xref: /llvm-project/lldb/include/lldb/Core/Module.h (revision 24feaab8380c69d5fa3eb8c21ef2d660913fd4a9)
1 //===-- Module.h ------------------------------------------------*- C++ -*-===//
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 #ifndef LLDB_CORE_MODULE_H
10 #define LLDB_CORE_MODULE_H
11 
12 #include "lldb/Core/Address.h"
13 #include "lldb/Core/ModuleList.h"
14 #include "lldb/Core/ModuleSpec.h"
15 #include "lldb/Symbol/ObjectFile.h"
16 #include "lldb/Symbol/SymbolContextScope.h"
17 #include "lldb/Symbol/TypeSystem.h"
18 #include "lldb/Target/PathMappingList.h"
19 #include "lldb/Target/Statistics.h"
20 #include "lldb/Utility/ArchSpec.h"
21 #include "lldb/Utility/ConstString.h"
22 #include "lldb/Utility/FileSpec.h"
23 #include "lldb/Utility/Status.h"
24 #include "lldb/Utility/UUID.h"
25 #include "lldb/Utility/XcodeSDK.h"
26 #include "lldb/lldb-defines.h"
27 #include "lldb/lldb-enumerations.h"
28 #include "lldb/lldb-forward.h"
29 #include "lldb/lldb-types.h"
30 
31 #include "llvm/ADT/DenseSet.h"
32 #include "llvm/ADT/STLFunctionalExtras.h"
33 #include "llvm/ADT/StableHashing.h"
34 #include "llvm/ADT/StringRef.h"
35 #include "llvm/Support/Chrono.h"
36 
37 #include <atomic>
38 #include <cstddef>
39 #include <cstdint>
40 #include <memory>
41 #include <mutex>
42 #include <optional>
43 #include <string>
44 #include <vector>
45 
46 namespace lldb_private {
47 class CompilerDeclContext;
48 class Function;
49 class Log;
50 class ObjectFile;
51 class RegularExpression;
52 class SectionList;
53 class Stream;
54 class Symbol;
55 class SymbolContext;
56 class SymbolContextList;
57 class SymbolFile;
58 class Symtab;
59 class Target;
60 class TypeList;
61 class TypeMap;
62 class VariableList;
63 
64 /// Options used by Module::FindFunctions. This cannot be a nested class
65 /// because it must be forward-declared in ModuleList.h.
66 struct ModuleFunctionSearchOptions {
67   /// Include the symbol table.
68   bool include_symbols = false;
69   /// Include inlined functions.
70   bool include_inlines = false;
71 };
72 
73 /// \class Module Module.h "lldb/Core/Module.h"
74 /// A class that describes an executable image and its associated
75 ///        object and symbol files.
76 ///
77 /// The module is designed to be able to select a single slice of an
78 /// executable image as it would appear on disk and during program execution.
79 ///
80 /// Modules control when and if information is parsed according to which
81 /// accessors are called. For example the object file (ObjectFile)
82 /// representation will only be parsed if the object file is requested using
83 /// the Module::GetObjectFile() is called. The debug symbols will only be
84 /// parsed if the symbol file (SymbolFile) is requested using the
85 /// Module::GetSymbolFile() method.
86 ///
87 /// The module will parse more detailed information as more queries are made.
88 class Module : public std::enable_shared_from_this<Module>,
89                public SymbolContextScope {
90 public:
91   class LookupInfo;
92   // Static functions that can track the lifetime of module objects. This is
93   // handy because we might have Module objects that are in shared pointers
94   // that aren't in the global module list (from ModuleList). If this is the
95   // case we need to know about it. The modules in the global list maintained
96   // by these functions can be viewed using the "target modules list" command
97   // using the "--global" (-g for short).
98   static size_t GetNumberAllocatedModules();
99 
100   static Module *GetAllocatedModuleAtIndex(size_t idx);
101 
102   static std::recursive_mutex &GetAllocationModuleCollectionMutex();
103 
104   /// Construct with file specification and architecture.
105   ///
106   /// Clients that wish to share modules with other targets should use
107   /// ModuleList::GetSharedModule().
108   ///
109   /// \param[in] file_spec
110   ///     The file specification for the on disk representation of
111   ///     this executable image.
112   ///
113   /// \param[in] arch
114   ///     The architecture to set as the current architecture in
115   ///     this module.
116   ///
117   /// \param[in] object_name
118   ///     The name of an object in a module used to extract a module
119   ///     within a module (.a files and modules that contain multiple
120   ///     architectures).
121   ///
122   /// \param[in] object_offset
123   ///     The offset within an existing module used to extract a
124   ///     module within a module (.a files and modules that contain
125   ///     multiple architectures).
126   Module(
127       const FileSpec &file_spec, const ArchSpec &arch,
128       ConstString object_name = ConstString(), lldb::offset_t object_offset = 0,
129       const llvm::sys::TimePoint<> &object_mod_time = llvm::sys::TimePoint<>());
130 
131   Module(const ModuleSpec &module_spec);
132 
133   template <typename ObjFilePlugin, typename... Args>
134   static lldb::ModuleSP CreateModuleFromObjectFile(Args &&...args) {
135     // Must create a module and place it into a shared pointer before we can
136     // create an object file since it has a std::weak_ptr back to the module,
137     // so we need to control the creation carefully in this static function
138     lldb::ModuleSP module_sp(new Module());
139     module_sp->m_objfile_sp =
140         std::make_shared<ObjFilePlugin>(module_sp, std::forward<Args>(args)...);
141     module_sp->m_did_load_objfile.store(true, std::memory_order_relaxed);
142 
143     // Once we get the object file, set module ArchSpec to the one we get from
144     // the object file. If the object file does not have an architecture, we
145     // consider the creation a failure.
