1 //===- Driver.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 #include "Driver.h"
10 #include "Config.h"
11 #include "ICF.h"
12 #include "InputFiles.h"
13 #include "LTO.h"
14 #include "MarkLive.h"
15 #include "ObjC.h"
16 #include "OutputSection.h"
17 #include "OutputSegment.h"
18 #include "SectionPriorities.h"
19 #include "SymbolTable.h"
20 #include "Symbols.h"
21 #include "SyntheticSections.h"
22 #include "Target.h"
23 #include "UnwindInfoSection.h"
24 #include "Writer.h"
25 
26 #include "lld/Common/Args.h"
27 #include "lld/Common/CommonLinkerContext.h"
28 #include "lld/Common/Driver.h"
29 #include "lld/Common/ErrorHandler.h"
30 #include "lld/Common/LLVM.h"
31 #include "lld/Common/Memory.h"
32 #include "lld/Common/Reproduce.h"
33 #include "lld/Common/Version.h"
34 #include "llvm/ADT/DenseSet.h"
35 #include "llvm/ADT/StringExtras.h"
36 #include "llvm/ADT/StringRef.h"
37 #include "llvm/BinaryFormat/MachO.h"
38 #include "llvm/BinaryFormat/Magic.h"
39 #include "llvm/Config/llvm-config.h"
40 #include "llvm/LTO/LTO.h"
41 #include "llvm/Object/Archive.h"
42 #include "llvm/Option/ArgList.h"
43 #include "llvm/Support/CommandLine.h"
44 #include "llvm/Support/FileSystem.h"
45 #include "llvm/Support/MemoryBuffer.h"
46 #include "llvm/Support/Parallel.h"
47 #include "llvm/Support/Path.h"
48 #include "llvm/Support/TarWriter.h"
49 #include "llvm/Support/TargetSelect.h"
50 #include "llvm/Support/TimeProfiler.h"
51 #include "llvm/TargetParser/Host.h"
52 #include "llvm/TextAPI/PackedVersion.h"
53 
54 #include <algorithm>
55 
56 using namespace llvm;
57 using namespace llvm::MachO;
58 using namespace llvm::object;
59 using namespace llvm::opt;
60 using namespace llvm::sys;
61 using namespace lld;
62 using namespace lld::macho;
63 
64 std::unique_ptr<Configuration> macho::config;
65 std::unique_ptr<DependencyTracker> macho::depTracker;
66 
getOutputType(const InputArgList & args)67 static HeaderFileType getOutputType(const InputArgList &args) {
68   // TODO: -r, -dylinker, -preload...
69   Arg *outputArg = args.getLastArg(OPT_bundle, OPT_dylib, OPT_execute);
70   if (outputArg == nullptr)
71     return MH_EXECUTE;
72 
73   switch (outputArg->getOption().getID()) {
74   case OPT_bundle:
75     return MH_BUNDLE;
76   case OPT_dylib:
77     return MH_DYLIB;
78   case OPT_execute:
79     return MH_EXECUTE;
80   default:
81     llvm_unreachable("internal error");
82   }
83 }
84 
85 static DenseMap<CachedHashStringRef, StringRef> resolvedLibraries;
findLibrary(StringRef name)86 static std::optional<StringRef> findLibrary(StringRef name) {
87   CachedHashStringRef key(name);
88   auto entry = resolvedLibraries.find(key);
89   if (entry != resolvedLibraries.end())
90     return entry->second;
91 
92   auto doFind = [&] {
93     // Special case for Csu support files required for Mac OS X 10.7 and older
94     // (crt1.o)
95     if (name.ends_with(".o"))
96       return findPathCombination(name, config->librarySearchPaths, {""});
97     if (config->searchDylibsFirst) {
98       if (std::optional<StringRef> path =
99               findPathCombination("lib" + name, config->librarySearchPaths,
100                                   {".tbd", ".dylib", ".so"}))
101         return path;
102       return findPathCombination("lib" + name, config->librarySearchPaths,
103                                  {".a"});
104     }
105     return findPathCombination("lib" + name, config->librarySearchPaths,
106                                {".tbd", ".dylib", ".so", ".a"});
107   };
108 
109   std::optional<StringRef> path = doFind();
110   if (path)
111     resolvedLibraries[key] = *path;
112 
113   return path;
114 }
115 
116 static DenseMap<CachedHashStringRef, StringRef> resolvedFrameworks;
findFramework(StringRef name)117 static std::optional<StringRef> findFramework(StringRef name) {
118   CachedHashStringRef key(name);
119   auto entry = resolvedFrameworks.find(key);
120   if (entry != resolvedFrameworks.end())
121     return entry->second;
122 
123   SmallString<260> symlink;
124   StringRef suffix;
125   std::tie(name, suffix) = name.split(",");
126   for (StringRef dir : config->frameworkSearchPaths) {
127     symlink = dir;
128     path::append(symlink, name + ".framework", name);
129 
130     if (!suffix.empty()) {
131       // NOTE: we must resolve the symlink before trying the suffixes, because
132       // there are no symlinks for the suffixed paths.
133       SmallString<260> location;
134       if (!fs::real_path(symlink, location)) {
135         // only append suffix if realpath() succeeds
136         Twine suffixed = location + suffix;
137         if (fs::exists(suffixed))
138           return resolvedFrameworks[key] = saver().save(suffixed.str());
139       }
140       // Suffix lookup failed, fall through to the no-suffix case.
141     }
142 
143     if (std::optional<StringRef> path = resolveDylibPath(symlink.str()))
144       return resolvedFrameworks[key] = *path;
145   }
146   return {};
147 }
148 
warnIfNotDirectory(StringRef option,StringRef path)149 static bool warnIfNotDirectory(StringRef option, StringRef path) {
150   if (!fs::exists(path)) {
151     warn("directory not found for option -" + option + path);
152     return false;
153   } else if (!fs::is_directory(path)) {
154     warn("option -" + option + path + " references a non-directory path");
155     return false;
156   }
157   return true;
158 }
159 
160 static std::vector<StringRef>
getSearchPaths(unsigned optionCode,InputArgList & args,const std::vector<StringRef> & roots,const SmallVector<StringRef,2> & systemPaths)161 getSearchPaths(unsigned optionCode, InputArgList &args,
162                const std::vector<StringRef> &roots,
163                const SmallVector<StringRef, 2> &systemPaths) {
164   std::vector<StringRef> paths;
165   StringRef optionLetter{optionCode == OPT_F ? "F" : "L"};
166   for (StringRef path : args::getStrings(args, optionCode)) {
167     // NOTE: only absolute paths are re-rooted to syslibroot(s)
168     bool found = false;
169     if (path::is_absolute(path, path::Style::posix)) {
170       for (StringRef root : roots) {
171         SmallString<261> buffer(root);
172         path::append(buffer, path);
173         // Do not warn about paths that are computed via the syslib roots
174         if (fs::is_directory(buffer)) {
175           paths.push_back(saver().save(buffer.str()));
176           found = true;
177         }
178       }
179     }
180     if (!found && warnIfNotDirectory(optionLetter, path))
181       paths.push_back(path);
182   }
183 
184   // `-Z` suppresses the standard "system" search paths.
185   if (args.hasArg(OPT_Z))
186     return paths;
187 
188   for (const StringRef &path : systemPaths) {
189     for (const StringRef &root : roots) {
190       SmallString<261> buffer(root);
191       path::append(buffer, path);
192       if (fs::is_directory(buffer))
193         paths.push_back(saver().save(buffer.str()));
194     }
195   }
196   return paths;
197 }
198 
getSystemLibraryRoots(InputArgList & args)199 static std::vector<StringRef> getSystemLibraryRoots(InputArgList &args) {
200   std::vector<StringRef> roots;
201   for (const Arg *arg : args.filtered(OPT_syslibroot))
202     roots.push_back(arg->getValue());
203   // NOTE: the final `-syslibroot` being `/` will ignore all roots
204   if (!roots.empty() && roots.back() == "/")
205     roots.clear();
206   // NOTE: roots can never be empty - add an empty root to simplify the library
207   // and framework search path computation.
208   if (roots.empty())
209     roots.emplace_back("");
210   return roots;
211 }
212 
213 static std::vector<StringRef>
getLibrarySearchPaths(InputArgList & args,const std::vector<StringRef> & roots)214 getLibrarySearchPaths(InputArgList &args, const std::vector<StringRef> &roots) {
215   return getSearchPaths(OPT_L, args, roots, {"/usr/lib", "/usr/local/lib"});
216 }
217 
218 static std::vector<StringRef>
getFrameworkSearchPaths(InputArgList & args,const std::vector<StringRef> & roots)219 getFrameworkSearchPaths(InputArgList &args,
220                         const std::vector<StringRef> &roots) {
221   return getSearchPaths(OPT_F, args, roots,
222                         {"/Library/Frameworks", "/System/Library/Frameworks"});
223 }
224 
getLTOCachePolicy(InputArgList & args)225 static llvm::CachePruningPolicy getLTOCachePolicy(InputArgList &args) {
226   SmallString<128> ltoPolicy;
227   auto add = [&ltoPolicy](Twine val) {
228     if (!ltoPolicy.empty())
229       ltoPolicy += ":";
230     val.toVector(ltoPolicy);
231   };
232   for (const Arg *arg :
233        args.filtered(OPT_thinlto_cache_policy_eq, OPT_prune_interval_lto,
234                      OPT_prune_after_lto, OPT_max_relative_cache_size_lto)) {
235     switch (arg->getOption().getID()) {
236     case OPT_thinlto_cache_policy_eq:
237       add(arg->getValue());
238       break;
239     case OPT_prune_interval_lto:
240       if (!strcmp("-1", arg->getValue()))
241         add("prune_interval=87600h"); // 10 years
242       else
243         add(Twine("prune_interval=") + arg->getValue() + "s");
244       break;
245     case OPT_prune_after_lto:
246       add(Twine("prune_after=") + arg->getValue() + "s");
247       break;
248     case OPT_max_relative_cache_size_lto:
249       add(Twine("cache_size=") + arg->getValue() + "%");
250       break;
251     }
252   }
253   return CHECK(parseCachePruningPolicy(ltoPolicy), "invalid LTO cache policy");
254 }
255 
256 // What caused a given library to be loaded. Only relevant for archives.
257 // Note that this does not tell us *how* we should load the library, i.e.
258 // whether we should do it lazily or eagerly (AKA force loading). The "how" is
259 // decided within addFile().
260 enum class LoadType {
261   CommandLine,      // Library was passed as a regular CLI argument
262   CommandLineForce, // Library was passed via `-force_load`
263   LCLinkerOption,   // Library was passed via LC_LINKER_OPTIONS
264 };
265 
266 struct ArchiveFileInfo {
267   ArchiveFile *file;
268   bool isCommandLineLoad;
269 };
270 
271 static DenseMap<StringRef, ArchiveFileInfo> loadedArchives;
272 
addFile(StringRef path,LoadType loadType,bool isLazy=false,bool isExplicit=true,bool isBundleLoader=false,bool isForceHidden=false)273 static InputFile *addFile(StringRef path, LoadType loadType,
274                           bool isLazy = false, bool isExplicit = true,
275                           bool isBundleLoader = false,
276                           bool isForceHidden = false) {
277   std::optional<MemoryBufferRef> buffer = readFile(path);
278   if (!buffer)
279     return nullptr;
280   MemoryBufferRef mbref = *buffer;
281   InputFile *newFile = nullptr;
282 
283   file_magic magic = identify_magic(mbref.getBuffer());
284   switch (magic) {
285   case file_magic::archive: {
286     bool isCommandLineLoad = loadType != LoadType::LCLinkerOption;
287     // Avoid loading archives twice. If the archives are being force-loaded,
288     // loading them twice would create duplicate symbol errors. In the
289     // non-force-loading case, this is just a minor performance optimization.
