nfp: flower: verify pre-tunnel rules
authorJohn Hurley <john.hurley@netronome.com>
Sun, 4 Aug 2019 15:09:09 +0000 (16:09 +0100)
committerDavid S. Miller <davem@davemloft.net>
Tue, 6 Aug 2019 21:24:22 +0000 (14:24 -0700)
Pre-tunnel rules must direct packets to an internal port based on L2
information. Rules that egress to an internal port are already indicated
by a non-NULL device in its nfp_fl_payload struct. Verfiy the rest of the
match fields indicate that the rule is a pre-tunnel rule. This requires a
full match on the destination MAC address, an option VLAN field, and no
specific matches on other lower layer fields (with the exception of L4
proto and flags).

If a rule is identified as a pre-tunnel rule then mark it for offload to
the pre-tunnel table. Similarly, remove it from the pre-tunnel table on
rule deletion. The actual offloading of these commands is left to a
following patch.

Signed-off-by: John Hurley <john.hurley@netronome.com>
Reviewed-by: Simon Horman <simon.horman@netronome.com>
Acked-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
drivers/net/ethernet/netronome/nfp/flower/main.h
drivers/net/ethernet/netronome/nfp/flower/offload.c
drivers/net/ethernet/netronome/nfp/flower/tunnel_conf.c

index 6e9de4e..c3aa415 100644 (file)
@@ -283,6 +283,7 @@ struct nfp_fl_payload {
        bool in_hw;
        struct {
                struct net_device *dev;
+               __be16 vlan_tci;
        } pre_tun_rule;
 };
 
@@ -419,4 +420,8 @@ void
 nfp_flower_non_repr_priv_put(struct nfp_app *app, struct net_device *netdev);
 u32 nfp_flower_get_port_id_from_netdev(struct nfp_app *app,
                                       struct net_device *netdev);
+int nfp_flower_xmit_pre_tun_flow(struct nfp_app *app,
+                                struct nfp_fl_payload *flow);
+int nfp_flower_xmit_pre_tun_del_flow(struct nfp_app *app,
+                                    struct nfp_fl_payload *flow);
 #endif
index fba802a..ff8a9f1 100644 (file)
         NFP_FLOWER_LAYER_IPV4 | \
         NFP_FLOWER_LAYER_IPV6)
 
