net/i40e: fix flow director config after flow validate
[dpdk.git] / drivers / net / i40e / i40e_fdir.c
index d59399a..da089ba 100644 (file)
@@ -11,7 +11,7 @@
 #include <stdarg.h>
 
 #include <rte_ether.h>
-#include <rte_ethdev_driver.h>
+#include <ethdev_driver.h>
 #include <rte_log.h>
 #include <rte_memzone.h>
 #include <rte_malloc.h>
@@ -21,6 +21,7 @@
 #include <rte_tcp.h>
 #include <rte_sctp.h>
 #include <rte_hash_crc.h>
+#include <rte_bitmap.h>
 
 #include "i40e_logs.h"
 #include "base/i40e_type.h"
        (1ULL << RTE_ETH_FLOW_NONFRAG_IPV6_OTHER) | \
        (1ULL << RTE_ETH_FLOW_L2_PAYLOAD))
 
-static int i40e_fdir_filter_programming(struct i40e_pf *pf,
-                       enum i40e_filter_pctype pctype,
-                       const struct rte_eth_fdir_filter *filter,
-                       bool add);
 static int i40e_fdir_filter_convert(const struct i40e_fdir_filter_conf *input,
                         struct i40e_fdir_filter *filter);
 static struct i40e_fdir_filter *
@@ -99,7 +96,7 @@ static int
 i40e_flow_fdir_filter_programming(struct i40e_pf *pf,
                                  enum i40e_filter_pctype pctype,
                                  const struct i40e_fdir_filter_conf *filter,
-                                 bool add);
+                                 bool add, bool wait_status);
 
 static int
 i40e_fdir_rx_queue_init(struct i40e_rx_queue *rxq)
@@ -119,7 +116,7 @@ i40e_fdir_rx_queue_init(struct i40e_rx_queue *rxq)
 #endif
        rx_ctx.dtype = i40e_header_split_none;
        rx_ctx.hsplit_0 = I40E_HEADER_SPLIT_NONE;
-       rx_ctx.rxmax = RTE_ETHER_MAX_LEN;
+       rx_ctx.rxmax = I40E_ETH_MAX_LEN;
        rx_ctx.tphrdesc_ena = 1;
        rx_ctx.tphwdesc_ena = 1;
        rx_ctx.tphdata_ena = 1;
@@ -163,6 +160,7 @@ i40e_fdir_setup(struct i40e_pf *pf)
        char z_name[RTE_MEMZONE_NAMESIZE];
        const struct rte_memzone *mz = NULL;
        struct rte_eth_dev *eth_dev = pf->adapter->eth_dev;
+       uint16_t i;
 
        if ((pf->flags & I40E_FLAG_FDIR) == 0) {
                PMD_INIT_LOG(ERR, "HW doesn't support FDIR");
@@ -179,6 +177,7 @@ i40e_fdir_setup(struct i40e_pf *pf)
                PMD_DRV_LOG(INFO, "FDIR initialization has been done.");
                return I40E_SUCCESS;
        }
+
        /* make new FDIR VSI */
        vsi = i40e_vsi_setup(pf, I40E_VSI_FDIR, pf->main_vsi, 0);
        if (!vsi) {
@@ -228,22 +227,37 @@ i40e_fdir_setup(struct i40e_pf *pf)
                goto fail_mem;
        }
 
+       /* enable FDIR MSIX interrupt */
+       vsi->nb_used_qps = 1;
+       i40e_vsi_queues_bind_intr(vsi, I40E_ITR_INDEX_NONE);
+       i40e_vsi_enable_queues_intr(vsi);
+
        /* reserve memory for the fdir programming packet */
        snprintf(z_name, sizeof(z_name), "%s_%s_%d",
                        eth_dev->device->driver->name,
                        I40E_FDIR_MZ_NAME,
                        eth_dev->data->port_id);
-       mz = i40e_memzone_reserve(z_name, I40E_FDIR_PKT_LEN, SOCKET_ID_ANY);
+       mz = i40e_memzone_reserve(z_name, I40E_FDIR_PKT_LEN *
+                       I40E_FDIR_PRG_PKT_CNT, SOCKET_ID_ANY);
        if (!mz) {
                PMD_DRV_LOG(ERR, "Cannot init memzone for "
                                 "flow director program packet.");
                err = I40E_ERR_NO_MEMORY;
                goto fail_mem;
        }
-       pf->fdir.prg_pkt = mz->addr;
-       pf->fdir.dma_addr = mz->iova;
+
+       for (i = 0; i < I40E_FDIR_PRG_PKT_CNT; i++) {
+               pf->fdir.prg_pkt[i] = (uint8_t *)mz->addr +
+                       I40E_FDIR_PKT_LEN * i;
+               pf->fdir.dma_addr[i] = mz->iova +
+                       I40E_FDIR_PKT_LEN * i;
+       }
 
        pf->fdir.match_counter_index = I40E_COUNTER_INDEX_FDIR(hw->pf_id);
+       pf->fdir.fdir_actual_cnt = 0;
+       pf->fdir.fdir_guarantee_free_space =
+               pf->fdir.fdir_guarantee_total_space;
+
        PMD_DRV_LOG(INFO, "FDIR setup successfully, with programming queue %u.",
                    vsi->base_queue);
        return I40E_SUCCESS;
@@ -269,19 +283,28 @@ i40e_fdir_teardown(struct i40e_pf *pf)
 {
        struct i40e_hw *hw = I40E_PF_TO_HW(pf);
        struct i40e_vsi *vsi;
+       struct rte_eth_dev *dev = pf->adapter->eth_dev;
 
        vsi = pf->fdir.fdir_vsi;
        if (!vsi)
                return;
+
+       /* disable FDIR MSIX interrupt */
+       i40e_vsi_queues_unbind_intr(vsi);
+       i40e_vsi_disable_queues_intr(vsi);
+
        int err = i40e_switch_tx_queue(hw, vsi->base_queue, FALSE);
        if (err)
                PMD_DRV_LOG(DEBUG, "Failed to do FDIR TX switch off");
        err = i40e_switch_rx_queue(hw, vsi->base_queue, FALSE);
        if (err)
                PMD_DRV_LOG(DEBUG, "Failed to do FDIR RX switch off");
+
        i40e_dev_rx_queue_release(pf->fdir.rxq);
+       rte_eth_dma_zone_free(dev, "fdir_rx_ring", pf->fdir.rxq->queue_id);
        pf->fdir.rxq = NULL;
        i40e_dev_tx_queue_release(pf->fdir.txq);
+       rte_eth_dma_zone_free(dev, "fdir_tx_ring", pf->fdir.txq->queue_id);
        pf->fdir.txq = NULL;
        i40e_vsi_release(vsi);
        pf->fdir.fdir_vsi = NULL;
@@ -332,6 +355,7 @@ i40e_init_flx_pld(struct i40e_pf *pf)
                        I40E_PRTQF_FLX_PIT(index + 1), 0x0000FC29);/*non-used*/
                I40E_WRITE_REG(hw,
                        I40E_PRTQF_FLX_PIT(index + 2), 0x0000FC2A);/*non-used*/
+               pf->fdir.flex_pit_flag[i] = 0;
        }
 
        /* initialize the masks */
@@ -704,263 +728,6 @@ i40e_fdir_configure(struct rte_eth_dev *dev)
        return ret;
 }
 