290     // We don't take a reference to cachedFile here because the
291     // loadArchiveMember() call below may recursively call addFile() and
292     // invalidate this reference.
293     auto entry = loadedArchives.find(path);
294 
295     ArchiveFile *file;
296     if (entry == loadedArchives.end()) {
297       // No cached archive, we need to create a new one
298       std::unique_ptr<object::Archive> archive = CHECK(
299           object::Archive::create(mbref), path + ": failed to parse archive");
300 
301       if (!archive->isEmpty() && !archive->hasSymbolTable())
302         error(path + ": archive has no index; run ranlib to add one");
303       file = make<ArchiveFile>(std::move(archive), isForceHidden);
304     } else {
305       file = entry->second.file;
306       // Command-line loads take precedence. If file is previously loaded via
307       // command line, or is loaded via LC_LINKER_OPTION and being loaded via
308       // LC_LINKER_OPTION again, using the cached archive is enough.
309       if (entry->second.isCommandLineLoad || !isCommandLineLoad)
310         return file;
311     }
312 
313     bool isLCLinkerForceLoad = loadType == LoadType::LCLinkerOption &&
314                                config->forceLoadSwift &&
315                                path::filename(path).starts_with("libswift");
316     if ((isCommandLineLoad && config->allLoad) ||
317         loadType == LoadType::CommandLineForce || isLCLinkerForceLoad) {
318       if (readFile(path)) {
319         Error e = Error::success();
320         for (const object::Archive::Child &c : file->getArchive().children(e)) {
321           StringRef reason;
322           switch (loadType) {
323             case LoadType::LCLinkerOption:
324               reason = "LC_LINKER_OPTION";
325               break;
326             case LoadType::CommandLineForce:
327               reason = "-force_load";
328               break;
329             case LoadType::CommandLine:
330               reason = "-all_load";
331               break;
332           }
333           if (Error e = file->fetch(c, reason))
334             error(toString(file) + ": " + reason +
335                   " failed to load archive member: " + toString(std::move(e)));
336         }
337         if (e)
338           error(toString(file) +
339                 ": Archive::children failed: " + toString(std::move(e)));
340       }
341     } else if (isCommandLineLoad && config->forceLoadObjC) {
342       for (const object::Archive::Symbol &sym : file->getArchive().symbols())
343         if (sym.getName().starts_with(objc::klass))
344           file->fetch(sym);
345 
346       // TODO: no need to look for ObjC sections for a given archive member if
347       // we already found that it contains an ObjC symbol.
348       if (readFile(path)) {
349         Error e = Error::success();
350         for (const object::Archive::Child &c : file->getArchive().children(e)) {
351           Expected<MemoryBufferRef> mb = c.getMemoryBufferRef();
352           if (!mb || !hasObjCSection(*mb))
353             continue;
354           if (Error e = file->fetch(c, "-ObjC"))
355             error(toString(file) + ": -ObjC failed to load archive member: " +
356                   toString(std::move(e)));
357         }
358         if (e)
359           error(toString(file) +
360                 ": Archive::children failed: " + toString(std::move(e)));
361       }
362     }
363 
364     file->addLazySymbols();
365     loadedArchives[path] = ArchiveFileInfo{file, isCommandLineLoad};
366     newFile = file;
367     break;
368   }
369   case file_magic::macho_object:
370     newFile = make<ObjFile>(mbref, getModTime(path), "", isLazy);
371     break;
372   case file_magic::macho_dynamically_linked_shared_lib:
373   case file_magic::macho_dynamically_linked_shared_lib_stub:
374   case file_magic::tapi_file:
375     if (DylibFile *dylibFile =
376             loadDylib(mbref, nullptr, /*isBundleLoader=*/false, isExplicit))
377       newFile = dylibFile;
378     break;
379   case file_magic::bitcode:
380     newFile = make<BitcodeFile>(mbref, "", 0, isLazy);
381     break;
382   case file_magic::macho_executable:
383   case file_magic::macho_bundle:
384     // We only allow executable and bundle type here if it is used
385     // as a bundle loader.
386     if (!isBundleLoader)
387       error(path + ": unhandled file type");
388     if (DylibFile *dylibFile = loadDylib(mbref, nullptr, isBundleLoader))
389       newFile = dylibFile;
390     break;
391   default:
392     error(path + ": unhandled file type");
393   }
394   if (newFile && !isa<DylibFile>(newFile)) {
395     if ((isa<ObjFile>(newFile) || isa<BitcodeFile>(newFile)) && newFile->lazy &&
396         config->forceLoadObjC) {
397       for (Symbol *sym : newFile->symbols)
398         if (sym && sym->getName().starts_with(objc::klass)) {
399           extract(*newFile, "-ObjC");
400           break;
401         }
402       if (newFile->lazy && hasObjCSection(mbref))
403         extract(*newFile, "-ObjC");
404     }
405 
406     // printArchiveMemberLoad() prints both .a and .o names, so no need to
407     // print the .a name here. Similarly skip lazy files.
408     if (config->printEachFile && magic != file_magic::archive && !isLazy)
409       message(toString(newFile));
410     inputFiles.insert(newFile);
411   }
412   return newFile;
413 }
414 
415 static std::vector<StringRef> missingAutolinkWarnings;
addLibrary(StringRef name,bool isNeeded,bool isWeak,bool isReexport,bool isHidden,bool isExplicit,LoadType loadType)416 static void addLibrary(StringRef name, bool isNeeded, bool isWeak,
417                        bool isReexport, bool isHidden, bool isExplicit,
418                        LoadType loadType) {
419   if (std::optional<StringRef> path = findLibrary(name)) {
420     if (auto *dylibFile = dyn_cast_or_null<DylibFile>(
421             addFile(*path, loadType, /*isLazy=*/false, isExplicit,
422                     /*isBundleLoader=*/false, isHidden))) {
423       if (isNeeded)
424         dylibFile->forceNeeded = true;
425       if (isWeak)
426         dylibFile->forceWeakImport = true;
427       if (isReexport) {
428         config->hasReexports = true;
429         dylibFile->reexport = true;
430       }
431     }
432     return;
433   }
434   if (loadType == LoadType::LCLinkerOption) {
435     missingAutolinkWarnings.push_back(
436         saver().save("auto-linked library not found for -l" + name));
437     return;
438   }
439   error("library not found for -l" + name);
440 }
441 
442 static DenseSet<StringRef> loadedObjectFrameworks;
addFramework(StringRef name,bool isNeeded,bool isWeak,bool isReexport,bool isExplicit,LoadType loadType)443 static void addFramework(StringRef name, bool isNeeded, bool isWeak,
444                          bool isReexport, bool isExplicit, LoadType loadType) {
445   if (std::optional<StringRef> path = findFramework(name)) {
446     if (loadedObjectFrameworks.contains(*path))
447       return;
448 
449     InputFile *file =
450         addFile(*path, loadType, /*isLazy=*/false, isExplicit, false);
451     if (auto *dylibFile = dyn_cast_or_null<DylibFile>(file)) {
452       if (isNeeded)
453         dylibFile->forceNeeded = true;
454       if (isWeak)
455         dylibFile->forceWeakImport = true;
456       if (isReexport) {
457         config->hasReexports = true;
458         dylibFile->reexport = true;
459       }
460     } else if (isa_and_nonnull<ObjFile>(file) ||
461                isa_and_nonnull<BitcodeFile>(file)) {
462       // Cache frameworks containing object or bitcode files to avoid duplicate
463       // symbols. Frameworks containing static archives are cached separately
464       // in addFile() to share caching with libraries, and frameworks
465       // containing dylibs should allow overwriting of attributes such as
466       // forceNeeded by subsequent loads
467       loadedObjectFrameworks.insert(*path);
468     }
469     return;
470   }
471   if (loadType == LoadType::LCLinkerOption) {
472     missingAutolinkWarnings.push_back(
473         saver().save("auto-linked framework not found for -framework " + name));
474     return;
475   }
476   error("framework not found for -framework " + name);
477 }
478 
479 // Parses LC_LINKER_OPTION contents, which can add additional command line
480 // flags. This directly parses the flags instead of using the standard argument
481 // parser to improve performance.
parseLCLinkerOption(llvm::SmallVectorImpl<StringRef> & LCLinkerOptions,InputFile * f,unsigned argc,StringRef data)482 void macho::parseLCLinkerOption(
483     llvm::SmallVectorImpl<StringRef> &LCLinkerOptions, InputFile *f,
484     unsigned argc, StringRef data) {
485   if (config->ignoreAutoLink)
486     return;
487 
488   SmallVector<StringRef, 4> argv;
489   size_t offset = 0;
490   for (unsigned i = 0; i < argc && offset < data.size(); ++i) {
491     argv.push_back(data.data() + offset);
492     offset += strlen(data.data() + offset) + 1;
493   }
494   if (argv.size() != argc || offset > data.size())
495     fatal(toString(f) + ": invalid LC_LINKER_OPTION");
496 
497   unsigned i = 0;
498   StringRef arg = argv[i];
499   if (arg.consume_front("-l")) {
500     if (config->ignoreAutoLinkOptions.contains(arg))
501       return;
502   } else if (arg == "-framework") {
503     StringRef name = argv[++i];
504     if (config->ignoreAutoLinkOptions.contains(name))
505       return;
506   } else {
507     error(arg + " is not allowed in LC_LINKER_OPTION");
508   }
509 
510   LCLinkerOptions.append(argv);
511 }
512 
resolveLCLinkerOptions()513 void macho::resolveLCLinkerOptions() {
514   while (!unprocessedLCLinkerOptions.empty()) {
515     SmallVector<StringRef> LCLinkerOptions(unprocessedLCLinkerOptions);
516     unprocessedLCLinkerOptions.clear();
517 
518     for (unsigned i = 0; i < LCLinkerOptions.size(); ++i) {
519       StringRef arg = LCLinkerOptions[i];
520       if (arg.consume_front("-l")) {
521         assert(!config->ignoreAutoLinkOptions.contains(arg));
522         addLibrary(arg, /*isNeeded=*/false, /*isWeak=*/false,
523                    /*isReexport=*/false, /*isHidden=*/false,
524                    /*isExplicit=*/false, LoadType::LCLinkerOption);
525       } else if (arg == "-framework") {
526         StringRef name = LCLinkerOptions[++i];
527         assert(!config->ignoreAutoLinkOptions.contains(name));
528         addFramework(name, /*isNeeded=*/false, /*isWeak=*/false,
529                      /*isReexport=*/false, /*isExplicit=*/false,
530                      LoadType::LCLinkerOption);
531       } else {
532         error(arg + " is not allowed in LC_LINKER_OPTION");
533       }
534     }
535   }
536 }
537 
addFileList(StringRef path,bool isLazy)538 static void addFileList(StringRef path, bool isLazy) {
539   std::optional<MemoryBufferRef> buffer = readFile(path);
540   if (!buffer)
541     return;
542   MemoryBufferRef mbref = *buffer;
543   for (StringRef path : args::getLines(mbref))
544     addFile(rerootPath(path), LoadType::CommandLine, isLazy);
545 }
546 
547 // We expect sub-library names of the form "libfoo", which will match a dylib
548 // with a path of .*/libfoo.{dylib, tbd}.
