I am having issues with boost::interprocess::ipcdetail::spin_condition::do_wait_timed deadlocking.
The issue occurs when a thread that is waiting due to a call to interprocess_condition_any::wait() terminates unexpectedly and interprocess_condition_any::notify_all() is called (from any process). The cause of the deadlock is not that
the terminating thread held a lock – it doesn’t because it is waiting, but that spin_condition::m_num_waiters is never decremented to zero because one of the threads is no longer waiting and so the enter mutex is not freed.
I have tried BOOST_INTERPROCESS_ENABLE_TIMEOUT_WHEN_LOCKING, however as far as I can see this only tells me that the deadlock is there.
Are there any suggestions how I can recover from spin_condition not freeing the enter mutex? Or perhaps some way of avoiding the issue?
Senior Software Engineer