146     ArchSpec arch = module_sp->m_objfile_sp->GetArchitecture();
147     if (!arch)
148       return nullptr;
149     module_sp->m_arch = arch;
150 
151     // Also copy the object file's FileSpec.
152     module_sp->m_file = module_sp->m_objfile_sp->GetFileSpec();
153     return module_sp;
154   }
155 
156   /// Destructor.
157   ~Module() override;
158 
159   bool MatchesModuleSpec(const ModuleSpec &module_ref);
160 
161   /// Set the load address for all sections in a module to be the file address
162   /// plus \a slide.
163   ///
164   /// Many times a module will be loaded in a target with a constant offset
165   /// applied to all top level sections. This function can set the load
166   /// address for all top level sections to be the section file address +
167   /// offset.
168   ///
169   /// \param[in] target
170   ///     The target in which to apply the section load addresses.
171   ///
172   /// \param[in] value
173   ///     if \a value_is_offset is true, then value is the offset to
174   ///     apply to all file addresses for all top level sections in
175   ///     the object file as each section load address is being set.
176   ///     If \a value_is_offset is false, then "value" is the new
177   ///     absolute base address for the image.
178   ///
179   /// \param[in] value_is_offset
180   ///     If \b true, then \a value is an offset to apply to each
181   ///     file address of each top level section.
182   ///     If \b false, then \a value is the image base address that
183   ///     will be used to rigidly slide all loadable sections.
184   ///
185   /// \param[out] changed
186   ///     If any section load addresses were changed in \a target,
187   ///     then \a changed will be set to \b true. Else \a changed
188   ///     will be set to false. This allows this function to be
189   ///     called multiple times on the same module for the same
190   ///     target. If the module hasn't moved, then \a changed will
191   ///     be false and no module updated notification will need to
192   ///     be sent out.
193   ///
194   /// \return
195   ///     /b True if any sections were successfully loaded in \a target,
196   ///     /b false otherwise.
197   bool SetLoadAddress(Target &target, lldb::addr_t value, bool value_is_offset,
198                       bool &changed);
199 
200   /// \copydoc SymbolContextScope::CalculateSymbolContext(SymbolContext*)
201   ///
202   /// \see SymbolContextScope
203   void CalculateSymbolContext(SymbolContext *sc) override;
204 
205   lldb::ModuleSP CalculateSymbolContextModule() override;
206 
207   void
208   GetDescription(llvm::raw_ostream &s,
209                  lldb::DescriptionLevel level = lldb::eDescriptionLevelFull);
210 
211   /// Get the module path and object name.
212   ///
213   /// Modules can refer to object files. In this case the specification is
214   /// simple and would return the path to the file:
215   ///
216   ///     "/usr/lib/foo.dylib"
217   ///
218   /// Modules can be .o files inside of a BSD archive (.a file). In this case,
219   /// the object specification will look like:
220   ///
221   ///     "/usr/lib/foo.a(bar.o)"
222   ///
223   /// There are many places where logging wants to log this fully qualified
224   /// specification, so we centralize this functionality here.
225   ///
226   /// \return
227   ///     The object path + object name if there is one.
228   std::string GetSpecificationDescription() const;
229 
230   /// Dump a description of this object to a Stream.
231   ///
232   /// Dump a description of the contents of this object to the supplied stream
233   /// \a s. The dumped content will be only what has been loaded or parsed up
234   /// to this point at which this function is called, so this is a good way to
235   /// see what has been parsed in a module.
236   ///
237   /// \param[in] s
238   ///     The stream to which to dump the object description.
239   void Dump(Stream *s);
240 
241   /// \copydoc SymbolContextScope::DumpSymbolContext(Stream*)
242   ///
243   /// \see SymbolContextScope
244   void DumpSymbolContext(Stream *s) override;
245 
246   /// Find a symbol in the object file's symbol table.
247   ///
248   /// \param[in] name
249   ///     The name of the symbol that we are looking for.
250   ///
251   /// \param[in] symbol_type
252   ///     If set to eSymbolTypeAny, find a symbol of any type that
253   ///     has a name that matches \a name. If set to any other valid
254   ///     SymbolType enumeration value, then search only for
255   ///     symbols that match \a symbol_type.
256   ///
257   /// \return
258   ///     Returns a valid symbol pointer if a symbol was found,
259   ///     nullptr otherwise.
260   const Symbol *FindFirstSymbolWithNameAndType(
261       ConstString name, lldb::SymbolType symbol_type = lldb::eSymbolTypeAny);
262 
263   void FindSymbolsWithNameAndType(ConstString name,
264                                   lldb::SymbolType symbol_type,
265                                   SymbolContextList &sc_list);
266 
267   void FindSymbolsMatchingRegExAndType(
268       const RegularExpression &regex, lldb::SymbolType symbol_type,
269       SymbolContextList &sc_list,
270       Mangled::NamePreference mangling_preference = Mangled::ePreferDemangled);
271 
272   /// Find a function symbols in the object file's symbol table.
273   ///
274   /// \param[in] name
275   ///     The name of the symbol that we are looking for.
276   ///
277   /// \param[in] name_type_mask
278   ///     A mask that has one or more bitwise OR'ed values from the
279   ///     lldb::FunctionNameType enumeration type that indicate what
280   ///     kind of names we are looking for.
281   ///
282   /// \param[out] sc_list
283   ///     A list to append any matching symbol contexts to.
284   void FindFunctionSymbols(ConstString name, uint32_t name_type_mask,
285                            SymbolContextList &sc_list);
286 
287   /// Find compile units by partial or full path.
288   ///
289   /// Finds all compile units that match \a path in all of the modules and
290   /// returns the results in \a sc_list.
291   ///
292   /// \param[in] path
293   ///     The name of the function we are looking for.
294   ///
295   /// \param[out] sc_list
296   ///     A symbol context list that gets filled in with all of the
297   ///     matches.
298   void FindCompileUnits(const FileSpec &path, SymbolContextList &sc_list);
299 
300   /// Find functions by lookup info.
301   ///
302   /// If the function is an inlined function, it will have a block,
303   /// representing the inlined function, and the function will be the
304   /// containing function.  If it is not inlined, then the block will be NULL.
305   ///
306   /// \param[in] lookup_info
307   ///     The lookup info of the function we are looking for.
308   ///
309   /// \param[out] sc_list
310   ///     A symbol context list that gets filled in with all of the
311   ///     matches.
312   void FindFunctions(const LookupInfo &lookup_info,
313                      const CompilerDeclContext &parent_decl_ctx,
314                      const ModuleFunctionSearchOptions &options,
315                      SymbolContextList &sc_list);
316 
317   /// Find functions by name.
318   ///
319   /// If the function is an inlined function, it will have a block,
320   /// representing the inlined function, and the function will be the
321   /// containing function.  If it is not inlined, then the block will be NULL.
322   ///
323   /// \param[in] name
324   ///     The name of the function we are looking for.
325   ///
326   /// \param[in] name_type_mask
327   ///     A bit mask of bits that indicate what kind of names should
328   ///     be used when doing the lookup. Bits include fully qualified
329   ///     names, base names, C++ methods, or ObjC selectors.
330   ///     See FunctionNameType for more details.
331   ///
332   /// \param[out] sc_list
333   ///     A symbol context list that gets filled in with all of the
334   ///     matches.
335   void FindFunctions(ConstString name,
336                      const CompilerDeclContext &parent_decl_ctx,
337                      lldb::FunctionNameType name_type_mask,
338                      const ModuleFunctionSearchOptions &options,
339                      SymbolContextList &sc_list);
340 
341   /// Find functions by compiler context.
342   void FindFunctions(llvm::ArrayRef<CompilerContext> compiler_ctx,
343                      lldb::FunctionNameType name_type_mask,
344                      const ModuleFunctionSearchOptions &options,
345                      SymbolContextList &sc_list);
346 
347   /// Find functions by name.
348   ///
349   /// If the function is an inlined function, it will have a block,
350   /// representing the inlined function, and the function will be the
351   /// containing function.  If it is not inlined, then the block will be NULL.
352   ///
353   /// \param[in] regex
354   ///     A regular expression to use when matching the name.
355   ///
356   /// \param[out] sc_list
357   ///     A symbol context list that gets filled in with all of the
358   ///     matches.
359   void FindFunctions(const RegularExpression &regex,
360                      const ModuleFunctionSearchOptions &options,
361                      SymbolContextList &sc_list);
362 
363   /// Find addresses by file/line
364   ///
365   /// \param[in] target_sp
366   ///     The target the addresses are desired for.
367   ///
368   /// \param[in] file
369   ///     Source file to locate.
370   ///
371   /// \param[in] line
372   ///     Source line to locate.
373   ///
374   /// \param[in] function
375   ///	    Optional filter function. Addresses within this function will be
376   ///     added to the 'local' list. All others will be added to the 'extern'
377   ///     list.
378   ///
379   /// \param[out] output_local
380   ///     All matching addresses within 'function'
381   ///
382   /// \param[out] output_extern
383   ///     All matching addresses not within 'function'
384   void FindAddressesForLine(const lldb::TargetSP target_sp,
385                             const FileSpec &file, uint32_t line,
386                             Function *function,
387                             std::vector<Address> &output_local,
388                             std::vector<Address> &output_extern);
389 
390   /// Find global and static variables by name.
391   ///
392   /// \param[in] name
393   ///     The name of the global or static variable we are looking
394   ///     for.
395   ///
396   /// \param[in] parent_decl_ctx
397   ///     If valid, a decl context that results must exist within
398   ///
399   /// \param[in] max_matches
400   ///     Allow the number of matches to be limited to \a
401   ///     max_matches. Specify UINT32_MAX to get all possible matches.
402   ///
403   /// \param[in] variable_list
404   ///     A list of variables that gets the matches appended to.
405   ///
406   void FindGlobalVariables(ConstString name,
407                            const CompilerDeclContext &parent_decl_ctx,
408                            size_t max_matches, VariableList &variable_list);
409 
410   /// Find global and static variables by regular expression.
411   ///
412   /// \param[in] regex
413   ///     A regular expression to use when matching the name.
414   ///
415   /// \param[in] max_matches
416   ///     Allow the number of matches to be limited to \a
417   ///     max_matches. Specify UINT32_MAX to get all possible matches.
418   ///
419   /// \param[in] variable_list
420   ///     A list of variables that gets the matches appended to.
421   ///
422   void FindGlobalVariables(const RegularExpression &regex, size_t max_matches,
423                            VariableList &variable_list);
424 
425   /// Find types using a type-matching object that contains all search
426   /// parameters.
427   ///
428   /// \see lldb_private::TypeQuery
429   ///
430   /// \param[in] query
431   ///     A type matching object that contains all of the details of the type
432   ///     search.
433   ///
434   /// \param[in] results
435   ///     Any matching types will be populated into the \a results object using
436   ///     TypeMap::InsertUnique(...).
