Re: fast data access mmu miss after harddisk upgrade
On Sat, 23 Apr 2005 23:11:25 -0700 (PDT)
<foo_bar_baz_boo-deb@yahoo.com> wrote:
> --- Jurij Smakov <jurij@wooyd.org> wrote:
> > This is a pretty old heisenbug, which cannot be reproduced reliably
> > (except on SunFire 280R). It has been reported numerous times before,
> > both
> > on this list and in BTS. See bugs #267428, #299043, #301503 and
> > probably
> > others.
>
> Can anybody tell me what this error message means on a technical
> hardware level? I am not comprehending it and googling it did not lead
> me to anything either.
It means that the kernel either:
1) dereferenced a bad pointer, such as a NULL or something
2) or it didn't remap the kernel properly into the TLBs
right before taking over the trap table from the firwmare
Sometimes adding "-p" to the kernel boot command line can
get you come debugging messages, but not always.
It is a low level crash, and usually the most useful
piece of information you can provide the developers is
the output of ".registers" from the OBP "ok" prompt
when this kind of crash happens. The PC value dumped
in particular will say where in SILO or the kernel that
the crash occurs.
Reply to: