net/hinic: add flow validation operations
authorXiaoyun Wang <cloud.wangxiaoyun@huawei.com>
Thu, 10 Oct 2019 14:51:52 +0000 (22:51 +0800)
committerFerruh Yigit <ferruh.yigit@intel.com>
Wed, 23 Oct 2019 14:43:08 +0000 (16:43 +0200)
This patch is to validate the filter rules, which includes
ntuple filter, ethertype filter and fdir filter. The packets
type that supported are BGP,VRRP,LACP,ARP and ICMP.

Signed-off-by: Xiaoyun Wang <cloud.wangxiaoyun@huawei.com>
doc/guides/rel_notes/release_19_11.rst
drivers/net/hinic/Makefile
drivers/net/hinic/hinic_pmd_ethdev.c
drivers/net/hinic/hinic_pmd_ethdev.h
drivers/net/hinic/hinic_pmd_flow.c [new file with mode: 0644]
drivers/net/hinic/meson.build

index db0aceb..60e098a 100644 (file)
@@ -118,6 +118,7 @@ New Features
   * Enabled SR-IOV - Partially supported at this point, VFIO only.
   * Supported VLAN filter and VLAN offload.
   * Supported Unicast MAC filter and Multicast MAC filter.
+  * Supported Flow director for LACP, VRRP, BGP and so on.
 
 * **Added Marvell NITROX symmetric crypto PMD.**
 
index 20a338e..b78fd8d 100644 (file)
@@ -60,6 +60,7 @@ SRCS-$(CONFIG_RTE_LIBRTE_HINIC_PMD) += hinic_pmd_niccfg.c
 SRCS-$(CONFIG_RTE_LIBRTE_HINIC_PMD) += hinic_pmd_nicio.c
 SRCS-$(CONFIG_RTE_LIBRTE_HINIC_PMD) += hinic_pmd_wq.c
 SRCS-$(CONFIG_RTE_LIBRTE_HINIC_PMD) += hinic_pmd_mbox.c
+SRCS-$(CONFIG_RTE_LIBRTE_HINIC_PMD) += hinic_pmd_flow.c
 
 SRCS-$(CONFIG_RTE_LIBRTE_HINIC_PMD) += hinic_pmd_ethdev.c
 SRCS-$(CONFIG_RTE_LIBRTE_HINIC_PMD) += hinic_pmd_rx.c
index 91b4f98..f2f0fa9 100644 (file)
@@ -2285,6 +2285,46 @@ allmulti:
        return 0;
 }
 
+/**
+ * DPDK callback to manage filter operations
+ *
+ * @param dev
+ *   Pointer to Ethernet device structure.
+ * @param filter_type
+ *   Filter type.
+ * @param filter_op
+ *   Operation to perform.
+ * @param arg
+ *   Pointer to operation-specific structure.
+ *
+ * @return
+ *   0 on success, negative errno value on failure.
+ */
+static int hinic_dev_filter_ctrl(struct rte_eth_dev *dev,
+                    enum rte_filter_type filter_type,
+                    enum rte_filter_op filter_op,
+                    void *arg)
+{
+       struct hinic_nic_dev *nic_dev = HINIC_ETH_DEV_TO_PRIVATE_NIC_DEV(dev);
+       int func_id = hinic_global_func_id(nic_dev->hwdev);
+
+       switch (filter_type) {
+       case RTE_ETH_FILTER_GENERIC:
+               if (filter_op != RTE_ETH_FILTER_GET)
+                       return -EINVAL;
+               *(const void **)arg = &hinic_flow_ops;
+               break;
+       default:
+               PMD_DRV_LOG(INFO, "Filter type (%d) not supported",
+                       filter_type);
+               return -EINVAL;
+       }
+
+       PMD_DRV_LOG(INFO, "Set filter_ctrl succeed, func_id: 0x%x, filter_type: 0x%x,"
+                       "filter_op: 0x%x.", func_id, filter_type, filter_op);
+       return 0;
+}
+
 static int hinic_set_default_pause_feature(struct hinic_nic_dev *nic_dev)
 {
        struct nic_pause_config pause_config = {0};
@@ -2736,6 +2776,7 @@ static const struct eth_dev_ops hinic_pmd_ops = {
        .mac_addr_remove               = hinic_mac_addr_remove,
        .mac_addr_add                  = hinic_mac_addr_add,
        .set_mc_addr_list              = hinic_set_mc_addr_list,
+       .filter_ctrl                   = hinic_dev_filter_ctrl,
 };
 
 static const struct eth_dev_ops hinic_pmd_vf_ops = {
@@ -2767,6 +2808,7 @@ static const struct eth_dev_ops hinic_pmd_vf_ops = {
        .mac_addr_remove               = hinic_mac_addr_remove,
        .mac_addr_add                  = hinic_mac_addr_add,
        .set_mc_addr_list              = hinic_set_mc_addr_list,
+       .filter_ctrl                   = hinic_dev_filter_ctrl,
 };
 
 static int hinic_func_init(struct rte_eth_dev *eth_dev)
index b4f93ad..58a38d9 100644 (file)
@@ -38,6 +38,30 @@ enum hinic_dev_status {
        HINIC_DEV_INTR_EN,
 };
 
+/* Information about the fdir mode. */
+struct hinic_hw_fdir_mask {
+       uint32_t src_ipv4_mask;
+       uint32_t dst_ipv4_mask;
+       uint16_t src_port_mask;
+       uint16_t dst_port_mask;
+};
+
+/* Flow Director attribute */
+struct hinic_atr_input {
+       u32 dst_ip;
+       u32 src_ip;
+       u16 src_port;
+       u16 dst_port;
+};
+
+struct hinic_fdir_rule {
+       struct hinic_hw_fdir_mask mask;
+       struct hinic_atr_input hinic_fdir; /* key of fdir filter */
+       uint8_t queue; /* queue assigned when matched */
+};
+
+extern const struct rte_flow_ops hinic_flow_ops;
+
 /* hinic nic_device */
 struct hinic_nic_dev {
        /* hardware device */
diff --git a/drivers/net/hinic/hinic_pmd_flow.c b/drivers/net/hinic/hinic_pmd_flow.c
new file mode 100644 (file)
index 0000000..cf9f105
--- /dev/null
@@ -0,0 +1,1172 @@
+/* SPDX-License-Identifier: BSD-3-Clause
+ * Copyright(c) 2017 Huawei Technologies Co., Ltd
+ */
+
+#include <stdio.h>
+#include <errno.h>
+#include <stdint.h>
+#include <string.h>
+#include <unistd.h>
+
+#include <rte_byteorder.h>
+#include <rte_common.h>
+#include <rte_ether.h>
+#include <rte_ethdev.h>
+#include <rte_malloc.h>
+#include <rte_flow.h>
+#include <rte_flow_driver.h>
+#include "base/hinic_compat.h"
+#include "base/hinic_pmd_hwdev.h"
+#include "base/hinic_pmd_hwif.h"
+#include "base/hinic_pmd_wq.h"
+#include "base/hinic_pmd_cmdq.h"
+#include "base/hinic_pmd_niccfg.h"
+#include "hinic_pmd_ethdev.h"
+
+#ifndef UINT8_MAX
+#define UINT8_MAX          (u8)(~((u8)0))      /* 0xFF               */
+#define UINT16_MAX         (u16)(~((u16)0))    /* 0xFFFF             */
+#define UINT32_MAX         (u32)(~((u32)0))    /* 0xFFFFFFFF         */
+#define UINT64_MAX         (u64)(~((u64)0))    /* 0xFFFFFFFFFFFFFFFF */
+#define ASCII_MAX          (0x7F)
+#endif
+
+#define HINIC_MIN_N_TUPLE_PRIO         1
+#define HINIC_MAX_N_TUPLE_PRIO         7
+
+/**
+ * Endless loop will never happen with below assumption
+ * 1. there is at least one no-void item(END)
+ * 2. cur is before END.
