13ca95b02SDimitry Andric //===-CachePruning.cpp - LLVM Cache Directory Pruning ---------------------===//
23ca95b02SDimitry Andric //
33ca95b02SDimitry Andric //                     The LLVM Compiler Infrastructure
43ca95b02SDimitry Andric //
53ca95b02SDimitry Andric // This file is distributed under the University of Illinois Open Source
63ca95b02SDimitry Andric // License. See LICENSE.TXT for details.
73ca95b02SDimitry Andric //
83ca95b02SDimitry Andric //===----------------------------------------------------------------------===//
93ca95b02SDimitry Andric //
103ca95b02SDimitry Andric // This file implements the pruning of a directory based on least recently used.
113ca95b02SDimitry Andric //
123ca95b02SDimitry Andric //===----------------------------------------------------------------------===//
133ca95b02SDimitry Andric 
143ca95b02SDimitry Andric #include "llvm/Support/CachePruning.h"
153ca95b02SDimitry Andric 
163ca95b02SDimitry Andric #include "llvm/Support/Debug.h"
173ca95b02SDimitry Andric #include "llvm/Support/Errc.h"
187a7e6055SDimitry Andric #include "llvm/Support/Error.h"
193ca95b02SDimitry Andric #include "llvm/Support/FileSystem.h"
203ca95b02SDimitry Andric #include "llvm/Support/Path.h"
213ca95b02SDimitry Andric #include "llvm/Support/raw_ostream.h"
223ca95b02SDimitry Andric 
233ca95b02SDimitry Andric #define DEBUG_TYPE "cache-pruning"
243ca95b02SDimitry Andric 
253ca95b02SDimitry Andric #include <set>
26d88c1a5aSDimitry Andric #include <system_error>
273ca95b02SDimitry Andric 
283ca95b02SDimitry Andric using namespace llvm;
293ca95b02SDimitry Andric 
30*b5893f02SDimitry Andric namespace {
31*b5893f02SDimitry Andric struct FileInfo {
32*b5893f02SDimitry Andric   sys::TimePoint<> Time;
33*b5893f02SDimitry Andric   uint64_t Size;
34*b5893f02SDimitry Andric   std::string Path;
35*b5893f02SDimitry Andric 
36*b5893f02SDimitry Andric   /// Used to determine which files to prune first. Also used to determine
37*b5893f02SDimitry Andric   /// set membership, so must take into account all fields.
operator <__anon904fa4980111::FileInfo38*b5893f02SDimitry Andric   bool operator<(const FileInfo &Other) const {
39*b5893f02SDimitry Andric     if (Time < Other.Time)
40*b5893f02SDimitry Andric       return true;
41*b5893f02SDimitry Andric     else if (Other.Time < Time)
42*b5893f02SDimitry Andric       return false;
43*b5893f02SDimitry Andric     if (Other.Size < Size)
44*b5893f02SDimitry Andric       return true;
45*b5893f02SDimitry Andric     else if (Size < Other.Size)
46*b5893f02SDimitry Andric       return false;
47*b5893f02SDimitry Andric     return Path < Other.Path;
48*b5893f02SDimitry Andric   }
49*b5893f02SDimitry Andric };
50*b5893f02SDimitry Andric } // anonymous namespace
51*b5893f02SDimitry Andric 
523ca95b02SDimitry Andric /// Write a new timestamp file with the given path. This is used for the pruning
533ca95b02SDimitry Andric /// interval option.
writeTimestampFile(StringRef TimestampFile)543ca95b02SDimitry Andric static void writeTimestampFile(StringRef TimestampFile) {
553ca95b02SDimitry Andric   std::error_code EC;
563ca95b02SDimitry Andric   raw_fd_ostream Out(TimestampFile.str(), EC, sys::fs::F_None);
573ca95b02SDimitry Andric }
583ca95b02SDimitry Andric 
parseDuration(StringRef Duration)597a7e6055SDimitry Andric static Expected<std::chrono::seconds> parseDuration(StringRef Duration) {
607a7e6055SDimitry Andric   if (Duration.empty())
617a7e6055SDimitry Andric     return make_error<StringError>("Duration must not be empty",
627a7e6055SDimitry Andric                                    inconvertibleErrorCode());
637a7e6055SDimitry Andric 
647a7e6055SDimitry Andric   StringRef NumStr = Duration.slice(0, Duration.size()-1);
657a7e6055SDimitry Andric   uint64_t Num;
667a7e6055SDimitry Andric   if (NumStr.getAsInteger(0, Num))
677a7e6055SDimitry Andric     return make_error<StringError>("'" + NumStr + "' not an integer",
687a7e6055SDimitry Andric                                    inconvertibleErrorCode());
697a7e6055SDimitry Andric 
707a7e6055SDimitry Andric   switch (Duration.back()) {
717a7e6055SDimitry Andric   case 's':
727a7e6055SDimitry Andric     return std::chrono::seconds(Num);
737a7e6055SDimitry Andric   case 'm':
747a7e6055SDimitry Andric     return std::chrono::minutes(Num);
757a7e6055SDimitry Andric   case 'h':
767a7e6055SDimitry Andric     return std::chrono::hours(Num);
777a7e6055SDimitry Andric   default:
787a7e6055SDimitry Andric     return make_error<StringError>("'" + Duration +
797a7e6055SDimitry Andric                                        "' must end with one of 's', 'm' or 'h'",
807a7e6055SDimitry Andric                                    inconvertibleErrorCode());
817a7e6055SDimitry Andric   }
827a7e6055SDimitry Andric }
837a7e6055SDimitry Andric 
847a7e6055SDimitry Andric Expected<CachePruningPolicy>
parseCachePruningPolicy(StringRef PolicyStr)857a7e6055SDimitry Andric llvm::parseCachePruningPolicy(StringRef PolicyStr) {
867a7e6055SDimitry Andric   CachePruningPolicy Policy;
877a7e6055SDimitry Andric   std::pair<StringRef, StringRef> P = {"", PolicyStr};
887a7e6055SDimitry Andric   while (!P.second.empty()) {
897a7e6055SDimitry Andric     P = P.second.split(':');
907a7e6055SDimitry Andric 
917a7e6055SDimitry Andric     StringRef Key, Value;
927a7e6055SDimitry Andric     std::tie(Key, Value) = P.first.split('=');
937a7e6055SDimitry Andric     if (Key == "prune_interval") {
947a7e6055SDimitry Andric       auto DurationOrErr = parseDuration(Value);
957a7e6055SDimitry Andric       if (!DurationOrErr)
967a7e6055SDimitry Andric         return DurationOrErr.takeError();
977a7e6055SDimitry Andric       Policy.Interval = *DurationOrErr;
987a7e6055SDimitry Andric     } else if (Key == "prune_after") {
997a7e6055SDimitry Andric       auto DurationOrErr = parseDuration(Value);
1007a7e6055SDimitry Andric       if (!DurationOrErr)
1017a7e6055SDimitry Andric         return DurationOrErr.takeError();
1027a7e6055SDimitry Andric       Policy.Expiration = *DurationOrErr;
1037a7e6055SDimitry Andric     } else if (Key == "cache_size") {
1047a7e6055SDimitry Andric       if (Value.back() != '%')
1057a7e6055SDimitry Andric         return make_error<StringError>("'" + Value + "' must be a percentage",
1067a7e6055SDimitry Andric                                        inconvertibleErrorCode());
107edd7eaddSDimitry Andric       StringRef SizeStr = Value.drop_back();
1087a7e6055SDimitry Andric       uint64_t Size;
1097a7e6055SDimitry Andric       if (SizeStr.getAsInteger(0, Size))
1107a7e6055SDimitry Andric         return make_error<StringError>("'" + SizeStr + "' not an integer",
1117a7e6055SDimitry Andric                                        inconvertibleErrorCode());
1127a7e6055SDimitry Andric       if (Size > 100)
1137a7e6055SDimitry Andric         return make_error<StringError>("'" + SizeStr +
1147a7e6055SDimitry Andric                                            "' must be between 0 and 100",
1157a7e6055SDimitry Andric                                        inconvertibleErrorCode());
116edd7eaddSDimitry Andric       Policy.MaxSizePercentageOfAvailableSpace = Size;