549 // XXX ld64 seems to ignore the extension entirely when matching sub-libraries;
550 // I'm not sure what the use case for that is.
markReexport(StringRef searchName,ArrayRef<StringRef> extensions)551 static bool markReexport(StringRef searchName, ArrayRef<StringRef> extensions) {
552   for (InputFile *file : inputFiles) {
553     if (auto *dylibFile = dyn_cast<DylibFile>(file)) {
554       StringRef filename = path::filename(dylibFile->getName());
555       if (filename.consume_front(searchName) &&
556           (filename.empty() || llvm::is_contained(extensions, filename))) {
557         dylibFile->reexport = true;
558         return true;
559       }
560     }
561   }
562   return false;
563 }
564 
565 // This function is called on startup. We need this for LTO since
566 // LTO calls LLVM functions to compile bitcode files to native code.
567 // Technically this can be delayed until we read bitcode files, but
568 // we don't bother to do lazily because the initialization is fast.
initLLVM()569 static void initLLVM() {
570   InitializeAllTargets();
571   InitializeAllTargetMCs();
572   InitializeAllAsmPrinters();
573   InitializeAllAsmParsers();
574 }
575 
compileBitcodeFiles()576 static bool compileBitcodeFiles() {
577   TimeTraceScope timeScope("LTO");
578   auto *lto = make<BitcodeCompiler>();
579   for (InputFile *file : inputFiles)
580     if (auto *bitcodeFile = dyn_cast<BitcodeFile>(file))
581       if (!file->lazy)
582         lto->add(*bitcodeFile);
583 
584   std::vector<ObjFile *> compiled = lto->compile();
585   for (ObjFile *file : compiled)
586     inputFiles.insert(file);
587 
588   return !compiled.empty();
589 }
590 
591 // Replaces common symbols with defined symbols residing in __common sections.
592 // This function must be called after all symbol names are resolved (i.e. after
593 // all InputFiles have been loaded.) As a result, later operations won't see
594 // any CommonSymbols.
replaceCommonSymbols()595 static void replaceCommonSymbols() {
596   TimeTraceScope timeScope("Replace common symbols");
597   ConcatOutputSection *osec = nullptr;
598   for (Symbol *sym : symtab->getSymbols()) {
599     auto *common = dyn_cast<CommonSymbol>(sym);
600     if (common == nullptr)
601       continue;
602 
603     // Casting to size_t will truncate large values on 32-bit architectures,
604     // but it's not really worth supporting the linking of 64-bit programs on
605     // 32-bit archs.
606     ArrayRef<uint8_t> data = {nullptr, static_cast<size_t>(common->size)};
607     // FIXME avoid creating one Section per symbol?
608     auto *section =
609         make<Section>(common->getFile(), segment_names::data,
610                       section_names::common, S_ZEROFILL, /*addr=*/0);
611     auto *isec = make<ConcatInputSection>(*section, data, common->align);
612     if (!osec)
613       osec = ConcatOutputSection::getOrCreateForInput(isec);
614     isec->parent = osec;
615     inputSections.push_back(isec);
616 
617     // FIXME: CommonSymbol should store isReferencedDynamically, noDeadStrip
618     // and pass them on here.
619     replaceSymbol<Defined>(
620         sym, sym->getName(), common->getFile(), isec, /*value=*/0, common->size,
621         /*isWeakDef=*/false, /*isExternal=*/true, common->privateExtern,
622         /*includeInSymtab=*/true, /*isReferencedDynamically=*/false,
623         /*noDeadStrip=*/false);
624   }
625 }
626 
initializeSectionRenameMap()627 static void initializeSectionRenameMap() {
628   if (config->dataConst) {
629     SmallVector<StringRef> v{section_names::got,
630                              section_names::authGot,
631                              section_names::authPtr,
632                              section_names::nonLazySymbolPtr,
633                              section_names::const_,
634                              section_names::cfString,
635                              section_names::moduleInitFunc,
636                              section_names::moduleTermFunc,
637                              section_names::objcClassList,
638                              section_names::objcNonLazyClassList,
639                              section_names::objcCatList,
640                              section_names::objcNonLazyCatList,
641                              section_names::objcProtoList,
642                              section_names::objCImageInfo};
643     for (StringRef s : v)
644       config->sectionRenameMap[{segment_names::data, s}] = {
645           segment_names::dataConst, s};
646   }
647   config->sectionRenameMap[{segment_names::text, section_names::staticInit}] = {
648       segment_names::text, section_names::text};
649   config->sectionRenameMap[{segment_names::import, section_names::pointers}] = {
650       config->dataConst ? segment_names::dataConst : segment_names::data,
651       section_names::nonLazySymbolPtr};
652 }
653 
toLowerDash(char x)654 static inline char toLowerDash(char x) {
655   if (x >= 'A' && x <= 'Z')
656     return x - 'A' + 'a';
657   else if (x == ' ')
658     return '-';
659   return x;
660 }
661 
lowerDash(StringRef s)662 static std::string lowerDash(StringRef s) {
663   return std::string(map_iterator(s.begin(), toLowerDash),
664                      map_iterator(s.end(), toLowerDash));
665 }
666 
667 struct PlatformVersion {
668   PlatformType platform = PLATFORM_UNKNOWN;
669   llvm::VersionTuple minimum;
670   llvm::VersionTuple sdk;
671 };
672 
parsePlatformVersion(const Arg * arg)673 static PlatformVersion parsePlatformVersion(const Arg *arg) {
674   assert(arg->getOption().getID() == OPT_platform_version);
675   StringRef platformStr = arg->getValue(0);
676   StringRef minVersionStr = arg->getValue(1);
677   StringRef sdkVersionStr = arg->getValue(2);
678 
679   PlatformVersion platformVersion;
680 
681   // TODO(compnerd) see if we can generate this case list via XMACROS
682   platformVersion.platform =
683       StringSwitch<PlatformType>(lowerDash(platformStr))
684           .Cases("macos", "1", PLATFORM_MACOS)
685           .Cases("ios", "2", PLATFORM_IOS)
686           .Cases("tvos", "3", PLATFORM_TVOS)
687           .Cases("watchos", "4", PLATFORM_WATCHOS)
688           .Cases("bridgeos", "5", PLATFORM_BRIDGEOS)
689           .Cases("mac-catalyst", "6", PLATFORM_MACCATALYST)
690           .Cases("ios-simulator", "7", PLATFORM_IOSSIMULATOR)
691           .Cases("tvos-simulator", "8", PLATFORM_TVOSSIMULATOR)
692           .Cases("watchos-simulator", "9", PLATFORM_WATCHOSSIMULATOR)
693           .Cases("driverkit", "10", PLATFORM_DRIVERKIT)
694           .Default(PLATFORM_UNKNOWN);
695   if (platformVersion.platform == PLATFORM_UNKNOWN)
696     error(Twine("malformed platform: ") + platformStr);
697   // TODO: check validity of version strings, which varies by platform
698   // NOTE: ld64 accepts version strings with 5 components
699   // llvm::VersionTuple accepts no more than 4 components
700   // Has Apple ever published version strings with 5 components?
701   if (platformVersion.minimum.tryParse(minVersionStr))
702     error(Twine("malformed minimum version: ") + minVersionStr);
703   if (platformVersion.sdk.tryParse(sdkVersionStr))
704     error(Twine("malformed sdk version: ") + sdkVersionStr);
705   return platformVersion;
706 }
707 
708 // Has the side-effect of setting Config::platformInfo and
709 // potentially Config::secondaryPlatformInfo.
setPlatformVersions(StringRef archName,const ArgList & args)710 static void setPlatformVersions(StringRef archName, const ArgList &args) {
711   std::map<PlatformType, PlatformVersion> platformVersions;
712   const PlatformVersion *lastVersionInfo = nullptr;
713   for (const Arg *arg : args.filtered(OPT_platform_version)) {
714     PlatformVersion version = parsePlatformVersion(arg);
715 
716     // For each platform, the last flag wins:
717     // `-platform_version macos 2 3 -platform_version macos 4 5` has the same
718     // effect as just passing `-platform_version macos 4 5`.
719     // FIXME: ld64 warns on multiple flags for one platform. Should we?
720     platformVersions[version.platform] = version;
721     lastVersionInfo = &platformVersions[version.platform];
722   }
723 
724   if (platformVersions.empty()) {
725     error("must specify -platform_version");
726     return;
727   }
728   if (platformVersions.size() > 2) {
729     error("must specify -platform_version at most twice");
730     return;
731   }
732   if (platformVersions.size() == 2) {
733     bool isZipperedCatalyst = platformVersions.count(PLATFORM_MACOS) &&
734                               platformVersions.count(PLATFORM_MACCATALYST);
735 
736     if (!isZipperedCatalyst) {
737       error("lld supports writing zippered outputs only for "
738             "macos and mac-catalyst");
739     } else if (config->outputType != MH_DYLIB &&
740                config->outputType != MH_BUNDLE) {
741       error("writing zippered outputs only valid for -dylib and -bundle");
742     }
743 
744     config->platformInfo = {
745         MachO::Target(getArchitectureFromName(archName), PLATFORM_MACOS,
746                       platformVersions[PLATFORM_MACOS].minimum),
747         platformVersions[PLATFORM_MACOS].sdk};
748     config->secondaryPlatformInfo = {
749         MachO::Target(getArchitectureFromName(archName), PLATFORM_MACCATALYST,
750                       platformVersions[PLATFORM_MACCATALYST].minimum),
751         platformVersions[PLATFORM_MACCATALYST].sdk};
752     return;
753   }
754 
755   config->platformInfo = {MachO::Target(getArchitectureFromName(archName),
756                                         lastVersionInfo->platform,
757                                         lastVersionInfo->minimum),
758                           lastVersionInfo->sdk};
759 }
760 
761 // Has the side-effect of setting Config::target.
