There is a new version of uBoot checked into SVN. Please follow the
instructions at http://casper.berkeley.edu/wiki/ROACH_kernel_uboot_update
for the update procedure. Note that you will need to reset your
environment and reprogram your eeprom for all the new changes to be
effected.
It updates a few things, but the primary change to the user is the USB
boot procedure. In future, ROACH will no longer boot off flash, then
look for a USB stick and pivot root. The user must explicitly select
the boot method (soloboot, usbboot, netboot) and update the initboot
environment variable appropriately to make it the default boot option.
In order to boot off USB now, you will need to copy the root
filesystem onto an EXT2 or 3 partition on a USB flash drive. It should
be the first partition on the drive. USB-powered HDDs generally don't
work because they draw too much power. Booting off USB is still a
little experimental. We have found that sometimes after a warm reset,
the USB won't mount the filesystem. Different models of USB stick
exhibit this problem to different degrees, but a cold-reboot usually
solves it. It is not a hardware fault, because uBoot can always read
it successfully. It appears to be a problems with the Linux driver.
Be sure to close all open files and do a clean shutdown (shutdown -h)
before powering down the board or you risk corrupting the filesystem.
Jason Manley
Digital Engineer
SKA-SA, KAT Pinelands office
Unit 12, Lonsdale Building
Lonsdale Road
Pinelands
7405
South Africa
Cell: +27 82 662 7726
Work: +27 21 531 7282
Fax: + 27 21 531 9761
On 25 May 2009, at 09:58, Jason Manley wrote:
I suspect this warning is due to the fact that we're pivoting root,
with runlevels and initscripts defined by the onboard flash's
filesystem, not the Debian root filesystem.
As outlined in my earlier email, I think we will be changing this to
use a stand-alone, native EXT2 filesystem on the USB stick. Then
ROACH will be a bog-standard Debian install with BORPH extensions.
We will try this today, and if it works as expected, upload the
necessary changes to CASPER SVN for you to try tomorrow.
Jason
On 25 May 2009, at 09:08, <[email protected]> <[email protected]>
wrote:
Hi Jason:
We have found what the problem is.
I have to use reboot command before I shutdown or reboot Roach. Or
the file system might be broken. I think this is why.
But why reboot could not work? I am told the runlevel could not
determined. Following information is given:
roach:~# reboot
WARNING: could not determine runlevel - doing soft reboot
(it's better to use shutdown instead of reboot from the command line)
shutdown: timeout opening/writing control channel /dev/initctl
init: timeout opening/writing control channel /dev/initctl
Thanks
Wan
-----Original Message-----
From: Jason Manley [mailto:[email protected]]
Sent: Monday, 25 May 2009 4:35 PM
To: Cheng, Wan (ATNF, Marsfield)
Cc: [email protected]
Subject: Re: [casper] Roach issue: Stale NFS file handle error is
reported
We're looking into this. I think we might change things to be a
little
less obscure; avoid the whole loopback filesystem and root pivot.
Perhaps boot natively off an EXT2/3 partition on the USB stick. We
boot our boards using NFS, so haven't seen this problem ourselves.
I will get back to you before tomorrow.
Jason
On 25 May 2009, at 02:16, <[email protected]> <[email protected]>
wrote:
Hi Jason:
Any idea?
Thanks
Wan
-----Original Message-----
From: Cheng, Wan (ATNF, Marsfield)
Sent: Friday, 22 May 2009 4:48 PM
To: 'Jason Manley'
Cc: [email protected]
Subject: RE: [casper] Roach issue: Stale NFS file handle error is
reported
Hi Jason:
Thanks for you great help. I reformat my USB stick and copy a fresh
Roach core into it.
Then I insert the following configuration into /etc/network/
interfaces with VI.
auto eth0
iface eth0 inet static
address 130.155.199.211
netmask 255.255.240.0
gateway 130.155.192.2
Then I reboot the Roach again.
The network could not be up automatically as normal. And I still
receive
ls: /etc/network/run/ifstate: Stale NFS file handle
When I run ls /etc/network/run/ifstate
The customised setting could be erased after I use a fresh core. But
this problem still there and repeatable.
And I don't think the file system is in RO because I can see my
configuration in interfaces file after rebooting roach.
I could not remember why the OS is crash. I will write it down when
it crashes again. I think it was crashed when I use
Apt-get install and apt-get -f install. I received no available disk
space error. I guess something in the Roach board flash are also
affected. But they could not be resumed by reprogramming the Uboot
and OS Image.
Any idea?
Thanks
Wan
-----Original Message-----
From: Jason Manley [mailto:[email protected]]
Sent: Friday, 22 May 2009 4:12 PM
To: Cheng, Wan (ATNF, Marsfield)
Cc: [email protected]
Subject: Re: [casper] Roach issue: Stale NFS file handle error is
reported
If your problem is with the filesystem, then this has nothing to do
with your flash (which hosts uboot, the kernel and busybox). You
will
need to reformat your USB stick and put a fresh copy of the "roach"
file on there. This will erase all your customised settings
though. I
suspect you've managed to corrupt your filesystem and it's now
mounting RO or something.
I am concerned about your statement of the OS crashing. Why did it
crash? Linux is pretty stable and doesn't usually crash easily. Is
this a problem with the ROACH, or was it due to a user-action?
FWIW, if you want to update your flash, see the step-by-step guide
here: http://casper.berkeley.edu/wiki/ROACH_kernel_uboot_update
Note that we are working on a new uBoot image as I write this email
and there will be an update released later today or on monday. The
primary change is to optimise the clock speeds around the PPC.
Jason
On 22 May 2009, at 08:01, <[email protected]> wrote:
Hi Jason:
Thanks for your great help. But even auot eth0 is added, the
network
still could not come up automatically.
And my USB file system is FAT32 R/W. I am not sure what the file
system is for the on board flash.
I agree with you that this should be a filesystem problem.
I think this problem is popped up only when the OS is crashed for a
few times. But this problem could not be fixed even I update the
Uboot and the image on the Roach board. Is there anyway I can erase
all the on board flash and reprogram them again?
Thanks
Wan
-----Original Message-----
From: Jason Manley [mailto:[email protected]]
Sent: Friday, 22 May 2009 3:52 PM
To: Cheng, Wan (ATNF, Marsfield)
Cc: [email protected]
Subject: Re: [casper] Roach issue: Stale NFS file handle error is
reported
This is complaining about the statefile (ifstate) not being
writeable.
This is probably a problem with your filesystem. Are you using an
NFS
filesystem? is it mounted R/W, or Read-only?
BTW, now that I've seen your interfaces file, I suspect the reason
your interface is not coming up automatically is because you've
left
out the "auto eth0" line...
auto eth0
iface eth0 inet static
address 192.168.1.36
netmask 255.255.255.0
gateway 192.168.1.1
Jason
On 22 May 2009, at 07:40, <[email protected]> <[email protected]>
wrote:
Hi:
In order to set a static IP address to roach Ethernet port, I
insert
the following configuration to the /etc/network/interfaces:
iface eth0 inet static
address 130.155.199.211
netmask 255.255.240.0
gateway 130.155.192.2
dns-nameservers 130.155.194.32 130.155.192.40
Then I
ifdown eth0
ifup eth0
So the Ethernet port works properly.
But when I restart the Roach, I could not run ifdown and ifup
command anymore. A failure is reported:
ifdown: failed to open statefile /etc/network/run/ifstate: Stale
NFS
file handl
Then I run ls /etc/network/run/ifstate, similar error is reported:
ls: /etc/network/run/ifstate: Stale NFS file handle
This is repeatable.
Anybody has any idea?
Thanks
Wan
