Merge tag 'tty-3.20-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/tty
[deliverable/linux.git] / net / openvswitch / datapath.c
index b07349e82d788dba64e768cf3ea1da77b08a9652..ae5e77cdc0ca1f34ff7f9c99d65ba0c8bda9ace6 100644 (file)
@@ -65,6 +65,8 @@ static struct genl_family dp_packet_genl_family;
 static struct genl_family dp_flow_genl_family;
 static struct genl_family dp_datapath_genl_family;
 
+static const struct nla_policy flow_policy[];
+
 static const struct genl_multicast_group ovs_dp_flow_multicast_group = {
        .name = OVS_FLOW_MCGROUP,
 };
@@ -419,7 +421,7 @@ static int queue_userspace_packet(struct datapath *dp, struct sk_buff *skb,
        if (!dp_ifindex)
                return -ENODEV;
 
-       if (vlan_tx_tag_present(skb)) {
+       if (skb_vlan_tag_present(skb)) {
                nskb = skb_clone(skb, GFP_ATOMIC);
                if (!nskb)
                        return -ENOMEM;
@@ -461,10 +463,8 @@ static int queue_userspace_packet(struct datapath *dp, struct sk_buff *skb,
                             0, upcall_info->cmd);
        upcall->dp_ifindex = dp_ifindex;
 
-       nla = nla_nest_start(user_skb, OVS_PACKET_ATTR_KEY);
-       err = ovs_nla_put_flow(key, key, user_skb);
+       err = ovs_nla_put_key(key, key, OVS_PACKET_ATTR_KEY, false, user_skb);
        BUG_ON(err);
-       nla_nest_end(user_skb, nla);
 
        if (upcall_info->userdata)
                __nla_put(user_skb, OVS_PACKET_ATTR_USERDATA,
@@ -664,46 +664,48 @@ static void get_dp_stats(const struct datapath *dp, struct ovs_dp_stats *stats,
        }
 }
 
-static size_t ovs_flow_cmd_msg_size(const struct sw_flow_actions *acts)
+static bool should_fill_key(const struct sw_flow_id *sfid, uint32_t ufid_flags)
 {
-       return NLMSG_ALIGN(sizeof(struct ovs_header))
-               + nla_total_size(ovs_key_attr_size()) /* OVS_FLOW_ATTR_KEY */
-               + nla_total_size(ovs_key_attr_size()) /* OVS_FLOW_ATTR_MASK */
-               + nla_total_size(sizeof(struct ovs_flow_stats)) /* OVS_FLOW_ATTR_STATS */
-               + nla_total_size(1) /* OVS_FLOW_ATTR_TCP_FLAGS */
-               + nla_total_size(8) /* OVS_FLOW_ATTR_USED */
-               + nla_total_size(acts->actions_len); /* OVS_FLOW_ATTR_ACTIONS */
+       return ovs_identifier_is_ufid(sfid) &&
+              !(ufid_flags & OVS_UFID_F_OMIT_KEY);
 }
 
-/* Called with ovs_mutex or RCU read lock. */
-static int ovs_flow_cmd_fill_match(const struct sw_flow *flow,
-                                  struct sk_buff *skb)
+static bool should_fill_mask(uint32_t ufid_flags)
 {
-       struct nlattr *nla;
-       int err;
+       return !(ufid_flags & OVS_UFID_F_OMIT_MASK);
+}
 
-       /* Fill flow key. */
-       nla = nla_nest_start(skb, OVS_FLOW_ATTR_KEY);
-       if (!nla)
-               return -EMSGSIZE;
+static bool should_fill_actions(uint32_t ufid_flags)
+{
+       return !(ufid_flags & OVS_UFID_F_OMIT_ACTIONS);
+}
 
-       err = ovs_nla_put_flow(&flow->unmasked_key, &flow->unmasked_key, skb);
-       if (err)
-               return err;
+static size_t ovs_flow_cmd_msg_size(const struct sw_flow_actions *acts,
+                                   const struct sw_flow_id *sfid,
+                                   uint32_t ufid_flags)
+{
+       size_t len = NLMSG_ALIGN(sizeof(struct ovs_header));
 
-       nla_nest_end(skb, nla);
+       /* OVS_FLOW_ATTR_UFID */
+       if (sfid && ovs_identifier_is_ufid(sfid))
+               len += nla_total_size(sfid->ufid_len);
 
