/*
- * Copyright (c) 2007-2012 Nicira, Inc.
+ * Copyright (c) 2007-2015 Nicira, Inc.
*
* This program is free software; you can redistribute it and/or
* modify it under the terms of version 2 of the GNU General Public
#include <linux/skbuff.h>
#include <linux/openvswitch.h>
-#include <net/llc.h>
+#include <net/ip_tunnels.h>
+#include <net/rtnetlink.h>
-#include "checksum.h"
#include "datapath.h"
-#include "vlan.h"
+#include "gso.h"
+#include "vport.h"
#include "vport-internal_dev.h"
#include "vport-netdev.h"
-#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,37) && \
- !defined(HAVE_VLAN_BUG_WORKAROUND)
-#include <linux/module.h>
+static struct vport_ops ovs_netdev_vport_ops;
-static int vlan_tso __read_mostly;
-module_param(vlan_tso, int, 0644);
-MODULE_PARM_DESC(vlan_tso, "Enable TSO for VLAN packets");
+/* Must be called with rcu_read_lock. */
+void netdev_port_receive(struct sk_buff *skb, struct ip_tunnel_info *tun_info)
+{
+ struct vport *vport;
+
+ vport = ovs_netdev_get_vport(skb->dev);
+ if (unlikely(!vport))
+ goto error;
+
+ if (unlikely(skb_warn_if_lro(skb)))
+ goto error;
+
+ /* Make our own copy of the packet. Otherwise we will mangle the
+ * packet for anyone who came before us (e.g. tcpdump via AF_PACKET).
+ */
+ skb = skb_share_check(skb, GFP_ATOMIC);
+ if (unlikely(!skb))
+ return;
+
+ skb_push(skb, ETH_HLEN);
+ ovs_skb_postpush_rcsum(skb, skb->data, ETH_HLEN);
+ ovs_vport_receive(vport, skb, tun_info);
+ return;
+error:
+ kfree_skb(skb);
+}
+
+#ifndef HAVE_METADATA_DST
+#define port_receive(skb) netdev_port_receive(skb, NULL)
#else
-#define vlan_tso true
+#define port_receive(skb) netdev_port_receive(skb, skb_tunnel_info(skb))
#endif
-static void netdev_port_receive(struct vport *vport, struct sk_buff *skb);
-
-#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,39)
+#if defined HAVE_RX_HANDLER_PSKB /* 2.6.39 and above or backports */
/* Called with rcu_read_lock and bottom-halves disabled. */
static rx_handler_result_t netdev_frame_hook(struct sk_buff **pskb)
{
struct sk_buff *skb = *pskb;
- struct vport *vport;
if (unlikely(skb->pkt_type == PACKET_LOOPBACK))
return RX_HANDLER_PASS;
- vport = ovs_netdev_get_vport(skb->dev);
-
- netdev_port_receive(vport, skb);
-
+ port_receive(skb);
return RX_HANDLER_CONSUMED;
}
#elif LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,36) || \
/* Called with rcu_read_lock and bottom-halves disabled. */
static struct sk_buff *netdev_frame_hook(struct sk_buff *skb)
{
- struct vport *vport;
-
if (unlikely(skb->pkt_type == PACKET_LOOPBACK))
return skb;
- vport = ovs_netdev_get_vport(skb->dev);
-
- netdev_port_receive(vport, skb);
-
+ port_receive(skb);
return NULL;
}
-#elif LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22)
+#elif LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,32)
/*
* Used as br_handle_frame_hook. (Cannot run bridge at the same time, even on
* different set of devices!)
static struct sk_buff *netdev_frame_hook(struct net_bridge_port *p,
struct sk_buff *skb)
{
- netdev_port_receive((struct vport *)p, skb);
+ port_receive(skb);
return NULL;
}
-#elif LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0)
-/*
- * Used as br_handle_frame_hook. (Cannot run bridge at the same time, even on
- * different set of devices!)
- */
-/* Called with rcu_read_lock and bottom-halves disabled. */
-static int netdev_frame_hook(struct net_bridge_port *p, struct sk_buff **pskb)
-{
- netdev_port_receive((struct vport *)p, *pskb);
- return 1;
-}
#else
#error
#endif
-#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,36) || \
- defined HAVE_RHEL_OVS_HOOK
-static int netdev_init(void) { return 0; }
-static void netdev_exit(void) { }
-#else
-static int port_count;
-
-static void netdev_init(void)
-{
- port_count++;
- if (port_count > 1)
- return;
-
- /* Hook into callback used by the bridge to intercept packets.
