Change in vdsm[master]: virt: add logic for POWER cpu xml element
danken at redhat.com
danken at redhat.com
Tue Jul 7 15:40:35 UTC 2015
Dan Kenigsberg has posted comments on this change.
Change subject: virt: add logic for POWER cpu xml element
......................................................................
Patch Set 4:
(2 comments)
https://gerrit.ovirt.org/#/c/42736/4//COMMIT_MSG
Commit Message:
Line 7: virt: add logic for POWER cpu xml element
Line 8:
Line 9: Current flow in appendCpu caused ppc to be skipped, generating empty
Line 10: element (<cpu/>). In order to specify cpu model on ppc, we have to use
Line 11: libvirt's host-model element and power8 model.
can you point to the libvirt-side discussion about this compulsory use of host-model?
At the very least, state the limitation that this puts on migration: migration from a new host to and older one can fail in unexpected ways inside the guest.
Line 12:
Line 13: This is currently the only way of safely adding CPU to ppc64le host.
Line 14:
Line 15: Change-Id: I9303b76904ef1344508136343104b37c09d2a370
https://gerrit.ovirt.org/#/c/42736/4/vdsm/virt/vmxml.py
File vdsm/virt/vmxml.py:
Line 366: elif feature[0] == '-':
Line 367: featureAttrs['policy'] = 'disable'
Line 368: cpu.appendChildWithArgs('feature', **featureAttrs)
Line 369: elif self.arch in (caps.Architecture.POWER,):
Line 370: cpu.setAttrs(mode='host-model')
I think that it is safer to send model=hostModel from Engine, instead of ignoring Engine's request in Vdsm.
In the future, when we have a more stable definition of guest CPU models, we can use them without a change.
Line 371:
Line 372: features = self.conf.get('cpuType', 'power8').split(',')
Line 373: model = features[0]
Line 374:
--
To view, visit https://gerrit.ovirt.org/42736
To unsubscribe, visit https://gerrit.ovirt.org/settings
Gerrit-MessageType: comment
Gerrit-Change-Id: I9303b76904ef1344508136343104b37c09d2a370
Gerrit-PatchSet: 4
Gerrit-Project: vdsm
Gerrit-Branch: master
Gerrit-Owner: Martin Polednik <mpolednik at redhat.com>
Gerrit-Reviewer: Dan Kenigsberg <danken at redhat.com>
Gerrit-Reviewer: Francesco Romani <fromani at redhat.com>
Gerrit-Reviewer: Jenkins CI
Gerrit-Reviewer: Martin Polednik <mpolednik at redhat.com>
Gerrit-Reviewer: Michal Skrivanek <mskrivan at redhat.com>
Gerrit-Reviewer: automation at ovirt.org
Gerrit-HasComments: Yes
More information about the vdsm-patches
mailing list