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