net/ice: add flow director create and destroy
authorYahui Cao <yahui.cao@intel.com>
Fri, 18 Oct 2019 11:15:56 +0000 (19:15 +0800)
committerFerruh Yigit <ferruh.yigit@intel.com>
Wed, 23 Oct 2019 14:43:10 +0000 (16:43 +0200)
Add ice_create_fdir_filter to create a rule. If a flow is matched by
flow director filter, filter rule will be set to HW. For now common
pattern and queue/passthru/drop/mark actions are supported.

Signed-off-by: Yahui Cao <yahui.cao@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
Reviewed-by: Xiaolong Ye <xiaolong.ye@intel.com>
doc/guides/rel_notes/release_19_11.rst
drivers/net/ice/ice_ethdev.h
drivers/net/ice/ice_fdir_filter.c

index 0d11284..24470a3 100644 (file)
@@ -113,6 +113,7 @@ New Features
   * Added support for device-specific DDP package loading.
   * Added support for handling Receive Flex Descriptor.
   * Added support for protocol extraction on per Rx queue.
+  * Added support for Flow Director filter based on generic filter framework.
   * Added support for the ``RTE_ETH_DEV_CLOSE_REMOVE`` flag.
   * Generic filter enhancement
     - Supported pipeline mode.
index 2060ad5..914e065 100644 (file)
@@ -247,6 +247,11 @@ TAILQ_HEAD(ice_flow_list, rte_flow);
 struct ice_flow_parser_node;
 TAILQ_HEAD(ice_parser_list, ice_flow_parser_node);
 
+struct ice_fdir_filter_conf {
+       struct ice_fdir_fltr input;
+       uint64_t input_set;
+};
+
 /**
  *  A structure used to define fields of a FDIR related info.
  */
@@ -256,6 +261,7 @@ struct ice_fdir_info {
        struct ice_rx_queue *rxq;
        void *prg_pkt;                 /* memory for fdir program packet */
        uint64_t dma_addr;             /* physic address of packet memory*/
+       struct ice_fdir_filter_conf conf;
 };
 
 struct ice_pf {
index 60eb27e..9d4875c 100644 (file)
@@ -7,6 +7,55 @@
 #include "ice_rxtx.h"
 #include "ice_generic_flow.h"
 
+#define ICE_FDIR_IPV6_TC_OFFSET                20
+#define ICE_IPV6_TC_MASK               (0xFF << ICE_FDIR_IPV6_TC_OFFSET)
+
+#define ICE_FDIR_INSET_ETH_IPV4 (\
+       ICE_INSET_DMAC | \
+       ICE_INSET_IPV4_SRC | ICE_INSET_IPV4_DST | ICE_INSET_IPV4_TOS | \
+       ICE_INSET_IPV4_TTL | ICE_INSET_IPV4_PROTO)
+
+#define ICE_FDIR_INSET_ETH_IPV4_UDP (\
+       ICE_FDIR_INSET_ETH_IPV4 | \
+       ICE_INSET_UDP_SRC_PORT | ICE_INSET_UDP_DST_PORT)
+
+#define ICE_FDIR_INSET_ETH_IPV4_TCP (\
+       ICE_FDIR_INSET_ETH_IPV4 | \
+       ICE_INSET_TCP_SRC_PORT | ICE_INSET_TCP_DST_PORT)
+
+#define ICE_FDIR_INSET_ETH_IPV4_SCTP (\
+       ICE_FDIR_INSET_ETH_IPV4 | \
+       ICE_INSET_SCTP_SRC_PORT | ICE_INSET_SCTP_DST_PORT)
+
+#define ICE_FDIR_INSET_ETH_IPV6 (\
+       ICE_INSET_IPV6_SRC | ICE_INSET_IPV6_DST | ICE_INSET_IPV6_TC | \
+       ICE_INSET_IPV6_HOP_LIMIT | ICE_INSET_IPV6_NEXT_HDR)
+
+#define ICE_FDIR_INSET_ETH_IPV6_UDP (\
+       ICE_FDIR_INSET_ETH_IPV6 | \
+       ICE_INSET_UDP_SRC_PORT | ICE_INSET_UDP_DST_PORT)
+
+#define ICE_FDIR_INSET_ETH_IPV6_TCP (\
+       ICE_FDIR_INSET_ETH_IPV6 | \
+       ICE_INSET_TCP_SRC_PORT | ICE_INSET_TCP_DST_PORT)
+
+#define ICE_FDIR_INSET_ETH_IPV6_SCTP (\
+       ICE_FDIR_INSET_ETH_IPV6 | \
+       ICE_INSET_SCTP_SRC_PORT | ICE_INSET_SCTP_DST_PORT)
+
+static struct ice_pattern_match_item ice_fdir_pattern[] = {
+       {pattern_eth_ipv4,             ICE_FDIR_INSET_ETH_IPV4,              ICE_INSET_NONE},
+       {pattern_eth_ipv4_udp,         ICE_FDIR_INSET_ETH_IPV4_UDP,          ICE_INSET_NONE},
+       {pattern_eth_ipv4_tcp,         ICE_FDIR_INSET_ETH_IPV4_TCP,          ICE_INSET_NONE},
+       {pattern_eth_ipv4_sctp,        ICE_FDIR_INSET_ETH_IPV4_SCTP,         ICE_INSET_NONE},
+       {pattern_eth_ipv6,             ICE_FDIR_INSET_ETH_IPV6,              ICE_INSET_NONE},
+       {pattern_eth_ipv6_udp,         ICE_FDIR_INSET_ETH_IPV6_UDP,          ICE_INSET_NONE},
+       {pattern_eth_ipv6_tcp,         ICE_FDIR_INSET_ETH_IPV6_TCP,          ICE_INSET_NONE},
+       {pattern_eth_ipv6_sctp,        ICE_FDIR_INSET_ETH_IPV6_SCTP,         ICE_INSET_NONE},
+};
+
+static struct ice_flow_parser ice_fdir_parser;
+
 static const struct rte_memzone *
 ice_memzone_reserve(const char *name, uint32_t len, int socket_id)
 {