117edd7eaddSDimitry Andric     } else if (Key == "cache_size_bytes") {
118edd7eaddSDimitry Andric       uint64_t Mult = 1;
119edd7eaddSDimitry Andric       switch (tolower(Value.back())) {
120edd7eaddSDimitry Andric       case 'k':
121edd7eaddSDimitry Andric         Mult = 1024;
122edd7eaddSDimitry Andric         Value = Value.drop_back();
123edd7eaddSDimitry Andric         break;
124edd7eaddSDimitry Andric       case 'm':
125edd7eaddSDimitry Andric         Mult = 1024 * 1024;
126edd7eaddSDimitry Andric         Value = Value.drop_back();
127edd7eaddSDimitry Andric         break;
128edd7eaddSDimitry Andric       case 'g':
129edd7eaddSDimitry Andric         Mult = 1024 * 1024 * 1024;
130edd7eaddSDimitry Andric         Value = Value.drop_back();
131edd7eaddSDimitry Andric         break;
132edd7eaddSDimitry Andric       }
133edd7eaddSDimitry Andric       uint64_t Size;
134edd7eaddSDimitry Andric       if (Value.getAsInteger(0, Size))
135edd7eaddSDimitry Andric         return make_error<StringError>("'" + Value + "' not an integer",
136edd7eaddSDimitry Andric                                        inconvertibleErrorCode());
137edd7eaddSDimitry Andric       Policy.MaxSizeBytes = Size * Mult;
1382cab237bSDimitry Andric     } else if (Key == "cache_size_files") {
1392cab237bSDimitry Andric       if (Value.getAsInteger(0, Policy.MaxSizeFiles))
1402cab237bSDimitry Andric         return make_error<StringError>("'" + Value + "' not an integer",
1412cab237bSDimitry Andric                                        inconvertibleErrorCode());
1427a7e6055SDimitry Andric     } else {
1437a7e6055SDimitry Andric       return make_error<StringError>("Unknown key: '" + Key + "'",
1447a7e6055SDimitry Andric                                      inconvertibleErrorCode());
1457a7e6055SDimitry Andric     }
1467a7e6055SDimitry Andric   }
1477a7e6055SDimitry Andric 
1487a7e6055SDimitry Andric   return Policy;
1497a7e6055SDimitry Andric }
1507a7e6055SDimitry Andric 
1513ca95b02SDimitry Andric /// Prune the cache of files that haven't been accessed in a long time.
pruneCache(StringRef Path,CachePruningPolicy Policy)1527a7e6055SDimitry Andric bool llvm::pruneCache(StringRef Path, CachePruningPolicy Policy) {
153d88c1a5aSDimitry Andric   using namespace std::chrono;
154d88c1a5aSDimitry Andric 
1553ca95b02SDimitry Andric   if (Path.empty())
1563ca95b02SDimitry Andric     return false;
1573ca95b02SDimitry Andric 
1583ca95b02SDimitry Andric   bool isPathDir;
1593ca95b02SDimitry Andric   if (sys::fs::is_directory(Path, isPathDir))
1603ca95b02SDimitry Andric     return false;
1613ca95b02SDimitry Andric 
1623ca95b02SDimitry Andric   if (!isPathDir)
1633ca95b02SDimitry Andric     return false;
1643ca95b02SDimitry Andric 
165edd7eaddSDimitry Andric   Policy.MaxSizePercentageOfAvailableSpace =
166edd7eaddSDimitry Andric       std::min(Policy.MaxSizePercentageOfAvailableSpace, 100u);
1677a7e6055SDimitry Andric 
1687a7e6055SDimitry Andric   if (Policy.Expiration == seconds(0) &&
169edd7eaddSDimitry Andric       Policy.MaxSizePercentageOfAvailableSpace == 0 &&
1702cab237bSDimitry Andric       Policy.MaxSizeBytes == 0 && Policy.MaxSizeFiles == 0) {
1714ba319b5SDimitry Andric     LLVM_DEBUG(dbgs() << "No pruning settings set, exit early\n");
1723ca95b02SDimitry Andric     // Nothing will be pruned, early exit
1733ca95b02SDimitry Andric     return false;
1743ca95b02SDimitry Andric   }
1753ca95b02SDimitry Andric 
1763ca95b02SDimitry Andric   // Try to stat() the timestamp file.
1773ca95b02SDimitry Andric   SmallString<128> TimestampFile(Path);
1783ca95b02SDimitry Andric   sys::path::append(TimestampFile, "llvmcache.timestamp");
1793ca95b02SDimitry Andric   sys::fs::file_status FileStatus;
180d88c1a5aSDimitry Andric   const auto CurrentTime = system_clock::now();
1813ca95b02SDimitry Andric   if (auto EC = sys::fs::status(TimestampFile, FileStatus)) {
1823ca95b02SDimitry Andric     if (EC == errc::no_such_file_or_directory) {
1833ca95b02SDimitry Andric       // If the timestamp file wasn't there, create one now.
1843ca95b02SDimitry Andric       writeTimestampFile(TimestampFile);
1853ca95b02SDimitry Andric     } else {
1863ca95b02SDimitry Andric       // Unknown error?
