/*
- * Copyright (c) 2010 Nicira Networks.
- * Distributed under the terms of the GNU GPL version 2.
+ * Copyright (c) 2007-2012 Nicira, Inc.
*
- * Significant portions of this file may be copied from parts of the Linux
- * kernel, by Linus Torvalds and others.
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of version 2 of the GNU General Public
+ * License as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ * General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA
*/
+#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
+
#include <linux/if_arp.h>
#include <linux/if_bridge.h>
#include <linux/if_vlan.h>
#include <linux/llc.h>
#include <linux/rtnetlink.h>
#include <linux/skbuff.h>
+#include <linux/openvswitch.h>
#include <net/llc.h>
#include "datapath.h"
+#include "vlan.h"
#include "vport-internal_dev.h"
#include "vport-netdev.h"
-#include "compat.h"
+static void netdev_port_receive(struct vport *vport, struct sk_buff *skb);
-/* If the native device stats aren't 64 bit use the vport stats tracking instead. */
-#define USE_VPORT_STATS (sizeof(((struct net_device_stats *)0)->rx_bytes) < sizeof(u64))
+#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,39)
+/* 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;
-static void netdev_port_receive(struct vport *vport, struct sk_buff *skb);
+ if (unlikely(skb->pkt_type == PACKET_LOOPBACK))
+ return RX_HANDLER_PASS;
+
+ vport = ovs_netdev_get_vport(skb->dev);
+
+ netdev_port_receive(vport, skb);
-#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,36)
+ return RX_HANDLER_CONSUMED;
+}
+#elif LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,36) || \
+ defined HAVE_RHEL_OVS_HOOK
/* Called with rcu_read_lock and bottom-halves disabled. */
static struct sk_buff *netdev_frame_hook(struct sk_buff *skb)
{
if (unlikely(skb->pkt_type == PACKET_LOOPBACK))
return skb;
- vport = netdev_get_vport(skb->dev);
+ vport = ovs_netdev_get_vport(skb->dev);
netdev_port_receive(vport, 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!)
netdev_port_receive((struct vport *)p, 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)
-static int netdev_init(void) { return 0; }
-static void netdev_exit(void) { }
-#else
-static int netdev_init(void)
+static struct net_device *get_dpdev(struct datapath *dp)
{
- /* Hook into callback used by the bridge to intercept packets.
- * Parasites we are. */
- br_handle_frame_hook = netdev_frame_hook;
-
- return 0;
-}
+ struct vport *local;
-static void netdev_exit(void)
-{
- br_handle_frame_hook = NULL;
+ local = ovs_vport_ovsl(dp, OVSP_LOCAL);
+ BUG_ON(!local);
+ return netdev_vport_priv(local)->dev;
}
-#endif
static struct vport *netdev_create(const struct vport_parms *parms)
{
struct netdev_vport *netdev_vport;
int err;
- vport = vport_alloc(sizeof(struct netdev_vport), &netdev_vport_ops, parms);
+ 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(&init_net, parms->name);
+ netdev_vport->dev = dev_get_by_name(ovs_dp_get_net(vport->dp), parms->name);
if (!netdev_vport->dev) {
err = -ENODEV;
goto error_free_vport;
if (netdev_vport->dev->flags & IFF_LOOPBACK ||
netdev_vport->dev->type != ARPHRD_ETHER ||
- is_internal_dev(netdev_vport->dev)) {
+ ovs_is_internal_dev(netdev_vport->dev)) {
err = -EINVAL;
goto error_put;
}
- /* If we are using the vport stats layer initialize it to the current
- * values so we are roughly consistent with the device stats. */
- if (USE_VPORT_STATS) {
- struct rtnl_link_stats64 stats;
+ rtnl_lock();
+ err = netdev_master_upper_dev_link(netdev_vport->dev,
+ get_dpdev(vport->dp));
+ if (err)
+ goto error_unlock;
- err = netdev_get_stats(vport, &stats);
- if (!err)
- vport_set_stats(vport, &stats);
- }
+ err = netdev_rx_handler_register(netdev_vport->dev, netdev_frame_hook,
+ vport);
+ if (err)
+ goto error_master_upper_dev_unlink;
+
+ dev_set_promiscuity(netdev_vport->dev, 1);
+ netdev_vport->dev->priv_flags |= IFF_OVS_DATAPATH;
+ rtnl_unlock();
return vport;
+error_master_upper_dev_unlink:
+ netdev_upper_dev_unlink(netdev_vport->dev, get_dpdev(vport->dp));
+error_unlock:
+ rtnl_unlock();
error_put:
dev_put(netdev_vport->dev);
error_free_vport:
- vport_free(vport);
+ ovs_vport_free(vport);
error:
return ERR_PTR(err);
}
-static int netdev_destroy(struct vport *vport)
+static void free_port_rcu(struct rcu_head *rcu)
{
- struct netdev_vport *netdev_vport = netdev_vport_priv(vport);
+ struct netdev_vport *netdev_vport = container_of(rcu,
+ struct netdev_vport, rcu);
dev_put(netdev_vport->dev);
- vport_free(vport);