+#define NFP_FLOWER_PRE_TUN_RULE_FIELDS \
+       (NFP_FLOWER_LAYER_PORT | \
+        NFP_FLOWER_LAYER_MAC | \
+        NFP_FLOWER_LAYER_IPV4)
+
 struct nfp_flower_merge_check {
        union {
                struct {
@@ -1012,7 +1017,89 @@ nfp_flower_validate_pre_tun_rule(struct nfp_app *app,
                                 struct nfp_fl_payload *flow,
                                 struct netlink_ext_ack *extack)
 {
-       return -EOPNOTSUPP;
+       struct nfp_flower_meta_tci *meta_tci;
+       struct nfp_flower_mac_mpls *mac;
+       struct nfp_fl_act_head *act;
+       u8 *mask = flow->mask_data;
+       bool vlan = false;
+       int act_offset;
+       u8 key_layer;
+
+       meta_tci = (struct nfp_flower_meta_tci *)flow->unmasked_data;
+       if (meta_tci->tci & cpu_to_be16(NFP_FLOWER_MASK_VLAN_PRESENT)) {
+               u16 vlan_tci = be16_to_cpu(meta_tci->tci);
+
+               vlan_tci &= ~NFP_FLOWER_MASK_VLAN_PRESENT;
+               flow->pre_tun_rule.vlan_tci = cpu_to_be16(vlan_tci);
+               vlan = true;
+       } else {
+               flow->pre_tun_rule.vlan_tci = cpu_to_be16(0xffff);
+       }
+
+       key_layer = meta_tci->nfp_flow_key_layer;
+       if (key_layer & ~NFP_FLOWER_PRE_TUN_RULE_FIELDS) {
+               NL_SET_ERR_MSG_MOD(extack, "unsupported pre-tunnel rule: too many match fields");
+               return -EOPNOTSUPP;
+       }
+
+       if (!(key_layer & NFP_FLOWER_LAYER_MAC)) {
+               NL_SET_ERR_MSG_MOD(extack, "unsupported pre-tunnel rule: MAC fields match required");
+               return -EOPNOTSUPP;
+       }
+
+       /* Skip fields known to exist. */
+       mask += sizeof(struct nfp_flower_meta_tci);
+       mask += sizeof(struct nfp_flower_in_port);
+
+       /* Ensure destination MAC address is fully matched. */
+       mac = (struct nfp_flower_mac_mpls *)mask;
+       if (!is_broadcast_ether_addr(&mac->mac_dst[0])) {
+               NL_SET_ERR_MSG_MOD(extack, "unsupported pre-tunnel rule: dest MAC field must not be masked");
+               return -EOPNOTSUPP;
+       }
+
+       if (key_layer & NFP_FLOWER_LAYER_IPV4) {
+               int ip_flags = offsetof(struct nfp_flower_ipv4, ip_ext.flags);
+               int ip_proto = offsetof(struct nfp_flower_ipv4, ip_ext.proto);
+               int i;
+
+               mask += sizeof(struct nfp_flower_mac_mpls);
+
+               /* Ensure proto and flags are the only IP layer fields. */
+               for (i = 0; i < sizeof(struct nfp_flower_ipv4); i++)
+                       if (mask[i] && i != ip_flags && i != ip_proto) {
+                               NL_SET_ERR_MSG_MOD(extack, "unsupported pre-tunnel rule: only flags and proto can be matched in ip header");
+                               return -EOPNOTSUPP;
+                       }
+       }
+
+       /* Action must be a single egress or pop_vlan and egress. */
+       act_offset = 0;
+       act = (struct nfp_fl_act_head *)&flow->action_data[act_offset];
+       if (vlan) {
+               if (act->jump_id != NFP_FL_ACTION_OPCODE_POP_VLAN) {
+                       NL_SET_ERR_MSG_MOD(extack, "unsupported pre-tunnel rule: match on VLAN must have VLAN pop as first action");
+                       return -EOPNOTSUPP;
+               }
+
+               act_offset += act->len_lw << NFP_FL_LW_SIZ;
+               act = (struct nfp_fl_act_head *)&flow->action_data[act_offset];
+       }
+
+       if (act->jump_id != NFP_FL_ACTION_OPCODE_OUTPUT) {
+               NL_SET_ERR_MSG_MOD(extack, "unsupported pre-tunnel rule: non egress action detected where egress was expected");
+               return -EOPNOTSUPP;
+       }
+
+       act_offset += act->len_lw << NFP_FL_LW_SIZ;
+
+       /* Ensure there are no more actions after egress. */
+       if (act_offset != flow->meta.act_len) {
+               NL_SET_ERR_MSG_MOD(extack, "unsupported pre-tunnel rule: egress is not the last action");
+               return -EOPNOTSUPP;
+       }
+
+       return 0;
 }
 
 /**
@@ -1083,8 +1170,11 @@ nfp_flower_add_offload(struct nfp_app *app, struct net_device *netdev,
                goto err_release_metadata;
        }
 
-       err = nfp_flower_xmit_flow(app, flow_pay,
-                                  NFP_FLOWER_CMSG_TYPE_FLOW_ADD);
+       if (flow_pay->pre_tun_rule.dev)
+               err = nfp_flower_xmit_pre_tun_flow(app, flow_pay);
+       else
+               err = nfp_flower_xmit_flow(app, flow_pay,
+                                          NFP_FLOWER_CMSG_TYPE_FLOW_ADD);
        if (err)
                goto err_remove_rhash;
 
@@ -1226,8 +1316,11 @@ nfp_flower_del_offload(struct nfp_app *app, struct net_device *netdev,
                goto err_free_merge_flow;
        }
 
-       err = nfp_flower_xmit_flow(app, nfp_flow,
-                                  NFP_FLOWER_CMSG_TYPE_FLOW_DEL);
+       if (nfp_flow->pre_tun_rule.dev)
+               err = nfp_flower_xmit_pre_tun_del_flow(app, nfp_flow);
+       else
+               err = nfp_flower_xmit_flow(app, nfp_flow,
+                                          NFP_FLOWER_CMSG_TYPE_FLOW_DEL);
        /* Fall through on error. */
 
 err_free_merge_flow:
index a7a80f4..ef59481 100644 (file)
@@ -832,6 +832,18 @@ int nfp_tunnel_mac_event_handler(struct nfp_app *app,
        return NOTIFY_OK;
 }
 
+int nfp_flower_xmit_pre_tun_flow(struct nfp_app *app,
+                                struct nfp_fl_payload *flow)
+{
+       return -EOPNOTSUPP;
+}
+
+int nfp_flower_xmit_pre_tun_del_flow(struct nfp_app *app,
+                                    struct nfp_fl_payload *flow)
+{
+       return -EOPNOTSUPP;
+}
+
 int nfp_tunnel_config_start(struct nfp_app *app)
 {
        struct nfp_flower_priv *priv = app->priv;