class 销毁时出现分段错误，boost::timer 作为 class 的成员定期调用
Segmentation fault on class destruction with boost::timer as a member of the class with periodic invocation
我正在开发一个简单的 class，它在创建时会安排一个周期性计时器来调用其方法之一。该方法是虚拟的，因此派生的 classes 可以用他们需要的任何周期性工作来重载它。
然而，在我对此 class 的测试中，我随机遇到了分段错误并且无法弄清楚原因。这是好的和坏的输出的代码和示例：
#include <boost/thread/mutex.hpp>
#include <boost/thread/lock_guard.hpp>
#include <boost/asio/steady_timer.hpp>
#include <boost/chrono.hpp>
#include <boost/enable_shared_from_this.hpp>
#include <boost/function.hpp>
#include <boost/atomic.hpp>
#include <boost/make_shared.hpp>
#include <boost/bind.hpp>
//******************************************************************************
class PeriodicImpl;
class Periodic {
public:
Periodic(boost::asio::io_service& io, unsigned int periodMs);
~Periodic();
virtual unsigned int periodicInvocation() = 0;
private:
boost::shared_ptr<PeriodicImpl> pimpl_;
};
//******************************************************************************
class PeriodicImpl : public boost::enable_shared_from_this<PeriodicImpl>
{
public:
PeriodicImpl(boost::asio::io_service& io, unsigned int periodMs,
boost::function<unsigned int(void)> workFunc);
~PeriodicImpl();
void setupTimer(unsigned int intervalMs);
boost::atomic<bool> isRunning_;
unsigned int periodMs_;
boost::asio::io_service& io_;
boost::function<unsigned int(void)> workFunc_;
boost::asio::steady_timer timer_;
};
//******************************************************************************
Periodic::Periodic(boost::asio::io_service& io, unsigned int periodMs):
pimpl_(boost::make_shared<PeriodicImpl>(io, periodMs, boost::bind(&Periodic::periodicInvocation, this)))
{
std::cout << "periodic ctor " << pimpl_.use_count() << std::endl;
pimpl_->setupTimer(periodMs);
}
Periodic::~Periodic()
{
std::cout << "periodic dtor " << pimpl_.use_count() << std::endl;
pimpl_->isRunning_ = false;
pimpl_->timer_.cancel();
std::cout << "periodic dtor end " << pimpl_.use_count() << std::endl;
}
//******************************************************************************
PeriodicImpl::PeriodicImpl(boost::asio::io_service& io, unsigned int periodMs,
boost::function<unsigned int(void)> workFunc):
isRunning_(true),
io_(io), periodMs_(periodMs), workFunc_(workFunc), timer_(io_)
{
}
PeriodicImpl::~PeriodicImpl()
{
std::cout << "periodic impl dtor" << std::endl;
}
void
PeriodicImpl::setupTimer(unsigned int intervalMs)
{
std::cout << "schedule new " << intervalMs << std::endl;
boost::shared_ptr<PeriodicImpl> self(shared_from_this());
timer_.expires_from_now(boost::chrono::milliseconds(intervalMs));
timer_.async_wait([self, this](const boost::system::error_code& e){
std::cout << "hello invoke" << std::endl;
if (!e)
{
if (isRunning_)
{
std::cout << "invoking" << std::endl;
unsigned int nextIntervalMs = workFunc_();
if (nextIntervalMs)
setupTimer(nextIntervalMs);
}
else
std::cout << "invoke not running" << std::endl;
}
else
std::cout << "invoke cancel" << std::endl;
});
std::cout << "scheduled " << self.use_count() << std::endl;
}
//******************************************************************************
class PeriodicTest : public Periodic
{
public:
PeriodicTest(boost::asio::io_service& io, unsigned int periodMs):
Periodic(io, periodMs), periodMs_(periodMs), workCounter_(0){}
~PeriodicTest(){
std::cout << "periodic test dtor" << std::endl;
}
unsigned int periodicInvocation() {
std::cout << "invocation " << workCounter_ << std::endl;
