virtio: set low features early on load

virtio migrates the low 32 feature bits twice, the first copy is there
for compatibility but ever since
019a3edbb25f1571e876f8af1ce4c55412939e5d: ("virtio: make features 64bit
wide") it's ignored on load. This is wrong since virtio_net_load tests
self announcement and guest offloads before the second copy including
high feature bits is loaded.  This means that self announcement, control
vq and guest offloads are all broken after migration.

Fix it up by loading low feature bits: somewhat ugly since high and low
bits become out of sync temporarily, but seems unavoidable for
compatibility.  The right thing to do for new features is probably to
test the host features, anyway.

Fixes: 019a3edbb25f1571e876f8af1ce4c55412939e5d
    ("virtio: make features 64bit wide")
Cc: qemu-stable@nongnu.org
Reported-by: Robin Geuze <robing@transip.nl>
Tested-by: Robin Geuze <robing@transip.nl>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

diff --git a/hw/virtio/virtio.c b/hw/virtio/virtio.c
index 7ed06eafa687a01c3c0d8f7620e8fc6daf85954f..18153d5a3995e3a1c9bfad2a408dbca602a5ef18 100644 (file)
--- a/hw/virtio/virtio.c
+++ b/hw/virtio/virtio.c
@@ -1499,6 +1499,16 @@ int virtio_load(VirtIODevice *vdev, QEMUFile *f, int version_id)
     }
     qemu_get_be32s(f, &features);
 
+    /*
+     * Temporarily set guest_features low bits - needed by
+     * virtio net load code testing for VIRTIO_NET_F_CTRL_GUEST_OFFLOADS
+     * VIRTIO_NET_F_GUEST_ANNOUNCE and VIRTIO_NET_F_CTRL_VQ.
+     *
+     * Note: devices should always test host features in future - don't create
+     * new dependencies like this.
+     */
+    vdev->guest_features = features;
+
     config_len = qemu_get_be32(f);
 
     /*