panic: pmap_enter: attempted pmap_enter on 4MB page

[Barney Wolff]

On Fri, Oct 17, 2003 at 09:51:57AM -0700, Peter Wemm wrote:
> 
> AHA!  origpte being 0xd0d0d0d0 means that something really came unstuck
> because that is the fill pattern that userland malloc(3) uses.  The
> 4MB page thing is a red herring, it just happens that PG_PS (0x80) is
> a set bit in the fill pattern.
> 
> Now to find the source. :-(  Are you getting this repeatably?  If somebody
> is able to (relatively) easily provoke this, there are a few things I'd
> like to try (shotgun diagnostics, but its better than nothing).

I also have a dual-athlon system that gets this panic on current.
The problem shows up first as sig 10 & 11 on builds (world/kernel/ports)
and eventual panic on pmap_enter on 4mb page.  I KNOW it's not hardware,
because a -current kernel from 9/24/03 is rock-solid (hours of port
builds, make -j16 buildworld) and so is RELENG_5_1.  The current kernel
from yesterday did not even get through installworld without a sig11.

I'm pretty sure it's not ata, because my system has scsi disks, and only
an idle ata cdrom.

I have pentiumpro as my cputype, because I also use this system to
build for my laptop - is that a conceivable clue?  But for experiments
we can set it to anything.

It's not in production, so if you want to try stuff, I'm willing to
give you a login or take direction.  Right now it's running RELENG_5_1
but I also have yesterday's -head on there and a local cvs on another
system updated twice a day.

-- 
Barney Wolff         http://www.databus.com/bwresume.pdf
I'm available by contract or FT, in the NYC metro area or via the 'Net.