-static inline int
-i40e_fdir_fill_eth_ip_head(const struct rte_eth_fdir_input *fdir_input,
-                          unsigned char *raw_pkt,
-                          bool vlan)
-{
-       static uint8_t vlan_frame[] = {0x81, 0, 0, 0};
-       uint16_t *ether_type;
-       uint8_t len = 2 * sizeof(struct rte_ether_addr);
-       struct rte_ipv4_hdr *ip;
-       struct rte_ipv6_hdr *ip6;
-       static const uint8_t next_proto[] = {
-               [RTE_ETH_FLOW_FRAG_IPV4] = IPPROTO_IP,
-               [RTE_ETH_FLOW_NONFRAG_IPV4_TCP] = IPPROTO_TCP,
-               [RTE_ETH_FLOW_NONFRAG_IPV4_UDP] = IPPROTO_UDP,
-               [RTE_ETH_FLOW_NONFRAG_IPV4_SCTP] = IPPROTO_SCTP,
-               [RTE_ETH_FLOW_NONFRAG_IPV4_OTHER] = IPPROTO_IP,
-               [RTE_ETH_FLOW_FRAG_IPV6] = IPPROTO_NONE,
-               [RTE_ETH_FLOW_NONFRAG_IPV6_TCP] = IPPROTO_TCP,
-               [RTE_ETH_FLOW_NONFRAG_IPV6_UDP] = IPPROTO_UDP,
-               [RTE_ETH_FLOW_NONFRAG_IPV6_SCTP] = IPPROTO_SCTP,
-               [RTE_ETH_FLOW_NONFRAG_IPV6_OTHER] = IPPROTO_NONE,
-       };
-
-       raw_pkt += 2 * sizeof(struct rte_ether_addr);
-       if (vlan && fdir_input->flow_ext.vlan_tci) {
-               rte_memcpy(raw_pkt, vlan_frame, sizeof(vlan_frame));
-               rte_memcpy(raw_pkt + sizeof(uint16_t),
-                          &fdir_input->flow_ext.vlan_tci,
-                          sizeof(uint16_t));
-               raw_pkt += sizeof(vlan_frame);
-               len += sizeof(vlan_frame);
-       }
-       ether_type = (uint16_t *)raw_pkt;
-       raw_pkt += sizeof(uint16_t);
-       len += sizeof(uint16_t);
-
-       switch (fdir_input->flow_type) {
-       case RTE_ETH_FLOW_L2_PAYLOAD:
-               *ether_type = fdir_input->flow.l2_flow.ether_type;
-               break;
-       case RTE_ETH_FLOW_NONFRAG_IPV4_TCP:
-       case RTE_ETH_FLOW_NONFRAG_IPV4_UDP:
-       case RTE_ETH_FLOW_NONFRAG_IPV4_SCTP:
-       case RTE_ETH_FLOW_NONFRAG_IPV4_OTHER:
-       case RTE_ETH_FLOW_FRAG_IPV4:
-               ip = (struct rte_ipv4_hdr *)raw_pkt;
-
-               *ether_type = rte_cpu_to_be_16(RTE_ETHER_TYPE_IPV4);
-               ip->version_ihl = I40E_FDIR_IP_DEFAULT_VERSION_IHL;
-               /* set len to by default */
-               ip->total_length = rte_cpu_to_be_16(I40E_FDIR_IP_DEFAULT_LEN);
-               ip->next_proto_id = fdir_input->flow.ip4_flow.proto ?
-                                       fdir_input->flow.ip4_flow.proto :
-                                       next_proto[fdir_input->flow_type];
-               ip->time_to_live = fdir_input->flow.ip4_flow.ttl ?
-                                       fdir_input->flow.ip4_flow.ttl :
-                                       I40E_FDIR_IP_DEFAULT_TTL;
-               ip->type_of_service = fdir_input->flow.ip4_flow.tos;
-               /*
-                * The source and destination fields in the transmitted packet
-                * need to be presented in a reversed order with respect
-                * to the expected received packets.
-                */
-               ip->src_addr = fdir_input->flow.ip4_flow.dst_ip;
-               ip->dst_addr = fdir_input->flow.ip4_flow.src_ip;
-               len += sizeof(struct rte_ipv4_hdr);
-               break;
-       case RTE_ETH_FLOW_NONFRAG_IPV6_TCP:
-       case RTE_ETH_FLOW_NONFRAG_IPV6_UDP:
-       case RTE_ETH_FLOW_NONFRAG_IPV6_SCTP:
-       case RTE_ETH_FLOW_NONFRAG_IPV6_OTHER:
-       case RTE_ETH_FLOW_FRAG_IPV6:
-               ip6 = (struct rte_ipv6_hdr *)raw_pkt;
-
-               *ether_type = rte_cpu_to_be_16(RTE_ETHER_TYPE_IPV6);
-               ip6->vtc_flow =
-                       rte_cpu_to_be_32(I40E_FDIR_IPv6_DEFAULT_VTC_FLOW |
-                                        (fdir_input->flow.ipv6_flow.tc <<
-                                         I40E_FDIR_IPv6_TC_OFFSET));
-               ip6->payload_len =
-                       rte_cpu_to_be_16(I40E_FDIR_IPv6_PAYLOAD_LEN);
-               ip6->proto = fdir_input->flow.ipv6_flow.proto ?
-                                       fdir_input->flow.ipv6_flow.proto :
-                                       next_proto[fdir_input->flow_type];
-               ip6->hop_limits = fdir_input->flow.ipv6_flow.hop_limits ?
-                                       fdir_input->flow.ipv6_flow.hop_limits :
-                                       I40E_FDIR_IPv6_DEFAULT_HOP_LIMITS;
-               /*
-                * The source and destination fields in the transmitted packet
-                * need to be presented in a reversed order with respect
-                * to the expected received packets.
-                */
-               rte_memcpy(&(ip6->src_addr),
-                          &(fdir_input->flow.ipv6_flow.dst_ip),
-                          IPV6_ADDR_LEN);
-               rte_memcpy(&(ip6->dst_addr),
-                          &(fdir_input->flow.ipv6_flow.src_ip),
-                          IPV6_ADDR_LEN);
-               len += sizeof(struct rte_ipv6_hdr);
-               break;
-       default:
-               PMD_DRV_LOG(ERR, "unknown flow type %u.",
-                           fdir_input->flow_type);
-               return -1;
-       }
-       return len;
-}
-
-
-/*
- * i40e_fdir_construct_pkt - construct packet based on fields in input
- * @pf: board private structure
- * @fdir_input: input set of the flow director entry
- * @raw_pkt: a packet to be constructed
- */
-static int
-i40e_fdir_construct_pkt(struct i40e_pf *pf,
-                            const struct rte_eth_fdir_input *fdir_input,
-                            unsigned char *raw_pkt)
-{
-       unsigned char *payload, *ptr;
-       struct rte_udp_hdr *udp;
-       struct rte_tcp_hdr *tcp;
-       struct rte_sctp_hdr *sctp;
-       uint8_t size, dst = 0;
-       uint8_t i, pit_idx, set_idx = I40E_FLXPLD_L4_IDX; /* use l4 by default*/
-       int len;
-
-       /* fill the ethernet and IP head */
-       len = i40e_fdir_fill_eth_ip_head(fdir_input, raw_pkt,
-                                        !!fdir_input->flow_ext.vlan_tci);
-       if (len < 0)
-               return -EINVAL;
-
-       /* fill the L4 head */
-       switch (fdir_input->flow_type) {
-       case RTE_ETH_FLOW_NONFRAG_IPV4_UDP:
-               udp = (struct rte_udp_hdr *)(raw_pkt + len);
-               payload = (unsigned char *)udp + sizeof(struct rte_udp_hdr);
-               /*
-                * The source and destination fields in the transmitted packet
-                * need to be presented in a reversed order with respect
-                * to the expected received packets.
-                */
-               udp->src_port = fdir_input->flow.udp4_flow.dst_port;
-               udp->dst_port = fdir_input->flow.udp4_flow.src_port;
-               udp->dgram_len = rte_cpu_to_be_16(I40E_FDIR_UDP_DEFAULT_LEN);
-               break;
-
-       case RTE_ETH_FLOW_NONFRAG_IPV4_TCP:
-               tcp = (struct rte_tcp_hdr *)(raw_pkt + len);
-               payload = (unsigned char *)tcp + sizeof(struct rte_tcp_hdr);
-               /*
-                * The source and destination fields in the transmitted packet
-                * need to be presented in a reversed order with respect
-                * to the expected received packets.
-                */
-               tcp->src_port = fdir_input->flow.tcp4_flow.dst_port;
-               tcp->dst_port = fdir_input->flow.tcp4_flow.src_port;
-               tcp->data_off = I40E_FDIR_TCP_DEFAULT_DATAOFF;
-               break;
-
-       case RTE_ETH_FLOW_NONFRAG_IPV4_SCTP:
-               sctp = (struct rte_sctp_hdr *)(raw_pkt + len);
-               payload = (unsigned char *)sctp + sizeof(struct rte_sctp_hdr);
-               /*
-                * The source and destination fields in the transmitted packet
-                * need to be presented in a reversed order with respect
-                * to the expected received packets.
-                */
-               sctp->src_port = fdir_input->flow.sctp4_flow.dst_port;
-               sctp->dst_port = fdir_input->flow.sctp4_flow.src_port;
-               sctp->tag = fdir_input->flow.sctp4_flow.verify_tag;
-               break;
-
-       case RTE_ETH_FLOW_NONFRAG_IPV4_OTHER:
-       case RTE_ETH_FLOW_FRAG_IPV4:
-               payload = raw_pkt + len;
-               set_idx = I40E_FLXPLD_L3_IDX;
-               break;
-
-       case RTE_ETH_FLOW_NONFRAG_IPV6_UDP:
-               udp = (struct rte_udp_hdr *)(raw_pkt + len);
-               payload = (unsigned char *)udp + sizeof(struct rte_udp_hdr);
-               /*
-                * The source and destination fields in the transmitted packet
-                * need to be presented in a reversed order with respect
-                * to the expected received packets.
-                */
-               udp->src_port = fdir_input->flow.udp6_flow.dst_port;
-               udp->dst_port = fdir_input->flow.udp6_flow.src_port;
-               udp->dgram_len = rte_cpu_to_be_16(I40E_FDIR_IPv6_PAYLOAD_LEN);
-               break;
-
-       case RTE_ETH_FLOW_NONFRAG_IPV6_TCP:
-               tcp = (struct rte_tcp_hdr *)(raw_pkt + len);
-               payload = (unsigned char *)tcp + sizeof(struct rte_tcp_hdr);
-               /*
-                * The source and destination fields in the transmitted packet
-                * need to be presented in a reversed order with respect
-                * to the expected received packets.
-                */
-               tcp->data_off = I40E_FDIR_TCP_DEFAULT_DATAOFF;
-               tcp->src_port = fdir_input->flow.udp6_flow.dst_port;
-               tcp->dst_port = fdir_input->flow.udp6_flow.src_port;
-               break;
-
-       case RTE_ETH_FLOW_NONFRAG_IPV6_SCTP:
-               sctp = (struct rte_sctp_hdr *)(raw_pkt + len);
-               payload = (unsigned char *)sctp + sizeof(struct rte_sctp_hdr);
-               /*
-                * The source and destination fields in the transmitted packet
-                * need to be presented in a reversed order with respect
-                * to the expected received packets.
-                */
-               sctp->src_port = fdir_input->flow.sctp6_flow.dst_port;
-               sctp->dst_port = fdir_input->flow.sctp6_flow.src_port;
-               sctp->tag = fdir_input->flow.sctp6_flow.verify_tag;
-               break;
-
-       case RTE_ETH_FLOW_NONFRAG_IPV6_OTHER:
-       case RTE_ETH_FLOW_FRAG_IPV6:
-               payload = raw_pkt + len;
-               set_idx = I40E_FLXPLD_L3_IDX;
-               break;
-       case RTE_ETH_FLOW_L2_PAYLOAD:
-               payload = raw_pkt + len;
-               /*
-                * ARP packet is a special case on which the payload
-                * starts after the whole ARP header
-                */
-               if (fdir_input->flow.l2_flow.ether_type ==
-                               rte_cpu_to_be_16(RTE_ETHER_TYPE_ARP))
-                       payload += sizeof(struct rte_arp_hdr);
-               set_idx = I40E_FLXPLD_L2_IDX;
-               break;
-       default:
-               PMD_DRV_LOG(ERR, "unknown flow type %u.", fdir_input->flow_type);
-               return -EINVAL;
-       }
-
-       /* fill the flexbytes to payload */
-       for (i = 0; i < I40E_MAX_FLXPLD_FIED; i++) {
-               pit_idx = set_idx * I40E_MAX_FLXPLD_FIED + i;
-               size = pf->fdir.flex_set[pit_idx].size;
-               if (size == 0)
-                       continue;
-               dst = pf->fdir.flex_set[pit_idx].dst_offset * sizeof(uint16_t);
-               ptr = payload +
-                       pf->fdir.flex_set[pit_idx].src_offset * sizeof(uint16_t);
-               rte_memcpy(ptr,
-                                &fdir_input->flow_ext.flexbytes[dst],
-                                size * sizeof(uint16_t));
-       }
-
-       return 0;
-}
 