-
- return 0;
-}
-
-static int netdev_attach(struct vport *vport)
-{
- struct netdev_vport *netdev_vport = netdev_vport_priv(vport);
- int err;
-
- err = netdev_rx_handler_register(netdev_vport->dev, netdev_frame_hook,
- vport);
- if (err)
- return err;
-
- dev_set_promiscuity(netdev_vport->dev, 1);
- dev_disable_lro(netdev_vport->dev);
- netdev_vport->dev->priv_flags |= IFF_OVS_DATAPATH;
-
- return 0;
+ ovs_vport_free(vport_from_priv(netdev_vport));
}
-static int netdev_detach(struct vport *vport)
+void ovs_netdev_detach_dev(struct vport *vport)
{
struct netdev_vport *netdev_vport = netdev_vport_priv(vport);
+ ASSERT_RTNL();
netdev_vport->dev->priv_flags &= ~IFF_OVS_DATAPATH;
netdev_rx_handler_unregister(netdev_vport->dev);
+ netdev_upper_dev_unlink(netdev_vport->dev,
+ netdev_master_upper_dev_get(netdev_vport->dev));
dev_set_promiscuity(netdev_vport->dev, -1);
-
- return 0;
-}
-
-int netdev_set_mtu(struct vport *vport, int mtu)
-{
- struct netdev_vport *netdev_vport = netdev_vport_priv(vport);
- return dev_set_mtu(netdev_vport->dev, mtu);
}
-int netdev_set_addr(struct vport *vport, const unsigned char *addr)
+static void netdev_destroy(struct vport *vport)
{
struct netdev_vport *netdev_vport = netdev_vport_priv(vport);
- struct sockaddr sa;
- sa.sa_family = ARPHRD_ETHER;
- memcpy(sa.sa_data, addr, ETH_ALEN);
+ rtnl_lock();
+ if (ovs_netdev_get_vport(netdev_vport->dev))
+ ovs_netdev_detach_dev(vport);
+ rtnl_unlock();
- return dev_set_mac_address(netdev_vport->dev, &sa);
+ call_rcu(&netdev_vport->rcu, free_port_rcu);
}
-const char *netdev_get_name(const struct vport *vport)
+const char *ovs_netdev_get_name(const struct vport *vport)
{
const struct netdev_vport *netdev_vport = netdev_vport_priv(vport);
return netdev_vport->dev->name;
}
-const unsigned char *netdev_get_addr(const struct vport *vport)
-{
- const struct netdev_vport *netdev_vport = netdev_vport_priv(vport);
- return netdev_vport->dev->dev_addr;
-}
-
-struct kobject *netdev_get_kobj(const struct vport *vport)
-{
- const struct netdev_vport *netdev_vport = netdev_vport_priv(vport);
- return &netdev_vport->dev->NETDEV_DEV_MEMBER.kobj;
-}
-
-int netdev_get_stats(const struct vport *vport, struct rtnl_link_stats64 *stats)
-{
- const struct netdev_vport *netdev_vport = netdev_vport_priv(vport);
- dev_get_stats(netdev_vport->dev, stats);
- return 0;
-}
-
-unsigned netdev_get_dev_flags(const struct vport *vport)
-{
- const struct netdev_vport *netdev_vport = netdev_vport_priv(vport);
- return dev_get_flags(netdev_vport->dev);
-}
-
-int netdev_is_running(const struct vport *vport)
-{
- const struct netdev_vport *netdev_vport = netdev_vport_priv(vport);
- return netif_running(netdev_vport->dev);
-}
-
-unsigned char netdev_get_operstate(const struct vport *vport)
-{
- const struct netdev_vport *netdev_vport = netdev_vport_priv(vport);
- return netdev_vport->dev->operstate;
-}
-
-int netdev_get_ifindex(const struct vport *vport)
-{
- const struct netdev_vport *netdev_vport = netdev_vport_priv(vport);
- return netdev_vport->dev->ifindex;
-}
-
-int netdev_get_iflink(const struct vport *vport)
-{
- const struct netdev_vport *netdev_vport = netdev_vport_priv(vport);
- return netdev_vport->dev->iflink;
-}
-
-int netdev_get_mtu(const struct vport *vport)
-{
- const struct netdev_vport *netdev_vport = netdev_vport_priv(vport);
- return netdev_vport->dev->mtu;
-}
-
/* 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
if (unlikely(!skb))
return;
- skb_warn_if_lro(skb);
-
skb_push(skb, ETH_HLEN);
- compute_ip_summed(skb, false);
+ ovs_skb_postpush_rcsum(skb, skb->data, ETH_HLEN);
- vport_receive(vport, skb);
+ ovs_vport_receive(vport, skb, NULL);
+ return;
+
+error:
+ kfree_skb(skb);
+}
+
+static unsigned int packet_length(const struct sk_buff *skb)
+{
+ unsigned int length = skb->len - ETH_HLEN;
+
+ if (skb->protocol == htons(ETH_P_8021Q))
+ length -= VLAN_HLEN;
+
+ return length;
}
static int netdev_send(struct vport *vport, struct sk_buff *skb)
{
struct netdev_vport *netdev_vport = netdev_vport_priv(vport);
- int len = skb->len;
+ 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);
+ len = skb->len;
dev_queue_xmit(skb);
return len;
+
+drop:
+ kfree_skb(skb);
+ return 0;
}
/* Returns null if this device is not attached to a datapath. */
-struct vport *netdev_get_vport(struct net_device *dev)
+struct vport *ovs_netdev_get_vport(struct net_device *dev)
{
-#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,36)
- /* XXX: The bridge code may have registered the data.
