Knowledge Base

I have a functionality where two threads use a common resource. In a particular scenario in this, a thread locks a mutex and terminates when it is hit a signal while other thread continues to starve on this mutex indefinitely. I am trying to find a solution for this.

As a workaround, I have set a flag on locking the mutex and reset the flag when the mutex is unlocked. On the signal handler code, I have checked whether the flag is set. If so, I unlocked the mutex and exit from the thread. I think it is not a better solution.

I have read about robust mutex in which if the thread owning the mutex terminates with it locked, any other thread that is trying to lock it will unblock with error code EOWNERDEAD. By properly handling this error, the mutex deadlock problem can be solved. But I think it is not supported as per POSIX standard. Does QNX provide any posix extension to support similar type of mutex?

I searched through QNX APIs for mutex and didn't find any solution except the timed lock. But I am not sure if locking the mutex in every situation with a timeout is a better idea.

Recommendationfirst use
http://www.qnx.com/developers/docs/6.4.1/neutrino/lib_ref/s/syncmutexevent.html
[Description:

The SyncMutexEvent() is a kernel call that attaches a specified event to a mutex pointed to by sync. You use SyncMutexRevive() to revive a DEAD mutex. Normally, a mutex will be placed in the DEAD state when the memory that was used to lock the mutex gets unmapped. One of the ways this may happen is when a process dies while holding the mutex in a shared memory. ]

Get the event and revive the mutex.

There is a seperate sample for SyncMutexRevive() here in our knowledgebase: