On Dec 2, Matthew Jacob wrote
> 
> 
> Care to elaborate on this? The number of targets is a property of the bus.
> The number of luns supported (as opposed to available) is a property of
> the transport layer's (HBA's) ability to cope. In any given runstate post
> configuratin I would expect neither of these to change. I mean, you don't
> all of a sudden say "I feel like going from narrow to wide", or "I feel
> that I have too much work to do. I think I'll stop worrying about luns
> 2 through 31,". If these parameters change, then a whole reconfig of that
> branch of the tree has to occur, including rescan/reconfig of any target
> drivers. 
> 
> So- I must be misunderstanding you? Whaddya mean?

Maybe I was not very clear. Of course this won't change at runtime.
I was talking about the default: if it's null, assume it can support 8.
One day there will be more drivers supporting 16 targets than 8.

> 
> > In any case, having 2 different sementics for target and lun is not
> > a good thing.
> 
> Yes, I was queasy about that a bit, but I guess you convinced me to go the
> extra 3cm. for NetBSD on this.... :-)
> 

Good :)

--
Manuel Bouyer, LIP6, Universite Paris VI.           Manuel.Bouyer@lip6.fr
--