)
To: None <gavan@coolfactor.org>
From: Chris Budd <budd_iq80321@hotmail.com>
List: port-arm
Date: 01/23/2004 13:59:17
>From: Gavan Fantom <gavan@coolfactor.org>
>Date: Fri, 23 Jan 2004 13:53:28 +0000 (GMT)
>
> > Realizing that the version on the CD is old (NetBSD 1.6F (IQ80321) #0:
>Thu
> > Aug 22 17:18:41 EDT 2002), I grabbed the 1.6.1 release (NetBSD 1.6.1
> > (IQ80321) #0: Mon Apr 7 22:31:19 UTC 2003) from NetBSD.org. Now I do
>not
> > get the panic, but I still have problems with no console.
>
>1.6F is an (old) version of the trunk; 1.6.1 is cut from the release
>branch. You probably want to be looking at a later snapshot of current
>from releng.netbsd.org.
>
I have now tried:
NetBSD 1.6ZG (IQ80321) #0: Sun Jan 4 08:32:30 UTC 2004
NetBSD 1.6.2_RC4 (IQ80321) #0: Fri Jan 9 08:38:47 UTC 2004
1.6ZG did not get very far before it hit a different panic. 1.6.2_RC4 was
similiar to 1.6.1; no console, and no panic.
I have noticed in the Ethernet captures, that the iq80321 was looking for a
/sbin/mount_mfs file; it is not in the 1.6.1 package, nor the 1.6.2_RC4
package. If I blindly copy the mount_mfs from the 1.6F package, they get
the same type of panic.
Why is mount_mfs missing from base.tgz?
>init should be creating a memory filesystem on /dev and executing MAKEDEV.
>Perhaps you could verify whether this is in fact happening using ddb?
>
Not sure what ddb can tell me. I know that when mount_mfs is there, it
reads the MAKEDEV files, and probably creates the /dev/console in RAM, but
it hits the panic. I included the panic, ps, t, and t/t below.
If the mount_mfs file is not there, such as with 1.6.1 and 1.6.2_RC4, I
cannot get ddb since there is no console on which to send a break.
>Alternatively, you could use MAKEDEV to create the device files on your
>NFS server.
>
I tried this previously. If I manually make a /dev/console using "sh
MAKEDEV std; chmod 777 console" on the NFS server, then every thirty seconds
the iq80321 overwrites the /dev/console file on the NFS server with some
ASCII text:
Jan 23 20:29:15 init: can't get /dev/console for controlling terminal:
Inappropriate ioctl for device
My two big questions are:
1. Why isn't mount_mfs in the evbarm binary packages?
2. If I manually copy mount_mfs from 1.6F, then I assume the iq80321
attempts to build a mfs /dev/console based on the reads from the NFS server.
However, why do I get the panic? Could it be something with the PCI slot
in which this iq80321 lives?
I am too new, so the trace below does not mean much to me, but maybe someone
will recognize something.
Thanks,
Chris.
panic: Prefetch abort in non-USR mode (frame=0xc7b13d38 PC=0x00000008)
Stopped in pid 12 (sh) at cpu_Debugger+0x10: ldmdb r11, {r11,
r13,
r
15}
db>
db>
db> ps
PID PPID PGRP UID S FLAGS COMMAND
WAIT
>12 1 1 0 7 0x4004 sh
11 1 11 0 3 0x84 mount_mfs
mfsidl
9 0 0 0 3 0x20204 aiodoned
aiodone
8 0 0 0 3 0x20204 ioflush
syncer
7 0 0 0 3 0x20204 reaper
reaper
6 0 0 0 3 0x20204 pagedaemon
pgdaemo
5 0 0 0 3 0x20284 nfsio
nfsidl
4 0 0 0 3 0x20284 nfsio
nfsidl
3 0 0 0 3 0x20284 nfsio
nfsidl
2 0 0 0 3 0x20284 nfsio
nfsidl
1 0 1 0 3 0x4084 init
wait
0 -1 0 0 3 0x20204 swapper
schedul
db> t
cpu_Debugger+0xc
scp=0xc0314e28 rlv=0xc028b7ec (panic+0xdc)
rsp=0xc7b13ce8 rfp=0xc7b13cfc
panic+0x10
scp=0xc028b720 rlv=0xc031bc10 (prefetch_abort_handler+0xac)
rsp=0xc7b13d10 rfp=0xc7b13d34
prefetch_abort_handler+0xc
scp=0xc031bb70 rlv=0xc031b1dc (exception_exit)
rsp=0xc7b13d38 rfp=0x00000000
r8=0x00000000 r7=0x00000000
r6=0x00000000 r5=0x00000000 r4=0x00000000
db> t/t c
trace: pid 12 at 0xc7b13f80
0x2
scp=0x00000002 rlv=0x00000000 ( 0)
rsp=0x0000000d rfp=0xc7b13fb4[u]vm_fault(0xc03a46f0, fffff000, 1, 0)
->
e
Unhandled trap (frame = 0xc7b13a80)
Data abort: 'Translation fault (section)' status=005 address=fffffffa
PC=c031527
4
Stopped in pid 12 (sh) at db_stack_trace_p
rint+0x170: ldr r7, [r4, r2, lsl #2]
db>
_________________________________________________________________
Check out the coupons and bargains on MSN Offers!
http://shopping.msn.com/softcontent/softcontent.aspx?scmId=1418