437   void FindTypes(const TypeQuery &query, TypeResults &results);
438 
439   /// Get const accessor for the module architecture.
440   ///
441   /// \return
442   ///     A const reference to the architecture object.
443   const ArchSpec &GetArchitecture() const;
444 
445   /// Get const accessor for the module file specification.
446   ///
447   /// This function returns the file for the module on the host system that is
448   /// running LLDB. This can differ from the path on the platform since we
449   /// might be doing remote debugging.
450   ///
451   /// \return
452   ///     A const reference to the file specification object.
453   const FileSpec &GetFileSpec() const { return m_file; }
454 
455   /// Get accessor for the module platform file specification.
456   ///
457   /// Platform file refers to the path of the module as it is known on the
458   /// remote system on which it is being debugged. For local debugging this is
459   /// always the same as Module::GetFileSpec(). But remote debugging might
460   /// mention a file "/usr/lib/liba.dylib" which might be locally downloaded
461   /// and cached. In this case the platform file could be something like:
462   /// "/tmp/lldb/platform-cache/remote.host.computer/usr/lib/liba.dylib" The
463   /// file could also be cached in a local developer kit directory.
464   ///
465   /// \return
466   ///     A const reference to the file specification object.
467   const FileSpec &GetPlatformFileSpec() const {
468     if (m_platform_file)
469       return m_platform_file;
470     return m_file;
471   }
472 
473   void SetPlatformFileSpec(const FileSpec &file) { m_platform_file = file; }
474 
475   const FileSpec &GetRemoteInstallFileSpec() const {
476     return m_remote_install_file;
477   }
478 
479   void SetRemoteInstallFileSpec(const FileSpec &file) {
480     m_remote_install_file = file;
481   }
482 
483   const FileSpec &GetSymbolFileFileSpec() const { return m_symfile_spec; }
484 
485   void PreloadSymbols();
486 
487   void SetSymbolFileFileSpec(const FileSpec &file);
488 
489   const llvm::sys::TimePoint<> &GetModificationTime() const {
490     return m_mod_time;
491   }
492 
493   const llvm::sys::TimePoint<> &GetObjectModificationTime() const {
494     return m_object_mod_time;
495   }
496 
497   /// This callback will be called by SymbolFile implementations when
498   /// parsing a compile unit that contains SDK information.
499   /// \param sysroot will be added to the path remapping dictionary.
500   void RegisterXcodeSDK(llvm::StringRef sdk, llvm::StringRef sysroot);
501 
502   /// Tells whether this module is capable of being the main executable for a
503   /// process.
504   ///
505   /// \return
506   ///     \b true if it is, \b false otherwise.
507   bool IsExecutable();
508 
509   /// Tells whether this module has been loaded in the target passed in. This
510   /// call doesn't distinguish between whether the module is loaded by the
511   /// dynamic loader, or by a "target module add" type call.
512   ///
513   /// \param[in] target
514   ///    The target to check whether this is loaded in.
515   ///
516   /// \return
517   ///     \b true if it is, \b false otherwise.
518   bool IsLoadedInTarget(Target *target);
519 
520   bool LoadScriptingResourceInTarget(Target *target, Status &error,
521                                      Stream &feedback_stream);
522 
523   /// Get the number of compile units for this module.
524   ///
525   /// \return
526   ///     The number of compile units that the symbol vendor plug-in
527   ///     finds.
528   size_t GetNumCompileUnits();
529 
530   lldb::CompUnitSP GetCompileUnitAtIndex(size_t idx);
531 
532   ConstString GetObjectName() const;
533 
534   uint64_t GetObjectOffset() const { return m_object_offset; }
535 
536   /// Get the object file representation for the current architecture.
537   ///
538   /// If the object file has not been located or parsed yet, this function
539   /// will find the best ObjectFile plug-in that can parse Module::m_file.
540   ///
541   /// \return
542   ///     If Module::m_file does not exist, or no plug-in was found
543   ///     that can parse the file, or the object file doesn't contain
544   ///     the current architecture in Module::m_arch, nullptr will be
545   ///     returned, else a valid object file interface will be
546   ///     returned. The returned pointer is owned by this object and
547   ///     remains valid as long as the object is around.
548   virtual ObjectFile *GetObjectFile();
549 
550   /// Get the unified section list for the module. This is the section list
551   /// created by the module's object file and any debug info and symbol files
552   /// created by the symbol vendor.
553   ///
554   /// If the symbol vendor has not been loaded yet, this function will return
555   /// the section list for the object file.
556   ///
557   /// \return
558   ///     Unified module section list.
559   virtual SectionList *GetSectionList();
560 
561   /// Notify the module that the file addresses for the Sections have been
562   /// updated.
563   ///
564   /// If the Section file addresses for a module are updated, this method
565   /// should be called.  Any parts of the module, object file, or symbol file
566   /// that has cached those file addresses must invalidate or update its
567   /// cache.
568   virtual void SectionFileAddressesChanged();
569 
570   /// Returns a reference to the UnwindTable for this Module
571   ///
572   /// The UnwindTable contains FuncUnwinders objects for any function in this
573   /// Module.  If a FuncUnwinders object hasn't been created yet (i.e. the
574   /// function has yet to be unwound in a stack walk), it will be created when
575   /// requested.  Specifically, we do not create FuncUnwinders objects for
576   /// functions until they are needed.
577   ///
578   /// \return
579   ///     Returns the unwind table for this module. If this object has no
580   ///     associated object file, an empty UnwindTable is returned.
581   UnwindTable &GetUnwindTable();
582 
583   llvm::VersionTuple GetVersion();
584 
585   /// Load an object file from memory.
