kern/70931: panic: page fault at end of boot on Athlon 64 / Asus K8V

Karl Swartz karl at
Thu Aug 26 09:40:59 PDT 2004

> > Will do, but I doubt that's related.  With the 4.10 twe driver bugs in
> > mind, I physically yanked the card from the system and tried booting
> > from floppies.  Same panic.
> Well, it looks like your panic happens, when accessing the disks, so i
> suspected the twe controller.

I suspected it as well, though the panic also happens when accessing the
MFS filesystem after booting from floppy.  With or without the twe card
in the system.  I thought that was convincing.  :-)

> Do you by chance have any other Disks available to try?

No, but I did try fiddling with memory sticks again.  One of them is
clearly bad -- with just it, the system won't even get through the BIOS,
and with the other one I've gotten all the way to a running system.

However, it's not out of the woods yet as I've seen several panics
during shutdown.  Here's one that I caught on paper:

   syncing disks... 1
   vm_page_free: pindex(57), busy(0), PG_BUSY(1), hold(0)
   panic: vm_page_free: freeing free page
   Uptime: 3m3s
   twe0: Cannot delete unit. error = 16

I'm going to run Memtest86 on the box overnight, and also try 5.3-BETA
for which I ran out of time last evening.

> > Do you have a preference for x86 or amd64?  I'll try both if I can.  I'm
> > also going to run more hardware diagnostics and re-check all of the BIOS
> > settings.
> There are several people including myself, who run FreeBSD-am64 on the
> K8V without problems. So if you don't have the need to run i386, give
> amd64 a try.

I'll give it a whirl.  However, this machine is destined to be my main
web and mail server so I want something very stable.  That's why I have
been planning to use 4.10 rather than 5.x, which in turn forces i386.
I mainly bought the K8V for lots of ECC memory, not because I need the
horsepower, though if 5.3(-BETA) is stable and the amd64 buys enough
performance it could be interesting for web site upgrades.  One of mine
( is quite busy and enough CPU could enable some interesting

More tonight.

 -- Karl

More information about the freebsd-bugs mailing list