[PATCH v3 net-next 00/10] openvswitch: Conntrack integration improvements.

[PATCH v3 net-next 00/10] openvswitch: Conntrack integration improvements.

[Jarno Rajahalme]

This series improves the conntrack integration code in the openvswitch
module by fixing outdated comments (patch 1), bugs (patches 2, 3, and
7), clarifying code (patches 4, 5, and 6), improving performance
(patch 10), and adding new features enabling better translation from
firewall admission policy to network configuration requested by user
communities (patches 8 and 9).

v3: Rebase to the current net-next, add the comment only changing
    patch 1 and reshuffle some of the patches as requested by Joe.

Jarno Rajahalme (10):
  openvswitch: Fix comments for skb->_nfct
  openvswitch: Use inverted tuple in ovs_ct_find_existing() if NATted.
  openvswitch: Do not trigger events for unconfirmed connections.
  openvswitch: Unionize ovs_key_ct_label with a u32 array.
  openvswitch: Simplify labels length logic.
  openvswitch: Refactor labels initialization.
  openvswitch: Inherit master's labels.
  openvswitch: Add original direction conntrack tuple to sw_flow_key.
  openvswitch: Add force commit.
  openvswitch: Pack struct sw_flow_key.

 include/uapi/linux/openvswitch.h |  33 ++++-
 net/openvswitch/actions.c        |   2 +
 net/openvswitch/conntrack.c      | 298 ++++++++++++++++++++++++++++++---------
 net/openvswitch/conntrack.h      |  14 +-
 net/openvswitch/flow.c           |  34 ++++-
 net/openvswitch/flow.h           |  55 ++++++--
 net/openvswitch/flow_netlink.c   |  92 +++++++++---
 net/openvswitch/flow_netlink.h   |   7 +-
 8 files changed, 422 insertions(+), 113 deletions(-)

-- 
2.1.4

[PATCH v3 net-next 01/10] openvswitch: Fix comments for skb->_nfct

[Jarno Rajahalme]

Fix comments referring to skb 'nfct' and 'nfctinfo' fields now that
they are combined into '_nfct'.

Signed-off-by: Jarno Rajahalme <jarno@ovn.org>
---
 net/openvswitch/conntrack.c | 14 +++++++-------
 1 file changed, 7 insertions(+), 7 deletions(-)

diff --git a/net/openvswitch/conntrack.c b/net/openvswitch/conntrack.c
index fbffe0e..5de6d12 100644
--- a/net/openvswitch/conntrack.c
+++ b/net/openvswitch/conntrack.c
@@ -157,7 +157,7 @@ static void __ovs_ct_update_key(struct sw_flow_key *key, u8 state,
 	ovs_ct_get_labels(ct, &key->ct.labels);
 }
 
-/* Update 'key' based on skb->nfct.  If 'post_ct' is true, then OVS has
+/* Update 'key' based on skb->_nfct.  If 'post_ct' is true, then OVS has
  * previously sent the packet to conntrack via the ct action.  If
  * 'keep_nat_flags' is true, the existing NAT flags retained, else they are
  * initialized from the connection status.
@@ -421,12 +421,12 @@ ovs_ct_get_info(const struct nf_conntrack_tuple_hash *h)
 
 /* Find an existing connection which this packet belongs to without
  * re-attributing statistics or modifying the connection state.  This allows an
- * skb->nfct lost due to an upcall to be recovered during actions execution.
+ * skb->_nfct lost due to an upcall to be recovered during actions execution.
  *
  * Must be called with rcu_read_lock.
  *
- * On success, populates skb->nfct and skb->nfctinfo, and returns the
- * connection.  Returns NULL if there is no existing entry.
+ * On success, populates skb->_nfct and returns the connection.  Returns NULL
+ * if there is no existing entry.
  */
 static struct nf_conn *
 ovs_ct_find_existing(struct net *net, const struct nf_conntrack_zone *zone,
@@ -464,7 +464,7 @@ ovs_ct_find_existing(struct net *net, const struct nf_conntrack_zone *zone,
 	return ct;
 }
 
-/* Determine whether skb->nfct is equal to the result of conntrack lookup. */
+/* Determine whether skb->_nfct is equal to the result of conntrack lookup. */
 static bool skb_nfct_cached(struct net *net,
 			    const struct sw_flow_key *key,
 			    const struct ovs_conntrack_info *info,
@@ -475,7 +475,7 @@ static bool skb_nfct_cached(struct net *net,
 
 	ct = nf_ct_get(skb, &ctinfo);
 	/* If no ct, check if we have evidence that an existing conntrack entry
-	 * might be found for this skb.  This happens when we lose a skb->nfct
+	 * might be found for this skb.  This happens when we lose a skb->_nfct
 	 * due to an upcall.  If the connection was not confirmed, it is not
 	 * cached and needs to be run through conntrack again.
 	 */
@@ -699,7 +699,7 @@ static int ovs_ct_nat(struct net *net, struct sw_flow_key *key,
 /* Pass 'skb' through conntrack in 'net', using zone configured in 'info', if
  * not done already.  Update key with new CT state after passing the packet
  * through conntrack.
- * Note that if the packet is deemed invalid by conntrack, skb->nfct will be
+ * Note that if the packet is deemed invalid by conntrack, skb->_nfct will be
  * set to NULL and 0 will be returned.
  */
 static int __ovs_ct_lookup(struct net *net, struct sw_flow_key *key,
-- 
2.1.4

[PATCH v3 net-next 02/10] openvswitch: Use inverted tuple in ovs_ct_find_existing() if NATted.

[Jarno Rajahalme]

The conntrack lookup for existing connections fails to invert the
packet 5-tuple for NATted packets, and therefore fails to find the
existing conntrack entry.  Conntrack only stores 5-tuples for incoming
packets, and there are various situations where a lookup on a packet
that has already been transformed by NAT needs to be made.  Looking up
an existing conntrack entry upon executing packet received from the
userspace is one of them.

This patch fixes ovs_ct_find_existing() to invert the packet 5-tuple
for the conntrack lookup whenever the packet has already been
transformed by conntrack from its input form as evidenced by one of
the NAT flags being set in the conntrack state metadata.

Fixes: 05752523e565 ("openvswitch: Interface with NAT.")
Signed-off-by: Jarno Rajahalme <jarno@ovn.org>
---
 net/openvswitch/conntrack.c | 24 ++++++++++++++++++++++--
 1 file changed, 22 insertions(+), 2 deletions(-)

diff --git a/net/openvswitch/conntrack.c b/net/openvswitch/conntrack.c
index 5de6d12..4df9a54 100644
--- a/net/openvswitch/conntrack.c
+++ b/net/openvswitch/conntrack.c
@@ -430,7 +430,7 @@ ovs_ct_get_info(const struct nf_conntrack_tuple_hash *h)
  */
 static struct nf_conn *
 ovs_ct_find_existing(struct net *net, const struct nf_conntrack_zone *zone,
-		     u8 l3num, struct sk_buff *skb)
+		     u8 l3num, struct sk_buff *skb, bool natted)
 {
 	struct nf_conntrack_l3proto *l3proto;
 	struct nf_conntrack_l4proto *l4proto;
@@ -453,6 +453,17 @@ ovs_ct_find_existing(struct net *net, const struct nf_conntrack_zone *zone,
 		return NULL;
 	}
 
+	/* Must invert the tuple if skb has been transformed by NAT. */
+	if (natted) {
+		struct nf_conntrack_tuple inverse;
+
+		if (!nf_ct_invert_tuple(&inverse, &tuple, l3proto, l4proto)) {
+			pr_debug("ovs_ct_find_existing: Inversion failed!\n");
+			return NULL;
+		}
+		tuple = inverse;
+	}
+
 	/* look for tuple match */
 	h = nf_conntrack_find_get(net, zone, &tuple);
 	if (!h)
@@ -460,6 +471,13 @@ ovs_ct_find_existing(struct net *net, const struct nf_conntrack_zone *zone,
 
 	ct = nf_ct_tuplehash_to_ctrack(h);
 
+	/* Inverted packet tuple matches the reverse direction conntrack tuple,
+	 * select the other tuplehash to get the right 'ctinfo' bits for this
+	 * packet.
+	 */
+	if (natted)
+		h = &ct->tuplehash[!h->tuple.dst.dir];
+
 	nf_ct_set(skb, ct, ovs_ct_get_info(h));
 	return ct;
 }
@@ -482,7 +500,9 @@ static bool skb_nfct_cached(struct net *net,
 	if (!ct && key->ct.state & OVS_CS_F_TRACKED &&
 	    !(key->ct.state & OVS_CS_F_INVALID) &&
 	    key->ct.zone == info->zone.id)
-		ct = ovs_ct_find_existing(net, &info->zone, info->family, skb);
+		ct = ovs_ct_find_existing(net, &info->zone, info->family, skb,
+					  !!(key->ct.state
+					     & OVS_CS_F_NAT_MASK));
 	if (!ct)
 		return false;
 	if (!net_eq(net, read_pnet(&ct->ct_net)))
-- 
2.1.4

[PATCH v3 net-next 03/10] openvswitch: Do not trigger events for unconfirmed connections.

[Jarno Rajahalme]

Receiving change events before the 'new' event for the connection has
been received can be confusing.  Avoid triggering change events for
setting conntrack mark or labels before the conntrack entry has been
confirmed.

