AW: [En-Nut-Discussion] Again a thread question

[Ole Reinhardt]

Hi,

> I'd vote against a more advanced scheduling
> algorithm, because this costs additional time
> where it hurts most: In the context switching.

Ok, no problem. Id was even more an academic question than a real need
of mine.

> Your worker thread may do things later, where
> it has to wait for something. If not, we may
> add to the endless discussion about cooperative
> and preemptive scheduling. In my view, there is
> _finally_ not much difference between a timed
> scheduler and using NutSleep().

Right. Moth things are working quite well with coop scheduling.

> API. On the other hand I don't see much difficulties
> in adding something like
> 
> NutThreadMurder(NUTTHREADINFO *td)

Ok, let me sleep some nights over this problem. Perhaps I'll implement
something like your NutThreadMurder... Perhaps one could also implement
an asynchronous signal handler. With that kind of mechanism a thread
could get a signal to free all it's allocated memory before killing him.

Regards,

Ole
-- 
kernel concepts    Tel: +49-271-771091-14
Dreisbachstr. 24   Fax: +49-271-771091-19
D-57250 Netphen    E+ : +49-177-7420433
--