Re: nfs-root booting works for 2.2.* client, but not 2.4.19
On Sat, Sep 21, 2002 at 08:54:15PM -0300, Jorge L. deLyra wrote:
> This is a bit confusing. This is neither a fail-to-mount-root panic nor a
> cant-find-init panic, it's an Ooops, a processing error within the kernel.
Right. And fiddling some more, I actually get a little further:
> The partition check is before mounting the root and before init comes in.
> Here is the sequence from my box, with 2.4.19:
>
> -----------------------------------------------------------------------------
> ...
> Partition check:
> hda: hda1 hda2 hda3 hda4 < hda5 hda6 >
> hdb: hdb1
> hdc: hdc1
> hdd: hdd1 hdd2
> <yours crashes here, right?>
> NET4: Linux TCP/IP 1.0 for NET4.0
> IP Protocols: ICMP, UDP, TCP, IGMP
> IP: routing cache hash table of 2048 buckets, 16Kbytes
> TCP: Hash tables configured (established 16384 bind 16384)
> Linux IP multicast router 0.06 plus PIM-SM
> NET4: Unix domain sockets 1.0/SMP for Linux NET4.0.
> VFS: Mounted root (ext2 filesystem) readonly.
> <root mounted, init come in below>
I now crash right here. The last good lines are
VFS: Mounted root (nfs filesystem)
Freeing unused kernel memory: 76k freed
Undable to handle kernel paging request at vritual address 14d58d54
printing eip:
[ gooblehoo ... ]
Ans you're right -- that is exactly where 2.2.2* pass control to init. The
next line is 'Activating swap'. It starts to smell like a memory issue.
The 2.4.* kernel is only barely larger than the 2.2.* ones, 770k to 710k.
> -----------------------------------------------------------------------------
>
> Looks like the problem is with setting up the network. The message looks
I think I got over that.
> like a memory problem. Are you sure the kernel is guessing the correct
> amount of RAM? In any case, here is a method we use here which will allow
It wasn't, but specifying the (correct, and only) 32mb doesn't help. It
could still be a memory issues as I don't have a swap partition or file
on the thin client. But then 2.2.2(0|1) work...
I also see the authenticated nfs request in the server's log.
> you to cut-and-paste the kernel boot messages into an email message. It is
> a bit complicated but usefull for a lot of things: use a serial console.
>
> 1) Enable the serial console on the kernel, boot with the parameter
> console=ttyS0,9600 (or some other speed that works for you).
>
> 2) Have available some other machine with X11 running. Run seyon in an X11
> session, attached to some serial port. Configure seyon for the correct
> speed, 1 bit, no parity, CR translations, etc. You can test this using
> it as a terminal in some working machine where you put a getty in one
> of the serial ports. You have to change /etc/inittab for this:
>
> # Example of how to put a getty on a serial line (for a terminal)
> T0:23:respawn:/sbin/getty -L ttyS0 9600 vt100
>
> 3) Build a simple 3-wire crossed serial cable and interconnect the two
> ports. Boot the node, you will be able to see the whole kernel boot
> procedure within you X11 seyon window. Just cut and paste...
I used to have one of those, but I don't think I still do...
Thanks for your help.
Dirk
> Note that your X11 session could be anywhere, not necessarily in the
> console of the machine with the serial port connected to the node. You log
> into the machine via the network and open the seyon window anywhere. Lilo
> and Etherboot also can be configured to use the serial port. You can use
> seyon as a terminal, reboot a node in the server root and look at the
> whole boot process (except the node's hardware cycling, of course) from
> the confort of your office.
> Cheers,
>
> ----------------------------------------------------------------
> Jorge L. deLyra, Associate Professor of Physics
> The University of Sao Paulo, IFUSP-DFMA
> For more information: finger delyra@latt.if.usp.br
> ----------------------------------------------------------------
>
--
Good judgement comes from experience; experience comes from bad judgement.
-- Fred Brooks
Reply to: