martin@duskware.de wrote:

> I have no clue if this could (negatively) affect the other booters - could
> someone test, please?

http://cvsweb.netbsd.org/bsdweb.cgi/src/sys/arch/sgimips/stand/common/boot.c#rev1.10

"under certain circumstances?"

I thought it had had some trouble to load kernels without symbol table,
but last time I tried it on IP32 I couldn't reproduce it...

Anyway, current bootinfo code of sgimips boot seems a bit strange.
IMHO, it's better to allocate bootinfo data in a single region
like pmax and arc so that bootinfo data would be preserved even
after boot.
---
Izumi Tsutsui