FileManager.cpp revision 475ddb4547ce35706ce1ff54b14bdcfdc51954c2 
1///===--- FileManager.cpp - File System Probing and Caching ----------------===//
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 implements the FileManager interface.
11//
12//===----------------------------------------------------------------------===//
13//
14// TODO: This should index all interesting directories with dirent calls.
15//  getdirentries ?
16//  opendir/readdir_r/closedir ?
17//
18//===----------------------------------------------------------------------===//
19
20#include "clang/Basic/FileManager.h"
21#include "llvm/ADT/SmallString.h"
22#include "llvm/Support/raw_ostream.h"
23#include "llvm/System/Path.h"
24#include "llvm/Config/config.h"
25#include <map>
26#include <set>
27#include <string>
28using namespace clang;
29
30// FIXME: Enhance libsystem to support inode and other fields.
31#include <sys/stat.h>
32
33#if defined(_MSC_VER)
34#define S_ISDIR(s) (_S_IFDIR & s)
35#endif
36
37/// NON_EXISTENT_DIR - A special value distinct from null that is used to
38/// represent a dir name that doesn't exist on the disk.
39#define NON_EXISTENT_DIR reinterpret_cast<DirectoryEntry*>((intptr_t)-1)
40
41//===----------------------------------------------------------------------===//
42// Windows.
43//===----------------------------------------------------------------------===//
44
45#ifdef LLVM_ON_WIN32
46
47#define IS_DIR_SEPARATOR_CHAR(x) ((x) == '/' || (x) == '\\')
48
49namespace {
50  static std::string GetFullPath(const char *relPath) {
51    char *absPathStrPtr = _fullpath(NULL, relPath, 0);
52    assert(absPathStrPtr && "_fullpath() returned NULL!");
53
54    std::string absPath(absPathStrPtr);
55
56    free(absPathStrPtr);
57    return absPath;
58  }
59}
60
61class FileManager::UniqueDirContainer {
62  /// UniqueDirs - Cache from full path to existing directories/files.
63  ///
64  llvm::StringMap<DirectoryEntry> UniqueDirs;
65
66public:
67  DirectoryEntry &getDirectory(const char *Name, struct stat &StatBuf) {
68    std::string FullPath(GetFullPath(Name));
69    return UniqueDirs.GetOrCreateValue(
70                              FullPath.c_str(),
71                              FullPath.c_str() + FullPath.size()
72                                                                ).getValue();
73  }
74
75  size_t size() { return UniqueDirs.size(); }
76};
77
78class FileManager::UniqueFileContainer {
79  /// UniqueFiles - Cache from full path to existing directories/files.
80  ///
81  llvm::StringMap<FileEntry, llvm::BumpPtrAllocator> UniqueFiles;
82
83public:
84  FileEntry &getFile(const char *Name, struct stat &StatBuf) {
85    std::string FullPath(GetFullPath(Name));
86    return UniqueFiles.GetOrCreateValue(
87                               FullPath.c_str(),
88                               FullPath.c_str() + FullPath.size()
89                                                                 ).getValue();
90  }
91
92  size_t size() { return UniqueFiles.size(); }
93};
94
95//===----------------------------------------------------------------------===//
96// Unix-like Systems.
97//===----------------------------------------------------------------------===//
98
99#else
100
101#define IS_DIR_SEPARATOR_CHAR(x) ((x) == '/')
102
103class FileManager::UniqueDirContainer {
104  /// UniqueDirs - Cache from ID's to existing directories/files.
105  ///
106  std::map<std::pair<dev_t, ino_t>, DirectoryEntry> UniqueDirs;
107
108public:
109  DirectoryEntry &getDirectory(const char *Name, struct stat &StatBuf) {
110    return UniqueDirs[std::make_pair(StatBuf.st_dev, StatBuf.st_ino)];
111  }
112
113  size_t size() { return UniqueDirs.size(); }
114};
115
116class FileManager::UniqueFileContainer {
117  /// UniqueFiles - Cache from ID's to existing directories/files.
118  ///
119  std::set<FileEntry> UniqueFiles;
120
121public:
122  FileEntry &getFile(const char *Name, struct stat &StatBuf) {
123    return
124      const_cast<FileEntry&>(
125                    *UniqueFiles.insert(FileEntry(StatBuf.st_dev,
126                                                  StatBuf.st_ino,
127                                                  StatBuf.st_mode)).first);
128  }
129
130  size_t size() { return UniqueFiles.size(); }
131};
132
133#endif
134
135//===----------------------------------------------------------------------===//
136// Common logic.