Fixes: 182e3042e15d ("openvswitch: Allow matching on conntrack mark")
Fixes: c2ac66735870 ("openvswitch: Allow matching on conntrack label")
Signed-off-by: Jarno Rajahalme <jarno@ovn.org>
---
 net/openvswitch/conntrack.c | 28 ++++++++++++++++++++++------
 1 file changed, 22 insertions(+), 6 deletions(-)

diff --git a/net/openvswitch/conntrack.c b/net/openvswitch/conntrack.c
index 4df9a54..a6ff374 100644
--- a/net/openvswitch/conntrack.c
+++ b/net/openvswitch/conntrack.c
@@ -245,7 +245,8 @@ static int ovs_ct_set_mark(struct sk_buff *skb, struct sw_flow_key *key,
 	new_mark = ct_mark | (ct->mark & ~(mask));
 	if (ct->mark != new_mark) {
 		ct->mark = new_mark;
-		nf_conntrack_event_cache(IPCT_MARK, ct);
+		if (nf_ct_is_confirmed(ct))
+			nf_conntrack_event_cache(IPCT_MARK, ct);
 		key->ct.mark = new_mark;
 	}
 
@@ -262,7 +263,6 @@ static int ovs_ct_set_labels(struct sk_buff *skb, struct sw_flow_key *key,
 	enum ip_conntrack_info ctinfo;
 	struct nf_conn_labels *cl;
 	struct nf_conn *ct;
-	int err;
 
 	/* The connection could be invalid, in which case set_label is no-op.*/
 	ct = nf_ct_get(skb, &ctinfo);
@@ -277,10 +277,26 @@ static int ovs_ct_set_labels(struct sk_buff *skb, struct sw_flow_key *key,
 	if (!cl || sizeof(cl->bits) < OVS_CT_LABELS_LEN)
 		return -ENOSPC;
 
-	err = nf_connlabels_replace(ct, (u32 *)labels, (u32 *)mask,
-				    OVS_CT_LABELS_LEN / sizeof(u32));
-	if (err)
-		return err;
+	if (nf_ct_is_confirmed(ct)) {
+		/* Triggers a change event, which makes sense only for
+		 * confirmed connections.
+		 */
+		int err = nf_connlabels_replace(ct, (u32 *)labels, (u32 *)mask,
+						OVS_CT_LABELS_LEN / sizeof(u32));
+		if (err)
+			return err;
+	} else {
+		u32 *dst = (u32 *)cl->bits;
+		const u32 *msk = (const u32 *)mask->ct_labels;
+		const u32 *lbl = (const u32 *)labels->ct_labels;
+		int i;
+
+		/* No-one else has access to the non-confirmed entry, copy
+		 * labels over, keeping any bits we are not explicitly setting.
+		 */
+		for (i = 0; i < OVS_CT_LABELS_LEN / sizeof(u32); i++)
+			dst[i] = (dst[i] & ~msk[i]) | (lbl[i] & msk[i]);
+	}
 
 	ovs_ct_get_labels(ct, &key->ct.labels);
 	return 0;
-- 
2.1.4

[PATCH v3 net-next 04/10] openvswitch: Unionize ovs_key_ct_label with a u32 array.

[Jarno Rajahalme]

Make the array of labels in struct ovs_key_ct_label an union, adding a
u32 array of the same byte size as the existing u8 array.  It is
faster to loop through the labels 32 bits at the time, which is also
the alignment of netlink attributes.

Signed-off-by: Jarno Rajahalme <jarno@ovn.org>
---
 include/uapi/linux/openvswitch.h |  8 ++++++--
 net/openvswitch/conntrack.c      | 15 ++++++++-------
 2 files changed, 14 insertions(+), 9 deletions(-)

diff --git a/include/uapi/linux/openvswitch.h b/include/uapi/linux/openvswitch.h
index 375d812..96aee34 100644
--- a/include/uapi/linux/openvswitch.h
+++ b/include/uapi/linux/openvswitch.h
@@ -446,9 +446,13 @@ struct ovs_key_nd {
 	__u8	nd_tll[ETH_ALEN];
 };
 
-#define OVS_CT_LABELS_LEN	16
+#define OVS_CT_LABELS_LEN_32	4
+#define OVS_CT_LABELS_LEN	(OVS_CT_LABELS_LEN_32 * sizeof(__u32))
 struct ovs_key_ct_labels {
-	__u8	ct_labels[OVS_CT_LABELS_LEN];
+	union {
+		__u8	ct_labels[OVS_CT_LABELS_LEN];
+		__u32	ct_labels_32[OVS_CT_LABELS_LEN_32];
+	};
 };
 
 /* OVS_KEY_ATTR_CT_STATE flags */
diff --git a/net/openvswitch/conntrack.c b/net/openvswitch/conntrack.c
index a6ff374..f23934c 100644
--- a/net/openvswitch/conntrack.c
+++ b/net/openvswitch/conntrack.c
@@ -281,20 +281,21 @@ static int ovs_ct_set_labels(struct sk_buff *skb, struct sw_flow_key *key,
 		/* Triggers a change event, which makes sense only for
 		 * confirmed connections.
 		 */
-		int err = nf_connlabels_replace(ct, (u32 *)labels, (u32 *)mask,
-						OVS_CT_LABELS_LEN / sizeof(u32));
+		int err = nf_connlabels_replace(ct, labels->ct_labels_32,
+						mask->ct_labels_32,
+						OVS_CT_LABELS_LEN_32);
 		if (err)
 			return err;
 	} else {
 		u32 *dst = (u32 *)cl->bits;
-		const u32 *msk = (const u32 *)mask->ct_labels;
-		const u32 *lbl = (const u32 *)labels->ct_labels;
+		const u32 *msk = mask->ct_labels_32;
+		const u32 *lbl = labels->ct_labels_32;
 		int i;
 
 		/* No-one else has access to the non-confirmed entry, copy
 		 * labels over, keeping any bits we are not explicitly setting.
 		 */
-		for (i = 0; i < OVS_CT_LABELS_LEN / sizeof(u32); i++)
+		for (i = 0; i < OVS_CT_LABELS_LEN_32; i++)
 			dst[i] = (dst[i] & ~msk[i]) | (lbl[i] & msk[i]);
 	}
 
@@ -866,8 +867,8 @@ static bool labels_nonzero(const struct ovs_key_ct_labels *labels)
 {
 	size_t i;
 
-	for (i = 0; i < sizeof(*labels); i++)
-		if (labels->ct_labels[i])
+	for (i = 0; i < OVS_CT_LABELS_LEN_32; i++)
+		if (labels->ct_labels_32[i])
 			return true;
 
 	return false;
-- 
2.1.4

[PATCH v3 net-next 05/10] openvswitch: Simplify labels length logic.

[Jarno Rajahalme]

Since 23014011ba42 ("netfilter: conntrack: support a fixed size of 128
distinct labels"), the size of conntrack labels extension has fixed to
128 bits, so we do not need to check for labels sizes shorter than 128
at run-time.  This patch simplifies labels length logic accordingly,
but allows the conntrack labels size to be increased in the future
without breaking the build.  In the event of conntrack labels
increasing in size OVS would still be able to deal with the 128 first
label bits.

Suggested-by: Joe Stringer <joe@ovn.org>
Signed-off-by: Jarno Rajahalme <jarno@ovn.org>
---
 net/openvswitch/conntrack.c | 22 +++++++++++-----------
 1 file changed, 11 insertions(+), 11 deletions(-)

diff --git a/net/openvswitch/conntrack.c b/net/openvswitch/conntrack.c
index f23934c..c7db4da 100644
--- a/net/openvswitch/conntrack.c
+++ b/net/openvswitch/conntrack.c
@@ -129,22 +129,22 @@ static u32 ovs_ct_get_mark(const struct nf_conn *ct)
 #endif
 }
 
+/* Guard against conntrack labels max size shrinking below 128 bits. */
+#if NF_CT_LABELS_MAX_SIZE < 16
+#error NF_CT_LABELS_MAX_SIZE must be at least 16 bytes
+#endif
+
 static void ovs_ct_get_labels(const struct nf_conn *ct,
 			      struct ovs_key_ct_labels *labels)
 {
 	struct nf_conn_labels *cl = ct ? nf_ct_labels_find(ct) : NULL;
 
-	if (cl) {
-		size_t len = sizeof(cl->bits);
-
-		if (len > OVS_CT_LABELS_LEN)
-			len = OVS_CT_LABELS_LEN;
-		else if (len < OVS_CT_LABELS_LEN)
-			memset(labels, 0, OVS_CT_LABELS_LEN);
-		memcpy(labels, cl->bits, len);
-	} else {
+	if (cl)
+		memcpy(labels, cl->bits,
+		       sizeof(cl->bits) > OVS_CT_LABELS_LEN
+		       ? OVS_CT_LABELS_LEN : sizeof(cl->bits));
+	else
 		memset(labels, 0, OVS_CT_LABELS_LEN);
-	}
 }
 
 static void __ovs_ct_update_key(struct sw_flow_key *key, u8 state,
@@ -274,7 +274,7 @@ static int ovs_ct_set_labels(struct sk_buff *skb, struct sw_flow_key *key,
 		nf_ct_labels_ext_add(ct);
 		cl = nf_ct_labels_find(ct);
 	}
-	if (!cl || sizeof(cl->bits) < OVS_CT_LABELS_LEN)
+	if (!cl)
 		return -ENOSPC;
 
 	if (nf_ct_is_confirmed(ct)) {
-- 
2.1.4

[PATCH v3 net-next 06/10] openvswitch: Refactor labels initialization.

[Jarno Rajahalme]

Refactoring conntrack labels initialization makes chenges in later
patches easier to review.

Signed-off-by: Jarno Rajahalme <jarno@ovn.org>
---
 net/openvswitch/conntrack.c | 104 ++++++++++++++++++++++++++------------------
 1 file changed, 62 insertions(+), 42 deletions(-)

diff --git a/net/openvswitch/conntrack.c b/net/openvswitch/conntrack.c
index c7db4da..0e038ee 100644
--- a/net/openvswitch/conntrack.c
+++ b/net/openvswitch/conntrack.c
@@ -229,19 +229,12 @@ int ovs_ct_put_key(const struct sw_flow_key *key, struct sk_buff *skb)
 	return 0;
 }
 
-static int ovs_ct_set_mark(struct sk_buff *skb, struct sw_flow_key *key,
+static int ovs_ct_set_mark(struct nf_conn *ct, struct sw_flow_key *key,
 			   u32 ct_mark, u32 mask)
 {
 #if IS_ENABLED(CONFIG_NF_CONNTRACK_MARK)
-	enum ip_conntrack_info ctinfo;
-	struct nf_conn *ct;
 	u32 new_mark;
 
-	/* The connection could be invalid, in which case set_mark is no-op. */
-	ct = nf_ct_get(skb, &ctinfo);
-	if (!ct)
-		return 0;
-
 	new_mark = ct_mark | (ct->mark & ~(mask));
 	if (ct->mark != new_mark) {
 		ct->mark = new_mark;
@@ -256,50 +249,66 @@ static int ovs_ct_set_mark(struct sk_buff *skb, struct sw_flow_key *key,
 #endif
 }
 
-static int ovs_ct_set_labels(struct sk_buff *skb, struct sw_flow_key *key,
-			     const struct ovs_key_ct_labels *labels,
-			     const struct ovs_key_ct_labels *mask)
+static struct nf_conn_labels *ovs_ct_get_conn_labels(struct nf_conn *ct)
 {
-	enum ip_conntrack_info ctinfo;
 	struct nf_conn_labels *cl;
-	struct nf_conn *ct;
-
-	/* The connection could be invalid, in which case set_label is no-op.*/
-	ct = nf_ct_get(skb, &ctinfo);
-	if (!ct)
-		return 0;
 
 	cl = nf_ct_labels_find(ct);
 	if (!cl) {
 		nf_ct_labels_ext_add(ct);
 		cl = nf_ct_labels_find(ct);
 	}
+
+	return cl;
+}
+
+/* Initialize labels for a new, to be committed conntrack entry.  Note that
+ * since the new connection is not yet confirmed, and thus no-one else has
+ * access to it's labels, we simply write them over.  Also, we refrain from
+ * triggering events, as receiving change events before the create event would
+ * be confusing.
+ */
+static int ovs_ct_init_labels(struct nf_conn *ct, struct sw_flow_key *key,
+			      const struct ovs_key_ct_labels *labels,
+			      const struct ovs_key_ct_labels *mask)
+{
+	struct nf_conn_labels *cl;
+	u32 *dst;
+	int i;
+
+	cl = ovs_ct_get_conn_labels(ct);
 	if (!cl)
 		return -ENOSPC;
 
