[PATCH net-next 0/3] vxlan: IPv6 fill_metadata_dst support

[PATCH net-next 0/3] vxlan: IPv6 fill_metadata_dst support

[Jiri Benc]

This adds IPv6 support to ndo_fill_metadata_dst in vxlan together with
restructuring to avoid duplicate code.

Jiri Benc (3):
  vxlan: consolidate output route calculation
  vxlan: move IPv6 outpute route calculation to a function
  vxlan: support ndo_fill_metadata_dst also for IPv6

 drivers/net/vxlan.c | 140 ++++++++++++++++++++++++++++++++++------------------
 1 file changed, 91 insertions(+), 49 deletions(-)

-- 
1.8.3.1

[PATCH net-next 1/3] vxlan: consolidate output route calculation

[Jiri Benc]

The code for output route lookup is duplicated for ndo_start_xmit and
ndo_fill_metadata_dst. Move it to a common function.

Signed-off-by: Jiri Benc <jbenc@redhat.com>
---
 drivers/net/vxlan.c | 49 +++++++++++++++++++++++++++++--------------------
 1 file changed, 29 insertions(+), 20 deletions(-)

diff --git a/drivers/net/vxlan.c b/drivers/net/vxlan.c
index 6369a5734d4c..085d0c5b82b7 100644
--- a/drivers/net/vxlan.c
+++ b/drivers/net/vxlan.c
@@ -1848,6 +1848,27 @@ static int vxlan_xmit_skb(struct rtable *rt, struct sock *sk, struct sk_buff *sk
 				   !(vxflags & VXLAN_F_UDP_CSUM));
 }
 
+static struct rtable *vxlan_get_route(struct vxlan_dev *vxlan,
+				      struct sk_buff *skb, int oif, u8 tos,
+				      __be32 daddr, __be32 *saddr)
+{
+	struct rtable *rt = NULL;
+	struct flowi4 fl4;
+
+	memset(&fl4, 0, sizeof(fl4));
+	fl4.flowi4_oif = oif;
+	fl4.flowi4_tos = RT_TOS(tos);
+	fl4.flowi4_mark = skb->mark;
+	fl4.flowi4_proto = IPPROTO_UDP;
+	fl4.daddr = daddr;
+	fl4.saddr = vxlan->cfg.saddr.sin.sin_addr.s_addr;
+
+	rt = ip_route_output_key(vxlan->net, &fl4);
+	if (!IS_ERR(rt))
+		*saddr = fl4.saddr;
+	return rt;
+}
+
 /* Bypass encapsulation if the destination is local */
 static void vxlan_encap_bypass(struct sk_buff *skb, struct vxlan_dev *src_vxlan,
 			       struct vxlan_dev *dst_vxlan)
@@ -1901,7 +1922,6 @@ static void vxlan_xmit_one(struct sk_buff *skb, struct net_device *dev,
 	struct sock *sk;
 	struct rtable *rt = NULL;
 	const struct iphdr *old_iph;
-	struct flowi4 fl4;
 	union vxlan_addr *dst;
 	union vxlan_addr remote_ip;
 	struct vxlan_metadata _md;
@@ -1973,6 +1993,8 @@ static void vxlan_xmit_one(struct sk_buff *skb, struct net_device *dev,
 	}
 
 	if (dst->sa.sa_family == AF_INET) {
+		__be32 saddr;
+
 		if (!vxlan->vn4_sock)
 			goto drop;
 		sk = vxlan->vn4_sock->sock->sk;
@@ -1980,15 +2002,9 @@ static void vxlan_xmit_one(struct sk_buff *skb, struct net_device *dev,
 		if (info && (info->key.tun_flags & TUNNEL_DONT_FRAGMENT))
 			df = htons(IP_DF);
 
-		memset(&fl4, 0, sizeof(fl4));
-		fl4.flowi4_oif = rdst ? rdst->remote_ifindex : 0;
-		fl4.flowi4_tos = RT_TOS(tos);
-		fl4.flowi4_mark = skb->mark;
-		fl4.flowi4_proto = IPPROTO_UDP;
-		fl4.daddr = dst->sin.sin_addr.s_addr;
-		fl4.saddr = vxlan->cfg.saddr.sin.sin_addr.s_addr;
-
-		rt = ip_route_output_key(vxlan->net, &fl4);
+		rt = vxlan_get_route(vxlan, skb,
+				     rdst ? rdst->remote_ifindex : 0, tos,
+				     dst->sin.sin_addr.s_addr, &saddr);
 		if (IS_ERR(rt)) {
 			netdev_dbg(dev, "no route to %pI4\n",
 				   &dst->sin.sin_addr.s_addr);
@@ -2020,7 +2036,7 @@ static void vxlan_xmit_one(struct sk_buff *skb, struct net_device *dev,
 