586   ///
587   /// If available, the size of the object file in memory may be passed to
588   /// avoid additional round trips to process memory. If the size is not
589   /// provided, a default value is used. This value should be large enough to
590   /// enable the ObjectFile plugins to read the header of the object file
591   /// without going back to the process.
592   ///
593   /// \return
594   ///     The object file loaded from memory or nullptr, if the operation
595   ///     failed (see the `error` for more information in that case).
596   ObjectFile *GetMemoryObjectFile(const lldb::ProcessSP &process_sp,
597                                   lldb::addr_t header_addr, Status &error,
598                                   size_t size_to_read = 512);
599 
600   /// Get the module's symbol file
601   ///
602   /// If the symbol file has already been loaded, this function returns it. All
603   /// arguments are ignored. If the symbol file has not been located yet, and
604   /// the can_create argument is false, the function returns nullptr. If
605   /// can_create is true, this function will find the best SymbolFile plug-in
606   /// that can use the current object file. feedback_strm, if not null, is used
607   /// to report the details of the search process.
608   virtual SymbolFile *GetSymbolFile(bool can_create = true,
609                                     Stream *feedback_strm = nullptr);
610 
611   Symtab *GetSymtab();
612 
613   /// Get a reference to the UUID value contained in this object.
614   ///
615   /// If the executable image file doesn't not have a UUID value built into
616   /// the file format, an MD5 checksum of the entire file, or slice of the
617   /// file for the current architecture should be used.
618   ///
619   /// \return
620   ///     A const pointer to the internal copy of the UUID value in
621   ///     this module if this module has a valid UUID value, NULL
622   ///     otherwise.
623   const lldb_private::UUID &GetUUID();
624 
625   /// A debugging function that will cause everything in a module to
626   /// be parsed.
627   ///
628   /// All compile units will be parsed, along with all globals and static
629   /// variables and all functions for those compile units. All types, scopes,
630   /// local variables, static variables, global variables, and line tables
631   /// will be parsed. This can be used prior to dumping a module to see a
632   /// complete list of the resulting debug information that gets parsed, or as
633   /// a debug function to ensure that the module can consume all of the debug
634   /// data the symbol vendor provides.
635   void ParseAllDebugSymbols();
636 
637   bool ResolveFileAddress(lldb::addr_t vm_addr, Address &so_addr);
638 
639   /// Resolve the symbol context for the given address.
640   ///
641   /// Tries to resolve the matching symbol context based on a lookup from the
642   /// current symbol vendor.  If the lazy lookup fails, an attempt is made to
643   /// parse the eh_frame section to handle stripped symbols.  If this fails,
644   /// an attempt is made to resolve the symbol to the previous address to
645   /// handle the case of a function with a tail call.
646   ///
647   /// Use properties of the modified SymbolContext to inspect any resolved
648   /// target, module, compilation unit, symbol, function, function block or
649   /// line entry.  Use the return value to determine which of these properties
650   /// have been modified.
651   ///
652   /// \param[in] so_addr
653   ///     A load address to resolve.
654   ///
655   /// \param[in] resolve_scope
656   ///     The scope that should be resolved (see SymbolContext::Scope).
657   ///     A combination of flags from the enumeration SymbolContextItem
658   ///     requesting a resolution depth.  Note that the flags that are
659   ///     actually resolved may be a superset of the requested flags.
660   ///     For instance, eSymbolContextSymbol requires resolution of
661   ///     eSymbolContextModule, and eSymbolContextFunction requires
662   ///     eSymbolContextSymbol.
663   ///
664   /// \param[out] sc
665   ///     The SymbolContext that is modified based on symbol resolution.
666   ///
667   /// \param[in] resolve_tail_call_address
668   ///     Determines if so_addr should resolve to a symbol in the case
669   ///     of a function whose last instruction is a call.  In this case,
670   ///     the PC can be one past the address range of the function.
671   ///
672   /// \return
673   ///     The scope that has been resolved (see SymbolContext::Scope).
674   ///
675   /// \see SymbolContext::Scope
676   uint32_t ResolveSymbolContextForAddress(
677       const Address &so_addr, lldb::SymbolContextItem resolve_scope,
678       SymbolContext &sc, bool resolve_tail_call_address = false);
679 
680   /// Resolve items in the symbol context for a given file and line.
681   ///
682   /// Tries to resolve \a file_path and \a line to a list of matching symbol
683   /// contexts.
684   ///
685   /// The line table entries contains addresses that can be used to further
686   /// resolve the values in each match: the function, block, symbol. Care
687   /// should be taken to minimize the amount of information that is requested
688   /// to only what is needed -- typically the module, compile unit, line table
689   /// and line table entry are sufficient.
690   ///
691   /// \param[in] file_path
692   ///     A path to a source file to match. If \a file_path does not
693   ///     specify a directory, then this query will match all files
694   ///     whose base filename matches. If \a file_path does specify
695   ///     a directory, the fullpath to the file must match.
696   ///
697   /// \param[in] line
698   ///     The source line to match, or zero if just the compile unit
699   ///     should be resolved.
700   ///
701   /// \param[in] check_inlines
702   ///     Check for inline file and line number matches. This option
703   ///     should be used sparingly as it will cause all line tables
704   ///     for every compile unit to be parsed and searched for
705   ///     matching inline file entries.
706   ///
707   /// \param[in] resolve_scope
708   ///     The scope that should be resolved (see
709   ///     SymbolContext::Scope).
710   ///
711   /// \param[out] sc_list
712   ///     A symbol context list that gets matching symbols contexts
713   ///     appended to.
714   ///
715   /// \return
716   ///     The number of matches that were added to \a sc_list.