-	if (nf_ct_is_confirmed(ct)) {
-		/* Triggers a change event, which makes sense only for
-		 * confirmed connections.
-		 */
-		int err = nf_connlabels_replace(ct, labels->ct_labels_32,
-						mask->ct_labels_32,
-						OVS_CT_LABELS_LEN_32);
-		if (err)
-			return err;
-	} else {
-		u32 *dst = (u32 *)cl->bits;
-		const u32 *msk = mask->ct_labels_32;
-		const u32 *lbl = labels->ct_labels_32;
-		int i;
+	dst = (u32 *)cl->bits;
+	for (i = 0; i < OVS_CT_LABELS_LEN_32; i++)
+		dst[i] = (dst[i] & ~mask->ct_labels_32[i]) |
+			(labels->ct_labels_32[i] & mask->ct_labels_32[i]);
 
-		/* No-one else has access to the non-confirmed entry, copy
-		 * labels over, keeping any bits we are not explicitly setting.
-		 */
-		for (i = 0; i < OVS_CT_LABELS_LEN_32; i++)
-			dst[i] = (dst[i] & ~msk[i]) | (lbl[i] & msk[i]);
-	}
+	memcpy(&key->ct.labels, cl->bits, OVS_CT_LABELS_LEN);
+
+	return 0;
+}
+
+static int ovs_ct_set_labels(struct nf_conn *ct, struct sw_flow_key *key,
+			     const struct ovs_key_ct_labels *labels,
+			     const struct ovs_key_ct_labels *mask)
+{
+	struct nf_conn_labels *cl;
+	int err;
+
+	cl = ovs_ct_get_conn_labels(ct);
+	if (!cl)
+		return -ENOSPC;
+
+	err = nf_connlabels_replace(ct, labels->ct_labels_32,
+				    mask->ct_labels_32,
+				    OVS_CT_LABELS_LEN_32);
+	if (err)
+		return err;
+
+	memcpy(&key->ct.labels, cl->bits, OVS_CT_LABELS_LEN);
 
-	ovs_ct_get_labels(ct, &key->ct.labels);
 	return 0;
 }
 
@@ -879,25 +888,36 @@ static int ovs_ct_commit(struct net *net, struct sw_flow_key *key,
 			 const struct ovs_conntrack_info *info,
 			 struct sk_buff *skb)
 {
+	enum ip_conntrack_info ctinfo;
+	struct nf_conn *ct;
 	int err;
 
 	err = __ovs_ct_lookup(net, key, info, skb);
 	if (err)
 		return err;
 
+	/* The connection could be invalid, in which case this is a no-op.*/
+	ct = nf_ct_get(skb, &ctinfo);
+	if (!ct)
+		return 0;
+
 	/* Apply changes before confirming the connection so that the initial
 	 * conntrack NEW netlink event carries the values given in the CT
 	 * action.
 	 */
 	if (info->mark.mask) {
-		err = ovs_ct_set_mark(skb, key, info->mark.value,
+		err = ovs_ct_set_mark(ct, key, info->mark.value,
 				      info->mark.mask);
 		if (err)
 			return err;
 	}
 	if (labels_nonzero(&info->labels.mask)) {
-		err = ovs_ct_set_labels(skb, key, &info->labels.value,
-					&info->labels.mask);
+		if (!nf_ct_is_confirmed(ct))
+			err = ovs_ct_init_labels(ct, key, &info->labels.value,
+						 &info->labels.mask);
+		else
+			err = ovs_ct_set_labels(ct, key, &info->labels.value,
+						&info->labels.mask);
 		if (err)
 			return err;
 	}
-- 
2.1.4

[PATCH v3 net-next 07/10] openvswitch: Inherit master's labels.

[Jarno Rajahalme]

We avoid calling into nf_conntrack_in() for expected connections, as
that would remove the expectation that we want to stick around until
we are ready to commit the connection.  Instead, we do a lookup in the
expectation table directly.  However, after a successful expectation
lookup we have set the flow key label field from the master
connection, whereas nf_conntrack_in() does not do this.  This leads to
master's labels being inherited after an expectation lookup, but those
labels not being inherited after the corresponding conntrack action
with a commit flag.

This patch resolves the problem by changing the commit code path to
also inherit the master's labels to the expected connection.
Resolving this conflict in favor or inheriting the labels allows more
information be passed from the master connection to related
connections, which would otherwise be much harder if the 32 bits in
the connmark are not enough.  Labels can still be set explicitly, so
this change only affects the default values of the labels in presense
of a master connection.

Fixes: 7f8a436eaa2c ("openvswitch: Add conntrack action")
Signed-off-by: Jarno Rajahalme <jarno@ovn.org>
---
 net/openvswitch/conntrack.c | 45 +++++++++++++++++++++++++++++++--------------
 1 file changed, 31 insertions(+), 14 deletions(-)

diff --git a/net/openvswitch/conntrack.c b/net/openvswitch/conntrack.c
index 0e038ee..5fbadcd 100644
--- a/net/openvswitch/conntrack.c
+++ b/net/openvswitch/conntrack.c
@@ -73,6 +73,8 @@ struct ovs_conntrack_info {
 #endif
 };
 
+static bool labels_nonzero(const struct ovs_key_ct_labels *labels);
+
 static void __ovs_ct_free_action(struct ovs_conntrack_info *ct_info);
 
 static u16 key_to_nfproto(const struct sw_flow_key *key)
@@ -272,18 +274,32 @@ static int ovs_ct_init_labels(struct nf_conn *ct, struct sw_flow_key *key,
 			      const struct ovs_key_ct_labels *labels,
 			      const struct ovs_key_ct_labels *mask)
 {
-	struct nf_conn_labels *cl;
-	u32 *dst;
-	int i;
+	struct nf_conn_labels *cl, *master_cl;
+	bool have_mask = labels_nonzero(mask);
+
+	/* Inherit master's labels to the related connection? */
+	master_cl = (ct->master) ? nf_ct_labels_find(ct->master) : NULL;
+
+	if (!master_cl && !have_mask)
+		return 0;   /* Nothing to do. */
 
 	cl = ovs_ct_get_conn_labels(ct);
 	if (!cl)
 		return -ENOSPC;
 
-	dst = (u32 *)cl->bits;
-	for (i = 0; i < OVS_CT_LABELS_LEN_32; i++)
-		dst[i] = (dst[i] & ~mask->ct_labels_32[i]) |
-			(labels->ct_labels_32[i] & mask->ct_labels_32[i]);
+	/* Inherit the master's labels, if any. */
+	if (master_cl)
+		*cl = *master_cl;
+
+	if (have_mask) {
+		u32 *dst = (u32 *)cl->bits;
+		int i;
+
+		for (i = 0; i < OVS_CT_LABELS_LEN_32; i++)
+			dst[i] = (dst[i] & ~mask->ct_labels_32[i]) |
+				(labels->ct_labels_32[i]
+				 & mask->ct_labels_32[i]);
+	}
 
 	memcpy(&key->ct.labels, cl->bits, OVS_CT_LABELS_LEN);
 
@@ -911,13 +927,14 @@ static int ovs_ct_commit(struct net *net, struct sw_flow_key *key,
 		if (err)
 			return err;
 	}
-	if (labels_nonzero(&info->labels.mask)) {
-		if (!nf_ct_is_confirmed(ct))
-			err = ovs_ct_init_labels(ct, key, &info->labels.value,
-						 &info->labels.mask);
-		else
-			err = ovs_ct_set_labels(ct, key, &info->labels.value,
-						&info->labels.mask);
+	if (!nf_ct_is_confirmed(ct)) {
+		err = ovs_ct_init_labels(ct, key, &info->labels.value,
+					 &info->labels.mask);
+		if (err)
+			return err;
+	} else if (labels_nonzero(&info->labels.mask)) {
+		err = ovs_ct_set_labels(ct, key, &info->labels.value,
+					&info->labels.mask);
 		if (err)
 			return err;
 	}
-- 
2.1.4

[PATCH v3 net-next 08/10] openvswitch: Add original direction conntrack tuple to sw_flow_key.

[Jarno Rajahalme]

Add the fields of the conntrack original direction 5-tuple to struct
sw_flow_key.  The new fields are initially marked as non-existent, and
are populated whenever a conntrack action is executed and either finds
or generates a conntrack entry.  This means that these fields exist
for all packets that were not rejected by conntrack as untrackable.

The original tuple fields in the sw_flow_key are filled from the
original direction tuple of the conntrack entry relating to the
current packet, or from the original direction tuple of the master
conntrack entry, if the current conntrack entry has a master.
Generally, expected connections of connections having an assigned
helper (e.g., FTP), have a master conntrack entry.

The main purpose of the new conntrack original tuple fields is to
allow matching on them for policy decision purposes, with the premise
that the admissibility of tracked connections reply packets (as well
as original direction packets), and both direction packets of any
related connections may be based on ACL rules applying to the master
connection's original direction 5-tuple.  This also makes it easier to
make policy decisions when the actual packet headers might have been
transformed by NAT, as the original direction 5-tuple represents the
packet headers before any such transformation.

When using the original direction 5-tuple the admissibility of return
and/or related packets need not be based on the mere existence of a
conntrack entry, allowing separation of admission policy from the
established conntrack state.  While existence of a conntrack entry is
required for admission of the return or related packets, policy
changes can render connections that were initially admitted to be
rejected or dropped afterwards.  If the admission of the return and
related packets was based on mere conntrack state (e.g., connection
being in an established state), a policy change that would make the
connection rejected or dropped would need to find and delete all
conntrack entries affected by such a change.  When using the original
direction 5-tuple matching the affected conntrack entries can be
allowed to time out instead, as the established state of the
connection would not need to be the basis for packet admission any
more.

It should be noted that the directionality of related connections may
be the same or different than that of the master connection, and
neither the original direction 5-tuple nor the conntrack state bits
carry this information.  If needed, the directionality of the master
connection can be stored in master's conntrack mark or labels, which
are automatically inherited by the expected related connections.

The fact that neither ARP not ND packets are trackable by conntrack
allows mutual exclusion between ARP/ND and the new conntrack original
tuple fields.  Hence, the IP addresses are overlaid in union with ARP
and ND fields.  This allows the sw_flow_key to not grow much due to
this patch, but it also means that we must be careful to never use the
new key fields with ARP or ND packets.  ARP is easy to distinguish and
keep mutually exclusive based on the ethernet type, but ND being an
ICMPv6 protocol requires a bit more attention.

