On 31/08/05, John Maddock <john@johnmaddock.co.uk> wrote:
Matt, can you distil this test case a little more please:
* Get rid of the macros so we can read the code :-) * Don't use Boost.Test for the tests, it's *not* thread safe, and bad things may be happening in there. * Don't declare variables as function-static: it's not thread safe. I'm not sure if this is an issue or not from reading the code, with those macros in place it's hard to see which bits are getting called by multiple threads.
John, Hopefully this is a little more readable. My set up is gcc 3.3.5 on Suse 9.3 pro x86_64 with dual opterons. Hope this helps, matt. matt@zomojo.com ___________________________ #include <iostream> #include <boost/thread/thread.hpp> #include <boost/thread/read_write_mutex.hpp> using namespace boost; // these definitions deadlock read_write_mutex guard( read_write_scheduling_policy::writer_priority ); typedef read_write_mutex::scoped_write_lock test_lock_type; // these definitions work as expected // mutex guard; // typedef mutex::scoped_lock test_lock_type; long global_count = 0; long loop_count_for_test = 100000; void do_work() { for (long i=0; i< loop_count_for_test; ++i ) { test_lock_type lk( guard ); ++global_count; } } int main() { thread_group pool; int number_of_threads = 100; for (int i=0; i < number_of_threads; ++i) { pool.create_thread( do_work ); } pool.join_all(); //result check std::cout << "We should get here without a deadlock\n"; std::cout << "global count = " << global_count << "\n"; std::cout << "global count should be = " << loop_count_for_test * number_of_threads << "\n"; return 0; }