |
| 93 | if (Node.Tag != "mmap" ) |
| 94 | return false; |
| 95 | std::optional<MMap> ParsedMMap = parseMMap(Element: Node); |
| 96 | if (!ParsedMMap) |
| 97 | return true; |
| 98 | |
| 99 | if (const MMap *M = getOverlappingMMap(Map: *ParsedMMap)) { |
| 100 | WithColor::error(OS&: errs()) |
| 101 | << formatv(Fmt: "overlapping mmap: #{0:x} [{1:x}-{2:x}]\n" , Vals: M->Mod->ID, |
| 102 | Vals: M->Addr, Vals: M->Addr + M->Size - 1); |
| 103 | reportLocation(Loc: Node.Fields[0].begin()); |
| 104 | return true; |
| 105 | } |
| 106 | |
| 107 | auto Res = MMaps.emplace(args&: ParsedMMap->Addr, args: std::move(*ParsedMMap)); |
| 108 | assert(Res.second && "Overlap check should ensure emplace succeeds." ); |
| 109 | MMap &MMap = Res.first->second; |
| 110 | |
| 111 | if (!MIL || MIL->Mod != MMap.Mod) { |
| 112 | endAnyModuleInfoLine(); |
| 113 | for (const MarkupNode &Node : DeferredNodes) |
| 114 | filterNode(Node); |
| 115 | beginModuleInfoLine(M: MMap.Mod); |
| 116 | OS << "; adds" ; |
| 117 | } |
| 118 | MIL->MMaps.push_back(Elt: &MMap); |
| 119 | return true; |
| 120 | } |
| 121 | |
| 122 | bool MarkupFilter::tryReset(const MarkupNode &Node, |
| 123 | const SmallVector<MarkupNode> &DeferredNodes) { |
| 124 | if (Node.Tag != "reset" ) |
| 125 | return false; |
| 126 | if (!checkNumFields(Element: Node, Size: 0)) |
| 127 | return true; |
| 128 | |
| 129 | if (!Modules.empty() || !MMaps.empty()) { |
| 130 | endAnyModuleInfoLine(); |
| 131 | for (const MarkupNode &Node : DeferredNodes) |
| 132 | filterNode(Node); |
| 133 | printRawElement(Element: Node); |
| 134 | OS << lineEnding(); |
| 135 | |
| 136 | Modules.clear(); |
| 137 | MMaps.clear(); |
| 138 | } |
| 139 | return true; |
| 140 | } |
| 141 | |
| 142 | bool MarkupFilter::tryModule(const MarkupNode &Node, |
| 143 | const SmallVector<MarkupNode> &DeferredNodes) { |
| 144 | if (Node.Tag != "module" ) |
| 145 | return false; |
| 146 | std::optional<Module> ParsedModule = parseModule(Element: Node); |
| 147 | if (!ParsedModule) |
| 148 | return true; |
| 149 | |
| 150 | auto Res = Modules.try_emplace( |
| 151 | Key: ParsedModule->ID, Args: std::make_unique<Module>(args: std::move(*ParsedModule))); |
| 152 | if (!Res.second) { |
| 153 | WithColor::error(OS&: errs()) << "duplicate module ID\n" ; |
| 154 | reportLocation(Loc: Node.Fields[0].begin()); |
| 155 | return true; |
| 156 | } |
| 157 | Module &Module = *Res.first->second; |
| 158 | |
| 159 | endAnyModuleInfoLine(); |
| 160 | for (const MarkupNode &Node : DeferredNodes) |
| 161 | filterNode(Node); |
| 162 | beginModuleInfoLine(M: &Module); |
| 163 | OS << "; BuildID=" ; |
| 164 | printValue(Value: toHex(Input: Module.BuildID, /*LowerCase=*/true)); |
| 165 | return true; |
| 166 | } |
| 167 | |
| 168 | void MarkupFilter::beginModuleInfoLine(const Module *M) { |
| 169 | highlight(); |
| 170 | OS << "[[[ELF module" ; |
| 171 | printValue(Value: formatv(Fmt: " #{0:x} " , Vals: M->ID)); |
| 172 | OS << '"'; |
| 173 | printValue(Value: M->Name); |
| 174 | OS << '"'; |
| 175 | MIL = ModuleInfoLine{.Mod: M}; |
| 176 | } |
| 177 | |
| 178 | void MarkupFilter::endAnyModuleInfoLine() { |