@@ -363,7 +412,7 @@ ice_fdir_input_set_parse(uint64_t inset, enum ice_flow_field *field)
        }
 }
 
-static int __rte_unused
+static int
 ice_fdir_input_set_conf(struct ice_pf *pf, enum ice_fltr_ptype flow,
                        uint64_t input_set, bool is_tunnel)
 {
@@ -457,7 +506,7 @@ ice_fdir_input_set_conf(struct ice_pf *pf, enum ice_fltr_ptype flow,
        }
 }
 
-static void __rte_unused
+static void
 ice_fdir_cnt_update(struct ice_pf *pf, enum ice_fltr_ptype ptype,
                    bool is_tunnel, bool add)
 {
@@ -476,8 +525,13 @@ static int
 ice_fdir_init(struct ice_adapter *ad)
 {
        struct ice_pf *pf = &ad->pf;
+       int ret;
+
+       ret = ice_fdir_setup(pf);
+       if (ret)
+               return ret;
 
-       return ice_fdir_setup(pf);
+       return ice_register_parser(&ice_fdir_parser, ad);
 }
 
 static void
@@ -485,15 +539,534 @@ ice_fdir_uninit(struct ice_adapter *ad)
 {
        struct ice_pf *pf = &ad->pf;
 
+       ice_unregister_parser(&ice_fdir_parser, ad);
+
        ice_fdir_teardown(pf);
 }
 
+static int
+ice_fdir_add_del_filter(struct ice_pf *pf,
+                       struct ice_fdir_filter_conf *filter,
+                       bool add)
+{
+       struct ice_fltr_desc desc;
+       struct ice_hw *hw = ICE_PF_TO_HW(pf);
+       unsigned char *pkt = (unsigned char *)pf->fdir.prg_pkt;
+       int ret;
+
+       filter->input.dest_vsi = pf->main_vsi->idx;
+
+       memset(&desc, 0, sizeof(desc));
+       ice_fdir_get_prgm_desc(hw, &filter->input, &desc, add);
+
+       memset(pkt, 0, ICE_FDIR_PKT_LEN);
+       ret = ice_fdir_get_prgm_pkt(&filter->input, pkt, false);
+       if (ret) {
+               PMD_DRV_LOG(ERR, "Generate dummy packet failed");
+               return -EINVAL;
+       }
+
+       return ice_fdir_programming(pf, &desc);
+}
+
+static int
+ice_fdir_create_filter(struct ice_adapter *ad,
+                      struct rte_flow *flow,
+                      void *meta,
+                      struct rte_flow_error *error)
+{
+       struct ice_pf *pf = &ad->pf;
+       struct ice_fdir_filter_conf *filter = meta;
+       struct ice_fdir_filter_conf *rule;
+       int ret;
+
+       rule = rte_zmalloc("fdir_entry", sizeof(*rule), 0);
+       if (!rule) {
+               rte_flow_error_set(error, ENOMEM,
+                                  RTE_FLOW_ERROR_TYPE_HANDLE, NULL,
+                                  "Failed to allocate memory");
+               return -rte_errno;
+       }
+
+       ret = ice_fdir_input_set_conf(pf, filter->input.flow_type,
+                       filter->input_set, false);
+       if (ret) {
+               rte_flow_error_set(error, -ret,
+                                  RTE_FLOW_ERROR_TYPE_HANDLE, NULL,
+                                  "Profile configure failed.");
+               goto free_entry;
+       }
+
+       ret = ice_fdir_add_del_filter(pf, filter, true);
+       if (ret) {
+               rte_flow_error_set(error, -ret,
+                                  RTE_FLOW_ERROR_TYPE_HANDLE, NULL,
+                                  "Add filter rule failed.");
+               goto free_entry;
+       }
+
+       rte_memcpy(rule, filter, sizeof(*rule));
+       flow->rule = rule;
+       ice_fdir_cnt_update(pf, filter->input.flow_type, false, true);
+       return 0;
+
+free_entry:
+       rte_free(rule);
+       return -rte_errno;
+}
+
+static int
+ice_fdir_destroy_filter(struct ice_adapter *ad,
+                       struct rte_flow *flow,
+                       struct rte_flow_error *error)
+{
+       struct ice_pf *pf = &ad->pf;
+       struct ice_fdir_filter_conf *filter;
+       int ret;
+
+       filter = (struct ice_fdir_filter_conf *)flow->rule;
+
+       ret = ice_fdir_add_del_filter(pf, filter, false);
+       if (ret) {
+               rte_flow_error_set(error, -ret,
+                                  RTE_FLOW_ERROR_TYPE_HANDLE, NULL,
+                                  "Del filter rule failed.");
+               return -rte_errno;
+       }
+
+       ice_fdir_cnt_update(pf, filter->input.flow_type, false, false);
+       flow->rule = NULL;
+
+       rte_free(filter);
+
+       return 0;
+}
+
 static struct ice_flow_engine ice_fdir_engine = {