137//===----------------------------------------------------------------------===//
138
139FileManager::FileManager()
140  : UniqueDirs(*new UniqueDirContainer),
141    UniqueFiles(*new UniqueFileContainer),
142    DirEntries(64), FileEntries(64), NextFileUID(0) {
143  NumDirLookups = NumFileLookups = 0;
144  NumDirCacheMisses = NumFileCacheMisses = 0;
145}
146
147FileManager::~FileManager() {
148  delete &UniqueDirs;
149  delete &UniqueFiles;
150  for (llvm::SmallVectorImpl<FileEntry *>::iterator
151         V = VirtualFileEntries.begin(),
152         VEnd = VirtualFileEntries.end();
153       V != VEnd;
154       ++V)
155    delete *V;
156}
157
158void FileManager::addStatCache(StatSysCallCache *statCache, bool AtBeginning) {
159  assert(statCache && "No stat cache provided?");
160  if (AtBeginning || StatCache.get() == 0) {
161    statCache->setNextStatCache(StatCache.take());
162    StatCache.reset(statCache);
163    return;
164  }
165
166  StatSysCallCache *LastCache = StatCache.get();
167  while (LastCache->getNextStatCache())
168    LastCache = LastCache->getNextStatCache();
169
170  LastCache->setNextStatCache(statCache);
171}
172
173void FileManager::removeStatCache(StatSysCallCache *statCache) {
174  if (!statCache)
175    return;
176
177  if (StatCache.get() == statCache) {
178    // This is the first stat cache.
179    StatCache.reset(StatCache->takeNextStatCache());
180    return;
181  }
182
183  // Find the stat cache in the list.
184  StatSysCallCache *PrevCache = StatCache.get();
185  while (PrevCache && PrevCache->getNextStatCache() != statCache)
186    PrevCache = PrevCache->getNextStatCache();
187  if (PrevCache)
188    PrevCache->setNextStatCache(statCache->getNextStatCache());
189  else
190    assert(false && "Stat cache not found for removal");
191}
192
193/// \brief Retrieve the directory that the given file name resides in.
194static const DirectoryEntry *getDirectoryFromFile(FileManager &FileMgr,
195                                                  const char *NameStart,
196                                                  const char *NameEnd) {
197  // Figure out what directory it is in.   If the string contains a / in it,
198  // strip off everything after it.
199  // FIXME: this logic should be in sys::Path.
200  const char *SlashPos = NameEnd-1;
201  while (SlashPos >= NameStart && !IS_DIR_SEPARATOR_CHAR(SlashPos[0]))
202    --SlashPos;
203  // Ignore duplicate //'s.
204  while (SlashPos > NameStart && IS_DIR_SEPARATOR_CHAR(SlashPos[-1]))
205    --SlashPos;
206
207  if (SlashPos < NameStart) {
208    // Use the current directory if file has no path component.
209    const char *Name = ".";
210    return FileMgr.getDirectory(Name, Name+1);
211  } else if (SlashPos == NameEnd-1)
212    return 0;       // If filename ends with a /, it's a directory.
213  else
214    return FileMgr.getDirectory(NameStart, SlashPos);
215}
216
217/// getDirectory - Lookup, cache, and verify the specified directory.  This
218/// returns null if the directory doesn't exist.
219///
220const DirectoryEntry *FileManager::getDirectory(const char *NameStart,
221                                                const char *NameEnd) {
222  ++NumDirLookups;
223  llvm::StringMapEntry<DirectoryEntry *> &NamedDirEnt =
224    DirEntries.GetOrCreateValue(NameStart, NameEnd);
225
226  // See if there is already an entry in the map.
227  if (NamedDirEnt.getValue())
228    return NamedDirEnt.getValue() == NON_EXISTENT_DIR
229              ? 0 : NamedDirEnt.getValue();
230
231  ++NumDirCacheMisses;
232
233  // By default, initialize it to invalid.
234  NamedDirEnt.setValue(NON_EXISTENT_DIR);
235
236  // Get the null-terminated directory name as stored as the key of the
237  // DirEntries map.
238  const char *InterndDirName = NamedDirEnt.getKeyData();
239
240  // Check to see if the directory exists.
241  struct stat StatBuf;
242  if (stat_cached(InterndDirName, &StatBuf) ||   // Error stat'ing.
243      !S_ISDIR(StatBuf.st_mode))          // Not a directory?
244    return 0;
245
246  // It exists.  See if we have already opened a directory with the same inode.
247  // This occurs when one dir is symlinked to another, for example.
248  DirectoryEntry &UDE = UniqueDirs.getDirectory(InterndDirName, StatBuf);
249
250  NamedDirEnt.setValue(&UDE);
251  if (UDE.getName()) // Already have an entry with this inode, return it.
252    return &UDE;
253
254  // Otherwise, we don't have this directory yet, add it.  We use the string
255  // key from the DirEntries map as the string.
256  UDE.Name  = InterndDirName;
257  return &UDE;
258}
259
260/// NON_EXISTENT_FILE - A special value distinct from null that is used to
261/// represent a filename that doesn't exist on the disk.
262#define NON_EXISTENT_FILE reinterpret_cast<FileEntry*>((intptr_t)-1)
263
264/// getFile - Lookup, cache, and verify the specified file.  This returns null
265/// if the file doesn't exist.
266///
267const FileEntry *FileManager::getFile(const char *NameStart,
268                                      const char *NameEnd) {
269  ++NumFileLookups;
270
271  // See if there is already an entry in the map.