createTargetInfo(InputArgList & args)762 static TargetInfo *createTargetInfo(InputArgList &args) {
763   StringRef archName = args.getLastArgValue(OPT_arch);
764   if (archName.empty()) {
765     error("must specify -arch");
766     return nullptr;
767   }
768 
769   setPlatformVersions(archName, args);
770   auto [cpuType, cpuSubtype] = getCPUTypeFromArchitecture(config->arch());
771   switch (cpuType) {
772   case CPU_TYPE_X86_64:
773     return createX86_64TargetInfo();
774   case CPU_TYPE_ARM64:
775     return createARM64TargetInfo();
776   case CPU_TYPE_ARM64_32:
777     return createARM64_32TargetInfo();
778   default:
779     error("missing or unsupported -arch " + archName);
780     return nullptr;
781   }
782 }
783 
784 static UndefinedSymbolTreatment
getUndefinedSymbolTreatment(const ArgList & args)785 getUndefinedSymbolTreatment(const ArgList &args) {
786   StringRef treatmentStr = args.getLastArgValue(OPT_undefined);
787   auto treatment =
788       StringSwitch<UndefinedSymbolTreatment>(treatmentStr)
789           .Cases("error", "", UndefinedSymbolTreatment::error)
790           .Case("warning", UndefinedSymbolTreatment::warning)
791           .Case("suppress", UndefinedSymbolTreatment::suppress)
792           .Case("dynamic_lookup", UndefinedSymbolTreatment::dynamic_lookup)
793           .Default(UndefinedSymbolTreatment::unknown);
794   if (treatment == UndefinedSymbolTreatment::unknown) {
795     warn(Twine("unknown -undefined TREATMENT '") + treatmentStr +
796          "', defaulting to 'error'");
797     treatment = UndefinedSymbolTreatment::error;
798   } else if (config->namespaceKind == NamespaceKind::twolevel &&
799              (treatment == UndefinedSymbolTreatment::warning ||
800               treatment == UndefinedSymbolTreatment::suppress)) {
801     if (treatment == UndefinedSymbolTreatment::warning)
802       fatal("'-undefined warning' only valid with '-flat_namespace'");
803     else
804       fatal("'-undefined suppress' only valid with '-flat_namespace'");
805     treatment = UndefinedSymbolTreatment::error;
806   }
807   return treatment;
808 }
809 
getICFLevel(const ArgList & args)810 static ICFLevel getICFLevel(const ArgList &args) {
811   StringRef icfLevelStr = args.getLastArgValue(OPT_icf_eq);
812   auto icfLevel = StringSwitch<ICFLevel>(icfLevelStr)
813                       .Cases("none", "", ICFLevel::none)
814                       .Case("safe", ICFLevel::safe)
815                       .Case("all", ICFLevel::all)
816                       .Default(ICFLevel::unknown);
817   if (icfLevel == ICFLevel::unknown) {
818     warn(Twine("unknown --icf=OPTION `") + icfLevelStr +
819          "', defaulting to `none'");
820     icfLevel = ICFLevel::none;
821   }
822   return icfLevel;
823 }
824 
getObjCStubsMode(const ArgList & args)825 static ObjCStubsMode getObjCStubsMode(const ArgList &args) {
826   const Arg *arg = args.getLastArg(OPT_objc_stubs_fast, OPT_objc_stubs_small);
827   if (!arg)
828     return ObjCStubsMode::fast;
829 
830   if (arg->getOption().getID() == OPT_objc_stubs_small) {
831     if (is_contained({AK_arm64e, AK_arm64}, config->arch()))
832       return ObjCStubsMode::small;
833     else
834       warn("-objc_stubs_small is not yet implemented, defaulting to "
835            "-objc_stubs_fast");
836   }
837   return ObjCStubsMode::fast;
838 }
839 
warnIfDeprecatedOption(const Option & opt)840 static void warnIfDeprecatedOption(const Option &opt) {
841   if (!opt.getGroup().isValid())
842     return;
843   if (opt.getGroup().getID() == OPT_grp_deprecated) {
844     warn("Option `" + opt.getPrefixedName() + "' is deprecated in ld64:");
845     warn(opt.getHelpText());
846   }
847 }
848 
warnIfUnimplementedOption(const Option & opt)849 static void warnIfUnimplementedOption(const Option &opt) {
850   if (!opt.getGroup().isValid() || !opt.hasFlag(DriverFlag::HelpHidden))
851     return;
852   switch (opt.getGroup().getID()) {
853   case OPT_grp_deprecated:
854     // warn about deprecated options elsewhere
855     break;
856   case OPT_grp_undocumented:
857     warn("Option `" + opt.getPrefixedName() +
858          "' is undocumented. Should lld implement it?");
859     break;
860   case OPT_grp_obsolete:
861     warn("Option `" + opt.getPrefixedName() +
862          "' is obsolete. Please modernize your usage.");
863     break;
864   case OPT_grp_ignored:
865     warn("Option `" + opt.getPrefixedName() + "' is ignored.");
866     break;
867   case OPT_grp_ignored_silently:
868     break;
869   default:
870     warn("Option `" + opt.getPrefixedName() +
871          "' is not yet implemented. Stay tuned...");
872     break;
873   }
874 }
875 
getReproduceOption(InputArgList & args)876 static const char *getReproduceOption(InputArgList &args) {
877   if (const Arg *arg = args.getLastArg(OPT_reproduce))
878     return arg->getValue();
879   return getenv("LLD_REPRODUCE");
880 }
881 
882 // Parse options of the form "old;new".
getOldNewOptions(opt::InputArgList & args,unsigned id)883 static std::pair<StringRef, StringRef> getOldNewOptions(opt::InputArgList &args,
884                                                         unsigned id) {
885   auto *arg = args.getLastArg(id);
886   if (!arg)
887     return {"", ""};
888 
889   StringRef s = arg->getValue();
890   std::pair<StringRef, StringRef> ret = s.split(';');
891   if (ret.second.empty())
892     error(arg->getSpelling() + " expects 'old;new' format, but got " + s);
893   return ret;
894 }
895 
896 // Parse options of the form "old;new[;extra]".
897 static std::tuple<StringRef, StringRef, StringRef>
getOldNewOptionsExtra(opt::InputArgList & args,unsigned id)898 getOldNewOptionsExtra(opt::InputArgList &args, unsigned id) {
899   auto [oldDir, second] = getOldNewOptions(args, id);
900   auto [newDir, extraDir] = second.split(';');
901   return {oldDir, newDir, extraDir};
902 }
903 
parseClangOption(StringRef opt,const Twine & msg)904 static void parseClangOption(StringRef opt, const Twine &msg) {
905   std::string err;
906   raw_string_ostream os(err);
907 
908   const char *argv[] = {"lld", opt.data()};
909   if (cl::ParseCommandLineOptions(2, argv, "", &os))
910     return;
911   os.flush();
912   error(msg + ": " + StringRef(err).trim());
913 }
914 
parseDylibVersion(const ArgList & args,unsigned id)915 static uint32_t parseDylibVersion(const ArgList &args, unsigned id) {
916   const Arg *arg = args.getLastArg(id);
917   if (!arg)
918     return 0;
919 
920   if (config->outputType != MH_DYLIB) {
921     error(arg->getAsString(args) + ": only valid with -dylib");
922     return 0;
923   }
924 
925   PackedVersion version;
926   if (!version.parse32(arg->getValue())) {
927     error(arg->getAsString(args) + ": malformed version");
928     return 0;
929   }
930 
931   return version.rawValue();
932 }
933 
parseProtection(StringRef protStr)934 static uint32_t parseProtection(StringRef protStr) {
935   uint32_t prot = 0;
936   for (char c : protStr) {
937     switch (c) {
938     case 'r':
939       prot |= VM_PROT_READ;
940       break;
941     case 'w':
942       prot |= VM_PROT_WRITE;
943       break;
944     case 'x':
945       prot |= VM_PROT_EXECUTE;
946       break;
947     case '-':
948       break;
949     default:
950       error("unknown -segprot letter '" + Twine(c) + "' in " + protStr);
951       return 0;
952     }
953   }
954   return prot;
955 }
956 
parseSectAlign(const opt::InputArgList & args)957 static std::vector<SectionAlign> parseSectAlign(const opt::InputArgList &args) {
958   std::vector<SectionAlign> sectAligns;
959   for (const Arg *arg : args.filtered(OPT_sectalign)) {
960     StringRef segName = arg->getValue(0);
961     StringRef sectName = arg->getValue(1);
962     StringRef alignStr = arg->getValue(2);
963     alignStr.consume_front_insensitive("0x");
964     uint32_t align;
965     if (alignStr.getAsInteger(16, align)) {
966       error("-sectalign: failed to parse '" + StringRef(arg->getValue(2)) +
967             "' as number");
968       continue;
969     }
970     if (!isPowerOf2_32(align)) {
971       error("-sectalign: '" + StringRef(arg->getValue(2)) +
972             "' (in base 16) not a power of two");
973       continue;
974     }
975     sectAligns.push_back({segName, sectName, align});
976   }
977   return sectAligns;
978 }
979 
removeSimulator(PlatformType platform)980 PlatformType macho::removeSimulator(PlatformType platform) {
981   switch (platform) {
982   case PLATFORM_IOSSIMULATOR:
983     return PLATFORM_IOS;
984   case PLATFORM_TVOSSIMULATOR:
985     return PLATFORM_TVOS;
986   case PLATFORM_WATCHOSSIMULATOR:
987     return PLATFORM_WATCHOS;
988   default:
989     return platform;
990   }
991 }
992 
supportsNoPie()993 static bool supportsNoPie() {
994   return !(config->arch() == AK_arm64 || config->arch() == AK_arm64e ||
995            config->arch() == AK_arm64_32);
996 }
997 
shouldAdhocSignByDefault(Architecture arch,PlatformType platform)998 static bool shouldAdhocSignByDefault(Architecture arch, PlatformType platform) {
999   if (arch != AK_arm64 && arch != AK_arm64e)
1000     return false;
1001 
1002   return platform == PLATFORM_MACOS || platform == PLATFORM_IOSSIMULATOR ||
1003          platform == PLATFORM_TVOSSIMULATOR ||
1004          platform == PLATFORM_WATCHOSSIMULATOR;
1005 }
1006 
dataConstDefault(const InputArgList & args)1007 static bool dataConstDefault(const InputArgList &args) {
1008   static const std::array<std::pair<PlatformType, VersionTuple>, 5> minVersion =
1009       {{{PLATFORM_MACOS, VersionTuple(10, 15)},
1010         {PLATFORM_IOS, VersionTuple(13, 0)},
1011         {PLATFORM_TVOS, VersionTuple(13, 0)},
1012         {PLATFORM_WATCHOS, VersionTuple(6, 0)},
1013         {PLATFORM_BRIDGEOS, VersionTuple(4, 0)}}};
1014   PlatformType platform = removeSimulator(config->platformInfo.target.Platform);
1015   auto it = llvm::find_if(minVersion,
1016                           [&](const auto &p) { return p.first == platform; });
1017   if (it != minVersion.end())
1018     if (config->platformInfo.target.MinDeployment < it->second)
1019       return false;
1020 
1021   switch (config->outputType) {
1022   case MH_EXECUTE:
1023     return !(args.hasArg(OPT_no_pie) && supportsNoPie());
1024   case MH_BUNDLE:
1025     // FIXME: return false when -final_name ...
1026     // has prefix "/System/Library/UserEventPlugins/"
1027     // or matches "/usr/libexec/locationd" "/usr/libexec/terminusd"
1028     return true;
1029   case MH_DYLIB:
1030     return true;
1031   case MH_OBJECT:
1032     return false;
1033   default:
1034     llvm_unreachable(
1035         "unsupported output type for determining data-const default");
1036   }
1037   return false;
1038 }
1039 
shouldEmitChainedFixups(const InputArgList & args)1040 static bool shouldEmitChainedFixups(const InputArgList &args) {
1041   const Arg *arg = args.getLastArg(OPT_fixup_chains, OPT_no_fixup_chains);
1042   if (arg && arg->getOption().matches(OPT_no_fixup_chains))
1043     return false;
1044 
1045   bool isRequested = arg != nullptr;
1046 
1047   // Version numbers taken from the Xcode 13.3 release notes.
1048   static const std::array<std::pair<PlatformType, VersionTuple>, 4> minVersion =
1049       {{{PLATFORM_MACOS, VersionTuple(11, 0)},
1050         {PLATFORM_IOS, VersionTuple(13, 4)},
1051         {PLATFORM_TVOS, VersionTuple(14, 0)},
1052         {PLATFORM_WATCHOS, VersionTuple(7, 0)}}};
1053   PlatformType platform = removeSimulator(config->platformInfo.target.Platform);
1054   auto it = llvm::find_if(minVersion,
1055                           [&](const auto &p) { return p.first == platform; });
1056   if (it != minVersion.end() &&
1057       it->second > config->platformInfo.target.MinDeployment) {
1058     if (!isRequested)
1059       return false;
1060 
1061     warn("-fixup_chains requires " + getPlatformName(config->platform()) + " " +
1062          it->second.getAsString() + ", which is newer than target minimum of " +
1063          config->platformInfo.target.MinDeployment.getAsString());
1064   }
1065 
1066   if (!is_contained({AK_x86_64, AK_x86_64h, AK_arm64}, config->arch())) {
1067     if (isRequested)
1068       error("-fixup_chains is only supported on x86_64 and arm64 targets");
1069     return false;
1070   }
1071 
1072   if (!config->isPic) {
1073     if (isRequested)
1074       error("-fixup_chains is incompatible with -no_pie");
1075     return false;
1076   }
1077 
1078   // TODO: Enable by default once stable.
