> From: Jason Thorpe <thorpej@nas.nasa.gov>
>
> On Sat, 11 Jul 1998 15:39:06 -0700 
>  "William J. Coldwell" <billc@warped.com> wrote:
>
>  > I exhibit this behavior on a Multia with -current.  It looks like
>  > we can track it down to something done right around the snapshot
>  > build.  It acts like the booter is jumping to a wrong address and
>  > just resets the machine.. over and over again :)
>
> I suspect this is related to the IDE hack Ross made for the axppci33
> systype.

Could be...perhaps there is something undocumented and explosive at
io port 0x26e or 26f.

Try putting a "return;" after line 121 in

	/sys/arch/alpha/alpha/dec_axppci_33.c

Or, feed this entire mail message (no need to edit out anything) to patch(1).

If that fixes it, you might try a different mod: first, restore to the
"bad" version but change NSIO_PORT on line 89 to the one proposed in the
comment on that line, and let me know what happens either way.

Index: dec_axppci_33.c
===================================================================
RCS file: /cvsroot/src/sys/arch/alpha/alpha/dec_axppci_33.c,v
retrieving revision 1.38
diff -u -w -r1.38 dec_axppci_33.c
--- dec_axppci_33.c	1998/07/07 08:49:12	1.38
+++ dec_axppci_33.c	1998/07/12 00:19:49
@@ -120,6 +120,8 @@
 	platform.cons_init = dec_axppci_33_cons_init;
 	platform.device_register = dec_axppci_33_device_register;
 
+	return;
+
 	lcp = lca_preinit();
 	iot = &lcp->lc_iot;
 	if (bus_space_map(iot, NSIO_PORT, NSIO_SIZE, 0, &nsio))