workCounter_++;
return periodMs_;
}
unsigned int periodMs_;
unsigned int workCounter_;
};
//******************************************************************************
void main()
{
boost::asio::io_service io;
boost::shared_ptr<boost::asio::io_service::work> work(new boost::asio::io_service::work(io));
boost::thread t([&io](){
io.run();
});
unsigned int workCounter = 0;
{
PeriodicTest p(io, 50);
boost::this_thread::sleep_for(boost::chrono::milliseconds(550));
workCounter = p.workCounter_;
}
work.reset();
//EXPECT_EQ(10, workCounter);
}
好的输出：
hello invoke
invoking
invocation 9
schedule new 50
scheduled 5
periodic test dtor
periodic dtor 2
periodic dtor end 2
hello invoke
invoke cancel
periodic impl dtor
错误输出：
hello invoke
invoking
invocation 9
schedule new 50
scheduled 5
periodic test dtor
periodic dtor 2
periodic dtor end 2
periodic impl dtor
Segmentation fault: 11
显然，发生分段错误是因为 PeriodicImpl 被破坏，因此它的计时器 timer_。但是计时器仍然被安排 - 这导致 SEGFAULT。我不明白为什么在这种情况下调用 PeriodicImpl 析构函数，因为 shared_ptr 到 PeriodicImpl 在 setupTimer 调用期间被复制到作为计时器处理函数传递的 lambda 而这应该保留 PeriodicImpl 的副本并防止析构函数调用。
有什么想法吗？
我运行你的program.Regreattably，编译失败。 我添加到你的程序中，修改如下代码：
timer_.expires_from_now(boost::chrono::milliseconds(intervalMs));
修改：
timer_.expires_from_now(std::chrono::milliseconds(intervalMs));
所以，我得到的结果与你的 "Good output" 相同，但结果与你的 "Bad output" 不同。
原来问题完全不在被质疑的代码中，而是在测试它的代码中。
我通过 运行 ulimit -c unlimited 启用保存核心转储文件，然后使用 lldb 读取它：
$ lldb bin/tests/test-segment-controller -c /cores/core.75876
(lldb) bt all
* thread #1: tid = 0x0000, 0x00007fff8eb800f9 libsystem_malloc. dylib`szone_malloc_should_clear + 2642, stop reason = signal SIGSTOP
* frame #0: 0x00007fff8eb800f9 libsystem_malloc.dylib`szone_malloc_should_clear + 2642
frame #1: 0x00007fff8eb7f667 libsystem_malloc.dylib`malloc_zone_malloc + 71
frame #2: 0x00007fff8eb7e187 libsystem_malloc.dylib`malloc + 42
frame #3: 0x00007fff9569923e libc++abi.dylib`operator new(unsigned long) + 30
frame #4: 0x000000010da4b516 test-periodic`testing::Message::Message( this=0x00007fff521e8450) + 38 at gtest.cc:946
frame #5: 0x000000010da4a645 test-periodic`testing::Message::Message( this=0x00007fff521e8450) + 21 at gtest.cc:946
frame #6: 0x000000010da6c027 test-periodic`std::string testing::internal::StreamableToString<long long>(streamable=0x00007fff521e84b0) + 39 at gtest-message.h:244
frame #7: 0x000000010da558e8 test- periodic`testing::internal::PrettyUnitTestResultPrinter::OnTestEnd( this=0x00007fe733421570, test_info=0x00007fe7334211c0) + 216 at gtest.cc:3141
frame #8: 0x000000010da56a28 test- periodic`testing::internal::TestEventRepeater::OnTestEnd( this=0x00007fe733421520, parameter=0x00007fe7334211c0) + 136 at gtest.cc:3321
frame #9: 0x000000010da53957 test-periodic`testing::TestInfo::Run( this=0x00007fe7334211c0) + 343 at gtest.cc:2667
frame #10: 0x000000010da540c7 test-periodic`testing::TestCase::Run( this=0x00007fe733421660) + 231 at gtest.cc:2774
frame #11: 0x000000010da5b5d6 test- periodic`testing::internal::UnitTestImpl::RunAllTests(this=0x00007fe733421310) + 726 at gtest.cc:4649
frame #12: 0x000000010da83263 test-periodic`bool testing::internal::HandleSehExceptionsInMethodIfSupported< testing::internal::UnitTestImpl, bool>(object=0x00007fe733421310, method=0x000000010da5b300, location="auxiliary test code (environments or event listeners)")(), char const*) + 131 at gtest.cc:2402