Signed-off-by: Jarno Rajahalme <jarno@ovn.org>
---
 include/uapi/linux/openvswitch.h | 20 +++++++++-
 net/openvswitch/actions.c        |  2 +
 net/openvswitch/conntrack.c      | 86 +++++++++++++++++++++++++++++++++++++---
 net/openvswitch/conntrack.h      | 10 ++++-
 net/openvswitch/flow.c           | 34 +++++++++++++---
 net/openvswitch/flow.h           | 49 ++++++++++++++++++-----
 net/openvswitch/flow_netlink.c   | 85 +++++++++++++++++++++++++++++----------
 net/openvswitch/flow_netlink.h   |  7 +++-
 8 files changed, 246 insertions(+), 47 deletions(-)

diff --git a/include/uapi/linux/openvswitch.h b/include/uapi/linux/openvswitch.h
index 96aee34..90af8b8 100644
--- a/include/uapi/linux/openvswitch.h
+++ b/include/uapi/linux/openvswitch.h
@@ -1,6 +1,6 @@
 
 /*
- * Copyright (c) 2007-2013 Nicira, Inc.
+ * Copyright (c) 2007-2017 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
@@ -331,6 +331,8 @@ enum ovs_key_attr {
 	OVS_KEY_ATTR_CT_ZONE,	/* u16 connection tracking zone. */
 	OVS_KEY_ATTR_CT_MARK,	/* u32 connection tracking mark */
 	OVS_KEY_ATTR_CT_LABELS,	/* 16-octet connection tracking label */
+	OVS_KEY_ATTR_CT_ORIG_TUPLE_IPV4,   /* struct ovs_key_ct_tuple_ipv4 */
+	OVS_KEY_ATTR_CT_ORIG_TUPLE_IPV6,   /* struct ovs_key_ct_tuple_ipv6 */
 
 #ifdef __KERNEL__
 	OVS_KEY_ATTR_TUNNEL_INFO,  /* struct ip_tunnel_info */
@@ -472,6 +474,22 @@ struct ovs_key_ct_labels {
 
 #define OVS_CS_F_NAT_MASK (OVS_CS_F_SRC_NAT | OVS_CS_F_DST_NAT)
 
+struct ovs_key_ct_tuple_ipv4 {
+	__be32 ipv4_src;
+	__be32 ipv4_dst;
+	__be16 src_port;
+	__be16 dst_port;
+	__u8   ipv4_proto;
+};
+
+struct ovs_key_ct_tuple_ipv6 {
+	__be32 ipv6_src[4];
+	__be32 ipv6_dst[4];
+	__be16 src_port;
+	__be16 dst_port;
+	__u8   ipv6_proto;
+};
+
 /**
  * enum ovs_flow_attr - attributes for %OVS_FLOW_* commands.
  * @OVS_FLOW_ATTR_KEY: Nested %OVS_KEY_ATTR_* attributes specifying the flow
diff --git a/net/openvswitch/actions.c b/net/openvswitch/actions.c
index efa9a88..b1beb2b 100644
--- a/net/openvswitch/actions.c
+++ b/net/openvswitch/actions.c
@@ -1074,6 +1074,8 @@ static int execute_masked_set_action(struct sk_buff *skb,
 	case OVS_KEY_ATTR_CT_ZONE:
 	case OVS_KEY_ATTR_CT_MARK:
 	case OVS_KEY_ATTR_CT_LABELS:
+	case OVS_KEY_ATTR_CT_ORIG_TUPLE_IPV4:
+	case OVS_KEY_ATTR_CT_ORIG_TUPLE_IPV6:
 		err = -EINVAL;
 		break;
 	}
diff --git a/net/openvswitch/conntrack.c b/net/openvswitch/conntrack.c
index 5fbadcd..8685bcd 100644
--- a/net/openvswitch/conntrack.c
+++ b/net/openvswitch/conntrack.c
@@ -149,6 +149,20 @@ static void ovs_ct_get_labels(const struct nf_conn *ct,
 		memset(labels, 0, OVS_CT_LABELS_LEN);
 }
 
+static void __ovs_ct_update_key_orig_tp(struct sw_flow_key *key,
+					const struct nf_conntrack_tuple *orig,
+					u8 icmp_proto)
+{
+	key->ct.orig_proto = orig->dst.protonum;
+	if (orig->dst.protonum == icmp_proto) {
+		key->ct.orig_tp.src = htons(orig->dst.u.icmp.type);
+		key->ct.orig_tp.dst = htons(orig->dst.u.icmp.code);
+	} else {
+		key->ct.orig_tp.src = orig->src.u.all;
+		key->ct.orig_tp.dst = orig->dst.u.all;
+	}
+}
+
 static void __ovs_ct_update_key(struct sw_flow_key *key, u8 state,
 				const struct nf_conntrack_zone *zone,
 				const struct nf_conn *ct)
@@ -157,6 +171,35 @@ static void __ovs_ct_update_key(struct sw_flow_key *key, u8 state,
 	key->ct.zone = zone->id;
 	key->ct.mark = ovs_ct_get_mark(ct);
 	ovs_ct_get_labels(ct, &key->ct.labels);
+
+	if (ct) {
+		const struct nf_conntrack_tuple *orig;
+
+		/* Use the master if we have one. */
+		if (ct->master)
+			ct = ct->master;
+		orig = &ct->tuplehash[IP_CT_DIR_ORIGINAL].tuple;
+
+		/* IP version must match with the master connection. */
+		if (key->eth.type == htons(ETH_P_IP) &&
+		    nf_ct_l3num(ct) == NFPROTO_IPV4) {
+			key->ipv4.ct_orig.src = orig->src.u3.ip;
+			key->ipv4.ct_orig.dst = orig->dst.u3.ip;
+			__ovs_ct_update_key_orig_tp(key, orig, IPPROTO_ICMP);
+			return;
+		} else if (key->eth.type == htons(ETH_P_IPV6) &&
+			   !sw_flow_key_is_nd(key) &&
+			   nf_ct_l3num(ct) == NFPROTO_IPV6) {
+			key->ipv6.ct_orig.src = orig->src.u3.in6;
+			key->ipv6.ct_orig.dst = orig->dst.u3.in6;
+			__ovs_ct_update_key_orig_tp(key, orig, NEXTHDR_ICMP);
+			return;
+		}
+	}
+	/* Clear 'ct.orig_proto' to mark the non-existence of conntrack
+	 * original direction key fields.
+	 */
+	key->ct.orig_proto = 0;
 }
 
 /* Update 'key' based on skb->_nfct.  If 'post_ct' is true, then OVS has
@@ -210,24 +253,55 @@ void ovs_ct_fill_key(const struct sk_buff *skb, struct sw_flow_key *key)
 	ovs_ct_update_key(skb, NULL, key, false, false);
 }
 
-int ovs_ct_put_key(const struct sw_flow_key *key, struct sk_buff *skb)
+#define IN6_ADDR_INITIALIZER(ADDR) \
+	{ (ADDR).s6_addr32[0], (ADDR).s6_addr32[1], \
+	  (ADDR).s6_addr32[2], (ADDR).s6_addr32[3] }
+
+int ovs_ct_put_key(const struct sw_flow_key *swkey,
+		   const struct sw_flow_key *output, struct sk_buff *skb)
 {
-	if (nla_put_u32(skb, OVS_KEY_ATTR_CT_STATE, key->ct.state))
+	if (nla_put_u32(skb, OVS_KEY_ATTR_CT_STATE, output->ct.state))
 		return -EMSGSIZE;
 
 	if (IS_ENABLED(CONFIG_NF_CONNTRACK_ZONES) &&
-	    nla_put_u16(skb, OVS_KEY_ATTR_CT_ZONE, key->ct.zone))
+	    nla_put_u16(skb, OVS_KEY_ATTR_CT_ZONE, output->ct.zone))
 		return -EMSGSIZE;
 
 	if (IS_ENABLED(CONFIG_NF_CONNTRACK_MARK) &&
-	    nla_put_u32(skb, OVS_KEY_ATTR_CT_MARK, key->ct.mark))
+	    nla_put_u32(skb, OVS_KEY_ATTR_CT_MARK, output->ct.mark))
 		return -EMSGSIZE;
 
 	if (IS_ENABLED(CONFIG_NF_CONNTRACK_LABELS) &&
-	    nla_put(skb, OVS_KEY_ATTR_CT_LABELS, sizeof(key->ct.labels),
-		    &key->ct.labels))
+	    nla_put(skb, OVS_KEY_ATTR_CT_LABELS, sizeof(output->ct.labels),
+		    &output->ct.labels))
 		return -EMSGSIZE;
 
+	if (swkey->ct.orig_proto) {
+		if (swkey->eth.type == htons(ETH_P_IP)) {
+			struct ovs_key_ct_tuple_ipv4 orig = {
+				output->ipv4.ct_orig.src,
+				output->ipv4.ct_orig.dst,
+				output->ct.orig_tp.src,
+				output->ct.orig_tp.dst,
+				output->ct.orig_proto,
+			};
+			if (nla_put(skb, OVS_KEY_ATTR_CT_ORIG_TUPLE_IPV4,
+				    sizeof(orig), &orig))
+				return -EMSGSIZE;
+		} else if (swkey->eth.type == htons(ETH_P_IPV6)) {
+			struct ovs_key_ct_tuple_ipv6 orig = {
+				IN6_ADDR_INITIALIZER(output->ipv6.ct_orig.src),
+				IN6_ADDR_INITIALIZER(output->ipv6.ct_orig.dst),
+				output->ct.orig_tp.src,
+				output->ct.orig_tp.dst,
+				output->ct.orig_proto,
+			};
+			if (nla_put(skb, OVS_KEY_ATTR_CT_ORIG_TUPLE_IPV6,
+				    sizeof(orig), &orig))
+				return -EMSGSIZE;
+		}
+	}
+
 	return 0;
 }
 
diff --git a/net/openvswitch/conntrack.h b/net/openvswitch/conntrack.h
index 8f6230b..9e92445 100644
--- a/net/openvswitch/conntrack.h
+++ b/net/openvswitch/conntrack.h
@@ -32,7 +32,8 @@ int ovs_ct_execute(struct net *, struct sk_buff *, struct sw_flow_key *,
 		   const struct ovs_conntrack_info *);
 
 void ovs_ct_fill_key(const struct sk_buff *skb, struct sw_flow_key *key);
-int ovs_ct_put_key(const struct sw_flow_key *key, struct sk_buff *skb);
+int ovs_ct_put_key(const struct sw_flow_key *swkey,
+		   const struct sw_flow_key *output, struct sk_buff *skb);
 void ovs_ct_free_action(const struct nlattr *a);
 
 #define CT_SUPPORTED_MASK (OVS_CS_F_NEW | OVS_CS_F_ESTABLISHED | \
@@ -79,9 +80,14 @@ static inline void ovs_ct_fill_key(const struct sk_buff *skb,
 	key->ct.zone = 0;
 	key->ct.mark = 0;
 	memset(&key->ct.labels, 0, sizeof(key->ct.labels));
+	/* Clear 'ct.orig_proto' to mark the non-existence of original
+	 * direction key fields.
+	 */
+	key->ct.orig_proto = 0;
 }
 
