datapath: Get packet metadata from userspace in odp_packet_cmd_execute().

Until now, the tun_id and in_port have been lost when a packet is sent from
the kernel to userspace and then back to the kernel.  I didn't think that
this was a problem, but recent behavior made me look closer and see that
it makes a difference if sFlow is turned on or if an
ODP_ATTR_ACTION_CONTROLLER action is present.  We could possibly kluge
around those, but for future-proofing it seems better to pass the packet
metadata from userspace to the kernel.  That is what this commit does.

This commit introduces a user-kernel protocol break.  We could avoid that,
if it is desirable, by making ODP_PACKET_ATTR_KEY optional for
ODP_PACKET_CMD_EXECUTE commands.

Signed-off-by: Ben Pfaff <blp@nicira.com>
Acked-by: Jesse Gross <jesse@nicira.com>

3 files changed, 84 insertions, 17 deletions

diff --git a/datapath/datapath.c b/datapath/datapath.c
index 728012249..d607315ed 100644
--- a/datapath/datapath.c
+++ b/datapath/datapath.c
@@ -655,7 +655,8 @@ static int odp_packet_cmd_execute(struct sk_buff *skb, struct genl_info *info)
 	int key_len;
 
 	err = -EINVAL;
-	if (!a[ODP_PACKET_ATTR_PACKET] || !a[ODP_PACKET_ATTR_ACTIONS] ||
+	if (!a[ODP_PACKET_ATTR_PACKET] || !a[ODP_PACKET_ATTR_KEY] ||
+	    !a[ODP_PACKET_ATTR_ACTIONS] ||
 	    nla_len(a[ODP_PACKET_ATTR_PACKET]) < ETH_HLEN)
 		goto err;
 
@@ -694,6 +695,12 @@ static int odp_packet_cmd_execute(struct sk_buff *skb, struct genl_info *info)
 		goto err_flow_put;
 	flow->tbl_node.hash = flow_hash(&flow->key, key_len);
 
+	err = flow_metadata_from_nlattrs(&flow->key.eth.in_port,
+					 &flow->key.eth.tun_id,
+					 a[ODP_PACKET_ATTR_KEY]);
+	if (err)
+		goto err_flow_put;
+
 	acts = flow_actions_alloc(a[ODP_PACKET_ATTR_ACTIONS]);
 	err = PTR_ERR(acts);
 	if (IS_ERR(acts))
@@ -725,6 +732,7 @@ err:
 
 static const struct nla_policy packet_policy[ODP_PACKET_ATTR_MAX + 1] = {
 	[ODP_PACKET_ATTR_PACKET] = { .type = NLA_UNSPEC },
+	[ODP_PACKET_ATTR_KEY] = { .type = NLA_NESTED },
 	[ODP_PACKET_ATTR_ACTIONS] = { .type = NLA_NESTED },
 };
 
diff --git a/datapath/flow.c b/datapath/flow.c
index 2b80c6d82..d181cdefb 100644
--- a/datapath/flow.c
+++ b/datapath/flow.c
@@ -597,6 +597,23 @@ int flow_cmp(const struct tbl_node *node, void *key2_, int len)
 	return !memcmp(key1, key2, len);
 }
 
+/* The size of the argument for each %ODP_KEY_ATTR_* Netlink attribute.  */
+static const u32 key_lens[ODP_KEY_ATTR_MAX + 1] = {
+	[ODP_KEY_ATTR_TUN_ID] = 8,
+	[ODP_KEY_ATTR_IN_PORT] = 4,
+	[ODP_KEY_ATTR_ETHERNET] = sizeof(struct odp_key_ethernet),
+	[ODP_KEY_ATTR_8021Q] = sizeof(struct odp_key_8021q),
+	[ODP_KEY_ATTR_ETHERTYPE] = 2,
+	[ODP_KEY_ATTR_IPV4] = sizeof(struct odp_key_ipv4),
+	[ODP_KEY_ATTR_IPV6] = sizeof(struct odp_key_ipv6),
+	[ODP_KEY_ATTR_TCP] = sizeof(struct odp_key_tcp),
+	[ODP_KEY_ATTR_UDP] = sizeof(struct odp_key_udp),
+	[ODP_KEY_ATTR_ICMP] = sizeof(struct odp_key_icmp),
+	[ODP_KEY_ATTR_ICMPV6] = sizeof(struct odp_key_icmpv6),
+	[ODP_KEY_ATTR_ARP] = sizeof(struct odp_key_arp),
+	[ODP_KEY_ATTR_ND] = sizeof(struct odp_key_nd),
+};
+
 /**
  * flow_from_nlattrs - parses Netlink attributes into a flow key.
  * @swkey: receives the extracted flow key.
@@ -625,22 +642,6 @@ int flow_from_nlattrs(struct sw_flow_key *swkey, int *key_lenp,
 