| 179 | if (!MIL) |
| 180 | return; |
| 181 | llvm::stable_sort(Range&: MIL->MMaps, C: [](const MMap *A, const MMap *B) { |
| 182 | return A->Addr < B->Addr; |
| 183 | }); |
| 184 | for (const MMap *M : MIL->MMaps) { |
| 185 | OS << (M == MIL->MMaps.front() ? ' ' : ','); |
| 186 | OS << '['; |
| 187 | printValue(Value: formatv(Fmt: "{0:x}" , Vals: M->Addr)); |
| 188 | OS << '-'; |
| 189 | printValue(Value: formatv(Fmt: "{0:x}" , Vals: M->Addr + M->Size - 1)); |
| 190 | OS << "](" ; |
| 191 | printValue(Value: M->Mode); |
| 192 | OS << ')'; |
| 193 | } |
| 194 | OS << "]]]" << lineEnding(); |
| 195 | restoreColor(); |
| 196 | MIL.reset(); |
| 197 | } |
| 198 | |
| 199 | // Handle a node that is known not to be a contextual element. |
| 200 | void MarkupFilter::filterNode(const MarkupNode &Node) { |
| 201 | if (!checkTag(Node)) |
| 202 | return; |
| 203 | if (tryPresentation(Node)) |
| 204 | return; |
| 205 | if (trySGR(Node)) |
| 206 | return; |
| 207 | |
| 208 | OS << Node.Text; |
| 209 | } |
| 210 | |
| 211 | bool MarkupFilter::tryPresentation(const MarkupNode &Node) { |
| 212 | if (trySymbol(Node)) |
| 213 | return true; |
| 214 | if (tryPC(Node)) |
| 215 | return true; |
| 216 | if (tryBackTrace(Node)) |
| 217 | return true; |
| 218 | return tryData(Node); |
| 219 | } |
| 220 | |
| 221 | bool MarkupFilter::trySymbol(const MarkupNode &Node) { |
| 222 | if (Node.Tag != "symbol" ) |
| 223 | return false; |
| 224 | if (!checkNumFields(Element: Node, Size: 1)) |
| 225 | return true; |
| 226 | |
| 227 | highlight(); |
| 228 | OS << llvm::demangle(MangledName: Node.Fields.front().str()); |
| 229 | restoreColor(); |
| 230 | return true; |
| 231 | } |
| 232 | |
| 233 | bool MarkupFilter::tryPC(const MarkupNode &Node) { |
| 234 | if (Node.Tag != "pc" ) |
| 235 | return false; |
| 236 | if (!checkNumFieldsAtLeast(Element: Node, Size: 1)) |
| 237 | return true; |
| 238 | warnNumFieldsAtMost(Element: Node, Size: 2); |
| 239 | |
| 240 | std::optional<uint64_t> Addr = parseAddr(Str: Node.Fields[0]); |
| 241 | if (!Addr) |
| 242 | return true; |
| 243 | |
| 244 | // PC addresses that aren't part of a backtrace are assumed to be precise code |
| 245 | // locations. |
| 246 | PCType Type = PCType::PreciseCode; |
| 247 | if (Node.Fields.size() == 2) { |
| 248 | std::optional<PCType> ParsedType = parsePCType(Str: Node.Fields[1]); |
| 249 | if (!ParsedType) |
| 250 | return true; |
| 251 | Type = *ParsedType; |
| 252 | } |
| 253 | *Addr = adjustAddr(Addr: *Addr, Type); |
| 254 | |
| 255 | const MMap *MMap = getContainingMMap(Addr: *Addr); |
| 256 | if (!MMap) { |
| 257 | WithColor::error() << "no mmap covers address\n" ; |
| 258 | reportLocation(Loc: Node.Fields[0].begin()); |
| 259 | printRawElement(Element: Node); |
| 260 | return true; |
| 261 | } |
| 262 | |
| 263 | Expected<DILineInfo> LI = Symbolizer.symbolizeCode( |
| 264 | BuildID: MMap->Mod->BuildID, ModuleOffset: {.Address: MMap->getModuleRelativeAddr(Addr: *Addr)}); |
| 265 | if (!LI) { |
| 266 | WithColor::defaultErrorHandler(Err: LI.takeError()); |
| 267 | printRawElement(Element: Node); |
