virtio: harsher barriers for rpmsg.

We were cheating with our barriers; using the smp ones rather than the
real device ones.  That was fine, until rpmsg came along, which is
used to talk to a real device (a non-SMP CPU).

Unfortunately, just putting back the real barriers (reverting
d57ed95d) causes a performance regression on virtio-pci.  In
particular, Amos reports netbench's TCP_RR over virtio_net CPU
utilization increased up to 35% while throughput went down by up to
14%.

By comparison, this branch is in the noise.

Reference: https://lkml.org/lkml/2011/12/11/22

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

1 files changed, 5 insertions, 3 deletions

diff --git a/drivers/lguest/lguest_device.c b/drivers/lguest/lguest_device.c
index 595d73197016..6a1d6447b864 100644
--- a/drivers/lguest/lguest_device.c
+++ b/drivers/lguest/lguest_device.c
@@ -292,10 +292,12 @@ static struct virtqueue *lg_find_vq(struct virtio_device *vdev,
 
         /*
          * OK, tell virtio_ring.c to set up a virtqueue now we know its size
-         * and we've got a pointer to its pages.
+         * and we've got a pointer to its pages.  Note that we set weak_barriers
+         * to 'true': the host just a(nother) SMP CPU, so we only need inter-cpu
+         * barriers.
          */
-        vq = vring_new_virtqueue(lvq->config.num, LGUEST_VRING_ALIGN,
-                                 vdev, lvq->pages, lg_notify, callback, name);
+        vq = vring_new_virtqueue(lvq->config.num, LGUEST_VRING_ALIGN, vdev,
+                                 true, lvq->pages, lg_notify, callback, name);
         if (!vq) {
                 err = -ENOMEM;
                 goto unmap;