X-Git-Url: http://git.cascardo.info/?a=blobdiff_plain;f=datapath%2Fvport-netdev.c;h=a1df79b2b9dd770afca437134b8a5df525a377af;hb=HEAD;hp=a85a9b5fa557c092427af27493b627ee5e3079f0;hpb=431585099e82a88a23d5dec28138b5a3a8bb11da;p=cascardo%2Fovs.git

diff --git a/datapath/vport-netdev.c b/datapath/vport-netdev.c
index a85a9b5fa..a1df79b2b 100644
--- a/datapath/vport-netdev.c
+++ b/datapath/vport-netdev.c
@@ -1,11 +1,23 @@
 /*
- * Copyright (c) 2010 Nicira Networks.
- * Distributed under the terms of the GNU GPL version 2.
+ * Copyright (c) 2007-2015 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>
@@ -13,36 +25,76 @@
 #include <linux/llc.h>
 #include <linux/rtnetlink.h>
 #include <linux/skbuff.h>
+#include <linux/openvswitch.h>
 
-#include <net/llc.h>
+#include <net/ip_tunnels.h>
+#include <net/rtnetlink.h>
 
 #include "datapath.h"
+#include "gso.h"
+#include "vport.h"
 #include "vport-internal_dev.h"
 #include "vport-netdev.h"
 
-#include "compat.h"
+static struct vport_ops ovs_netdev_vport_ops;
+
+/* 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;
 
-/* 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))
+	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);
+}
 
-static void netdev_port_receive(struct vport *vport, struct sk_buff *skb);
+#ifndef HAVE_METADATA_DST
+#define port_receive(skb)  netdev_port_receive(skb, NULL)
+#else
+#define port_receive(skb)  netdev_port_receive(skb, skb_tunnel_info(skb))
+#endif
 
-#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,36)
+#if defined HAVE_RX_HANDLER_PSKB  /* 2.6.39 and above or backports */
 /* Called with rcu_read_lock and bottom-halves disabled. */
-static struct sk_buff *netdev_frame_hook(struct sk_buff *skb)
+static rx_handler_result_t netdev_frame_hook(struct sk_buff **pskb)
 {
-	struct vport *vport;
+	struct sk_buff *skb = *pskb;
 
 	if (unlikely(skb->pkt_type == PACKET_LOOPBACK))
-		return skb;
+		return RX_HANDLER_PASS;
 
-	vport = (struct vport *)rcu_dereference(skb->dev->br_port);
-
-	netdev_port_receive(vport, skb);
+	port_receive(skb);
+	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;
 
+	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!)
@@ -51,304 +103,192 @@ static struct sk_buff *netdev_frame_hook(struct sk_buff *skb)
 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)
-static int netdev_init(void) { return 0; }
-static void netdev_exit(void) { }
-#else
-static int netdev_init(void)
+static struct net_device *get_dpdev(const struct datapath *dp)
 {
-	/* Hook into callback used by the bridge to intercept packets.
-	 * Parasites we are. */
-	br_handle_frame_hook = netdev_frame_hook;
+	struct vport *local;
 
-	return 0;
+	local = ovs_vport_ovsl(dp, OVSP_LOCAL);
+	BUG_ON(!local);
+	return local->dev;
 }
 
-static void netdev_exit(void)
+struct vport *ovs_netdev_link(struct vport *vport, const char *name)
 {
-	br_handle_frame_hook = NULL;
-}
-#endif
-
-static struct vport *netdev_create(const char *name, const void __user *config)
-{
-	struct vport *vport;
-	struct netdev_vport *netdev_vport;
 	int err;
 
-	vport = vport_alloc(sizeof(struct netdev_vport), &netdev_vport_ops);
-	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, 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 ||
-	    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;
 	}
 
-	if (netdev_vport->dev->br_port) {
-		err = -EBUSY;
-		goto error_put;
-	}
+	rtnl_lock();
+	err = netdev_master_upper_dev_link(vport->dev,
+					   get_dpdev(vport->dp));
+	if (err)
+		goto error_unlock;
 
-	/* 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 odp_vport_stats stats;
+	err = netdev_rx_handler_register(vport->dev, netdev_frame_hook,
+					 vport);
+	if (err)
+		goto error_master_upper_dev_unlink;
 
-		err = netdev_get_stats(vport, &stats);
-		if (!err)
-			vport_set_stats(vport, &stats);
-	}
+	dev_disable_lro(vport->dev);
+	dev_set_promiscuity(vport->dev, 1);
+	vport->dev->priv_flags |= IFF_OVS_DATAPATH;
+	rtnl_unlock();
 
 	return vport;
 