+ */
+static inline const struct rte_flow_item *
+next_no_void_pattern(const struct rte_flow_item pattern[],
+               const struct rte_flow_item *cur)
+{
+       const struct rte_flow_item *next =
+               cur ? cur + 1 : &pattern[0];
+       while (1) {
+               if (next->type != RTE_FLOW_ITEM_TYPE_VOID)
+                       return next;
+               next++;
+       }
+}
+
+static inline const struct rte_flow_action *
+next_no_void_action(const struct rte_flow_action actions[],
+               const struct rte_flow_action *cur)
+{
+       const struct rte_flow_action *next =
+               cur ? cur + 1 : &actions[0];
+       while (1) {
+               if (next->type != RTE_FLOW_ACTION_TYPE_VOID)
+                       return next;
+               next++;
+       }
+}
+
+static int hinic_check_ethertype_attr_ele(const struct rte_flow_attr *attr,
+                                       struct rte_flow_error *error)
+{
+       /* Must be input direction */
+       if (!attr->ingress) {
+               rte_flow_error_set(error, EINVAL,
+                       RTE_FLOW_ERROR_TYPE_ATTR_INGRESS,
+                       attr, "Only support ingress.");
+               return -rte_errno;
+       }
+
+       if (attr->egress) {
+               rte_flow_error_set(error, EINVAL,
+                               RTE_FLOW_ERROR_TYPE_ATTR_EGRESS,
+                               attr, "Not support egress.");
+               return -rte_errno;
+       }
+
+       if (attr->priority) {
+               rte_flow_error_set(error, EINVAL,
+                               RTE_FLOW_ERROR_TYPE_ATTR_PRIORITY,
+                               attr, "Not support priority.");
+               return -rte_errno;
+       }
+
+       if (attr->group) {
+               rte_flow_error_set(error, EINVAL,
+                               RTE_FLOW_ERROR_TYPE_ATTR_GROUP,
+                               attr, "Not support group.");
+               return -rte_errno;
+       }
+
+       return 0;
+}
+
+static int hinic_check_filter_arg(const struct rte_flow_attr *attr,
+                               const struct rte_flow_item *pattern,
+                               const struct rte_flow_action *actions,
+                               struct rte_flow_error *error)
+{
+       if (!pattern) {
+               rte_flow_error_set(error, EINVAL, RTE_FLOW_ERROR_TYPE_ITEM_NUM,
+                               NULL, "NULL pattern.");
+               return -rte_errno;
+       }
+
+       if (!actions) {
+               rte_flow_error_set(error, EINVAL,
+                               RTE_FLOW_ERROR_TYPE_ACTION_NUM,
+                               NULL, "NULL action.");
+               return -rte_errno;
+       }
+
+       if (!attr) {
+               rte_flow_error_set(error, EINVAL, RTE_FLOW_ERROR_TYPE_ATTR,
+                                  NULL, "NULL attribute.");
+               return -rte_errno;
+       }
+
+       return 0;
+}
+
+static int hinic_check_ethertype_first_item(const struct rte_flow_item *item,
+                                       struct rte_flow_error *error)
+{
+       /* The first non-void item should be MAC */
+       if (item->type != RTE_FLOW_ITEM_TYPE_ETH) {
+               rte_flow_error_set(error, EINVAL,
+                       RTE_FLOW_ERROR_TYPE_ITEM,
+                       item, "Not supported by ethertype filter");
+               return -rte_errno;
+       }
+
+       /* Not supported last point for range */
+       if (item->last) {
+               rte_flow_error_set(error, EINVAL,
+                       RTE_FLOW_ERROR_TYPE_UNSPECIFIED,
+                       item, "Not supported last point for range");
+               return -rte_errno;
+       }
+
+       /* Get the MAC info. */
+       if (!item->spec || !item->mask) {
+               rte_flow_error_set(error, EINVAL,
+                               RTE_FLOW_ERROR_TYPE_ITEM,
+                               item, "Not supported by ethertype filter");
+               return -rte_errno;
+       }
+       return 0;
+}
+
+static int
+hinic_parse_ethertype_aciton(const struct rte_flow_action *actions,
+                       const struct rte_flow_action *act,
+                       const struct rte_flow_action_queue *act_q,
+                       struct rte_eth_ethertype_filter *filter,
+                       struct rte_flow_error *error)
+{
+       /* Parse action */
+       act = next_no_void_action(actions, NULL);
+       if (act->type != RTE_FLOW_ACTION_TYPE_QUEUE &&
+               act->type != RTE_FLOW_ACTION_TYPE_DROP) {
+               rte_flow_error_set(error, EINVAL,
+                               RTE_FLOW_ERROR_TYPE_ACTION,
+                               act, "Not supported action.");
+               return -rte_errno;
+       }
+
+       if (act->type == RTE_FLOW_ACTION_TYPE_QUEUE) {
+               act_q = (const struct rte_flow_action_queue *)act->conf;
+               filter->queue = act_q->index;
+       } else {
+               filter->flags |= RTE_ETHTYPE_FLAGS_DROP;
+       }
+
+       /* Check if the next non-void item is END */
+       act = next_no_void_action(actions, act);
+       if (act->type != RTE_FLOW_ACTION_TYPE_END) {
+               rte_flow_error_set(error, EINVAL,
+                               RTE_FLOW_ERROR_TYPE_ACTION,
+                               act, "Not supported action.");
+               return -rte_errno;
+       }
+
+       return 0;
+}
+
+/**
+ * Parse the rule to see if it is a ethertype rule.
+ * And get the ethertype filter info BTW.
+ * pattern:
+ * The first not void item can be ETH.