1873ca95b02SDimitry Andric       return false;
1883ca95b02SDimitry Andric     }
1893ca95b02SDimitry Andric   } else {
190da09e106SDimitry Andric     if (!Policy.Interval)
191da09e106SDimitry Andric       return false;
1922cab237bSDimitry Andric     if (Policy.Interval != seconds(0)) {
1933ca95b02SDimitry Andric       // Check whether the time stamp is older than our pruning interval.
1943ca95b02SDimitry Andric       // If not, do nothing.
195d88c1a5aSDimitry Andric       const auto TimeStampModTime = FileStatus.getLastModificationTime();
1963ca95b02SDimitry Andric       auto TimeStampAge = CurrentTime - TimeStampModTime;
197da09e106SDimitry Andric       if (TimeStampAge <= *Policy.Interval) {
1984ba319b5SDimitry Andric         LLVM_DEBUG(dbgs() << "Timestamp file too recent ("
199d88c1a5aSDimitry Andric                           << duration_cast<seconds>(TimeStampAge).count()
2003ca95b02SDimitry Andric                           << "s old), do not prune.\n");
2013ca95b02SDimitry Andric         return false;
2023ca95b02SDimitry Andric       }
2033ca95b02SDimitry Andric     }
2043ca95b02SDimitry Andric     // Write a new timestamp file so that nobody else attempts to prune.
2053ca95b02SDimitry Andric     // There is a benign race condition here, if two processes happen to
2063ca95b02SDimitry Andric     // notice at the same time that the timestamp is out-of-date.
2073ca95b02SDimitry Andric     writeTimestampFile(TimestampFile);
2083ca95b02SDimitry Andric   }
2093ca95b02SDimitry Andric 
210*b5893f02SDimitry Andric   // Keep track of files to delete to get below the size limit.
211*b5893f02SDimitry Andric   // Order by time of last use so that recently used files are preserved.
212*b5893f02SDimitry Andric   std::set<FileInfo> FileInfos;
2133ca95b02SDimitry Andric   uint64_t TotalSize = 0;
2143ca95b02SDimitry Andric 
2153ca95b02SDimitry Andric   // Walk the entire directory cache, looking for unused files.
2163ca95b02SDimitry Andric   std::error_code EC;
2173ca95b02SDimitry Andric   SmallString<128> CachePathNative;
2183ca95b02SDimitry Andric   sys::path::native(Path, CachePathNative);
2193ca95b02SDimitry Andric   // Walk all of the files within this directory.
2203ca95b02SDimitry Andric   for (sys::fs::directory_iterator File(CachePathNative, EC), FileEnd;
2213ca95b02SDimitry Andric        File != FileEnd && !EC; File.increment(EC)) {
2227a7e6055SDimitry Andric     // Ignore any files not beginning with the string "llvmcache-". This
2237a7e6055SDimitry Andric     // includes the timestamp file as well as any files created by the user.
2247a7e6055SDimitry Andric     // This acts as a safeguard against data loss if the user specifies the
2257a7e6055SDimitry Andric     // wrong directory as their cache directory.
2267a7e6055SDimitry Andric     if (!sys::path::filename(File->path()).startswith("llvmcache-"))
2273ca95b02SDimitry Andric       continue;
2283ca95b02SDimitry Andric 
2293ca95b02SDimitry Andric     // Look at this file. If we can't stat it, there's nothing interesting
2303ca95b02SDimitry Andric     // there.
2312cab237bSDimitry Andric     ErrorOr<sys::fs::basic_file_status> StatusOrErr = File->status();
2322cab237bSDimitry Andric     if (!StatusOrErr) {
2334ba319b5SDimitry Andric       LLVM_DEBUG(dbgs() << "Ignore " << File->path() << " (can't stat)\n");
2343ca95b02SDimitry Andric       continue;
2353ca95b02SDimitry Andric     }
2363ca95b02SDimitry Andric 
2373ca95b02SDimitry Andric     // If the file hasn't been used recently enough, delete it
2382cab237bSDimitry Andric     const auto FileAccessTime = StatusOrErr->getLastAccessedTime();
2393ca95b02SDimitry Andric     auto FileAge = CurrentTime - FileAccessTime;
2402cab237bSDimitry Andric     if (Policy.Expiration != seconds(0) && FileAge > Policy.Expiration) {
2414ba319b5SDimitry Andric       LLVM_DEBUG(dbgs() << "Remove " << File->path() << " ("
2424ba319b5SDimitry Andric                         << duration_cast<seconds>(FileAge).count()
2434ba319b5SDimitry Andric                         << "s old)\n");
2443ca95b02SDimitry Andric       sys::fs::remove(File->path());
2453ca95b02SDimitry Andric       continue;
2463ca95b02SDimitry Andric     }
2473ca95b02SDimitry Andric 
2483ca95b02SDimitry Andric     // Leave it here for now, but add it to the list of size-based pruning.
2492cab237bSDimitry Andric     TotalSize += StatusOrErr->getSize();
250*b5893f02SDimitry Andric     FileInfos.insert({FileAccessTime, StatusOrErr->getSize(), File->path()});
2513ca95b02SDimitry Andric   }
2523ca95b02SDimitry Andric 
253*b5893f02SDimitry Andric   auto FileInfo = FileInfos.begin();
254*b5893f02SDimitry Andric   size_t NumFiles = FileInfos.size();
2552cab237bSDimitry Andric 
2562cab237bSDimitry Andric   auto RemoveCacheFile = [&]() {
2572cab237bSDimitry Andric     // Remove the file.
258*b5893f02SDimitry Andric     sys::fs::remove(FileInfo->Path);
2592cab237bSDimitry Andric     // Update size
260*b5893f02SDimitry Andric     TotalSize -= FileInfo->Size;
2612cab237bSDimitry Andric     NumFiles--;
262*b5893f02SDimitry Andric     LLVM_DEBUG(dbgs() << " - Remove " << FileInfo->Path << " (size "
263*b5893f02SDimitry Andric                       << FileInfo->Size << "), new occupancy is " << TotalSize
264*b5893f02SDimitry Andric                       << "%\n");
265*b5893f02SDimitry Andric     ++FileInfo;
2662cab237bSDimitry Andric   };
2672cab237bSDimitry Andric 
2682cab237bSDimitry Andric   // Prune for number of files.
2692cab237bSDimitry Andric   if (Policy.MaxSizeFiles)
2702cab237bSDimitry Andric     while (NumFiles > Policy.MaxSizeFiles)
2712cab237bSDimitry Andric       RemoveCacheFile();
2722cab237bSDimitry Andric 
2733ca95b02SDimitry Andric   // Prune for size now if needed
2742cab237bSDimitry Andric   if (Policy.MaxSizePercentageOfAvailableSpace > 0 || Policy.MaxSizeBytes > 0) {
2753ca95b02SDimitry Andric     auto ErrOrSpaceInfo = sys::fs::disk_space(Path);
2763ca95b02SDimitry Andric     if (!ErrOrSpaceInfo) {
2773ca95b02SDimitry Andric       report_fatal_error("Can't get available size");
2783ca95b02SDimitry Andric     }
2793ca95b02SDimitry Andric     sys::fs::space_info SpaceInfo = ErrOrSpaceInfo.get();
2803ca95b02SDimitry Andric     auto AvailableSpace = TotalSize + SpaceInfo.free;
281edd7eaddSDimitry Andric 
282edd7eaddSDimitry Andric     if (Policy.MaxSizePercentageOfAvailableSpace == 0)
283edd7eaddSDimitry Andric       Policy.MaxSizePercentageOfAvailableSpace = 100;
284edd7eaddSDimitry Andric     if (Policy.MaxSizeBytes == 0)
285edd7eaddSDimitry Andric       Policy.MaxSizeBytes = AvailableSpace;
286edd7eaddSDimitry Andric     auto TotalSizeTarget = std::min<uint64_t>(
287edd7eaddSDimitry Andric         AvailableSpace * Policy.MaxSizePercentageOfAvailableSpace / 100ull,
288edd7eaddSDimitry Andric         Policy.MaxSizeBytes);
289edd7eaddSDimitry Andric 
2904ba319b5SDimitry Andric     LLVM_DEBUG(dbgs() << "Occupancy: " << ((100 * TotalSize) / AvailableSpace)
2914ba319b5SDimitry Andric                       << "% target is: "
2924ba319b5SDimitry Andric                       << Policy.MaxSizePercentageOfAvailableSpace << "%, "
2934ba319b5SDimitry Andric                       << Policy.MaxSizeBytes << " bytes\n");
294edd7eaddSDimitry Andric 
2952cab237bSDimitry Andric     // Remove the oldest accessed files first, till we get below the threshold.
296*b5893f02SDimitry Andric     while (TotalSize > TotalSizeTarget && FileInfo != FileInfos.end())
2972cab237bSDimitry Andric       RemoveCacheFile();
2983ca95b02SDimitry Andric   }
2993ca95b02SDimitry Andric   return true;
3003ca95b02SDimitry Andric }
301