From owner-freebsd-current  Tue Jan  7 20:57:26 1997
Return-Path: <owner-current>
Received: (from root@localhost)
          by freefall.freebsd.org (8.8.4/8.8.4) id UAA12467
          for current-outgoing; Tue, 7 Jan 1997 20:57:26 -0800 (PST)
Received: from rocky.mt.sri.com (rocky.mt.sri.com [206.127.76.100])
          by freefall.freebsd.org (8.8.4/8.8.4) with ESMTP id UAA12459
          for <current@freebsd.org>; Tue, 7 Jan 1997 20:57:24 -0800 (PST)
Received: (from nate@localhost) by rocky.mt.sri.com (8.7.5/8.7.3) id VAA29298; Tue, 7 Jan 1997 21:57:05 -0700 (MST)
Date: Tue, 7 Jan 1997 21:57:05 -0700 (MST)
Message-Id: <199701080457.VAA29298@rocky.mt.sri.com>
From: Nate Williams <nate@mt.sri.com>
To: Bruce Evans <bde@zeta.org.au>
Cc: current@freebsd.org, nate@mt.sri.com
Subject: Re: Kernel driver advice
In-Reply-To: <199701080202.NAA13412@godzilla.zeta.org.au>
References: <199701080202.NAA13412@godzilla.zeta.org.au>
Sender: owner-current@freebsd.org
X-Loop: FreeBSD.org
Precedence: bulk

> >Is there a better/different way of registering the need for an interrupt
> >and *NOT* being an ISA device?  How do the PCI devices grab an
> >interrupt?
> 
> No.  The PCI devices just grab an interrupt.  They are initialized
> before ISA devices, so this sort of works.  However, the ISA conflict
> checking doesn't know about resources grabbed by PCI devices.  If an
> ISA probe succeeds, then isa.c attempts to grab the interrupt.  If the
> interrupt is already allocated, then the allocation isn't changed and
> the error code is ignored, leaving the ISA driver unattached from the
> interrupt.

This isn't much of a solution, since it's the current way things are.
The PCIC controller grabs interrupt 3, and when sio1 tries to use it
fails (w/out the user knowing it).

> >Finally, is there a way to request the list of used/unused IRQ's in the
> >system at a point in time?  I'd like to be able to check if a particular
> 
> Attempt to allocate all IRQs and put back the ones that you get but don't
> want.

*laugh* 

Who keeps track of allocated interrupts?  Would it be possible/useful to
add a 'give me what has already been allocated' kind of function?
Instead of alloc_intr() something obvious like intr_alloced()? *grin*



Nate