FileManager.cpp revision efc4be242b8c8f076ad4addd3598c4853ec50ac7 
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 "clang/Basic/FileSystemOptions.h"
22#include "llvm/ADT/SmallString.h"
23#include "llvm/ADT/StringExtras.h"
24#include "llvm/Support/MemoryBuffer.h"
25#include "llvm/Support/raw_ostream.h"
26#include "llvm/System/Path.h"
27#include "llvm/Config/config.h"
28#include <map>
29#include <set>
30#include <string>
31using namespace clang;
32
33// FIXME: Enhance libsystem to support inode and other fields.
34#include <sys/stat.h>
35
36#if defined(_MSC_VER)
37#define S_ISDIR(s) (_S_IFDIR & s)
38#endif
39
40/// NON_EXISTENT_DIR - A special value distinct from null that is used to
41/// represent a dir name that doesn't exist on the disk.
42#define NON_EXISTENT_DIR reinterpret_cast<DirectoryEntry*>((intptr_t)-1)
43
44//===----------------------------------------------------------------------===//
45// Windows.
46//===----------------------------------------------------------------------===//
47
48#ifdef LLVM_ON_WIN32
49
50#define IS_DIR_SEPARATOR_CHAR(x) ((x) == '/' || (x) == '\\')
51
52namespace {
53  static std::string GetFullPath(const char *relPath) {
54    char *absPathStrPtr = _fullpath(NULL, relPath, 0);
55    assert(absPathStrPtr && "_fullpath() returned NULL!");
56
57    std::string absPath(absPathStrPtr);
58
59    free(absPathStrPtr);
60    return absPath;
61  }
62}
63
64class FileManager::UniqueDirContainer {
65  /// UniqueDirs - Cache from full path to existing directories/files.
66  ///
67  llvm::StringMap<DirectoryEntry> UniqueDirs;
68
69public:
70  DirectoryEntry &getDirectory(const char *Name, struct stat &StatBuf) {
71    std::string FullPath(GetFullPath(Name));
72    return UniqueDirs.GetOrCreateValue(
73                              FullPath.c_str(),
74                              FullPath.c_str() + FullPath.size()
75                                                                ).getValue();
76  }
77
78  size_t size() { return UniqueDirs.size(); }
79};
80
81class FileManager::UniqueFileContainer {
82  /// UniqueFiles - Cache from full path to existing directories/files.
83  ///
84  llvm::StringMap<FileEntry, llvm::BumpPtrAllocator> UniqueFiles;
85
86public:
87  FileEntry &getFile(const char *Name, struct stat &StatBuf) {
88    std::string FullPath(GetFullPath(Name));
89
90    // LowercaseString because Windows filesystem is case insensitive.
91    FullPath = llvm::LowercaseString(FullPath);
92    return UniqueFiles.GetOrCreateValue(
93                               FullPath.c_str(),
94                               FullPath.c_str() + FullPath.size()
95                                                                 ).getValue();
96  }
97
98  size_t size() { return UniqueFiles.size(); }
99};
100
101//===----------------------------------------------------------------------===//
102// Unix-like Systems.
103//===----------------------------------------------------------------------===//
104
105#else
106
107#define IS_DIR_SEPARATOR_CHAR(x) ((x) == '/')
108
109class FileManager::UniqueDirContainer {
110  /// UniqueDirs - Cache from ID's to existing directories/files.
111  ///
112  std::map<std::pair<dev_t, ino_t>, DirectoryEntry> UniqueDirs;
113
114public:
115  DirectoryEntry &getDirectory(const char *Name, struct stat &StatBuf) {
116    return UniqueDirs[std::make_pair(StatBuf.st_dev, StatBuf.st_ino)];
117  }
118
119  size_t size() { return UniqueDirs.size(); }
120};
121
122class FileManager::UniqueFileContainer {
123  /// UniqueFiles - Cache from ID's to existing directories/files.
124  ///
125  std::set<FileEntry> UniqueFiles;
126
127public:
128  FileEntry &getFile(const char *Name, struct stat &StatBuf) {
129    return
130      const_cast<FileEntry&>(
131                    *UniqueFiles.insert(FileEntry(StatBuf.st_dev,
132                                                  StatBuf.st_ino,
133                                                  StatBuf.st_mode)).first);
134  }
135
136  size_t size() { return UniqueFiles.size(); }
137};
138
139#endif
140
141//===----------------------------------------------------------------------===//
142// Common logic.
143//===----------------------------------------------------------------------===//
144
145FileManager::FileManager()
146  : UniqueDirs(*new UniqueDirContainer),
147    UniqueFiles(*new UniqueFileContainer),
148    DirEntries(64), FileEntries(64), NextFileUID(0) {
149  NumDirLookups = NumFileLookups = 0;
150  NumDirCacheMisses = NumFileCacheMisses = 0;
151}
152
153FileManager::~FileManager() {
154  delete &UniqueDirs;
155  delete &UniqueFiles;
156  for (llvm::SmallVectorImpl<FileEntry *>::iterator
157         V = VirtualFileEntries.begin(),
158         VEnd = VirtualFileEntries.end();
159       V != VEnd;
160       ++V)
161    delete *V;
162}
163
164void FileManager::addStatCache(StatSysCallCache *statCache, bool AtBeginning) {
165  assert(statCache && "No stat cache provided?");
166  if (AtBeginning || StatCache.get() == 0) {
167    statCache->setNextStatCache(StatCache.take());
168    StatCache.reset(statCache);
169    return;
170  }
171
172  StatSysCallCache *LastCache = StatCache.get();
173  while (LastCache->getNextStatCache())
174    LastCache = LastCache->getNextStatCache();
175
176  LastCache->setNextStatCache(statCache);
177}
178
179void FileManager::removeStatCache(StatSysCallCache *statCache) {
180  if (!statCache)
181    return;
182
183  if (StatCache.get() == statCache) {
184    // This is the first stat cache.
185    StatCache.reset(StatCache->takeNextStatCache());
186    return;
187  }
188
189  // Find the stat cache in the list.
190  StatSysCallCache *PrevCache = StatCache.get();
191  while (PrevCache && PrevCache->getNextStatCache() != statCache)
192    PrevCache = PrevCache->getNextStatCache();
193  if (PrevCache)
194    PrevCache->setNextStatCache(statCache->getNextStatCache());
195  else
196    assert(false && "Stat cache not found for removal");
197}
198
199/// \brief Retrieve the directory that the given file name resides in.
200static const DirectoryEntry *getDirectoryFromFile(FileManager &FileMgr,
201                                                  llvm::StringRef Filename,
202                                      const FileSystemOptions &FileSystemOpts) {
203  // Figure out what directory it is in.   If the string contains a / in it,
204  // strip off everything after it.
205  // FIXME: this logic should be in sys::Path.
206  size_t SlashPos = Filename.size();
207  while (SlashPos != 0 && !IS_DIR_SEPARATOR_CHAR(Filename[SlashPos-1]))
208    --SlashPos;
209
210  // Use the current directory if file has no path component.
211  if (SlashPos == 0)
212    return FileMgr.getDirectory(".", FileSystemOpts);
213
214  if (SlashPos == Filename.size()-1)
215    return 0;       // If filename ends with a /, it's a directory.
216
217  // Ignore repeated //'s.
218  while (SlashPos != 0 && IS_DIR_SEPARATOR_CHAR(Filename[SlashPos-1]))
219    --SlashPos;
220
221  return FileMgr.getDirectory(Filename.substr(0, SlashPos), FileSystemOpts);
222}
223
224/// getDirectory - Lookup, cache, and verify the specified directory.  This
225/// returns null if the directory doesn't exist.
226///
227const DirectoryEntry *FileManager::getDirectory(llvm::StringRef Filename,
228                                      const FileSystemOptions &FileSystemOpts) {
229  // stat doesn't like trailing separators (at least on Windows).
230  if (Filename.size() > 1 && IS_DIR_SEPARATOR_CHAR(Filename.back()))
231    Filename = Filename.substr(0, Filename.size()-1);
232
233  ++NumDirLookups;
234  llvm::StringMapEntry<DirectoryEntry *> &NamedDirEnt =
235    DirEntries.GetOrCreateValue(Filename);
236
237  // See if there is already an entry in the map.
238  if (NamedDirEnt.getValue())
239    return NamedDirEnt.getValue() == NON_EXISTENT_DIR
240              ? 0 : NamedDirEnt.getValue();
241
242  ++NumDirCacheMisses;
243
244  // By default, initialize it to invalid.
245  NamedDirEnt.setValue(NON_EXISTENT_DIR);
246
247  // Get the null-terminated directory name as stored as the key of the
248  // DirEntries map.
249  const char *InterndDirName = NamedDirEnt.getKeyData();
250
251  // Check to see if the directory exists.
252  struct stat StatBuf;
253  if (stat_cached(InterndDirName, &StatBuf, FileSystemOpts) ||   // Error stat'ing.
254      !S_ISDIR(StatBuf.st_mode))          // Not a directory?
255    return 0;
256
257  // It exists.  See if we have already opened a directory with the same inode.
258  // This occurs when one dir is symlinked to another, for example.
259  DirectoryEntry &UDE = UniqueDirs.getDirectory(InterndDirName, StatBuf);
260
261  NamedDirEnt.setValue(&UDE);
262  if (UDE.getName()) // Already have an entry with this inode, return it.
263    return &UDE;
264
265  // Otherwise, we don't have this directory yet, add it.  We use the string
266  // key from the DirEntries map as the string.
267  UDE.Name  = InterndDirName;
268  return &UDE;
269}
270
271/// NON_EXISTENT_FILE - A special value distinct from null that is used to
272/// represent a filename that doesn't exist on the disk.
273#define NON_EXISTENT_FILE reinterpret_cast<FileEntry*>((intptr_t)-1)
274
275/// getFile - Lookup, cache, and verify the specified file.  This returns null
276/// if the file doesn't exist.
277///
278const FileEntry *FileManager::getFile(llvm::StringRef Filename,
279                                      const FileSystemOptions &FileSystemOpts) {
280  ++NumFileLookups;
281
282  // See if there is already an entry in the map.
283  llvm::StringMapEntry<FileEntry *> &NamedFileEnt =
284    FileEntries.GetOrCreateValue(Filename);
285
286  // See if there is already an entry in the map.
287  if (NamedFileEnt.getValue())
288    return NamedFileEnt.getValue() == NON_EXISTENT_FILE
289                 ? 0 : NamedFileEnt.getValue();
290
291  ++NumFileCacheMisses;
292
293  // By default, initialize it to invalid.
294  NamedFileEnt.setValue(NON_EXISTENT_FILE);
295
296
297  // Get the null-terminated file name as stored as the key of the
298  // FileEntries map.
299  const char *InterndFileName = NamedFileEnt.getKeyData();
300
301  const DirectoryEntry *DirInfo
302    = getDirectoryFromFile(*this, Filename, FileSystemOpts);
303  if (DirInfo == 0)  // Directory doesn't exist, file can't exist.
304    return 0;
305
306  // FIXME: Use the directory info to prune this, before doing the stat syscall.
307  // FIXME: This will reduce the # syscalls.
308
309  // Nope, there isn't.  Check to see if the file exists.
310  struct stat StatBuf;
311  //llvm::errs() << "STATING: " << Filename;
312  if (stat_cached(InterndFileName, &StatBuf, FileSystemOpts) ||   // Error stat'ing.
313        S_ISDIR(StatBuf.st_mode)) {           // A directory?
314    // If this file doesn't exist, we leave a null in FileEntries for this path.
315    //llvm::errs() << ": Not existing\n";
316    return 0;
317  }
318  //llvm::errs() << ": exists\n";
319
320  // It exists.  See if we have already opened a file with the same inode.
321  // This occurs when one dir is symlinked to another, for example.
322  FileEntry &UFE = UniqueFiles.getFile(InterndFileName, StatBuf);
323
324  NamedFileEnt.setValue(&UFE);
325  if (UFE.getName())  // Already have an entry with this inode, return it.
326    return &UFE;
327
328  // Otherwise, we don't have this directory yet, add it.
329  // FIXME: Change the name to be a char* that points back to the 'FileEntries'
330  // key.
331  UFE.Name    = InterndFileName;
332  UFE.Size    = StatBuf.st_size;
333  UFE.ModTime = StatBuf.st_mtime;
334  UFE.Dir     = DirInfo;
335  UFE.UID     = NextFileUID++;
336  return &UFE;
337}
338
339const FileEntry *
340FileManager::getVirtualFile(llvm::StringRef Filename, off_t Size,
341                            time_t ModificationTime,
342                            const FileSystemOptions &FileSystemOpts) {
343  ++NumFileLookups;
344
345  // See if there is already an entry in the map.
346  llvm::StringMapEntry<FileEntry *> &NamedFileEnt =
347    FileEntries.GetOrCreateValue(Filename);
348
349  // See if there is already an entry in the map.
350  if (NamedFileEnt.getValue())
351    return NamedFileEnt.getValue() == NON_EXISTENT_FILE
352                 ? 0 : NamedFileEnt.getValue();
353
354  ++NumFileCacheMisses;
355
356  // By default, initialize it to invalid.