| 268 | return true; |
| 269 | } |
| 270 | if (!*LI) { |
| 271 | printRawElement(Element: Node); |
| 272 | return true; |
| 273 | } |
| 274 | |
| 275 | highlight(); |
| 276 | printValue(Value: LI->FunctionName); |
| 277 | OS << '['; |
| 278 | printValue(Value: LI->FileName); |
| 279 | OS << ':'; |
| 280 | printValue(Value: Twine(LI->Line)); |
| 281 | OS << ']'; |
| 282 | restoreColor(); |
| 283 | return true; |
| 284 | } |
| 285 | |
| 286 | bool MarkupFilter::tryBackTrace(const MarkupNode &Node) { |
| 287 | if (Node.Tag != "bt" ) |
| 288 | return false; |
| 289 | if (!checkNumFieldsAtLeast(Element: Node, Size: 2)) |
| 290 | return true; |
| 291 | warnNumFieldsAtMost(Element: Node, Size: 3); |
| 292 | |
| 293 | std::optional<uint64_t> = parseFrameNumber(Str: Node.Fields[0]); |
| 294 | if (!FrameNumber) |
| 295 | return true; |
| 296 | |
| 297 | std::optional<uint64_t> Addr = parseAddr(Str: Node.Fields[1]); |
| 298 | if (!Addr) |
| 299 | return true; |
| 300 | |
| 301 | // Backtrace addresses are assumed to be return addresses by default. |
| 302 | PCType Type = PCType::ReturnAddress; |
| 303 | if (Node.Fields.size() == 3) { |
| 304 | std::optional<PCType> ParsedType = parsePCType(Str: Node.Fields[2]); |
| 305 | if (!ParsedType) |
| 306 | return true; |
| 307 | Type = *ParsedType; |
| 308 | } |
| 309 | *Addr = adjustAddr(Addr: *Addr, Type); |
| 310 | |
| 311 | const MMap *MMap = getContainingMMap(Addr: *Addr); |
| 312 | if (!MMap) { |
| 313 | WithColor::error() << "no mmap covers address\n" ; |
| 314 | reportLocation(Loc: Node.Fields[0].begin()); |
| 315 | printRawElement(Element: Node); |
| 316 | return true; |
| 317 | } |
| 318 | uint64_t MRA = MMap->getModuleRelativeAddr(Addr: *Addr); |
| 319 | |
| 320 | Expected<DIInliningInfo> II = |
| 321 | Symbolizer.symbolizeInlinedCode(BuildID: MMap->Mod->BuildID, ModuleOffset: {.Address: MRA}); |
| 322 | if (!II) { |
| 323 | WithColor::defaultErrorHandler(Err: II.takeError()); |
| 324 | printRawElement(Element: Node); |
| 325 | return true; |
| 326 | } |
| 327 | |
| 328 | highlight(); |
| 329 | for (unsigned I = 0, E = II->getNumberOfFrames(); I != E; ++I) { |
| 330 | auto = formatv(Fmt: "{0, +6}" , Vals: formatv(Fmt: "#{0}" , Vals&: FrameNumber)).sstr<16>(); |
| 331 | // Don't highlight the # sign as a value. |
| 332 | size_t NumberIdx = Header.find(Str: "#" ) + 1; |
| 333 | OS << Header.substr(Start: 0, N: NumberIdx); |
| 334 | printValue(Value: Header.substr(Start: NumberIdx)); |
| 335 | if (I == E - 1) { |
| 336 | OS << " " ; |
| 337 | } else { |
| 338 | OS << '.'; |
| 339 | printValue(Value: formatv(Fmt: "{0, -2}" , Vals: I + 1)); |
| 340 | } |
| 341 | printValue(Value: formatv(Fmt: " {0:x16} " , Vals&: *Addr)); |
| 342 | |
| 343 | DILineInfo LI = II->getFrame(Index: I); |
| 344 | if (LI) { |
| 345 | printValue(Value: LI.FunctionName); |
| 346 | OS << ' '; |
| 347 | printValue(Value: LI.FileName); |
| 348 | OS << ':'; |
| 349 | printValue(Value: Twine(LI.Line)); |
| 350 | OS << ':'; |
| 351 | printValue(Value: Twine(LI.Column)); |
| 352 | OS << ' '; |
| 353 | } |
| 354 | OS << '('; |
| 355 | printValue(Value: MMap->Mod->Name); |