 static struct i40e_customized_pctype *
 i40e_flow_fdir_find_customized_pctype(struct i40e_pf *pf, uint8_t pctype)
@@ -1522,6 +1289,17 @@ i40e_check_fdir_programming_status(struct i40e_rx_queue *rxq)
        return ret;
 }
 
+static inline void
+i40e_fdir_programming_status_cleanup(struct i40e_rx_queue *rxq)
+{
+       uint16_t retry_count = 0;
+
+       /* capture the previous error report(if any) from rx ring */
+       while ((i40e_check_fdir_programming_status(rxq) < 0) &&
+                       (++retry_count < I40E_FDIR_NUM_RX_DESC))
+               PMD_DRV_LOG(INFO, "error report captured.");
+}
+
 static int
 i40e_fdir_filter_convert(const struct i40e_fdir_filter_conf *input,
                         struct i40e_fdir_filter *filter)
@@ -1562,6 +1340,7 @@ static int
 i40e_sw_fdir_filter_insert(struct i40e_pf *pf, struct i40e_fdir_filter *filter)
 {
        struct i40e_fdir_info *fdir_info = &pf->fdir;
+       struct i40e_fdir_filter *hash_filter;
        int ret;
 
        if (filter->fdir.input.flow_ext.pkt_template)
@@ -1577,9 +1356,14 @@ i40e_sw_fdir_filter_insert(struct i40e_pf *pf, struct i40e_fdir_filter *filter)
                            ret);
                return ret;
        }
-       fdir_info->hash_map[ret] = filter;
 
-       TAILQ_INSERT_TAIL(&fdir_info->fdir_list, filter, rules);
+       if (fdir_info->hash_map[ret])
+               return -1;
+
+       hash_filter = &fdir_info->fdir_filter_array[ret];
+       rte_memcpy(hash_filter, filter, sizeof(*filter));
+       fdir_info->hash_map[ret] = hash_filter;
+       TAILQ_INSERT_TAIL(&fdir_info->fdir_list, hash_filter, rules);
 
        return 0;
 }
@@ -1608,71 +1392,316 @@ i40e_sw_fdir_filter_del(struct i40e_pf *pf, struct i40e_fdir_input *input)
        fdir_info->hash_map[ret] = NULL;
 
        TAILQ_REMOVE(&fdir_info->fdir_list, filter, rules);
-       rte_free(filter);
 
        return 0;
 }
 
-/*
- * i40e_add_del_fdir_filter - add or remove a flow director filter.
- * @pf: board private structure
- * @filter: fdir filter entry
- * @add: 0 - delete, 1 - add
- */
-int
-i40e_add_del_fdir_filter(struct rte_eth_dev *dev,
-                        const struct rte_eth_fdir_filter *filter,
-                        bool add)
+struct rte_flow *
+i40e_fdir_entry_pool_get(struct i40e_fdir_info *fdir_info)
 {
-       struct i40e_hw *hw = I40E_DEV_PRIVATE_TO_HW(dev->data->dev_private);
-       struct i40e_pf *pf = I40E_DEV_PRIVATE_TO_PF(dev->data->dev_private);
-       unsigned char *pkt = (unsigned char *)pf->fdir.prg_pkt;
-       enum i40e_filter_pctype pctype;
-       int ret = 0;
+       struct rte_flow *flow = NULL;
+       uint64_t slab = 0;
+       uint32_t pos = 0;
+       uint32_t i = 0;
+       int ret;
 
