User mode Linux Issues

dannystaple · 03-09-2007, 05:37 PM

I am using User Mode Linux with an Ubuntu Edgy Guest on an Ubuntu Edgy host. I am using a guest kernel built from sources - 2.6.21, with an ext3 root_fs sparse image created with dd and debootstrap, and a swap image that is not sparse.

My starting command line for the UML VM is the following:

Code:

./linux-2.6.21 con=/dev/null con0=null,fd:2 con1=fd:0,fd:1 ubd0=./root_fs ubd1=swap eth0=tuntap,,,192.168.1.3 mem=256M

My first issue is a lock-up. When I start up the UML instance (using the tuntap helper), it locks up for a few minutes just after the last arp line of the tuntap setup. During this lockup, it is using none of the host CPU, and there is no massive harddisk activity. Any ideas what it is waiting for?

The very next line when it stops waiting is:

Code:

* Filesystem type 'usbfs' is not supported. Skipping mount.

I am not sure if it is related or not. I am not an expert on init v or the kernel. I did contact the UML IRC community, but it seems a little quiet right now.

My second issue, which may, or may not be related to the first is that the second device there - ubd1 shows up in the kernel bootup, but swap is not mounted. No entry /dev/ubdb is created for it by udev (although ubda is created for the root_fs) but it does show a line in /proc/discstat.

If I use the uml_mconsole to configure the device post boot with "config ubd2=swap" then udev picks it up, the entry is made, and I can mount it with swapon with no problems. If I try ubd2 in the start command, the same issue occurs, only to manually add it in I would have to use ubd1 in the config command.
The one message I see about ubdb (the swap space when passed in as ubd0) is the following:

Code:

ubdb: unknown partition table

However, since I see the same message for the root_fs, I am also not convinced that this log message is relevant.

Swap is in the guests fstab with the following line -

Code:

/dev/ubdb       none            swap    sw              0       0

So if anyone can shed any light on the two problems, I would be pretty chuffed. I am working away at them myself anyway, and will post back if I get anywhere. Time to go read the udev manual...

dannystaple · 03-09-2007, 06:30 PM

It turns out that the two problems were actually related. The command line I was using had some stuff to redirect the ports for the xterms that usually pop up to represent the ttys. I found this by paring down the command line to the basics and seeing it all worked fine.

The command line that worked now is this:

Code:

./linux-2.6.21 con=null con0=null,fd:2 con1=fd:0,fd:1 ubd0=./root_fs ubd1=swap e
th0=tuntap,,,192.168.1.3 mem=256M

It appears that having con=/dev/null there instead of con=null (which is what it should have been) seems to have caused all this trouble. How odd.