Tom Ivar Helbekkmo wrote:

> Lord Isildur <mrfusion@umbar.vaxpower.org> writes:
> 
> > You're not using a CMD scsi-mscp adaptor are you? what kind of
> > problems are you having with using local disks?
> 
> 'fraid so.  It's a newer CMD this time, a 223, and trying to use it in
> the 4000/500 is just a total no-go, just like with the 220 that I've
> got in the KA650-based box that now runs 4.3BSD-Reno.
> 
> These controllers work with Ultrix, Reno, 2.11BSD (on PDP-11), VMS,
> and, until somewhere in the vicinity of a year ago, NetBSD.  I've
> studied the code, looking for changes between code I know worked and
> the current state of the drivers, and I can't see anything that should
> be able to cause the problems.
> 
> NetBSD can't even boot properly from CMD controllers, and the code
> involved in *that* is simple enough that I really thought I should be
> able to find something, but no...  Can it be a compiler problem?  A
> bug related to __volatile__ in the newer GCC, perhaps?  (Just a wild
> guess, of course, but it's the time of wild guesses by now.)

I don't really follow port-vax that closely - has vax switched to
ELF yet?  When I did the hp300 switch, there was one particular PROM
structure in the boot code where the padding differences between a.out
and ELF made a difference.  One thing to try is a simple program that
includes the right headers and prints the sizeof() of each structure
that you might be suspicious about, then compile with both a.out and ELF
toolchains and look for differences.

Simon.
--
Simon Burge                            <simonb@wasabisystems.com>
NetBSD CDs, Support and Service:    http://www.wasabisystems.com/