- * Parasites we are. */
- br_handle_frame_hook = netdev_frame_hook;
-
- return;
-}
-
-static void netdev_exit(void)
-{
- port_count--;
- if (port_count > 0)
- return;
-
- br_handle_frame_hook = NULL;
-}
-#endif
-
-static struct net_device *get_dpdev(struct datapath *dp)
+static struct net_device *get_dpdev(const struct datapath *dp)
{
struct vport *local;
local = ovs_vport_ovsl(dp, OVSP_LOCAL);
BUG_ON(!local);
- return netdev_vport_priv(local)->dev;
+ return local->dev;
}
-static struct vport *netdev_create(const struct vport_parms *parms)
+struct vport *ovs_netdev_link(struct vport *vport, const char *name)
{
- struct vport *vport;
- struct netdev_vport *netdev_vport;
int err;
- vport = ovs_vport_alloc(sizeof(struct netdev_vport),
- &ovs_netdev_vport_ops, parms);
- if (IS_ERR(vport)) {
- err = PTR_ERR(vport);
- goto error;
- }
-
- netdev_vport = netdev_vport_priv(vport);
-
- netdev_vport->dev = dev_get_by_name(ovs_dp_get_net(vport->dp), parms->name);
- if (!netdev_vport->dev) {
+ vport->dev = dev_get_by_name(ovs_dp_get_net(vport->dp), name);
+ if (!vport->dev) {
err = -ENODEV;
goto error_free_vport;
}
- if (netdev_vport->dev->flags & IFF_LOOPBACK ||
- netdev_vport->dev->type != ARPHRD_ETHER ||
- ovs_is_internal_dev(netdev_vport->dev)) {
+ if (vport->dev->flags & IFF_LOOPBACK ||
+ vport->dev->type != ARPHRD_ETHER ||
+ ovs_is_internal_dev(vport->dev)) {
err = -EINVAL;
goto error_put;
}
rtnl_lock();
- err = netdev_master_upper_dev_link(netdev_vport->dev,
+ err = netdev_master_upper_dev_link(vport->dev,
get_dpdev(vport->dp));
if (err)
goto error_unlock;
- err = netdev_rx_handler_register(netdev_vport->dev, netdev_frame_hook,
+ err = netdev_rx_handler_register(vport->dev, netdev_frame_hook,
vport);
if (err)
goto error_master_upper_dev_unlink;
- dev_set_promiscuity(netdev_vport->dev, 1);
-#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,24)
- dev_disable_lro(netdev_vport->dev);
-#endif
- netdev_vport->dev->priv_flags |= IFF_OVS_DATAPATH;
+ dev_disable_lro(vport->dev);
+ dev_set_promiscuity(vport->dev, 1);
+ vport->dev->priv_flags |= IFF_OVS_DATAPATH;
rtnl_unlock();
- netdev_init();
return vport;
error_master_upper_dev_unlink:
- netdev_upper_dev_unlink(netdev_vport->dev, get_dpdev(vport->dp));
+ netdev_upper_dev_unlink(vport->dev, get_dpdev(vport->dp));
error_unlock:
rtnl_unlock();
error_put:
- dev_put(netdev_vport->dev);
+ dev_put(vport->dev);
error_free_vport:
ovs_vport_free(vport);
-error:
return ERR_PTR(err);
}
+EXPORT_SYMBOL_GPL(ovs_netdev_link);
-static void free_port_rcu(struct rcu_head *rcu)
-{
- struct netdev_vport *netdev_vport = container_of(rcu,
- struct netdev_vport, rcu);
-
- dev_put(netdev_vport->dev);
- ovs_vport_free(vport_from_priv(netdev_vport));
-}
-
-static void netdev_destroy(struct vport *vport)
+static struct vport *netdev_create(const struct vport_parms *parms)
{
- struct netdev_vport *netdev_vport = netdev_vport_priv(vport);
+ struct vport *vport;
- netdev_exit();
- rtnl_lock();
- netdev_vport->dev->priv_flags &= ~IFF_OVS_DATAPATH;
- netdev_rx_handler_unregister(netdev_vport->dev);
