[英]boost::function deallocation segmentation fault in thread pool
我正在嘗試創建一個阻塞主線程的線程池,直到它的所有子線程都完成。 現實世界的用例是一個“控制器”進程,它產生獨立的進程供用戶與之交互。
不幸的是,當main
退出時,遇到了segmentation fault。 我無法弄清楚這個分段錯誤的原因。
我編寫了一個Process
類,它只不過是打開一個 shell 腳本(稱為waiter.sh
,其中包含一個sleep 5
)並等待pid
退出。 初始化Process
類,然后將Wait()
方法放置在線程池中的線程之一中。
~thread_pool()
時會出現問題。 std::queue
無法正確釋放傳遞給它的boost::function
,即使對Process
的引用仍然有效。
#include <sys/types.h>
#include <sys/wait.h>
#include <spawn.h>
#include <queue>
#include <boost/bind.hpp>
#include <boost/thread.hpp>
extern char **environ;
class Process {
private:
pid_t pid;
int status;
public:
Process() : status(0), pid(-1) {
}
~Process() {
std::cout << "calling ~Process" << std::endl;
}
void Spawn(char **argv) {
// want spawn posix and wait for th epid to return
status = posix_spawn(&pid, "waiter.sh", NULL, NULL, argv, environ);
if (status != 0) {
perror("unable to spawn");
return;
}
}
void Wait() {
std::cout << "spawned proc with " << pid << std::endl;
waitpid(pid, &status, 0);
// wait(&pid);
std::cout << "wait complete" << std::endl;
}
};
下面是thread_pool
類。 這是從這個問題的公認答案中松散地改編的
class thread_pool {
private:
std::queue<boost::function<void() >> tasks;
boost::thread_group threads;
std::size_t available;
boost::mutex mutex;
boost::condition_variable condition;
bool running;
public:
thread_pool(std::size_t pool_size) : available(pool_size), running(true) {
std::cout << "creating " << pool_size << " threads" << std::endl;
for (std::size_t i = 0; i < available; ++i) {
threads.create_thread(boost::bind(&thread_pool::pool_main, this));
}
}
~thread_pool() {
std::cout << "~thread_pool" << std::endl;
{
boost::unique_lock<boost::mutex> lock(mutex);
running = false;
condition.notify_all();
}
try {
threads.join_all();
} catch (const std::exception &) {
// supress exceptions
}
}
template <typename Task>
void run_task(Task task) {
boost::unique_lock<boost::mutex> lock(mutex);
if (0 == available) {
return; //\todo err
}
--available;
tasks.push(boost::function<void()>(task));
condition.notify_one();
return;
}
private:
void pool_main() {
// wait on condition variable while the task is empty and the pool is still
// running
boost::unique_lock<boost::mutex> lock(mutex);
while (tasks.empty() && running) {
condition.wait(lock);
}
// copy task locally and remove from the queue. this is
// done within it's own scope so that the task object is destructed
// immediately after running the task. This is useful in the
// event that the function contains shared_ptr arguments
// bound via 'bind'
{
auto task = tasks.front();
tasks.pop();
lock.unlock();
// run the task
try {
std::cout << "running task" << std::endl;
task();
} catch (const std::exception &) {
// supress
}
}
// task has finished so increment count of availabe threads
lock.lock();
++available;
}
};
這里是主要的:
int main() {
// input arguments are not required
char *argv[] = {NULL};
Process process;
process.Spawn(argv);
thread_pool pool(5);
pool.run_task(boost::bind(&Process::Wait, &process));
return 0;
}
這個的輸出是
creating 5 threads
~thread_pool
I am waiting... (from waiting.sh)
running task
spawned proc with 2573
running task
running task
running task
running task
wait complete
Segmentation fault (core dumped)
這是堆棧跟蹤:
Starting program: /home/jandreau/NetBeansProjects/Controller/dist/Debug/GNU- Linux/controller
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
creating 5 threads
[New Thread 0x7ffff691d700 (LWP 2600)]
[New Thread 0x7ffff611c700 (LWP 2601)]
[New Thread 0x7ffff591b700 (LWP 2602)]
[New Thread 0x7ffff511a700 (LWP 2603)]
[New Thread 0x7ffff4919700 (LWP 2604)]
~thread_pool
running task
running task
spawned proc with 2599
[Thread 0x7ffff611c700 (LWP 2601) exited]
running task
[Thread 0x7ffff591b700 (LWP 2602) exited]
running task
[Thread 0x7ffff511a700 (LWP 2603) exited]
running task
[Thread 0x7ffff4919700 (LWP 2604) exited]
I am waiting...
wait complete
[Thread 0x7ffff691d700 (LWP 2600) exited]
Thread 1 "controller" received signal SIGSEGV, Segmentation fault.