        .init = ice_fdir_init,
        .uninit = ice_fdir_uninit,
+       .create = ice_fdir_create_filter,
+       .destroy = ice_fdir_destroy_filter,
        .type = ICE_FLOW_ENGINE_FDIR,
 };
 
+static int
+ice_fdir_parse_action(struct ice_adapter *ad,
+                     const struct rte_flow_action actions[],
+                     struct rte_flow_error *error,
+                     struct ice_fdir_filter_conf *filter)
+{
+       struct ice_pf *pf = &ad->pf;
+       const struct rte_flow_action_queue *act_q;
+       const struct rte_flow_action_mark *mark_spec = NULL;
+       uint32_t dest_num = 0;
+       uint32_t mark_num = 0;
+
+       for (; actions->type != RTE_FLOW_ACTION_TYPE_END; actions++) {
+               switch (actions->type) {
+               case RTE_FLOW_ACTION_TYPE_VOID:
+                       break;
+               case RTE_FLOW_ACTION_TYPE_QUEUE:
+                       dest_num++;
+
+                       act_q = actions->conf;
+                       filter->input.q_index = act_q->index;
+                       if (filter->input.q_index >=
+                                       pf->dev_data->nb_rx_queues) {
+                               rte_flow_error_set(error, EINVAL,
+                                                  RTE_FLOW_ERROR_TYPE_ACTION,
+                                                  actions,
+                                                  "Invalid queue for FDIR.");
+                               return -rte_errno;
+                       }
+                       filter->input.dest_ctl =
+                               ICE_FLTR_PRGM_DESC_DEST_DIRECT_PKT_QINDEX;
+                       break;
+               case RTE_FLOW_ACTION_TYPE_DROP:
+                       dest_num++;
+
+                       filter->input.dest_ctl =
+                               ICE_FLTR_PRGM_DESC_DEST_DROP_PKT;
+                       break;
+               case RTE_FLOW_ACTION_TYPE_PASSTHRU:
+                       dest_num++;
+
+                       filter->input.dest_ctl =
+                               ICE_FLTR_PRGM_DESC_DEST_DIRECT_PKT_QINDEX;
+                       filter->input.q_index = 0;
+                       break;
+               case RTE_FLOW_ACTION_TYPE_MARK:
+                       mark_num++;
+
+                       mark_spec = actions->conf;
+                       filter->input.fltr_id = mark_spec->id;
+                       break;
+               default:
+                       rte_flow_error_set(error, EINVAL,
+                                  RTE_FLOW_ERROR_TYPE_ACTION, actions,
+                                  "Invalid action.");
+                       return -rte_errno;
+               }
+       }
+
+       if (dest_num == 0 || dest_num >= 2) {
+               rte_flow_error_set(error, EINVAL,
+                          RTE_FLOW_ERROR_TYPE_ACTION, actions,
+                          "Unsupported action combination");
+               return -rte_errno;
+       }
+
+       if (mark_num >= 2) {
+               rte_flow_error_set(error, EINVAL,
+                          RTE_FLOW_ERROR_TYPE_ACTION, actions,
+                          "Too many mark actions");
+               return -rte_errno;
+       }
+
+       return 0;
+}
+
+static int
+ice_fdir_parse_pattern(__rte_unused struct ice_adapter *ad,
+                      const struct rte_flow_item pattern[],
+                      struct rte_flow_error *error,
+                      struct ice_fdir_filter_conf *filter)
+{
+       const struct rte_flow_item *item = pattern;
+       enum rte_flow_item_type item_type;
+       enum rte_flow_item_type l3 = RTE_FLOW_ITEM_TYPE_END;
+       const struct rte_flow_item_eth *eth_spec, *eth_mask;
+       const struct rte_flow_item_ipv4 *ipv4_spec, *ipv4_mask;
+       const struct rte_flow_item_ipv6 *ipv6_spec, *ipv6_mask;
+       const struct rte_flow_item_tcp *tcp_spec, *tcp_mask;
+       const struct rte_flow_item_udp *udp_spec, *udp_mask;
+       const struct rte_flow_item_sctp *sctp_spec, *sctp_mask;
+       uint64_t input_set = ICE_INSET_NONE;
+       uint8_t flow_type = ICE_FLTR_PTYPE_NONF_NONE;
+       uint8_t  ipv6_addr_mask[16] = {