+ * The next not void item must be END.
+ * action:
+ * The first not void action should be QUEUE.
+ * The next not void action should be END.
+ * pattern example:
+ * ITEM                Spec                    Mask
+ * ETH         type    0x0807          0xFFFF
+ * END
+ * other members in mask and spec should set to 0x00.
+ * item->last should be NULL.
+ */
+static int
+cons_parse_ethertype_filter(const struct rte_flow_attr *attr,
+                       const struct rte_flow_item *pattern,
+                       const struct rte_flow_action *actions,
+                       struct rte_eth_ethertype_filter *filter,
+                       struct rte_flow_error *error)
+{
+       const struct rte_flow_item *item;
+       const struct rte_flow_action *act = NULL;
+       const struct rte_flow_item_eth *eth_spec;
+       const struct rte_flow_item_eth *eth_mask;
+       const struct rte_flow_action_queue *act_q = NULL;
+
+       if (hinic_check_filter_arg(attr, pattern, actions, error))
+               return -rte_errno;
+
+       item = next_no_void_pattern(pattern, NULL);
+       if (hinic_check_ethertype_first_item(item, error))
+               return -rte_errno;
+
+       eth_spec = (const struct rte_flow_item_eth *)item->spec;
+       eth_mask = (const struct rte_flow_item_eth *)item->mask;
+
+       /*
+        * Mask bits of source MAC address must be full of 0.
+        * Mask bits of destination MAC address must be full
+        * of 1 or full of 0.
+        */
+       if (!rte_is_zero_ether_addr(&eth_mask->src) ||
+           (!rte_is_zero_ether_addr(&eth_mask->dst) &&
+            !rte_is_broadcast_ether_addr(&eth_mask->dst))) {
+               rte_flow_error_set(error, EINVAL, RTE_FLOW_ERROR_TYPE_ITEM,
+                               item, "Invalid ether address mask");
+               return -rte_errno;
+       }
+
+       if ((eth_mask->type & UINT16_MAX) != UINT16_MAX) {
+               rte_flow_error_set(error, EINVAL, RTE_FLOW_ERROR_TYPE_ITEM,
+                               item, "Invalid ethertype mask");
+               return -rte_errno;
+       }
+
+       /*
+        * If mask bits of destination MAC address
+        * are full of 1, set RTE_ETHTYPE_FLAGS_MAC.
+        */
+       if (rte_is_broadcast_ether_addr(&eth_mask->dst)) {
+               filter->mac_addr = eth_spec->dst;
+               filter->flags |= RTE_ETHTYPE_FLAGS_MAC;
+       } else {
+               filter->flags &= ~RTE_ETHTYPE_FLAGS_MAC;
+       }
+       filter->ether_type = rte_be_to_cpu_16(eth_spec->type);
+
+       /* Check if the next non-void item is END. */
+       item = next_no_void_pattern(pattern, item);
+       if (item->type != RTE_FLOW_ITEM_TYPE_END) {
+               rte_flow_error_set(error, EINVAL, RTE_FLOW_ERROR_TYPE_ITEM,
+                       item, "Not supported by ethertype filter.");
+               return -rte_errno;
+       }
+
+       if (hinic_parse_ethertype_aciton(actions, act, act_q, filter, error))
+               return -rte_errno;
+
+       if (hinic_check_ethertype_attr_ele(attr, error))
+               return -rte_errno;
+
+       return 0;
+}
+
+static int
+hinic_parse_ethertype_filter(struct rte_eth_dev *dev,
+                       const struct rte_flow_attr *attr,
+                       const struct rte_flow_item pattern[],
+                       const struct rte_flow_action actions[],
+                       struct rte_eth_ethertype_filter *filter,
+                       struct rte_flow_error *error)
+{
+       if (cons_parse_ethertype_filter(attr, pattern, actions, filter, error))
+               return -rte_errno;
+
+       /* NIC doesn't support MAC address. */
+       if (filter->flags & RTE_ETHTYPE_FLAGS_MAC) {
+               memset(filter, 0, sizeof(struct rte_eth_ethertype_filter));
+               rte_flow_error_set(error, EINVAL,
+                       RTE_FLOW_ERROR_TYPE_ITEM,
+                       NULL, "Not supported by ethertype filter");
+               return -rte_errno;
+       }
+
+       if (filter->queue >= dev->data->nb_rx_queues) {
+               memset(filter, 0, sizeof(struct rte_eth_ethertype_filter));
+               rte_flow_error_set(error, EINVAL,
+                       RTE_FLOW_ERROR_TYPE_ITEM,
+                       NULL, "Queue index much too big");
+               return -rte_errno;
+       }
+
+       if (filter->ether_type == RTE_ETHER_TYPE_IPV4 ||
+               filter->ether_type == RTE_ETHER_TYPE_IPV6) {
+               memset(filter, 0, sizeof(struct rte_eth_ethertype_filter));
+               rte_flow_error_set(error, EINVAL,
+                       RTE_FLOW_ERROR_TYPE_ITEM,
+                       NULL, "IPv4/IPv6 not supported by ethertype filter");
+               return -rte_errno;
+       }
+
+       if (filter->flags & RTE_ETHTYPE_FLAGS_DROP) {
+               memset(filter, 0, sizeof(struct rte_eth_ethertype_filter));
+               rte_flow_error_set(error, EINVAL,
+                       RTE_FLOW_ERROR_TYPE_ITEM,
+                       NULL, "Drop option is unsupported");
+               return -rte_errno;
+       }
+
+       /* Hinic only support LACP/ARP for ether type */
+       if (filter->ether_type != RTE_ETHER_TYPE_SLOW &&
+               filter->ether_type != RTE_ETHER_TYPE_ARP) {
+               memset(filter, 0, sizeof(struct rte_eth_ethertype_filter));
+               rte_flow_error_set(error, EINVAL,
+                       RTE_FLOW_ERROR_TYPE_ITEM, NULL,
+                       "only lacp/arp type supported by ethertype filter");
+               return -rte_errno;
+       }
+
+       return 0;
+}
+
+static int hinic_check_ntuple_attr_ele(const struct rte_flow_attr *attr,
+                               struct rte_eth_ntuple_filter *filter,
+                               struct rte_flow_error *error)
+{
+       /* Must be input direction */
+       if (!attr->ingress) {
+               memset(filter, 0, sizeof(struct rte_eth_ntuple_filter));
+               rte_flow_error_set(error, EINVAL,
+                                  RTE_FLOW_ERROR_TYPE_ATTR_INGRESS,
+                                  attr, "Only support ingress.");
+               return -rte_errno;
+       }
+
+       if (attr->egress) {
+               memset(filter, 0, sizeof(struct rte_eth_ntuple_filter));
+               rte_flow_error_set(error, EINVAL,
+                                  RTE_FLOW_ERROR_TYPE_ATTR_EGRESS,
+                                  attr, "Not support egress.");
+               return -rte_errno;
+       }
+
+       if (attr->priority > 0xFFFF) {
+               memset(filter, 0, sizeof(struct rte_eth_ntuple_filter));
+               rte_flow_error_set(error, EINVAL,
+                                  RTE_FLOW_ERROR_TYPE_ATTR_PRIORITY,
+                                  attr, "Error priority.");
+               return -rte_errno;
+       }
+
+       if (attr->priority < HINIC_MIN_N_TUPLE_PRIO ||
+                   attr->priority > HINIC_MAX_N_TUPLE_PRIO)
+               filter->priority = 1;
+       else
+               filter->priority = (uint16_t)attr->priority;
+
+       return 0;
+}
+
+static int
+hinic_check_ntuple_act_ele(__rte_unused const struct rte_flow_item *item,
+                       const struct rte_flow_action actions[],
+                       struct rte_eth_ntuple_filter *filter,
+                       struct rte_flow_error *error)
+{
+       const struct rte_flow_action *act;
+       /*
+        * n-tuple only supports forwarding,
+        * check if the first not void action is QUEUE.