0x000000000040f482 in boost::detail::function::basic_vtable0<void>::clear (
this=0xa393935322068, functor=...)
at /usr/include/boost/function/function_template.hpp:509
509 if (base.manager)
(gdb) where
#0 0x000000000040f482 in boost::detail::function::basic_vtable0<void>::clear (
this=0xa393935322068, functor=...)
at /usr/include/boost/function/function_template.hpp:509
#1 0x000000000040e263 in boost::function0<void>::clear (this=0x62ef50)
at /usr/include/boost/function/function_template.hpp:883
#2 0x000000000040cf20 in boost::function0<void>::~function0 (this=0x62ef50,
__in_chrg=<optimized out>)
at /usr/include/boost/function/function_template.hpp:765
#3 0x000000000040b28e in boost::function<void ()>::~function() (
this=0x62ef50, __in_chrg=<optimized out>)
at /usr/include/boost/function/function_template.hpp:1056
#4 0x000000000041193a in std::_Destroy<boost::function<void ()> >(boost::function<void ()>*) (__pointer=0x62ef50)
at /usr/include/c++/5/bits/stl_construct.h:93
#5 0x00000000004112df in std::_Destroy_aux<false>::__destroy<boost::function<void ()>*>(boost::function<void ()>*, boost::function<void ()>*) (
__first=0x62ef50, __last=0x62ed50)
at /usr/include/c++/5/bits/stl_construct.h:103
#6 0x0000000000410d16 in std::_Destroy<boost::function<void ()>*>(boost::function<void ()>*, boost::function<void ()>*) (__first=0x62edd0, __last=0x62ed50)
at /usr/include/c++/5/bits/stl_construct.h:126
#7 0x0000000000410608 in std::_Destroy<boost::function<void ()>*, boost::function<void ()> >(boost::function<void ()>*, boost::function<void ()>*, std::allocat---Type <return> to continue, or q <return> to quit---
or<boost::function<void ()> >&) (__first=0x62edd0, __last=0x62ed50)
at /usr/include/c++/5/bits/stl_construct.h:151
#8 0x000000000040fac5 in std::deque<boost::function<void ()>, std::allocator<boost::function<void ()> > >::_M_destroy_data_aux(std::_Deque_iterator<boost::function<void ()>, boost::function<void ()>&, boost::function<void ()>*>, std::_Deque_iterator<boost::function<void ()>, boost::function<void ()>&, boost::function<void ()>*>) (this=0x7fffffffdaf0, __first=..., __last=...)
at /usr/include/c++/5/bits/deque.tcc:845
#9 0x000000000040e6e4 in std::deque<boost::function<void ()>, std::allocator<boost::function<void ()> > >::_M_destroy_data(std::_Deque_iterator<boost::function<void ()>, boost::function<void ()>&, boost::function<void ()>*>, std::_Deque_iterator<boost::function<void ()>, boost::function<void ()>&, boost::function<void ()>*>, std::allocator<boost::function<void ()> > const&) (
this=0x7fffffffdaf0, __first=..., __last=...)
at /usr/include/c++/5/bits/stl_deque.h:2037
#10 0x000000000040d0c8 in std::deque<boost::function<void ()>, std::allocator<boost::function<void ()> > >::~deque() (this=0x7fffffffdaf0,
__in_chrg=<optimized out>) at /usr/include/c++/5/bits/stl_deque.h:1039
#11 0x000000000040b3ce in std::queue<boost::function<void ()>, std::deque<boost::function<void ()>, std::allocator<boost::function<void ()> > > >::~queue() (
this=0x7fffffffdaf0, __in_chrg=<optimized out>)
at /usr/include/c++/5/bits/stl_queue.h:96
#12 0x000000000040b6c0 in thread_pool::~thread_pool (this=0x7fffffffdaf0,
---Type <return> to continue, or q <return> to quit---
__in_chrg=<optimized out>) at main.cpp:63
#13 0x0000000000408b60 in main () at main.cpp:140
我對此感到困惑,因為Process
尚未超出范圍,我將boost::function<void()>
的副本傳遞給線程池進行處理。
有任何想法嗎?
堆棧跟蹤表明您正在銷毀未正確初始化的std::function
(例如,某些隨機內存位置被視為std::function
)或您正在銷毀std::function
兩次。
問題是您的程序只推送一次tasks
,但彈出五次,因此您從空雙端隊列中刪除元素,這是未定義的行為。
如果running
為 false,則pool_main
的while
循環終止,即使deque
為空, running
也可能為 false。 然后你無條件pop
。 您可以考慮按如下方式更正pool_main
:
void pool_main() {
// wait on condition variable
// while the task is empty and the pool is still
// running
boost::unique_lock<boost::mutex> lock(mutex);
while (tasks.empty() && running) {
condition.wait(lock);
}
// copy task locally and remove from the queue. this is
// done within it's own scope so that the task object is destructed
// immediately after running the task. This is useful in the
// event that the function contains shared_ptr arguments
// bound via 'bind'
if (!tasks.empty ()) { // <--- !!!!!!!!!!!!!!!!!!!!!!!!
auto task = tasks.front();
tasks.pop();
lock.unlock();
// run the task
try {
std::cout << "running task" << std::endl;
task();
} catch (const std::exception &) {
// supress
}
}
// task has finished so increment count of availabe threads
lock.lock();
++available;
};
但是,我不確定有關available
的邏輯是否正確。 不應available
在開始任務的處理遞減和遞增它完成時(因此被內改變pool_main
,並僅中新引入的if
從句)?
你似乎沒有為
extern char **environ;
任何地方。 雖然那不會是鏈接錯誤嗎?
將其減少到最小的復制案例將有很大幫助。 這里有很多代碼可能不是重現問題所必需的。
還有,這是什么:
// supress exceptions
如果您在加入線程時遇到異常,那么您可能還沒有加入所有線程,並且在不加入它們的情況下清理線程會在主退出后導致錯誤。
聲明:本站的技術帖子網頁,遵循CC BY-SA 4.0協議,如果您需要轉載,請注明本站網址或者原文地址。任何問題請咨詢:yoyou2525@163.com.