-static inline int ovs_ct_put_key(const struct sw_flow_key *key,
+static inline int ovs_ct_put_key(const struct sw_flow_key *swkey,
+				 const struct sw_flow_key *output,
 				 struct sk_buff *skb)
 {
 	return 0;
diff --git a/net/openvswitch/flow.c b/net/openvswitch/flow.c
index 2c0a00f..9d4bb8e 100644
--- a/net/openvswitch/flow.c
+++ b/net/openvswitch/flow.c
@@ -765,7 +765,7 @@ static int key_extract_mac_proto(struct sk_buff *skb)
 int ovs_flow_key_extract(const struct ip_tunnel_info *tun_info,
 			 struct sk_buff *skb, struct sw_flow_key *key)
 {
-	int res;
+	int res, err;
 
 	/* Extract metadata from packet. */
 	if (tun_info) {
@@ -792,7 +792,6 @@ int ovs_flow_key_extract(const struct ip_tunnel_info *tun_info,
 	key->phy.priority = skb->priority;
 	key->phy.in_port = OVS_CB(skb)->input_vport->port_no;
 	key->phy.skb_mark = skb->mark;
-	ovs_ct_fill_key(skb, key);
 	key->ovs_flow_hash = 0;
 	res = key_extract_mac_proto(skb);
 	if (res < 0)
@@ -800,17 +799,26 @@ int ovs_flow_key_extract(const struct ip_tunnel_info *tun_info,
 	key->mac_proto = res;
 	key->recirc_id = 0;
 
-	return key_extract(skb, key);
+	err = key_extract(skb, key);
+	if (!err)
+		ovs_ct_fill_key(skb, key);   /* Must be after key_extract(). */
+	return err;
 }
 
 int ovs_flow_key_extract_userspace(struct net *net, const struct nlattr *attr,
 				   struct sk_buff *skb,
 				   struct sw_flow_key *key, bool log)
 {
+	const struct nlattr *a[OVS_KEY_ATTR_MAX + 1];
+	u64 attrs = 0;
 	int err;
 
+	err = parse_flow_nlattrs(attr, a, &attrs, log);
+	if (err)
+		return -EINVAL;
+
 	/* Extract metadata from netlink attributes. */
-	err = ovs_nla_get_flow_metadata(net, attr, key, log);
+	err = ovs_nla_get_flow_metadata(net, a, attrs, key, log);
 	if (err)
 		return err;
 
@@ -824,5 +832,21 @@ int ovs_flow_key_extract_userspace(struct net *net, const struct nlattr *attr,
 	 */
 
 	skb->protocol = key->eth.type;
-	return key_extract(skb, key);
+	err = key_extract(skb, key);
+	if (err)
+		return err;
+
+	/* Check that we have conntrack original direction tuple metadata only
+	 * for packets for which it makes sense.  Otherwise the key may be
+	 * corrupted due to overlapping key fields.
+	 */
+	if (attrs & (1 << OVS_KEY_ATTR_CT_ORIG_TUPLE_IPV4) &&
+	    key->eth.type != htons(ETH_P_IP))
+		return -EINVAL;
+	if (attrs & (1 << OVS_KEY_ATTR_CT_ORIG_TUPLE_IPV6) &&
+	    (key->eth.type != htons(ETH_P_IPV6) ||
+	     sw_flow_key_is_nd(key)))
+		return -EINVAL;
+
+	return 0;
 }
diff --git a/net/openvswitch/flow.h b/net/openvswitch/flow.h
index f61cae7..76e05b2 100644
--- a/net/openvswitch/flow.h
+++ b/net/openvswitch/flow.h
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2007-2014 Nicira, Inc.
+ * Copyright (c) 2007-2017 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
@@ -107,10 +107,16 @@ struct sw_flow_key {
 				__be32 src;	/* IP source address. */
 				__be32 dst;	/* IP destination address. */
 			} addr;
-			struct {
-				u8 sha[ETH_ALEN];	/* ARP source hardware address. */
-				u8 tha[ETH_ALEN];	/* ARP target hardware address. */
-			} arp;
+			union {
+				struct {
+					__be32 src;
+					__be32 dst;
+				} ct_orig;	/* Conntrack original direction fields. */
+				struct {
+					u8 sha[ETH_ALEN];	/* ARP source hardware address. */
+					u8 tha[ETH_ALEN];	/* ARP target hardware address. */
+				} arp;
+			};
 		} ipv4;
 		struct {
 			struct {
@@ -118,23 +124,44 @@ struct sw_flow_key {
 				struct in6_addr dst;	/* IPv6 destination address. */
 			} addr;
 			__be32 label;			/* IPv6 flow label. */
-			struct {
-				struct in6_addr target;	/* ND target address. */
-				u8 sll[ETH_ALEN];	/* ND source link layer address. */
-				u8 tll[ETH_ALEN];	/* ND target link layer address. */
-			} nd;
+			union {
+				struct {
+					struct in6_addr src;
+					struct in6_addr dst;
+				} ct_orig;	/* Conntrack original direction fields. */
+				struct {
+					struct in6_addr target;	/* ND target address. */
+					u8 sll[ETH_ALEN];	/* ND source link layer address. */
+					u8 tll[ETH_ALEN];	/* ND target link layer address. */
+				} nd;
+			};
 		} ipv6;
 	};
 	struct {
 		/* Connection tracking fields. */
+		u8 state;
+		u8 orig_proto;		/* CT orig tuple IP protocol. */
 		u16 zone;
 		u32 mark;
-		u8 state;
+		struct {
+			__be16 src;	/* CT orig tuple tp src port. */
+			__be16 dst;	/* CT orig tuple tp dst port. */
+		} orig_tp;
+
 		struct ovs_key_ct_labels labels;
 	} ct;
 
 } __aligned(BITS_PER_LONG/8); /* Ensure that we can do comparisons as longs. */
 
+static inline bool sw_flow_key_is_nd(const struct sw_flow_key *key)
+{
+	return key->eth.type == htons(ETH_P_IPV6) &&
+		key->ip.proto == NEXTHDR_ICMP &&
+		key->tp.dst == 0 &&
+		(key->tp.src == htons(NDISC_NEIGHBOUR_SOLICITATION) ||
+		 key->tp.src == htons(NDISC_NEIGHBOUR_ADVERTISEMENT));
+}
+
 struct sw_flow_key_range {
 	unsigned short int start;
 	unsigned short int end;
diff --git a/net/openvswitch/flow_netlink.c b/net/openvswitch/flow_netlink.c
index c87d359..989f38f 100644
--- a/net/openvswitch/flow_netlink.c
+++ b/net/openvswitch/flow_netlink.c
@@ -129,7 +129,9 @@ static bool match_validate(const struct sw_flow_match *match,
 	/* The following mask attributes allowed only if they
 	 * pass the validation tests. */
 	mask_allowed &= ~((1 << OVS_KEY_ATTR_IPV4)
+			| (1 << OVS_KEY_ATTR_CT_ORIG_TUPLE_IPV4)
 			| (1 << OVS_KEY_ATTR_IPV6)
+			| (1 << OVS_KEY_ATTR_CT_ORIG_TUPLE_IPV6)
 			| (1 << OVS_KEY_ATTR_TCP)
 			| (1 << OVS_KEY_ATTR_TCP_FLAGS)
 			| (1 << OVS_KEY_ATTR_UDP)
@@ -161,8 +163,10 @@ static bool match_validate(const struct sw_flow_match *match,
 
 	if (match->key->eth.type == htons(ETH_P_IP)) {
 		key_expected |= 1 << OVS_KEY_ATTR_IPV4;
-		if (match->mask && (match->mask->key.eth.type == htons(0xffff)))
+		if (match->mask && match->mask->key.eth.type == htons(0xffff)) {
 			mask_allowed |= 1 << OVS_KEY_ATTR_IPV4;
+			mask_allowed |= 1 << OVS_KEY_ATTR_CT_ORIG_TUPLE_IPV4;
+		}
 
 		if (match->key->ip.frag != OVS_FRAG_TYPE_LATER) {
 			if (match->key->ip.proto == IPPROTO_UDP) {
@@ -196,8 +200,10 @@ static bool match_validate(const struct sw_flow_match *match,
 
 	if (match->key->eth.type == htons(ETH_P_IPV6)) {
 		key_expected |= 1 << OVS_KEY_ATTR_IPV6;
-		if (match->mask && (match->mask->key.eth.type == htons(0xffff)))
+		if (match->mask && match->mask->key.eth.type == htons(0xffff)) {
 			mask_allowed |= 1 << OVS_KEY_ATTR_IPV6;
+			mask_allowed |= 1 << OVS_KEY_ATTR_CT_ORIG_TUPLE_IPV6;
+		}
 
 		if (match->key->ip.frag != OVS_FRAG_TYPE_LATER) {
 			if (match->key->ip.proto == IPPROTO_UDP) {
@@ -230,6 +236,12 @@ static bool match_validate(const struct sw_flow_match *match,
 						htons(NDISC_NEIGHBOUR_SOLICITATION) ||
 				    match->key->tp.src == htons(NDISC_NEIGHBOUR_ADVERTISEMENT)) {
 					key_expected |= 1 << OVS_KEY_ATTR_ND;
+					/* Original direction conntrack tuple
+					 * uses the same space as the ND fields
+					 * in the key, so both are not allowed
+					 * at the same time.
+					 */
+					mask_allowed &= ~(1ULL << OVS_KEY_ATTR_CT_ORIG_TUPLE_IPV6);
 					if (match->mask && (match->mask->key.tp.src == htons(0xff)))
 						mask_allowed |= 1 << OVS_KEY_ATTR_ND;
 				}
@@ -282,7 +294,7 @@ size_t ovs_key_attr_size(void)
 	/* Whenever adding new OVS_KEY_ FIELDS, we should consider
 	 * updating this function.
 	 */
-	BUILD_BUG_ON(OVS_KEY_ATTR_TUNNEL_INFO != 26);
+	BUILD_BUG_ON(OVS_KEY_ATTR_TUNNEL_INFO != 28);
 
 	return    nla_total_size(4)   /* OVS_KEY_ATTR_PRIORITY */
 		+ nla_total_size(0)   /* OVS_KEY_ATTR_TUNNEL */
@@ -295,6 +307,7 @@ size_t ovs_key_attr_size(void)
 		+ nla_total_size(2)   /* OVS_KEY_ATTR_CT_ZONE */
 		+ nla_total_size(4)   /* OVS_KEY_ATTR_CT_MARK */
 		+ nla_total_size(16)  /* OVS_KEY_ATTR_CT_LABELS */
+		+ nla_total_size(40)  /* OVS_KEY_ATTR_CT_ORIG_TUPLE_IPV6 */
 		+ nla_total_size(12)  /* OVS_KEY_ATTR_ETHERNET */
 		+ nla_total_size(2)   /* OVS_KEY_ATTR_ETHERTYPE */
 		+ nla_total_size(4)   /* OVS_KEY_ATTR_VLAN */
@@ -355,6 +368,10 @@ static const struct ovs_len_tbl ovs_key_lens[OVS_KEY_ATTR_MAX + 1] = {
 	[OVS_KEY_ATTR_CT_ZONE]	 = { .len = sizeof(u16) },
 	[OVS_KEY_ATTR_CT_MARK]	 = { .len = sizeof(u32) },
 	[OVS_KEY_ATTR_CT_LABELS] = { .len = sizeof(struct ovs_key_ct_labels) },
+	[OVS_KEY_ATTR_CT_ORIG_TUPLE_IPV4] = {
+		.len = sizeof(struct ovs_key_ct_tuple_ipv4) },
+	[OVS_KEY_ATTR_CT_ORIG_TUPLE_IPV6] = {
+		.len = sizeof(struct ovs_key_ct_tuple_ipv6) },
 };
 
 static bool check_attr_len(unsigned int attr_len, unsigned int expected_len)
@@ -430,9 +447,8 @@ static int parse_flow_mask_nlattrs(const struct nlattr *attr,
 	return __parse_flow_nlattrs(attr, a, attrsp, log, true);
 }
 
