net/mlx5: support modifying VLAN priority on VLAN header
authorMoti Haimovsky <motih@mellanox.com>
Mon, 9 Sep 2019 15:56:47 +0000 (18:56 +0300)
committerFerruh Yigit <ferruh.yigit@intel.com>
Fri, 20 Sep 2019 08:19:41 +0000 (10:19 +0200)
This commit adds support for modifying the VLAN priority (PCP) field
in about-to-be-pushed VLAN header.
This feature can only modify the PCP field of a new VLAN header yet
to be pushed. It does not support modifying an existing or already
pushed VLAN headers.

Signed-off-by: Moti Haimovsky <motih@mellanox.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@mellanox.com>
doc/guides/nics/mlx5.rst
doc/guides/rel_notes/release_19_11.rst
drivers/net/mlx5/mlx5_flow_dv.c

index 875a87f..01215f4 100644 (file)
@@ -132,6 +132,8 @@ Limitations
 
 - VLAN push offload is not supported on ingress traffic.
 
+- VLAN set PCP offload is not supported on existing headers.
+
 - A multi segment packet must have not more segments than reported by dev_infos_get()
   in tx_desc_lim.nb_seg_max field. This value depends on maximal supported Tx descriptor
   size and ``txq_inline_min`` settings and may be from 2 (worst case forced by maximal
@@ -1042,7 +1044,8 @@ Supported hardware offloads
    +-----------------------+-----------------+-----------------+
    | | VLAN                | | DPDK 19.11    | | DPDK 19.11    |
    | | (of_pop_vlan /      | | OFED 4.6-4    | | OFED 4.6-4    |
-   | | of_push_vlan)       | | ConnectX-5    | | ConnectX-5    |
+   | | of_push_vlan /      | | ConnectX-5    | | ConnectX-5    |
+   | | of_set_vlan_pcp)    |                 |                 |
    +-----------------------+-----------------+-----------------+
 
 Notes for testpmd
index 20718b9..761ddfb 100644 (file)
@@ -252,4 +252,5 @@ Tested Platforms
 
   * Added support for VLAN pop flow offload command.
   * Added support for VLAN push flow offload command.
+  * Added support for VLAN set PCP offload command.
 
index 2780735..73459a0 100644 (file)
@@ -973,6 +973,47 @@ flow_dv_validate_action_push_vlan(uint64_t action_flags,
        return 0;
 }
 
+/**
+ * Validate the set VLAN PCP.
+ *
+ * @param[in] action_flags
+ *   Holds the actions detected until now.
+ * @param[in] actions
+ *   Pointer to the list of actions remaining in the flow rule.
+ * @param[in] attr
+ *   Pointer to flow attributes
+ * @param[out] error
+ *   Pointer to error structure.
+ *
+ * @return
+ *   0 on success, a negative errno value otherwise and rte_errno is set.
+ */
+static int
+flow_dv_validate_action_set_vlan_pcp(uint64_t action_flags,
+                                    const struct rte_flow_action actions[],
+                                    struct rte_flow_error *error)
+{
+       const struct rte_flow_action *action = actions;
+       const struct rte_flow_action_of_set_vlan_pcp *conf = action->conf;
+
+       if (conf->vlan_pcp > 7)
+               return rte_flow_error_set(error, EINVAL,
+                                         RTE_FLOW_ERROR_TYPE_ACTION, action,
+                                         "VLAN PCP value is too big");
+       if (mlx5_flow_find_action(actions,
+                                 RTE_FLOW_ACTION_TYPE_OF_PUSH_VLAN) == NULL)
+               return rte_flow_error_set(error, ENOTSUP,
+                                         RTE_FLOW_ERROR_TYPE_ACTION, action,
+                                         "set VLAN PCP can only be used "
+                                         "with push VLAN action");
+       if (action_flags & MLX5_FLOW_ACTION_OF_PUSH_VLAN)
+               return rte_flow_error_set(error, ENOTSUP,
+                                         RTE_FLOW_ERROR_TYPE_ACTION, action,
+                                         "set VLAN PCP action must precede "
+                                         "the push VLAN action");
+       return 0;
+}
+
 /**
  * Validate count action.
  *
@@ -3396,6 +3437,13 @@ flow_dv_validate(struct rte_eth_dev *dev, const struct rte_flow_attr *attr,
                        action_flags |= MLX5_FLOW_ACTION_OF_PUSH_VLAN;
                        ++actions_n;
                        break;
+               case RTE_FLOW_ACTION_TYPE_OF_SET_VLAN_PCP:
+                       ret = flow_dv_validate_action_set_vlan_pcp
+                                               (action_flags, actions, error);
+                       if (ret < 0)
+                               return ret;
+                       /* Count PCP with push_vlan command. */
+                       break;
                case RTE_FLOW_ACTION_TYPE_VXLAN_ENCAP:
                case RTE_FLOW_ACTION_TYPE_NVGRE_ENCAP:
                        ret = flow_dv_validate_action_l2_encap(action_flags,
@@ -4989,6 +5037,7 @@ flow_dv_translate(struct rte_eth_dev *dev,
        uint8_t next_protocol = 0xff;
        struct rte_vlan_hdr vlan = { 0 };
        bool vlan_inherited = false;
+       uint16_t vlan_tci;
 
        flow->group = attr->group;
        if (attr->transfer)
@@ -5118,6 +5167,19 @@ cnt_err:
                                           dev_flow->dv.push_vlan_res->action;
                        action_flags |= MLX5_FLOW_ACTION_OF_PUSH_VLAN;
                        break;
+               case RTE_FLOW_ACTION_TYPE_OF_SET_VLAN_PCP:
+                       if (!vlan_inherited) {
+                               flow_dev_get_vlan_info_from_items(items, &vlan);
+                               vlan_inherited = true;
+                       }
+                       vlan_tci =
+                           ((const struct rte_flow_action_of_set_vlan_pcp *)
+                                                      actions->conf)->vlan_pcp;
+                       vlan_tci = vlan_tci << MLX5DV_FLOW_VLAN_PCP_SHIFT;
+                       vlan.vlan_tci &= ~MLX5DV_FLOW_VLAN_PCP_MASK;
+                       vlan.vlan_tci |= vlan_tci;
+                       /* Push VLAN command will use this value */
+                       break;
                case RTE_FLOW_ACTION_TYPE_VXLAN_ENCAP:
                case RTE_FLOW_ACTION_TYPE_NVGRE_ENCAP:
                        if (flow_dv_create_action_l2_encap(dev, actions,