 		tos = ip_tunnel_ecn_encap(tos, old_iph, skb);
 		ttl = ttl ? : ip4_dst_hoplimit(&rt->dst);
-		err = vxlan_xmit_skb(rt, sk, skb, fl4.saddr,
+		err = vxlan_xmit_skb(rt, sk, skb, saddr,
 				     dst->sin.sin_addr.s_addr, tos, ttl, df,
 				     src_port, dst_port, htonl(vni << 8), md,
 				     !net_eq(vxlan->net, dev_net(vxlan->dev)),
@@ -2366,20 +2382,13 @@ static int egress_ipv4_tun_info(struct net_device *dev, struct sk_buff *skb,
 {
 	struct vxlan_dev *vxlan = netdev_priv(dev);
 	struct rtable *rt;
-	struct flowi4 fl4;
 
-	memset(&fl4, 0, sizeof(fl4));
-	fl4.flowi4_tos = RT_TOS(info->key.tos);
-	fl4.flowi4_mark = skb->mark;
-	fl4.flowi4_proto = IPPROTO_UDP;
-	fl4.daddr = info->key.u.ipv4.dst;
-
-	rt = ip_route_output_key(vxlan->net, &fl4);
+	rt = vxlan_get_route(vxlan, skb, 0, info->key.tos,
+			     info->key.u.ipv4.dst, &info->key.u.ipv4.src);
 	if (IS_ERR(rt))
 		return PTR_ERR(rt);
 	ip_rt_put(rt);
 
-	info->key.u.ipv4.src = fl4.saddr;
 	info->key.tp_src = sport;
 	info->key.tp_dst = dport;
 	return 0;
-- 
1.8.3.1

[PATCH net-next 2/3] vxlan: move IPv6 outpute route calculation to a function

[Jiri Benc]

Will be used also for ndo_fill_metadata_dst.

Signed-off-by: Jiri Benc <jbenc@redhat.com>
---
 drivers/net/vxlan.c | 45 +++++++++++++++++++++++++++++++++++----------
 1 file changed, 35 insertions(+), 10 deletions(-)

diff --git a/drivers/net/vxlan.c b/drivers/net/vxlan.c
index 085d0c5b82b7..2945cf7f9c73 100644
--- a/drivers/net/vxlan.c
+++ b/drivers/net/vxlan.c
@@ -1869,6 +1869,35 @@ static struct rtable *vxlan_get_route(struct vxlan_dev *vxlan,
 	return rt;
 }
 
+
+#if IS_ENABLED(CONFIG_IPV6)
+static struct dst_entry *vxlan6_get_route(struct vxlan_dev *vxlan,
+					  struct sk_buff *skb, int oif, u8 tos,
+					  const struct in6_addr *daddr,
+					  struct in6_addr *saddr)
+{
+	struct dst_entry *ndst;
+	struct flowi6 fl6;
+	int err;
+
+	memset(&fl6, 0, sizeof(fl6));
+	fl6.flowi6_oif = oif;
+	fl6.daddr = *daddr;
+	fl6.saddr = vxlan->cfg.saddr.sin6.sin6_addr;
+	fl6.flowi6_mark = skb->mark;
+	fl6.flowi6_proto = IPPROTO_UDP;
+
+	err = ipv6_stub->ipv6_dst_lookup(vxlan->net,
+					 vxlan->vn6_sock->sock->sk,
+					 &ndst, &fl6);
+	if (err < 0)
+		return ERR_PTR(err);
+
+	*saddr = fl6.saddr;
+	return ndst;
+}
+#endif
+
 /* Bypass encapsulation if the destination is local */
 static void vxlan_encap_bypass(struct sk_buff *skb, struct vxlan_dev *src_vxlan,
 			       struct vxlan_dev *dst_vxlan)
@@ -2051,21 +2080,17 @@ static void vxlan_xmit_one(struct sk_buff *skb, struct net_device *dev,
 #if IS_ENABLED(CONFIG_IPV6)
 	} else {
 		struct dst_entry *ndst;
-		struct flowi6 fl6;
+		struct in6_addr saddr;
 		u32 rt6i_flags;
 
 		if (!vxlan->vn6_sock)
 			goto drop;
 		sk = vxlan->vn6_sock->sock->sk;
 