+        */
+       act = next_no_void_action(actions, NULL);
+       if (act->type != RTE_FLOW_ACTION_TYPE_QUEUE) {
+               memset(filter, 0, sizeof(struct rte_eth_ntuple_filter));
+               rte_flow_error_set(error, EINVAL,
+                       RTE_FLOW_ERROR_TYPE_ACTION,
+                       act, "Flow action type is not QUEUE.");
+               return -rte_errno;
+       }
+       filter->queue =
+               ((const struct rte_flow_action_queue *)act->conf)->index;
+
+       /* Check if the next not void item is END */
+       act = next_no_void_action(actions, act);
+       if (act->type != RTE_FLOW_ACTION_TYPE_END) {
+               memset(filter, 0, sizeof(struct rte_eth_ntuple_filter));
+               rte_flow_error_set(error, EINVAL,
+                       RTE_FLOW_ERROR_TYPE_ACTION,
+                       act, "Next not void item is not END.");
+               return -rte_errno;
+       }
+
+       return 0;
+}
+
+static int hinic_ntuple_item_check_ether(const struct rte_flow_item **ipv4_item,
+                                       const struct rte_flow_item pattern[],
+                                       struct rte_flow_error *error)
+{
+       const struct rte_flow_item *item;
+
+       /* The first not void item can be MAC or IPv4 */
+       item = next_no_void_pattern(pattern, NULL);
+
+       if (item->type != RTE_FLOW_ITEM_TYPE_ETH &&
+               item->type != RTE_FLOW_ITEM_TYPE_IPV4) {
+               rte_flow_error_set(error, EINVAL,
+                       RTE_FLOW_ERROR_TYPE_ITEM,
+                       item, "Not supported by ntuple filter");
+               return -rte_errno;
+       }
+
+       /* Skip Ethernet */
+       if (item->type == RTE_FLOW_ITEM_TYPE_ETH) {
+               /* Not supported last point for range */
+               if (item->last) {
+                       rte_flow_error_set(error,
+                               EINVAL,
+                               RTE_FLOW_ERROR_TYPE_UNSPECIFIED,
+                               item, "Not supported last point for range");
+                       return -rte_errno;
+               }
+               /* if the first item is MAC, the content should be NULL */
+               if (item->spec || item->mask) {
+                       rte_flow_error_set(error, EINVAL,
+                               RTE_FLOW_ERROR_TYPE_ITEM,
+                               item, "Not supported by ntuple filter");
+                       return -rte_errno;
+               }
+               /* check if the next not void item is IPv4 */
+               item = next_no_void_pattern(pattern, item);
+               if (item->type != RTE_FLOW_ITEM_TYPE_IPV4) {
+                       rte_flow_error_set(error,
+                               EINVAL, RTE_FLOW_ERROR_TYPE_ITEM,
+                               item, "Not supported by ntuple filter");
+                       return -rte_errno;
+               }
+       }
+
+       *ipv4_item = item;
+       return 0;
+}
+
+static int
+hinic_ntuple_item_check_ipv4(const struct rte_flow_item **in_out_item,
+                       const struct rte_flow_item pattern[],
+                       struct rte_eth_ntuple_filter *filter,
+                       struct rte_flow_error *error)
+{
+       const struct rte_flow_item_ipv4 *ipv4_spec;
+       const struct rte_flow_item_ipv4 *ipv4_mask;
+       const struct rte_flow_item *item = *in_out_item;
+
+       /* Get the IPv4 info */
+       if (!item->spec || !item->mask) {
+               rte_flow_error_set(error, EINVAL,
+                       RTE_FLOW_ERROR_TYPE_ITEM,
+                       item, "Invalid ntuple mask");
+               return -rte_errno;
+       }
+       /* Not supported last point for range */
+       if (item->last) {
+               rte_flow_error_set(error, EINVAL,
+                       RTE_FLOW_ERROR_TYPE_UNSPECIFIED,
+                       item, "Not supported last point for range");
+               return -rte_errno;
+       }
+
+       ipv4_mask = (const struct rte_flow_item_ipv4 *)item->mask;
+       /*
+        * Only support src & dst addresses, protocol,
+        * others should be masked.
+        */
+       if (ipv4_mask->hdr.version_ihl ||
+               ipv4_mask->hdr.type_of_service ||
+               ipv4_mask->hdr.total_length ||
+               ipv4_mask->hdr.packet_id ||
+               ipv4_mask->hdr.fragment_offset ||
+               ipv4_mask->hdr.time_to_live ||
+               ipv4_mask->hdr.hdr_checksum ||
+               !ipv4_mask->hdr.next_proto_id) {
+               rte_flow_error_set(error,
+                       EINVAL, RTE_FLOW_ERROR_TYPE_ITEM,
+                       item, "Not supported by ntuple filter");
+               return -rte_errno;
+       }
+
+       filter->dst_ip_mask = ipv4_mask->hdr.dst_addr;
+       filter->src_ip_mask = ipv4_mask->hdr.src_addr;
+       filter->proto_mask = ipv4_mask->hdr.next_proto_id;
+
+       ipv4_spec = (const struct rte_flow_item_ipv4 *)item->spec;
+       filter->dst_ip = ipv4_spec->hdr.dst_addr;
+       filter->src_ip = ipv4_spec->hdr.src_addr;
+       filter->proto  = ipv4_spec->hdr.next_proto_id;
+
+       /* Get next no void item */
+       *in_out_item = next_no_void_pattern(pattern, item);
+       return 0;
+}
+
+static int hinic_ntuple_item_check_l4(const struct rte_flow_item **in_out_item,
+                               const struct rte_flow_item pattern[],
+                               struct rte_eth_ntuple_filter *filter,
+                               struct rte_flow_error *error)
+{
+       const struct rte_flow_item_tcp *tcp_spec;
+       const struct rte_flow_item_tcp *tcp_mask;
+       const struct rte_flow_item_icmp *icmp_mask;
+       const struct rte_flow_item *item = *in_out_item;
+       u32 ntuple_filter_size = sizeof(struct rte_eth_ntuple_filter);
+
+       if (item->type == RTE_FLOW_ITEM_TYPE_END)
+               return 0;
+
+       /* Get TCP or UDP info */
+       if (item->type != RTE_FLOW_ITEM_TYPE_END &&
+               (!item->spec || !item->mask)) {
+               memset(filter, 0, ntuple_filter_size);
+               rte_flow_error_set(error, EINVAL,
+                       RTE_FLOW_ERROR_TYPE_ITEM,
+                       item, "Invalid ntuple mask");
+               return -rte_errno;
+       }
+
+       /* Not supported last point for range */
+       if (item->last) {
+               memset(filter, 0, ntuple_filter_size);
+               rte_flow_error_set(error, EINVAL,
+                       RTE_FLOW_ERROR_TYPE_UNSPECIFIED,
+                       item, "Not supported last point for range");
+               return -rte_errno;
+       }
+
+       if (item->type == RTE_FLOW_ITEM_TYPE_TCP) {
+               tcp_mask = (const struct rte_flow_item_tcp *)item->mask;
+
+               /*
+                * Only support src & dst ports, tcp flags,
+                * others should be masked.
+                */
+               if (tcp_mask->hdr.sent_seq ||
+                       tcp_mask->hdr.recv_ack ||
+                       tcp_mask->hdr.data_off ||
+                       tcp_mask->hdr.rx_win ||
+                       tcp_mask->hdr.cksum ||
+                       tcp_mask->hdr.tcp_urp) {
+                       memset(filter, 0, ntuple_filter_size);
+                       rte_flow_error_set(error, EINVAL,
+                               RTE_FLOW_ERROR_TYPE_ITEM,
+                               item, "Not supported by ntuple filter");
+                       return -rte_errno;
+               }
+
+               filter->dst_port_mask  = tcp_mask->hdr.dst_port;
+               filter->src_port_mask  = tcp_mask->hdr.src_port;
+               if (tcp_mask->hdr.tcp_flags == 0xFF) {
+                       filter->flags |= RTE_NTUPLE_FLAGS_TCP_FLAG;
+               } else if (!tcp_mask->hdr.tcp_flags) {
+                       filter->flags &= ~RTE_NTUPLE_FLAGS_TCP_FLAG;
+               } else {
+                       memset(filter, 0, ntuple_filter_size);
+                       rte_flow_error_set(error, EINVAL,
+                               RTE_FLOW_ERROR_TYPE_ITEM,
+                               item, "Not supported by ntuple filter");
+                       return -rte_errno;
+               }
+
+               tcp_spec = (const struct rte_flow_item_tcp *)item->spec;
+               filter->dst_port  = tcp_spec->hdr.dst_port;
+               filter->src_port  = tcp_spec->hdr.src_port;
+               filter->tcp_flags = tcp_spec->hdr.tcp_flags;
+       } else if (item->type == RTE_FLOW_ITEM_TYPE_ICMP) {
+               icmp_mask = (const struct rte_flow_item_icmp *)item->mask;
+
+               /* ICMP all should be masked. */
+               if (icmp_mask->hdr.icmp_cksum ||
+                       icmp_mask->hdr.icmp_ident ||
+                       icmp_mask->hdr.icmp_seq_nb ||
+                       icmp_mask->hdr.icmp_type ||
+                       icmp_mask->hdr.icmp_code) {
+                       memset(filter, 0, ntuple_filter_size);
+                       rte_flow_error_set(error, EINVAL,
+                               RTE_FLOW_ERROR_TYPE_ITEM,
+                               item, "Not supported by ntuple filter");
+                       return -rte_errno;
+               }
+       }
+
+       /* Get next no void item */
+       *in_out_item = next_no_void_pattern(pattern, item);
+       return 0;
+}
+
+static int hinic_ntuple_item_check_end(const struct rte_flow_item *item,
+                                       struct rte_eth_ntuple_filter *filter,
+                                       struct rte_flow_error *error)
+{
+       /* Check if the next not void item is END */
+       if (item->type != RTE_FLOW_ITEM_TYPE_END) {
+               memset(filter, 0, sizeof(struct rte_eth_ntuple_filter));
+               rte_flow_error_set(error, EINVAL,
+                       RTE_FLOW_ERROR_TYPE_ITEM,
+                       item, "Not supported by ntuple filter");
+               return -rte_errno;
+       }
+       return 0;
+}
+
+static int hinic_check_ntuple_item_ele(const struct rte_flow_item *item,
+                                       const struct rte_flow_item pattern[],
+                                       struct rte_eth_ntuple_filter *filter,
+                                       struct rte_flow_error *error)
+{
+       if (hinic_ntuple_item_check_ether(&item, pattern, error) ||
+               hinic_ntuple_item_check_ipv4(&item, pattern, filter, error) ||
+               hinic_ntuple_item_check_l4(&item, pattern, filter, error) ||
+               hinic_ntuple_item_check_end(item, filter, error))
+               return -rte_errno;
+
+       return 0;
+}
+
+/**
+ * Parse the rule to see if it is a n-tuple rule.
+ * And get the n-tuple filter info BTW.
+ * pattern:
+ * The first not void item can be ETH or IPV4.
+ * The second not void item must be IPV4 if the first one is ETH.
+ * The third not void item must be UDP or TCP.
+ * The next not void item must be END.
+ * action:
+ * The first not void action should be QUEUE.
+ * The next not void action should be END.
+ * pattern example:
+ * ITEM                Spec                    Mask
+ * ETH         NULL                    NULL
+ * IPV4                src_addr 192.168.1.20   0xFFFFFFFF
+ *             dst_addr 192.167.3.50   0xFFFFFFFF
+ *             next_proto_id   17      0xFF
+ * UDP/TCP/    src_port        80      0xFFFF
+ * SCTP                dst_port        80      0xFFFF
+ * END
+ * other members in mask and spec should set to 0x00.
+ * item->last should be NULL.
+ * Please aware there's an asumption for all the parsers.
+ * rte_flow_item is using big endian, rte_flow_attr and
+ * rte_flow_action are using CPU order.
+ * Because the pattern is used to describe the packets,
+ * normally the packets should use network order.
