[clang] 0ffa29f - [clang][modules] Timestamp PCM files when writing (#112452)
via cfe-commits
cfe-commits at lists.llvm.org
Tue Oct 22 15:08:06 PDT 2024
Author: Jan Svoboda
Date: 2024-10-22T15:08:02-07:00
New Revision: 0ffa29fe8152e247eea87017e8c5aeedc6329c15
URL: https://github.com/llvm/llvm-project/commit/0ffa29fe8152e247eea87017e8c5aeedc6329c15
DIFF: https://github.com/llvm/llvm-project/commit/0ffa29fe8152e247eea87017e8c5aeedc6329c15.diff
LOG: [clang][modules] Timestamp PCM files when writing (#112452)
Clang uses timestamp files to track the last time an implicitly-built
PCM file was verified to be up-to-date with regard to its inputs. With
`-fbuild-session-{file,timestamp}=` and
`-fmodules-validate-once-per-build-session` this reduces the number of
times a PCM file is checked per "build session".
The behavior I'm seeing with the current scheme is that when lots of
Clang instances wait for the same PCM to be built, they race to validate
it as soon as the file lock gets released, causing lots of concurrent
IO.
This patch makes it so that the timestamp is written by the same Clang
instance responsible for building the PCM while still holding the lock.
This makes it so that whenever a PCM file gets compiled, it's never
re-validated in the same build session.
I believe this is as sound as the current scheme. One thing to be aware
of is that there might be a time interval between accessing input file N
and writing the timestamp file, where changes to input files 0..<N would
not result in a rebuild. Since this is the case current scheme too, I'm
not too concerned about that.
I've seen this speed up `clang-scan-deps` by ~27%.
Added:
Modified:
clang/include/clang/Serialization/ModuleFile.h
clang/lib/Serialization/ASTCommon.cpp
clang/lib/Serialization/ASTCommon.h
clang/lib/Serialization/ASTReader.cpp
clang/lib/Serialization/ASTWriter.cpp
clang/lib/Serialization/ModuleManager.cpp
Removed:
################################################################################
diff --git a/clang/include/clang/Serialization/ModuleFile.h b/clang/include/clang/Serialization/ModuleFile.h
index 30e7f6b3e57bd8..29d3354d07a129 100644
--- a/clang/include/clang/Serialization/ModuleFile.h
+++ b/clang/include/clang/Serialization/ModuleFile.h
@@ -15,6 +15,7 @@
#define LLVM_CLANG_SERIALIZATION_MODULEFILE_H
#include "clang/Basic/FileManager.h"
+#include "clang/Basic/LLVM.h"
#include "clang/Basic/Module.h"
#include "clang/Basic/SourceLocation.h"
#include "clang/Serialization/ASTBitCodes.h"
@@ -144,8 +145,8 @@ class ModuleFile {
/// The base directory of the module.
std::string BaseDirectory;
- std::string getTimestampFilename() const {
- return FileName + ".timestamp";
+ static std::string getTimestampFilename(StringRef FileName) {
+ return (FileName + ".timestamp").str();
}
/// The original source file name that was used to build the
diff --git a/clang/lib/Serialization/ASTCommon.cpp b/clang/lib/Serialization/ASTCommon.cpp
index ab4923de6346f8..ec18e84255ca8e 100644
--- a/clang/lib/Serialization/ASTCommon.cpp
+++ b/clang/lib/Serialization/ASTCommon.cpp
@@ -15,7 +15,10 @@
#include "clang/AST/DeclObjC.h"
#include "clang/Basic/IdentifierTable.h"
#include "clang/Serialization/ASTDeserializationListener.h"
+#include "clang/Serialization/ModuleFile.h"
#include "llvm/Support/DJB.h"
+#include "llvm/Support/FileSystem.h"
+#include "llvm/Support/raw_ostream.h"
using namespace clang;
@@ -503,3 +506,15 @@ bool serialization::needsAnonymousDeclarationNumber(const NamedDecl *D) {
return false;
return isa<TagDecl, FieldDecl>(D);
}
+
+void serialization::updateModuleTimestamp(StringRef ModuleFilename) {
+ // Overwrite the timestamp file contents so that file's mtime changes.
+ std::error_code EC;
+ llvm::raw_fd_ostream OS(ModuleFile::getTimestampFilename(ModuleFilename), EC,
+ llvm::sys::fs::OF_TextWithCRLF);
+ if (EC)
+ return;
+ OS << "Timestamp file\n";
+ OS.close();
+ OS.clear_error(); // Avoid triggering a fatal error.
