Fatal Trap 12 problem after mergemaster crash

Josef Karthauser joe at tao.org.uk
Sun Jan 13 22:42:33 GMT 2002

Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

Ok, it looks like something has got corrupted in the filesystem.
My guess would be the linux.ko loadable kernel module.

What you should do is boot the system into single user mode, fsck
the disks, mount them and then do a 'make world' to freshen up the
binaries on the system.

Here's what you do.

At the boot loader prompt, the one that counts down in seconds, hit
any key to break into the load process and type 'boot -s'. This
will book the kernel but remain in single user mode. You should
end up with a request to enter the name of the shell that you want.
Hit carriage return and accept the default.

Next fsck the drives by running 'fsck -p' and wait until it's

Now mount the drives; 'mount -a' from the command line.

You should now be good for a 'cd /usr/src' followed by a 'make

Once it's finished, make and reinstall a kernel, the way that you
normally do it, and reboot the box in the normal way.

This should fix things up for you.  Of course you can probably get
away with just the kernel rebuild, or maybe just a kernel install
if you've already got a built kernel.  You milage may vary, but you
can always ask here if you get into trouble.


On Sat, Jan 12, 2002 at 07:24:51PM -0000, CB wrote:
> Yesterday I successfully did "make buildworld" and "make installworld" on=
 cvsup'd files to upgrade from FreeBSD 4.0 to 4.1 but haven't been able to =
boot with the upgradfed system yet.
> Following the guidance in the current manual #19.4.10 of "Using make worl=
d" I ran "mergemaster" to update files not updated by "make world".
> Sadly this crashed the machine - the symptom being a  screen full of what=
 looked like program source code, and any keypress just causing another lin=
e of code to scroll up onto the screen with an inverse (black text on white=
) byte count in the bottom left hand corner.
> I could not find any way to get out of this, and eventually power-cycled =
the box. (I probably should have been wary as the handbook guidance may hav=
e changed since version 4.0 to 4.1 upgrades)
> Now I can't get it to boot up properly. After the usual final bootup mess=
age before the login prompt, which is:
> Initial rc.i386 initialisation: linux
> launching the linux emulator, instead of the login prompt I get:
> Fatal trap 12: page fault while in kernel mode
> fault virtual address =3D 0x3
> fault code    =3D supervisor read, page not present
> instruction number  =3D 0x8:0xc0287e74
> stack pointer   =3D 0x10:0xc870798c
> frame pointer   =3D 0x10:0xc870798c
> code segment   =3D base 0x0, limit 0xfffff, type 0x1b
>       =3D DPL 0, pres 1, def32 1, gran 1
> processor eflags  =3D interrupt enabled, resume, IOPL =3D 0
> current process   =3D 167 (sh)
> interrupt mask   =3D none
> trap number    =3D 12
> panic: page fault
> syncing disks... 42 42 35 21 6
> done
> Uptime: 2m57s
> Automatic reboot in 15 seconds - press a key on the console to abort
> --
> It may not be helping that the system is effectively in mid-upgrade. Tryi=
ng to boot from a saved /kernel.prev gets to:
> mounting root from ufs:/dev/ad0s1a
> WARNING: / was not properly dismounted
> swapon: adding /dev/ad0s1b as swap device
> Automatic reboot in progress...
> /dev/ad0s1a: 1648 files ......
> it sticks there for long enough to allow me to transcribe that far, with =
the hard drive continuously active, then it continues the boot process befo=
re giving the Fatal Trap 12 message above. trying a boot with /kernel.GENER=
IC does the same thing, though I notice it is identified as 4.0 in the boot=
up script.
> Does anybody know what the just-made Version 4.1 kernel would be called a=
nd/or where it would be saved at this stage of the upgrade process? When fi=
rst rooted the system says in cannot find /kernel or /kernel.old and offers=
 me an "ok" prompt from where I can boot with a named kernel file.
> Much hoping I don't have to trash the system back to a total wipeout and =
> Many thanks for any help.
> Charles

Content-Type: application/pgp-signature
Content-Disposition: inline

Version: GnuPG v1.0.6 (FreeBSD)
Comment: For info see http://www.gnupg.org



More information about the Ukfreebsd mailing list