[interfacekit] Re: BScreen (again)

"Ingo Weinhold" <bonefish@xxxxxxxxxxxxxxx> wrote:
> On Mon, 28 Jul 2003 19:35:43 +0200 CEST "Axel Dörfler" <axeld@pinc-
> software.de> wrote:
> > Indeed, the other safe way to do this would be using delete_sem() :
> > -)
> Right. It's a bit unpractical in for this purpose, though. :-)

Yup :)

> > Perhaps we want to add a way to release all waiting threads from a 
> > semaphore to release_sem(), by settings a magic release count (-1?) 
> > or provide a special flag.
> Yep, that would be nice.

I'll try to remember it :)

> > We could also find out how R5 solves this, if it does.
> Feel free... ;-)

Since the vertical blank is generally generated as an interrupt, I 
would guess, they just release the semaphore in a spinlock protected 
section - there you can safely call get_sem_count().
Anyway, I think a solution that would also work in userland would be 
very nice.

Adios...
   Axel.

Other related posts: