Thanks guys - the good news is that I can boot the PV DomU with your fixes:
title Fedora (PV boot, Xen kernel, 2952)
kernel /vmlinuz-2.6.21-2952.fc8xen ro root=LABEL=/ rhgb quiet
name = "fc8-32-001"
memory = 1024
vcpus = 1
vif = [ 'mac=00:0f:b5:ae:4f:2a, bridge=xenbr0' ]
disk = [ 'phy:/dev/VG_Guests/FC8-32-001,hda,w' ];
boot = "cd"
sdl = 1
vnc = 0
I got rid of the root line with no problems. I couldn't boot at all with
the original ramdisk (I got a "mount: could not find filesystem
'dev/root'" error), but had no problem with the xenblk and
The *bad* news is, well, bad. The DomU appears to be Ok at first sight,
but crashes when I run a specific program (wine, which is a
Windows-on-Unix emulator [yes, I know that's crazy, but I need it for
testing]). I get a kernel bug:
kernel BUG at include/linux/tracehook.h:369!
invalid opcode: 0000 [#1]
Should I bother reporting this? The Dom0 appears flakey after this so I
need to restart the entire machine.
I could live with this, by going back to the HVM guest, but this is
non-trivial. The PV boot messed up various things, and an HVM create
leaves me locked out of the DomU:
1 - The console doesn't work (but it never has on my HVM DomU's; I get a
"xenconsole: could not read tty from store" error, with nothing useful
in the qemu log)
2 - networking isn't started; I get the normal eth0 startup failed
message, but there's no subsequent attempt to startup eth1/2/etc
3 - The sdl window hangs up at the anacron boot message, which is normal
on the FC8 Dom0 (after FC8 messes up the serial console and inittab).
I'm slowly working my way through this lot, with some (Ok, major)
irritation. FC8 (or Xen? I don't know) replaced the DomU inittab with
one it obviously thought was better, which left the DomU in run level 3.
fixing that has helped. Network next....
Xen-users mailing list