On Wed, Sep 17, 2008 at 09:52:08AM -0400, Ben Guthro wrote:
> I think the most likely place for exposing CPU flags would be in the
> capabilities XML format. We do in fact already expose 3 flags there,
> PAE, VMX and SVM.
This looks like all of the info that I need - I guess I overlooked this part of the code
prior to my message.
The problem that I see currently is that despite the capabilities string containing the
"host" tag, with this cpu info - it is up to each hypervisor driver to implement
the broadcasting of host features.
For example - Xen will report PAE,VMX,SVM - but my ovirt node running KVM currently only
reports the following:
<host>
<cpu>
<arch>x86_64</arch>
</cpu>
</host>
It seems to me that it might be useful for some sort of "node" info
driver, where we might be able to share code for hypervisor independent
info about the physical machine it is running on.
We already have a place for this useful reusable node info code - the
cunningly named src/nodeinfo.c :-)
We should not have to re-write the scanning of /proc/cpuinfo in every
hypervisor driver, IMHO.
Totally agreed , this is why we put the node stuff in a separate nodeinfo.c
file even though only QEMU driver currently uses it. Which reminds me that
I need to hook it into the LXC driver.
Daniel
--
|: Red Hat, Engineering, London -o-
http://people.redhat.com/berrange/ :|
|:
http://libvirt.org -o-
http://virt-manager.org -o-
http://ovirt.org :|
|:
http://autobuild.org -o-
http://search.cpan.org/~danberr/ :|
|: GnuPG: 7D3B9505 -o- F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 :|