127814980SMehdi Amini //===-CachePruning.cpp - LLVM Cache Directory Pruning ---------------------===//
227814980SMehdi Amini //
32946cd70SChandler Carruth // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
42946cd70SChandler Carruth // See https://llvm.org/LICENSE.txt for license information.
52946cd70SChandler Carruth // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
627814980SMehdi Amini //
727814980SMehdi Amini //===----------------------------------------------------------------------===//
827814980SMehdi Amini //
927814980SMehdi Amini // This file implements the pruning of a directory based on least recently used.
1027814980SMehdi Amini //
1127814980SMehdi Amini //===----------------------------------------------------------------------===//
1227814980SMehdi Amini 
1327814980SMehdi Amini #include "llvm/Support/CachePruning.h"
14a30f5fd9SSimon Pilgrim #include "llvm/ADT/StringRef.h"
15045d4754SMehdi Amini #include "llvm/Support/Debug.h"
16f5e36ad8SNAKAMURA Takumi #include "llvm/Support/Errc.h"
17d1eac7bcSPeter Collingbourne #include "llvm/Support/Error.h"
1827814980SMehdi Amini #include "llvm/Support/FileSystem.h"
1927814980SMehdi Amini #include "llvm/Support/Path.h"
2027814980SMehdi Amini #include "llvm/Support/raw_ostream.h"
2127814980SMehdi Amini 
22045d4754SMehdi Amini #define DEBUG_TYPE "cache-pruning"
23045d4754SMehdi Amini 
2427814980SMehdi Amini #include <set>
252ec8b150SVassil Vassilev #include <system_error>
2627814980SMehdi Amini 
2727814980SMehdi Amini using namespace llvm;
2827814980SMehdi Amini 
29481d224bSBob Haarman namespace {
30481d224bSBob Haarman struct FileInfo {
31481d224bSBob Haarman   sys::TimePoint<> Time;
32481d224bSBob Haarman   uint64_t Size;
33481d224bSBob Haarman   std::string Path;
34481d224bSBob Haarman 
35481d224bSBob Haarman   /// Used to determine which files to prune first. Also used to determine
36481d224bSBob Haarman   /// set membership, so must take into account all fields.
operator <__anon4d91df230111::FileInfo37481d224bSBob Haarman   bool operator<(const FileInfo &Other) const {
38a0f9c4f7SFangrui Song     return std::tie(Time, Other.Size, Path) <
39a0f9c4f7SFangrui Song            std::tie(Other.Time, Size, Other.Path);
40481d224bSBob Haarman   }
41481d224bSBob Haarman };
42481d224bSBob Haarman } // anonymous namespace
43481d224bSBob Haarman 
4427814980SMehdi Amini /// Write a new timestamp file with the given path. This is used for the pruning
4527814980SMehdi Amini /// interval option.
writeTimestampFile(StringRef TimestampFile)4627814980SMehdi Amini static void writeTimestampFile(StringRef TimestampFile) {
4727814980SMehdi Amini   std::error_code EC;
48d9b948b6SFangrui Song   raw_fd_ostream Out(TimestampFile.str(), EC, sys::fs::OF_None);
4927814980SMehdi Amini }
5027814980SMehdi Amini 
parseDuration(StringRef Duration)51d1eac7bcSPeter Collingbourne static Expected<std::chrono::seconds> parseDuration(StringRef Duration) {
52d1eac7bcSPeter Collingbourne   if (Duration.empty())
53d1eac7bcSPeter Collingbourne     return make_error<StringError>("Duration must not be empty",
54d1eac7bcSPeter Collingbourne                                    inconvertibleErrorCode());
55d1eac7bcSPeter Collingbourne 
56d1eac7bcSPeter Collingbourne   StringRef NumStr = Duration.slice(0, Duration.size()-1);
57d1eac7bcSPeter Collingbourne   uint64_t Num;
58d1eac7bcSPeter Collingbourne   if (NumStr.getAsInteger(0, Num))
59d1eac7bcSPeter Collingbourne     return make_error<StringError>("'" + NumStr + "' not an integer",
60d1eac7bcSPeter Collingbourne                                    inconvertibleErrorCode());
61d1eac7bcSPeter Collingbourne 
62d1eac7bcSPeter Collingbourne   switch (Duration.back()) {
63d1eac7bcSPeter Collingbourne   case 's':
