1 //===--- GlobalModuleIndex.h - Global Module Index --------------*- C++ -*-===//
2 //
3 //                     The LLVM Compiler Infrastructure
4 //
5 // This file is distributed under the University of Illinois Open Source
6 // License. See LICENSE.TXT for details.
7 //
8 //===----------------------------------------------------------------------===//
9 //
10 // This file defines the GlobalModuleIndex class, which manages a global index
11 // containing all of the identifiers known to the various modules within a given
12 // subdirectory of the module cache. It is used to improve the performance of
13 // queries such as "do any modules know about this identifier?"
14 //
15 //===----------------------------------------------------------------------===//
16 #ifndef LLVM_CLANG_SERIALIZATION_GLOBALMODULEINDEX_H
17 #define LLVM_CLANG_SERIALIZATION_GLOBALMODULEINDEX_H
18 
19 #include "llvm/ADT/DenseMap.h"
20 #include "llvm/ADT/SmallPtrSet.h"
21 #include "llvm/ADT/SmallVector.h"
22 #include "llvm/ADT/StringMap.h"
23 #include "llvm/ADT/StringRef.h"
24 #include <memory>
25 #include <utility>
26 
27 namespace llvm {
28 class BitstreamCursor;
29 class MemoryBuffer;
30 }
31 
32 namespace clang {
33 
34 class DirectoryEntry;
35 class FileEntry;
36 class FileManager;
37 class IdentifierIterator;
38 class PCHContainerOperations;
39 class PCHContainerReader;
40 
41 namespace serialization {
42   class ModuleFile;
43 }
44 
45 /// A global index for a set of module files, providing information about
46 /// the identifiers within those module files.
47 ///
48 /// The global index is an aid for name lookup into modules, offering a central
49 /// place where one can look for identifiers determine which
50 /// module files contain any information about that identifier. This
51 /// allows the client to restrict the search to only those module files known
52 /// to have a information about that identifier, improving performance. Moreover,
53 /// the global module index may know about module files that have not been
54 /// imported, and can be queried to determine which modules the current
55 /// translation could or should load to fix a problem.
56 class GlobalModuleIndex {
57   using ModuleFile = serialization::ModuleFile;
58 
59   /// Buffer containing the index file, which is lazily accessed so long
60   /// as the global module index is live.
61   std::unique_ptr<llvm::MemoryBuffer> Buffer;
62 
63   /// The hash table.
64   ///
65   /// This pointer actually points to a IdentifierIndexTable object,
66   /// but that type is only accessible within the implementation of
67   /// GlobalModuleIndex.
68   void *IdentifierIndex;
69 
70   /// Information about a given module file.
71   struct ModuleInfo {
ModuleInfoModuleInfo72     ModuleInfo() : File(), Size(), ModTime() { }
73 
74     /// The module file, once it has been resolved.
75     ModuleFile *File;
76 
77     /// The module file name.
78     std::string FileName;
79 
80     /// Size of the module file at the time the global index was built.
81     off_t Size;
82 
83     /// Modification time of the module file at the time the global
84     /// index was built.
85     time_t ModTime;
86 
87     /// The module IDs on which this module directly depends.
88     /// FIXME: We don't really need a vector here.
89     llvm::SmallVector<unsigned, 4> Dependencies;
90   };
91 
92   /// A mapping from module IDs to information about each module.
93   ///
94   /// This vector may have gaps, if module files have been removed or have
95   /// been updated since the index was built. A gap is indicated by an empty
96   /// file name.
97   llvm::SmallVector<ModuleInfo, 16> Modules;
98 
99   /// Lazily-populated mapping from module files to their
100   /// corresponding index into the \c Modules vector.
101   llvm::DenseMap<ModuleFile *, unsigned> ModulesByFile;
102 
103   /// The set of modules that have not yet been resolved.
104   ///
105   /// The string is just the name of the module itself, which maps to the
106   /// module ID.
107   llvm::StringMap<unsigned> UnresolvedModules;
108 
109   /// The number of identifier lookups we performed.
110   unsigned NumIdentifierLookups;
111 
112   /// The number of identifier lookup hits, where we recognize the
113   /// identifier.
114   unsigned NumIdentifierLookupHits;
115 
116   /// Internal constructor. Use \c readIndex() to read an index.
117   explicit GlobalModuleIndex(std::unique_ptr<llvm::MemoryBuffer> Buffer,
118                              llvm::BitstreamCursor Cursor);
119 
120   GlobalModuleIndex(const GlobalModuleIndex &) = delete;
121   GlobalModuleIndex &operator=(const GlobalModuleIndex &) = delete;
122 
123 public:
124   ~GlobalModuleIndex();
125 
126   /// An error code returned when trying to read an index.
127   enum ErrorCode {
128     /// No error occurred.
129     EC_None,
130     /// No index was found.
131     EC_NotFound,
132     /// Some other process is currently building the index; it is not
133     /// available yet.
134     EC_Building,
135     /// There was an unspecified I/O error reading or writing the index.
136     EC_IOError
137   };
138 
139   /// Read a global index file for the given directory.
140   ///
141   /// \param Path The path to the specific module cache where the module files
142   /// for the intended configuration reside.
143   ///
144   /// \returns A pair containing the global module index (if it exists) and
145   /// the error code.
146   static std::pair<GlobalModuleIndex *, ErrorCode>
147   readIndex(llvm::StringRef Path);
148 
149   /// Returns an iterator for identifiers stored in the index table.
150   ///
151   /// The caller accepts ownership of the returned object.
152   IdentifierIterator *createIdentifierIterator() const;
153 
154   /// Retrieve the set of modules that have up-to-date indexes.
155   ///
156   /// \param ModuleFiles Will be populated with the set of module files that
157   /// have been indexed.
158   void getKnownModules(llvm::SmallVectorImpl<ModuleFile *> &ModuleFiles);
159 
160   /// Retrieve the set of module files on which the given module file
161   /// directly depends.
162   void getModuleDependencies(ModuleFile *File,
163                              llvm::SmallVectorImpl<ModuleFile *> &Dependencies);
164 
165   /// A set of module files in which we found a result.
166   typedef llvm::SmallPtrSet<ModuleFile *, 4> HitSet;
167 
168   /// Look for all of the module files with information about the given
169   /// identifier, e.g., a global function, variable, or type with that name.
170   ///
171   /// \param Name The identifier to look for.
172   ///
173   /// \param Hits Will be populated with the set of module files that have
174   /// information about this name.
175   ///
176   /// \returns true if the identifier is known to the index, false otherwise.
177   bool lookupIdentifier(llvm::StringRef Name, HitSet &Hits);
178 
179   /// Note that the given module file has been loaded.
180   ///
181   /// \returns false if the global module index has information about this
182   /// module file, and true otherwise.
183   bool loadedModuleFile(ModuleFile *File);
184 
185   /// Print statistics to standard error.
186   void printStats();
187 
188   /// Print debugging view to standard error.
189   void dump();
190 
191   /// Write a global index into the given
192   ///
193   /// \param FileMgr The file manager to use to load module files.
194   /// \param PCHContainerRdr - The PCHContainerOperations to use for loading and
195   /// creating modules.
196   /// \param Path The path to the directory containing module files, into
197   /// which the global index will be written.
198   static ErrorCode writeIndex(FileManager &FileMgr,
199                               const PCHContainerReader &PCHContainerRdr,
200                               llvm::StringRef Path);
201 };
202 }
203 
204 #endif
205