717   ///
718   /// \see SymbolContext::Scope
719   uint32_t ResolveSymbolContextForFilePath(
720       const char *file_path, uint32_t line, bool check_inlines,
721       lldb::SymbolContextItem resolve_scope, SymbolContextList &sc_list);
722 
723   /// Resolve items in the symbol context for a given file and line.
724   ///
725   /// Tries to resolve \a file_spec and \a line to a list of matching symbol
726   /// contexts.
727   ///
728   /// The line table entries contains addresses that can be used to further
729   /// resolve the values in each match: the function, block, symbol. Care
730   /// should be taken to minimize the amount of information that is requested
731   /// to only what is needed -- typically the module, compile unit, line table
732   /// and line table entry are sufficient.
733   ///
734   /// \param[in] file_spec
735   ///     A file spec to a source file to match. If \a file_path does
736   ///     not specify a directory, then this query will match all
737   ///     files whose base filename matches. If \a file_path does
738   ///     specify a directory, the fullpath to the file must match.
739   ///
740   /// \param[in] line
741   ///     The source line to match, or zero if just the compile unit
742   ///     should be resolved.
743   ///
744   /// \param[in] check_inlines
745   ///     Check for inline file and line number matches. This option
746   ///     should be used sparingly as it will cause all line tables
747   ///     for every compile unit to be parsed and searched for
748   ///     matching inline file entries.
749   ///
750   /// \param[in] resolve_scope
751   ///     The scope that should be resolved (see
752   ///     SymbolContext::Scope).
753   ///
754   /// \param[out] sc_list
755   ///     A symbol context list that gets filled in with all of the
756   ///     matches.
757   ///
758   /// \return
759   ///     A integer that contains SymbolContext::Scope bits set for
760   ///     each item that was successfully resolved.
761   ///
762   /// \see SymbolContext::Scope
763   uint32_t ResolveSymbolContextsForFileSpec(
764       const FileSpec &file_spec, uint32_t line, bool check_inlines,
765       lldb::SymbolContextItem resolve_scope, SymbolContextList &sc_list);
766 
767   void SetFileSpecAndObjectName(const FileSpec &file, ConstString object_name);
768 
769   bool GetIsDynamicLinkEditor();
770 
771   llvm::Expected<lldb::TypeSystemSP>
772   GetTypeSystemForLanguage(lldb::LanguageType language);
773 
774   /// Call \p callback for each \p TypeSystem in this \p Module.
775   /// Return true from callback to keep iterating, false to stop iterating.
776   void ForEachTypeSystem(llvm::function_ref<bool(lldb::TypeSystemSP)> callback);
777 
778   // Special error functions that can do printf style formatting that will
779   // prepend the message with something appropriate for this module (like the
780   // architecture, path and object name (if any)). This centralizes code so
781   // that everyone doesn't need to format their error and log messages on their
782   // own and keeps the output a bit more consistent.
783   template <typename... Args>
784   void LogMessage(Log *log, const char *format, Args &&...args) {
785     LogMessage(log, llvm::formatv(format, std::forward<Args>(args)...));
786   }
787 
788   template <typename... Args>
789   void LogMessageVerboseBacktrace(Log *log, const char *format,
790                                   Args &&...args) {
791     LogMessageVerboseBacktrace(
792         log, llvm::formatv(format, std::forward<Args>(args)...));
793   }
794 
795   template <typename... Args>
796   void ReportWarning(const char *format, Args &&...args) {
797     ReportWarning(llvm::formatv(format, std::forward<Args>(args)...));
798   }
799 
800   template <typename... Args>
801   void ReportError(const char *format, Args &&...args) {
802     ReportError(llvm::formatv(format, std::forward<Args>(args)...));
803   }
804 
805   // Only report an error once when the module is first detected to be modified
806   // so we don't spam the console with many messages.
807   template <typename... Args>
808   void ReportErrorIfModifyDetected(const char *format, Args &&...args) {
809     ReportErrorIfModifyDetected(
810         llvm::formatv(format, std::forward<Args>(args)...));
811   }
812 
813   void ReportWarningOptimization(std::optional<lldb::user_id_t> debugger_id);
814 
815   void
816   ReportWarningUnsupportedLanguage(lldb::LanguageType language,
817                                    std::optional<lldb::user_id_t> debugger_id);
818 
819   // Return true if the file backing this module has changed since the module
820   // was originally created  since we saved the initial file modification time
821   // when the module first gets created.
822   bool FileHasChanged() const;
823 
824   // SymbolFile and ObjectFile member objects should lock the
825   // module mutex to avoid deadlocks.
826   std::recursive_mutex &GetMutex() const { return m_mutex; }
827 
828   PathMappingList &GetSourceMappingList() { return m_source_mappings; }
829 
830   const PathMappingList &GetSourceMappingList() const {
831     return m_source_mappings;
832   }
833 
834   /// Finds a source file given a file spec using the module source path
835   /// remappings (if any).
836   ///
837   /// Tries to resolve \a orig_spec by checking the module source path
838   /// remappings. It makes sure the file exists, so this call can be expensive
839   /// if the remappings are on a network file system, so use this function
840   /// sparingly (not in a tight debug info parsing loop).
841   ///
842   /// \param[in] orig_spec
843   ///     The original source file path to try and remap.
844   ///
845   /// \param[out] new_spec
846   ///     The newly remapped filespec that is guaranteed to exist.
847   ///
848   /// \return
849   ///     /b true if \a orig_spec was successfully located and
850   ///     \a new_spec is filled in with an existing file spec,
851   ///     \b false otherwise.