| 356 | OS << "+" ; |
| 357 | printValue(Value: formatv(Fmt: "{0:x}" , Vals&: MRA)); |
| 358 | OS << ')'; |
| 359 | if (I != E - 1) |
| 360 | OS << lineEnding(); |
| 361 | } |
| 362 | restoreColor(); |
| 363 | return true; |
| 364 | } |
| 365 | |
| 366 | bool MarkupFilter::tryData(const MarkupNode &Node) { |
| 367 | if (Node.Tag != "data" ) |
| 368 | return false; |
| 369 | if (!checkNumFields(Element: Node, Size: 1)) |
| 370 | return true; |
| 371 | std::optional<uint64_t> Addr = parseAddr(Str: Node.Fields[0]); |
| 372 | if (!Addr) |
| 373 | return true; |
| 374 | |
| 375 | const MMap *MMap = getContainingMMap(Addr: *Addr); |
| 376 | if (!MMap) { |
| 377 | WithColor::error() << "no mmap covers address\n" ; |
| 378 | reportLocation(Loc: Node.Fields[0].begin()); |
| 379 | printRawElement(Element: Node); |
| 380 | return true; |
| 381 | } |
| 382 | |
| 383 | Expected<DIGlobal> Symbol = Symbolizer.symbolizeData( |
| 384 | BuildID: MMap->Mod->BuildID, ModuleOffset: {.Address: MMap->getModuleRelativeAddr(Addr: *Addr)}); |
| 385 | if (!Symbol) { |
| 386 | WithColor::defaultErrorHandler(Err: Symbol.takeError()); |
| 387 | printRawElement(Element: Node); |
| 388 | return true; |
| 389 | } |
| 390 | |
| 391 | highlight(); |
| 392 | OS << Symbol->Name; |
| 393 | restoreColor(); |
| 394 | return true; |
| 395 | } |
| 396 | |
| 397 | bool MarkupFilter::trySGR(const MarkupNode &Node) { |
| 398 | if (Node.Text == "\033[0m" ) { |
| 399 | resetColor(); |
| 400 | return true; |
| 401 | } |
| 402 | if (Node.Text == "\033[1m" ) { |
| 403 | Bold = true; |
| 404 | if (ColorsEnabled) |
| 405 | OS.changeColor(Color: raw_ostream::Colors::SAVEDCOLOR, Bold); |
| 406 | return true; |
| 407 | } |
| 408 | auto SGRColor = StringSwitch<std::optional<raw_ostream::Colors>>(Node.Text) |
| 409 | .Case(S: "\033[30m" , Value: raw_ostream::Colors::BLACK) |
| 410 | .Case(S: "\033[31m" , Value: raw_ostream::Colors::RED) |
| 411 | .Case(S: "\033[32m" , Value: raw_ostream::Colors::GREEN) |
| 412 | .Case(S: "\033[33m" , Value: raw_ostream::Colors::YELLOW) |
| 413 | .Case(S: "\033[34m" , Value: raw_ostream::Colors::BLUE) |
| 414 | .Case(S: "\033[35m" , Value: raw_ostream::Colors::MAGENTA) |
| 415 | .Case(S: "\033[36m" , Value: raw_ostream::Colors::CYAN) |
| 416 | .Case(S: "\033[37m" , Value: raw_ostream::Colors::WHITE) |
| 417 | .Default(Value: std::nullopt); |
| 418 | if (SGRColor) { |
| 419 | Color = *SGRColor; |
| 420 | if (ColorsEnabled) |
| 421 | OS.changeColor(Color: *Color); |
| 422 | return true; |
| 423 | } |
| 424 | |
| 425 | return false; |
| 426 | } |
| 427 | |
| 428 | // Begin highlighting text by picking a different color than the current color |
| 429 | // state. |
| 430 | void MarkupFilter::highlight() { |
| 431 | if (!ColorsEnabled) |
| 432 | return; |
| 433 | OS.changeColor(Color: Color == raw_ostream::Colors::BLUE ? raw_ostream::Colors::CYAN |
| 434 | : raw_ostream::Colors::BLUE, |
| 435 | Bold); |
| 436 | } |
| 437 | |
| 438 | // Begin highlighting a field within a highlighted markup string. |
| 439 | void MarkupFilter::highlightValue() { |
| 440 | if (!ColorsEnabled) |
