On 04/22/2015 05:17 AM, Daniel P. Berrange wrote: > On Tue, Apr 21, 2015 at 02:34:09PM -0600, Eric Blake wrote: >> On 04/21/2015 02:17 PM, Cole Robinson wrote: >>> --- >>> docs/schemas/capability.rng | 1 + >>> 1 file changed, 1 insertion(+) >>> >>> diff --git a/docs/schemas/capability.rng b/docs/schemas/capability.rng >>> index 5f3ec70..3868ee2 100644 >>> --- a/docs/schemas/capability.rng >>> +++ b/docs/schemas/capability.rng >>> @@ -266,6 +266,7 @@ >>> <value>hvm</value> <!-- unmodified OS --> >>> <value>exe</value> <!-- For container based virt --> >>> <value>uml</value> <!-- user mode linux --> >>> + <value>aix</value> <!-- used by phyp driver --> >> >> ACK > > Actually I don't think this is correct - I think it is another case > of the phyp driver being rather confused. > > The phyp driver doesn't use the def->os.type anywhere in its code as > far as I can tell. > > The place where 'aix' is used is bizarrely in the phypDomainAttachDevice > method where it calls virDomainDefNew() and then sets os.type to 'aix' > before calling virDomainDeviceDefParse. This is all just bizarre, but > the 'aix' type doesn't appear to ever leak out to the end user in the > XML unless I'm missing something. So I don't think we should add this > to the RNG. Indeed thanks for looking, I should have looked closer. I just sent a patch that drops AIX entirely and switches phyp to use OSTYPE_LINUX (?) which is what it advertises via capabilities - Cole -- libvir-list mailing list libvir-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/libvir-list