-static int parse_flow_nlattrs(const struct nlattr *attr,
-			      const struct nlattr *a[], u64 *attrsp,
-			      bool log)
+int parse_flow_nlattrs(const struct nlattr *attr, const struct nlattr *a[],
+		       u64 *attrsp, bool log)
 {
 	return __parse_flow_nlattrs(attr, a, attrsp, log, false);
 }
@@ -1082,6 +1098,34 @@ static int metadata_from_nlattrs(struct net *net, struct sw_flow_match *match,
 				   sizeof(*cl), is_mask);
 		*attrs &= ~(1ULL << OVS_KEY_ATTR_CT_LABELS);
 	}
+	if (*attrs & (1ULL << OVS_KEY_ATTR_CT_ORIG_TUPLE_IPV4)) {
+		const struct ovs_key_ct_tuple_ipv4 *ct;
+
+		ct = nla_data(a[OVS_KEY_ATTR_CT_ORIG_TUPLE_IPV4]);
+
+		SW_FLOW_KEY_PUT(match, ipv4.ct_orig.src, ct->ipv4_src, is_mask);
+		SW_FLOW_KEY_PUT(match, ipv4.ct_orig.dst, ct->ipv4_dst, is_mask);
+		SW_FLOW_KEY_PUT(match, ct.orig_tp.src, ct->src_port, is_mask);
+		SW_FLOW_KEY_PUT(match, ct.orig_tp.dst, ct->dst_port, is_mask);
+		SW_FLOW_KEY_PUT(match, ct.orig_proto, ct->ipv4_proto, is_mask);
+		*attrs &= ~(1ULL << OVS_KEY_ATTR_CT_ORIG_TUPLE_IPV4);
+	}
+	if (*attrs & (1ULL << OVS_KEY_ATTR_CT_ORIG_TUPLE_IPV6)) {
+		const struct ovs_key_ct_tuple_ipv6 *ct;
+
+		ct = nla_data(a[OVS_KEY_ATTR_CT_ORIG_TUPLE_IPV6]);
+
+		SW_FLOW_KEY_MEMCPY(match, ipv6.ct_orig.src, &ct->ipv6_src,
+				   sizeof(match->key->ipv6.ct_orig.src),
+				   is_mask);
+		SW_FLOW_KEY_MEMCPY(match, ipv6.ct_orig.dst, &ct->ipv6_dst,
+				   sizeof(match->key->ipv6.ct_orig.dst),
+				   is_mask);
+		SW_FLOW_KEY_PUT(match, ct.orig_tp.src, ct->src_port, is_mask);
+		SW_FLOW_KEY_PUT(match, ct.orig_tp.dst, ct->dst_port, is_mask);
+		SW_FLOW_KEY_PUT(match, ct.orig_proto, ct->ipv6_proto, is_mask);
+		*attrs &= ~(1ULL << OVS_KEY_ATTR_CT_ORIG_TUPLE_IPV6);
+	}
 
 	/* For layer 3 packets the Ethernet type is provided
 	 * and treated as metadata but no MAC addresses are provided.
@@ -1493,9 +1537,12 @@ u32 ovs_nla_get_ufid_flags(const struct nlattr *attr)
 
 /**
  * ovs_nla_get_flow_metadata - parses Netlink attributes into a flow key.
- * @key: Receives extracted in_port, priority, tun_key and skb_mark.
- * @attr: Netlink attribute holding nested %OVS_KEY_ATTR_* Netlink attribute
- * sequence.
+ * @net: Network namespace.
+ * @key: Receives extracted in_port, priority, tun_key, skb_mark and conntrack
+ * metadata.
+ * @a: Array of netlink attributes holding parsed %OVS_KEY_ATTR_* Netlink
+ * attributes.
+ * @attrs: Bit mask for the netlink attributes included in @a.
  * @log: Boolean to allow kernel error logging.  Normally true, but when
  * probing for feature compatibility this should be passed in as false to
  * suppress unnecessary error logging.
@@ -1504,25 +1551,23 @@ u32 ovs_nla_get_ufid_flags(const struct nlattr *attr)
  * take the same form accepted by flow_from_nlattrs(), but only enough of it to
  * get the metadata, that is, the parts of the flow key that cannot be
  * extracted from the packet itself.
+ *
+ * This must be called before the packet key fields are filled in 'key'.
  */
 
-int ovs_nla_get_flow_metadata(struct net *net, const struct nlattr *attr,
-			      struct sw_flow_key *key,
-			      bool log)
+int ovs_nla_get_flow_metadata(struct net *net,
+			      const struct nlattr *a[OVS_KEY_ATTR_MAX + 1],
+			      u64 attrs, struct sw_flow_key *key, bool log)
 {
-	const struct nlattr *a[OVS_KEY_ATTR_MAX + 1];
 	struct sw_flow_match match;
-	u64 attrs = 0;
-	int err;
-
-	err = parse_flow_nlattrs(attr, a, &attrs, log);
-	if (err)
-		return -EINVAL;
 
 	memset(&match, 0, sizeof(match));
 	match.key = key;
 
 	memset(&key->ct, 0, sizeof(key->ct));
+	memset(&key->ipv4.ct_orig, 0, sizeof(key->ipv4.ct_orig));
+	memset(&key->ipv6.ct_orig, 0, sizeof(key->ipv6.ct_orig));
+
 	key->phy.in_port = DP_MAX_PORTS;
 
 	return metadata_from_nlattrs(net, &match, &attrs, a, false, log);
@@ -1584,7 +1629,7 @@ static int __ovs_nla_put_key(const struct sw_flow_key *swkey,
 	if (nla_put_u32(skb, OVS_KEY_ATTR_SKB_MARK, output->phy.skb_mark))
 		goto nla_put_failure;
 
-	if (ovs_ct_put_key(output, skb))
+	if (ovs_ct_put_key(swkey, output, skb))
 		goto nla_put_failure;
 
 	if (ovs_key_mac_proto(swkey) == MAC_PROTO_ETHERNET) {
diff --git a/net/openvswitch/flow_netlink.h b/net/openvswitch/flow_netlink.h
index 45f9769..929c665 100644
--- a/net/openvswitch/flow_netlink.h
+++ b/net/openvswitch/flow_netlink.h
@@ -46,8 +46,11 @@ void ovs_match_init(struct sw_flow_match *match,
 
 int ovs_nla_put_key(const struct sw_flow_key *, const struct sw_flow_key *,
 		    int attr, bool is_mask, struct sk_buff *);
-int ovs_nla_get_flow_metadata(struct net *, const struct nlattr *,
-			      struct sw_flow_key *, bool log);
+int parse_flow_nlattrs(const struct nlattr *attr, const struct nlattr *a[],
+		       u64 *attrsp, bool log);
+int ovs_nla_get_flow_metadata(struct net *net,
+			      const struct nlattr *a[OVS_KEY_ATTR_MAX + 1],
+			      u64 attrs, struct sw_flow_key *key, bool log);
 
 int ovs_nla_put_identifier(const struct sw_flow *flow, struct sk_buff *skb);
 int ovs_nla_put_masked_key(const struct sw_flow *flow, struct sk_buff *skb);
-- 
2.1.4

[PATCH v3 net-next 09/10] openvswitch: Add force commit.

[Jarno Rajahalme]

Stateful network admission policy may allow connections to one
direction and reject connections initiated in the other direction.
After policy change it is possible that for a new connection an
overlapping conntrack entry already exists, where the original
direction of the existing connection is opposed to the new
connection's initial packet.

Most importantly, conntrack state relating to the current packet gets
the "reply" designation based on whether the original direction tuple
or the reply direction tuple matched.  If this "directionality" is
wrong w.r.t. to the stateful network admission policy it may happen
that packets in neither direction are correctly admitted.

This patch adds a new "force commit" option to the OVS conntrack
action that checks the original direction of an existing conntrack
entry.  If that direction is opposed to the current packet, the
existing conntrack entry is deleted and a new one is subsequently
created in the correct direction.

