Subject: assigning IRQ's on ISA buses
To: None <tech-kern@NetBSD.ORG>
From: Michael C. Richardson <mcr@sandelman.ottawa.on.ca>
List: tech-kern
Date: 10/04/1997 14:18:24
-----BEGIN PGP SIGNED MESSAGE-----


  Thanks to those to pointed out the difference between multi and
region. It wasn't obvious from reading the comments in bus.h, but it
makes total sense now. Perry points out that there is now a man page,
if only I bothered to update my /usr/share/man/man9 from -current more
often. I should have known, and I did try to RTFM before asking... The
man page is very good too.  
  
  I have looked through man pages on this, but feel free to RTFM me
again :-)

  I would like to wildcard an IRQ on an ISA card. The card has a
software configurable IRQ, and no NVRAM, so the kernel config file
becomes the ultimate authority. Is there someway to have autoconf
assign my driver an IRQ? I realize that it might be in use on the ISA
bus by another (unconfigured, or not yet configured) device, or might
have been assigned by the BIOS to a PCI device (would the autoconf
know anything about this?). 

  In the meantime, I guess I have to hardcode the irq.

  I'm still looking for a decent explanation of the broken indirect
problem. From what I can see, if _BROKEN_INDIRECT is defined, then a
lot of drivers treat their void * match argument as a "struct device"
even though that isn't what is passed by isascan().

   :!mcr!:            |  Network security programming, currently
   Michael Richardson | on contract with SSH IPSEC (http://www.ssh.fi/)
 WWW: mcr@sandelman.ottawa.on.ca. PGP key available.


  

-----BEGIN PGP SIGNATURE-----
Version: 2.6.3ia
Charset: latin1
Comment: Processed by Mailcrypt 3.4, an Emacs/PGP interface

iQB1AwUBNDaIbqZpLyXYhL+BAQGdHQL+OqPDFld9IVPvUo588jPVG6IwwyhB9ieb
55+PJXH+MIr4xVBZprTeBOawyD067sVsWOipSi7nu7gWAYwUXS90z4LzC+SvCjZt
I7eJojAf5qo+fjEhQ0aLiN1oYsV5syfA
=h2bd
-----END PGP SIGNATURE-----