-       /* Fill flow mask. */
-       nla = nla_nest_start(skb, OVS_FLOW_ATTR_MASK);
-       if (!nla)
-               return -EMSGSIZE;
+       /* OVS_FLOW_ATTR_KEY */
+       if (!sfid || should_fill_key(sfid, ufid_flags))
+               len += nla_total_size(ovs_key_attr_size());
 
-       err = ovs_nla_put_flow(&flow->key, &flow->mask->key, skb);
-       if (err)
-               return err;
+       /* OVS_FLOW_ATTR_MASK */
+       if (should_fill_mask(ufid_flags))
+               len += nla_total_size(ovs_key_attr_size());
 
-       nla_nest_end(skb, nla);
-       return 0;
+       /* OVS_FLOW_ATTR_ACTIONS */
+       if (should_fill_actions(ufid_flags))
+               len += nla_total_size(acts->actions_len);
+
+       return len
+               + nla_total_size(sizeof(struct ovs_flow_stats)) /* OVS_FLOW_ATTR_STATS */
+               + nla_total_size(1) /* OVS_FLOW_ATTR_TCP_FLAGS */
+               + nla_total_size(8); /* OVS_FLOW_ATTR_USED */
 }
 
 /* Called with ovs_mutex or RCU read lock. */
@@ -774,7 +776,7 @@ static int ovs_flow_cmd_fill_actions(const struct sw_flow *flow,
 /* Called with ovs_mutex or RCU read lock. */
 static int ovs_flow_cmd_fill_info(const struct sw_flow *flow, int dp_ifindex,
                                  struct sk_buff *skb, u32 portid,
-                                 u32 seq, u32 flags, u8 cmd)
+                                 u32 seq, u32 flags, u8 cmd, u32 ufid_flags)
 {
        const int skb_orig_len = skb->len;
        struct ovs_header *ovs_header;
@@ -787,19 +789,34 @@ static int ovs_flow_cmd_fill_info(const struct sw_flow *flow, int dp_ifindex,
 
        ovs_header->dp_ifindex = dp_ifindex;
 
-       err = ovs_flow_cmd_fill_match(flow, skb);
+       err = ovs_nla_put_identifier(flow, skb);
        if (err)
                goto error;
 
+       if (should_fill_key(&flow->id, ufid_flags)) {
+               err = ovs_nla_put_masked_key(flow, skb);
+               if (err)
+                       goto error;
+       }
+
+       if (should_fill_mask(ufid_flags)) {
+               err = ovs_nla_put_mask(flow, skb);
+               if (err)
+                       goto error;
+       }
+
        err = ovs_flow_cmd_fill_stats(flow, skb);
        if (err)
                goto error;
 
-       err = ovs_flow_cmd_fill_actions(flow, skb, skb_orig_len);
-       if (err)
-               goto error;
+       if (should_fill_actions(ufid_flags)) {
+               err = ovs_flow_cmd_fill_actions(flow, skb, skb_orig_len);
+               if (err)
+                       goto error;
+       }
 
-       return genlmsg_end(skb, ovs_header);
+       genlmsg_end(skb, ovs_header);
+       return 0;
 
 error:
        genlmsg_cancel(skb, ovs_header);
@@ -808,15 +825,19 @@ error:
 
 /* May not be called with RCU read lock. */
 static struct sk_buff *ovs_flow_cmd_alloc_info(const struct sw_flow_actions *acts,
+                                              const struct sw_flow_id *sfid,
                                               struct genl_info *info,
-                                              bool always)
+                                              bool always,
+                                              uint32_t ufid_flags)
 {
        struct sk_buff *skb;
+       size_t len;
 
        if (!always && !ovs_must_notify(&dp_flow_genl_family, info, 0))
                return NULL;
 
-       skb = genlmsg_new_unicast(ovs_flow_cmd_msg_size(acts), info, GFP_KERNEL);
+       len = ovs_flow_cmd_msg_size(acts, sfid, ufid_flags);
+       skb = genlmsg_new_unicast(len, info, GFP_KERNEL);
        if (!skb)
                return ERR_PTR(-ENOMEM);
 
