Merge branch 'tunnel-offload-future-proof'

Alexander Duyck says:

====================
Future-proof tunnel offload handlers

These patches are meant to address two things.  First we are currently
using the ndo_add/del_vxlan_port calls with VXLAN-GPE tunnels and we
cannot really support that as it is likely to cause more harm than
good since VXLAN-GPE can support tunnels without a MAC address on the
inner header.

As such we need to add a new offload to advertise this, but in doing so it
would mean introducing 3 new functions for the driver to request the ports,
and then for the tunnel to push the changes to add and delete the ports to
the device.  However instead of taking that approach I think it would be
much better if we just made one common function for fetching the ports, and
provided a generic means to push the tunnels to the device.  So in order to
make this work this patch set does several things.

First it merges the existing VXLAN and GENEVE functionality into one set of
functions and passes an enum in order to specify the type of tunnel we want
to offload.  By doing this we only have to extend this enum in the future
if we want to add additional types.

Second it goes through the drivers replacing all of the tunnel specific
offload calls with implementations that support the generic calls so that
we can drop the VXLAN and GENEVE specific calls entirely.

Finally I go through in the last patch and replace the VXLAN specific
offload request that was being used for VXLAN-GPE with one that specifies
if we want to offload VXLAN or VXLAN-GPE so that the hardware can decide if
it can actually support it or not.

I also ended up with some minor clean-up built into the driver patches for
this.  Most of it is to either fix misuse of build flags, specifying a type
to ignore instead of the type that should be used, or in the case of ixgbe
I actually moved a rtnl_lock/unlock in order to avoid taking it unless it
was actually needed.

v2:
I did my best to remove the word "offload" from any of the calls or
notifiers as this isn't really an offload.  It
 is a workaround for the fact that the drivers don't provide basic features
like CHECKSUM_COMPLETE.  I also added a disclaimer to the section defining
the function prototypes explaining that these are essentially workarounds.

I ended up going through and stripping all of the VXLAN and GENEVE build
flags from the drivers.  There isn't much point in carrying them.  In
addition I dropped the use of the vxlan.h or geneve.h header files in favor
of udp_tunnel.h in the cases where a driver didn't need anything from
either of those headers.

I updated the tunnel add/del functions so that they pass a udp_tunnel_info
structure instead of a list of arguments.  This way we should be able to
add additional information in the future with little impact on the other
drivers.

I updated bnxt so that it doesn't use a hard-coded port number for GENEVE.

I have been able to test mlx4e, mlx5e, and i40e and verified functionality
on these drivers.  Though there are patches for the net tree I submitted
due to unrelated bugs I found in the mlx4e and i40e/i40evf drivers.

v3:
Fixed a typo that caused us to add geneve port when we should have been
deleting it.

Ended up dropping geneve and vxlan wrappers for
udp_tunnel_notify_rx_add/del_port and instead just called them directly.

Updated comments for functions to call out RTNL instead of port lock.

Updated patch description to remove changes that were moved into a second
patch.

Rebased on latest net-next to fix merge conflict on bnxt driver.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

4 files changed, 63 insertions, 59 deletions

diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h
index 890158e99159..e84d9d23c2d5 100644
--- a/include/linux/netdevice.h
+++ b/include/linux/netdevice.h
@@ -61,6 +61,8 @@ struct wireless_dev;
 /* 802.15.4 specific */
 struct wpan_dev;
 struct mpls_dev;
+/* UDP Tunnel offloads */
+struct udp_tunnel_info;
 
