net/virtio: reinitialize device when configuring
authorOlivier Matz <olivier.matz@6wind.com>
Thu, 13 Oct 2016 14:16:02 +0000 (16:16 +0200)
committerThomas Monjalon <thomas.monjalon@6wind.com>
Thu, 13 Oct 2016 18:30:33 +0000 (20:30 +0200)
Add the ability to reset the virtio device in the configure callback
if the features flag changed since previous reset. This will be possible
with the introduction of offload support in next commits.

Signed-off-by: Olivier Matz <olivier.matz@6wind.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Acked-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
drivers/net/virtio/virtio_ethdev.c
drivers/net/virtio/virtio_pci.h

index 9abcfc9..7048f84 100644 (file)
@@ -1063,14 +1063,13 @@ virtio_vlan_filter_set(struct rte_eth_dev *dev, uint16_t vlan_id, int on)
 }
 
 static int
-virtio_negotiate_features(struct virtio_hw *hw)
+virtio_negotiate_features(struct virtio_hw *hw, uint64_t req_features)
 {
        uint64_t host_features;
 
        /* Prepare guest_features: feature that driver wants to support */
-       hw->guest_features = VIRTIO_PMD_GUEST_FEATURES;
        PMD_INIT_LOG(DEBUG, "guest_features before negotiate = %" PRIx64,
-               hw->guest_features);
+               req_features);
 
        /* Read device(host) feature bits */
        host_features = hw->vtpci_ops->get_features(hw);
@@ -1081,6 +1080,7 @@ virtio_negotiate_features(struct virtio_hw *hw)
         * Negotiate features: Subset of device feature bits are written back
         * guest feature bits.
         */
+       hw->guest_features = req_features;
        hw->guest_features = vtpci_negotiate_features(hw, host_features);
        PMD_INIT_LOG(DEBUG, "features after negotiate = %" PRIx64,
                hw->guest_features);
@@ -1099,6 +1099,8 @@ virtio_negotiate_features(struct virtio_hw *hw)
                }
        }
 
+       hw->req_guest_features = req_features;
+
        return 0;
 }
 
@@ -1139,8 +1141,9 @@ rx_func_get(struct rte_eth_dev *eth_dev)
                eth_dev->rx_pkt_burst = &virtio_recv_pkts;
 }
 
+/* reset device and renegotiate features if needed */
 static int
-virtio_init_device(struct rte_eth_dev *eth_dev)
+virtio_init_device(struct rte_eth_dev *eth_dev, uint64_t req_features)
 {
        struct virtio_hw *hw = eth_dev->data->dev_private;
        struct virtio_net_config *config;
@@ -1155,7 +1158,7 @@ virtio_init_device(struct rte_eth_dev *eth_dev)
 
        /* Tell the host we've known how to drive the device. */
        vtpci_set_status(hw, VIRTIO_CONFIG_STATUS_DRIVER);
-       if (virtio_negotiate_features(hw) < 0)
+       if (virtio_negotiate_features(hw, req_features) < 0)
                return -1;
 
        /* If host does not support status then disable LSC */
@@ -1276,8 +1279,8 @@ eth_virtio_dev_init(struct rte_eth_dev *eth_dev)
 
        eth_dev->data->dev_flags = dev_flags;
 
-       /* reset device and negotiate features */
-       ret = virtio_init_device(eth_dev);
+       /* reset device and negotiate default features */
+       ret = virtio_init_device(eth_dev, VIRTIO_PMD_GUEST_FEATURES);
        if (ret < 0)
                return ret;
 
@@ -1359,6 +1362,7 @@ virtio_dev_configure(struct rte_eth_dev *dev)
 {
        const struct rte_eth_rxmode *rxmode = &dev->data->dev_conf.rxmode;
        struct virtio_hw *hw = dev->data->dev_private;
+       uint64_t req_features;
        int ret;
 
        PMD_INIT_LOG(DEBUG, "configure");
@@ -1368,6 +1372,14 @@ virtio_dev_configure(struct rte_eth_dev *dev)
                return -EINVAL;
        }
 
+       req_features = VIRTIO_PMD_GUEST_FEATURES;
+       /* if request features changed, reinit the device */
+       if (req_features != hw->req_guest_features) {
+               ret = virtio_init_device(dev, req_features);
+               if (ret < 0)
+                       return ret;
+       }
+
        /* Setup and start control queue */
        if (vtpci_with_feature(hw, VIRTIO_NET_F_CTRL_VQ)) {
                ret = virtio_dev_cq_queue_setup(dev,
index 38168e0..0c5ed31 100644 (file)
@@ -246,6 +246,7 @@ struct virtio_net_config;
 struct virtio_hw {
        struct virtnet_ctl *cvq;
        struct rte_pci_ioport io;
+       uint64_t    req_guest_features;
        uint64_t    guest_features;
        uint32_t    max_queue_pairs;
        uint16_t    vtnet_hdr_size;