Signed-off-by: Jarno Rajahalme <jarno@ovn.org>
---
 include/uapi/linux/openvswitch.h |  5 +++++
 net/openvswitch/conntrack.c      | 26 ++++++++++++++++++++++++--
 2 files changed, 29 insertions(+), 2 deletions(-)

diff --git a/include/uapi/linux/openvswitch.h b/include/uapi/linux/openvswitch.h
index 90af8b8..7f41f7d 100644
--- a/include/uapi/linux/openvswitch.h
+++ b/include/uapi/linux/openvswitch.h
@@ -674,6 +674,10 @@ struct ovs_action_hash {
  * @OVS_CT_ATTR_HELPER: variable length string defining conntrack ALG.
  * @OVS_CT_ATTR_NAT: Nested OVS_NAT_ATTR_* for performing L3 network address
  * translation (NAT) on the packet.
+ * @OVS_CT_ATTR_FORCE_COMMIT: Like %OVS_CT_ATTR_COMMIT, but instead of doing
+ * nothing if the connection is already committed will check that the current
+ * packet is in conntrack entry's original direction.  If directionality does
+ * not match, will delete the existing conntrack entry and commit a new one.
  */
 enum ovs_ct_attr {
 	OVS_CT_ATTR_UNSPEC,
@@ -684,6 +688,7 @@ enum ovs_ct_attr {
 	OVS_CT_ATTR_HELPER,     /* netlink helper to assist detection of
 				   related connections. */
 	OVS_CT_ATTR_NAT,        /* Nested OVS_NAT_ATTR_* */
+	OVS_CT_ATTR_FORCE_COMMIT,  /* No argument */
 	__OVS_CT_ATTR_MAX
 };
 
diff --git a/net/openvswitch/conntrack.c b/net/openvswitch/conntrack.c
index 8685bcd..de47782 100644
--- a/net/openvswitch/conntrack.c
+++ b/net/openvswitch/conntrack.c
@@ -65,6 +65,7 @@ struct ovs_conntrack_info {
 	struct nf_conn *ct;
 	u8 commit : 1;
 	u8 nat : 3;                 /* enum ovs_ct_nat */
+	u8 force : 1;
 	u16 family;
 	struct md_mark mark;
 	struct md_labels labels;
@@ -615,10 +616,13 @@ static bool skb_nfct_cached(struct net *net,
 	 */
 	if (!ct && key->ct.state & OVS_CS_F_TRACKED &&
 	    !(key->ct.state & OVS_CS_F_INVALID) &&
-	    key->ct.zone == info->zone.id)
+	    key->ct.zone == info->zone.id) {
 		ct = ovs_ct_find_existing(net, &info->zone, info->family, skb,
 					  !!(key->ct.state
 					     & OVS_CS_F_NAT_MASK));
+		if (ct)
+			nf_ct_get(skb, &ctinfo);
+	}
 	if (!ct)
 		return false;
 	if (!net_eq(net, read_pnet(&ct->ct_net)))
@@ -632,6 +636,18 @@ static bool skb_nfct_cached(struct net *net,
 		if (help && rcu_access_pointer(help->helper) != info->helper)
 			return false;
 	}
+	/* Force conntrack entry direction to the current packet? */
+	if (info->force && CTINFO2DIR(ctinfo) != IP_CT_DIR_ORIGINAL) {
+		/* Delete the conntrack entry if confirmed, else just release
+		 * the reference.
+		 */
+		if (nf_ct_is_confirmed(ct))
+			nf_ct_delete(ct, 0, 0);
+		else
+			nf_conntrack_put(&ct->ct_general);
+		nf_ct_set(skb, NULL, 0);
+		return false;
+	}
 
 	return true;
 }
@@ -1209,6 +1225,7 @@ static int parse_nat(const struct nlattr *attr,
 
 static const struct ovs_ct_len_tbl ovs_ct_attr_lens[OVS_CT_ATTR_MAX + 1] = {
 	[OVS_CT_ATTR_COMMIT]	= { .minlen = 0, .maxlen = 0 },
+	[OVS_CT_ATTR_FORCE_COMMIT]	= { .minlen = 0, .maxlen = 0 },
 	[OVS_CT_ATTR_ZONE]	= { .minlen = sizeof(u16),
 				    .maxlen = sizeof(u16) },
 	[OVS_CT_ATTR_MARK]	= { .minlen = sizeof(struct md_mark),
@@ -1248,6 +1265,9 @@ static int parse_ct(const struct nlattr *attr, struct ovs_conntrack_info *info,
 		}
 
 		switch (type) {
+		case OVS_CT_ATTR_FORCE_COMMIT:
+			info->force = true;
+			/* fall through. */
 		case OVS_CT_ATTR_COMMIT:
 			info->commit = true;
 			break;
@@ -1474,7 +1494,9 @@ int ovs_ct_action_to_attr(const struct ovs_conntrack_info *ct_info,
 	if (!start)
 		return -EMSGSIZE;
 
-	if (ct_info->commit && nla_put_flag(skb, OVS_CT_ATTR_COMMIT))
+	if (ct_info->commit && nla_put_flag(skb, ct_info->force
+					    ? OVS_CT_ATTR_FORCE_COMMIT
+					    : OVS_CT_ATTR_COMMIT))
 		return -EMSGSIZE;
 	if (IS_ENABLED(CONFIG_NF_CONNTRACK_ZONES) &&
 	    nla_put_u16(skb, OVS_CT_ATTR_ZONE, ct_info->zone.id))
-- 
2.1.4

[PATCH v3 net-next 10/10] openvswitch: Pack struct sw_flow_key.

[Jarno Rajahalme]

struct sw_flow_key has two 16-bit holes. Move the most matched
conntrack match fields there.  In some typical cases this reduces the
size of the key that needs to be hashed into half and into one cache
line.

Signed-off-by: Jarno Rajahalme <jarno@ovn.org>
---
 net/openvswitch/conntrack.c    | 40 ++++++++++++++++++++--------------------
 net/openvswitch/conntrack.h    |  8 ++++----
 net/openvswitch/flow.h         | 14 ++++++++------
 net/openvswitch/flow_netlink.c | 11 +++++++----
 4 files changed, 39 insertions(+), 34 deletions(-)

diff --git a/net/openvswitch/conntrack.c b/net/openvswitch/conntrack.c
index de47782..a12825e 100644
--- a/net/openvswitch/conntrack.c
+++ b/net/openvswitch/conntrack.c
@@ -154,7 +154,7 @@ static void __ovs_ct_update_key_orig_tp(struct sw_flow_key *key,
 					const struct nf_conntrack_tuple *orig,
 					u8 icmp_proto)
 {
-	key->ct.orig_proto = orig->dst.protonum;
+	key->ct_orig_proto = orig->dst.protonum;
 	if (orig->dst.protonum == icmp_proto) {
 		key->ct.orig_tp.src = htons(orig->dst.u.icmp.type);
 		key->ct.orig_tp.dst = htons(orig->dst.u.icmp.code);
@@ -168,8 +168,8 @@ static void __ovs_ct_update_key(struct sw_flow_key *key, u8 state,
 				const struct nf_conntrack_zone *zone,
 				const struct nf_conn *ct)
 {
-	key->ct.state = state;
-	key->ct.zone = zone->id;
+	key->ct_state = state;
+	key->ct_zone = zone->id;
 	key->ct.mark = ovs_ct_get_mark(ct);
 	ovs_ct_get_labels(ct, &key->ct.labels);
 
@@ -197,10 +197,10 @@ static void __ovs_ct_update_key(struct sw_flow_key *key, u8 state,
 			return;
 		}
 	}
-	/* Clear 'ct.orig_proto' to mark the non-existence of conntrack
+	/* Clear 'ct_orig_proto' to mark the non-existence of conntrack
 	 * original direction key fields.
 	 */
-	key->ct.orig_proto = 0;
+	key->ct_orig_proto = 0;
 }
 
 /* Update 'key' based on skb->_nfct.  If 'post_ct' is true, then OVS has
@@ -230,7 +230,7 @@ static void ovs_ct_update_key(const struct sk_buff *skb,
 		if (ct->master)
 			state |= OVS_CS_F_RELATED;
 		if (keep_nat_flags) {
-			state |= key->ct.state & OVS_CS_F_NAT_MASK;
+			state |= key->ct_state & OVS_CS_F_NAT_MASK;
 		} else {
 			if (ct->status & IPS_SRC_NAT)
 				state |= OVS_CS_F_SRC_NAT;
@@ -261,11 +261,11 @@ void ovs_ct_fill_key(const struct sk_buff *skb, struct sw_flow_key *key)
 int ovs_ct_put_key(const struct sw_flow_key *swkey,
 		   const struct sw_flow_key *output, struct sk_buff *skb)
 {
-	if (nla_put_u32(skb, OVS_KEY_ATTR_CT_STATE, output->ct.state))
+	if (nla_put_u32(skb, OVS_KEY_ATTR_CT_STATE, output->ct_state))
 		return -EMSGSIZE;
 
 	if (IS_ENABLED(CONFIG_NF_CONNTRACK_ZONES) &&
-	    nla_put_u16(skb, OVS_KEY_ATTR_CT_ZONE, output->ct.zone))
+	    nla_put_u16(skb, OVS_KEY_ATTR_CT_ZONE, output->ct_zone))
 		return -EMSGSIZE;
 
 	if (IS_ENABLED(CONFIG_NF_CONNTRACK_MARK) &&
@@ -277,14 +277,14 @@ int ovs_ct_put_key(const struct sw_flow_key *swkey,
 		    &output->ct.labels))
 		return -EMSGSIZE;
 
-	if (swkey->ct.orig_proto) {
+	if (swkey->ct_orig_proto) {
 		if (swkey->eth.type == htons(ETH_P_IP)) {
 			struct ovs_key_ct_tuple_ipv4 orig = {
 				output->ipv4.ct_orig.src,
 				output->ipv4.ct_orig.dst,
 				output->ct.orig_tp.src,
 				output->ct.orig_tp.dst,
-				output->ct.orig_proto,
+				output->ct_orig_proto,
 			};
 			if (nla_put(skb, OVS_KEY_ATTR_CT_ORIG_TUPLE_IPV4,
 				    sizeof(orig), &orig))
@@ -295,7 +295,7 @@ int ovs_ct_put_key(const struct sw_flow_key *swkey,
 				IN6_ADDR_INITIALIZER(output->ipv6.ct_orig.dst),
 				output->ct.orig_tp.src,
 				output->ct.orig_tp.dst,
-				output->ct.orig_proto,
+				output->ct_orig_proto,
 			};
 			if (nla_put(skb, OVS_KEY_ATTR_CT_ORIG_TUPLE_IPV6,
 				    sizeof(orig), &orig))
@@ -614,11 +614,11 @@ static bool skb_nfct_cached(struct net *net,
 	 * due to an upcall.  If the connection was not confirmed, it is not
 	 * cached and needs to be run through conntrack again.
 	 */
-	if (!ct && key->ct.state & OVS_CS_F_TRACKED &&
-	    !(key->ct.state & OVS_CS_F_INVALID) &&
-	    key->ct.zone == info->zone.id) {
+	if (!ct && key->ct_state & OVS_CS_F_TRACKED &&
+	    !(key->ct_state & OVS_CS_F_INVALID) &&
+	    key->ct_zone == info->zone.id) {
 		ct = ovs_ct_find_existing(net, &info->zone, info->family, skb,
-					  !!(key->ct.state
+					  !!(key->ct_state
 					     & OVS_CS_F_NAT_MASK));
 		if (ct)
 			nf_ct_get(skb, &ctinfo);
@@ -742,7 +742,7 @@ static void ovs_nat_update_key(struct sw_flow_key *key,
 	if (maniptype == NF_NAT_MANIP_SRC) {
 		__be16 src;
 
-		key->ct.state |= OVS_CS_F_SRC_NAT;
+		key->ct_state |= OVS_CS_F_SRC_NAT;
 		if (key->eth.type == htons(ETH_P_IP))
 			key->ipv4.addr.src = ip_hdr(skb)->saddr;
 		else if (key->eth.type == htons(ETH_P_IPV6))
@@ -764,7 +764,7 @@ static void ovs_nat_update_key(struct sw_flow_key *key,
 	} else {
 		__be16 dst;
 