@@ -827,19 +848,19 @@ static struct sk_buff *ovs_flow_cmd_alloc_info(const struct sw_flow_actions *act
 static struct sk_buff *ovs_flow_cmd_build_info(const struct sw_flow *flow,
                                               int dp_ifindex,
                                               struct genl_info *info, u8 cmd,
-                                              bool always)
+                                              bool always, u32 ufid_flags)
 {
        struct sk_buff *skb;
        int retval;
 
-       skb = ovs_flow_cmd_alloc_info(ovsl_dereference(flow->sf_acts), info,
-                                     always);
+       skb = ovs_flow_cmd_alloc_info(ovsl_dereference(flow->sf_acts),
+                                     &flow->id, info, always, ufid_flags);
        if (IS_ERR_OR_NULL(skb))
                return skb;
 
        retval = ovs_flow_cmd_fill_info(flow, dp_ifindex, skb,
                                        info->snd_portid, info->snd_seq, 0,
-                                       cmd);
+                                       cmd, ufid_flags);
        BUG_ON(retval < 0);
        return skb;
 }
@@ -848,12 +869,14 @@ static int ovs_flow_cmd_new(struct sk_buff *skb, struct genl_info *info)
 {
        struct nlattr **a = info->attrs;
        struct ovs_header *ovs_header = info->userhdr;
-       struct sw_flow *flow, *new_flow;
+       struct sw_flow *flow = NULL, *new_flow;
        struct sw_flow_mask mask;
        struct sk_buff *reply;
        struct datapath *dp;
+       struct sw_flow_key key;
        struct sw_flow_actions *acts;
        struct sw_flow_match match;
+       u32 ufid_flags = ovs_nla_get_ufid_flags(a[OVS_FLOW_ATTR_UFID_FLAGS]);
        int error;
        bool log = !a[OVS_FLOW_ATTR_PROBE];
 
@@ -878,13 +901,19 @@ static int ovs_flow_cmd_new(struct sk_buff *skb, struct genl_info *info)
        }
 
        /* Extract key. */
-       ovs_match_init(&match, &new_flow->unmasked_key, &mask);
+       ovs_match_init(&match, &key, &mask);
        error = ovs_nla_get_match(&match, a[OVS_FLOW_ATTR_KEY],
                                  a[OVS_FLOW_ATTR_MASK], log);
        if (error)
                goto err_kfree_flow;
 
-       ovs_flow_mask_key(&new_flow->key, &new_flow->unmasked_key, &mask);
+       ovs_flow_mask_key(&new_flow->key, &key, &mask);
+
+       /* Extract flow identifier. */
+       error = ovs_nla_get_identifier(&new_flow->id, a[OVS_FLOW_ATTR_UFID],
+                                      &key, log);
+       if (error)
+               goto err_kfree_flow;
 
        /* Validate actions. */
        error = ovs_nla_copy_actions(a[OVS_FLOW_ATTR_ACTIONS], &new_flow->key,
@@ -894,7 +923,8 @@ static int ovs_flow_cmd_new(struct sk_buff *skb, struct genl_info *info)
                goto err_kfree_flow;
        }
 
-       reply = ovs_flow_cmd_alloc_info(acts, info, false);
+       reply = ovs_flow_cmd_alloc_info(acts, &new_flow->id, info, false,
+                                       ufid_flags);
        if (IS_ERR(reply)) {
                error = PTR_ERR(reply);
                goto err_kfree_acts;
@@ -906,8 +936,12 @@ static int ovs_flow_cmd_new(struct sk_buff *skb, struct genl_info *info)
                error = -ENODEV;
                goto err_unlock_ovs;
        }
+
        /* Check if this is a duplicate flow */
-       flow = ovs_flow_tbl_lookup(&dp->table, &new_flow->unmasked_key);
+       if (ovs_identifier_is_ufid(&new_flow->id))
+               flow = ovs_flow_tbl_lookup_ufid(&dp->table, &new_flow->id);
+       if (!flow)
+               flow = ovs_flow_tbl_lookup(&dp->table, &key);
        if (likely(!flow)) {
                rcu_assign_pointer(new_flow->sf_acts, acts);
 
