[Qemu-devel] [PATCHv2 14/14] virtio-net: enable mrg buf header in tap on linux

["Michael S. Tsirkin" <mst@redhat.com>]

Modern linux supports arbitrary header size,
which makes it possible to pass mrg buf header
to tap directly without iovec mangling.
Use this capability when it is there.

This removes the need to deal with it in
vhost-net as we do now.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
---
 hw/vhost_net.c  | 13 -------------
 hw/virtio-net.c | 26 ++++++++++++++++++--------
 2 files changed, 18 insertions(+), 21 deletions(-)

diff --git a/hw/vhost_net.c b/hw/vhost_net.c
index df2c4a3..8241601 100644
--- a/hw/vhost_net.c
+++ b/hw/vhost_net.c
@@ -150,10 +150,6 @@ int vhost_net_start(struct vhost_net *net,
     if (r < 0) {
         goto fail_notifiers;
     }
-    if (net->dev.acked_features & (1 << VIRTIO_NET_F_MRG_RXBUF)) {
-        tap_set_vnet_hdr_len(net->nc,
-                             sizeof(struct virtio_net_hdr_mrg_rxbuf));
-    }
 
     r = vhost_dev_start(&net->dev, dev);
     if (r < 0) {
@@ -179,9 +175,6 @@ fail:
     }
     net->nc->info->poll(net->nc, true);
     vhost_dev_stop(&net->dev, dev);
-    if (net->dev.acked_features & (1 << VIRTIO_NET_F_MRG_RXBUF)) {
-        tap_set_vnet_hdr_len(net->nc, sizeof(struct virtio_net_hdr));
-    }
 fail_start:
     vhost_dev_disable_notifiers(&net->dev, dev);
 fail_notifiers:
@@ -199,18 +192,12 @@ void vhost_net_stop(struct vhost_net *net,
     }
     net->nc->info->poll(net->nc, true);
     vhost_dev_stop(&net->dev, dev);
-    if (net->dev.acked_features & (1 << VIRTIO_NET_F_MRG_RXBUF)) {
-        tap_set_vnet_hdr_len(net->nc, sizeof(struct virtio_net_hdr));
-    }
     vhost_dev_disable_notifiers(&net->dev, dev);
 }
 
 void vhost_net_cleanup(struct vhost_net *net)
 {
     vhost_dev_cleanup(&net->dev);
-    if (net->dev.acked_features & (1 << VIRTIO_NET_F_MRG_RXBUF)) {
-        tap_set_vnet_hdr_len(net->nc, sizeof(struct virtio_net_hdr));
-    }
     g_free(net);
 }
 #else
diff --git a/hw/virtio-net.c b/hw/virtio-net.c
index 86207a1..c2a6bbf 100644
--- a/hw/virtio-net.c
+++ b/hw/virtio-net.c
@@ -228,6 +228,20 @@ static int peer_has_ufo(VirtIONet *n)
     return n->has_ufo;
 }
 
+static void virtio_net_set_mrg_rx_bufs(VirtIONet *n, int mergeable_rx_bufs)
+{
+    n->mergeable_rx_bufs = mergeable_rx_bufs;
+
+    n->guest_hdr_len = n->mergeable_rx_bufs ?
+        sizeof(struct virtio_net_hdr_mrg_rxbuf) : sizeof(struct virtio_net_hdr);
+
+    if (peer_has_vnet_hdr(n) &&
+        tap_has_vnet_hdr_len(n->nic->nc.peer, n->guest_hdr_len)) {
+        tap_set_vnet_hdr_len(n->nic->nc.peer, n->guest_hdr_len);
+        n->host_hdr_len = n->guest_hdr_len;
+    }
+}
+
 static uint32_t virtio_net_get_features(VirtIODevice *vdev, uint32_t features)
 {
     VirtIONet *n = to_virtio_net(vdev);
@@ -280,9 +294,7 @@ static void virtio_net_set_features(VirtIODevice *vdev, uint32_t features)
 {
     VirtIONet *n = to_virtio_net(vdev);
 
-    n->mergeable_rx_bufs = !!(features & (1 << VIRTIO_NET_F_MRG_RXBUF));
-    n->guest_hdr_len = n->mergeable_rx_bufs ?
-        sizeof(struct virtio_net_hdr_mrg_rxbuf) : sizeof(struct virtio_net_hdr);
+    virtio_net_set_mrg_rx_bufs(n, !!(features & (1 << VIRTIO_NET_F_MRG_RXBUF)));
 
     if (n->has_vnet_hdr) {
         tap_set_offload(n->nic->nc.peer,
@@ -898,9 +910,8 @@ static int virtio_net_load(QEMUFile *f, void *opaque, int version_id)
 
     qemu_get_buffer(f, n->mac, ETH_ALEN);
     n->tx_waiting = qemu_get_be32(f);
-    n->mergeable_rx_bufs = qemu_get_be32(f);
-    n->guest_hdr_len = n->mergeable_rx_bufs ?
-        sizeof(struct virtio_net_hdr_mrg_rxbuf) : sizeof(struct virtio_net_hdr);
+
+    virtio_net_set_mrg_rx_bufs(n, qemu_get_be32(f));
 
     if (version_id >= 3)
         n->status = qemu_get_be16(f);
@@ -1043,8 +1054,7 @@ VirtIODevice *virtio_net_init(DeviceState *dev, NICConf *conf,
 
     n->tx_waiting = 0;
     n->tx_burst = net->txburst;
-    n->mergeable_rx_bufs = 0;
-    n->guest_hdr_len = sizeof(struct virtio_net_hdr);
+    virtio_net_set_mrg_rx_bufs(n, 0);
     n->promisc = 1; /* for compatibility */
 
     n->mac_table.macs = g_malloc0(MAC_TABLE_ENTRIES * ETH_ALEN);
-- 
MST