virtio: acknowledge all features before access

The feature negotiation was designed in a way that makes it possible for devices to know which config fields will be accessed by drivers. This is broken since commit 404123c2db79 ("virtio: allow drivers to validate features") with fallout in at least block and net. We have a partial work-around in commit 2f9a174f918e ("virtio: write back F_VERSION_1 before validate") which at least lets devices find out which format should config space have, but this is a partial fix: guests should not access config space without acknowledging features since otherwise we'll never be able to change the config space format. To fix, split finalize_features from virtio_finalize_features and call finalize_features with all feature bits before validation, and then - if validation changed any bits - once again after. Since virtio_finalize_features no longer writes out features rename it to virtio_features_ok - since that is what it does: checks that features are ok with the device. As a side effect, this also reduces the amount of hypervisor accesses - we now only acknowledge features once unless we are clearing any features when validating (which is uncommon). IRC I think that this was more or less always the intent in the spec but unfortunately the way the spec is worded does not say this explicitly, I plan to address this at the spec level, too.

MSMichael S. Tsirkin committed 4 years ago