852   bool FindSourceFile(const FileSpec &orig_spec, FileSpec &new_spec) const;
853 
854   /// Remaps a source file given \a path into \a new_path.
855   ///
856   /// Remaps \a path if any source remappings match. This function does NOT
857   /// stat the file system so it can be used in tight loops where debug info
858   /// is being parsed.
859   ///
860   /// \param[in] path
861   ///     The original source file path to try and remap.
862   ///
863   /// \return
864   ///     The newly remapped filespec that is may or may not exist if
865   ///     \a path was successfully located.
866   std::optional<std::string> RemapSourceFile(llvm::StringRef path) const;
867   bool RemapSourceFile(const char *, std::string &) const = delete;
868 
869   /// Update the ArchSpec to a more specific variant.
870   bool MergeArchitecture(const ArchSpec &arch_spec);
871 
872   /// Accessor for the symbol table parse time metric.
873   ///
874   /// The value is returned as a reference to allow it to be updated by the
875   /// ElapsedTime RAII object.
876   StatsDuration &GetSymtabParseTime() { return m_symtab_parse_time; }
877 
878   /// Accessor for the symbol table index time metric.
879   ///
880   /// The value is returned as a reference to allow it to be updated by the
881   /// ElapsedTime RAII object.
882   StatsDuration &GetSymtabIndexTime() { return m_symtab_index_time; }
883 
884   void ResetStatistics();
885 
886   /// \class LookupInfo Module.h "lldb/Core/Module.h"
887   /// A class that encapsulates name lookup information.
888   ///
889   /// Users can type a wide variety of partial names when setting breakpoints
890   /// by name or when looking for functions by name. The SymbolFile object is
891   /// only required to implement name lookup for function basenames and for
892   /// fully mangled names. This means if the user types in a partial name, we
893   /// must reduce this to a name lookup that will work with all SymbolFile
894   /// objects. So we might reduce a name lookup to look for a basename, and then
895   /// prune out any results that don't match.
896   ///
897   /// The "m_name" member variable represents the name as it was typed by the
898   /// user. "m_lookup_name" will be the name we actually search for through
899   /// the symbol or objects files. Lanaguage is included in case we need to
900   /// filter results by language at a later date. The "m_name_type_mask"
901   /// member variable tells us what kinds of names we are looking for and can
902   /// help us prune out unwanted results.
903   ///
904   /// Function lookups are done in Module.cpp, ModuleList.cpp and in
905   /// BreakpointResolverName.cpp and they all now use this class to do lookups
906   /// correctly.
907   class LookupInfo {
908   public:
909     LookupInfo() = default;
910 
911     LookupInfo(ConstString name, lldb::FunctionNameType name_type_mask,
912                lldb::LanguageType language);
913 
914     ConstString GetName() const { return m_name; }
915 
916     void SetName(ConstString name) { m_name = name; }
917 
918     ConstString GetLookupName() const { return m_lookup_name; }
919 
920     void SetLookupName(ConstString name) { m_lookup_name = name; }
921 
922     lldb::FunctionNameType GetNameTypeMask() const { return m_name_type_mask; }
923 
924     void SetNameTypeMask(lldb::FunctionNameType mask) {
925       m_name_type_mask = mask;
926     }
927 
928     lldb::LanguageType GetLanguageType() const { return m_language; }
929 
930     bool NameMatchesLookupInfo(
931         ConstString function_name,
932         lldb::LanguageType language_type = lldb::eLanguageTypeUnknown) const;
933 
934     void Prune(SymbolContextList &sc_list, size_t start_idx) const;
935 
936   protected:
937     /// What the user originally typed
938     ConstString m_name;
939 
940     /// The actual name will lookup when calling in the object or symbol file
941     ConstString m_lookup_name;
942 
943     /// Limit matches to only be for this language
944     lldb::LanguageType m_language = lldb::eLanguageTypeUnknown;
945 
946     /// One or more bits from lldb::FunctionNameType that indicate what kind of
947     /// names we are looking for
948     lldb::FunctionNameType m_name_type_mask = lldb::eFunctionNameTypeNone;
949 
950     ///< If \b true, then demangled names that match will need to contain
951     ///< "m_name" in order to be considered a match
952     bool m_match_name_after_lookup = false;
953   };
954 
955   /// Get a unique hash for this module.
956   ///
957   /// The hash should be enough to identify the file on disk and the
958   /// architecture of the file. If the module represents an object inside of a
959   /// file, then the hash should include the object name and object offset to
960   /// ensure a unique hash. Some examples:
961   /// - just a regular object file (mach-o, elf, coff, etc) should create a hash
962   /// - a universal mach-o file that contains to multiple architectures,
963   ///   each architecture slice should have a unique hash even though they come
964   ///   from the same file
965   /// - a .o file inside of a BSD archive. Each .o file will have an object name
966   ///   and object offset that should produce a unique hash. The object offset
967   ///   is needed as BSD archive files can contain multiple .o files that have
968   ///   the same name.
969   uint32_t Hash();
970 
971   /// Get a unique cache key for the current module.
972   ///
973   /// The cache key must be unique for a file on disk and not change if the file
974   /// is updated. This allows cache data to use this key as a prefix and as
975   /// files are modified in disk, we will overwrite the cache files. If one file
976   /// can contain multiple files, like a universal mach-o file or like a BSD
977   /// archive, the cache key must contain enough information to differentiate
978   /// these different files.
979   std::string GetCacheKey();
980 
981   /// Get the global index file cache.
982   ///
983   /// LLDB can cache data for a module between runs. This cache directory can be
984   /// used to stored data that previously was manually created each time you debug.