1079   return isRequested;
1080 }
1081 
clear()1082 void SymbolPatterns::clear() {
1083   literals.clear();
1084   globs.clear();
1085 }
1086 
insert(StringRef symbolName)1087 void SymbolPatterns::insert(StringRef symbolName) {
1088   if (symbolName.find_first_of("*?[]") == StringRef::npos)
1089     literals.insert(CachedHashStringRef(symbolName));
1090   else if (Expected<GlobPattern> pattern = GlobPattern::create(symbolName))
1091     globs.emplace_back(*pattern);
1092   else
1093     error("invalid symbol-name pattern: " + symbolName);
1094 }
1095 
matchLiteral(StringRef symbolName) const1096 bool SymbolPatterns::matchLiteral(StringRef symbolName) const {
1097   return literals.contains(CachedHashStringRef(symbolName));
1098 }
1099 
matchGlob(StringRef symbolName) const1100 bool SymbolPatterns::matchGlob(StringRef symbolName) const {
1101   for (const GlobPattern &glob : globs)
1102     if (glob.match(symbolName))
1103       return true;
1104   return false;
1105 }
1106 
match(StringRef symbolName) const1107 bool SymbolPatterns::match(StringRef symbolName) const {
1108   return matchLiteral(symbolName) || matchGlob(symbolName);
1109 }
1110 
parseSymbolPatternsFile(const Arg * arg,SymbolPatterns & symbolPatterns)1111 static void parseSymbolPatternsFile(const Arg *arg,
1112                                     SymbolPatterns &symbolPatterns) {
1113   StringRef path = arg->getValue();
1114   std::optional<MemoryBufferRef> buffer = readFile(path);
1115   if (!buffer) {
1116     error("Could not read symbol file: " + path);
1117     return;
1118   }
1119   MemoryBufferRef mbref = *buffer;
1120   for (StringRef line : args::getLines(mbref)) {
1121     line = line.take_until([](char c) { return c == '#'; }).trim();
1122     if (!line.empty())
1123       symbolPatterns.insert(line);
1124   }
1125 }
1126 
handleSymbolPatterns(InputArgList & args,SymbolPatterns & symbolPatterns,unsigned singleOptionCode,unsigned listFileOptionCode)1127 static void handleSymbolPatterns(InputArgList &args,
1128                                  SymbolPatterns &symbolPatterns,
1129                                  unsigned singleOptionCode,
1130                                  unsigned listFileOptionCode) {
1131   for (const Arg *arg : args.filtered(singleOptionCode))
1132     symbolPatterns.insert(arg->getValue());
1133   for (const Arg *arg : args.filtered(listFileOptionCode))
1134     parseSymbolPatternsFile(arg, symbolPatterns);
1135 }
1136 
createFiles(const InputArgList & args)1137 static void createFiles(const InputArgList &args) {
1138   TimeTraceScope timeScope("Load input files");
1139   // This loop should be reserved for options whose exact ordering matters.
1140   // Other options should be handled via filtered() and/or getLastArg().
1141   bool isLazy = false;
1142   for (const Arg *arg : args) {
1143     const Option &opt = arg->getOption();
1144     warnIfDeprecatedOption(opt);
1145     warnIfUnimplementedOption(opt);
1146 
1147     switch (opt.getID()) {
1148     case OPT_INPUT:
1149       addFile(rerootPath(arg->getValue()), LoadType::CommandLine, isLazy);
1150       break;
1151     case OPT_needed_library:
1152       if (auto *dylibFile = dyn_cast_or_null<DylibFile>(
1153               addFile(rerootPath(arg->getValue()), LoadType::CommandLine)))
1154         dylibFile->forceNeeded = true;
1155       break;
1156     case OPT_reexport_library:
1157       if (auto *dylibFile = dyn_cast_or_null<DylibFile>(
1158               addFile(rerootPath(arg->getValue()), LoadType::CommandLine))) {
1159         config->hasReexports = true;
1160         dylibFile->reexport = true;
1161       }
1162       break;
1163     case OPT_weak_library:
1164       if (auto *dylibFile = dyn_cast_or_null<DylibFile>(
1165               addFile(rerootPath(arg->getValue()), LoadType::CommandLine)))
1166         dylibFile->forceWeakImport = true;
1167       break;
1168     case OPT_filelist:
1169       addFileList(arg->getValue(), isLazy);
1170       break;
1171     case OPT_force_load:
1172       addFile(rerootPath(arg->getValue()), LoadType::CommandLineForce);
1173       break;
1174     case OPT_load_hidden:
1175       addFile(rerootPath(arg->getValue()), LoadType::CommandLine,
1176               /*isLazy=*/false, /*isExplicit=*/true, /*isBundleLoader=*/false,
1177               /*isForceHidden=*/true);
1178       break;
1179     case OPT_l:
1180     case OPT_needed_l:
1181     case OPT_reexport_l:
1182     case OPT_weak_l:
1183     case OPT_hidden_l:
1184       addLibrary(arg->getValue(), opt.getID() == OPT_needed_l,
1185                  opt.getID() == OPT_weak_l, opt.getID() == OPT_reexport_l,
1186                  opt.getID() == OPT_hidden_l,
1187                  /*isExplicit=*/true, LoadType::CommandLine);
1188       break;
1189     case OPT_framework:
1190     case OPT_needed_framework:
1191     case OPT_reexport_framework:
1192     case OPT_weak_framework:
1193       addFramework(arg->getValue(), opt.getID() == OPT_needed_framework,
1194                    opt.getID() == OPT_weak_framework,
1195                    opt.getID() == OPT_reexport_framework, /*isExplicit=*/true,
1196                    LoadType::CommandLine);
1197       break;
1198     case OPT_start_lib:
1199       if (isLazy)
1200         error("nested --start-lib");
1201       isLazy = true;
1202       break;
1203     case OPT_end_lib:
1204       if (!isLazy)
1205         error("stray --end-lib");
1206       isLazy = false;
1207       break;
1208     default:
1209       break;
1210     }
1211   }
1212 }
1213 
gatherInputSections()1214 static void gatherInputSections() {
1215   TimeTraceScope timeScope("Gathering input sections");
1216   int inputOrder = 0;
1217   for (const InputFile *file : inputFiles) {
1218     for (const Section *section : file->sections) {
1219       // Compact unwind entries require special handling elsewhere. (In
1220       // contrast, EH frames are handled like regular ConcatInputSections.)
1221       if (section->name == section_names::compactUnwind)
1222         continue;
1223       ConcatOutputSection *osec = nullptr;
1224       for (const Subsection &subsection : section->subsections) {
1225         if (auto *isec = dyn_cast<ConcatInputSection>(subsection.isec)) {
1226           if (isec->isCoalescedWeak())
1227             continue;
1228           if (config->emitInitOffsets &&
1229               sectionType(isec->getFlags()) == S_MOD_INIT_FUNC_POINTERS) {
1230             in.initOffsets->addInput(isec);
1231             continue;
1232           }
1233           isec->outSecOff = inputOrder++;
1234           if (!osec)
1235             osec = ConcatOutputSection::getOrCreateForInput(isec);
1236           isec->parent = osec;
1237           inputSections.push_back(isec);
1238         } else if (auto *isec =
1239                        dyn_cast<CStringInputSection>(subsection.isec)) {
1240           if (isec->getName() == section_names::objcMethname) {
1241             if (in.objcMethnameSection->inputOrder == UnspecifiedInputOrder)
1242               in.objcMethnameSection->inputOrder = inputOrder++;
1243             in.objcMethnameSection->addInput(isec);
1244           } else {
1245             if (in.cStringSection->inputOrder == UnspecifiedInputOrder)
1246               in.cStringSection->inputOrder = inputOrder++;
1247             in.cStringSection->addInput(isec);
1248           }
1249         } else if (auto *isec =
1250                        dyn_cast<WordLiteralInputSection>(subsection.isec)) {
1251           if (in.wordLiteralSection->inputOrder == UnspecifiedInputOrder)
1252             in.wordLiteralSection->inputOrder = inputOrder++;
1253           in.wordLiteralSection->addInput(isec);
1254         } else {
1255           llvm_unreachable("unexpected input section kind");
1256         }
1257       }
1258     }
1259     if (!file->objCImageInfo.empty())
1260       in.objCImageInfo->addFile(file);
1261   }
1262   assert(inputOrder <= UnspecifiedInputOrder);
1263 }
1264 
foldIdenticalLiterals()1265 static void foldIdenticalLiterals() {
1266   TimeTraceScope timeScope("Fold identical literals");
1267   // We always create a cStringSection, regardless of whether dedupLiterals is
1268   // true. If it isn't, we simply create a non-deduplicating CStringSection.
1269   // Either way, we must unconditionally finalize it here.
1270   in.cStringSection->finalizeContents();
1271   in.objcMethnameSection->finalizeContents();
1272   in.wordLiteralSection->finalizeContents();
1273 }
1274 
addSynthenticMethnames()1275 static void addSynthenticMethnames() {
1276   std::string &data = *make<std::string>();
1277   llvm::raw_string_ostream os(data);
1278   const int prefixLength = ObjCStubsSection::symbolPrefix.size();
1279   for (Symbol *sym : symtab->getSymbols())
1280     if (isa<Undefined>(sym))
1281       if (sym->getName().starts_with(ObjCStubsSection::symbolPrefix))
1282         os << sym->getName().drop_front(prefixLength) << '\0';
1283 
1284   if (data.empty())
1285     return;
1286 
1287   const auto *buf = reinterpret_cast<const uint8_t *>(data.c_str());
1288   Section &section = *make<Section>(/*file=*/nullptr, segment_names::text,
1289                                     section_names::objcMethname,
1290                                     S_CSTRING_LITERALS, /*addr=*/0);
1291 
1292   auto *isec =
1293       make<CStringInputSection>(section, ArrayRef<uint8_t>{buf, data.size()},
1294                                 /*align=*/1, /*dedupLiterals=*/true);
1295   isec->splitIntoPieces();
1296   for (auto &piece : isec->pieces)
1297     piece.live = true;
1298   section.subsections.push_back({0, isec});
1299   in.objcMethnameSection->addInput(isec);
1300   in.objcMethnameSection->isec->markLive(0);
1301 }
1302 
referenceStubBinder()1303 static void referenceStubBinder() {
1304   bool needsStubHelper = config->outputType == MH_DYLIB ||
1305                          config->outputType == MH_EXECUTE ||
1306                          config->outputType == MH_BUNDLE;
1307   if (!needsStubHelper || !symtab->find("dyld_stub_binder"))
1308     return;
1309 
1310   // dyld_stub_binder is used by dyld to resolve lazy bindings. This code here
1311   // adds a opportunistic reference to dyld_stub_binder if it happens to exist.
1312   // dyld_stub_binder is in libSystem.dylib, which is usually linked in. This
1313   // isn't needed for correctness, but the presence of that symbol suppresses
1314   // "no symbols" diagnostics from `nm`.
1315   // StubHelperSection::setUp() adds a reference and errors out if
1316   // dyld_stub_binder doesn't exist in case it is actually needed.