@@ -923,7 +957,8 @@ static int ovs_flow_cmd_new(struct sk_buff *skb, struct genl_info *info)
                                                       ovs_header->dp_ifindex,
                                                       reply, info->snd_portid,
                                                       info->snd_seq, 0,
-                                                      OVS_FLOW_CMD_NEW);
+                                                      OVS_FLOW_CMD_NEW,
+                                                      ufid_flags);
                        BUG_ON(error < 0);
                }
                ovs_unlock();
@@ -941,10 +976,15 @@ static int ovs_flow_cmd_new(struct sk_buff *skb, struct genl_info *info)
                        error = -EEXIST;
                        goto err_unlock_ovs;
                }
-               /* The unmasked key has to be the same for flow updates. */
-               if (unlikely(!ovs_flow_cmp_unmasked_key(flow, &match))) {
-                       /* Look for any overlapping flow. */
-                       flow = ovs_flow_tbl_lookup_exact(&dp->table, &match);
+               /* The flow identifier has to be the same for flow updates.
+                * Look for any overlapping flow.
+                */
+               if (unlikely(!ovs_flow_cmp(flow, &match))) {
+                       if (ovs_identifier_is_key(&flow->id))
+                               flow = ovs_flow_tbl_lookup_exact(&dp->table,
+                                                                &match);
+                       else /* UFID matches but key is different */
+                               flow = NULL;
                        if (!flow) {
                                error = -ENOENT;
                                goto err_unlock_ovs;
@@ -959,7 +999,8 @@ static int ovs_flow_cmd_new(struct sk_buff *skb, struct genl_info *info)
                                                       ovs_header->dp_ifindex,
                                                       reply, info->snd_portid,
                                                       info->snd_seq, 0,
-                                                      OVS_FLOW_CMD_NEW);
+                                                      OVS_FLOW_CMD_NEW,
+                                                      ufid_flags);
                        BUG_ON(error < 0);
                }
                ovs_unlock();
@@ -1015,8 +1056,11 @@ static int ovs_flow_cmd_set(struct sk_buff *skb, struct genl_info *info)
        struct datapath *dp;
        struct sw_flow_actions *old_acts = NULL, *acts = NULL;
        struct sw_flow_match match;
+       struct sw_flow_id sfid;
+       u32 ufid_flags = ovs_nla_get_ufid_flags(a[OVS_FLOW_ATTR_UFID_FLAGS]);
        int error;
        bool log = !a[OVS_FLOW_ATTR_PROBE];
+       bool ufid_present;
 
        /* Extract key. */
        error = -EINVAL;
@@ -1025,6 +1069,7 @@ static int ovs_flow_cmd_set(struct sk_buff *skb, struct genl_info *info)
                goto error;
        }
 
+       ufid_present = ovs_nla_get_ufid(&sfid, a[OVS_FLOW_ATTR_UFID], log);
        ovs_match_init(&match, &key, &mask);
        error = ovs_nla_get_match(&match, a[OVS_FLOW_ATTR_KEY],
                                  a[OVS_FLOW_ATTR_MASK], log);
@@ -1041,7 +1086,8 @@ static int ovs_flow_cmd_set(struct sk_buff *skb, struct genl_info *info)
                }
 
                /* Can allocate before locking if have acts. */
-               reply = ovs_flow_cmd_alloc_info(acts, info, false);
+               reply = ovs_flow_cmd_alloc_info(acts, &sfid, info, false,
+                                               ufid_flags);
                if (IS_ERR(reply)) {
                        error = PTR_ERR(reply);
                        goto err_kfree_acts;
@@ -1055,7 +1101,10 @@ static int ovs_flow_cmd_set(struct sk_buff *skb, struct genl_info *info)
                goto err_unlock_ovs;
        }
        /* Check that the flow exists. */
-       flow = ovs_flow_tbl_lookup_exact(&dp->table, &match);
+       if (ufid_present)
+               flow = ovs_flow_tbl_lookup_ufid(&dp->table, &sfid);
+       else
+               flow = ovs_flow_tbl_lookup_exact(&dp->table, &match);
        if (unlikely(!flow)) {
                error = -ENOENT;
                goto err_unlock_ovs;
@@ -1071,13 +1120,16 @@ static int ovs_flow_cmd_set(struct sk_buff *skb, struct genl_info *info)
                                                       ovs_header->dp_ifindex,
                                                       reply, info->snd_portid,
                                                       info->snd_seq, 0,
-                                                      OVS_FLOW_CMD_NEW);
+                                                      OVS_FLOW_CMD_NEW,
+                                                      ufid_flags);
                        BUG_ON(error < 0);
                }
        } else {
                /* Could not alloc without acts before locking. */
                reply = ovs_flow_cmd_build_info(flow, ovs_header->dp_ifindex,
-                                               info, OVS_FLOW_CMD_NEW, false);
+                                               info, OVS_FLOW_CMD_NEW, false,
+                                               ufid_flags);
+
                if (unlikely(IS_ERR(reply))) {
                        error = PTR_ERR(reply);
                        goto err_unlock_ovs;
@@ -1114,17 +1166,22 @@ static int ovs_flow_cmd_get(struct sk_buff *skb, struct genl_info *info)
        struct sw_flow *flow;
        struct datapath *dp;
        struct sw_flow_match match;
-       int err;
+       struct sw_flow_id ufid;
+       u32 ufid_flags = ovs_nla_get_ufid_flags(a[OVS_FLOW_ATTR_UFID_FLAGS]);
+       int err = 0;
        bool log = !a[OVS_FLOW_ATTR_PROBE];
+       bool ufid_present;
 