+}
diff --git a/clang/lib/Serialization/ASTCommon.h b/clang/lib/Serialization/ASTCommon.h
index 0230908d3e0528..2a765eafe08951 100644
--- a/clang/lib/Serialization/ASTCommon.h
+++ b/clang/lib/Serialization/ASTCommon.h
@@ -15,6 +15,7 @@
#include "clang/AST/ASTContext.h"
#include "clang/AST/DeclFriend.h"
+#include "clang/Basic/LLVM.h"
#include "clang/Serialization/ASTBitCodes.h"
namespace clang {
@@ -100,6 +101,8 @@ inline bool isPartOfPerModuleInitializer(const Decl *D) {
return false;
}
+void updateModuleTimestamp(StringRef ModuleFilename);
+
} // namespace serialization
} // namespace clang
diff --git a/clang/lib/Serialization/ASTReader.cpp b/clang/lib/Serialization/ASTReader.cpp
index 60b708067dc597..7d9170e7f0b479 100644
--- a/clang/lib/Serialization/ASTReader.cpp
+++ b/clang/lib/Serialization/ASTReader.cpp
@@ -4416,19 +4416,6 @@ bool ASTReader::isGlobalIndexUnavailable() const {
!hasGlobalIndex() && TriedLoadingGlobalIndex;
}
-static void updateModuleTimestamp(ModuleFile &MF) {
- // Overwrite the timestamp file contents so that file's mtime changes.
- std::string TimestampFilename = MF.getTimestampFilename();
- std::error_code EC;
- llvm::raw_fd_ostream OS(TimestampFilename, EC,
- llvm::sys::fs::OF_TextWithCRLF);
- if (EC)
- return;
- OS << "Timestamp file\n";
- OS.close();
- OS.clear_error(); // Avoid triggering a fatal error.
-}
-
/// Given a cursor at the start of an AST file, scan ahead and drop the
/// cursor into the start of the given block ID, returning false on success and
/// true on failure.
@@ -4707,7 +4694,7 @@ ASTReader::ASTReadResult ASTReader::ReadAST(StringRef FileName, ModuleKind Type,
ImportedModule &M = Loaded[I];
if (M.Mod->Kind == MK_ImplicitModule &&
M.Mod->InputFilesValidationTimestamp < HSOpts.BuildSessionTimestamp)
- updateModuleTimestamp(*M.Mod);
+ updateModuleTimestamp(M.Mod->FileName);
}
}
diff --git a/clang/lib/Serialization/ASTWriter.cpp b/clang/lib/Serialization/ASTWriter.cpp
index 938d7b525cb959..494890284d2f2c 100644
--- a/clang/lib/Serialization/ASTWriter.cpp
+++ b/clang/lib/Serialization/ASTWriter.cpp
@@ -4905,6 +4905,12 @@ ASTFileSignature ASTWriter::WriteAST(Sema &SemaRef, StringRef OutputFile,
this->BaseDirectory.clear();
WritingAST = false;
+
+ if (WritingModule && SemaRef.PP.getHeaderSearchInfo()
+ .getHeaderSearchOpts()
+ .ModulesValidateOncePerBuildSession)
+ updateModuleTimestamp(OutputFile);
+
if (ShouldCacheASTInMemory) {
// Construct MemoryBuffer and update buffer manager.
ModuleCache.addBuiltPCM(OutputFile,
diff --git a/clang/lib/Serialization/ModuleManager.cpp b/clang/lib/Serialization/ModuleManager.cpp
index e74a16b6368028..ba78c9ef5af67f 100644
--- a/clang/lib/Serialization/ModuleManager.cpp
+++ b/clang/lib/Serialization/ModuleManager.cpp
@@ -170,7 +170,8 @@ ModuleManager::addModule(StringRef FileName, ModuleKind Type,
NewModule->InputFilesValidationTimestamp = 0;
if (NewModule->Kind == MK_ImplicitModule) {
- std::string TimestampFilename = NewModule->getTimestampFilename();
+ std::string TimestampFilename =
+ ModuleFile::getTimestampFilename(NewModule->FileName);
llvm::vfs::Status Status;
// A cached stat value would be fine as well.
if (!FileMgr.getNoncachedStatValue(TimestampFilename, Status))
More information about the cfe-commits
mailing list