1317   symtab->addUndefined("dyld_stub_binder", /*file=*/nullptr, /*isWeak=*/false);
1318 }
1319 
createAliases()1320 static void createAliases() {
1321   for (const auto &pair : config->aliasedSymbols) {
1322     if (const auto &sym = symtab->find(pair.first)) {
1323       if (const auto &defined = dyn_cast<Defined>(sym)) {
1324         symtab->aliasDefined(defined, pair.second, defined->getFile())
1325             ->noDeadStrip = true;
1326       } else {
1327         error("TODO: support aliasing to symbols of kind " +
1328               Twine(sym->kind()));
1329       }
1330     } else {
1331       warn("undefined base symbol '" + pair.first + "' for alias '" +
1332            pair.second + "'\n");
1333     }
1334   }
1335 
1336   for (const InputFile *file : inputFiles) {
1337     if (auto *objFile = dyn_cast<ObjFile>(file)) {
1338       for (const AliasSymbol *alias : objFile->aliases) {
1339         if (const auto &aliased = symtab->find(alias->getAliasedName())) {
1340           if (const auto &defined = dyn_cast<Defined>(aliased)) {
1341             symtab->aliasDefined(defined, alias->getName(), alias->getFile(),
1342                                  alias->privateExtern);
1343           } else {
1344             // Common, dylib, and undefined symbols are all valid alias
1345             // referents (undefineds can become valid Defined symbols later on
1346             // in the link.)
1347             error("TODO: support aliasing to symbols of kind " +
1348                   Twine(aliased->kind()));
1349           }
1350         } else {
1351           // This shouldn't happen since MC generates undefined symbols to
1352           // represent the alias referents. Thus we fatal() instead of just
1353           // warning here.
1354           fatal("unable to find alias referent " + alias->getAliasedName() +
1355                 " for " + alias->getName());
1356         }
1357       }
1358     }
1359   }
1360 }
1361 
handleExplicitExports()1362 static void handleExplicitExports() {
1363   static constexpr int kMaxWarnings = 3;
1364   if (config->hasExplicitExports) {
1365     std::atomic<uint64_t> warningsCount{0};
1366     parallelForEach(symtab->getSymbols(), [&warningsCount](Symbol *sym) {
1367       if (auto *defined = dyn_cast<Defined>(sym)) {
1368         if (config->exportedSymbols.match(sym->getName())) {
1369           if (defined->privateExtern) {
1370             if (defined->weakDefCanBeHidden) {
1371               // weak_def_can_be_hidden symbols behave similarly to
1372               // private_extern symbols in most cases, except for when
1373               // it is explicitly exported.
1374               // The former can be exported but the latter cannot.
1375               defined->privateExtern = false;
1376             } else {
1377               // Only print the first 3 warnings verbosely, and
1378               // shorten the rest to avoid crowding logs.
1379               if (warningsCount.fetch_add(1, std::memory_order_relaxed) <
1380                   kMaxWarnings)
1381                 warn("cannot export hidden symbol " + toString(*defined) +
1382                      "\n>>> defined in " + toString(defined->getFile()));
1383             }
1384           }
1385         } else {
1386           defined->privateExtern = true;
1387         }
1388       } else if (auto *dysym = dyn_cast<DylibSymbol>(sym)) {
1389         dysym->shouldReexport = config->exportedSymbols.match(sym->getName());
1390       }
1391     });
1392     if (warningsCount > kMaxWarnings)
1393       warn("<... " + Twine(warningsCount - kMaxWarnings) +
1394            " more similar warnings...>");
1395   } else if (!config->unexportedSymbols.empty()) {
1396     parallelForEach(symtab->getSymbols(), [](Symbol *sym) {
1397       if (auto *defined = dyn_cast<Defined>(sym))
1398         if (config->unexportedSymbols.match(defined->getName()))
1399           defined->privateExtern = true;
1400     });
1401   }
1402 }
1403 
1404 namespace lld {
1405 namespace macho {
link(ArrayRef<const char * > argsArr,llvm::raw_ostream & stdoutOS,llvm::raw_ostream & stderrOS,bool exitEarly,bool disableOutput)1406 bool link(ArrayRef<const char *> argsArr, llvm::raw_ostream &stdoutOS,
1407           llvm::raw_ostream &stderrOS, bool exitEarly, bool disableOutput) {
1408   // This driver-specific context will be freed later by lldMain().
1409   auto *ctx = new CommonLinkerContext;
1410 
1411   ctx->e.initialize(stdoutOS, stderrOS, exitEarly, disableOutput);
1412   ctx->e.cleanupCallback = []() {
1413     resolvedFrameworks.clear();
1414     resolvedLibraries.clear();
1415     cachedReads.clear();
1416     concatOutputSections.clear();
1417     inputFiles.clear();
1418     inputSections.clear();
1419     loadedArchives.clear();
1420     loadedObjectFrameworks.clear();
1421     missingAutolinkWarnings.clear();
1422     syntheticSections.clear();
1423     thunkMap.clear();
1424     unprocessedLCLinkerOptions.clear();
1425 
1426     firstTLVDataSection = nullptr;
1427     tar = nullptr;
1428     memset(&in, 0, sizeof(in));
1429 
1430     resetLoadedDylibs();
1431     resetOutputSegments();
1432     resetWriter();
1433     InputFile::resetIdCount();
1434   };
1435 
1436   ctx->e.logName = args::getFilenameWithoutExe(argsArr[0]);
1437 
1438   MachOOptTable parser;
1439   InputArgList args = parser.parse(argsArr.slice(1));
1440 
1441   ctx->e.errorLimitExceededMsg = "too many errors emitted, stopping now "
1442                                  "(use --error-limit=0 to see all errors)";
1443   ctx->e.errorLimit = args::getInteger(args, OPT_error_limit_eq, 20);
1444   ctx->e.verbose = args.hasArg(OPT_verbose);
1445 
1446   if (args.hasArg(OPT_help_hidden)) {
1447     parser.printHelp(argsArr[0], /*showHidden=*/true);
1448     return true;
1449   }
1450   if (args.hasArg(OPT_help)) {
1451     parser.printHelp(argsArr[0], /*showHidden=*/false);
1452     return true;
1453   }
1454   if (args.hasArg(OPT_version)) {
1455     message(getLLDVersion());
1456     return true;
1457   }
1458 
1459   config = std::make_unique<Configuration>();
1460   symtab = std::make_unique<SymbolTable>();
1461   config->outputType = getOutputType(args);
1462   target = createTargetInfo(args);
1463   depTracker = std::make_unique<DependencyTracker>(
1464       args.getLastArgValue(OPT_dependency_info));
1465 
1466   config->ltoo = args::getInteger(args, OPT_lto_O, 2);
1467   if (config->ltoo > 3)
1468     error("--lto-O: invalid optimization level: " + Twine(config->ltoo));
1469   unsigned ltoCgo =
1470       args::getInteger(args, OPT_lto_CGO, args::getCGOptLevel(config->ltoo));
1471   if (auto level = CodeGenOpt::getLevel(ltoCgo))
1472     config->ltoCgo = *level;
1473   else
1474     error("--lto-CGO: invalid codegen optimization level: " + Twine(ltoCgo));
1475 
1476   if (errorCount())
1477     return false;
1478 
1479   if (args.hasArg(OPT_pagezero_size)) {
1480     uint64_t pagezeroSize = args::getHex(args, OPT_pagezero_size, 0);
1481 
1482     // ld64 does something really weird. It attempts to realign the value to the
1483     // page size, but assumes the page size is 4K. This doesn't work with most
1484     // of Apple's ARM64 devices, which use a page size of 16K. This means that
1485     // it will first 4K align it by rounding down, then round up to 16K.  This
1486     // probably only happened because no one using this arg with anything other
1487     // then 0, so no one checked if it did what is what it says it does.
1488 
1489     // So we are not copying this weird behavior and doing the it in a logical
1490     // way, by always rounding down to page size.
1491     if (!isAligned(Align(target->getPageSize()), pagezeroSize)) {
1492       pagezeroSize -= pagezeroSize % target->getPageSize();
1493       warn("__PAGEZERO size is not page aligned, rounding down to 0x" +
1494            Twine::utohexstr(pagezeroSize));
1495     }
1496 
1497     target->pageZeroSize = pagezeroSize;
1498   }
1499 
1500   config->osoPrefix = args.getLastArgValue(OPT_oso_prefix);
1501   if (!config->osoPrefix.empty()) {
1502     // Expand special characters, such as ".", "..", or  "~", if present.
1503     // Note: LD64 only expands "." and not other special characters.
1504     // That seems silly to imitate so we will not try to follow it, but rather
1505     // just use real_path() to do it.
1506 
1507     // The max path length is 4096, in theory. However that seems quite long
1508     // and seems unlikely that any one would want to strip everything from the
1509     // path. Hence we've picked a reasonably large number here.
1510     SmallString<1024> expanded;
1511     if (!fs::real_path(config->osoPrefix, expanded,
1512                        /*expand_tilde=*/true)) {
1513       // Note: LD64 expands "." to be `<current_dir>/`
1514       // (ie., it has a slash suffix) whereas real_path() doesn't.
1515       // So we have to append '/' to be consistent.
1516       StringRef sep = sys::path::get_separator();
1517       // real_path removes trailing slashes as part of the normalization, but
1518       // these are meaningful for our text based stripping
1519       if (config->osoPrefix.equals(".") || config->osoPrefix.ends_with(sep))
1520         expanded += sep;
1521       config->osoPrefix = saver().save(expanded.str());
1522     }
1523   }
1524 
1525   bool pie = args.hasFlag(OPT_pie, OPT_no_pie, true);
1526   if (!supportsNoPie() && !pie) {
1527     warn("-no_pie ignored for arm64");
1528     pie = true;
1529   }
1530 
1531   config->isPic = config->outputType == MH_DYLIB ||
1532                   config->outputType == MH_BUNDLE ||
1533                   (config->outputType == MH_EXECUTE && pie);
1534 
1535   // Must be set before any InputSections and Symbols are created.
1536   config->deadStrip = args.hasArg(OPT_dead_strip);
1537 
1538   config->systemLibraryRoots = getSystemLibraryRoots(args);
1539   if (const char *path = getReproduceOption(args)) {
1540     // Note that --reproduce is a debug option so you can ignore it
1541     // if you are trying to understand the whole picture of the code.
