On Mon, Mar 30, 2009 at 03:00:55PM +0100, Daniel P. Berrange wrote:
This patch implements PCI device passthrough for Xen.
In pci.c there is a fair bit of refactoring to better support both
pcistub.ko and pciback.ko. The latter has two extra sysfs files,
add_slot and remove_slot for associating the driver with a specific
PCI device domain:bus:slot:func address, whereas pcistub can only
associate based on vendor:product ID
The Xen driver gets an implementation of the dettach/reattach APIs
for node devices.
The XML <-> SEXPR conversion gets adapted to cope with PCI devices
The XM <-> XML conversion gets adapted to load/save PCI device info
in /etc/xen config files
Okay, I also wondered about the 2 modprobe commands test, their order
and what the Run() command return value < 0 really meant (man modprobe
doesn't document any return value at least on Fedora 9 !).
So I'm a bit worried about the portability of that specific test,
maybe something more appropriate than just the return value should be
used to detect failure.
Still this can't generate regression since it's a new feature, so
fine by me, ACK , but I'm still wondering a bit :-)
Daniel
--
Daniel Veillard | libxml Gnome XML XSLT toolkit
http://xmlsoft.org/
daniel(a)veillard.com | Rpmfind RPM search engine
http://rpmfind.net/
http://veillard.com/ | virtualization library
http://libvirt.org/