357  NamedFileEnt.setValue(NON_EXISTENT_FILE);
358
359  const DirectoryEntry *DirInfo
360    = getDirectoryFromFile(*this, Filename, FileSystemOpts);
361  if (DirInfo == 0)  // Directory doesn't exist, file can't exist.
362    return 0;
363
364  FileEntry *UFE = new FileEntry();
365  VirtualFileEntries.push_back(UFE);
366  NamedFileEnt.setValue(UFE);
367
368  UFE->Name    = NamedFileEnt.getKeyData();
369  UFE->Size    = Size;
370  UFE->ModTime = ModificationTime;
371  UFE->Dir     = DirInfo;
372  UFE->UID     = NextFileUID++;
373
374  // If this virtual file resolves to a file, also map that file to the
375  // newly-created file entry.
376  const char *InterndFileName = NamedFileEnt.getKeyData();
377  struct stat StatBuf;
378  if (!stat_cached(InterndFileName, &StatBuf, FileSystemOpts) &&
379      !S_ISDIR(StatBuf.st_mode)) {
380    llvm::sys::Path FilePath(InterndFileName);
381    FilePath.makeAbsolute();
382    FileEntries[FilePath.str()] = UFE;
383  }
384
385  return UFE;
386}
387
388llvm::MemoryBuffer *FileManager::
389getBufferForFile(const char *FilenameStart, const char *FilenameEnd,
390                 const FileSystemOptions &FileSystemOpts,
391                 std::string *ErrorStr,
392                 int64_t FileSize,
393                 struct stat *FileInfo) {
394  assert(FilenameEnd[0] == 0);
395  if (FileSystemOpts.WorkingDir.empty())
396    return llvm::MemoryBuffer::getFile(FilenameStart, ErrorStr,
397                                       FileSize, FileInfo);
398  llvm::sys::Path FilePath(llvm::StringRef(FilenameStart,
399                                           FilenameEnd-FilenameStart));
400  FixupRelativePath(FilePath, FileSystemOpts);
401
402  return llvm::MemoryBuffer::getFile(FilePath.c_str(), ErrorStr,
403                                     FileSize, FileInfo);
404}
405
406int FileManager::stat_cached(const char *path, struct stat *buf,
407                             const FileSystemOptions &FileSystemOpts) {
408  if (FileSystemOpts.WorkingDir.empty())
409    return StatCache.get() ? StatCache->stat(path, buf) : stat(path, buf);
410
411  llvm::sys::Path FilePath(path);
412  FixupRelativePath(FilePath, FileSystemOpts);
413
414  return StatCache.get() ? StatCache->stat(FilePath.c_str(), buf)
415                         : stat(FilePath.c_str(), buf);
416}
417
418void FileManager::FixupRelativePath(llvm::sys::Path &path,
419                                    const FileSystemOptions &FSOpts) {
420  if (!FSOpts.WorkingDir.empty() && !path.isAbsolute()) {
421    llvm::sys::Path NewPath(FSOpts.WorkingDir);
422    NewPath.appendComponent(path.str());
423    path = NewPath;
424  }
425}
426
427void FileManager::PrintStats() const {
428  llvm::errs() << "\n*** File Manager Stats:\n";
429  llvm::errs() << UniqueFiles.size() << " files found, "
430               << UniqueDirs.size() << " dirs found.\n";
431  llvm::errs() << NumDirLookups << " dir lookups, "
432               << NumDirCacheMisses << " dir cache misses.\n";
433  llvm::errs() << NumFileLookups << " file lookups, "
434               << NumFileCacheMisses << " file cache misses.\n";
435
436  //llvm::errs() << PagesMapped << BytesOfPagesMapped << FSLookups;
437}
438
439int MemorizeStatCalls::stat(const char *path, struct stat *buf) {
440  int result = StatSysCallCache::stat(path, buf);
441
442  // Do not cache failed stats, it is easy to construct common inconsistent
443  // situations if we do, and they are not important for PCH performance (which
444  // currently only needs the stats to construct the initial FileManager
445  // entries).
446  if (result != 0)
447    return result;
448
449  // Cache file 'stat' results and directories with absolutely paths.
450  if (!S_ISDIR(buf->st_mode) || llvm::sys::Path(path).isAbsolute())
451    StatCalls[path] = StatResult(result, *buf);
452
453  return result;
454}
455