In message <20060320233030.GA10353@vaasje.org>Frank van der Linden writes
>Hi folks,
>
>There have been some issues with what looks like interrupt routing with
>NetBSD/amd64 and some motherboards, most notably nforce4.

[... diuscussion of problem, approach to fixing...]

>The files are in
>	ftp://ftp.netbsd.org/pub/NetBSD/misc/fvdl/acpi
>		/i386
>			/netbsd.i386.acpi.bz2
>			/netbsd.i386.acpilap.bz2
>			/netbsd.i386.acpimp.bz2
>		/amd64
>			/netbsd.amd64.acpi.bz2
>			/netbsd.amd64.acpimp.bz2

Hi Frank,

wow, thanks very much!

Over the next day or two, I promise to try these on a variety of
(single-processor, single-dual-core) socket-939 motherboards, and if
time permits, also on various socket-940 (single-processor,
dual-single-processor) which have exhibited slightly different
problems than with the socket-939 machines.

BTW, I can confirm that stock FreeBSD-6.0/amd64 kernels use ACPI to
configure PCI/PCI-express devices, and don't seem subject to the
various known problems in Netbsd-3/netbsd-current with interrupts and
nforce4, on any of the machines i've tried. So there's grounds to hope
your ACPI-based kernels will be a robust, complete fix.


Last: assuming you've come up with a good fix, what are the chances of
pulling the final fixes up to the netbsd-3 branch?

thanks,
--Jonathan