Reproduceable kernel panic on boot

[Terry Lambert]

"Cliff L. Biffle" wrote:
> Fatal trap 12: page fault while in kernel mode
> fault virtual address   = 0x0
> fault code              = supervisor write, page not present
> instruction pointer     = 0x8:0xc1c1bd75

NULL pointer dereference.  You probably want to post a traceback.
If you have a debug kernel available, you probably want to gdb -k
it, and then look at the code at 0xc1c1bd75; that will give you
the exact line with the error.


> I'd like to attach my kernel configuration, but have no working kernel on the
> machine.  Yay.  It's basically GENERIC configured for I586 with at least the
> following added:
> options GEOM_AES, GEOM_BDE, GEOM_BSD, GEOM_GPT, GEOM_MBR, GEOM_VOL
> options PSM_RESETAFTERSUSPEND
> options CPU_SUSP_HLT, CPU_WT_ALLOC, NO_F00F_HACK, NO_MEMORY_HOLE
> 
> The machine is a K6-2; I've uncommented the CFLAGS and COPTFLAGS in make.conf
> but left them unchanged, and set cputype to k6-2.

It's generally a bad idea to set the CPU type (see the recent
discussion on -current about the guy who shot his foot off).
In general, GCC tends to generate buggy code if you set the
CPU type; of course, if you can live with buggy code...

> 
> While I am using ccache, I've cleared and reset my caches and the problem
> still occurs.
> 
> Please, help!  Now I get to see if I remember how to revert the entire tree to
> a couple days ago. :-)

Try also DISABLE_PSE and DISABLE_PG_G.  AMD K6's are known to
have the same CPU bug as Intel Xeons, P3's, and P4's.  This is
probably not what's happening in this case, as that's generally
not a NULL pointer dereference problem, but it could have been
in the page table code (to know for sure, you would have to post
a traceback).

-- Terry