+ */
+static int
+cons_parse_ntuple_filter(const struct rte_flow_attr *attr,
+                       const struct rte_flow_item pattern[],
+                       const struct rte_flow_action actions[],
+                       struct rte_eth_ntuple_filter *filter,
+                       struct rte_flow_error *error)
+{
+       const struct rte_flow_item *item = NULL;
+
+       if (hinic_check_filter_arg(attr, pattern, actions, error))
+               return -rte_errno;
+
+       if (hinic_check_ntuple_item_ele(item, pattern, filter, error))
+               return -rte_errno;
+
+       if (hinic_check_ntuple_act_ele(item, actions, filter, error))
+               return -rte_errno;
+
+       if (hinic_check_ntuple_attr_ele(attr, filter, error))
+               return -rte_errno;
+
+       return 0;
+}
+
+static int
+hinic_parse_ntuple_filter(struct rte_eth_dev *dev,
+                       const struct rte_flow_attr *attr,
+                       const struct rte_flow_item pattern[],
+                       const struct rte_flow_action actions[],
+                       struct rte_eth_ntuple_filter *filter,
+                       struct rte_flow_error *error)
+{
+       int ret;
+
+       ret = cons_parse_ntuple_filter(attr, pattern, actions, filter, error);
+       if (ret)
+               return ret;
+
+       /* Hinic doesn't support tcp flags */
+       if (filter->flags & RTE_NTUPLE_FLAGS_TCP_FLAG) {
+               memset(filter, 0, sizeof(struct rte_eth_ntuple_filter));
+               rte_flow_error_set(error, EINVAL,
+                                  RTE_FLOW_ERROR_TYPE_ITEM,
+                                  NULL, "Not supported by ntuple filter");
+               return -rte_errno;
+       }
+
+       /* Hinic doesn't support many priorities */
+       if (filter->priority < HINIC_MIN_N_TUPLE_PRIO ||
+           filter->priority > HINIC_MAX_N_TUPLE_PRIO) {
+               memset(filter, 0, sizeof(struct rte_eth_ntuple_filter));
+               rte_flow_error_set(error, EINVAL,
+                       RTE_FLOW_ERROR_TYPE_ITEM,
+                       NULL, "Priority not supported by ntuple filter");
+               return -rte_errno;
+       }
+
+       if (filter->queue >= dev->data->nb_rx_queues)
+               return -rte_errno;
+
+       /* Fixed value for hinic */
+       filter->flags = RTE_5TUPLE_FLAGS;
+       return 0;
+}
+
+static int hinic_normal_item_check_ether(const struct rte_flow_item **ip_item,
+                                       const struct rte_flow_item pattern[],
+                                       struct rte_flow_error *error)
+{
+       const struct rte_flow_item *item;
+
+       /* The first not void item can be MAC or IPv4  or TCP or UDP */
+       item = next_no_void_pattern(pattern, NULL);
+
+       if (item->type != RTE_FLOW_ITEM_TYPE_ETH &&
+               item->type != RTE_FLOW_ITEM_TYPE_IPV4 &&
+               item->type != RTE_FLOW_ITEM_TYPE_TCP &&
+               item->type != RTE_FLOW_ITEM_TYPE_UDP) {
+               rte_flow_error_set(error, EINVAL,
+                       RTE_FLOW_ERROR_TYPE_ITEM, item,
+                       "Not supported by fdir filter,support mac,ipv4,tcp,udp");
+               return -rte_errno;
+       }
+
+       /* Not supported last point for range */
+       if (item->last) {
+               rte_flow_error_set(error, EINVAL,
+                       RTE_FLOW_ERROR_TYPE_UNSPECIFIED, item,
+                       "Not supported last point for range");
+               return -rte_errno;
+       }
+
+       /* Skip Ethernet */
+       if (item->type == RTE_FLOW_ITEM_TYPE_ETH) {
+               /* All should be masked. */
+               if (item->spec || item->mask) {
+                       rte_flow_error_set(error, EINVAL,
+                               RTE_FLOW_ERROR_TYPE_ITEM,
+                               item, "Not supported by fdir filter,support mac");
+                       return -rte_errno;
+               }
+               /* Check if the next not void item is IPv4 */
+               item = next_no_void_pattern(pattern, item);
+               if (item->type != RTE_FLOW_ITEM_TYPE_IPV4) {
+                       rte_flow_error_set(error, EINVAL,
+                               RTE_FLOW_ERROR_TYPE_ITEM, item,
+                               "Not supported by fdir filter,support mac,ipv4");
+                       return -rte_errno;
+               }
+       }
+
+       *ip_item = item;
+       return 0;
+}
+
+static int hinic_normal_item_check_ip(const struct rte_flow_item **in_out_item,
+                               const struct rte_flow_item pattern[],
+                               struct hinic_fdir_rule *rule,
+                               struct rte_flow_error *error)
+{
+       const struct rte_flow_item_ipv4 *ipv4_spec;
+       const struct rte_flow_item_ipv4 *ipv4_mask;
+       const struct rte_flow_item *item = *in_out_item;
+
+       /* Get the IPv4 info */
+       if (item->type == RTE_FLOW_ITEM_TYPE_IPV4) {
+               /* Not supported last point for range */
+               if (item->last) {
+                       rte_flow_error_set(error, EINVAL,
+                               RTE_FLOW_ERROR_TYPE_UNSPECIFIED,
+                               item, "Not supported last point for range");
+                       return -rte_errno;
+               }
+
+               if (!item->mask) {
+                       memset(rule, 0, sizeof(struct hinic_fdir_rule));
+                       rte_flow_error_set(error, EINVAL,
+                               RTE_FLOW_ERROR_TYPE_ITEM,
+                               item, "Invalid fdir filter mask");
+                       return -rte_errno;
+               }
+
+               ipv4_mask = (const struct rte_flow_item_ipv4 *)item->mask;
+               /*
+                * Only support src & dst addresses,
+                * others should be masked.
+                */
+               if (ipv4_mask->hdr.version_ihl ||
+                       ipv4_mask->hdr.type_of_service ||
+                       ipv4_mask->hdr.total_length ||
+                       ipv4_mask->hdr.packet_id ||
+                       ipv4_mask->hdr.fragment_offset ||
+                       ipv4_mask->hdr.time_to_live ||
+                       ipv4_mask->hdr.next_proto_id ||
+                       ipv4_mask->hdr.hdr_checksum) {
+                       rte_flow_error_set(error,
+                               EINVAL, RTE_FLOW_ERROR_TYPE_ITEM, item,
+                               "Not supported by fdir filter, support src,dst ip");
+                       return -rte_errno;
+               }
+
+               rule->mask.dst_ipv4_mask = ipv4_mask->hdr.dst_addr;
+               rule->mask.src_ipv4_mask = ipv4_mask->hdr.src_addr;
+
+               if (item->spec) {
+                       ipv4_spec =
+                               (const struct rte_flow_item_ipv4 *)item->spec;
+                       rule->hinic_fdir.dst_ip = ipv4_spec->hdr.dst_addr;
+                       rule->hinic_fdir.src_ip = ipv4_spec->hdr.src_addr;
+               }
+
+               /*
+                * Check if the next not void item is
+                * TCP or UDP or END.
+                */
+               item = next_no_void_pattern(pattern, item);
+               if (item->type != RTE_FLOW_ITEM_TYPE_TCP &&
+                   item->type != RTE_FLOW_ITEM_TYPE_UDP &&
+                   item->type != RTE_FLOW_ITEM_TYPE_END) {
+                       memset(rule, 0, sizeof(struct hinic_fdir_rule));
+                       rte_flow_error_set(error, EINVAL,
+                               RTE_FLOW_ERROR_TYPE_ITEM, item,
+                               "Not supported by fdir filter, support tcp, udp, end");
+                       return -rte_errno;
+               }
+       }
+
+       *in_out_item = item;
+       return 0;
+}
+
+static int hinic_normal_item_check_l4(const struct rte_flow_item **in_out_item,
+                               const struct rte_flow_item pattern[],
+                               struct hinic_fdir_rule *rule,
+                               struct rte_flow_error *error)
+{
+       const struct rte_flow_item_tcp *tcp_spec;
+       const struct rte_flow_item_tcp *tcp_mask;
+       const struct rte_flow_item_udp *udp_spec;
+       const struct rte_flow_item_udp *udp_mask;
+       const struct rte_flow_item *item = *in_out_item;
+
+       if (item->type != RTE_FLOW_ITEM_TYPE_END) {
+               /* Not supported last point for range */
+               if (item->last) {
+                       rte_flow_error_set(error, EINVAL,
+                               RTE_FLOW_ERROR_TYPE_UNSPECIFIED,
+                               item, "Not supported last point for range");
+                       return -rte_errno;
+               }
+
+               /* Get TCP/UDP info */
+               if (item->type == RTE_FLOW_ITEM_TYPE_TCP) {
+                       /*
+                        * Only care about src & dst ports,
+                        * others should be masked.
+                        */
+                       if (!item->mask) {
+                               memset(rule, 0, sizeof(struct hinic_fdir_rule));
+                               rte_flow_error_set(error, EINVAL,
+                                       RTE_FLOW_ERROR_TYPE_ITEM, item,
+                                       "Not supported by fdir filter,support src,dst ports");
+                               return -rte_errno;
+                       }
+
+                       tcp_mask = (const struct rte_flow_item_tcp *)item->mask;
+                       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) {
+                               memset(rule, 0, sizeof(struct hinic_fdir_rule));
+                               rte_flow_error_set(error, EINVAL,
+                                       RTE_FLOW_ERROR_TYPE_ITEM,
+                                       item, "Not supported by fdir filter,support tcp");
+                               return -rte_errno;
+                       }
+
+                       rule->mask.src_port_mask = tcp_mask->hdr.src_port;
+                       rule->mask.dst_port_mask = tcp_mask->hdr.dst_port;
+
+                       if (item->spec) {
+                               tcp_spec =
+                                       (const struct rte_flow_item_tcp *)
+                                       item->spec;
+                               rule->hinic_fdir.src_port =
+                                       tcp_spec->hdr.src_port;
+                               rule->hinic_fdir.dst_port =
+                                       tcp_spec->hdr.dst_port;
+                       }
+
+               } else if (item->type == RTE_FLOW_ITEM_TYPE_UDP) {
+                       /*
+                        * Only care about src & dst ports,
+                        * others should be masked.
+                        */
+                       if (!item->mask) {
+                               memset(rule, 0, sizeof(struct hinic_fdir_rule));
+                               rte_flow_error_set(error, EINVAL,
+                                       RTE_FLOW_ERROR_TYPE_ITEM,
+                                       item, "Not supported by fdir filter,support src,dst ports");
+                               return -rte_errno;
+                       }
+
+                       udp_mask = (const struct rte_flow_item_udp *)item->mask;
+                       if (udp_mask->hdr.dgram_len ||
+                           udp_mask->hdr.dgram_cksum) {
+                               memset(rule, 0, sizeof(struct hinic_fdir_rule));
+                               rte_flow_error_set(error, EINVAL,
+                                       RTE_FLOW_ERROR_TYPE_ITEM,
+                                       item, "Not supported by fdir filter,support udp");
+                               return -rte_errno;
+                       }
+                       rule->mask.src_port_mask = udp_mask->hdr.src_port;
+                       rule->mask.dst_port_mask = udp_mask->hdr.dst_port;
+
+                       if (item->spec) {
+                               udp_spec =
+                                       (const struct rte_flow_item_udp *)
+                                       item->spec;
+                               rule->hinic_fdir.src_port =
+                                       udp_spec->hdr.src_port;
+                               rule->hinic_fdir.dst_port =
+                                       udp_spec->hdr.dst_port;
+                       }
+               } else {
+                       memset(rule, 0, sizeof(struct hinic_fdir_rule));
+                       rte_flow_error_set(error, EINVAL,
+                               RTE_FLOW_ERROR_TYPE_ITEM,
+                               item, "Not supported by fdir filter,support tcp/udp");
+                       return -rte_errno;
+               }
+
+               /* Get next no void item */
+               *in_out_item = next_no_void_pattern(pattern, item);
+       }
+
+       return 0;
+}
+
+static int hinic_normal_item_check_end(const struct rte_flow_item *item,
+                                       struct hinic_fdir_rule *rule,
+                                       struct rte_flow_error *error)
+{
+       /* Check if the next not void item is END */
+       if (item->type != RTE_FLOW_ITEM_TYPE_END) {
+               memset(rule, 0, sizeof(struct hinic_fdir_rule));
+               rte_flow_error_set(error, EINVAL,
+                       RTE_FLOW_ERROR_TYPE_ITEM,
+                       item, "Not supported by fdir filter,support end");
+               return -rte_errno;
+       }
+
+       return 0;
+}
+
+static int hinic_check_normal_item_ele(const struct rte_flow_item *item,
+                                       const struct rte_flow_item pattern[],
+                                       struct hinic_fdir_rule *rule,
+                                       struct rte_flow_error *error)
+{
+       if (hinic_normal_item_check_ether(&item, pattern, error) ||
+               hinic_normal_item_check_ip(&item, pattern, rule, error) ||
+               hinic_normal_item_check_l4(&item, pattern, rule, error) ||
+               hinic_normal_item_check_end(item, rule, error))
+               return -rte_errno;
+
+       return 0;
+}
+
+static int hinic_check_normal_attr_ele(const struct rte_flow_attr *attr,
+                                       struct hinic_fdir_rule *rule,
+                                       struct rte_flow_error *error)
+{
+       /* Must be input direction */
+       if (!attr->ingress) {
+               memset(rule, 0, sizeof(struct hinic_fdir_rule));
+               rte_flow_error_set(error, EINVAL,
+                                  RTE_FLOW_ERROR_TYPE_ATTR_INGRESS,
+                                  attr, "Only support ingress.");
+               return -rte_errno;
+       }
+
+       /* Not supported */
+       if (attr->egress) {
+               memset(rule, 0, sizeof(struct hinic_fdir_rule));
+               rte_flow_error_set(error, EINVAL,
+                                  RTE_FLOW_ERROR_TYPE_ATTR_EGRESS,
+                                  attr, "Not support egress.");
+               return -rte_errno;
+       }
+
+       /* Not supported */
+       if (attr->priority) {
+               memset(rule, 0, sizeof(struct hinic_fdir_rule));
+               rte_flow_error_set(error, EINVAL,
+                       RTE_FLOW_ERROR_TYPE_ATTR_PRIORITY,
+                       attr, "Not support priority.");
+               return -rte_errno;
+       }
+
+       return 0;
+}
+
+static int hinic_check_normal_act_ele(const struct rte_flow_item *item,
+                               const struct rte_flow_action actions[],
+                               struct hinic_fdir_rule *rule,
+                               struct rte_flow_error *error)
+{
+       const struct rte_flow_action *act;
+
+       /* Check if the first not void action is QUEUE */
+       act = next_no_void_action(actions, NULL);
+       if (act->type != RTE_FLOW_ACTION_TYPE_QUEUE) {
+               memset(rule, 0, sizeof(struct hinic_fdir_rule));
+               rte_flow_error_set(error, EINVAL, RTE_FLOW_ERROR_TYPE_ACTION,
+                       item, "Not supported action.");
+               return -rte_errno;
+       }
+
+       rule->queue = ((const struct rte_flow_action_queue *)act->conf)->index;
+
+       /* Check if the next not void item is END */
+       act = next_no_void_action(actions, act);
+       if (act->type != RTE_FLOW_ACTION_TYPE_END) {
+               memset(rule, 0, sizeof(struct hinic_fdir_rule));
+               rte_flow_error_set(error, EINVAL, RTE_FLOW_ERROR_TYPE_ACTION,
+                       act, "Not supported action.");
+               return -rte_errno;
+       }
+
+       return 0;
+}
+
+/**
+ * Parse the rule to see if it is a IP or MAC VLAN flow director rule.
+ * And get the flow director filter info BTW.
+ * UDP/TCP/SCTP PATTERN:
+ * The first not void item can be ETH or IPV4 or IPV6
+ * The second not void item must be IPV4 or IPV6 if the first one is ETH.
+ * The next not void item could be UDP or TCP(optional)
+ * The next not void item must be END.
+ * ACTION:
+ * The first not void action should be QUEUE.
+ * The second not void optional action should be MARK,
+ * mark_id is a uint32_t number.
+ * The next not void action should be END.
+ * UDP/TCP pattern example:
+ * ITEM          Spec                                      Mask
+ * ETH            NULL                                    NULL
+ * IPV4           src_addr  1.2.3.6                 0xFFFFFFFF
+ *                   dst_addr  1.2.3.5                 0xFFFFFFFF
+ * UDP/TCP    src_port  80                         0xFFFF
+ *                   dst_port  80                         0xFFFF
+ * END
+ * Other members in mask and spec should set to 0x00.
+ * Item->last should be NULL.
+ */
+static int
+hinic_parse_fdir_filter_normal(const struct rte_flow_attr *attr,
+                              const struct rte_flow_item pattern[],
+                              const struct rte_flow_action actions[],
+                              struct hinic_fdir_rule *rule,
+                              struct rte_flow_error *error)
+{
+       const struct rte_flow_item *item = NULL;
+
+       if (hinic_check_filter_arg(attr, pattern, actions, error))
+               return -rte_errno;
+
+       if (hinic_check_normal_item_ele(item, pattern, rule, error))
+               return -rte_errno;
+
+       if (hinic_check_normal_attr_ele(attr, rule, error))
+               return -rte_errno;
+
+       if (hinic_check_normal_act_ele(item, actions, rule, error))
+               return -rte_errno;
+
+       return 0;
+}
+
+static int
+hinic_parse_fdir_filter(struct rte_eth_dev *dev,
+                       const struct rte_flow_attr *attr,
+                       const struct rte_flow_item pattern[],
+                       const struct rte_flow_action actions[],
+                       struct hinic_fdir_rule *rule,
+                       struct rte_flow_error *error)
+{
+       int ret;
+
+       ret = hinic_parse_fdir_filter_normal(attr, pattern,
+                                               actions, rule, error);
+       if (ret)
+               return ret;
+
+       if (rule->queue >= dev->data->nb_rx_queues)
+               return -ENOTSUP;
+
+       return ret;
+}
+
+/**
+ * Check if the flow rule is supported by nic.
+ * It only checkes the format. Don't guarantee the rule can be programmed into
+ * the HW. Because there can be no enough room for the rule.
+ */
+static int hinic_flow_validate(struct rte_eth_dev *dev,
+                               const struct rte_flow_attr *attr,
+                               const struct rte_flow_item pattern[],
+                               const struct rte_flow_action actions[],
+                               struct rte_flow_error *error)
+{
+       struct rte_eth_ethertype_filter ethertype_filter;
+       struct rte_eth_ntuple_filter ntuple_filter;
+       struct hinic_fdir_rule fdir_rule;
+       int ret;
+
+       memset(&ntuple_filter, 0, sizeof(struct rte_eth_ntuple_filter));
+       ret = hinic_parse_ntuple_filter(dev, attr, pattern,
+                               actions, &ntuple_filter, error);
+       if (!ret)
+               return 0;
+
+       memset(&ethertype_filter, 0, sizeof(struct rte_eth_ethertype_filter));
+       ret = hinic_parse_ethertype_filter(dev, attr, pattern,
+                               actions, &ethertype_filter, error);
+
+       if (!ret)
+               return 0;
+
+       memset(&fdir_rule, 0, sizeof(struct hinic_fdir_rule));
+       ret = hinic_parse_fdir_filter(dev, attr, pattern,
+                               actions, &fdir_rule, error);
+
+       return ret;
+}
+
+const struct rte_flow_ops hinic_flow_ops = {
+       .validate = hinic_flow_validate,
+};
index 87c8d16..bc7e246 100644 (file)
@@ -8,6 +8,7 @@ sources = files(
        'hinic_pmd_ethdev.c',
        'hinic_pmd_rx.c',
        'hinic_pmd_tx.c',
+       'hinic_pmd_flow.c',
        )
 
 includes += include_directories('base')