- netdev_upper_dev_unlink(netdev_vport->dev, get_dpdev(vport->dp));
- dev_set_promiscuity(netdev_vport->dev, -1);
- rtnl_unlock();
+ vport = ovs_vport_alloc(0, &ovs_netdev_vport_ops, parms);
+ if (IS_ERR(vport))
+ return vport;
- call_rcu(&netdev_vport->rcu, free_port_rcu);
+ return ovs_netdev_link(vport, parms->name);
}
-const char *ovs_netdev_get_name(const struct vport *vport)
+static void vport_netdev_free(struct rcu_head *rcu)
{
- const struct netdev_vport *netdev_vport = netdev_vport_priv(vport);
- return netdev_vport->dev->name;
-}
+ struct vport *vport = container_of(rcu, struct vport, rcu);
-/* Must be called with rcu_read_lock. */
-static void netdev_port_receive(struct vport *vport, struct sk_buff *skb)
-{
- if (unlikely(!vport))
- goto error;
-
- if (unlikely(skb_warn_if_lro(skb)))
- goto error;
-
- /* Make our own copy of the packet. Otherwise we will mangle the
- * packet for anyone who came before us (e.g. tcpdump via AF_PACKET).
- * (No one comes after us, since we tell handle_bridge() that we took
- * the packet.) */
- skb = skb_share_check(skb, GFP_ATOMIC);
- if (unlikely(!skb))
- return;
-
- if (unlikely(compute_ip_summed(skb, false)))
- goto error;
-
- skb_push(skb, ETH_HLEN);
- ovs_skb_postpush_rcsum(skb, skb->data, ETH_HLEN);
-
- vlan_copy_skb_tci(skb);
-
- ovs_vport_receive(vport, skb, NULL);
- return;
-
-error:
- kfree_skb(skb);
+ if (vport->dev)
+ dev_put(vport->dev);
+ ovs_vport_free(vport);
}
-static unsigned int packet_length(const struct sk_buff *skb)
+void ovs_netdev_detach_dev(struct vport *vport)
{
- unsigned int length = skb->len - ETH_HLEN;
-
- if (skb->protocol == htons(ETH_P_8021Q))
- length -= VLAN_HLEN;
-
- return length;
+ ASSERT_RTNL();
+ vport->dev->priv_flags &= ~IFF_OVS_DATAPATH;
+ netdev_rx_handler_unregister(vport->dev);
+ netdev_upper_dev_unlink(vport->dev,
+ netdev_master_upper_dev_get(vport->dev));
+ dev_set_promiscuity(vport->dev, -1);
}
+EXPORT_SYMBOL_GPL(ovs_netdev_detach_dev);
-static bool dev_supports_vlan_tx(struct net_device *dev)
+static void netdev_destroy(struct vport *vport)
{
-#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,37)
- /* Software fallback means every device supports vlan_tci on TX. */
- return true;
-#elif defined(HAVE_VLAN_BUG_WORKAROUND)
- return dev->features & NETIF_F_HW_VLAN_TX;
-#else
- /* Assume that the driver is buggy. */
- return false;
-#endif
+ rtnl_lock();
+ if (vport->dev->priv_flags & IFF_OVS_DATAPATH)
+ ovs_netdev_detach_dev(vport);
+ rtnl_unlock();
+
+ call_rcu(&vport->rcu, vport_netdev_free);
}
-static int netdev_send(struct vport *vport, struct sk_buff *skb)
+void ovs_netdev_tunnel_destroy(struct vport *vport)
{
- struct netdev_vport *netdev_vport = netdev_vport_priv(vport);
- int mtu = netdev_vport->dev->mtu;
- int len;
-
- if (unlikely(packet_length(skb) > mtu && !skb_is_gso(skb))) {
- net_warn_ratelimited("%s: dropped over-mtu packet: %d > %d\n",
- netdev_vport->dev->name,
- packet_length(skb), mtu);
- goto drop;
- }
-
- skb->dev = netdev_vport->dev;