+               0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
+               0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF
+       };
+       uint32_t vtc_flow_cpu;
+
+
+       for (item = pattern; item->type != RTE_FLOW_ITEM_TYPE_END; item++) {
+               if (item->last) {
+                       rte_flow_error_set(error, EINVAL,
+                                       RTE_FLOW_ERROR_TYPE_ITEM,
+                                       item,
+                                       "Not support range");
+                       return -rte_errno;
+               }
+               item_type = item->type;
+
+               switch (item_type) {
+               case RTE_FLOW_ITEM_TYPE_ETH:
+                       eth_spec = item->spec;
+                       eth_mask = item->mask;
+
+                       if (eth_spec && eth_mask) {
+                               if (!rte_is_zero_ether_addr(&eth_spec->src) ||
+                                   !rte_is_zero_ether_addr(&eth_mask->src)) {
+                                       rte_flow_error_set(error, EINVAL,
+                                               RTE_FLOW_ERROR_TYPE_ITEM,
+                                               item,
+                                               "Src mac not support");
+                                       return -rte_errno;
+                               }
+
+                               if (!rte_is_broadcast_ether_addr(&eth_mask->dst)) {
+                                       rte_flow_error_set(error, EINVAL,
+                                               RTE_FLOW_ERROR_TYPE_ITEM,
+                                               item,
+                                               "Invalid mac addr mask");
+                                       return -rte_errno;
+                               }
+
+                               input_set |= ICE_INSET_DMAC;
+                               rte_memcpy(&filter->input.ext_data.dst_mac,
+                                          &eth_spec->dst,
+                                          RTE_ETHER_ADDR_LEN);
+                       }
+                       break;
+               case RTE_FLOW_ITEM_TYPE_IPV4:
+                       l3 = RTE_FLOW_ITEM_TYPE_IPV4;
+                       ipv4_spec = item->spec;
+                       ipv4_mask = item->mask;
+
+                       if (ipv4_spec && ipv4_mask) {
+                               /* Check IPv4 mask and update input set */
+                               if (ipv4_mask->hdr.version_ihl ||
+                                   ipv4_mask->hdr.total_length ||
+                                   ipv4_mask->hdr.packet_id ||
+                                   ipv4_mask->hdr.fragment_offset ||
+                                   ipv4_mask->hdr.hdr_checksum) {
+                                       rte_flow_error_set(error, EINVAL,
+                                                  RTE_FLOW_ERROR_TYPE_ITEM,
+                                                  item,
+                                                  "Invalid IPv4 mask.");
+                                       return -rte_errno;
+                               }
+                               if (ipv4_mask->hdr.src_addr == UINT32_MAX)
+                                       input_set |= ICE_INSET_IPV4_SRC;
+                               if (ipv4_mask->hdr.dst_addr == UINT32_MAX)
+                                       input_set |= ICE_INSET_IPV4_DST;
+                               if (ipv4_mask->hdr.type_of_service == UINT8_MAX)
+                                       input_set |= ICE_INSET_IPV4_TOS;
+                               if (ipv4_mask->hdr.time_to_live == UINT8_MAX)
+                                       input_set |= ICE_INSET_IPV4_TTL;
+                               if (ipv4_mask->hdr.next_proto_id == UINT8_MAX)
+                                       input_set |= ICE_INSET_IPV4_PROTO;
+
+                               filter->input.ip.v4.dst_ip =
+                                       ipv4_spec->hdr.src_addr;
+                               filter->input.ip.v4.src_ip =