64d1eac7bcSPeter Collingbourne     return std::chrono::seconds(Num);
65d1eac7bcSPeter Collingbourne   case 'm':
66d1eac7bcSPeter Collingbourne     return std::chrono::minutes(Num);
67d1eac7bcSPeter Collingbourne   case 'h':
68d1eac7bcSPeter Collingbourne     return std::chrono::hours(Num);
69d1eac7bcSPeter Collingbourne   default:
70d1eac7bcSPeter Collingbourne     return make_error<StringError>("'" + Duration +
71d1eac7bcSPeter Collingbourne                                        "' must end with one of 's', 'm' or 'h'",
72d1eac7bcSPeter Collingbourne                                    inconvertibleErrorCode());
73d1eac7bcSPeter Collingbourne   }
74d1eac7bcSPeter Collingbourne }
75d1eac7bcSPeter Collingbourne 
76d1eac7bcSPeter Collingbourne Expected<CachePruningPolicy>
parseCachePruningPolicy(StringRef PolicyStr)77d1eac7bcSPeter Collingbourne llvm::parseCachePruningPolicy(StringRef PolicyStr) {
78d1eac7bcSPeter Collingbourne   CachePruningPolicy Policy;
79d1eac7bcSPeter Collingbourne   std::pair<StringRef, StringRef> P = {"", PolicyStr};
80d1eac7bcSPeter Collingbourne   while (!P.second.empty()) {
81d1eac7bcSPeter Collingbourne     P = P.second.split(':');
82d1eac7bcSPeter Collingbourne 
83d1eac7bcSPeter Collingbourne     StringRef Key, Value;
84d1eac7bcSPeter Collingbourne     std::tie(Key, Value) = P.first.split('=');
85d1eac7bcSPeter Collingbourne     if (Key == "prune_interval") {
86d1eac7bcSPeter Collingbourne       auto DurationOrErr = parseDuration(Value);
87d1eac7bcSPeter Collingbourne       if (!DurationOrErr)
88255c6e1bSPeter Collingbourne         return DurationOrErr.takeError();
89d1eac7bcSPeter Collingbourne       Policy.Interval = *DurationOrErr;
90d1eac7bcSPeter Collingbourne     } else if (Key == "prune_after") {
91d1eac7bcSPeter Collingbourne       auto DurationOrErr = parseDuration(Value);
92d1eac7bcSPeter Collingbourne       if (!DurationOrErr)
93255c6e1bSPeter Collingbourne         return DurationOrErr.takeError();
94d1eac7bcSPeter Collingbourne       Policy.Expiration = *DurationOrErr;
95d1eac7bcSPeter Collingbourne     } else if (Key == "cache_size") {
96d1eac7bcSPeter Collingbourne       if (Value.back() != '%')
97d1eac7bcSPeter Collingbourne         return make_error<StringError>("'" + Value + "' must be a percentage",
98d1eac7bcSPeter Collingbourne                                        inconvertibleErrorCode());
998d292233SPeter Collingbourne       StringRef SizeStr = Value.drop_back();
100d1eac7bcSPeter Collingbourne       uint64_t Size;
101d1eac7bcSPeter Collingbourne       if (SizeStr.getAsInteger(0, Size))
102d1eac7bcSPeter Collingbourne         return make_error<StringError>("'" + SizeStr + "' not an integer",
103d1eac7bcSPeter Collingbourne                                        inconvertibleErrorCode());
104d1eac7bcSPeter Collingbourne       if (Size > 100)
105d1eac7bcSPeter Collingbourne         return make_error<StringError>("'" + SizeStr +
106d1eac7bcSPeter Collingbourne                                            "' must be between 0 and 100",
107d1eac7bcSPeter Collingbourne                                        inconvertibleErrorCode());
1088d292233SPeter Collingbourne       Policy.MaxSizePercentageOfAvailableSpace = Size;
1098d292233SPeter Collingbourne     } else if (Key == "cache_size_bytes") {
1108d292233SPeter Collingbourne       uint64_t Mult = 1;
11130aaa2f3SPeter Collingbourne       switch (tolower(Value.back())) {
1128d292233SPeter Collingbourne       case 'k':
1138d292233SPeter Collingbourne         Mult = 1024;
1148d292233SPeter Collingbourne         Value = Value.drop_back();
1158d292233SPeter Collingbourne         break;
1168d292233SPeter Collingbourne       case 'm':
1178d292233SPeter Collingbourne         Mult = 1024 * 1024;
1188d292233SPeter Collingbourne         Value = Value.drop_back();
