Why the mono_fdhandle_insert doesn't release the mutex lock after getting an error

[duduita]

While most of the methods in fdhandle.c release the mutex lock after getting an error, such as mono_fdhandle_try_insert, mono_fdhandle_lookup_and_ref, mono_fdhandle_close, etc... the mono_fdhandle_insert doesn't do it, so if for some reason I got a duplicated socket error, the mono_fdhandle_insert will fail calling the g_error without releasing the mutex lock. Consequently, after that I'm getting the error mono_os_mutex_trylock: pthread_mutex_trylock failed with "Error code '142'", since it seems that other threads cannot lock the mutex since it is already locked.

To summarize, I would like to understand if not releasing the mutex lock after getting an error in the mono_fdhandle_insert is intentional, in that case, why we cannot release the mutex as we do in the other functions. Since it's a core method, I guess that it should be made intentionally @luhenry.

[Xyncgas]

Maybe because after double failure no one is permitted to enter a crime scene

[duduita]

But, why do we use the mono_fdhandle_try_insert, which basically does the same as the mono_fdhandle_insert, but just releases the mutex lock in the case of failure, as can be seen on the w32file-unix.c. My naive question is why we cannot do the same on the w32socket.c.