[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Xen-users] PCI passthrough using xl instead of xm commands



Seeing the recomendation to migrate from xm to xl, I thought i'd give
it a go, I'm not concerned here with the functioning of the PCI
devices within the domU (which is giving me other problems) but just
with assigning the PCI devices to the domU.

My config file includes multiple PCI devices for passthrough, the
relevant section being

pci = [ '09:00.0', '09:01.0', '09:03.0', '02:00.0', '02:00.1',
'02:00.2', '02:00.3', '02:00.4', '02:00.5', '02:00.6', '02:00.7' ]

Fedora 16 provides xen-pciback as a module, so I unbind/bind the PCI
devices (09:*.0) and USB root devices (02:00.*) from their relevant
busses with a bash script

SLOTS=(0000:02:00.0 0000:02:00.2 0000:02:00.4 0000:02:00.6)
for i in ${SLOTS[@]}; do
  echo -n $i > /sys/bus/pci/drivers/ohci_hcd/unbind
done

SLOTS=(0000:02:00.1 0000:02:00.3 0000:02:00.5 0000:02:00.7)
for i in ${SLOTS[@]}; do
  echo -n $i > /sys/bus/pci/drivers/ehci_hcd/unbind
done

SLOTS=(0000:02:00.0 0000:02:00.1 0000:02:00.2 0000:02:00.3
0000:02:00.4 0000:02:00.5 0000:02:00.6 0000:02:00.7 0000:09:00.
0 0000:09:01.0 0000:09:03.0)
for i in ${SLOTS[@]}; do
  echo -n $i > /sys/bus/pci/drivers/pciback/new_slot
  echo -n $i > /sys/bus/pci/drivers/pciback/bind
done

Both "xm pci-list-assignable-devices" and "xl
pci-list-assignable-devices" show all eleven devices as available,
though the numerical ordering is different for them.

0000:09:00.0
0000:09:01.0
0000:09:03.0
0000:02:00.0
0000:02:00.1
0000:02:00.2
0000:02:00.3
0000:02:00.4
0000:02:00.5
0000:02:00.6
0000:02:00.7

Using "xm create" the domU is assigned all PCI devices, and "lspci"
within the domain sees them, after shutting down the domU, "xm list"
shows it gone cleanly

Name                                        ID   Mem VCPUs      State   Time(s)
Domain-0                                     0   512     4     r-----     73.1

But using "xl create" I receive

Parsing config file mythbe16.cfg
libxl: error: libxl_pci.c:754:libxl_device_pci_reset The kernel
doesn't support reset from sysfs for PCI device 0000:02:00.0
libxl: error: libxl_pci.c:754:libxl_device_pci_reset The kernel
doesn't support reset from sysfs for PCI device 0000:02:00.1
libxl: error: libxl_pci.c:754:libxl_device_pci_reset The kernel
doesn't support reset from sysfs for PCI device 0000:02:00.2
libxl: error: libxl_pci.c:754:libxl_device_pci_reset The kernel
doesn't support reset from sysfs for PCI device 0000:02:00.3
libxl: error: libxl_pci.c:754:libxl_device_pci_reset The kernel
doesn't support reset from sysfs for PCI device 0000:02:00.4
libxl: error: libxl_pci.c:754:libxl_device_pci_reset The kernel
doesn't support reset from sysfs for PCI device 0000:02:00.5
libxl: error: libxl_pci.c:754:libxl_device_pci_reset The kernel
doesn't support reset from sysfs for PCI device 0000:02:00.6
libxl: error: libxl_pci.c:754:libxl_device_pci_reset The kernel
doesn't support reset from sysfs for PCI device 0000:02:00.7

Running "lspci" in the domU shows only a single PCI device (09:00.00),
after shutting down the domain "xl list" shows it as some sort of
zombie domain

Name                                        ID   Mem VCPUs      State   Time(s)
Domain-0                                     0   512     4     r-----      72.8
(null)                                       2     0     1     --ps-d       7.0

Are there any new rules for PCI devices using xl?

Should I avoid mixing and matching xm and xl commands on the dom0?

_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-users


 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.