1542     Expected<std::unique_ptr<TarWriter>> errOrWriter =
1543         TarWriter::create(path, path::stem(path));
1544     if (errOrWriter) {
1545       tar = std::move(*errOrWriter);
1546       tar->append("response.txt", createResponseFile(args));
1547       tar->append("version.txt", getLLDVersion() + "\n");
1548     } else {
1549       error("--reproduce: " + toString(errOrWriter.takeError()));
1550     }
1551   }
1552 
1553   if (auto *arg = args.getLastArg(OPT_threads_eq)) {
1554     StringRef v(arg->getValue());
1555     unsigned threads = 0;
1556     if (!llvm::to_integer(v, threads, 0) || threads == 0)
1557       error(arg->getSpelling() + ": expected a positive integer, but got '" +
1558             arg->getValue() + "'");
1559     parallel::strategy = hardware_concurrency(threads);
1560     config->thinLTOJobs = v;
1561   }
1562   if (auto *arg = args.getLastArg(OPT_thinlto_jobs_eq))
1563     config->thinLTOJobs = arg->getValue();
1564   if (!get_threadpool_strategy(config->thinLTOJobs))
1565     error("--thinlto-jobs: invalid job count: " + config->thinLTOJobs);
1566 
1567   for (const Arg *arg : args.filtered(OPT_u)) {
1568     config->explicitUndefineds.push_back(symtab->addUndefined(
1569         arg->getValue(), /*file=*/nullptr, /*isWeakRef=*/false));
1570   }
1571 
1572   for (const Arg *arg : args.filtered(OPT_U))
1573     config->explicitDynamicLookups.insert(arg->getValue());
1574 
1575   config->mapFile = args.getLastArgValue(OPT_map);
1576   config->optimize = args::getInteger(args, OPT_O, 1);
1577   config->outputFile = args.getLastArgValue(OPT_o, "a.out");
1578   config->finalOutput =
1579       args.getLastArgValue(OPT_final_output, config->outputFile);
1580   config->astPaths = args.getAllArgValues(OPT_add_ast_path);
1581   config->headerPad = args::getHex(args, OPT_headerpad, /*Default=*/32);
1582   config->headerPadMaxInstallNames =
1583       args.hasArg(OPT_headerpad_max_install_names);
1584   config->printDylibSearch =
1585       args.hasArg(OPT_print_dylib_search) || getenv("RC_TRACE_DYLIB_SEARCHING");
1586   config->printEachFile = args.hasArg(OPT_t);
1587   config->printWhyLoad = args.hasArg(OPT_why_load);
1588   config->omitDebugInfo = args.hasArg(OPT_S);
1589   config->errorForArchMismatch = args.hasArg(OPT_arch_errors_fatal);
1590   if (const Arg *arg = args.getLastArg(OPT_bundle_loader)) {
1591     if (config->outputType != MH_BUNDLE)
1592       error("-bundle_loader can only be used with MachO bundle output");
1593     addFile(arg->getValue(), LoadType::CommandLine, /*isLazy=*/false,
1594             /*isExplicit=*/false, /*isBundleLoader=*/true);
1595   }
1596   for (auto *arg : args.filtered(OPT_dyld_env)) {
1597     StringRef envPair(arg->getValue());
1598     if (!envPair.contains('='))
1599       error("-dyld_env's argument is  malformed. Expected "
1600             "-dyld_env <ENV_VAR>=<VALUE>, got `" +
1601             envPair + "`");
1602     config->dyldEnvs.push_back(envPair);
1603   }
1604   if (!config->dyldEnvs.empty() && config->outputType != MH_EXECUTE)
1605     error("-dyld_env can only be used when creating executable output");
1606 
1607   if (const Arg *arg = args.getLastArg(OPT_umbrella)) {
1608     if (config->outputType != MH_DYLIB)
1609       warn("-umbrella used, but not creating dylib");
1610     config->umbrella = arg->getValue();
1611   }
1612   config->ltoObjPath = args.getLastArgValue(OPT_object_path_lto);
1613   config->thinLTOCacheDir = args.getLastArgValue(OPT_cache_path_lto);
1614   config->thinLTOCachePolicy = getLTOCachePolicy(args);
1615   config->thinLTOEmitImportsFiles = args.hasArg(OPT_thinlto_emit_imports_files);
1616   config->thinLTOEmitIndexFiles = args.hasArg(OPT_thinlto_emit_index_files) ||
1617                                   args.hasArg(OPT_thinlto_index_only) ||
1618                                   args.hasArg(OPT_thinlto_index_only_eq);
1619   config->thinLTOIndexOnly = args.hasArg(OPT_thinlto_index_only) ||
1620                              args.hasArg(OPT_thinlto_index_only_eq);
1621   config->thinLTOIndexOnlyArg = args.getLastArgValue(OPT_thinlto_index_only_eq);
1622   config->thinLTOObjectSuffixReplace =
1623       getOldNewOptions(args, OPT_thinlto_object_suffix_replace_eq);
1624   std::tie(config->thinLTOPrefixReplaceOld, config->thinLTOPrefixReplaceNew,
1625            config->thinLTOPrefixReplaceNativeObject) =
1626       getOldNewOptionsExtra(args, OPT_thinlto_prefix_replace_eq);
1627   if (config->thinLTOEmitIndexFiles && !config->thinLTOIndexOnly) {
1628     if (args.hasArg(OPT_thinlto_object_suffix_replace_eq))
1629       error("--thinlto-object-suffix-replace is not supported with "
1630             "--thinlto-emit-index-files");
1631     else if (args.hasArg(OPT_thinlto_prefix_replace_eq))
1632       error("--thinlto-prefix-replace is not supported with "
1633             "--thinlto-emit-index-files");
1634   }
1635   if (!config->thinLTOPrefixReplaceNativeObject.empty() &&
1636       config->thinLTOIndexOnlyArg.empty()) {
1637     error("--thinlto-prefix-replace=old_dir;new_dir;obj_dir must be used with "
1638           "--thinlto-index-only=");
1639   }
1640   config->runtimePaths = args::getStrings(args, OPT_rpath);
1641   config->allLoad = args.hasFlag(OPT_all_load, OPT_noall_load, false);
1642   config->archMultiple = args.hasArg(OPT_arch_multiple);
1643   config->applicationExtension = args.hasFlag(
1644       OPT_application_extension, OPT_no_application_extension, false);
1645   config->exportDynamic = args.hasArg(OPT_export_dynamic);
1646   config->forceLoadObjC = args.hasArg(OPT_ObjC);
1647   config->forceLoadSwift = args.hasArg(OPT_force_load_swift_libs);
1648   config->deadStripDylibs = args.hasArg(OPT_dead_strip_dylibs);
1649   config->demangle = args.hasArg(OPT_demangle);
1650   config->implicitDylibs = !args.hasArg(OPT_no_implicit_dylibs);
1651   config->emitFunctionStarts =
1652       args.hasFlag(OPT_function_starts, OPT_no_function_starts, true);
1653   config->emitDataInCodeInfo =
1654       args.hasFlag(OPT_data_in_code_info, OPT_no_data_in_code_info, true);
1655   config->emitChainedFixups = shouldEmitChainedFixups(args);
1656   config->emitInitOffsets =
1657       config->emitChainedFixups || args.hasArg(OPT_init_offsets);
1658   config->icfLevel = getICFLevel(args);
1659   config->dedupStrings =
1660       args.hasFlag(OPT_deduplicate_strings, OPT_no_deduplicate_strings, true);
1661   config->deadStripDuplicates = args.hasArg(OPT_dead_strip_duplicates);
1662   config->warnDylibInstallName = args.hasFlag(
1663       OPT_warn_dylib_install_name, OPT_no_warn_dylib_install_name, false);
1664   config->ignoreOptimizationHints = args.hasArg(OPT_ignore_optimization_hints);
1665   config->callGraphProfileSort = args.hasFlag(
1666       OPT_call_graph_profile_sort, OPT_no_call_graph_profile_sort, true);
1667   config->printSymbolOrder = args.getLastArgValue(OPT_print_symbol_order_eq);
1668   config->forceExactCpuSubtypeMatch =
1669       getenv("LD_DYLIB_CPU_SUBTYPES_MUST_MATCH");
1670   config->objcStubsMode = getObjCStubsMode(args);
1671   config->ignoreAutoLink = args.hasArg(OPT_ignore_auto_link);
1672   for (const Arg *arg : args.filtered(OPT_ignore_auto_link_option))
1673     config->ignoreAutoLinkOptions.insert(arg->getValue());
1674   config->strictAutoLink = args.hasArg(OPT_strict_auto_link);
1675   config->ltoDebugPassManager = args.hasArg(OPT_lto_debug_pass_manager);
1676   config->csProfileGenerate = args.hasArg(OPT_cs_profile_generate);
1677   config->csProfilePath = args.getLastArgValue(OPT_cs_profile_path);
1678   config->pgoWarnMismatch =
1679       args.hasFlag(OPT_pgo_warn_mismatch, OPT_no_pgo_warn_mismatch, true);
1680   config->generateUuid = !args.hasArg(OPT_no_uuid);
1681 
1682   for (const Arg *arg : args.filtered(OPT_alias)) {
1683     config->aliasedSymbols.push_back(
1684         std::make_pair(arg->getValue(0), arg->getValue(1)));
1685   }
1686 
1687   if (const char *zero = getenv("ZERO_AR_DATE"))
1688     config->zeroModTime = strcmp(zero, "0") != 0;
1689   if (args.getLastArg(OPT_reproducible))
1690     config->zeroModTime = true;
1691 
1692   std::array<PlatformType, 3> encryptablePlatforms{
1693       PLATFORM_IOS, PLATFORM_WATCHOS, PLATFORM_TVOS};
1694   config->emitEncryptionInfo =
1695       args.hasFlag(OPT_encryptable, OPT_no_encryption,
1696                    is_contained(encryptablePlatforms, config->platform()));
1697 
1698   if (const Arg *arg = args.getLastArg(OPT_install_name)) {
1699     if (config->warnDylibInstallName && config->outputType != MH_DYLIB)
1700       warn(
1701           arg->getAsString(args) +
1702           ": ignored, only has effect with -dylib [--warn-dylib-install-name]");
1703     else
1704       config->installName = arg->getValue();
1705   } else if (config->outputType == MH_DYLIB) {
1706     config->installName = config->finalOutput;
1707   }
1708 
1709   if (args.hasArg(OPT_mark_dead_strippable_dylib)) {
1710     if (config->outputType != MH_DYLIB)
1711       warn("-mark_dead_strippable_dylib: ignored, only has effect with -dylib");
1712     else
1713       config->markDeadStrippableDylib = true;
1714   }
1715 
1716   if (const Arg *arg = args.getLastArg(OPT_static, OPT_dynamic))
1717     config->staticLink = (arg->getOption().getID() == OPT_static);
1718 
1719   if (const Arg *arg =
1720           args.getLastArg(OPT_flat_namespace, OPT_twolevel_namespace))
1721     config->namespaceKind = arg->getOption().getID() == OPT_twolevel_namespace
1722                                 ? NamespaceKind::twolevel
1723                                 : NamespaceKind::flat;
1724 
1725   config->undefinedSymbolTreatment = getUndefinedSymbolTreatment(args);
1726 
1727   if (config->outputType == MH_EXECUTE)
1728     config->entry = symtab->addUndefined(args.getLastArgValue(OPT_e, "_main"),
1729                                          /*file=*/nullptr,
1730                                          /*isWeakRef=*/false);
1731 
1732   config->librarySearchPaths =
1733       getLibrarySearchPaths(args, config->systemLibraryRoots);
1734   config->frameworkSearchPaths =
1735       getFrameworkSearchPaths(args, config->systemLibraryRoots);
1736   if (const Arg *arg =
1737           args.getLastArg(OPT_search_paths_first, OPT_search_dylibs_first))
1738     config->searchDylibsFirst =
1739         arg->getOption().getID() == OPT_search_dylibs_first;
1740 
1741   config->dylibCompatibilityVersion =
1742       parseDylibVersion(args, OPT_compatibility_version);
1743   config->dylibCurrentVersion = parseDylibVersion(args, OPT_current_version);
1744 
1745   config->dataConst =
1746       args.hasFlag(OPT_data_const, OPT_no_data_const, dataConstDefault(args));
1747   // Populate config->sectionRenameMap with builtin default renames.
1748   // Options -rename_section and -rename_segment are able to override.
