In message <20020923175938.400A3687A@void.crufty.net>, "Simon J. Gerraty" write
s:
>Mostly FYI, but if anyone has any explaination that would be cool.

[ Kernel build with -mcpu=v8 gets this: ]

>Boot device: /iommu/sbus/espdma@4,8400000/esp@4,8800000/
>sd@3,0   File and args: netbsd.n -s
>>> NetBSD/sparc Secondary Boot, Revision 1.12
>>> (autobuild@tgm.daemon.org, Mon Sep  9 08:12:32 UTC 2002)
>Booting netbsd.n
>4477624Data Access Exception
>ok boot netbsd
>Resetting ... 

Same thing with my SparcLX, but I was using -msupersparc -O2 I was
originally told "get new boot blocks" and did that when my 2 year
old blocks wouldn't load my new kernel.  (How did I get 1.5.3_ALPHA
loaded with blocks that didn't support ELF, and why did 1.5.3 itself
break?)

Well, my bootblocks are the latest as of a couple of weeks ago,
and it's still happening with my -msupersparc built kernels.  So
Simon, no -msupersparc will not do any better, at least not by
itself without looking into compiler versions, other flags, etc.
also.  I too, am curious about this, and how much it might improve
the performance if at all.

Jon