985   /// Examples include debug information indexes, symbol tables, symbol table
986   /// indexes, and more.
987   ///
988   /// \returns
989   ///   If caching is enabled in the lldb settings, return a pointer to the data
990   ///   file cache. If caching is not enabled, return NULL.
991   static DataFileCache *GetIndexCache();
992 protected:
993   // Member Variables
994   mutable std::recursive_mutex m_mutex; ///< A mutex to keep this object happy
995                                         /// in multi-threaded environments.
996 
997   /// The modification time for this module when it was created.
998   llvm::sys::TimePoint<> m_mod_time;
999 
1000   ArchSpec m_arch; ///< The architecture for this module.
1001   UUID m_uuid; ///< Each module is assumed to have a unique identifier to help
1002                /// match it up to debug symbols.
1003   FileSpec m_file; ///< The file representation on disk for this module (if
1004                    /// there is one).
1005   FileSpec m_platform_file; ///< The path to the module on the platform on which
1006                             /// it is being debugged
1007   FileSpec m_remote_install_file; ///< If set when debugging on remote
1008                                   /// platforms, this module will be installed
1009                                   /// at this location
1010   FileSpec m_symfile_spec;   ///< If this path is valid, then this is the file
1011                              /// that _will_ be used as the symbol file for this
1012                              /// module
1013   ConstString m_object_name; ///< The name an object within this module that is
1014                              /// selected, or empty of the module is represented
1015                              /// by \a m_file.
1016   uint64_t m_object_offset = 0;
1017   llvm::sys::TimePoint<> m_object_mod_time;
1018 
1019   /// DataBuffer containing the module image, if it was provided at
1020   /// construction time. Otherwise the data will be retrieved by mapping
1021   /// one of the FileSpec members above.
1022   lldb::DataBufferSP m_data_sp;
1023 
1024   lldb::ObjectFileSP m_objfile_sp; ///< A shared pointer to the object file
1025                                    /// parser for this module as it may or may
1026                                    /// not be shared with the SymbolFile
1027   UnwindTable m_unwind_table;      ///< Table of FuncUnwinders
1028                                    /// objects created for this
1029                                    /// Module's functions
1030   lldb::SymbolVendorUP
1031       m_symfile_up; ///< A pointer to the symbol vendor for this module.
1032   std::vector<lldb::SymbolVendorUP>
1033       m_old_symfiles; ///< If anyone calls Module::SetSymbolFileFileSpec() and
1034                       /// changes the symbol file,
1035   ///< we need to keep all old symbol files around in case anyone has type
1036   /// references to them
1037   TypeSystemMap m_type_system_map; ///< A map of any type systems associated
1038                                    /// with this module
1039   /// Module specific source remappings for when you have debug info for a
1040   /// module that doesn't match where the sources currently are.
1041   PathMappingList m_source_mappings =
1042       ModuleList::GetGlobalModuleListProperties().GetSymlinkMappings();
1043 
1044   lldb::SectionListUP m_sections_up; ///< Unified section list for module that
1045                                      /// is used by the ObjectFile and
1046                                      /// ObjectFile instances for the debug info
1047 
1048   std::atomic<bool> m_did_load_objfile{false};
1049   std::atomic<bool> m_did_load_symfile{false};
1050   std::atomic<bool> m_did_set_uuid{false};
1051   mutable bool m_file_has_changed : 1,
1052       m_first_file_changed_log : 1; /// See if the module was modified after it
1053                                     /// was initially opened.
1054   /// We store a symbol table parse time duration here because we might have
1055   /// an object file and a symbol file which both have symbol tables. The parse
1056   /// time for the symbol tables can be aggregated here.
1057   StatsDuration m_symtab_parse_time;
1058   /// We store a symbol named index time duration here because we might have
1059   /// an object file and a symbol file which both have symbol tables. The parse
1060   /// time for the symbol tables can be aggregated here.
1061   StatsDuration m_symtab_index_time;
1062 
1063   /// A set of hashes of all warnings and errors, to avoid reporting them
1064   /// multiple times to the same Debugger.
1065   llvm::DenseMap<llvm::stable_hash, std::unique_ptr<std::once_flag>>
1066       m_shown_diagnostics;
1067   std::recursive_mutex m_diagnostic_mutex;
1068 
1069   void SymbolIndicesToSymbolContextList(Symtab *symtab,
1070                                         std::vector<uint32_t> &symbol_indexes,
1071                                         SymbolContextList &sc_list);
1072 
1073   bool SetArchitecture(const ArchSpec &new_arch);
1074 
1075   void SetUUID(const lldb_private::UUID &uuid);
1076 
1077   SectionList *GetUnifiedSectionList();
1078 
1079   friend class ModuleList;
1080   friend class ObjectFile;
1081   friend class SymbolFile;
1082 
1083 private:
1084   Module(); // Only used internally by CreateJITModule ()
1085 
1086   Module(const Module &) = delete;
1087   const Module &operator=(const Module &) = delete;
1088 
1089   void LogMessage(Log *log, const llvm::formatv_object_base &payload);
1090   void LogMessageVerboseBacktrace(Log *log,
1091                                   const llvm::formatv_object_base &payload);
1092   void ReportWarning(const llvm::formatv_object_base &payload);
1093   void ReportError(const llvm::formatv_object_base &payload);
1094   void ReportErrorIfModifyDetected(const llvm::formatv_object_base &payload);
1095   std::once_flag *GetDiagnosticOnceFlag(llvm::StringRef msg);
1096 };
1097 
1098 } // namespace lldb_private
1099 
1100 #endif // LLDB_CORE_MODULE_H
1101