1198d292233SPeter Collingbourne         break;
1208d292233SPeter Collingbourne       case 'g':
1218d292233SPeter Collingbourne         Mult = 1024 * 1024 * 1024;
1228d292233SPeter Collingbourne         Value = Value.drop_back();
1238d292233SPeter Collingbourne         break;
1248d292233SPeter Collingbourne       }
1258d292233SPeter Collingbourne       uint64_t Size;
1268d292233SPeter Collingbourne       if (Value.getAsInteger(0, Size))
1278d292233SPeter Collingbourne         return make_error<StringError>("'" + Value + "' not an integer",
1288d292233SPeter Collingbourne                                        inconvertibleErrorCode());
1298d292233SPeter Collingbourne       Policy.MaxSizeBytes = Size * Mult;
130048ac839SPeter Collingbourne     } else if (Key == "cache_size_files") {
131048ac839SPeter Collingbourne       if (Value.getAsInteger(0, Policy.MaxSizeFiles))
132048ac839SPeter Collingbourne         return make_error<StringError>("'" + Value + "' not an integer",
133048ac839SPeter Collingbourne                                        inconvertibleErrorCode());
134d1eac7bcSPeter Collingbourne     } else {
135d1eac7bcSPeter Collingbourne       return make_error<StringError>("Unknown key: '" + Key + "'",
136d1eac7bcSPeter Collingbourne                                      inconvertibleErrorCode());
137d1eac7bcSPeter Collingbourne     }
138d1eac7bcSPeter Collingbourne   }
139d1eac7bcSPeter Collingbourne 
140d1eac7bcSPeter Collingbourne   return Policy;
141d1eac7bcSPeter Collingbourne }
142d1eac7bcSPeter Collingbourne 
14327814980SMehdi Amini /// Prune the cache of files that haven't been accessed in a long time.
pruneCache(StringRef Path,CachePruningPolicy Policy)144cead56fbSPeter Collingbourne bool llvm::pruneCache(StringRef Path, CachePruningPolicy Policy) {
145757ca886SPavel Labath   using namespace std::chrono;
146757ca886SPavel Labath 
147721800d4SMehdi Amini   if (Path.empty())
148721800d4SMehdi Amini     return false;
149721800d4SMehdi Amini 
150721800d4SMehdi Amini   bool isPathDir;
151721800d4SMehdi Amini   if (sys::fs::is_directory(Path, isPathDir))
152721800d4SMehdi Amini     return false;
153721800d4SMehdi Amini 
154721800d4SMehdi Amini   if (!isPathDir)
155721800d4SMehdi Amini     return false;
15627814980SMehdi Amini 
1578d292233SPeter Collingbourne   Policy.MaxSizePercentageOfAvailableSpace =
1588d292233SPeter Collingbourne       std::min(Policy.MaxSizePercentageOfAvailableSpace, 100u);
159cead56fbSPeter Collingbourne 
160cead56fbSPeter Collingbourne   if (Policy.Expiration == seconds(0) &&
1618d292233SPeter Collingbourne       Policy.MaxSizePercentageOfAvailableSpace == 0 &&
162048ac839SPeter Collingbourne       Policy.MaxSizeBytes == 0 && Policy.MaxSizeFiles == 0) {
163d34e60caSNicola Zaghen     LLVM_DEBUG(dbgs() << "No pruning settings set, exit early\n");
16427814980SMehdi Amini     // Nothing will be pruned, early exit
16527814980SMehdi Amini     return false;
166045d4754SMehdi Amini   }
16727814980SMehdi Amini 
16827814980SMehdi Amini   // Try to stat() the timestamp file.
169721800d4SMehdi Amini   SmallString<128> TimestampFile(Path);
170721800d4SMehdi Amini   sys::path::append(TimestampFile, "llvmcache.timestamp");
17127814980SMehdi Amini   sys::fs::file_status FileStatus;
172bb534b15SBen Dunbobbin   const auto CurrentTime = system_clock::now();
173f5e36ad8SNAKAMURA Takumi   if (auto EC = sys::fs::status(TimestampFile, FileStatus)) {
174f5e36ad8SNAKAMURA Takumi     if (EC == errc::no_such_file_or_directory) {
17527814980SMehdi Amini       // If the timestamp file wasn't there, create one now.
17627814980SMehdi Amini       writeTimestampFile(TimestampFile);
17727814980SMehdi Amini     } else {
17827814980SMehdi Amini       // Unknown error?