 void netdev_set_default_ethtool_ops(struct net_device *dev,
 				    const struct ethtool_ops *ops);
@@ -1024,31 +1026,18 @@ struct tc_to_netdev {
  *	not implement this, it is assumed that the hw is not able to have
  *	multiple net devices on single physical port.
  *
- * void (*ndo_add_vxlan_port)(struct  net_device *dev,
- *			      sa_family_t sa_family, __be16 port);
- *	Called by vxlan to notify a driver about the UDP port and socket
- *	address family that vxlan is listening to. It is called only when
- *	a new port starts listening. The operation is protected by the
- *	vxlan_net->sock_lock.
- *
- * void (*ndo_add_geneve_port)(struct net_device *dev,
- *			       sa_family_t sa_family, __be16 port);
- *	Called by geneve to notify a driver about the UDP port and socket
- *	address family that geneve is listnening to. It is called only when
- *	a new port starts listening. The operation is protected by the
- *	geneve_net->sock_lock.
- *
- * void (*ndo_del_geneve_port)(struct net_device *dev,
- *			       sa_family_t sa_family, __be16 port);
- *	Called by geneve to notify the driver about a UDP port and socket
- *	address family that geneve is not listening to anymore. The operation
- *	is protected by the geneve_net->sock_lock.
- *
- * void (*ndo_del_vxlan_port)(struct  net_device *dev,
- *			      sa_family_t sa_family, __be16 port);
- *	Called by vxlan to notify the driver about a UDP port and socket
- *	address family that vxlan is not listening to anymore. The operation
- *	is protected by the vxlan_net->sock_lock.
+ * void (*ndo_udp_tunnel_add)(struct net_device *dev,
+ *			      struct udp_tunnel_info *ti);
+ *	Called by UDP tunnel to notify a driver about the UDP port and socket
+ *	address family that a UDP tunnel is listnening to. It is called only
+ *	when a new port starts listening. The operation is protected by the
+ *	RTNL.
+ *
+ * void (*ndo_udp_tunnel_del)(struct net_device *dev,
+ *			      struct udp_tunnel_info *ti);
+ *	Called by UDP tunnel to notify the driver about a UDP port and socket
+ *	address family that the UDP tunnel is not listening to anymore. The
+ *	operation is protected by the RTNL.
  *
  * void* (*ndo_dfwd_add_station)(struct net_device *pdev,
  *				 struct net_device *dev)
@@ -1257,18 +1246,10 @@ struct net_device_ops {
 							struct netdev_phys_item_id *ppid);
 	int			(*ndo_get_phys_port_name)(struct net_device *dev,
 							  char *name, size_t len);
-	void			(*ndo_add_vxlan_port)(struct  net_device *dev,
-						      sa_family_t sa_family,
-						      __be16 port);
-	void			(*ndo_del_vxlan_port)(struct  net_device *dev,
-						      sa_family_t sa_family,
-						      __be16 port);
-	void			(*ndo_add_geneve_port)(struct  net_device *dev,
-						       sa_family_t sa_family,
-						       __be16 port);
-	void			(*ndo_del_geneve_port)(struct  net_device *dev,
-						       sa_family_t sa_family,
-						       __be16 port);
+	void			(*ndo_udp_tunnel_add)(struct net_device *dev,
+						      struct udp_tunnel_info *ti);
+	void			(*ndo_udp_tunnel_del)(struct net_device *dev,
+						      struct udp_tunnel_info *ti);
 	void*			(*ndo_dfwd_add_station)(struct net_device *pdev,
 							struct net_device *dev);
 	void			(*ndo_dfwd_del_station)(struct net_device *pdev,
@@ -2255,8 +2236,7 @@ struct netdev_lag_lower_state_info {
 #define NETDEV_BONDING_INFO	0x0019
 #define NETDEV_PRECHANGEUPPER	0x001A
 #define NETDEV_CHANGELOWERSTATE	0x001B
-#define NETDEV_OFFLOAD_PUSH_VXLAN	0x001C
-#define NETDEV_OFFLOAD_PUSH_GENEVE	0x001D
+#define NETDEV_UDP_TUNNEL_PUSH_INFO	0x001C
 
 int register_netdevice_notifier(struct notifier_block *nb);
 int unregister_netdevice_notifier(struct notifier_block *nb);
diff --git a/include/net/geneve.h b/include/net/geneve.h
index cb544a530146..ec0327d4331b 100644
--- a/include/net/geneve.h
+++ b/include/net/geneve.h
@@ -1,10 +1,7 @@
 #ifndef __NET_GENEVE_H
 #define __NET_GENEVE_H  1
 
-#ifdef CONFIG_INET
 #include <net/udp_tunnel.h>
-#endif
-
 
 /* Geneve Header:
  *  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
@@ -62,12 +59,6 @@ struct genevehdr {
 	struct geneve_opt options[];
 };
 
-static inline void geneve_get_rx_port(struct net_device *netdev)
-{
-	ASSERT_RTNL();
-	call_netdevice_notifiers(NETDEV_OFFLOAD_PUSH_GENEVE, netdev);
-}
-
 #ifdef CONFIG_INET
 struct net_device *geneve_dev_create_fb(struct net *net, const char *name,
 					u8 name_assign_type, u16 dst_port);
diff --git a/include/net/udp_tunnel.h b/include/net/udp_tunnel.h
index 9d14f707e534..02c5be037451 100644
--- a/include/net/udp_tunnel.h
+++ b/include/net/udp_tunnel.h
@@ -84,6 +84,46 @@ struct udp_tunnel_sock_cfg {
 void setup_udp_tunnel_sock(struct net *net, struct socket *sock,
 			   struct udp_tunnel_sock_cfg *sock_cfg);
 