1749   initializeSectionRenameMap();
1750   // Reject every special character except '.' and '$'
1751   // TODO(gkm): verify that this is the proper set of invalid chars
1752   StringRef invalidNameChars("!\"#%&'()*+,-/:;<=>?@[\\]^`{|}~");
1753   auto validName = [invalidNameChars](StringRef s) {
1754     if (s.find_first_of(invalidNameChars) != StringRef::npos)
1755       error("invalid name for segment or section: " + s);
1756     return s;
1757   };
1758   for (const Arg *arg : args.filtered(OPT_rename_section)) {
1759     config->sectionRenameMap[{validName(arg->getValue(0)),
1760                               validName(arg->getValue(1))}] = {
1761         validName(arg->getValue(2)), validName(arg->getValue(3))};
1762   }
1763   for (const Arg *arg : args.filtered(OPT_rename_segment)) {
1764     config->segmentRenameMap[validName(arg->getValue(0))] =
1765         validName(arg->getValue(1));
1766   }
1767 
1768   config->sectionAlignments = parseSectAlign(args);
1769 
1770   for (const Arg *arg : args.filtered(OPT_segprot)) {
1771     StringRef segName = arg->getValue(0);
1772     uint32_t maxProt = parseProtection(arg->getValue(1));
1773     uint32_t initProt = parseProtection(arg->getValue(2));
1774     if (maxProt != initProt && config->arch() != AK_i386)
1775       error("invalid argument '" + arg->getAsString(args) +
1776             "': max and init must be the same for non-i386 archs");
1777     if (segName == segment_names::linkEdit)
1778       error("-segprot cannot be used to change __LINKEDIT's protections");
1779     config->segmentProtections.push_back({segName, maxProt, initProt});
1780   }
1781 
1782   config->hasExplicitExports =
1783       args.hasArg(OPT_no_exported_symbols) ||
1784       args.hasArgNoClaim(OPT_exported_symbol, OPT_exported_symbols_list);
1785   handleSymbolPatterns(args, config->exportedSymbols, OPT_exported_symbol,
1786                        OPT_exported_symbols_list);
1787   handleSymbolPatterns(args, config->unexportedSymbols, OPT_unexported_symbol,
1788                        OPT_unexported_symbols_list);
1789   if (config->hasExplicitExports && !config->unexportedSymbols.empty())
1790     error("cannot use both -exported_symbol* and -unexported_symbol* options");
1791 
1792   if (args.hasArg(OPT_no_exported_symbols) && !config->exportedSymbols.empty())
1793     error("cannot use both -exported_symbol* and -no_exported_symbols options");
1794 
1795   // Imitating LD64's:
1796   // -non_global_symbols_no_strip_list and -non_global_symbols_strip_list can't
1797   // both be present.
1798   // But -x can be used with either of these two, in which case, the last arg
1799   // takes effect.
1800   // (TODO: This is kind of confusing - considering disallowing using them
1801   // together for a more straightforward behaviour)
1802   {
1803     bool includeLocal = false;
1804     bool excludeLocal = false;
1805     for (const Arg *arg :
1806          args.filtered(OPT_x, OPT_non_global_symbols_no_strip_list,
1807                        OPT_non_global_symbols_strip_list)) {
1808       switch (arg->getOption().getID()) {
1809       case OPT_x:
1810         config->localSymbolsPresence = SymtabPresence::None;
1811         break;
1812       case OPT_non_global_symbols_no_strip_list:
1813         if (excludeLocal) {
1814           error("cannot use both -non_global_symbols_no_strip_list and "
1815                 "-non_global_symbols_strip_list");
1816         } else {
1817           includeLocal = true;
1818           config->localSymbolsPresence = SymtabPresence::SelectivelyIncluded;
1819           parseSymbolPatternsFile(arg, config->localSymbolPatterns);
1820         }
1821         break;
1822       case OPT_non_global_symbols_strip_list:
1823         if (includeLocal) {
1824           error("cannot use both -non_global_symbols_no_strip_list and "
1825                 "-non_global_symbols_strip_list");
1826         } else {
1827           excludeLocal = true;
1828           config->localSymbolsPresence = SymtabPresence::SelectivelyExcluded;
1829           parseSymbolPatternsFile(arg, config->localSymbolPatterns);
1830         }
1831         break;
1832       default:
1833         llvm_unreachable("unexpected option");
1834       }
1835     }
1836   }
1837   // Explicitly-exported literal symbols must be defined, but might
1838   // languish in an archive if unreferenced elsewhere or if they are in the
1839   // non-global strip list. Light a fire under those lazy symbols!
1840   for (const CachedHashStringRef &cachedName : config->exportedSymbols.literals)
1841     symtab->addUndefined(cachedName.val(), /*file=*/nullptr,
1842                          /*isWeakRef=*/false);
1843 
1844   for (const Arg *arg : args.filtered(OPT_why_live))
1845     config->whyLive.insert(arg->getValue());
1846   if (!config->whyLive.empty() && !config->deadStrip)
1847     warn("-why_live has no effect without -dead_strip, ignoring");
1848 
1849   config->saveTemps = args.hasArg(OPT_save_temps);
1850 
1851   config->adhocCodesign = args.hasFlag(
1852       OPT_adhoc_codesign, OPT_no_adhoc_codesign,
1853       shouldAdhocSignByDefault(config->arch(), config->platform()));
1854 
1855   if (args.hasArg(OPT_v)) {
1856     message(getLLDVersion(), lld::errs());
1857     message(StringRef("Library search paths:") +
1858                 (config->librarySearchPaths.empty()
1859                      ? ""
1860                      : "\n\t" + join(config->librarySearchPaths, "\n\t")),
1861             lld::errs());
1862     message(StringRef("Framework search paths:") +
1863                 (config->frameworkSearchPaths.empty()
1864                      ? ""
1865                      : "\n\t" + join(config->frameworkSearchPaths, "\n\t")),
1866             lld::errs());
1867   }
1868 
1869   config->progName = argsArr[0];
1870 
1871   config->timeTraceEnabled = args.hasArg(OPT_time_trace_eq);
1872   config->timeTraceGranularity =
1873       args::getInteger(args, OPT_time_trace_granularity_eq, 500);
1874 
1875   // Initialize time trace profiler.
1876   if (config->timeTraceEnabled)
1877     timeTraceProfilerInitialize(config->timeTraceGranularity, config->progName);
1878 
1879   {
1880     TimeTraceScope timeScope("ExecuteLinker");
1881 
1882     initLLVM(); // must be run before any call to addFile()
1883     createFiles(args);
1884 
1885     // Now that all dylibs have been loaded, search for those that should be
1886     // re-exported.
1887     {
1888       auto reexportHandler = [](const Arg *arg,
1889                                 const std::vector<StringRef> &extensions) {
1890         config->hasReexports = true;
1891         StringRef searchName = arg->getValue();
1892         if (!markReexport(searchName, extensions))
1893           error(arg->getSpelling() + " " + searchName +
1894                 " does not match a supplied dylib");
1895       };
1896       std::vector<StringRef> extensions = {".tbd"};
1897       for (const Arg *arg : args.filtered(OPT_sub_umbrella))
1898         reexportHandler(arg, extensions);
1899 
1900       extensions.push_back(".dylib");
1901       for (const Arg *arg : args.filtered(OPT_sub_library))
1902         reexportHandler(arg, extensions);
1903     }
1904 
1905     cl::ResetAllOptionOccurrences();
1906 
1907     // Parse LTO options.
1908     if (const Arg *arg = args.getLastArg(OPT_mcpu))
1909       parseClangOption(saver().save("-mcpu=" + StringRef(arg->getValue())),
1910                        arg->getSpelling());
1911 
1912     for (const Arg *arg : args.filtered(OPT_mllvm)) {
1913       parseClangOption(arg->getValue(), arg->getSpelling());
1914       config->mllvmOpts.emplace_back(arg->getValue());
1915     }
1916 
1917     createSyntheticSections();
1918     createSyntheticSymbols();
1919     addSynthenticMethnames();
1920 
1921     createAliases();
1922     // If we are in "explicit exports" mode, hide everything that isn't
1923     // explicitly exported. Do this before running LTO so that LTO can better
1924     // optimize.
1925     handleExplicitExports();
1926 
1927     bool didCompileBitcodeFiles = compileBitcodeFiles();
1928 
1929     resolveLCLinkerOptions();
1930 
1931     // If --thinlto-index-only is given, we should create only "index
1932     // files" and not object files. Index file creation is already done
1933     // in compileBitcodeFiles, so we are done if that's the case.
1934     if (config->thinLTOIndexOnly)
1935       return errorCount() == 0;
1936 
1937     // LTO may emit a non-hidden (extern) object file symbol even if the
1938     // corresponding bitcode symbol is hidden. In particular, this happens for
1939     // cross-module references to hidden symbols under ThinLTO. Thus, if we
1940     // compiled any bitcode files, we must redo the symbol hiding.
1941     if (didCompileBitcodeFiles)
1942       handleExplicitExports();
1943     replaceCommonSymbols();
1944 
1945     StringRef orderFile = args.getLastArgValue(OPT_order_file);
1946     if (!orderFile.empty())
1947       priorityBuilder.parseOrderFile(orderFile);
1948 
1949     referenceStubBinder();
1950 
1951     // FIXME: should terminate the link early based on errors encountered so
1952     // far?
1953 
1954     for (const Arg *arg : args.filtered(OPT_sectcreate)) {
1955       StringRef segName = arg->getValue(0);
1956       StringRef sectName = arg->getValue(1);
1957       StringRef fileName = arg->getValue(2);
1958       std::optional<MemoryBufferRef> buffer = readFile(fileName);
1959       if (buffer)
1960         inputFiles.insert(make<OpaqueFile>(*buffer, segName, sectName));
1961     }
1962 
1963     for (const Arg *arg : args.filtered(OPT_add_empty_section)) {
1964       StringRef segName = arg->getValue(0);
1965       StringRef sectName = arg->getValue(1);
1966       inputFiles.insert(make<OpaqueFile>(MemoryBufferRef(), segName, sectName));
1967     }
1968 
1969     gatherInputSections();
1970     if (config->callGraphProfileSort)
1971       priorityBuilder.extractCallGraphProfile();
1972 
1973     if (config->deadStrip)
1974       markLive();
1975 
1976     if (args.hasArg(OPT_check_category_conflicts))
1977       objc::checkCategories();
1978 
1979     // ICF assumes that all literals have been folded already, so we must run
1980     // foldIdenticalLiterals before foldIdenticalSections.
1981     foldIdenticalLiterals();
1982     if (config->icfLevel != ICFLevel::none) {
1983       if (config->icfLevel == ICFLevel::safe)
1984         markAddrSigSymbols();
1985       foldIdenticalSections(/*onlyCfStrings=*/false);
1986     } else if (config->dedupStrings) {
1987       foldIdenticalSections(/*onlyCfStrings=*/true);
1988     }
1989 
1990     // Write to an output file.
1991     if (target->wordSize == 8)
1992       writeResult<LP64>();
1993     else
1994       writeResult<ILP32>();
1995 
1996     depTracker->write(getLLDVersion(), inputFiles, config->outputFile);
1997   }
1998 
1999   if (config->timeTraceEnabled) {
2000     checkError(timeTraceProfilerWrite(
2001         args.getLastArgValue(OPT_time_trace_eq).str(), config->outputFile));
2002 
2003     timeTraceProfilerCleanup();
2004   }
2005 
2006   if (errorCount() != 0 || config->strictAutoLink)
2007     for (const auto &warning : missingAutolinkWarnings)
2008       warn(warning);
2009 
2010   return errorCount() == 0;
2011 }
2012 } // namespace macho
2013 } // namespace lld
2014