Re: [PATCH v2] tools/libxl: don't allow IOMMU usage with PoD

[Anthony PERARD <anthony.perard@xxxxxxxxxx>]

On Wed, Feb 16, 2022 at 10:12:03AM +0100, Roger Pau Monné wrote:
> On Tue, Feb 15, 2022 at 03:08:21PM +0000, Anthony PERARD wrote:
> > On Thu, Feb 03, 2022 at 03:32:11PM +0100, Roger Pau Monne wrote:
> > >      if (d_config->c_info.type != LIBXL_DOMAIN_TYPE_PV &&
> > > -        d_config->num_pcidevs && pod_enabled) {
> > > +        d_config->c_info.passthrough != LIBXL_PASSTHROUGH_DISABLED &&
> > > +        pod_enabled) {
> > >          ret = ERROR_INVAL;
> > > -        LOGD(ERROR, domid,
> > > -             "PCI device assignment for HVM guest failed due to PoD 
> > > enabled");
> > > +        LOGD(ERROR, domid, "IOMMU not supported together with PoD");
> > 
> > I'm not sure that this new error message is going to be good enough to
> > point out configuration issue for the guest.
> > 
> > One is going to set 'pci=["foo"]' or 'dtdev=["bar"]', which will enable
> > passthrough. Then they may get en error about IOMMU or PoD.
> > Should we maybe write something like this instead?
> > 
> >    "IOMMU or device passthrough not supported together with PoD"
> 
> The "or" seems weird to me: IOMMU is mandatory for device passthrough.
> Maybe:
> 
> "IOMMU required for device passthrough but not support together with PoD"

                                                 ^ supported ?
> Would that be OK?

Sound good, with that new error message: Reviewed-by: Anthony PERARD 
<anthony.perard@xxxxxxxxxx>

Thanks,

-- 
Anthony PERARD