| 441 | return; |
| 442 | OS.changeColor(Color: raw_ostream::Colors::GREEN, Bold); |
| 443 | } |
| 444 | |
| 445 | // Set the output stream's color to the current color and bold state of the SGR |
| 446 | // abstract machine. |
| 447 | void MarkupFilter::restoreColor() { |
| 448 | if (!ColorsEnabled) |
| 449 | return; |
| 450 | if (Color) { |
| 451 | OS.changeColor(Color: *Color, Bold); |
| 452 | } else { |
| 453 | OS.resetColor(); |
| 454 | if (Bold) |
| 455 | OS.changeColor(Color: raw_ostream::Colors::SAVEDCOLOR, Bold); |
| 456 | } |
| 457 | } |
| 458 | |
| 459 | // Set the SGR and output stream's color and bold states back to the default. |
| 460 | void MarkupFilter::resetColor() { |
| 461 | if (!Color && !Bold) |
| 462 | return; |
| 463 | Color.reset(); |
| 464 | Bold = false; |
| 465 | if (ColorsEnabled) |
| 466 | OS.resetColor(); |
| 467 | } |
| 468 | |
| 469 | void MarkupFilter::printRawElement(const MarkupNode &Element) { |
| 470 | highlight(); |
| 471 | OS << "[[[" ; |
| 472 | printValue(Value: Element.Tag); |
| 473 | for (StringRef Field : Element.Fields) { |
| 474 | OS << ':'; |
| 475 | printValue(Value: Field); |
| 476 | } |
| 477 | OS << "]]]" ; |
| 478 | restoreColor(); |
| 479 | } |
| 480 | |
| 481 | void MarkupFilter::printValue(Twine Value) { |
| 482 | highlightValue(); |
| 483 | OS << Value; |
| 484 | highlight(); |
| 485 | } |
| 486 | |
| 487 | // This macro helps reduce the amount of indirection done through Optional |
| 488 | // below, since the usual case upon returning a std::nullopt Optional is to |
| 489 | // return std::nullopt. |
| 490 | #define ASSIGN_OR_RETURN_NONE(TYPE, NAME, EXPR) \ |
| 491 | auto NAME##Opt = (EXPR); \ |
| 492 | if (!NAME##Opt) \ |
| 493 | return std::nullopt; \ |
| 494 | TYPE NAME = std::move(*NAME##Opt) |
| 495 | |
| 496 | std::optional<MarkupFilter::Module> |
| 497 | MarkupFilter::parseModule(const MarkupNode &Element) const { |
| 498 | if (!checkNumFieldsAtLeast(Element, Size: 3)) |
| 499 | return std::nullopt; |
| 500 | ASSIGN_OR_RETURN_NONE(uint64_t, ID, parseModuleID(Element.Fields[0])); |
| 501 | StringRef Name = Element.Fields[1]; |
| 502 | StringRef Type = Element.Fields[2]; |
| 503 | if (Type != "elf" ) { |
| 504 | WithColor::error() << "unknown module type\n" ; |
| 505 | reportLocation(Loc: Type.begin()); |
| 506 | return std::nullopt; |
| 507 | } |
| 508 | if (!checkNumFields(Element, Size: 4)) |
| 509 | return std::nullopt; |
| 510 | SmallVector<uint8_t> BuildID = parseBuildID(Str: Element.Fields[3]); |
| 511 | if (BuildID.empty()) |
| 512 | return std::nullopt; |
| 513 | return Module{.ID: ID, .Name: Name.str(), .BuildID: std::move(BuildID)}; |
| 514 | } |
| 515 | |
| 516 | std::optional<MarkupFilter::MMap> |
| 517 | MarkupFilter::parseMMap(const MarkupNode &Element) const { |
| 518 | if (!checkNumFieldsAtLeast(Element, Size: 3)) |
| 519 | return std::nullopt; |
| 520 | ASSIGN_OR_RETURN_NONE(uint64_t, Addr, parseAddr(Element.Fields[0])); |
| 521 | ASSIGN_OR_RETURN_NONE(uint64_t, Size, parseSize(Element.Fields[1])); |
| 522 | StringRef Type = Element.Fields[2]; |
| 523 | if (Type != "load" ) { |
