Subject: Re: uvm_fault
To: None <tech-kern@netbsd.org>
From: Marc Tooley <netbsdMLpostNO@spam.quake.ca>
List: current-users
Date: 02/21/2006 16:39:27
On Tuesday 21 February 2006 15:54, Malcolm Herbert wrote:
> On Tue, Feb 21, 2006 at 10:16:46PM +0100, mouss wrote:
>
> |>>uvm_fault(0xcace89a0, 0, 0, 1) -> 0xe
> |>>kernel: supervisor trap page fault, code = 0
> |>>Stopped in pid 543.1 (mountd) at 0: invalid address
>
> I have a similar setup with chroot and null mounts trying to compile
> pkgsrc stuff and got a number of these errors during my last build
> ... single CPU machine, NetBSD 3.0, RAIDFrame disks, 512M of memory
> ...
>
> Rather than type them out each time I just took a photo - they're at
> http://www.ergonaut.org/album/200601210000/

Mounting two union mounts overtop of cgd volumes (which themselves are 
built on top of just standard partitions) and then building a kernel in 
the union mount gives me a reliable kernel fault.

(gdb) target kcore netbsd.3.core
panic: lockmgr: locking against myself
#0  0x07df0000 in ?? ()
(gdb) bt
#0  0x07df0000 in ?? ()
#1  0xc02dfd63 in cpu_reboot ()
#2  0xc028d4ef in panic ()
#3  0xc0272bb5 in lockmgr ()
#4  0xc02b9aef in genfs_lock ()
#5  0xc02b8984 in VOP_LOCK ()
#6  0xc02b684e in vn_lock ()
#7  0xc01bc50d in union_dircache ()
#8  0xc01bc63a in union_readdirhook ()
#9  0xc02b6258 in vn_readdir ()
#10 0xc02b5948 in sys___getdents30 ()
#11 0xc02e6d3a in syscall_plain ()
(gdb) 

I was in the midst of building a debug kernel so I could (hopefully) 
capture these with a little more detail when I saw your note about null 
mounts. I wonder if our two problems are related?