[libvirt] [PATCH] PCI passthrough docs, tests and minor change

Daniel Veillard veillard at redhat.com
Mon Jan 12 11:52:45 UTC 2009


On Mon, Jan 12, 2009 at 11:23:51AM +0000, Daniel P. Berrange wrote:
> On Mon, Jan 12, 2009 at 11:17:52AM +0000, Mark McLoughlin wrote:
> > On Mon, 2009-01-12 at 11:30 +0100, Daniel Veillard wrote:
> > 
> > > +      For PCI devices the element carries 3 attributes allowing to designate
> > > +      the device as can be found with the <code>lspci</code> command, the
> > > +      <code>bus</code> attribute allows the hexadecimal values 0 to ff, the
> > > +      <code>slot</code> attribute allows the hexadecimal values 0 to 1f, and
> > > +      the <code>function</code> attribute allows the hexadecimal values 0 to
> > > +      7.</dd></dl>
> > 
> > Looks good, but probably want to recommend "virsh nodedev-list" rather
> > than lspci?
> 
> Definitely - this was the primary motivation for adding the node device
> APIs 

  I agree that it allows to work with remote URIs, but ...
except it doesn't seems to indicate what devices are actually there.
IMHO the current PCI pasthrough limitation is that it doesn't
work by name or by capability of the device, so you have to go fishing
to get what you look for, lspci allows that.  Unless you go though
all the nodedev-list results and run virsh nodedev-dumpxml on each.
  Is there a way to dump in one command all informations about the
PCI devices on the host.
  Oh and a way to list the capabilities available, to avoid guessing.

Daniel

-- 
Daniel Veillard      | libxml Gnome XML XSLT toolkit  http://xmlsoft.org/
daniel at veillard.com  | Rpmfind RPM search engine http://rpmfind.net/
http://veillard.com/ | virtualization library  http://libvirt.org/




More information about the libvir-list mailing list