I'm having some pretty ugly lockups with both the current xen/master and
xen/stable kernels in both Xen 3.4.2 and 4.0. These crashes are only
happening on one specific box. The specific hardware is:
mobo: Intel dp55kg
cpu: Intel Core i5 570 (quad-core)
ram: 4GB DDR3
nic: onboard e1000e
Everything (dom0, all domUs) is running 32bit. I'm not passing any PCI
cards through. Nothing fancy, here.
1. If dom0 receives multicast traffic from its domUs. This seems to only
happen if I have multiple domUs with at least one of them sending out
multicast traffic to an address that dom0 is listening on. Just write a
simple app to listen on a multicast socket and send traffic to that
channel on at least 1 domU, and dom0 kernel panics. The panic message
seems to always have 'invalid opcode: 0000', pid: 0, comm: swapper, EIP at
gnttab_copy_grant_pageon, with a call trace containing some variant of a
list of network related functions (net_tx_action, netif_page_release,
e1000_clean, etc).
2. If dom0 creates an XDMCP session to one of its domUs. This is really
strange. Even running w/out KMS using the plain old vesa driver, if you
turn off opaque window moving and drag a window around the dom0 kernel
panics. Also, if you let xscreensaver fade to black the dom0 kernel
panics.
Both these problems ONLY happen when the domUs in question are running on
the dom0 (ie, XDMCP into a domU hosted by a different dom0 works
perfectly). Also, both crashes happen regardless of how many vcpus dom0
has (ie, dom0_max_vcpus=1).
Any ideas?
-Mike
PS - Wouldn't you know, this particular box doesn't have any serial ports.
I'm overnighting a PCI serial cards so I can capture the panic messages
and send them to the list, too.
---
Michael D Labriola
Electric Boat
mlabriol@xxxxxxxx
401-848-8871 (desk)
401-848-8513 (lab)
401-316-9844 (cell)
_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel
|
Home