| 524 | WithColor::error() << "unknown mmap type\n" ; |
| 525 | reportLocation(Loc: Type.begin()); |
| 526 | return std::nullopt; |
| 527 | } |
| 528 | if (!checkNumFields(Element, Size: 6)) |
| 529 | return std::nullopt; |
| 530 | ASSIGN_OR_RETURN_NONE(uint64_t, ID, parseModuleID(Element.Fields[3])); |
| 531 | ASSIGN_OR_RETURN_NONE(std::string, Mode, parseMode(Element.Fields[4])); |
| 532 | auto It = Modules.find(Val: ID); |
| 533 | if (It == Modules.end()) { |
| 534 | WithColor::error() << "unknown module ID\n" ; |
| 535 | reportLocation(Loc: Element.Fields[3].begin()); |
| 536 | return std::nullopt; |
| 537 | } |
| 538 | ASSIGN_OR_RETURN_NONE(uint64_t, ModuleRelativeAddr, |
| 539 | parseAddr(Element.Fields[5])); |
| 540 | return MMap{.Addr: Addr, .Size: Size, .Mod: It->second.get(), .Mode: std::move(Mode), |
| 541 | .ModuleRelativeAddr: ModuleRelativeAddr}; |
| 542 | } |
| 543 | |
| 544 | // Parse an address (%p in the spec). |
| 545 | std::optional<uint64_t> MarkupFilter::parseAddr(StringRef Str) const { |
| 546 | if (Str.empty()) { |
| 547 | reportTypeError(Str, TypeName: "address" ); |
| 548 | return std::nullopt; |
| 549 | } |
| 550 | if (all_of(Range&: Str, P: [](char C) { return C == '0'; })) |
| 551 | return 0; |
| 552 | if (!Str.starts_with(Prefix: "0x" )) { |
| 553 | reportTypeError(Str, TypeName: "address" ); |
| 554 | return std::nullopt; |
| 555 | } |
| 556 | uint64_t Addr; |
| 557 | if (Str.drop_front(N: 2).getAsInteger(Radix: 16, Result&: Addr)) { |
| 558 | reportTypeError(Str, TypeName: "address" ); |
| 559 | return std::nullopt; |
| 560 | } |
| 561 | return Addr; |
| 562 | } |
| 563 | |
| 564 | // Parse a module ID (%i in the spec). |
| 565 | std::optional<uint64_t> MarkupFilter::parseModuleID(StringRef Str) const { |
| 566 | uint64_t ID; |
| 567 | if (Str.getAsInteger(Radix: 0, Result&: ID)) { |
| 568 | reportTypeError(Str, TypeName: "module ID" ); |
| 569 | return std::nullopt; |
| 570 | } |
| 571 | return ID; |
| 572 | } |
| 573 | |
| 574 | // Parse a size (%i in the spec). |
| 575 | std::optional<uint64_t> MarkupFilter::parseSize(StringRef Str) const { |
| 576 | uint64_t ID; |
| 577 | if (Str.getAsInteger(Radix: 0, Result&: ID)) { |
| 578 | reportTypeError(Str, TypeName: "size" ); |
| 579 | return std::nullopt; |
| 580 | } |
| 581 | return ID; |
| 582 | } |
| 583 | |
| 584 | // Parse a frame number (%i in the spec). |
| 585 | std::optional<uint64_t> MarkupFilter::(StringRef Str) const { |
| 586 | uint64_t ID; |
| 587 | if (Str.getAsInteger(Radix: 10, Result&: ID)) { |
| 588 | reportTypeError(Str, TypeName: "frame number" ); |
| 589 | return std::nullopt; |
| 590 | } |
| 591 | return ID; |
| 592 | } |
| 593 | |
| 594 | // Parse a build ID (%x in the spec). |
| 595 | object::BuildID MarkupFilter::parseBuildID(StringRef Str) const { |
| 596 | object::BuildID BID = llvm::object::parseBuildID(Str); |
| 597 | if (BID.empty()) |
| 598 | reportTypeError(Str, TypeName: "build ID" ); |
| 599 | return BID; |
| 600 | } |
| 601 | |
| 602 | // Parses the mode string for an mmap element. |
| 603 | std::optional<std::string> MarkupFilter::parseMode(StringRef Str) const { |