-       if (dev->data->dev_conf.fdir_conf.mode != RTE_FDIR_MODE_PERFECT) {
-               PMD_DRV_LOG(ERR, "FDIR is not enabled, please"
-                       " check the mode in fdir_conf.");
-               return -ENOTSUP;
+       if (fdir_info->fdir_actual_cnt >=
+                       fdir_info->fdir_space_size) {
+               PMD_DRV_LOG(ERR, "Fdir space full");
+               return NULL;
        }
 
-       pctype = i40e_flowtype_to_pctype(pf->adapter, filter->input.flow_type);
-       if (pctype == I40E_FILTER_PCTYPE_INVALID) {
-               PMD_DRV_LOG(ERR, "invalid flow_type input.");
-               return -EINVAL;
+       ret = rte_bitmap_scan(fdir_info->fdir_flow_pool.bitmap, &pos,
+                       &slab);
+
+       /* normally this won't happen as the fdir_actual_cnt should be
+        * same with the number of the set bits in fdir_flow_pool,
+        * but anyway handle this error condition here for safe
+        */
+       if (ret == 0) {
+               PMD_DRV_LOG(ERR, "fdir_actual_cnt out of sync");
+               return NULL;
        }
-       if (filter->action.rx_queue >= pf->dev_data->nb_rx_queues) {
-               PMD_DRV_LOG(ERR, "Invalid queue ID");
-               return -EINVAL;
+
+       i = rte_bsf64(slab);
+       pos += i;
+       rte_bitmap_clear(fdir_info->fdir_flow_pool.bitmap, pos);
+       flow = &fdir_info->fdir_flow_pool.pool[pos].flow;
+
+       memset(flow, 0, sizeof(struct rte_flow));
+
+       return flow;
+}
+
+void
+i40e_fdir_entry_pool_put(struct i40e_fdir_info *fdir_info,
+               struct rte_flow *flow)
+{
+       struct i40e_fdir_entry *f;
+
+       f = FLOW_TO_FLOW_BITMAP(flow);
+       rte_bitmap_set(fdir_info->fdir_flow_pool.bitmap, f->idx);
+}
+
+static int
+i40e_flow_store_flex_pit(struct i40e_pf *pf,
+                        struct i40e_fdir_flex_pit *flex_pit,
+                        enum i40e_flxpld_layer_idx layer_idx,
+                        uint8_t raw_id)
+{
+       uint8_t field_idx;
+
+       field_idx = layer_idx * I40E_MAX_FLXPLD_FIED + raw_id;
+       /* Check if the configuration is conflicted */
+       if (pf->fdir.flex_pit_flag[layer_idx] &&
+           (pf->fdir.flex_set[field_idx].src_offset != flex_pit->src_offset ||
+            pf->fdir.flex_set[field_idx].size != flex_pit->size ||
+            pf->fdir.flex_set[field_idx].dst_offset != flex_pit->dst_offset))
+               return -1;
+
+       /* Check if the configuration exists. */
+       if (pf->fdir.flex_pit_flag[layer_idx] &&
+           (pf->fdir.flex_set[field_idx].src_offset == flex_pit->src_offset &&
+            pf->fdir.flex_set[field_idx].size == flex_pit->size &&
+            pf->fdir.flex_set[field_idx].dst_offset == flex_pit->dst_offset))
+               return 1;
+
+       pf->fdir.flex_set[field_idx].src_offset =
+               flex_pit->src_offset;
+       pf->fdir.flex_set[field_idx].size =
+               flex_pit->size;
+       pf->fdir.flex_set[field_idx].dst_offset =
+               flex_pit->dst_offset;
+
+       return 0;
+}
+
+static void
+i40e_flow_set_fdir_flex_pit(struct i40e_pf *pf,
+                           enum i40e_flxpld_layer_idx layer_idx,
+                           uint8_t raw_id)
+{
+       struct i40e_hw *hw = I40E_PF_TO_HW(pf);
+       uint32_t flx_pit, flx_ort;
+       uint16_t min_next_off = 0;
+       uint8_t field_idx;
+       uint8_t i;
+
+       if (raw_id) {
+               flx_ort = (1 << I40E_GLQF_ORT_FLX_PAYLOAD_SHIFT) |
+                         (raw_id << I40E_GLQF_ORT_FIELD_CNT_SHIFT) |
+                         (layer_idx * I40E_MAX_FLXPLD_FIED);
+               I40E_WRITE_GLB_REG(hw, I40E_GLQF_ORT(33 + layer_idx), flx_ort);
        }
-       if (filter->input.flow_ext.is_vf &&
-               filter->input.flow_ext.dst_id >= pf->vf_num) {
-               PMD_DRV_LOG(ERR, "Invalid VF ID");
-               return -EINVAL;
+
+       /* Set flex pit */
+       for (i = 0; i < raw_id; i++) {
+               field_idx = layer_idx * I40E_MAX_FLXPLD_FIED + i;
+               flx_pit = MK_FLX_PIT(pf->fdir.flex_set[field_idx].src_offset,
+                                    pf->fdir.flex_set[field_idx].size,
+                                    pf->fdir.flex_set[field_idx].dst_offset);
+
+               I40E_WRITE_REG(hw, I40E_PRTQF_FLX_PIT(field_idx), flx_pit);
+               min_next_off = pf->fdir.flex_set[field_idx].src_offset +
+                       pf->fdir.flex_set[field_idx].size;
        }
 
-       memset(pkt, 0, I40E_FDIR_PKT_LEN);
+       for (; i < I40E_MAX_FLXPLD_FIED; i++) {
+               /* set the non-used register obeying register's constrain */
+               field_idx = layer_idx * I40E_MAX_FLXPLD_FIED + i;
+               flx_pit = MK_FLX_PIT(min_next_off, NONUSE_FLX_PIT_FSIZE,
+                                    NONUSE_FLX_PIT_DEST_OFF);
+               I40E_WRITE_REG(hw, I40E_PRTQF_FLX_PIT(field_idx), flx_pit);
+               min_next_off++;
+       }
+}
 
-       ret = i40e_fdir_construct_pkt(pf, &filter->input, pkt);
-       if (ret < 0) {
-               PMD_DRV_LOG(ERR, "construct packet for fdir fails.");
-               return ret;
+static int
+i40e_flow_store_flex_mask(struct i40e_pf *pf,
+                         enum i40e_filter_pctype pctype,
+                         uint8_t *mask)
+{
+       struct i40e_fdir_flex_mask flex_mask;
+       uint8_t nb_bitmask = 0;
+       uint16_t mask_tmp;
+       uint8_t i;
+
+       memset(&flex_mask, 0, sizeof(struct i40e_fdir_flex_mask));
+       for (i = 0; i < I40E_FDIR_MAX_FLEX_LEN; i += sizeof(uint16_t)) {
+               mask_tmp = I40E_WORD(mask[i], mask[i + 1]);
+               if (mask_tmp) {
+                       flex_mask.word_mask |=
+                               I40E_FLEX_WORD_MASK(i / sizeof(uint16_t));
+                       if (mask_tmp != UINT16_MAX) {
+                               flex_mask.bitmask[nb_bitmask].mask = ~mask_tmp;
+                               flex_mask.bitmask[nb_bitmask].offset =
+                                       i / sizeof(uint16_t);
+                               nb_bitmask++;
+                               if (nb_bitmask > I40E_FDIR_BITMASK_NUM_WORD)
+                                       return -1;
+                       }
+               }
        }
+       flex_mask.nb_bitmask = nb_bitmask;
 