 	prev_type = ODP_KEY_ATTR_UNSPEC;
 	nla_for_each_nested(nla, attr, rem) {
-		static const u32 key_lens[ODP_KEY_ATTR_MAX + 1] = {
-			[ODP_KEY_ATTR_TUN_ID] = 8,
-			[ODP_KEY_ATTR_IN_PORT] = 4,
-			[ODP_KEY_ATTR_ETHERNET] = sizeof(struct odp_key_ethernet),
-			[ODP_KEY_ATTR_8021Q] = sizeof(struct odp_key_8021q),
-			[ODP_KEY_ATTR_ETHERTYPE] = 2,
-			[ODP_KEY_ATTR_IPV4] = sizeof(struct odp_key_ipv4),
-			[ODP_KEY_ATTR_IPV6] = sizeof(struct odp_key_ipv6),
-			[ODP_KEY_ATTR_TCP] = sizeof(struct odp_key_tcp),
-			[ODP_KEY_ATTR_UDP] = sizeof(struct odp_key_udp),
-			[ODP_KEY_ATTR_ICMP] = sizeof(struct odp_key_icmp),
-			[ODP_KEY_ATTR_ICMPV6] = sizeof(struct odp_key_icmpv6),
-			[ODP_KEY_ATTR_ARP] = sizeof(struct odp_key_arp),
-			[ODP_KEY_ATTR_ND] = sizeof(struct odp_key_nd),
-		};
-
 		const struct odp_key_ethernet *eth_key;
 		const struct odp_key_8021q *q_key;
 		const struct odp_key_ipv4 *ipv4_key;
@@ -868,6 +869,62 @@ ok:
 	return error;
 }
 
+/**
+ * flow_metadata_from_nlattrs - parses Netlink attributes into a flow key.
+ * @in_port: receives the extracted input port.
+ * @tun_id: receives the extracted tunnel ID.
+ * @key: Netlink attribute holding nested %ODP_KEY_ATTR_* Netlink attribute
+ * sequence.
+ *
+ * This parses a series of Netlink attributes that form a flow key, which must
+ * 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.
+ */
+int flow_metadata_from_nlattrs(u16 *in_port, __be64 *tun_id,
+			       const struct nlattr *attr)
+{
+	const struct nlattr *nla;
+	u16 prev_type;
+	int rem;
+
+	*tun_id = 0;
+
+	prev_type = ODP_KEY_ATTR_UNSPEC;
+	nla_for_each_nested(nla, attr, rem) {
+                int type = nla_type(nla);
+
+                if (type > ODP_KEY_ATTR_MAX || nla_len(nla) != key_lens[type])
+                        return -EINVAL;
+
+		switch (TRANSITION(prev_type, type)) {
+		case TRANSITION(ODP_KEY_ATTR_UNSPEC, ODP_KEY_ATTR_TUN_ID):
+			*tun_id = nla_get_be64(nla);
+			break;
+
+		case TRANSITION(ODP_KEY_ATTR_UNSPEC, ODP_KEY_ATTR_IN_PORT):
+		case TRANSITION(ODP_KEY_ATTR_TUN_ID, ODP_KEY_ATTR_IN_PORT):
+			if (nla_get_u32(nla) >= DP_MAX_PORTS)
+				return -EINVAL;
+			*in_port = nla_get_u32(nla);
+			break;
+
+		default:
+			goto done;
+		}
+
+		prev_type = type;
+	}
+	if (rem)
+		return -EINVAL;
+
+done:
+	if (prev_type == ODP_KEY_ATTR_UNSPEC ||
+	    prev_type == ODP_KEY_ATTR_TUN_ID)
+		return -EINVAL;
+	return 0;
+}
+
 int flow_to_nlattrs(const struct sw_flow_key *swkey, struct sk_buff *skb)
 {
 	struct odp_key_ethernet *eth_key;
diff --git a/datapath/flow.h b/datapath/flow.h
index 1d75a0188..3cda5967c 100644
--- a/datapath/flow.h
+++ b/datapath/flow.h
@@ -152,6 +152,8 @@ int flow_cmp(const struct tbl_node *, void *target, int len);
 int flow_to_nlattrs(const struct sw_flow_key *, struct sk_buff *);
 int flow_from_nlattrs(struct sw_flow_key *swkey, int *key_lenp,
 		      const struct nlattr *);
+int flow_metadata_from_nlattrs(u16 *in_port, __be64 *tun_id,
+			       const struct nlattr *);
 
 static inline struct sw_flow *flow_cast(const struct tbl_node *node)
 {