+                                       ipv4_spec->hdr.dst_addr;
+                               filter->input.ip.v4.tos =
+                                       ipv4_spec->hdr.type_of_service;
+                               filter->input.ip.v4.ttl =
+                                       ipv4_spec->hdr.time_to_live;
+                               filter->input.ip.v4.proto =
+                                       ipv4_spec->hdr.next_proto_id;
+                       }
+
+                       flow_type = ICE_FLTR_PTYPE_NONF_IPV4_OTHER;
+                       break;
+               case RTE_FLOW_ITEM_TYPE_IPV6:
+                       l3 = RTE_FLOW_ITEM_TYPE_IPV6;
+                       ipv6_spec = item->spec;
+                       ipv6_mask = item->mask;
+
+                       if (ipv6_spec && ipv6_mask) {
+                               /* Check IPv6 mask and update input set */
+                               if (ipv6_mask->hdr.payload_len) {
+                                       rte_flow_error_set(error, EINVAL,
+                                                  RTE_FLOW_ERROR_TYPE_ITEM,
+                                                  item,
+                                                  "Invalid IPv6 mask");
+                                       return -rte_errno;
+                               }
+
+                               if (!memcmp(ipv6_mask->hdr.src_addr,
+                                           ipv6_addr_mask,
+                                           RTE_DIM(ipv6_mask->hdr.src_addr)))
+                                       input_set |= ICE_INSET_IPV6_SRC;
+                               if (!memcmp(ipv6_mask->hdr.dst_addr,
+                                           ipv6_addr_mask,
+                                           RTE_DIM(ipv6_mask->hdr.dst_addr)))
+                                       input_set |= ICE_INSET_IPV6_DST;
+
+                               if ((ipv6_mask->hdr.vtc_flow &
+                                    rte_cpu_to_be_32(ICE_IPV6_TC_MASK))
+                                   == rte_cpu_to_be_32(ICE_IPV6_TC_MASK))
+                                       input_set |= ICE_INSET_IPV6_TC;
+                               if (ipv6_mask->hdr.proto == UINT8_MAX)
+                                       input_set |= ICE_INSET_IPV6_NEXT_HDR;
+                               if (ipv6_mask->hdr.hop_limits == UINT8_MAX)
+                                       input_set |= ICE_INSET_IPV6_HOP_LIMIT;
+
+                               rte_memcpy(filter->input.ip.v6.dst_ip,
+                                          ipv6_spec->hdr.src_addr, 16);
+                               rte_memcpy(filter->input.ip.v6.src_ip,
+                                          ipv6_spec->hdr.dst_addr, 16);
+
+                               vtc_flow_cpu =
+                                     rte_be_to_cpu_32(ipv6_spec->hdr.vtc_flow);
+                               filter->input.ip.v6.tc =
+                                       (uint8_t)(vtc_flow_cpu >>
+                                                 ICE_FDIR_IPV6_TC_OFFSET);
+                               filter->input.ip.v6.proto =
+                                       ipv6_spec->hdr.proto;
+                               filter->input.ip.v6.hlim =
+                                       ipv6_spec->hdr.hop_limits;
+                       }
+
+                       flow_type = ICE_FLTR_PTYPE_NONF_IPV6_OTHER;
+                       break;
+               case RTE_FLOW_ITEM_TYPE_TCP:
+                       tcp_spec = item->spec;
+                       tcp_mask = item->mask;
+
+                       if (tcp_spec && tcp_mask) {
+                               /* Check TCP mask and update input set */
+                               if (tcp_mask->hdr.sent_seq ||
+                                   tcp_mask->hdr.recv_ack ||
+                                   tcp_mask->hdr.data_off ||
+                                   tcp_mask->hdr.tcp_flags ||
+                                   tcp_mask->hdr.rx_win ||
+                                   tcp_mask->hdr.cksum ||
+                                   tcp_mask->hdr.tcp_urp) {
