On Wed, Jun 24, 2020 at 03:55:59AM -0400, Michael S. Tsirkin wrote:
> On Fri, Jun 19, 2020 at 01:16:38PM +0100, Daniel P. Berrangé wrote:
> > On Fri, Jun 19, 2020 at 07:47:20AM -0400, Michael S. Tsirkin wrote:
> > > On Fri, Jun 19, 2020 at 07:46:14AM -0400, Michael S. Tsirkin wrote:
> > > > On Fri, Jun 19, 2020 at 11:12:45AM +0100, Daniel P. Berrangé wrote:
> > > > > On Fri, Jun 19, 2020 at 12:06:02PM +1000, David Gibson wrote:
> > > > > > The default behaviour for virtio devices is not to use the platforms normal
> > > > > > DMA paths, but instead to use the fact that it's running in a hypervisor
> > > > > > to directly access guest memory.  That doesn't work if the guest's memory
> > > > > > is protected from hypervisor access, such as with AMD's SEV or POWER's PEF.
> > > > > > 
> > > > > > So, if a host trust limitation mechanism is enabled, then apply the
> > > > > > iommu_platform=on option so it will go through normal DMA mechanisms.
> > > > > > Those will presumably have some way of marking memory as shared with the
> > > > > > hypervisor or hardware so that DMA will work.
> > > > > > 
> > > > > > Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
> > > > > > ---
> > > > > >  hw/core/machine.c | 11 +++++++++++
> > > > > >  1 file changed, 11 insertions(+)
> > > > > > 
> > > > > > diff --git a/hw/core/machine.c b/hw/core/machine.c
> > > > > > index a71792bc16..8dfc1bb3f8 100644
> > > > > > --- a/hw/core/machine.c
> > > > > > +++ b/hw/core/machine.c
> > > > > > @@ -28,6 +28,8 @@
> > > > > >  #include "hw/mem/nvdimm.h"
> > > > > >  #include "migration/vmstate.h"
> > > > > >  #include "exec/host-trust-limitation.h"
> > > > > > +#include "hw/virtio/virtio.h"
> > > > > > +#include "hw/virtio/virtio-pci.h"
> > > > > >  
> > > > > >  GlobalProperty hw_compat_5_0[] = {
> > > > > >      { "virtio-balloon-device", "page-poison", "false" },
> > > > > > @@ -1165,6 +1167,15 @@ void machine_run_board_init(MachineState *machine)
> > > > > >           * areas.
> > > > > >           */
> > > > > >          machine_set_mem_merge(OBJECT(machine), false, &error_abort);
> > > > > > +
> > > > > > +        /*
> > > > > > +         * Virtio devices can't count on directly accessing guest
> > > > > > +         * memory, so they need iommu_platform=on to use normal DMA
> > > > > > +         * mechanisms.  That requires disabling legacy virtio support
> > > > > > +         * for virtio pci devices
> > > > > > +         */
> > > > > > +        object_register_sugar_prop(TYPE_VIRTIO_PCI, "disable-legacy", "on");
> > > > > > +        object_register_sugar_prop(TYPE_VIRTIO_DEVICE, "iommu_platform", "on");
> > > > > >      }
> > > > > 
> > > > > Silently changing the user's request configuration like this is a bad idea.
> > > > > The "disable-legacy" option in particular is undesirable as that switches
> > > > > the device to virtio-1.0 only mode, which exposes a different PCI ID to
> > > > > the guest.
> > > > > 
> > > > > If some options are incompatible with encryption, then we should raise a
> > > > > fatal error at startup, so applications/admins are aware that their requested
> > > > > config is broken.
> > > >
> > > > Agreed - my suggestion is an on/off/auto property, auto value
> > > > changes automatically, on/off is validated.
> > > 
> > > In fact should we extend all bit properties to allow an auto value?
> > 
> > If "auto" was made the default that creates a similar headache, as to
> > preserve existing configuration semantics we expose to apps, libvirt
> > would need to find all the properties changed to use "auto" and manually
> > set them back to on/off explicitly.
> > 
> > Regards,
> > Daniel
> 
> It's QEMU's job to try and have more or less consistent semantics across
> versions. QEMU does not guarantee not to change any option defaults
> though.
> 
> My point is to add ability to differentiate between property values
> set by user and ones set by machine type for compatibility.

At which point are you looking to differentiate these?  The use of
sugar_prop() in my draft code accomplishes this already for the
purposes of resolving a final property value within qemu (an explicit
user set one takes precedence).

-- 
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson