openvswitch: kernel datapath clone action

Add 'clone' action to kernel datapath by using existing functions.
When actions within clone don't modify the current flow, the flow
key is not cloned before executing clone actions.

This is a follow up patch for this incomplete work:
https://patchwork.ozlabs.org/patch/722096/

v1 -> v2:
Refactor as advised by reviewer.

Signed-off-by: Yifeng Sun <pkusunyifeng@gmail.com>
Signed-off-by: Andy Zhou <azhou@ovn.org>
Acked-by: Pravin B Shelar <pshelar@ovn.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

diff --git a/include/linux/openvswitch.h b/include/linux/openvswitch.h
index e6b240b6196cae3a7dad22693af3857849fb0c8f..379affc63e24c64937c503f6e87afed63bbb8db9 100644 (file)
--- a/include/linux/openvswitch.h
+++ b/include/linux/openvswitch.h
@@ -21,4 +21,9 @@
 
 #include <uapi/linux/openvswitch.h>
 
+#define OVS_CLONE_ATTR_EXEC      0   /* Specify an u32 value. When nonzero,
+                                     * actions in clone will not change flow
+                                     * keys. False otherwise.
+                                     */
+
 #endif /* _LINUX_OPENVSWITCH_H */

diff --git a/include/uapi/linux/openvswitch.h b/include/uapi/linux/openvswitch.h
index 863aabaa5cc926f82667637eab6974a3dfb0199f..dbe0cbe4f1b72b9e0c4751791614e98b0825989f 100644 (file)
--- a/include/uapi/linux/openvswitch.h
+++ b/include/uapi/linux/openvswitch.h
@@ -840,6 +840,8 @@ struct ovs_action_push_eth {
  * @OVS_ACTION_ATTR_POP_NSH: pop the outermost NSH header off the packet.
  * @OVS_ACTION_ATTR_METER: Run packet through a meter, which may drop the
  * packet, or modify the packet (e.g., change the DSCP field).
+ * @OVS_ACTION_ATTR_CLONE: make a copy of the packet and execute a list of
+ * actions without affecting the original packet and key.
  *
  * Only a single header can be set with a single %OVS_ACTION_ATTR_SET.  Not all
  * fields within a header are modifiable, e.g. the IPv4 protocol and fragment
@@ -873,6 +875,7 @@ enum ovs_action_attr {
        OVS_ACTION_ATTR_PUSH_NSH,     /* Nested OVS_NSH_KEY_ATTR_*. */
        OVS_ACTION_ATTR_POP_NSH,      /* No argument. */
        OVS_ACTION_ATTR_METER,        /* u32 meter ID. */
+       OVS_ACTION_ATTR_CLONE,        /* Nested OVS_CLONE_ATTR_*.  */
 
        __OVS_ACTION_ATTR_MAX,        /* Nothing past this will be accepted
                                       * from userspace. */

diff --git a/net/openvswitch/actions.c b/net/openvswitch/actions.c
index 30a5df27116ec695d08bd4b57d8ee97b93822403..85ae53d8fd098b80e22a4b3ccc26f83be7b110c5 100644 (file)
--- a/net/openvswitch/actions.c
+++ b/net/openvswitch/actions.c
@@ -1057,6 +1057,28 @@ static int sample(struct datapath *dp, struct sk_buff *skb,
                             clone_flow_key);
 }
 
+/* When 'last' is true, clone() should always consume the 'skb'.
+ * Otherwise, clone() should keep 'skb' intact regardless what
+ * actions are executed within clone().
+ */
+static int clone(struct datapath *dp, struct sk_buff *skb,
+                struct sw_flow_key *key, const struct nlattr *attr,
+                bool last)
+{
+       struct nlattr *actions;
+       struct nlattr *clone_arg;
+       int rem = nla_len(attr);
+       bool dont_clone_flow_key;
+
+       /* The first action is always 'OVS_CLONE_ATTR_ARG'. */
+       clone_arg = nla_data(attr);
+       dont_clone_flow_key = nla_get_u32(clone_arg);
+       actions = nla_next(clone_arg, &rem);
+
+       return clone_execute(dp, skb, key, 0, actions, rem, last,
+                            !dont_clone_flow_key);
+}
+
 static void execute_hash(struct sk_buff *skb, struct sw_flow_key *key,
                         const struct nlattr *attr)
 {
@@ -1336,6 +1358,17 @@ static int do_execute_actions(struct datapath *dp, struct sk_buff *skb,
                                consume_skb(skb);
                                return 0;
                        }
+                       break;
+
+               case OVS_ACTION_ATTR_CLONE: {
+                       bool last = nla_is_last(a, rem);
+
+                       err = clone(dp, skb, key, a, last);
+                       if (last)
+                               return err;
+
+                       break;
+               }
                }
 
                if (unlikely(err)) {

diff --git a/net/openvswitch/flow_netlink.c b/net/openvswitch/flow_netlink.c
index 391c4073a6dcd7facffca0ccd4e7afe2d0a07d3f..a70097ecf33c2bf9e9df7b92c2359ab679ae6d7e 100644 (file)
--- a/net/openvswitch/flow_netlink.c
+++ b/net/openvswitch/flow_netlink.c
@@ -2460,6 +2460,40 @@ static int validate_and_copy_sample(struct net *net, const struct nlattr *attr,
        return 0;
 }
 
+static int validate_and_copy_clone(struct net *net,
+                                  const struct nlattr *attr,
+                                  const struct sw_flow_key *key,
+                                  struct sw_flow_actions **sfa,
+                                  __be16 eth_type, __be16 vlan_tci,
+                                  bool log, bool last)
+{
+       int start, err;
+       u32 exec;
+
+       if (nla_len(attr) && nla_len(attr) < NLA_HDRLEN)
+               return -EINVAL;
+
+       start = add_nested_action_start(sfa, OVS_ACTION_ATTR_CLONE, log);
+       if (start < 0)
+               return start;
+
+       exec = last || !actions_may_change_flow(attr);
+
+       err = ovs_nla_add_action(sfa, OVS_CLONE_ATTR_EXEC, &exec,
+                                sizeof(exec), log);
+       if (err)
+               return err;
+
+       err = __ovs_nla_copy_actions(net, attr, key, sfa,
+                                    eth_type, vlan_tci, log);
+       if (err)
+               return err;
+
+       add_nested_action_end(*sfa, start);
+
+       return 0;
+}
+
 void ovs_match_init(struct sw_flow_match *match,
                    struct sw_flow_key *key,
                    bool reset_key,
@@ -2849,6 +2883,7 @@ static int __ovs_nla_copy_actions(struct net *net, const struct nlattr *attr,
                        [OVS_ACTION_ATTR_PUSH_NSH] = (u32)-1,
                        [OVS_ACTION_ATTR_POP_NSH] = 0,
                        [OVS_ACTION_ATTR_METER] = sizeof(u32),
+                       [OVS_ACTION_ATTR_CLONE] = (u32)-1,
                };
                const struct ovs_action_push_vlan *vlan;
                int type = nla_type(a);
@@ -3038,6 +3073,18 @@ static int __ovs_nla_copy_actions(struct net *net, const struct nlattr *attr,
                        /* Non-existent meters are simply ignored.  */
                        break;
 
+               case OVS_ACTION_ATTR_CLONE: {
+                       bool last = nla_is_last(a, rem);
+
+                       err = validate_and_copy_clone(net, a, key, sfa,
+                                                     eth_type, vlan_tci,
+                                                     log, last);
+                       if (err)
+                               return err;
+                       skip_copy = true;
+                       break;
+               }
+
                default:
                        OVS_NLERR(log, "Unknown Action type %d", type);
                        return -EINVAL;
@@ -3116,6 +3163,26 @@ out:
        return err;
 }
 
+static int clone_action_to_attr(const struct nlattr *attr,
+                               struct sk_buff *skb)
+{
+       struct nlattr *start;
+       int err = 0, rem = nla_len(attr);
+
+       start = nla_nest_start(skb, OVS_ACTION_ATTR_CLONE);
+       if (!start)
+               return -EMSGSIZE;
+
+       err = ovs_nla_put_actions(nla_data(attr), rem, skb);
+
+       if (err)
+               nla_nest_cancel(skb, start);
+       else
+               nla_nest_end(skb, start);
+
+       return err;
+}
+
 static int set_action_to_attr(const struct nlattr *a, struct sk_buff *skb)
 {
        const struct nlattr *ovs_key = nla_data(a);
@@ -3204,6 +3271,12 @@ int ovs_nla_put_actions(const struct nlattr *attr, int len, struct sk_buff *skb)
                                return err;
                        break;
 
+               case OVS_ACTION_ATTR_CLONE:
+                       err = clone_action_to_attr(a, skb);
+                       if (err)
+                               return err;
+                       break;
+
                default:
                        if (nla_put(skb, type, nla_len(a), nla_data(a)))
                                return -EMSGSIZE;