+                                       rte_flow_error_set(error, EINVAL,
+                                                  RTE_FLOW_ERROR_TYPE_ITEM,
+                                                  item,
+                                                  "Invalid TCP mask");
+                                       return -rte_errno;
+                               }
+
+                               if (tcp_mask->hdr.src_port == UINT16_MAX)
+                                       input_set |= ICE_INSET_TCP_SRC_PORT;
+                               if (tcp_mask->hdr.dst_port == UINT16_MAX)
+                                       input_set |= ICE_INSET_TCP_DST_PORT;
+
+                               /* Get filter info */
+                               if (l3 == RTE_FLOW_ITEM_TYPE_IPV4) {
+                                       filter->input.ip.v4.dst_port =
+                                               tcp_spec->hdr.src_port;
+                                       filter->input.ip.v4.src_port =
+                                               tcp_spec->hdr.dst_port;
+                                       flow_type =
+                                               ICE_FLTR_PTYPE_NONF_IPV4_TCP;
+                               } else if (l3 == RTE_FLOW_ITEM_TYPE_IPV6) {
+                                       filter->input.ip.v6.dst_port =
+                                               tcp_spec->hdr.src_port;
+                                       filter->input.ip.v6.src_port =
+                                               tcp_spec->hdr.dst_port;
+                                       flow_type =
+                                               ICE_FLTR_PTYPE_NONF_IPV6_TCP;
+                               }
+                       }
+                       break;
+               case RTE_FLOW_ITEM_TYPE_UDP:
+                       udp_spec = item->spec;
+                       udp_mask = item->mask;
+
+                       if (udp_spec && udp_mask) {
+                               /* Check UDP mask and update input set*/
+                               if (udp_mask->hdr.dgram_len ||
+                                   udp_mask->hdr.dgram_cksum) {
+                                       rte_flow_error_set(error, EINVAL,
+                                                  RTE_FLOW_ERROR_TYPE_ITEM,
+                                                  item,
+                                                  "Invalid UDP mask");
+                                       return -rte_errno;
+                               }
+
+                               if (udp_mask->hdr.src_port == UINT16_MAX)
+                                       input_set |= ICE_INSET_UDP_SRC_PORT;
+                               if (udp_mask->hdr.dst_port == UINT16_MAX)
+                                       input_set |= ICE_INSET_UDP_DST_PORT;
+
+                               /* Get filter info */
+                               if (l3 == RTE_FLOW_ITEM_TYPE_IPV4) {
+                                       filter->input.ip.v4.dst_port =
+                                               udp_spec->hdr.src_port;
+                                       filter->input.ip.v4.src_port =
+                                               udp_spec->hdr.dst_port;
+                                       flow_type =
+                                               ICE_FLTR_PTYPE_NONF_IPV4_UDP;
+                               } else if (l3 == RTE_FLOW_ITEM_TYPE_IPV6) {
+                                       filter->input.ip.v6.src_port =
+                                               udp_spec->hdr.src_port;
+                                       filter->input.ip.v6.dst_port =
+                                               udp_spec->hdr.dst_port;
+                                       flow_type =
+                                               ICE_FLTR_PTYPE_NONF_IPV6_UDP;
+                               }
+                       }
+                       break;
+               case RTE_FLOW_ITEM_TYPE_SCTP:
+                       sctp_spec = item->spec;
+                       sctp_mask = item->mask;
+
+                       if (sctp_spec && sctp_mask) {
