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
27namespace llvm {
28class BitstreamCursor;
29class MemoryBuffer;
30}
31
32namespace clang {
33
34class DirectoryEntry;
35class FileEntry;
36class FileManager;
37class IdentifierIterator;
38class PCHContainerOperations;
39class PCHContainerReader;
40
41namespace serialization {
42  class ModuleFile;
43}
44
45using llvm::SmallVector;
46using llvm::SmallVectorImpl;
47using llvm::StringRef;
48using serialization::ModuleFile;
49
50/// \brief A global index for a set of module files, providing information about
51/// the identifiers within those module files.
52///
53/// The global index is an aid for name lookup into modules, offering a central
54/// place where one can look for identifiers determine which
55/// module files contain any information about that identifier. This
56/// allows the client to restrict the search to only those module files known
57/// to have a information about that identifier, improving performance. Moreover,
58/// the global module index may know about module files that have not been
59/// imported, and can be queried to determine which modules the current
60/// translation could or should load to fix a problem.
61class GlobalModuleIndex {
62  /// \brief Buffer containing the index file, which is lazily accessed so long
63  /// as the global module index is live.
64  std::unique_ptr<llvm::MemoryBuffer> Buffer;
65
66  /// \brief The hash table.
67  ///
68  /// This pointer actually points to a IdentifierIndexTable object,
69  /// but that type is only accessible within the implementation of
70  /// GlobalModuleIndex.
71  void *IdentifierIndex;
72
73  /// \brief Information about a given module file.
74  struct ModuleInfo {
75    ModuleInfo() : File(), Size(), ModTime() { }
76
77    /// \brief The module file, once it has been resolved.
78    ModuleFile *File;
79
80    /// \brief The module file name.
81    std::string FileName;
82
83    /// \brief Size of the module file at the time the global index was built.
84    off_t Size;
85
86    /// \brief Modification time of the module file at the time the global
87    /// index was built.
88    time_t ModTime;
89
90    /// \brief The module IDs on which this module directly depends.
91    /// FIXME: We don't really need a vector here.
92    llvm::SmallVector<unsigned, 4> Dependencies;
93  };
94
95  /// \brief A mapping from module IDs to information about each module.
96  ///
97  /// This vector may have gaps, if module files have been removed or have
98  /// been updated since the index was built. A gap is indicated by an empty
99  /// file name.
100  llvm::SmallVector<ModuleInfo, 16> Modules;
101
102  /// \brief Lazily-populated mapping from module files to their
103  /// corresponding index into the \c Modules vector.
104  llvm::DenseMap<ModuleFile *, unsigned> ModulesByFile;
105
106  /// \brief The set of modules that have not yet been resolved.
107  ///
108  /// The string is just the name of the module itself, which maps to the
109  /// module ID.
110  llvm::StringMap<unsigned> UnresolvedModules;
111
112  /// \brief The number of identifier lookups we performed.
113  unsigned NumIdentifierLookups;
114
115  /// \brief The number of identifier lookup hits, where we recognize the
116  /// identifier.
117  unsigned NumIdentifierLookupHits;
118
119  /// \brief Internal constructor. Use \c readIndex() to read an index.
120  explicit GlobalModuleIndex(std::unique_ptr<llvm::MemoryBuffer> Buffer,
121                             llvm::BitstreamCursor Cursor);
122
123  GlobalModuleIndex(const GlobalModuleIndex &) = delete;
124  GlobalModuleIndex &operator=(const GlobalModuleIndex &) = delete;
125
126public:
127  ~GlobalModuleIndex();
128
129  /// \brief An error code returned when trying to read an index.
130  enum ErrorCode {
131    /// \brief No error occurred.
132    EC_None,
133    /// \brief No index was found.
134    EC_NotFound,
135    /// \brief Some other process is currently building the index; it is not
136    /// available yet.
137    EC_Building,
138    /// \brief There was an unspecified I/O error reading or writing the index.
139    EC_IOError
140  };
141
142  /// \brief Read a global index file for the given directory.
143  ///
144  /// \param Path The path to the specific module cache where the module files
145  /// for the intended configuration reside.
146  ///
147  /// \returns A pair containing the global module index (if it exists) and
148  /// the error code.
149  static std::pair<GlobalModuleIndex *, ErrorCode>
150  readIndex(StringRef Path);
151
152  /// \brief Returns an iterator for identifiers stored in the index table.
153  ///
154  /// The caller accepts ownership of the returned object.
155  IdentifierIterator *createIdentifierIterator() const;
156
157  /// \brief Retrieve the set of modules that have up-to-date indexes.
158  ///
159  /// \param ModuleFiles Will be populated with the set of module files that
160  /// have been indexed.
161  void getKnownModules(SmallVectorImpl<ModuleFile *> &ModuleFiles);
162
163  /// \brief Retrieve the set of module files on which the given module file
164  /// directly depends.
165  void getModuleDependencies(ModuleFile *File,
166                             SmallVectorImpl<ModuleFile *> &Dependencies);
167
168  /// \brief A set of module files in which we found a result.
169  typedef llvm::SmallPtrSet<ModuleFile *, 4> HitSet;
170
171  /// \brief Look for all of the module files with information about the given
172  /// identifier, e.g., a global function, variable, or type with that name.
173  ///
174  /// \param Name The identifier to look for.
175  ///
176  /// \param Hits Will be populated with the set of module files that have
177  /// information about this name.
178  ///
179  /// \returns true if the identifier is known to the index, false otherwise.
180  bool lookupIdentifier(StringRef Name, HitSet &Hits);
181
182  /// \brief Note that the given module file has been loaded.
183  ///
184  /// \returns false if the global module index has information about this
185  /// module file, and true otherwise.
186  bool loadedModuleFile(ModuleFile *File);
187
188  /// \brief Print statistics to standard error.
189  void printStats();
190
191  /// \brief Print debugging view to standard error.
192  void dump();
193
194  /// \brief Write a global index into the given
195  ///
196  /// \param FileMgr The file manager to use to load module files.
197  /// \param PCHContainerRdr - The PCHContainerOperations to use for loading and
198  /// creating modules.
199  /// \param Path The path to the directory containing module files, into
200  /// which the global index will be written.
201  static ErrorCode writeIndex(FileManager &FileMgr,
202                              const PCHContainerReader &PCHContainerRdr,
203                              StringRef Path);
204};
205}
206
207#endif
208