[PATCH 9/9] score: Add and use _Objects_Release()

[Gedare Bloom]

On Thu, Jun 6, 2013 at 6:53 AM, Sebastian Huber
<sebastian.huber at embedded-brains.de> wrote:
> On 06/05/2013 05:22 PM, Sebastian Huber wrote:
>>
>> Add and use _Objects_Release_without_thread_dispatch().  These two
>> functions pair with the _Objects_Get() function.  This helps to
>> introduce object specific SMP locks to avoid lock contention.
>
>
> Maybe it is better to name it _Objects_Put().  So we have
>
> obj = _Objects_Get()
> do_something_with_it(obj)
> _Objects_Put(obj)
>
Yes I think "Put" is a better inverse operation to "Get". Though given
the following discussion, might it be better to say "Acquire/Release"?
Objects_Get already does a dispatch disable crit section, so the
semantic meaning probably makes sense.

> For SMP it is desirable to gradually introduce more fine grained locking.
> So we may add a SMP lock to Objects_Control and use new functions
> _Objects_Acquire() and _Objects_Release() to protect the object state.  For
> example a mutex obtain may use only the object specific lock if it is free
> and use additional locks if it has to perform a blocking operation which
> requires intervention of the scheduler.
>
Will there need to be new functions, or can "Get/Put" determine
whether the object needs to be locked/unlocked?

> --
> Sebastian Huber, embedded brains GmbH
>
> Address : Dornierstr. 4, D-82178 Puchheim, Germany
> Phone   : +49 89 189 47 41-16
> Fax     : +49 89 189 47 41-09
> E-Mail  : sebastian.huber at embedded-brains.de
> PGP     : Public key available on request.
>
> Diese Nachricht ist keine geschäftliche Mitteilung im Sinne des EHUG.
>
> _______________________________________________
> rtems-devel mailing list
> rtems-devel at rtems.org
> http://www.rtems.org/mailman/listinfo/rtems-devel