- forward_ip_summed(skb, true);
-
- if (vlan_tx_tag_present(skb) && !dev_supports_vlan_tx(skb->dev)) {
- int features;
-
- features = netif_skb_features(skb);
-
- if (!vlan_tso)
- features &= ~(NETIF_F_TSO | NETIF_F_TSO6 |
- NETIF_F_UFO | NETIF_F_FSO);
-
- if (netif_needs_gso(skb, features)) {
- struct sk_buff *nskb;
-
- nskb = skb_gso_segment(skb, features);
- if (!nskb) {
- if (unlikely(skb_cloned(skb) &&
- pskb_expand_head(skb, 0, 0, GFP_ATOMIC)))
- goto drop;
-
- skb_shinfo(skb)->gso_type &= ~SKB_GSO_DODGY;
- goto tag;
- }
-
- if (IS_ERR(nskb))
- goto drop;
- consume_skb(skb);
- skb = nskb;
-
- len = 0;
- do {
- nskb = skb->next;
- skb->next = NULL;
-
- skb = __vlan_put_tag(skb, skb->vlan_proto, vlan_tx_tag_get(skb));
- if (likely(skb)) {
- len += skb->len;
- vlan_set_tci(skb, 0);
- dev_queue_xmit(skb);
- }
-
- skb = nskb;
- } while (skb);
-
- return len;
- }
+ rtnl_lock();
+ if (vport->dev->priv_flags & IFF_OVS_DATAPATH)
+ ovs_netdev_detach_dev(vport);
-tag:
- skb = __vlan_put_tag(skb, skb->vlan_proto, vlan_tx_tag_get(skb));
- if (unlikely(!skb))
- return 0;
- vlan_set_tci(skb, 0);
- }
+ /* We can be invoked by both explicit vport deletion and
+ * underlying netdev deregistration; delete the link only
+ * if it's not already shutting down.
+ */
- len = skb->len;
- dev_queue_xmit(skb);
+ if (vport->dev->reg_state == NETREG_REGISTERED)
+ rtnl_delete_link(vport->dev);
- return len;
+ dev_put(vport->dev);
+ vport->dev = NULL;
+ rtnl_unlock();
-drop:
- kfree_skb(skb);
- return 0;
+ call_rcu(&vport->rcu, vport_netdev_free);
}
+EXPORT_SYMBOL_GPL(ovs_netdev_tunnel_destroy);
/* Returns null if this device is not attached to a datapath. */
struct vport *ovs_netdev_get_vport(struct net_device *dev)
{
-#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,36) || \
+#if defined HAVE_NETDEV_RX_HANDLER_REGISTER || \
defined HAVE_RHEL_OVS_HOOK
-#if IFF_OVS_DATAPATH != 0
+#ifdef HAVE_OVS_DATAPATH
if (likely(dev->priv_flags & IFF_OVS_DATAPATH))
#else
if (likely(rcu_access_pointer(dev->rx_handler) == netdev_frame_hook))
#endif
#ifdef HAVE_RHEL_OVS_HOOK
return (struct vport *)rcu_dereference_rtnl(dev->ax25_ptr);
+#else
+#ifdef HAVE_NET_DEVICE_EXTENDED
+ return (struct vport *)
+ rcu_dereference_rtnl(netdev_extended(dev)->rx_handler_data);
#else
return (struct vport *)rcu_dereference_rtnl(dev->rx_handler_data);
+#endif
#endif
else
return NULL;
#endif
}
-const struct vport_ops ovs_netdev_vport_ops = {
+static struct vport_ops ovs_netdev_vport_ops = {
.type = OVS_VPORT_TYPE_NETDEV,
.create = netdev_create,
.destroy = netdev_destroy,
- .get_name = ovs_netdev_get_name,
- .send = netdev_send,
+ .send = dev_queue_xmit,
};
-#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,36) && \
+int __init ovs_netdev_init(void)
+{
+ return ovs_vport_ops_register(&ovs_netdev_vport_ops);
+}
+
+void ovs_netdev_exit(void)
+{
+ ovs_vport_ops_unregister(&ovs_netdev_vport_ops);
+}
+
+#if !defined HAVE_NETDEV_RX_HANDLER_REGISTER && \
!defined HAVE_RHEL_OVS_HOOK
/*
* Enforces, mutual exclusion with the Linux bridge module, by declaring and