-       if (hw->mac.type == I40E_MAC_X722) {
-               /* get translated pctype value in fd pctype register */
-               pctype = (enum i40e_filter_pctype)i40e_read_rx_ctl(
-                       hw, I40E_GLQF_FD_PCTYPES((int)pctype));
+       if (pf->fdir.flex_mask_flag[pctype] &&
+           (memcmp(&flex_mask, &pf->fdir.flex_mask[pctype],
+                   sizeof(struct i40e_fdir_flex_mask))))
+               return -2;
+       else if (pf->fdir.flex_mask_flag[pctype] &&
+                !(memcmp(&flex_mask, &pf->fdir.flex_mask[pctype],
+                         sizeof(struct i40e_fdir_flex_mask))))
+               return 1;
+
+       memcpy(&pf->fdir.flex_mask[pctype], &flex_mask,
+              sizeof(struct i40e_fdir_flex_mask));
+       return 0;
+}
+
+static void
+i40e_flow_set_fdir_flex_msk(struct i40e_pf *pf,
+                           enum i40e_filter_pctype pctype)
+{
+       struct i40e_hw *hw = I40E_PF_TO_HW(pf);
+       struct i40e_fdir_flex_mask *flex_mask;
+       uint32_t flxinset, fd_mask;
+       uint8_t i;
+
+       /* Set flex mask */
+       flex_mask = &pf->fdir.flex_mask[pctype];
+       flxinset = (flex_mask->word_mask <<
+                   I40E_PRTQF_FD_FLXINSET_INSET_SHIFT) &
+               I40E_PRTQF_FD_FLXINSET_INSET_MASK;
+       i40e_write_rx_ctl(hw, I40E_PRTQF_FD_FLXINSET(pctype), flxinset);
+
+       for (i = 0; i < flex_mask->nb_bitmask; i++) {
+               fd_mask = (flex_mask->bitmask[i].mask <<
+                          I40E_PRTQF_FD_MSK_MASK_SHIFT) &
+                          I40E_PRTQF_FD_MSK_MASK_MASK;
+               fd_mask |= ((flex_mask->bitmask[i].offset +
+                            I40E_FLX_OFFSET_IN_FIELD_VECTOR) <<
+                           I40E_PRTQF_FD_MSK_OFFSET_SHIFT) &
+                               I40E_PRTQF_FD_MSK_OFFSET_MASK;
+               i40e_write_rx_ctl(hw, I40E_PRTQF_FD_MSK(pctype, i), fd_mask);
        }
 
-       ret = i40e_fdir_filter_programming(pf, pctype, filter, add);
-       if (ret < 0) {
-               PMD_DRV_LOG(ERR, "fdir programming fails for PCTYPE(%u).",
-                           pctype);
-               return ret;
+       pf->fdir.flex_mask_flag[pctype] = 1;
+}
+
+static int
+i40e_flow_set_fdir_inset(struct i40e_pf *pf,
+                        enum i40e_filter_pctype pctype,
+                        uint64_t input_set)
+{
+       uint32_t mask_reg[I40E_INSET_MASK_NUM_REG] = {0};
+       struct i40e_hw *hw = I40E_PF_TO_HW(pf);
+       uint64_t inset_reg = 0;
+       int i, num;
+
+       /* Check if the input set is valid */
+       if (i40e_validate_input_set(pctype, RTE_ETH_FILTER_FDIR,
+                                   input_set) != 0) {
+               PMD_DRV_LOG(ERR, "Invalid input set");
+               return -EINVAL;
        }
 
-       return ret;
+       /* Check if the configuration is conflicted */
+       if (pf->fdir.inset_flag[pctype] &&
+           memcmp(&pf->fdir.input_set[pctype], &input_set, sizeof(uint64_t)))
+               return -1;
+
+       if (pf->fdir.inset_flag[pctype] &&
+           !memcmp(&pf->fdir.input_set[pctype], &input_set, sizeof(uint64_t)))
+               return 0;
+
+       num = i40e_generate_inset_mask_reg(hw, input_set, mask_reg,
+                                                I40E_INSET_MASK_NUM_REG);
+       if (num < 0)
+               return -EINVAL;
+
+       if (pf->support_multi_driver) {
+               for (i = 0; i < num; i++)
+                       if (i40e_read_rx_ctl(hw,
+                                       I40E_GLQF_FD_MSK(i, pctype)) !=
+                                       mask_reg[i]) {
+                               PMD_DRV_LOG(ERR, "Input set setting is not"
+                                               " supported with"
+                                               " `support-multi-driver`"
+                                               " enabled!");
+                               return -EPERM;
+                       }
+               for (i = num; i < I40E_INSET_MASK_NUM_REG; i++)
+                       if (i40e_read_rx_ctl(hw,
+                                       I40E_GLQF_FD_MSK(i, pctype)) != 0) {
+                               PMD_DRV_LOG(ERR, "Input set setting is not"
+                                               " supported with"
+                                               " `support-multi-driver`"
+                                               " enabled!");
+                               return -EPERM;
+                       }
+
+       } else {
+               for (i = 0; i < num; i++)
+                       i40e_check_write_reg(hw, I40E_GLQF_FD_MSK(i, pctype),
+                               mask_reg[i]);
+               /*clear unused mask registers of the pctype */
+               for (i = num; i < I40E_INSET_MASK_NUM_REG; i++)
+                       i40e_check_write_reg(hw,
+                                       I40E_GLQF_FD_MSK(i, pctype), 0);
+       }
+
+       inset_reg |= i40e_translate_input_set_reg(hw->mac.type, input_set);
+
+       i40e_check_write_reg(hw, I40E_PRTQF_FD_INSET(pctype, 0),
+                            (uint32_t)(inset_reg & UINT32_MAX));
+       i40e_check_write_reg(hw, I40E_PRTQF_FD_INSET(pctype, 1),
+                            (uint32_t)((inset_reg >>
+                                        I40E_32_BIT_WIDTH) & UINT32_MAX));
+
+       I40E_WRITE_FLUSH(hw);
+
+       pf->fdir.input_set[pctype] = input_set;
+       pf->fdir.inset_flag[pctype] = 1;
+       return 0;
+}
+
+static inline unsigned char *
+i40e_find_available_buffer(struct rte_eth_dev *dev)
+{
+       struct i40e_pf *pf = I40E_DEV_PRIVATE_TO_PF(dev->data->dev_private);
+       struct i40e_fdir_info *fdir_info = &pf->fdir;
+       struct i40e_tx_queue *txq = pf->fdir.txq;
+
+       /* no available buffer
+        * search for more available buffers from the current
+        * descriptor, until an unavailable one
+        */
+       if (fdir_info->txq_available_buf_count <= 0) {
+               uint16_t tmp_tail;
+               volatile struct i40e_tx_desc *tmp_txdp;
+
+               tmp_tail = txq->tx_tail;
+               tmp_txdp = &txq->tx_ring[tmp_tail + 1];
+
+               do {
+                       if ((tmp_txdp->cmd_type_offset_bsz &
+                                       rte_cpu_to_le_64(I40E_TXD_QW1_DTYPE_MASK)) ==
+                                       rte_cpu_to_le_64(I40E_TX_DESC_DTYPE_DESC_DONE))
+                               fdir_info->txq_available_buf_count++;
+                       else
+                               break;
+
+                       tmp_tail += 2;
+                       if (tmp_tail >= txq->nb_tx_desc)
+                               tmp_tail = 0;
+               } while (tmp_tail != txq->tx_tail);
+       }
+
+       if (fdir_info->txq_available_buf_count > 0)
+               fdir_info->txq_available_buf_count--;
+       else
+               return NULL;
+       return (unsigned char *)fdir_info->prg_pkt[txq->tx_tail >> 1];
 }
 