+                               /* Check SCTP mask and update input set */
+                               if (sctp_mask->hdr.cksum) {
+                                       rte_flow_error_set(error, EINVAL,
+                                                  RTE_FLOW_ERROR_TYPE_ITEM,
+                                                  item,
+                                                  "Invalid UDP mask");
+                                       return -rte_errno;
+                               }
+
+                               if (sctp_mask->hdr.src_port == UINT16_MAX)
+                                       input_set |= ICE_INSET_SCTP_SRC_PORT;
+                               if (sctp_mask->hdr.dst_port == UINT16_MAX)
+                                       input_set |= ICE_INSET_SCTP_DST_PORT;
+
+                               /* Get filter info */
+                               if (l3 == RTE_FLOW_ITEM_TYPE_IPV4) {
+                                       filter->input.ip.v4.dst_port =
+                                               sctp_spec->hdr.src_port;
+                                       filter->input.ip.v4.src_port =
+                                               sctp_spec->hdr.dst_port;
+                                       flow_type =
+                                               ICE_FLTR_PTYPE_NONF_IPV4_SCTP;
+                               } else if (l3 == RTE_FLOW_ITEM_TYPE_IPV6) {
+                                       filter->input.ip.v6.dst_port =
+                                               sctp_spec->hdr.src_port;
+                                       filter->input.ip.v6.src_port =
+                                               sctp_spec->hdr.dst_port;
+                                       flow_type =
+                                               ICE_FLTR_PTYPE_NONF_IPV6_SCTP;
+                               }
+                       }
+                       break;
+               case RTE_FLOW_ITEM_TYPE_VOID:
+                       break;
+               default:
+                       rte_flow_error_set(error, EINVAL,
+                                  RTE_FLOW_ERROR_TYPE_ITEM,
+                                  item,
+                                  "Invalid pattern item.");
+                       return -rte_errno;
+               }
+       }
+
+       filter->input.flow_type = flow_type;
+       filter->input_set = input_set;
+
+       return 0;
+}
+
+static int
+ice_fdir_parse(struct ice_adapter *ad,
+              struct ice_pattern_match_item *array,
+              uint32_t array_len,
+              const struct rte_flow_item pattern[],
+              const struct rte_flow_action actions[],
+              void **meta,
+              struct rte_flow_error *error)
+{
+       struct ice_pf *pf = &ad->pf;
+       struct ice_fdir_filter_conf *filter = &pf->fdir.conf;
+       struct ice_pattern_match_item *item = NULL;
+       uint64_t input_set;
+       int ret;
+
+       memset(filter, 0, sizeof(*filter));
+       item = ice_search_pattern_match_item(pattern, array, array_len, error);
+       if (!item)
+               return -rte_errno;
+
+       ret = ice_fdir_parse_pattern(ad, pattern, error, filter);
+       if (ret)
+               return ret;
+       input_set = filter->input_set;
+       if (!input_set || input_set & ~item->input_set_mask) {
+               rte_flow_error_set(error, EINVAL,
+                                  RTE_FLOW_ERROR_TYPE_ITEM_SPEC,
+                                  pattern,
+                                  "Invalid input set");
+               return -rte_errno;
+       }
+
+       ret = ice_fdir_parse_action(ad, actions, error, filter);
+       if (ret)
+               return ret;
+
+       *meta = filter;
+
+       return 0;
+}
+
+static struct ice_flow_parser ice_fdir_parser = {
+       .engine = &ice_fdir_engine,
+       .array = ice_fdir_pattern,
+       .array_len = RTE_DIM(ice_fdir_pattern),
+       .parse_pattern_action = ice_fdir_parse,
+       .stage = ICE_FLOW_STAGE_DISTRIBUTOR,
+};
+
 RTE_INIT(ice_fdir_engine_register)
 {
        ice_register_flow_engine(&ice_fdir_engine);