Re: [libvirt] [PATCH v3 09/28] qemu: Probe "max" CPU model in TCG
On Thu, Feb 23, 2017 at 20:07:00 +0100, Pavel Hrdina wrote:
On Thu, Feb 23, 2017 at 03:15:07PM +0100, Jiri Denemark wrote:
> Querying "host" CPU model expansion only makes sense for KVM. QEMU 2.9.0
> introduces a new "max" CPU model which can be used to ask QEMU what the
> best CPU it can provide to a TCG domain is.
...
> static int
> virQEMUCapsInitCPUModelS390(virQEMUCapsPtr qemuCaps,
> + virDomainVirtType type,
> virCPUDefPtr cpu)
> {
> - qemuMonitorCPUModelInfoPtr modelInfo = qemuCaps->hostCPUModelInfo;
> + qemuMonitorCPUModelInfoPtr modelInfo;
> size_t i;
>
> - if (!modelInfo) {
> + if (type != VIR_DOMAIN_VIRT_KVM)
> + return -1;
It would be nice to have some error message in this case as well. Before
this patch the following error would be used because the "hostCPUModelInfo"
would be empty for TCG.
ACK with that fixed
Pavel
> +
> + if (!(modelInfo = qemuCaps->kvmCPUModelInfo)) {
> virReportError(VIR_ERR_INTERNAL_ERROR,
> _("missing host CPU model info from QEMU capabilities
"
> "for binary %s"),
Hmm, the code should be a bit different:
if (type == VIR_DOMAIN_VIRT_KVM)
modelInfo = qemuCaps->kvmCPUModelInfo;
else
modelInfo = qemuCaps->tcgCPUModelInfo;
if (!modelInfo) {
virReportError(VIR_ERR_INTERNAL_ERROR,
_("missing host CPU model info from QEMU capabilities "
"for binary %s"),
qemuCaps->binary);
return -1;
}
There's no reason to ignore tcgCPUModelInfo in case QEMU implements
support for "max" CPU model for TCG.
Jirka