 /**
@@ -1688,12 +1717,19 @@ i40e_flow_add_del_fdir_filter(struct rte_eth_dev *dev,
 {
        struct i40e_hw *hw = I40E_DEV_PRIVATE_TO_HW(dev->data->dev_private);
        struct i40e_pf *pf = I40E_DEV_PRIVATE_TO_PF(dev->data->dev_private);
-       unsigned char *pkt = (unsigned char *)pf->fdir.prg_pkt;
-       enum i40e_filter_pctype pctype;
+       enum i40e_flxpld_layer_idx layer_idx = I40E_FLXPLD_L2_IDX;
        struct i40e_fdir_info *fdir_info = &pf->fdir;
-       struct i40e_fdir_filter *fdir_filter, *node;
+       uint8_t flex_mask[I40E_FDIR_MAX_FLEX_LEN];
        struct i40e_fdir_filter check_filter; /* Check if the filter exists */
+       struct i40e_fdir_flex_pit flex_pit;
+       enum i40e_filter_pctype pctype;
+       struct i40e_fdir_filter *node;
+       unsigned char *pkt = NULL;
+       bool cfg_flex_pit = true;
+       bool wait_status = true;
+       uint8_t field_idx;
        int ret = 0;
+       int i;
 
        if (pf->fdir.fdir_vsi == NULL) {
                PMD_DRV_LOG(ERR, "FDIR is not enabled");
@@ -1724,25 +1760,105 @@ i40e_flow_add_del_fdir_filter(struct rte_eth_dev *dev,
        /* Check if there is the filter in SW list */
        memset(&check_filter, 0, sizeof(check_filter));
        i40e_fdir_filter_convert(filter, &check_filter);
-       node = i40e_sw_fdir_filter_lookup(fdir_info, &check_filter.fdir.input);
-       if (add && node) {
-               PMD_DRV_LOG(ERR,
-                           "Conflict with existing flow director rules!");
-               return -EINVAL;
-       }
 
-       if (!add && !node) {
-               PMD_DRV_LOG(ERR,
-                           "There's no corresponding flow firector filter!");
-               return -EINVAL;
+       if (add) {
+               if (filter->input.flow_ext.is_flex_flow) {
+                       ret = i40e_flow_set_fdir_inset(pf, pctype,
+                                       filter->input.flow_ext.input_set);
+                       if (ret == -1) {
+                               PMD_DRV_LOG(ERR, "Conflict with the"
+                                           " first rule's input set.");
+                               return -EINVAL;
+                       } else if (ret == -EINVAL) {
+                               PMD_DRV_LOG(ERR, "Invalid pattern mask.");
+                               return -EINVAL;
+                       }
+
+                       for (i = 0; i < filter->input.flow_ext.raw_id; i++) {
+                               layer_idx = filter->input.flow_ext.layer_idx;
+                               field_idx = layer_idx * I40E_MAX_FLXPLD_FIED + i;
+                               flex_pit = filter->input.flow_ext.flex_pit[field_idx];
+
+                               /* Store flex pit to SW */
+                               ret = i40e_flow_store_flex_pit(pf, &flex_pit,
+                                                              layer_idx, i);
+                               if (ret < 0) {
+                                       PMD_DRV_LOG(ERR, "Conflict with the"
+                                                   " first flexible rule.");
+                                       return -EINVAL;
+                               } else if (ret > 0) {
+                                       cfg_flex_pit = false;
+                               }
+                       }
+
+                       if (cfg_flex_pit)
+                               i40e_flow_set_fdir_flex_pit(pf, layer_idx,
+                                               filter->input.flow_ext.raw_id);
+
+                       /* Store flex mask to SW */
+                       for (i = 0; i < I40E_FDIR_MAX_FLEX_LEN; i++)
+                               flex_mask[i] =
+                                       filter->input.flow_ext.flex_mask[i];
+
+                       ret = i40e_flow_store_flex_mask(pf, pctype, flex_mask);
+                       if (ret == -1) {
+                               PMD_DRV_LOG(ERR, "Exceed maximal"
+                                           " number of bitmasks");
+                               return -EINVAL;
+                       } else if (ret == -2) {
+                               PMD_DRV_LOG(ERR, "Conflict with the"
+                                           " first flexible rule");
+                               return -EINVAL;
+                       } else if (ret == 0) {
+                               i40e_flow_set_fdir_flex_msk(pf, pctype);
+                       }
+               }
+
+               ret = i40e_sw_fdir_filter_insert(pf, &check_filter);
+               if (ret < 0) {
+                       PMD_DRV_LOG(ERR,
+                                   "Conflict with existing flow director rules!");
+                       return -EINVAL;
+               }
+
+               if (fdir_info->fdir_invalprio == 1 &&
+                               fdir_info->fdir_guarantee_free_space > 0)
+                       wait_status = false;
+       } else {
+               if (filter->input.flow_ext.is_flex_flow)
+                       layer_idx = filter->input.flow_ext.layer_idx;
+
+               node = i40e_sw_fdir_filter_lookup(fdir_info,
+                               &check_filter.fdir.input);
+               if (!node) {
+                       PMD_DRV_LOG(ERR,
+                                   "There's no corresponding flow firector filter!");
+                       return -EINVAL;
+               }
+
+               ret = i40e_sw_fdir_filter_del(pf, &node->fdir.input);
+               if (ret < 0) {
+                       PMD_DRV_LOG(ERR,
+                                       "Error deleting fdir rule from hash table!");
+                       return -EINVAL;
+               }
+
+               pf->fdir.flex_mask_flag[pctype] = 0;
+
+               if (fdir_info->fdir_invalprio == 1)
+                       wait_status = false;
        }
 
-       memset(pkt, 0, I40E_FDIR_PKT_LEN);
+       /* find a buffer to store the pkt */
+       pkt = i40e_find_available_buffer(dev);
+       if (pkt == NULL)
+               goto error_op;
 
+       memset(pkt, 0, I40E_FDIR_PKT_LEN);
        ret = i40e_flow_fdir_construct_pkt(pf, &filter->input, pkt);
        if (ret < 0) {
                PMD_DRV_LOG(ERR, "construct packet for fdir fails.");
-               return ret;
+               goto error_op;
        }
 
        if (hw->mac.type == I40E_MAC_X722) {
@@ -1751,165 +1867,48 @@ i40e_flow_add_del_fdir_filter(struct rte_eth_dev *dev,
                        hw, I40E_GLQF_FD_PCTYPES((int)pctype));
        }
 
-       ret = i40e_flow_fdir_filter_programming(pf, pctype, filter, add);
+       ret = i40e_flow_fdir_filter_programming(pf, pctype, filter, add,
+                       wait_status);
        if (ret < 0) {
                PMD_DRV_LOG(ERR, "fdir programming fails for PCTYPE(%u).",
                            pctype);
-               return ret;
+               goto error_op;
        }
 
-       if (add) {
-               fdir_filter = rte_zmalloc("fdir_filter",
-                                         sizeof(*fdir_filter), 0);
-               if (fdir_filter == NULL) {
-                       PMD_DRV_LOG(ERR, "Failed to alloc memory.");
-                       return -ENOMEM;
+       if (filter->input.flow_ext.is_flex_flow) {
+               if (add) {
+                       fdir_info->flex_flow_count[layer_idx]++;
+                       pf->fdir.flex_pit_flag[layer_idx] = 1;
+               } else {
+                       fdir_info->flex_flow_count[layer_idx]--;
+                       if (!fdir_info->flex_flow_count[layer_idx])
+                               pf->fdir.flex_pit_flag[layer_idx] = 0;
                }
+       }
 
