Il 06/03/2012 15:53, Michael S. Tsirkin ha scritto: > > The guest must already be prepared to see SG_IO support > > disappear from under its feet, for example if migration > > refers to a block device on the source and file-based > > storage on the destination; or more likely, if the source > > kernel allows (gasp) SG_IO on a partition and the destination > > does not. So, we can migrate safely even if the source > > had VIRTIO_BLK_F_SCSI and the destination does not. > > > > Signed-off-by: Paolo Bonzini <pbonz...@redhat.com> > > My first reaction is you want a new non guest > visible flag to control whether SG_IO fails on host. > guest visible ones must be consistent across migration.
So scsi=off/on would control VIRTIO_BLK_F_SCSI, while the new flag would cause requests to fail. Then it's simpler to do the other way round. Make the "new non guest-visible flag" be scsi=on/off and set VIRTIO_BLK_F_SCSI unconditionally as you suggested first. Paolo