| 604 | if (Str.empty()) { |
| 605 | reportTypeError(Str, TypeName: "mode" ); |
| 606 | return std::nullopt; |
| 607 | } |
| 608 | |
| 609 | // Pop off each of r/R, w/W, and x/X from the front, in that order. |
| 610 | StringRef Remainder = Str; |
| 611 | Remainder.consume_front_insensitive(Prefix: "r" ); |
| 612 | Remainder.consume_front_insensitive(Prefix: "w" ); |
| 613 | Remainder.consume_front_insensitive(Prefix: "x" ); |
| 614 | |
| 615 | // If anything remains, then the string wasn't a mode. |
| 616 | if (!Remainder.empty()) { |
| 617 | reportTypeError(Str, TypeName: "mode" ); |
| 618 | return std::nullopt; |
| 619 | } |
| 620 | |
| 621 | // Normalize the mode. |
| 622 | return Str.lower(); |
| 623 | } |
| 624 | |
| 625 | std::optional<MarkupFilter::PCType> |
| 626 | MarkupFilter::parsePCType(StringRef Str) const { |
| 627 | std::optional<MarkupFilter::PCType> Type = |
| 628 | StringSwitch<std::optional<MarkupFilter::PCType>>(Str) |
| 629 | .Case(S: "ra" , Value: MarkupFilter::PCType::ReturnAddress) |
| 630 | .Case(S: "pc" , Value: MarkupFilter::PCType::PreciseCode) |
| 631 | .Default(Value: std::nullopt); |
| 632 | if (!Type) |
| 633 | reportTypeError(Str, TypeName: "PC type" ); |
| 634 | return Type; |
| 635 | } |
| 636 | |
| 637 | bool MarkupFilter::checkTag(const MarkupNode &Node) const { |
| 638 | if (any_of(Range: Node.Tag, P: [](char C) { return C < 'a' || C > 'z'; })) { |
| 639 | WithColor::error(OS&: errs()) << "tags must be all lowercase characters\n" ; |
| 640 | reportLocation(Loc: Node.Tag.begin()); |
| 641 | return false; |
| 642 | } |
| 643 | return true; |
| 644 | } |
| 645 | |
| 646 | bool MarkupFilter::checkNumFields(const MarkupNode &Element, |
| 647 | size_t Size) const { |
| 648 | if (Element.Fields.size() != Size) { |
| 649 | bool Warn = Element.Fields.size() > Size; |
| 650 | WithColor(errs(), Warn ? HighlightColor::Warning : HighlightColor::Error) |
| 651 | << (Warn ? "warning: " : "error: " ) << "expected " << Size |
| 652 | << " field(s); found " << Element.Fields.size() << "\n" ; |
| 653 | reportLocation(Loc: Element.Tag.end()); |
| 654 | return Warn; |
| 655 | } |
| 656 | return true; |
| 657 | } |
| 658 | |
| 659 | bool MarkupFilter::checkNumFieldsAtLeast(const MarkupNode &Element, |
| 660 | size_t Size) const { |
| 661 | if (Element.Fields.size() < Size) { |
| 662 | WithColor::error(OS&: errs()) |
| 663 | << "expected at least " << Size << " field(s); found " |
| 664 | << Element.Fields.size() << "\n" ; |
| 665 | reportLocation(Loc: Element.Tag.end()); |
| 666 | return false; |
| 667 | } |
| 668 | return true; |
| 669 | } |
| 670 | |
| 671 | void MarkupFilter::warnNumFieldsAtMost(const MarkupNode &Element, |
| 672 | size_t Size) const { |
| 673 | if (Element.Fields.size() <= Size) |
| 674 | return; |
| 675 | WithColor::warning(OS&: errs()) |
| 676 | << "expected at most " << Size << " field(s); found " |
| 677 | << Element.Fields.size() << "\n" ; |
| 678 | reportLocation(Loc: Element.Tag.end()); |
| 679 | } |
| 680 | |
| 681 | void MarkupFilter::reportTypeError(StringRef Str, StringRef TypeName) const { |