-       if (!a[OVS_FLOW_ATTR_KEY]) {
+       ufid_present = ovs_nla_get_ufid(&ufid, a[OVS_FLOW_ATTR_UFID], log);
+       if (a[OVS_FLOW_ATTR_KEY]) {
+               ovs_match_init(&match, &key, NULL);
+               err = ovs_nla_get_match(&match, a[OVS_FLOW_ATTR_KEY], NULL,
+                                       log);
+       } else if (!ufid_present) {
                OVS_NLERR(log,
                          "Flow get message rejected, Key attribute missing.");
-               return -EINVAL;
+               err = -EINVAL;
        }
-
-       ovs_match_init(&match, &key, NULL);
-       err = ovs_nla_get_match(&match, a[OVS_FLOW_ATTR_KEY], NULL, log);
        if (err)
                return err;
 
@@ -1135,14 +1192,17 @@ static int ovs_flow_cmd_get(struct sk_buff *skb, struct genl_info *info)
                goto unlock;
        }
 
-       flow = ovs_flow_tbl_lookup_exact(&dp->table, &match);
+       if (ufid_present)
+               flow = ovs_flow_tbl_lookup_ufid(&dp->table, &ufid);
+       else
+               flow = ovs_flow_tbl_lookup_exact(&dp->table, &match);
        if (!flow) {
                err = -ENOENT;
                goto unlock;
        }
 
        reply = ovs_flow_cmd_build_info(flow, ovs_header->dp_ifindex, info,
-                                       OVS_FLOW_CMD_NEW, true);
+                                       OVS_FLOW_CMD_NEW, true, ufid_flags);
        if (IS_ERR(reply)) {
                err = PTR_ERR(reply);
                goto unlock;
@@ -1161,13 +1221,17 @@ static int ovs_flow_cmd_del(struct sk_buff *skb, struct genl_info *info)
        struct ovs_header *ovs_header = info->userhdr;
        struct sw_flow_key key;
        struct sk_buff *reply;
-       struct sw_flow *flow;
+       struct sw_flow *flow = NULL;
        struct datapath *dp;
        struct sw_flow_match match;
+       struct sw_flow_id ufid;
+       u32 ufid_flags = ovs_nla_get_ufid_flags(a[OVS_FLOW_ATTR_UFID_FLAGS]);
        int err;
        bool log = !a[OVS_FLOW_ATTR_PROBE];
+       bool ufid_present;
 
-       if (likely(a[OVS_FLOW_ATTR_KEY])) {
+       ufid_present = ovs_nla_get_ufid(&ufid, a[OVS_FLOW_ATTR_UFID], log);
+       if (a[OVS_FLOW_ATTR_KEY]) {
                ovs_match_init(&match, &key, NULL);
                err = ovs_nla_get_match(&match, a[OVS_FLOW_ATTR_KEY], NULL,
                                        log);
@@ -1182,12 +1246,15 @@ static int ovs_flow_cmd_del(struct sk_buff *skb, struct genl_info *info)
                goto unlock;
        }
 
-       if (unlikely(!a[OVS_FLOW_ATTR_KEY])) {
+       if (unlikely(!a[OVS_FLOW_ATTR_KEY] && !ufid_present)) {
                err = ovs_flow_tbl_flush(&dp->table);
                goto unlock;
        }
 