-		memset(&fl6, 0, sizeof(fl6));
-		fl6.flowi6_oif = rdst ? rdst->remote_ifindex : 0;
-		fl6.daddr = dst->sin6.sin6_addr;
-		fl6.saddr = vxlan->cfg.saddr.sin6.sin6_addr;
-		fl6.flowi6_mark = skb->mark;
-		fl6.flowi6_proto = IPPROTO_UDP;
-
-		if (ipv6_stub->ipv6_dst_lookup(vxlan->net, sk, &ndst, &fl6)) {
+		ndst = vxlan6_get_route(vxlan, skb,
+					rdst ? rdst->remote_ifindex : 0, tos,
+					&dst->sin6.sin6_addr, &saddr);
+		if (IS_ERR(ndst)) {
 			netdev_dbg(dev, "no route to %pI6\n",
 				   &dst->sin6.sin6_addr);
 			dev->stats.tx_carrier_errors++;
@@ -2097,7 +2122,7 @@ static void vxlan_xmit_one(struct sk_buff *skb, struct net_device *dev,
 		}
 
 		ttl = ttl ? : ip6_dst_hoplimit(ndst);
-		err = vxlan6_xmit_skb(ndst, sk, skb, dev, &fl6.saddr, &fl6.daddr,
+		err = vxlan6_xmit_skb(ndst, sk, skb, dev, &saddr, &dst->sin6.sin6_addr,
 				      0, ttl, src_port, dst_port, htonl(vni << 8), md,
 				      !net_eq(vxlan->net, dev_net(vxlan->dev)),
 				      flags);
-- 
1.8.3.1

[PATCH net-next 3/3] vxlan: support ndo_fill_metadata_dst also for IPv6

[Jiri Benc]

Fill the metadata correctly even when tunneling over IPv6. Also, check that
the provided metadata is of an address family that is supported by the
tunnel.

Signed-off-by: Jiri Benc <jbenc@redhat.com>
---
 drivers/net/vxlan.c | 50 +++++++++++++++++++++++++++++---------------------
 1 file changed, 29 insertions(+), 21 deletions(-)

diff --git a/drivers/net/vxlan.c b/drivers/net/vxlan.c
index 2945cf7f9c73..d2edc35cfdaf 100644
--- a/drivers/net/vxlan.c
+++ b/drivers/net/vxlan.c
@@ -2401,37 +2401,45 @@ static int vxlan_change_mtu(struct net_device *dev, int new_mtu)
 	return 0;
 }
 
-static int egress_ipv4_tun_info(struct net_device *dev, struct sk_buff *skb,
-				struct ip_tunnel_info *info,
-				__be16 sport, __be16 dport)
-{
-	struct vxlan_dev *vxlan = netdev_priv(dev);
-	struct rtable *rt;
-
-	rt = vxlan_get_route(vxlan, skb, 0, info->key.tos,
-			     info->key.u.ipv4.dst, &info->key.u.ipv4.src);
-	if (IS_ERR(rt))
-		return PTR_ERR(rt);
-	ip_rt_put(rt);
-
-	info->key.tp_src = sport;
-	info->key.tp_dst = dport;
-	return 0;
-}
-
 static int vxlan_fill_metadata_dst(struct net_device *dev, struct sk_buff *skb)
 {
 	struct vxlan_dev *vxlan = netdev_priv(dev);
 	struct ip_tunnel_info *info = skb_tunnel_info(skb);
+	struct dst_entry *ndst;
+	struct rtable *rt;
 	__be16 sport, dport;
 
 	sport = udp_flow_src_port(dev_net(dev), skb, vxlan->cfg.port_min,
 				  vxlan->cfg.port_max, true);
 	dport = info->key.tp_dst ? : vxlan->cfg.dst_port;
 
