NFS root
Dunge
Sorry I'm not here to help Sim.One development, I'm just asking
questions and try to learn by doing it.
After a while turned off, I decided to gives this board another life
attempt. I found out the speed is way too slow and this probably have
to do with the SDCard rootfs causing it's process to continually take
about 70% of the CPU. So I followed advices, and tried to create a NFS
rootfs.
I first installed and configured tftpd and nfsd on my debian machine.
I found out nfs was not built into the kernel, so I followed the
"BootLoader" wiki guide to update the kernel with the uImage posted in
the other thread "working uImage 2.6.24.7 with CONFIG_ROOT_NFS
support" using tftp. It worked fine. I then set the needed u-boot
bootargs variable and managed to actually got the thing booting on my
NFS root.
IP-Config: Complete:
device=eth0, addr=192.168.0.136, mask=255.255.255.0,
gw=192.168.0.1,
host=192.168.0.136, domain=, nis-domain=(none),
bootserver=192.168.0.86, rootserver=192.168.0.86, rootpath=
Looking up port of RPC 100003/2 on 192.168.0.86
Looking uð port of RPC 100005/1 on 192.168.0.86
VFS: Mounted root (nfs filesystem).
Freeing init memory: 120K
INIT: version 2.86 booting
Now the problem might seems a bit stupid and simpler than what I
already did, but I don't know how to cleanly fix this. At the end of
the boot log, I get this:
Setting up networking....
Configuring network interfaces...SIOCSIFHWADDR: Device or resource
busy - you ma
y need to down the interface
Failed to bring up eth0.
done.
nfs: server 192.168.0.86 not responding, still trying
So, from what I understand, it try to load up network interface, who
are already up, and then get disconnected from the NFS server because
the interface is shut down. My guess would be to do something to
prevent this in me /etc/init.d/ in my nfsroot folder, but I just don't
know what exactly.
Thx
Dunge
"Device buzy" warning, but still disconnect from the nfs server. I
tried to find out which file cause the problem in /etc/rcS.d/ but
can't find the problem.
Martin Guy
> At the end of the boot log, I get this:
>
> Setting up networking....
> Configuring network interfaces...SIOCSIFHWADDR: Device or resource
> Failed to bring up eth0.
> done.
This is normal with standard debian on nfsroot. Debian normally runs
from a local rootfs and expects to bring the ethernet up as part of
the boot process, but the ethernet was already configured during the
kernel boot (and is being used for tehe rootfs, which is why it's
busy!). I get exactly this message too but it continues past this
point and works fine.
> nfs: server 192.168.0.86 not responding, still trying
This is not normal. Does it always fail in the same way at the same point?
M
Dunge
> This is not normal. Does it always fail in the same way at the same point?
Yes it does. As I said, I tried to remove files in /etc/rcS.d/ one by
one to find out exactly where it disconnect, but the result seems
strange. If I remove either S55bootmisc.sh or S55random it continue
further, print "entering boot level 2" and then disconnect. These
files don't seem to have anything related. If I remove the file before
(S46mountnfs-bootclean.sh and S99stop-bootlogd-single), it react as
usual and disconnect before entering boot level 2. I don't understand
where the problem could be, my u-boot config, my nfs server config or
my nfsroot filesystem?
Martin Guy
> As I said, I tried to remove files in /etc/rcS.d/ one by
Well, the nfsroot is definitely working at first.
You could try booting single-user by adding
init=/bin/sh
to the end of the bootargs setting. That will launch /bin/sh instead
of /sbin/init and you can try running stuff from there to see if you
can provoke an NFS error. That will run none of the /etc/init.d stuff
(until you "exec /sbin/init" or run the scripts yourself manually).
I mean, I'm running the same thing with no problems. I have on the
server (Debian, nfs-kernel-server) 10.0.0.1 in /etc/exports:
/tftpboot/sim24 10.0.0.124(rw,no_root_squash,no_subtree_check)
M