17927814980SMehdi Amini       return false;
18027814980SMehdi Amini     }
18127814980SMehdi Amini   } else {
182bb534b15SBen Dunbobbin     if (!Policy.Interval)
183bb534b15SBen Dunbobbin       return false;
184cac52140SBen Dunbobbin     if (Policy.Interval != seconds(0)) {
18527814980SMehdi Amini       // Check whether the time stamp is older than our pruning interval.
18627814980SMehdi Amini       // If not, do nothing.
187bb534b15SBen Dunbobbin       const auto TimeStampModTime = FileStatus.getLastModificationTime();
188045d4754SMehdi Amini       auto TimeStampAge = CurrentTime - TimeStampModTime;
189bb534b15SBen Dunbobbin       if (TimeStampAge <= *Policy.Interval) {
190d34e60caSNicola Zaghen         LLVM_DEBUG(dbgs() << "Timestamp file too recent ("
191757ca886SPavel Labath                           << duration_cast<seconds>(TimeStampAge).count()
192045d4754SMehdi Amini                           << "s old), do not prune.\n");
19327814980SMehdi Amini         return false;
19427814980SMehdi Amini       }
195045d4754SMehdi Amini     }
19627814980SMehdi Amini     // Write a new timestamp file so that nobody else attempts to prune.
19727814980SMehdi Amini     // There is a benign race condition here, if two processes happen to
19827814980SMehdi Amini     // notice at the same time that the timestamp is out-of-date.
19927814980SMehdi Amini     writeTimestampFile(TimestampFile);
20027814980SMehdi Amini   }
20127814980SMehdi Amini 
202481d224bSBob Haarman   // Keep track of files to delete to get below the size limit.
203481d224bSBob Haarman   // Order by time of last use so that recently used files are preserved.
204481d224bSBob Haarman   std::set<FileInfo> FileInfos;
20527814980SMehdi Amini   uint64_t TotalSize = 0;
20627814980SMehdi Amini 
20727814980SMehdi Amini   // Walk the entire directory cache, looking for unused files.
20827814980SMehdi Amini   std::error_code EC;
20927814980SMehdi Amini   SmallString<128> CachePathNative;
21027814980SMehdi Amini   sys::path::native(Path, CachePathNative);
21127814980SMehdi Amini   // Walk all of the files within this directory.
21227814980SMehdi Amini   for (sys::fs::directory_iterator File(CachePathNative, EC), FileEnd;
21327814980SMehdi Amini        File != FileEnd && !EC; File.increment(EC)) {
214*ec877106SHans Wennborg     // Ignore filenames not beginning with "llvmcache-" or "Thin-". This
21525a17ba4SPeter Collingbourne     // includes the timestamp file as well as any files created by the user.
21625a17ba4SPeter Collingbourne     // This acts as a safeguard against data loss if the user specifies the
21725a17ba4SPeter Collingbourne     // wrong directory as their cache directory.
218*ec877106SHans Wennborg     StringRef filename = sys::path::filename(File->path());
219*ec877106SHans Wennborg     if (!filename.startswith("llvmcache-") && !filename.startswith("Thin-"))
22027814980SMehdi Amini       continue;
22127814980SMehdi Amini 
22227814980SMehdi Amini     // Look at this file. If we can't stat it, there's nothing interesting
22327814980SMehdi Amini     // there.
2240dfdb447SPeter Collingbourne     ErrorOr<sys::fs::basic_file_status> StatusOrErr = File->status();
2250dfdb447SPeter Collingbourne     if (!StatusOrErr) {
226d34e60caSNicola Zaghen       LLVM_DEBUG(dbgs() << "Ignore " << File->path() << " (can't stat)\n");
22727814980SMehdi Amini       continue;
228045d4754SMehdi Amini     }
22927814980SMehdi Amini 
23027814980SMehdi Amini     // If the file hasn't been used recently enough, delete it
2310dfdb447SPeter Collingbourne     const auto FileAccessTime = StatusOrErr->getLastAccessedTime();
232045d4754SMehdi Amini     auto FileAge = CurrentTime - FileAccessTime;
233048ac839SPeter Collingbourne     if (Policy.Expiration != seconds(0) && FileAge > Policy.Expiration) {
234d34e60caSNicola Zaghen       LLVM_DEBUG(dbgs() << "Remove " << File->path() << " ("
235d34e60caSNicola Zaghen                         << duration_cast<seconds>(FileAge).count()
236d34e60caSNicola Zaghen                         << "s old)\n");
23727814980SMehdi Amini       sys::fs::remove(File->path());
23827814980SMehdi Amini       continue;
23927814980SMehdi Amini     }
24027814980SMehdi Amini 
24127814980SMehdi Amini     // Leave it here for now, but add it to the list of size-based pruning.