-               rte_memcpy(fdir_filter, &check_filter, sizeof(check_filter));
-               ret = i40e_sw_fdir_filter_insert(pf, fdir_filter);
-               if (ret < 0)
-                       rte_free(fdir_filter);
+       if (add) {
+               fdir_info->fdir_actual_cnt++;
+               if (fdir_info->fdir_invalprio == 1 &&
+                               fdir_info->fdir_guarantee_free_space > 0)
+                       fdir_info->fdir_guarantee_free_space--;
        } else {
-               ret = i40e_sw_fdir_filter_del(pf, &node->fdir.input);
+               fdir_info->fdir_actual_cnt--;
+               if (fdir_info->fdir_invalprio == 1 &&
+                               fdir_info->fdir_guarantee_free_space <
+                               fdir_info->fdir_guarantee_total_space)
+                       fdir_info->fdir_guarantee_free_space++;
        }
 
        return ret;
-}
-
-/*
- * i40e_fdir_filter_programming - Program a flow director filter rule.
- * Is done by Flow Director Programming Descriptor followed by packet
- * structure that contains the filter fields need to match.
- * @pf: board private structure
- * @pctype: pctype
- * @filter: fdir filter entry
- * @add: 0 - delete, 1 - add
- */
-static int
-i40e_fdir_filter_programming(struct i40e_pf *pf,
-                       enum i40e_filter_pctype pctype,
-                       const struct rte_eth_fdir_filter *filter,
-                       bool add)
-{
-       struct i40e_tx_queue *txq = pf->fdir.txq;
-       struct i40e_rx_queue *rxq = pf->fdir.rxq;
-       const struct rte_eth_fdir_action *fdir_action = &filter->action;
-       volatile struct i40e_tx_desc *txdp;
-       volatile struct i40e_filter_program_desc *fdirdp;
-       uint32_t td_cmd;
-       uint16_t vsi_id, i;
-       uint8_t dest;
-
-       PMD_DRV_LOG(INFO, "filling filter programming descriptor.");
-       fdirdp = (volatile struct i40e_filter_program_desc *)
-                       (&(txq->tx_ring[txq->tx_tail]));
-
-       fdirdp->qindex_flex_ptype_vsi =
-                       rte_cpu_to_le_32((fdir_action->rx_queue <<
-                                         I40E_TXD_FLTR_QW0_QINDEX_SHIFT) &
-                                         I40E_TXD_FLTR_QW0_QINDEX_MASK);
-
-       fdirdp->qindex_flex_ptype_vsi |=
-                       rte_cpu_to_le_32((fdir_action->flex_off <<
-                                         I40E_TXD_FLTR_QW0_FLEXOFF_SHIFT) &
-                                         I40E_TXD_FLTR_QW0_FLEXOFF_MASK);
-
-       fdirdp->qindex_flex_ptype_vsi |=
-                       rte_cpu_to_le_32((pctype <<
-                                         I40E_TXD_FLTR_QW0_PCTYPE_SHIFT) &
-                                         I40E_TXD_FLTR_QW0_PCTYPE_MASK);
-
-       if (filter->input.flow_ext.is_vf)
-               vsi_id = pf->vfs[filter->input.flow_ext.dst_id].vsi->vsi_id;
-       else
-               /* Use LAN VSI Id by default */
-               vsi_id = pf->main_vsi->vsi_id;
-       fdirdp->qindex_flex_ptype_vsi |=
-               rte_cpu_to_le_32(((uint32_t)vsi_id <<
-                                 I40E_TXD_FLTR_QW0_DEST_VSI_SHIFT) &
-                                 I40E_TXD_FLTR_QW0_DEST_VSI_MASK);
-
-       fdirdp->dtype_cmd_cntindex =
-                       rte_cpu_to_le_32(I40E_TX_DESC_DTYPE_FILTER_PROG);
 
+error_op:
+       /* roll back */
        if (add)
-               fdirdp->dtype_cmd_cntindex |= rte_cpu_to_le_32(
-                               I40E_FILTER_PROGRAM_DESC_PCMD_ADD_UPDATE <<
-                               I40E_TXD_FLTR_QW1_PCMD_SHIFT);
+               i40e_sw_fdir_filter_del(pf, &check_filter.fdir.input);
        else
-               fdirdp->dtype_cmd_cntindex |= rte_cpu_to_le_32(
-                               I40E_FILTER_PROGRAM_DESC_PCMD_REMOVE <<
-                               I40E_TXD_FLTR_QW1_PCMD_SHIFT);
+               i40e_sw_fdir_filter_insert(pf, &check_filter);
 
-       if (fdir_action->behavior == RTE_ETH_FDIR_REJECT)
-               dest = I40E_FILTER_PROGRAM_DESC_DEST_DROP_PACKET;
-       else if (fdir_action->behavior == RTE_ETH_FDIR_ACCEPT)
-               dest = I40E_FILTER_PROGRAM_DESC_DEST_DIRECT_PACKET_QINDEX;
-       else if (fdir_action->behavior == RTE_ETH_FDIR_PASSTHRU)
-               dest = I40E_FILTER_PROGRAM_DESC_DEST_DIRECT_PACKET_OTHER;
-       else {
-               PMD_DRV_LOG(ERR, "Failed to program FDIR filter:"
-                           " unsupported fdir behavior.");
-               return -EINVAL;
-       }
-
-       fdirdp->dtype_cmd_cntindex |= rte_cpu_to_le_32((dest <<
-                               I40E_TXD_FLTR_QW1_DEST_SHIFT) &
-                               I40E_TXD_FLTR_QW1_DEST_MASK);
-
-       fdirdp->dtype_cmd_cntindex |=
-               rte_cpu_to_le_32((fdir_action->report_status<<
-                               I40E_TXD_FLTR_QW1_FD_STATUS_SHIFT) &
-                               I40E_TXD_FLTR_QW1_FD_STATUS_MASK);
-
-       fdirdp->dtype_cmd_cntindex |=
-                       rte_cpu_to_le_32(I40E_TXD_FLTR_QW1_CNT_ENA_MASK);
-       fdirdp->dtype_cmd_cntindex |=
-                       rte_cpu_to_le_32(
-                       ((uint32_t)pf->fdir.match_counter_index <<
-                       I40E_TXD_FLTR_QW1_CNTINDEX_SHIFT) &
-                       I40E_TXD_FLTR_QW1_CNTINDEX_MASK);
-
-       fdirdp->fd_id = rte_cpu_to_le_32(filter->soft_id);
-
-       PMD_DRV_LOG(INFO, "filling transmit descriptor.");
-       txdp = &(txq->tx_ring[txq->tx_tail + 1]);
-       txdp->buffer_addr = rte_cpu_to_le_64(pf->fdir.dma_addr);
-       td_cmd = I40E_TX_DESC_CMD_EOP |
-                I40E_TX_DESC_CMD_RS  |
-                I40E_TX_DESC_CMD_DUMMY;
-
-       txdp->cmd_type_offset_bsz =
-               i40e_build_ctob(td_cmd, 0, I40E_FDIR_PKT_LEN, 0);
-
-       txq->tx_tail += 2; /* set 2 descriptors above, fdirdp and txdp */
-       if (txq->tx_tail >= txq->nb_tx_desc)
-               txq->tx_tail = 0;
-       /* Update the tx tail register */
-       rte_wmb();
-       I40E_PCI_REG_WRITE(txq->qtx_tail, txq->tx_tail);
-       for (i = 0; i < I40E_FDIR_MAX_WAIT_US; i++) {
-               if ((txdp->cmd_type_offset_bsz &
-                               rte_cpu_to_le_64(I40E_TXD_QW1_DTYPE_MASK)) ==
-                               rte_cpu_to_le_64(I40E_TX_DESC_DTYPE_DESC_DONE))
-                       break;
-               rte_delay_us(1);
-       }
-       if (i >= I40E_FDIR_MAX_WAIT_US) {
-               PMD_DRV_LOG(ERR, "Failed to program FDIR filter:"
-                           " time out to get DD on tx queue.");
-               return -ETIMEDOUT;
-       }
-       /* totally delay 10 ms to check programming status*/
-       for (; i < I40E_FDIR_MAX_WAIT_US; i++) {
-               if (i40e_check_fdir_programming_status(rxq) >= 0)
-                       return 0;
-               rte_delay_us(1);
-       }
-       PMD_DRV_LOG(ERR,
-               "Failed to program FDIR filter: programming status reported.");
-       return -ETIMEDOUT;
+       return ret;
 }
 
 /*
@@ -1925,7 +1924,7 @@ static int
 i40e_flow_fdir_filter_programming(struct i40e_pf *pf,
                                  enum i40e_filter_pctype pctype,
                                  const struct i40e_fdir_filter_conf *filter,
-                                 bool add)
+                                 bool add, bool wait_status)
 {
        struct i40e_tx_queue *txq = pf->fdir.txq;
        struct i40e_rx_queue *rxq = pf->fdir.rxq;
@@ -1933,8 +1932,9 @@ i40e_flow_fdir_filter_programming(struct i40e_pf *pf,
        volatile struct i40e_tx_desc *txdp;
        volatile struct i40e_filter_program_desc *fdirdp;
        uint32_t td_cmd;
-       uint16_t vsi_id, i;
+       uint16_t vsi_id;
        uint8_t dest;
+       uint32_t i;
 