+error_master_upper_dev_unlink:
+	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:
-	vport_free(vport);
-error:
+	ovs_vport_free(vport);
 	return ERR_PTR(err);
 }
+EXPORT_SYMBOL_GPL(ovs_netdev_link);
 
-static int netdev_destroy(struct vport *vport)
-{
-	struct netdev_vport *netdev_vport = netdev_vport_priv(vport);
-
-	dev_put(netdev_vport->dev);
-	vport_free(vport);
-
-	return 0;
-}
-
-static int netdev_attach(struct vport *vport)
+static struct vport *netdev_create(const struct vport_parms *parms)
 {
-	struct netdev_vport *netdev_vport = netdev_vport_priv(vport);
-	int err;
-
-	rcu_assign_pointer(netdev_vport->dev->br_port,
-			   (struct net_bridge_port *)vport);
-	err = netdev_rx_handler_register(netdev_vport->dev, netdev_frame_hook,
-					 NULL);
-	if (err)
-		return err;
-
-	dev_set_promiscuity(netdev_vport->dev, 1);
-	dev_disable_lro(netdev_vport->dev);
-
-	return 0;
-}
-
-static int netdev_detach(struct vport *vport)
-{
-	struct netdev_vport *netdev_vport = netdev_vport_priv(vport);
+	struct vport *vport;
 
-	netdev_rx_handler_unregister(netdev_vport->dev);
-	rcu_assign_pointer(netdev_vport->dev->br_port, NULL);
-	dev_set_promiscuity(netdev_vport->dev, -1);
+	vport = ovs_vport_alloc(0, &ovs_netdev_vport_ops, parms);
+	if (IS_ERR(vport))
+		return vport;
 
-	return 0;
+	return ovs_netdev_link(vport, parms->name);
 }
 
-int netdev_set_mtu(struct vport *vport, int mtu)
+static void vport_netdev_free(struct rcu_head *rcu)
 {
-	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)
-{
-	struct netdev_vport *netdev_vport = netdev_vport_priv(vport);
-	struct sockaddr sa;
-
-	sa.sa_family = ARPHRD_ETHER;
-	memcpy(sa.sa_data, addr, ETH_ALEN);
-
-	return dev_set_mac_address(netdev_vport->dev, &sa);
-}
+	struct vport *vport = container_of(rcu, struct vport, rcu);
 
-const char *netdev_get_name(const struct vport *vport)
-{
-	const struct netdev_vport *netdev_vport = netdev_vport_priv(vport);
-	return netdev_vport->dev->name;
+	if (vport->dev)
+		dev_put(vport->dev);
+	ovs_vport_free(vport);
 }
 
-const unsigned char *netdev_get_addr(const struct vport *vport)
+void ovs_netdev_detach_dev(struct vport *vport)
 {
-	const struct netdev_vport *netdev_vport = netdev_vport_priv(vport);
-	return netdev_vport->dev->dev_addr;
+	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);
 
-struct kobject *netdev_get_kobj(const struct vport *vport)
+static void netdev_destroy(struct vport *vport)
 {
-	const struct netdev_vport *netdev_vport = netdev_vport_priv(vport);
-	return &netdev_vport->dev->NETDEV_DEV_MEMBER.kobj;
-}
+	rtnl_lock();
+	if (vport->dev->priv_flags & IFF_OVS_DATAPATH)
+		ovs_netdev_detach_dev(vport);
+	rtnl_unlock();
 
-int netdev_get_stats(const struct vport *vport, struct odp_vport_stats *stats)
-{
-	const struct netdev_vport *netdev_vport = netdev_vport_priv(vport);
-	const struct net_device_stats *netdev_stats;
-
-	netdev_stats = dev_get_stats(netdev_vport->dev);
-
-	stats->rx_bytes		= netdev_stats->rx_bytes;
-	stats->rx_packets	= netdev_stats->rx_packets;
-	stats->tx_bytes		= netdev_stats->tx_bytes;
-	stats->tx_packets	= netdev_stats->tx_packets;
-	stats->rx_dropped	= netdev_stats->rx_dropped;
-	stats->rx_errors	= netdev_stats->rx_errors;
-	stats->rx_frame_err	= netdev_stats->rx_frame_errors;
-	stats->rx_over_err	= netdev_stats->rx_over_errors;
-	stats->rx_crc_err	= netdev_stats->rx_crc_errors;
-	stats->tx_dropped	= netdev_stats->tx_dropped;
-	stats->tx_errors	= netdev_stats->tx_errors;
-	stats->collisions	= netdev_stats->collisions;
-
-	return 0;
+	call_rcu(&vport->rcu, vport_netdev_free);
 }
 