+/* -- List of parsable UDP tunnel types --
+ *
+ * Adding to this list will result in serious debate.  The main issue is
+ * that this list is essentially a list of workarounds for either poorly
+ * designed tunnels, or poorly designed device offloads.
+ *
+ * The parsing supported via these types should really be used for Rx
+ * traffic only as the network stack will have already inserted offsets for
+ * the location of the headers in the skb.  In addition any ports that are
+ * pushed should be kept within the namespace without leaking to other
+ * devices such as VFs or other ports on the same device.
+ *
+ * It is strongly encouraged to use CHECKSUM_COMPLETE for Rx to avoid the
+ * need to use this for Rx checksum offload.  It should not be necessary to
+ * call this function to perform Tx offloads on outgoing traffic.
+ */
+enum udp_parsable_tunnel_type {
+	UDP_TUNNEL_TYPE_VXLAN,		/* RFC 7348 */
+	UDP_TUNNEL_TYPE_GENEVE,		/* draft-ietf-nvo3-geneve */
+	UDP_TUNNEL_TYPE_VXLAN_GPE,	/* draft-ietf-nvo3-vxlan-gpe */
+};
+
+struct udp_tunnel_info {
+	unsigned short type;
+	sa_family_t sa_family;
+	__be16 port;
+};
+
+/* Notify network devices of offloadable types */
+void udp_tunnel_push_rx_port(struct net_device *dev, struct socket *sock,
+			     unsigned short type);
+void udp_tunnel_notify_add_rx_port(struct socket *sock, unsigned short type);
+void udp_tunnel_notify_del_rx_port(struct socket *sock, unsigned short type);
+
+static inline void udp_tunnel_get_rx_info(struct net_device *dev)
+{
+	ASSERT_RTNL();
+	call_netdevice_notifiers(NETDEV_UDP_TUNNEL_PUSH_INFO, dev);
+}
+
 /* Transmit the skb using UDP encapsulation. */
 void udp_tunnel_xmit_skb(struct rtable *rt, struct sock *sk, struct sk_buff *skb,
 			 __be32 src, __be32 dst, __u8 tos, __u8 ttl,
@@ -105,12 +145,14 @@ struct metadata_dst *udp_tun_rx_dst(struct sk_buff *skb, unsigned short family,
 				    __be16 flags, __be64 tunnel_id,
 				    int md_size);
 
+#ifdef CONFIG_INET
 static inline int udp_tunnel_handle_offloads(struct sk_buff *skb, bool udp_csum)
 {
 	int type = udp_csum ? SKB_GSO_UDP_TUNNEL_CSUM : SKB_GSO_UDP_TUNNEL;
 
 	return iptunnel_handle_offloads(skb, type);
 }
+#endif
 
 static inline void udp_tunnel_encap_enable(struct socket *sock)
 {
diff --git a/include/net/vxlan.h b/include/net/vxlan.h
index b8803165df91..b96d0360c095 100644
--- a/include/net/vxlan.h
+++ b/include/net/vxlan.h
@@ -1,13 +1,10 @@
 #ifndef __NET_VXLAN_H
 #define __NET_VXLAN_H 1
 
-#include <linux/ip.h>
-#include <linux/ipv6.h>
 #include <linux/if_vlan.h>
-#include <linux/skbuff.h>
-#include <linux/netdevice.h>
-#include <linux/udp.h>
+#include <net/udp_tunnel.h>
 #include <net/dst_metadata.h>
+#include <net/udp_tunnel.h>
 
 /* VXLAN protocol (RFC 7348) header:
  * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
@@ -392,12 +389,6 @@ static inline __be32 vxlan_compute_rco(unsigned int start, unsigned int offset)
 	return vni_field;
 }
 
-static inline void vxlan_get_rx_port(struct net_device *netdev)
-{
-	ASSERT_RTNL();
-	call_netdevice_notifiers(NETDEV_OFFLOAD_PUSH_VXLAN, netdev);
-}
-
 static inline unsigned short vxlan_get_sk_family(struct vxlan_sock *vs)
 {
 	return vs->sock->sk->sk_family;