network ioctl overhaul; testers?

[David Young <dyoung%pobox.com@localhost>]

I'm overhauling network ioctls.  My changes are mainly mechanical, but
they touch a lot of network code, more than I can test alone.  agr(4),
carp(4), ppp(4), slip(4), vlan(4), ISDN, ISO, AppleTalk need attention
from their users.

I hope that I can get some help with the testing.  I have built mac68k,
macppc, i386, and sparc64 GENERIC kernels:

amd64   ftp://cuw.ojctech.com/users/amd64-cf03f83c/netbsd
i386    ftp://cuw.ojctech.com/users/i386-1b4aa010/netbsd
mac68k  ftp://cuw.ojctech.com/users/mac68k-2f6a9af1/netbsd
macppc  ftp://cuw.ojctech.com/users/macppc-bf653ece/netbsd
sparc64 ftp://cuw.ojctech.com/users/sparc64-1e7a4d43/netbsd

If you need some other kernel, please place an order. :-)

Patches are here,
ftp://cuw.ojctech.com/users/patches-d2821638/ifioctl.patch

With my overhaul, I'm aiming to produce a regular hierarchy
of inheritance: drivers inherit/override ioctls in their class
(e.g., ether_ioctl).  Classes inherit/override the generic ioctls
(ifioctl_common) and the protocol ioctls, (*pr_usrreq)(... PRU_CONTROL
...).

Future changes will provide a generic way for pseudo-interfaces that
augment and aggregate other pseudo-interfaces (e.g., agr, bridge, carp,
vlan) to filter and modify the member interfaces' ioctls, since NetBSD
sorely needs that.

(Mixed with today's network ioctl patches are some that improve the
link-layer address assignment.  The latter patches depend on the former.)

Dave

-- 
David Young             OJC Technologies
dyoung%ojctech.com@localhost      Urbana, IL * (217) 278-3933 ext 24