-unsigned netdev_get_dev_flags(const struct vport *vport)
+void ovs_netdev_tunnel_destroy(struct vport *vport)
 {
-	const struct netdev_vport *netdev_vport = netdev_vport_priv(vport);
-	return dev_get_flags(netdev_vport->dev);
-}
+	rtnl_lock();
+	if (vport->dev->priv_flags & IFF_OVS_DATAPATH)
+		ovs_netdev_detach_dev(vport);
 
-int netdev_is_running(const struct vport *vport)
-{
-	const struct netdev_vport *netdev_vport = netdev_vport_priv(vport);
-	return netif_running(netdev_vport->dev);
-}
+	/* We can be invoked by both explicit vport deletion and
+	 * underlying netdev deregistration; delete the link only
+	 * if it's not already shutting down.
+	 */
 
-unsigned char netdev_get_operstate(const struct vport *vport)
-{
-	const struct netdev_vport *netdev_vport = netdev_vport_priv(vport);
-	return netdev_vport->dev->operstate;
-}
+	if (vport->dev->reg_state == NETREG_REGISTERED)
+		rtnl_delete_link(vport->dev);
 
-int netdev_get_ifindex(const struct vport *vport)
-{
-	const struct netdev_vport *netdev_vport = netdev_vport_priv(vport);
-	return netdev_vport->dev->ifindex;
-}
+	dev_put(vport->dev);
+	vport->dev = NULL;
+	rtnl_unlock();
 
-int netdev_get_iflink(const struct vport *vport)
-{
-	const struct netdev_vport *netdev_vport = netdev_vport_priv(vport);
-	return netdev_vport->dev->iflink;
+	call_rcu(&vport->rcu, vport_netdev_free);
 }
+EXPORT_SYMBOL_GPL(ovs_netdev_tunnel_destroy);
 
-int netdev_get_mtu(const struct vport *vport)
+/* Returns null if this device is not attached to a datapath. */
+struct vport *ovs_netdev_get_vport(struct net_device *dev)
 {
-	const struct netdev_vport *netdev_vport = netdev_vport_priv(vport);
-	return netdev_vport->dev->mtu;
+#if defined HAVE_NETDEV_RX_HANDLER_REGISTER || \
+    defined HAVE_RHEL_OVS_HOOK
+#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;
+#else
+	return (struct vport *)rcu_dereference_rtnl(dev->br_port);
+#endif
 }
 
-/* Must be called with rcu_read_lock. */
-static void netdev_port_receive(struct vport *vport, struct sk_buff *skb)
-{
-	/* 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;
-
-	skb_warn_if_lro(skb);
-
-	/* Push the Ethernet header back on. */
-	skb_push(skb, ETH_HLEN);
-	skb_reset_mac_header(skb);
-	compute_ip_summed(skb, false);
-
-	vport_receive(vport, skb);
-}
+static struct vport_ops ovs_netdev_vport_ops = {
+	.type		= OVS_VPORT_TYPE_NETDEV,
+	.create		= netdev_create,
+	.destroy	= netdev_destroy,
+	.send		= dev_queue_xmit,
+};
 
-static int netdev_send(struct vport *vport, struct sk_buff *skb)
+int __init ovs_netdev_init(void)
 {
-	struct netdev_vport *netdev_vport = netdev_vport_priv(vport);
-	int len = skb->len;
-
-	skb->dev = netdev_vport->dev;
-	forward_ip_summed(skb);
-	dev_queue_xmit(skb);
-
-	return len;
+	return ovs_vport_ops_register(&ovs_netdev_vport_ops);
 }
 
-/* Returns null if this device is not attached to a datapath. */
-struct vport *netdev_get_vport(struct net_device *dev)
+void ovs_netdev_exit(void)
 {
-	return (struct vport *)dev->br_port;
+	ovs_vport_ops_unregister(&ovs_netdev_vport_ops);
 }
 
-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,
-	.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,
-	.send		= netdev_send,
-};
-
+#if !defined HAVE_NETDEV_RX_HANDLER_REGISTER && \
+    !defined HAVE_RHEL_OVS_HOOK
 /*
- * 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.
  */
 typeof(br_should_route_hook) br_should_route_hook;
 EXPORT_SYMBOL(br_should_route_hook);
+#endif