mirror of
https://github.com/intel/llvm.git
synced 2026-01-27 06:06:34 +08:00
2694810153
Summary: So this patch just make sure that the thread is at least stated before we return from main. If we just detach then the thread may be actually be stated just after the process returned from main and it's calling atexit handers. Then the thread may try to create own function static variable and it will add new at exit handlers confusing libc. GLIBC before 2.27 had race in that case which corrupted atexit handlers list. Support for this use-case for other implementation is also unclear, so we can try just avoid that. PR40162 Reviewers: ruiu, espindola Subscribers: emaste, arichardson, jfb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D58246 llvm-svn: 354078
103 lines
3.5 KiB
C++
103 lines
3.5 KiB
C++
//===- Filesystem.cpp -----------------------------------------------------===//
|
|
//
|
|
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
|
|
// See https://llvm.org/LICENSE.txt for license information.
|
|
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
|
|
//
|
|
//===----------------------------------------------------------------------===//
|
|
//
|
|
// This file contains a few utility functions to handle files.
|
|
//
|
|
//===----------------------------------------------------------------------===//
|
|
|
|
#include "Filesystem.h"
|
|
#include "Config.h"
|
|
#include "lld/Common/Threads.h"
|
|
#include "llvm/Config/llvm-config.h"
|
|
#include "llvm/Support/FileOutputBuffer.h"
|
|
#include "llvm/Support/FileSystem.h"
|
|
#if LLVM_ON_UNIX
|
|
#include <unistd.h>
|
|
#endif
|
|
#include <thread>
|
|
|
|
using namespace llvm;
|
|
|
|
using namespace lld;
|
|
using namespace lld::elf;
|
|
|
|
// Removes a given file asynchronously. This is a performance hack,
|
|
// so remove this when operating systems are improved.
|
|
//
|
|
// On Linux (and probably on other Unix-like systems), unlink(2) is a
|
|
// noticeably slow system call. As of 2016, unlink takes 250
|
|
// milliseconds to remove a 1 GB file on ext4 filesystem on my machine.
|
|
//
|
|
// To create a new result file, we first remove existing file. So, if
|
|
// you repeatedly link a 1 GB program in a regular compile-link-debug
|
|
// cycle, every cycle wastes 250 milliseconds only to remove a file.
|
|
// Since LLD can link a 1 GB binary in about 5 seconds, that waste
|
|
// actually counts.
|
|
//
|
|
// This function spawns a background thread to remove the file.
|
|
// The calling thread returns almost immediately.
|
|
void elf::unlinkAsync(StringRef Path) {
|
|
// Removing a file is async on windows.
|
|
#if defined(_WIN32)
|
|
sys::fs::remove(Path);
|
|
#else
|
|
if (!ThreadsEnabled || !sys::fs::exists(Path) ||
|
|
!sys::fs::is_regular_file(Path))
|
|
return;
|
|
|
|
// We cannot just remove path from a different thread because we are now going
|
|
// to create path as a new file.
|
|
// Instead we open the file and unlink it on this thread. The unlink is fast
|
|
// since the open fd guarantees that it is not removing the last reference.
|
|
int FD;
|
|
std::error_code EC = sys::fs::openFileForRead(Path, FD);
|
|
sys::fs::remove(Path);
|
|
|
|
if (EC)
|
|
return;
|
|
|
|
// close and therefore remove TempPath in background.
|
|
std::mutex M;
|
|
std::condition_variable CV;
|
|
bool Started = false;
|
|
std::thread([&, FD] {
|
|
{
|
|
std::lock_guard<std::mutex> L(M);
|
|
Started = true;
|
|
CV.notify_all();
|
|
}
|
|
::close(FD);
|
|
}).detach();
|
|
|
|
// GLIBC 2.26 and earlier have race condition that crashes an entire process
|
|
// if the main thread calls exit(2) while other thread is starting up.
|
|
std::unique_lock<std::mutex> L(M);
|
|
CV.wait(L, [&] { return Started; });
|
|
#endif
|
|
}
|
|
|
|
// Simulate file creation to see if Path is writable.
|
|
//
|
|
// Determining whether a file is writable or not is amazingly hard,
|
|
// and after all the only reliable way of doing that is to actually
|
|
// create a file. But we don't want to do that in this function
|
|
// because LLD shouldn't update any file if it will end in a failure.
|
|
// We also don't want to reimplement heuristics to determine if a
|
|
// file is writable. So we'll let FileOutputBuffer do the work.
|
|
//
|
|
// FileOutputBuffer doesn't touch a desitnation file until commit()
|
|
// is called. We use that class without calling commit() to predict
|
|
// if the given file is writable.
|
|
std::error_code elf::tryCreateFile(StringRef Path) {
|
|
if (Path.empty())
|
|
return std::error_code();
|
|
if (Path == "-")
|
|
return std::error_code();
|
|
return errorToErrorCode(FileOutputBuffer::create(Path, 1).takeError());
|
|
}
|