February 2011 - openSUSE Virtual - openSUSE Mailing Lists

[opensuse-virtual] where to troubleshoot? drives attached to a DomU's passed-thru PCI card 'vanish' after problem in Guest (e.g., OOPS ...); *system* reboot seems required
by dev001x 25 Feb '11

25 Feb '11

hi i'm testing an 11.3 Dom0 with, rpm -qa | egrep -i "xen-4|kernel-xen" kernel-xen-2.6.37.1-4.1.x86_64 xen-4.0.2_02-102.1.x86_64 i've a DomU, also 11.3, same kernel version. a sata_sil24 card/driver have been hidden from Dom0, and i'm passing it through to the DomU, as I always have, with, pci = [ '05:06.0' ] extra = 'textmode=1 xencons=xvc0 noirqdebug elevator=noop swiotlb=128,force' at 1st Guest start, PCI appears, drives are seen, array can be started, etc. i.e., all's well. @ any 'problem' in DomU, e.g. a kernel OOPS, I shutdown/restart the guest -- and the PCI card appears in the guest, but NOT the drives. simply not there -- and afaict no error in dmesg. shutting down the guest, restarting xend @ Dom0, and restarting the guest is no help. card's passed through -- just no drives. restarting the Dom0 (system reboot), however, resets something -- and the next launch of Guest, all's back to normal -- the drives are, again, seen & addressable in the Guest. as the card, and attached drives, are hidden from the Dom0, i'd suspect that the DomU "handles" the finding of the card-attached drives, but clearly a restart of the the Guest is not sufficient to re-init that process. so, 1st question -- *where* might I begin to look for this problem? thanks. -- To unsubscribe, e-mail: opensuse-virtual+unsubscribe(a)opensuse.org For additional commands, e-mail: opensuse-virtual+help(a)opensuse.org

2 2

[opensuse-virtual] Re: [Xen-devel] 2.6.38-rc5-11-xen kernel oops when destroy DomU
by Jan Beulich 23 Feb '11

23 Feb '11

>>> On 23.02.11 at 03:15, excited man rupright <jasonsmr(a)hotmail.com> wrote: > Hello this is what happens I create a xen domU using ether of two methods xm > create or virt-manager, then destroy using ether of two methods (or shutdown > in the case of virt-manager. > > I revive a kernel oops upon destroying the DomU. > > SYSTEM > > openSUSE_Factory/openSUSE-11.4 Definitely the wrong list to report this on - opensuse-virtual(a)opensuse.org would be much better. > Feb 21 22:14:56 fisban kernel: [ 1131.615675] BUG: unable to handle kernel > paging request at 000060ffb0e5f9f4 > Feb 21 22:14:56 fisban kernel: [ 1131.616021] IP: [<ffffffff802f30fe>] > unbind_from_irq+0x11e/0x220 I fixed something in that area recently, but this didn't make it to the shared kernel repo yet. And you weren't told to use 2.6.38-rc on 11.4 anyway... Jan -- To unsubscribe, e-mail: opensuse-virtual+unsubscribe(a)opensuse.org For additional commands, e-mail: opensuse-virtual+help(a)opensuse.org

1 0

[opensuse-virtual] kernel-xen-conditional code in /etc/init.d startup scripts?
by dev001x 09 Feb '11

09 Feb '11

hi, what's the correct/recommended way to wrap conditional code in opensuse init.d scripts to exec on a -xen boot? e.g., i need to add some module exec stmts to /etc/init.d/boot.local that are specific to kernel-xen, that cause errors/warnings in boot logs -- an an occasional hang -- if they are exec'd at boot to a NON-xen (e.g., kernel-default) kernel. so, i want to check for status = {booted to kernel-xen}, and exec only if matched. should one check for current kernel via uname? something in /sys/... of /proc/... ? i understand there are many possible checks; the question is - which, if any, is the 'right' one in this case? thanks. -- To unsubscribe, e-mail: opensuse-virtual+unsubscribe(a)opensuse.org For additional commands, e-mail: opensuse-virtual+help(a)opensuse.org

3 3