2420dfdb447SPeter Collingbourne     TotalSize += StatusOrErr->getSize();
243481d224bSBob Haarman     FileInfos.insert({FileAccessTime, StatusOrErr->getSize(), File->path()});
24427814980SMehdi Amini   }
24527814980SMehdi Amini 
246481d224bSBob Haarman   auto FileInfo = FileInfos.begin();
247481d224bSBob Haarman   size_t NumFiles = FileInfos.size();
248048ac839SPeter Collingbourne 
249048ac839SPeter Collingbourne   auto RemoveCacheFile = [&]() {
250048ac839SPeter Collingbourne     // Remove the file.
251481d224bSBob Haarman     sys::fs::remove(FileInfo->Path);
252048ac839SPeter Collingbourne     // Update size
253481d224bSBob Haarman     TotalSize -= FileInfo->Size;
254048ac839SPeter Collingbourne     NumFiles--;
255481d224bSBob Haarman     LLVM_DEBUG(dbgs() << " - Remove " << FileInfo->Path << " (size "
256481d224bSBob Haarman                       << FileInfo->Size << "), new occupancy is " << TotalSize
257481d224bSBob Haarman                       << "%\n");
258481d224bSBob Haarman     ++FileInfo;
259048ac839SPeter Collingbourne   };
260048ac839SPeter Collingbourne 
261048ac839SPeter Collingbourne   // Prune for number of files.
262048ac839SPeter Collingbourne   if (Policy.MaxSizeFiles)
263048ac839SPeter Collingbourne     while (NumFiles > Policy.MaxSizeFiles)
264048ac839SPeter Collingbourne       RemoveCacheFile();
265048ac839SPeter Collingbourne 
26627814980SMehdi Amini   // Prune for size now if needed
267048ac839SPeter Collingbourne   if (Policy.MaxSizePercentageOfAvailableSpace > 0 || Policy.MaxSizeBytes > 0) {
26827814980SMehdi Amini     auto ErrOrSpaceInfo = sys::fs::disk_space(Path);
26927814980SMehdi Amini     if (!ErrOrSpaceInfo) {
27027814980SMehdi Amini       report_fatal_error("Can't get available size");
27127814980SMehdi Amini     }
27227814980SMehdi Amini     sys::fs::space_info SpaceInfo = ErrOrSpaceInfo.get();
27327814980SMehdi Amini     auto AvailableSpace = TotalSize + SpaceInfo.free;
2748d292233SPeter Collingbourne 
2758d292233SPeter Collingbourne     if (Policy.MaxSizePercentageOfAvailableSpace == 0)
2768d292233SPeter Collingbourne       Policy.MaxSizePercentageOfAvailableSpace = 100;
2778d292233SPeter Collingbourne     if (Policy.MaxSizeBytes == 0)
2788d292233SPeter Collingbourne       Policy.MaxSizeBytes = AvailableSpace;
2798d292233SPeter Collingbourne     auto TotalSizeTarget = std::min<uint64_t>(
2808d292233SPeter Collingbourne         AvailableSpace * Policy.MaxSizePercentageOfAvailableSpace / 100ull,
2818d292233SPeter Collingbourne         Policy.MaxSizeBytes);
2828d292233SPeter Collingbourne 
283d34e60caSNicola Zaghen     LLVM_DEBUG(dbgs() << "Occupancy: " << ((100 * TotalSize) / AvailableSpace)
284d34e60caSNicola Zaghen                       << "% target is: "
285d34e60caSNicola Zaghen                       << Policy.MaxSizePercentageOfAvailableSpace << "%, "
286d34e60caSNicola Zaghen                       << Policy.MaxSizeBytes << " bytes\n");
2878d292233SPeter Collingbourne 
288048ac839SPeter Collingbourne     // Remove the oldest accessed files first, till we get below the threshold.
289481d224bSBob Haarman     while (TotalSize > TotalSizeTarget && FileInfo != FileInfos.end())
290048ac839SPeter Collingbourne       RemoveCacheFile();
29127814980SMehdi Amini   }
29227814980SMehdi Amini   return true;
29327814980SMehdi Amini }
294