| 682 | WithColor::error(OS&: errs()) << "expected " << TypeName << "; found '" << Str |
| 683 | << "'\n" ; |
| 684 | reportLocation(Loc: Str.begin()); |
| 685 | } |
| 686 | |
| 687 | // Prints two lines that point out the given location in the current Line using |
| 688 | // a caret. The iterator must be within the bounds of the most recent line |
| 689 | // passed to beginLine(). |
| 690 | void MarkupFilter::reportLocation(StringRef::iterator Loc) const { |
| 691 | errs() << Line; |
| 692 | WithColor(errs().indent(NumSpaces: Loc - StringRef(Line).begin()), |
| 693 | HighlightColor::String) |
| 694 | << '^'; |
| 695 | errs() << '\n'; |
| 696 | } |
| 697 | |
| 698 | // Checks for an existing mmap that overlaps the given one and returns a |
| 699 | // pointer to one of them. |
| 700 | const MarkupFilter::MMap * |
| 701 | MarkupFilter::getOverlappingMMap(const MMap &Map) const { |
| 702 | // If the given map contains the start of another mmap, they overlap. |
| 703 | auto I = MMaps.upper_bound(x: Map.Addr); |
| 704 | if (I != MMaps.end() && Map.contains(Addr: I->second.Addr)) |
| 705 | return &I->second; |
| 706 | |
| 707 | // If no element starts inside the given mmap, the only possible overlap would |
| 708 | // be if the preceding mmap contains the start point of the given mmap. |
| 709 | if (I != MMaps.begin()) { |
| 710 | --I; |
| 711 | if (I->second.contains(Addr: Map.Addr)) |
| 712 | return &I->second; |
| 713 | } |
| 714 | return nullptr; |
| 715 | } |
| 716 | |
| 717 | // Returns the MMap that contains the given address or nullptr if none. |
| 718 | const MarkupFilter::MMap *MarkupFilter::getContainingMMap(uint64_t Addr) const { |
| 719 | // Find the first mmap starting >= Addr. |
| 720 | auto I = MMaps.lower_bound(x: Addr); |
| 721 | if (I != MMaps.end() && I->second.contains(Addr)) |
| 722 | return &I->second; |
| 723 | |
| 724 | // The previous mmap is the last one starting < Addr. |
| 725 | if (I == MMaps.begin()) |
| 726 | return nullptr; |
| 727 | --I; |
| 728 | return I->second.contains(Addr) ? &I->second : nullptr; |
| 729 | } |
| 730 | |
| 731 | uint64_t MarkupFilter::adjustAddr(uint64_t Addr, PCType Type) const { |
| 732 | // Decrementing return addresses by one moves them into the call instruction. |
| 733 | // The address doesn't have to be the start of the call instruction, just some |
| 734 | // byte on the inside. Subtracting one avoids needing detailed instruction |
| 735 | // length information here. |
| 736 | return Type == MarkupFilter::PCType::ReturnAddress ? Addr - 1 : Addr; |
| 737 | } |
| 738 | |
| 739 | StringRef MarkupFilter::lineEnding() const { |
| 740 | return StringRef(Line).ends_with(Suffix: "\r\n" ) ? "\r\n" : "\n" ; |
| 741 | } |
| 742 | |
| 743 | bool MarkupFilter::MMap::contains(uint64_t Addr) const { |
| 744 | return this->Addr <= Addr && Addr < this->Addr + Size; |
| 745 | } |
| 746 | |
| 747 | // Returns the module-relative address for a given virtual address. |
| 748 | uint64_t MarkupFilter::MMap::getModuleRelativeAddr(uint64_t Addr) const { |
| 749 | return Addr - this->Addr + ModuleRelativeAddr; |
| 750 | } |
| 751 | |