-       flow = ovs_flow_tbl_lookup_exact(&dp->table, &match);
+       if (ufid_present)
+               flow = ovs_flow_tbl_lookup_ufid(&dp->table, &ufid);
+       else
+               flow = ovs_flow_tbl_lookup_exact(&dp->table, &match);
        if (unlikely(!flow)) {
                err = -ENOENT;
                goto unlock;
@@ -1197,14 +1264,15 @@ static int ovs_flow_cmd_del(struct sk_buff *skb, struct genl_info *info)
        ovs_unlock();
 
        reply = ovs_flow_cmd_alloc_info((const struct sw_flow_actions __force *) flow->sf_acts,
-                                       info, false);
+                                       &flow->id, info, false, ufid_flags);
        if (likely(reply)) {
                if (likely(!IS_ERR(reply))) {
                        rcu_read_lock();        /*To keep RCU checker happy. */
                        err = ovs_flow_cmd_fill_info(flow, ovs_header->dp_ifindex,
                                                     reply, info->snd_portid,
                                                     info->snd_seq, 0,
-                                                    OVS_FLOW_CMD_DEL);
+                                                    OVS_FLOW_CMD_DEL,
+                                                    ufid_flags);
                        rcu_read_unlock();
                        BUG_ON(err < 0);
 
@@ -1223,9 +1291,18 @@ unlock:
 
 static int ovs_flow_cmd_dump(struct sk_buff *skb, struct netlink_callback *cb)
 {
+       struct nlattr *a[__OVS_FLOW_ATTR_MAX];
        struct ovs_header *ovs_header = genlmsg_data(nlmsg_data(cb->nlh));
        struct table_instance *ti;
        struct datapath *dp;
+       u32 ufid_flags;
+       int err;
+
+       err = genlmsg_parse(cb->nlh, &dp_flow_genl_family, a,
+                           OVS_FLOW_ATTR_MAX, flow_policy);
+       if (err)
+               return err;
+       ufid_flags = ovs_nla_get_ufid_flags(a[OVS_FLOW_ATTR_UFID_FLAGS]);
 
        rcu_read_lock();
        dp = get_dp_rcu(sock_net(skb->sk), ovs_header->dp_ifindex);
@@ -1248,7 +1325,7 @@ static int ovs_flow_cmd_dump(struct sk_buff *skb, struct netlink_callback *cb)
                if (ovs_flow_cmd_fill_info(flow, ovs_header->dp_ifindex, skb,
                                           NETLINK_CB(cb->skb).portid,
                                           cb->nlh->nlmsg_seq, NLM_F_MULTI,
-                                          OVS_FLOW_CMD_NEW) < 0)
+                                          OVS_FLOW_CMD_NEW, ufid_flags) < 0)
                        break;
 
                cb->args[0] = bucket;
@@ -1264,6 +1341,8 @@ static const struct nla_policy flow_policy[OVS_FLOW_ATTR_MAX + 1] = {
        [OVS_FLOW_ATTR_ACTIONS] = { .type = NLA_NESTED },
        [OVS_FLOW_ATTR_CLEAR] = { .type = NLA_FLAG },
        [OVS_FLOW_ATTR_PROBE] = { .type = NLA_FLAG },
+       [OVS_FLOW_ATTR_UFID] = { .type = NLA_UNSPEC, .len = 1 },
+       [OVS_FLOW_ATTR_UFID_FLAGS] = { .type = NLA_U32 },
 };
 
 static const struct genl_ops dp_flow_genl_ops[] = {
@@ -1349,7 +1428,8 @@ static int ovs_dp_cmd_fill_info(struct datapath *dp, struct sk_buff *skb,
        if (nla_put_u32(skb, OVS_DP_ATTR_USER_FEATURES, dp->user_features))
                goto nla_put_failure;
 
-       return genlmsg_end(skb, ovs_header);
+       genlmsg_end(skb, ovs_header);
+       return 0;
 
 nla_put_failure:
        genlmsg_cancel(skb, ovs_header);
@@ -1723,7 +1803,8 @@ static int ovs_vport_cmd_fill_info(struct vport *vport, struct sk_buff *skb,
        if (err == -EMSGSIZE)
                goto error;
 
-       return genlmsg_end(skb, ovs_header);
+       genlmsg_end(skb, ovs_header);
+       return 0;
 
 nla_put_failure:
        err = -EMSGSIZE;
This page took 0.031102 seconds and 5 git commands to generate.