-		key->ct.state |= OVS_CS_F_DST_NAT;
+		key->ct_state |= OVS_CS_F_DST_NAT;
 		if (key->eth.type == htons(ETH_P_IP))
 			key->ipv4.addr.dst = ip_hdr(skb)->daddr;
 		else if (key->eth.type == htons(ETH_P_IPV6))
@@ -888,7 +888,7 @@ static int __ovs_ct_lookup(struct net *net, struct sw_flow_key *key,
 		 * NAT after the nf_conntrack_in() call.  We can actually clear
 		 * the whole state, as it will be re-initialized below.
 		 */
-		key->ct.state = 0;
+		key->ct_state = 0;
 
 		/* Update the key, but keep the NAT flags. */
 		ovs_ct_update_key(skb, info, key, true, true);
@@ -904,9 +904,9 @@ static int __ovs_ct_lookup(struct net *net, struct sw_flow_key *key,
 		 *
 		 * NAT will be done only if the CT action has NAT, and only
 		 * once per packet (per zone), as guarded by the NAT bits in
-		 * the key->ct.state.
+		 * the key->ct_state.
 		 */
-		if (info->nat && !(key->ct.state & OVS_CS_F_NAT_MASK) &&
+		if (info->nat && !(key->ct_state & OVS_CS_F_NAT_MASK) &&
 		    (nf_ct_is_confirmed(ct) || info->commit) &&
 		    ovs_ct_nat(net, key, info, skb, ct, ctinfo) != NF_ACCEPT) {
 			return -EINVAL;
diff --git a/net/openvswitch/conntrack.h b/net/openvswitch/conntrack.h
index 9e92445..bc7efd1 100644
--- a/net/openvswitch/conntrack.h
+++ b/net/openvswitch/conntrack.h
@@ -76,14 +76,14 @@ static inline int ovs_ct_execute(struct net *net, struct sk_buff *skb,
 static inline void ovs_ct_fill_key(const struct sk_buff *skb,
 				   struct sw_flow_key *key)
 {
-	key->ct.state = 0;
-	key->ct.zone = 0;
+	key->ct_state = 0;
+	key->ct_zone = 0;
 	key->ct.mark = 0;
 	memset(&key->ct.labels, 0, sizeof(key->ct.labels));
-	/* Clear 'ct.orig_proto' to mark the non-existence of original
+	/* Clear 'ct_orig_proto' to mark the non-existence of original
 	 * direction key fields.
 	 */
-	key->ct.orig_proto = 0;
+	key->ct_orig_proto = 0;
 }
 
 static inline int ovs_ct_put_key(const struct sw_flow_key *swkey,
diff --git a/net/openvswitch/flow.h b/net/openvswitch/flow.h
index 76e05b2..a9bc1c8 100644
--- a/net/openvswitch/flow.h
+++ b/net/openvswitch/flow.h
@@ -85,6 +85,11 @@ struct sw_flow_key {
 		struct vlan_head cvlan;
 		__be16 type;		/* Ethernet frame type. */
 	} eth;
+	/* Filling a hole of two bytes. */
+	u8 ct_state;
+	u8 ct_orig_proto;		/* CT original direction tuple IP
+					 * protocol.
+					 */
 	union {
 		struct {
 			__be32 top_lse;	/* top label stack entry */
@@ -96,6 +101,7 @@ struct sw_flow_key {
 			u8     frag;	/* One of OVS_FRAG_TYPE_*. */
 		} ip;
 	};
+	u16 ct_zone;			/* Conntrack zone. */
 	struct {
 		__be16 src;		/* TCP/UDP/SCTP source port. */
 		__be16 dst;		/* TCP/UDP/SCTP destination port. */
@@ -138,16 +144,12 @@ struct sw_flow_key {
 		} ipv6;
 	};
 	struct {
-		/* Connection tracking fields. */
-		u8 state;
-		u8 orig_proto;		/* CT orig tuple IP protocol. */
-		u16 zone;
-		u32 mark;
+		/* Connection tracking fields not packed above. */
 		struct {
 			__be16 src;	/* CT orig tuple tp src port. */
 			__be16 dst;	/* CT orig tuple tp dst port. */
 		} orig_tp;
-
+		u32 mark;
 		struct ovs_key_ct_labels labels;
 	} ct;
 
diff --git a/net/openvswitch/flow_netlink.c b/net/openvswitch/flow_netlink.c
index 989f38f..6f5fa50 100644
--- a/net/openvswitch/flow_netlink.c
+++ b/net/openvswitch/flow_netlink.c
@@ -1072,14 +1072,14 @@ static int metadata_from_nlattrs(struct net *net, struct sw_flow_match *match,
 			return -EINVAL;
 		}
 
-		SW_FLOW_KEY_PUT(match, ct.state, ct_state, is_mask);
+		SW_FLOW_KEY_PUT(match, ct_state, ct_state, is_mask);
 		*attrs &= ~(1ULL << OVS_KEY_ATTR_CT_STATE);
 	}
 	if (*attrs & (1 << OVS_KEY_ATTR_CT_ZONE) &&
 	    ovs_ct_verify(net, OVS_KEY_ATTR_CT_ZONE)) {
 		u16 ct_zone = nla_get_u16(a[OVS_KEY_ATTR_CT_ZONE]);
 
-		SW_FLOW_KEY_PUT(match, ct.zone, ct_zone, is_mask);
+		SW_FLOW_KEY_PUT(match, ct_zone, ct_zone, is_mask);
 		*attrs &= ~(1ULL << OVS_KEY_ATTR_CT_ZONE);
 	}
 	if (*attrs & (1 << OVS_KEY_ATTR_CT_MARK) &&
@@ -1107,7 +1107,7 @@ static int metadata_from_nlattrs(struct net *net, struct sw_flow_match *match,
 		SW_FLOW_KEY_PUT(match, ipv4.ct_orig.dst, ct->ipv4_dst, is_mask);
 		SW_FLOW_KEY_PUT(match, ct.orig_tp.src, ct->src_port, is_mask);
 		SW_FLOW_KEY_PUT(match, ct.orig_tp.dst, ct->dst_port, is_mask);
-		SW_FLOW_KEY_PUT(match, ct.orig_proto, ct->ipv4_proto, is_mask);
+		SW_FLOW_KEY_PUT(match, ct_orig_proto, ct->ipv4_proto, is_mask);
 		*attrs &= ~(1ULL << OVS_KEY_ATTR_CT_ORIG_TUPLE_IPV4);
 	}
 	if (*attrs & (1ULL << OVS_KEY_ATTR_CT_ORIG_TUPLE_IPV6)) {
@@ -1123,7 +1123,7 @@ static int metadata_from_nlattrs(struct net *net, struct sw_flow_match *match,
 				   is_mask);
 		SW_FLOW_KEY_PUT(match, ct.orig_tp.src, ct->src_port, is_mask);
 		SW_FLOW_KEY_PUT(match, ct.orig_tp.dst, ct->dst_port, is_mask);
-		SW_FLOW_KEY_PUT(match, ct.orig_proto, ct->ipv6_proto, is_mask);
+		SW_FLOW_KEY_PUT(match, ct_orig_proto, ct->ipv6_proto, is_mask);
 		*attrs &= ~(1ULL << OVS_KEY_ATTR_CT_ORIG_TUPLE_IPV6);
 	}
 
@@ -1564,6 +1564,9 @@ int ovs_nla_get_flow_metadata(struct net *net,
 	memset(&match, 0, sizeof(match));
 	match.key = key;
 
+	key->ct_state = 0;
+	key->ct_zone = 0;
+	key->ct_orig_proto = 0;
 	memset(&key->ct, 0, sizeof(key->ct));
 	memset(&key->ipv4.ct_orig, 0, sizeof(key->ipv4.ct_orig));
 	memset(&key->ipv6.ct_orig, 0, sizeof(key->ipv6.ct_orig));
-- 
2.1.4

Re: [PATCH v3 net-next 00/10] openvswitch: Conntrack integration improvements.

[Pravin Shelar]

On Wed, Feb 8, 2017 at 5:30 PM, Jarno Rajahalme <jarno@ovn.org> wrote:
> This series improves the conntrack integration code in the openvswitch
> module by fixing outdated comments (patch 1), bugs (patches 2, 3, and
> 7), clarifying code (patches 4, 5, and 6), improving performance
> (patch 10), and adding new features enabling better translation from
> firewall admission policy to network configuration requested by user
> communities (patches 8 and 9).
>
> v3: Rebase to the current net-next, add the comment only changing
>     patch 1 and reshuffle some of the patches as requested by Joe.
>

All patches looks good to me.

Acked-by: Pravin B Shelar <pshelar@ovn.org>

Thanks.

Re: [PATCH v3 net-next 00/10] openvswitch: Conntrack integration improvements.

[Jarno Rajahalme]

> On Feb 9, 2017, at 8:44 AM, Pravin Shelar <pshelar@ovn.org> wrote:
> 
> On Wed, Feb 8, 2017 at 5:30 PM, Jarno Rajahalme <jarno@ovn.org> wrote:
>> This series improves the conntrack integration code in the openvswitch
>> module by fixing outdated comments (patch 1), bugs (patches 2, 3, and
>> 7), clarifying code (patches 4, 5, and 6), improving performance
>> (patch 10), and adding new features enabling better translation from
>> firewall admission policy to network configuration requested by user
>> communities (patches 8 and 9).
>> 
>> v3: Rebase to the current net-next, add the comment only changing
>>    patch 1 and reshuffle some of the patches as requested by Joe.
>> 
> 
> All patches looks good to me.
> 
> Acked-by: Pravin B Shelar <pshelar@ovn.org>
> 
> Thanks.

Thanks for the review!

Joe spotted some english language and coding style issues I had not addressed in v3, so I just posted a v4. I included Joes and yours Acks, but not sure if that is the proper protocol, or if you need to Ack the v4 again. Sorry for the inconvenience!

  Jarno

Re: [PATCH v3 net-next 00/10] openvswitch: Conntrack integration improvements.

[David Miller]

From: Pravin Shelar <pshelar@ovn.org>
Date: Thu, 9 Feb 2017 08:44:55 -0800

> On Wed, Feb 8, 2017 at 5:30 PM, Jarno Rajahalme <jarno@ovn.org> wrote:
>> This series improves the conntrack integration code in the openvswitch
>> module by fixing outdated comments (patch 1), bugs (patches 2, 3, and
>> 7), clarifying code (patches 4, 5, and 6), improving performance
>> (patch 10), and adding new features enabling better translation from
>> firewall admission policy to network configuration requested by user
>> communities (patches 8 and 9).
>>
>> v3: Rebase to the current net-next, add the comment only changing
>>     patch 1 and reshuffle some of the patches as requested by Joe.
>>
> 
> All patches looks good to me.
> 
> Acked-by: Pravin B Shelar <pshelar@ovn.org>

Series applied.