frame #13: 0x000000010da6cde1 test-periodic`bool testing::internal::HandleExceptionsInMethodIfSupported< testing::internal::UnitTestImpl, bool>(object=0x00007fe733421310, method=0x000000010da5b300, location="auxiliary test code (environments or event listeners)")(), char const*) + 113 at gtest.cc:2438
frame #14: 0x000000010da5b2a2 test-periodic`testing::UnitTest::Run( this=0x000000010dab18e8) + 210 at gtest.cc:4257
frame #15: 0x000000010da19541 test-periodic`RUN_ALL_TESTS() + 17 at gtest. h:2233
frame #16: 0x000000010da1818b test-periodic`main(argc=1, argv=0x00007fff521e88b8) + 43 at test-periodic.cc:57
frame #17: 0x00007fff9557b5c9 libdyld.dylib`start + 1
frame #18: 0x00007fff9557b5c9 libdyld.dylib`start + 1
thread #2: tid = 0x0001, 0x00007fff8ab404cd libsystem_pthread. dylib`_pthread_mutex_lock + 23, stop reason = signal SIGSTOP
frame #0: 0x00007fff8ab404cd libsystem_pthread.dylib`_pthread_mutex_lock + 23
frame #1: 0x000000010da1c8d5 test- periodic`boost::asio::detail::posix_mutex::lock(this=0x0000000000000030) + 21 at posix_mutex.hpp:52
frame #2: 0x000000010da1c883 test-periodic`boost::asio::detail::scoped_lock< boost::asio::detail::posix_mutex>::scoped_lock(this=0x000000010e4fac38, m=0x0000000000000030) + 51 at scoped_lock.hpp:46
frame #3: 0x000000010da1c79d test-periodic`boost::asio::detail::scoped_lock< boost::asio::detail::posix_mutex>::scoped_lock(this=0x000000010e4fac38, m=0x0000000000000030) + 29 at scoped_lock.hpp:45
frame #4: 0x000000010da227a7 test- periodic`boost::asio::detail::kqueue_reactor::run(this=0x0000000000000000, block=true, ops=0x000000010e4fbda8) + 103 at kqueue_reactor.ipp:355
frame #5: 0x000000010da2223c test- periodic`boost::asio::detail::task_io_service::do_run_one( this=0x00007fe733421900, lock=0x000000010e4fbd60, this_thread=0x000000010e4fbd98, ec=0x000000010e4fbe58) + 348 at task_io_service .ipp:368
frame #6: 0x000000010da21e9f test- periodic`boost::asio::detail::task_io_service::run(this=0x00007fe733421900, ec=0x000000010e4fbe58) + 303 at task_io_service.ipp:153
frame #7: 0x000000010da21d51 test-periodic`boost::asio::io_service::run( this=0x00007fff521e8338) + 49 at io_service.ipp:59
frame #8: 0x000000010da184b8 test- periodic`TestPeriodic_TestDestructionDifferentThread_Test::TestBody( this=0x00007fe733421e28)::$_0::operator()() const + 24 at test-periodic.cc:41
frame #9: 0x000000010da1846c test-periodic`boost::detail::thread_data< TestPeriodic_TestDestructionDifferentThread_Test::TestBody()::$_0>::run( this=0x00007fe733421c10) + 28 at thread.hpp:117
frame #10: 0x000000010da8849c test-periodic`boost::(anonymous namespace) ::thread_proxy(param=<unavailable>) + 124 at thread.cpp:164
frame #11: 0x00007fff8ab4305a libsystem_pthread.dylib`_pthread_body + 131
frame #12: 0x00007fff8ab42fd7 libsystem_pthread.dylib`_pthread_start + 176
frame #13: 0x00007fff8ab403ed libsystem_pthread.dylib`thread_start + 13
显然，线程 2 在尝试锁定已被破坏的互斥量时导致崩溃。但是，我没有使用任何互斥体，所以这一定是 io_service 内部的东西。如果 io_service 在其被销毁后仍在使用，则可能会发生这种情况。仔细观察我的 main() 函数，我注意到我创建的线程 t 悬空，即没有 join() 调用它。因此，这有时会造成 io 对象已被破坏（在 main 结束时）但线程 t 仍尝试使用它的情况。
因此，通过在 main() 函数末尾添加 t.join() 调用解决了问题：
void main()
{
boost::asio::io_service io;
boost::shared_ptr<boost::asio::io_service::work> work(new boost::asio::io_service::work(io));
boost::thread t([&io](){
io.run();
});
unsigned int workCounter = 0;
{
PeriodicTest p(io, 50);
boost::this_thread::sleep_for(boost::chrono::milliseconds(550));
workCounter = p.workCounter_;
}
work.reset();
t.join();
//EXPECT_EQ(10, workCounter);
}