- * So check that the handler pointer is the datapath's.
- * Once the merge is done and IFF_OVS_DATAPATH stops
- * being the same value as IFF_BRIDGE_PORT the check can
- * simply be netdev_vport->dev->priv_flags & IFF_OVS_DATAPATH. */
- if (rcu_dereference(dev->rx_handler) != netdev_frame_hook)
+#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,36) || \
+ defined HAVE_RHEL_OVS_HOOK
+#if IFF_OVS_DATAPATH != 0
+ 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
+ return (struct vport *)rcu_dereference_rtnl(dev->rx_handler_data);
+#endif
+ else
return NULL;
- return (struct vport *)rcu_dereference(dev->rx_handler_data);
#else
- return (struct vport *)rcu_dereference(dev->br_port);
+ return (struct vport *)rcu_dereference_rtnl(dev->br_port);
#endif
}
-const struct vport_ops netdev_vport_ops = {
- .type = "netdev",
- .flags = (VPORT_F_REQUIRED |
- (USE_VPORT_STATS ? VPORT_F_GEN_STATS : 0)),
- .init = netdev_init,
- .exit = netdev_exit,
+const struct vport_ops ovs_netdev_vport_ops = {
+ .type = OVS_VPORT_TYPE_NETDEV,
.create = netdev_create,
.destroy = netdev_destroy,
- .attach = netdev_attach,
- .detach = netdev_detach,
- .set_mtu = netdev_set_mtu,
- .set_addr = netdev_set_addr,
- .get_name = netdev_get_name,
- .get_addr = netdev_get_addr,
- .get_kobj = netdev_get_kobj,
- .get_stats = netdev_get_stats,
- .get_dev_flags = netdev_get_dev_flags,
- .is_running = netdev_is_running,
- .get_operstate = netdev_get_operstate,
- .get_ifindex = netdev_get_ifindex,
- .get_iflink = netdev_get_iflink,
- .get_mtu = netdev_get_mtu,
+ .get_name = ovs_netdev_get_name,
.send = netdev_send,
};
-#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,36)
+#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,36) && \
+ !defined HAVE_RHEL_OVS_HOOK
/*
- * In kernels earlier than 2.6.36, Open vSwitch cannot safely coexist with
- * the Linux bridge module on any released version of Linux, because there
- * is only a single bridge hook function and only a single br_port member
- * in struct net_device.
+ * Enforces, mutual exclusion with the Linux bridge module, by declaring and
+ * exporting br_should_route_hook. Because the bridge module also exports the
+ * same symbol, the module loader will refuse to load both modules at the same
+ * time (e.g. "bridge: exports duplicate symbol br_should_route_hook (owned by
+ * openvswitch)").
*
- * Declaring and exporting this symbol enforces mutual exclusion. The bridge
- * module also exports the same symbol, so the module loader will refuse to
- * load both modules at the same time (e.g. "bridge: exports duplicate symbol
- * br_should_route_hook (owned by openvswitch_mod)").
+ * Before Linux 2.6.36, Open vSwitch cannot safely coexist with the Linux
+ * bridge module, so openvswitch uses this macro in those versions. In
+ * Linux 2.6.36 and later, Open vSwitch can coexist with the bridge module.
*
* The use of "typeof" here avoids the need to track changes in the type of
* br_should_route_hook over various kernel versions.
git://git.onelab.eu / sliver-openvswitch.git / blobdiff