272  llvm::StringMapEntry<FileEntry *> &NamedFileEnt =
273    FileEntries.GetOrCreateValue(NameStart, NameEnd);
274
275  // See if there is already an entry in the map.
276  if (NamedFileEnt.getValue())
277    return NamedFileEnt.getValue() == NON_EXISTENT_FILE
278                 ? 0 : NamedFileEnt.getValue();
279
280  ++NumFileCacheMisses;
281
282  // By default, initialize it to invalid.
283  NamedFileEnt.setValue(NON_EXISTENT_FILE);
284
285
286  // Get the null-terminated file name as stored as the key of the
287  // FileEntries map.
288  const char *InterndFileName = NamedFileEnt.getKeyData();
289
290  const DirectoryEntry *DirInfo
291    = getDirectoryFromFile(*this, NameStart, NameEnd);
292  if (DirInfo == 0)  // Directory doesn't exist, file can't exist.
293    return 0;
294
295  // FIXME: Use the directory info to prune this, before doing the stat syscall.
296  // FIXME: This will reduce the # syscalls.
297
298  // Nope, there isn't.  Check to see if the file exists.
299  struct stat StatBuf;
300  //llvm::errs() << "STATING: " << Filename;
301  if (stat_cached(InterndFileName, &StatBuf) ||   // Error stat'ing.
302        S_ISDIR(StatBuf.st_mode)) {           // A directory?
303    // If this file doesn't exist, we leave a null in FileEntries for this path.
304    //llvm::errs() << ": Not existing\n";
305    return 0;
306  }
307  //llvm::errs() << ": exists\n";
308
309  // It exists.  See if we have already opened a file with the same inode.
310  // This occurs when one dir is symlinked to another, for example.
311  FileEntry &UFE = UniqueFiles.getFile(InterndFileName, StatBuf);
312
313  NamedFileEnt.setValue(&UFE);
314  if (UFE.getName())  // Already have an entry with this inode, return it.
315    return &UFE;
316
317  // Otherwise, we don't have this directory yet, add it.
318  // FIXME: Change the name to be a char* that points back to the 'FileEntries'
319  // key.
320  UFE.Name    = InterndFileName;
321  UFE.Size    = StatBuf.st_size;
322  UFE.ModTime = StatBuf.st_mtime;
323  UFE.Dir     = DirInfo;
324  UFE.UID     = NextFileUID++;
325  return &UFE;
326}
327
328const FileEntry *
329FileManager::getVirtualFile(const llvm::StringRef &Filename,
330                            off_t Size, time_t ModificationTime) {
331  const char *NameStart = Filename.begin(), *NameEnd = Filename.end();
332
333  ++NumFileLookups;
334
335  // See if there is already an entry in the map.
336  llvm::StringMapEntry<FileEntry *> &NamedFileEnt =
337    FileEntries.GetOrCreateValue(NameStart, NameEnd);
338
339  // See if there is already an entry in the map.
340  if (NamedFileEnt.getValue())
341    return NamedFileEnt.getValue() == NON_EXISTENT_FILE
342                 ? 0 : NamedFileEnt.getValue();
343
344  ++NumFileCacheMisses;
345
346  // By default, initialize it to invalid.
347  NamedFileEnt.setValue(NON_EXISTENT_FILE);
348
349  const DirectoryEntry *DirInfo
350    = getDirectoryFromFile(*this, NameStart, NameEnd);
351  if (DirInfo == 0)  // Directory doesn't exist, file can't exist.
352    return 0;
353
354  FileEntry *UFE = new FileEntry();
355  VirtualFileEntries.push_back(UFE);
356  NamedFileEnt.setValue(UFE);
357
358  UFE->Name    = NamedFileEnt.getKeyData();
359  UFE->Size    = Size;
360  UFE->ModTime = ModificationTime;
361  UFE->Dir     = DirInfo;
362  UFE->UID     = NextFileUID++;
363  return UFE;
364}
365
366void FileManager::PrintStats() const {
367  llvm::errs() << "\n*** File Manager Stats:\n";
368  llvm::errs() << UniqueFiles.size() << " files found, "
369               << UniqueDirs.size() << " dirs found.\n";
370  llvm::errs() << NumDirLookups << " dir lookups, "
371               << NumDirCacheMisses << " dir cache misses.\n";
372  llvm::errs() << NumFileLookups << " file lookups, "
373               << NumFileCacheMisses << " file cache misses.\n";
374
375  //llvm::errs() << PagesMapped << BytesOfPagesMapped << FSLookups;
376}
377
378int MemorizeStatCalls::stat(const char *path, struct stat *buf) {
379  int result = StatSysCallCache::stat(path, buf);
380
381  // Do not cache failed stats, it is easy to construct common inconsistent
382  // situations if we do, and they are not important for PCH performance (which
383  // currently only needs the stats to construct the initial FileManager
384  // entries).
385  if (result != 0)
386    return result;
387
388  // Cache file 'stat' results and directories with absolutely paths.
389  if (!S_ISDIR(buf->st_mode) || llvm::sys::Path(path).isAbsolute())
390    StatCalls[path] = StatResult(result, *buf);
391
392  return result;
393}
394