        PMD_DRV_LOG(INFO, "filling filter programming descriptor.");
        fdirdp = (volatile struct i40e_filter_program_desc *)
@@ -2009,7 +2009,8 @@ i40e_flow_fdir_filter_programming(struct i40e_pf *pf,
 
        PMD_DRV_LOG(INFO, "filling transmit descriptor.");
        txdp = &txq->tx_ring[txq->tx_tail + 1];
-       txdp->buffer_addr = rte_cpu_to_le_64(pf->fdir.dma_addr);
+       txdp->buffer_addr = rte_cpu_to_le_64(pf->fdir.dma_addr[txq->tx_tail >> 1]);
+
        td_cmd = I40E_TX_DESC_CMD_EOP |
                 I40E_TX_DESC_CMD_RS  |
                 I40E_TX_DESC_CMD_DUMMY;
@@ -2022,25 +2023,32 @@ i40e_flow_fdir_filter_programming(struct i40e_pf *pf,
                txq->tx_tail = 0;
        /* Update the tx tail register */
        rte_wmb();
+
+       /* fdir program rx queue cleanup */
+       i40e_fdir_programming_status_cleanup(rxq);
+
        I40E_PCI_REG_WRITE(txq->qtx_tail, txq->tx_tail);
-       for (i = 0; i < I40E_FDIR_MAX_WAIT_US; i++) {
-               if ((txdp->cmd_type_offset_bsz &
-                               rte_cpu_to_le_64(I40E_TXD_QW1_DTYPE_MASK)) ==
-                               rte_cpu_to_le_64(I40E_TX_DESC_DTYPE_DESC_DONE))
-                       break;
-               rte_delay_us(1);
-       }
-       if (i >= I40E_FDIR_MAX_WAIT_US) {
-               PMD_DRV_LOG(ERR,
-                   "Failed to program FDIR filter: time out to get DD on tx queue.");
-               return -ETIMEDOUT;
-       }
-       /* totally delay 10 ms to check programming status*/
-       rte_delay_us(I40E_FDIR_MAX_WAIT_US);
-       if (i40e_check_fdir_programming_status(rxq) < 0) {
-               PMD_DRV_LOG(ERR,
-                   "Failed to program FDIR filter: programming status reported.");
-               return -ETIMEDOUT;
+
+       if (wait_status) {
+               for (i = 0; i < I40E_FDIR_MAX_WAIT_US; i++) {
+                       if ((txdp->cmd_type_offset_bsz &
+                                       rte_cpu_to_le_64(I40E_TXD_QW1_DTYPE_MASK)) ==
+                                       rte_cpu_to_le_64(I40E_TX_DESC_DTYPE_DESC_DONE))
+                               break;
+                       rte_delay_us(1);
+               }
+               if (i >= I40E_FDIR_MAX_WAIT_US) {
+                       PMD_DRV_LOG(ERR,
+                           "Failed to program FDIR filter: time out to get DD on tx queue.");
+                       return -ETIMEDOUT;
+               }
+               /* totally delay 10 ms to check programming status*/
+               rte_delay_us(I40E_FDIR_MAX_WAIT_US);
+               if (i40e_check_fdir_programming_status(rxq) < 0) {
+                       PMD_DRV_LOG(ERR,
+                           "Failed to program FDIR filter: programming status reported.");
+                       return -ETIMEDOUT;
+               }
        }
 
        return 0;
@@ -2169,7 +2177,7 @@ i40e_fdir_info_get_flex_mask(struct i40e_pf *pf,
  * @fdir: a pointer to a structure of type *rte_eth_fdir_info* to be filled with
  *    the flow director information.
  */
-static void
+void
 i40e_fdir_info_get(struct rte_eth_dev *dev, struct rte_eth_fdir_info *fdir)
 {
        struct i40e_pf *pf = I40E_DEV_PRIVATE_TO_PF(dev->data->dev_private);
@@ -2214,7 +2222,7 @@ i40e_fdir_info_get(struct rte_eth_dev *dev, struct rte_eth_fdir_info *fdir)
  * @stat: a pointer to a structure of type *rte_eth_fdir_stats* to be filled with
  *    the flow director statistics.
  */
-static void
+void
 i40e_fdir_stats_get(struct rte_eth_dev *dev, struct rte_eth_fdir_stats *stat)
 {
        struct i40e_pf *pf = I40E_DEV_PRIVATE_TO_PF(dev->data->dev_private);
@@ -2230,87 +2238,6 @@ i40e_fdir_stats_get(struct rte_eth_dev *dev, struct rte_eth_fdir_stats *stat)
                            I40E_PFQF_FDSTAT_BEST_CNT_SHIFT);
 }
 
-static int
-i40e_fdir_filter_set(struct rte_eth_dev *dev,
-                    struct rte_eth_fdir_filter_info *info)
-{
-       struct i40e_pf *pf = I40E_DEV_PRIVATE_TO_PF(dev->data->dev_private);
-       int ret = 0;
-
-       if (!info) {
-               PMD_DRV_LOG(ERR, "Invalid pointer");
-               return -EFAULT;
-       }
-
-       switch (info->info_type) {
-       case RTE_ETH_FDIR_FILTER_INPUT_SET_SELECT:
-               ret = i40e_fdir_filter_inset_select(pf,
-                               &(info->info.input_set_conf));
-               break;
-       default:
-               PMD_DRV_LOG(ERR, "FD filter info type (%d) not supported",
-                           info->info_type);
-               return -EINVAL;
-       }
-
-       return ret;
-}
-
-/*
- * i40e_fdir_ctrl_func - deal with all operations on flow director.
- * @pf: board private structure
- * @filter_op:operation will be taken.
- * @arg: a pointer to specific structure corresponding to the filter_op
- */
-int
-i40e_fdir_ctrl_func(struct rte_eth_dev *dev,
-                      enum rte_filter_op filter_op,
-                      void *arg)
-{
-       struct i40e_pf *pf = I40E_DEV_PRIVATE_TO_PF(dev->data->dev_private);
-       int ret = 0;
-
-       if ((pf->flags & I40E_FLAG_FDIR) == 0)
-               return -ENOTSUP;
-
-       if (filter_op == RTE_ETH_FILTER_NOP)
-               return 0;
-
-       if (arg == NULL && filter_op != RTE_ETH_FILTER_FLUSH)
-               return -EINVAL;
-
-       switch (filter_op) {
-       case RTE_ETH_FILTER_ADD:
-               ret = i40e_add_del_fdir_filter(dev,
-                       (struct rte_eth_fdir_filter *)arg,
-                       TRUE);
-               break;
-       case RTE_ETH_FILTER_DELETE:
-               ret = i40e_add_del_fdir_filter(dev,
-                       (struct rte_eth_fdir_filter *)arg,
-                       FALSE);
-               break;
-       case RTE_ETH_FILTER_FLUSH:
-               ret = i40e_fdir_flush(dev);
-               break;
-       case RTE_ETH_FILTER_INFO:
-               i40e_fdir_info_get(dev, (struct rte_eth_fdir_info *)arg);
-               break;
-       case RTE_ETH_FILTER_SET:
-               ret = i40e_fdir_filter_set(dev,
-                       (struct rte_eth_fdir_filter_info *)arg);
-               break;
-       case RTE_ETH_FILTER_STATS:
-               i40e_fdir_stats_get(dev, (struct rte_eth_fdir_stats *)arg);
-               break;
-       default:
-               PMD_DRV_LOG(ERR, "unknown operation %u.", filter_op);
-               ret = -EINVAL;
-               break;
-       }
-       return ret;
-}
-
 /* Restore flow director filter */
 void
 i40e_fdir_filter_restore(struct i40e_pf *pf)