[Boost-users] [asio] Crash when posted handlers execute in one thread only?

Subject: [Boost-users] [asio] Crash when posted handlers execute in one thread only?
From: Sohail Somani (sohail_at_[hidden])
Date: 2014-04-05 10:49:18

Hi,

Not sure if there is a race condition in the code below, or if I am not 
stopping the service in a sanctioned manner but when run, the code 
consistently crashes on OSX 10.9 with Boost 1.55 and Clang. The only 
regular pattern was that the crash occurs when all the posted completion 
handlers execute in one thread only. When the completion handlers 
execute across both threads, there is no crash. This seems to me like 
there could be an initialization bug somewhere. I'll try and create a 
smaller test case.

Thanks for any ideas or suggestions.

The code:

#define LOG(msg) {                                      \
     boost::mutex::scoped_lock lock(log_mutex);          \
     std::cout << "[" << boost::this_thread::get_id()    \
               << "] " << msg << std::endl;              \
   }

void test() {
   boost::mutex log_mutex;

   LOG("Boost version: " << BOOST_LIB_VERSION);

   for(int ii = 0; ii < 100; ++ii) {
     LOG("Iteration: " << ii);
     boost::asio::io_service service;
     boost::asio::io_service::work work(service);

     boost::thread_group group;
     auto fun = [&](){
       LOG("Starting service");
       service.run();
       LOG("Completing service");
     };
     group.create_thread(fun);
     group.create_thread(fun);

     int ntasks = 10;
     volatile int val = 2;
     std::vector<volatile int> ints(ntasks,val);

     for(int ii = 0; ii < ntasks; ++ii) {
       service.post([&ints,ii,&log_mutex]() {
           LOG("Task: " << ii);
           ints[ii] = 1;
         });
     }

     service.post([&service,&log_mutex](){
         LOG("Stopping service");
         service.stop();
       });

     group.join_all();

     for(int ii = 0; ii < ntasks; ++ii) {
       if(ints[ii] != 1) {
         LOG(ii << "->" << ints[ii]);
       }
     }
   }

   LOG("Done test");
}

int main() {
   try {
     test();
   }
   catch(...) {
     std::cout << "Exception in main()" << std::endl;
   }
}

The output + stack trace:

...
[0x7fff7aeb9310] Iteration: 91
[0x104681000] Starting service
[0x104704000] Starting service
[0x104704000] Task: 0
[0x104704000] Task: 1
[0x104704000] Task: 2
[0x104704000] Task: 3
[0x104704000] Task: 4
[0x104704000] Task: 5
[0x104704000] Task: 6
[0x104704000] Task: 7
[0x104704000] Task: 8
[0x104704000] Task: 9
[0x104704000] Stopping service
[0x104704000] Completing service
[0x104681000] Completing service

Program received signal EXC_BAD_ACCESS, Could not access memory.
Reason: KERN_INVALID_ADDRESS at address: 0x0000000104703c57
0x00007fff8b06cdf3 in _pthread_cond_updateval ()
(gdb) where
#0  0x00007fff8b06cdf3 in _pthread_cond_updateval ()
#1  0x00007fff8b06c91e in _pthread_cond_signal ()
#2  0x00000001000048d0 in 
boost::asio::detail::posix_event::signal_and_unlock<boost::asio::detail::scoped_lock<boost::asio::detail::posix_mutex> 
 > () at build/boost/include/boost/asio/detail/posix_event.hpp:62
#3  0x00000001000048d0 in 
boost::asio::detail::task_io_service::wake_one_idle_thread_and_unlock () 
at build/boost/include/boost/asio/detail/task_io_service.hpp:484
#4  0x00000001000048d0 in 
boost::asio::detail::task_io_service::wake_one_thread_and_unlock 
(this=<value temporarily unavailable, due to optimizations>, lock=<value 
temporarily unavailable, due to optimizations>) at posix_event.hpp:493
#5  0x0000000100004776 in 
boost::asio::detail::scoped_lock<boost::asio::detail::posix_mutex>::~scoped_lock 
() at task_io_service.ipp:278
#6 
boost::asio::detail::scoped_lock<boost::asio::detail::posix_mutex>::~scoped_lock 
() at build/boost/include/boost/asio/detail/scoped_lock.hpp:52
#7  0x0000000100004776 in 
boost::asio::detail::task_io_service::post_immediate_completion 
(this=<value temporarily unavailable, due to optimizations>, op=<value 
temporarily unavailable, due to optimizations>, is_continuation=<value 
temporarily unavailable, due to optimizations>) at scoped_lock.hpp:279
#8  0x0000000100001952 in test () at task_io_service.hpp:70
#9  0x0000000100002ca0 in main () at app/exe/testcpp/main.cpp:64
(gdb) info threads
   2 0x00007fff985fc662 in kevent64 ()
* 1 0x00007fff8b06cdf3 in _pthread_cond_updateval ()