관리-도구
편집 파일: Utils.h
/* * Phusion Passenger - https://www.phusionpassenger.com/ * Copyright (c) 2010-2017 Phusion Holding B.V. * * "Passenger", "Phusion Passenger" and "Union Station" are registered * trademarks of Phusion Holding B.V. * * Permission is hereby granted, free of charge, to any person obtaining a copy * of this software and associated documentation files (the "Software"), to deal * in the Software without restriction, including without limitation the rights * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell * copies of the Software, and to permit persons to whom the Software is * furnished to do so, subject to the following conditions: * * The above copyright notice and this permission notice shall be included in * all copies or substantial portions of the Software. * * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN * THE SOFTWARE. */ #ifndef _PASSENGER_PROCESS_MANAGEMENT_UTILS_H_ #define _PASSENGER_PROCESS_MANAGEMENT_UTILS_H_ #include <sys/types.h> namespace Passenger { using namespace std; /** * Thread-safe and async-signal safe way to fork(). * * On Linux, the fork() glibc wrapper grabs a ptmalloc lock, so * if malloc causes a segfault then we can't fork. * http://sourceware.org/bugzilla/show_bug.cgi?id=4737 * * OS X apparently does something similar, except they use a * spinlock so it results in 100% CPU. See _cthread_fork_prepare() * at http://www.opensource.apple.com/source/Libc/Libc-166/threads.subproj/cthreads.c * However, since POSIX in OS X is implemented on top of a Mach layer, * calling asyncFork() can mess up the state of the Mach layer, causing * some POSIX functions to mysteriously fail. See * https://code.google.com/p/phusion-passenger/issues/detail?id=1094 * You should therefore not use asyncFork() unless you're in a signal * handler or if you only perform async-signal-safe stuff in the child. * * On 2017 October 9 with macOS 10.11 El Capitan, we also confirmed * a case in which the child process can get stuck indefinitely with 0% CPU. * If we we create a thread which performs memory allocation, and shortly * after thread creation we fork, then the child process gets stuck because * one of its pthread_atfork() handlers tries to allocate memory, which tries * to grab a lock which was already locked. This means that on macOS * we pretty much can never use regular fork() at all in a multithreaded * environment. * * As of 2018 May 16 with macOS 10.13 High Sierra, it was confirmed that the * use of asyncFork() can lead to the following messages to be printed * if the child process allocates memory: * * malloc: *** mach_vm_map(size=1048576) failed (error code=268435459) * malloc: *** error: can't allocate region securely * malloc: *** set a breakpoint in malloc_error_break to debug * * See https://github.com/phusion/passenger/issues/1193#issuecomment-389503928 */ pid_t asyncFork(); /** * Resets the current process's signal handler disposition and signal mask * to default values. One should call this every time one forks a child process; * non-default signal masks/handler dispositions can cause all kinds of weird quirks, * like waitpid() malfunctioning on macOS. * * This function is async-signal safe. */ void resetSignalHandlersAndMask(); /** * Disables malloc() debugging facilities on macOS. */ void disableMallocDebugging(); /** * Close all file descriptors that are higher than `lastToKeepOpen`. * * If you set `asyncSignalSafe` to true, then this function becomes fully async-signal, * through the use of asyncFork() instead of fork(). However, read the documentation * for asyncFork() to learn about its caveats. * * Also, regardless of whether `asyncSignalSafe` is true or not, this function is not * *thread* safe. Make sure there are no other threads running that might open file * descriptors, otherwise some file descriptors might not be closed even though they * should be. */ void closeAllFileDescriptors(int lastToKeepOpen, bool asyncSignalSafe = false); /** * Given a failed exec() syscall and its resulting errno * value, print an appropriate error message to STDERR. * * This function is async signal-safe. Its main intended use is to be the * default value for the `onExecFail` parameter for the * `runCommand()` and `runCommandAndCaptureOutput()` functions. */ void printExecError(const char **command, int errcode); void printExecError2(const char **command, int errcode, char *buf, size_t size); } // namespace Passenger #endif /* _PASSENGER_PROCESS_MANAGEMENT_UTILS_H_ */