-	if (ip_tunnel_info_af(info) == AF_INET)
-		return egress_ipv4_tun_info(dev, skb, info, sport, dport);
-	return -EINVAL;
+	if (ip_tunnel_info_af(info) == AF_INET) {
+		if (!vxlan->vn4_sock)
+			return -EINVAL;
+		rt = vxlan_get_route(vxlan, skb, 0, info->key.tos,
+				     info->key.u.ipv4.dst,
+				     &info->key.u.ipv4.src);
+		if (IS_ERR(rt))
+			return PTR_ERR(rt);
+		ip_rt_put(rt);
+	} else {
+		if (!IS_ENABLED(CONFIG_IPV6))
+			return -EPFNOSUPPORT;
+
+#if IS_ENABLED(CONFIG_IPV6)
+		if (!vxlan->vn6_sock)
+			return -EINVAL;
+		ndst = vxlan6_get_route(vxlan, skb, 0, info->key.tos,
+					&info->key.u.ipv6.dst,
+					&info->key.u.ipv6.src);
+		if (IS_ERR(ndst))
+			return PTR_ERR(ndst);
+		dst_release(ndst);
+#endif
+	}
+	info->key.tp_src = sport;
+	info->key.tp_dst = dport;
+	return 0;
 }
 
 static const struct net_device_ops vxlan_netdev_ops = {
-- 
1.8.3.1

Re: [PATCH net-next 3/3] vxlan: support ndo_fill_metadata_dst also for IPv6

[kbuild test robot]

[-- Attachment #1: Type: text/plain, Size: 1617 bytes --]

Hi Jiri,

[auto build test WARNING on net-next/master]

url:    https://github.com/0day-ci/linux/commits/Jiri-Benc/vxlan-IPv6-fill_metadata_dst-support/20151203-021449
config: x86_64-randconfig-i0-201548 (attached as .config)
reproduce:
        # save the attached .config to linux build tree
        make ARCH=x86_64 

All warnings (new ones prefixed by >>):

   drivers/net/vxlan.c: In function 'vxlan_fill_metadata_dst':
>> drivers/net/vxlan.c:2408:20: warning: unused variable 'ndst' [-Wunused-variable]
     struct dst_entry *ndst;
                       ^

vim +/ndst +2408 drivers/net/vxlan.c

  2392		if (dst->remote_ip.sa.sa_family == AF_INET6)
  2393			max_mtu = lowerdev->mtu - VXLAN6_HEADROOM;
  2394		else
  2395			max_mtu = lowerdev->mtu - VXLAN_HEADROOM;
  2396	
  2397		if (new_mtu < 68 || new_mtu > max_mtu)
  2398			return -EINVAL;
  2399	
  2400		dev->mtu = new_mtu;
  2401		return 0;
  2402	}
  2403	
  2404	static int vxlan_fill_metadata_dst(struct net_device *dev, struct sk_buff *skb)
  2405	{
  2406		struct vxlan_dev *vxlan = netdev_priv(dev);
  2407		struct ip_tunnel_info *info = skb_tunnel_info(skb);
> 2408		struct dst_entry *ndst;
  2409		struct rtable *rt;
  2410		__be16 sport, dport;
  2411	
  2412		sport = udp_flow_src_port(dev_net(dev), skb, vxlan->cfg.port_min,
  2413					  vxlan->cfg.port_max, true);
  2414		dport = info->key.tp_dst ? : vxlan->cfg.dst_port;
  2415	
  2416		if (ip_tunnel_info_af(info) == AF_INET) {

---
0-DAY kernel test infrastructure                Open Source Technology Center
https://lists.01.org/pipermail/kbuild-all                   Intel Corporation

[-- Attachment #2: .config.gz --]
[-- Type: application/octet-stream, Size: 27362 bytes --]

Re: [PATCH net-next 0/3] vxlan: IPv6 fill_metadata_dst support

[Pravin Shelar]

On Wed, Dec 2, 2015 at 10:12 AM, Jiri Benc <jbenc@redhat.com> wrote:
> This adds IPv6 support to ndo_fill_metadata_dst in vxlan together with
> restructuring to avoid duplicate code.
>

We need the ndo_fill_metadata_dst  IPv6 support in net branch, since
OVS IPv6 tunnel support is there already. Is there reason you targeted
this patch series for net-next?

Re: [PATCH net-next 0/3] vxlan: IPv6 fill_metadata_dst support

[Jiri Benc]

On Wed, 2 Dec 2015 15:02:22 -0800, Pravin Shelar wrote:
> We need the ndo_fill_metadata_dst  IPv6 support in net branch, since
> OVS IPv6 tunnel support is there already. Is there reason you targeted
> this patch series for net-next?

I guess you're right.

I need to resubmit anyway (seems I overlooked a warning when building
without IPv6), I'll submit patches 2 and 3 for net and patch 1 for
net-next (since it's a cleanup).

 Jiri

-- 
Jiri Benc