Re: Suspicious reentrant semaphore

["Dmitry A. Kazakov" <mailbox@dmitry-kazakov.de>]

Jano wrote:

> I'm developing a multitasking program that runs OK for several hours. At
> some indeterminate moment, a bunch of its tasks cease to respond, and I
> have a strong suspicion that the culprit is a semaphore package I've
> done.
> 
> The purpose is to have a semaphore which can be called more than a time
> _by the same task_, and that must be released the same amount of times.
> Nested calls, simply.
> 
> Leaving aside why I have done that (yes, poor design, I'm removing it
> anyway ;-), I would like to know if my package is right or really is
> there some race condition. Please comment on it. I know that
> Current_Task shouldn't be called inside entries (in fact Gnat doesn't
> allow it, I tried :-P) but doesn't complain about it being used in
> barriers. Anyways, I have a (maybe wrong) feel about that being related
> to the deadlock.

The requeue statement does the trick in case you want to use a parameter 
value in a barrier. You could add one more private entry and requeue to it 
if the task is not the owner:

entry P (Owner: Task_id) when True is
begin
   if Caller = Owner then
      In_use := In_use + 1;
   else
      requeue Alien_P with abort;
   end if;
end P;

entry Alien_P (Owner: Task_id) when In_use = 0 is
begin
   Caller := Owner;
   In_use := 1;
end Alien_P;

-- 
Regards,
Dmitry A. Kazakov
www.dmitry-kazakov.de