add prefix to cache line macros
[dpdk.git] / lib / librte_ether / rte_ethdev.c
index db013e5..95f2ceb 100644 (file)
@@ -1,35 +1,34 @@
 /*-
  *   BSD LICENSE
- * 
- *   Copyright(c) 2010-2012 Intel Corporation. All rights reserved.
+ *
+ *   Copyright(c) 2010-2014 Intel Corporation. All rights reserved.
  *   All rights reserved.
- * 
- *   Redistribution and use in source and binary forms, with or without 
- *   modification, are permitted provided that the following conditions 
+ *
+ *   Redistribution and use in source and binary forms, with or without
+ *   modification, are permitted provided that the following conditions
  *   are met:
- * 
- *     * Redistributions of source code must retain the above copyright 
+ *
+ *     * Redistributions of source code must retain the above copyright
  *       notice, this list of conditions and the following disclaimer.
- *     * Redistributions in binary form must reproduce the above copyright 
- *       notice, this list of conditions and the following disclaimer in 
- *       the documentation and/or other materials provided with the 
+ *     * Redistributions in binary form must reproduce the above copyright
+ *       notice, this list of conditions and the following disclaimer in
+ *       the documentation and/or other materials provided with the
  *       distribution.
- *     * Neither the name of Intel Corporation nor the names of its 
- *       contributors may be used to endorse or promote products derived 
+ *     * Neither the name of Intel Corporation nor the names of its
+ *       contributors may be used to endorse or promote products derived
  *       from this software without specific prior written permission.
- * 
- *   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS 
- *   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT 
- *   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR 
- *   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT 
- *   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, 
- *   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT 
- *   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 
- *   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 
- *   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 
- *   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE 
+ *
+ *   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ *   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ *   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ *   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ *   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ *   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ *   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ *   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ *   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ *   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
  *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
- * 
  */
 
 #include <sys/types.h>
@@ -42,6 +41,7 @@
 #include <errno.h>
 #include <stdint.h>
 #include <inttypes.h>
+#include <netinet/in.h>
 
 #include <rte_byteorder.h>
 #include <rte_log.h>
@@ -65,6 +65,7 @@
 #include <rte_mbuf.h>
 #include <rte_errno.h>
 #include <rte_spinlock.h>
+#include <rte_string_fns.h>
 
 #include "rte_ether.h"
 #include "rte_ethdev.h"
@@ -113,6 +114,48 @@ static uint8_t nb_ports = 0;
 /* spinlock for eth device callbacks */
 static rte_spinlock_t rte_eth_dev_cb_lock = RTE_SPINLOCK_INITIALIZER;
 
+/* store statistics names and its offset in stats structure  */
+struct rte_eth_xstats_name_off {
+       char name[RTE_ETH_XSTATS_NAME_SIZE];
+       unsigned offset;
+};
+
+static struct rte_eth_xstats_name_off rte_stats_strings[] = {
+        {"rx_packets", offsetof(struct rte_eth_stats, ipackets)},
+        {"tx_packets", offsetof(struct rte_eth_stats, opackets)},
+        {"rx_bytes", offsetof(struct rte_eth_stats, ibytes)},
+        {"tx_bytes", offsetof(struct rte_eth_stats, obytes)},
+        {"tx_errors", offsetof(struct rte_eth_stats, oerrors)},
+        {"rx_missed_errors", offsetof(struct rte_eth_stats, imissed)},
+        {"rx_crc_errors", offsetof(struct rte_eth_stats, ibadcrc)},
+        {"rx_bad_length_errors", offsetof(struct rte_eth_stats, ibadlen)},
+        {"rx_errors", offsetof(struct rte_eth_stats, ierrors)},
+        {"alloc_rx_buff_failed", offsetof(struct rte_eth_stats, rx_nombuf)},
+        {"fdir_match", offsetof(struct rte_eth_stats, fdirmatch)},
+        {"fdir_miss", offsetof(struct rte_eth_stats, fdirmiss)},
+        {"tx_flow_control_xon", offsetof(struct rte_eth_stats, tx_pause_xon)},
+        {"rx_flow_control_xon", offsetof(struct rte_eth_stats, rx_pause_xon)},
+        {"tx_flow_control_xoff", offsetof(struct rte_eth_stats, tx_pause_xoff)},
+        {"rx_flow_control_xoff", offsetof(struct rte_eth_stats, rx_pause_xoff)},
+};
+#define RTE_NB_STATS (sizeof(rte_stats_strings) / sizeof(rte_stats_strings[0]))
+
+static struct rte_eth_xstats_name_off rte_rxq_stats_strings[] = {
+       {"rx_packets", offsetof(struct rte_eth_stats, q_ipackets)},
+       {"rx_bytes", offsetof(struct rte_eth_stats, q_ibytes)},
+};
+#define RTE_NB_RXQ_STATS (sizeof(rte_rxq_stats_strings) /      \
+               sizeof(rte_rxq_stats_strings[0]))
+
+static struct rte_eth_xstats_name_off rte_txq_stats_strings[] = {
+       {"tx_packets", offsetof(struct rte_eth_stats, q_opackets)},
+       {"tx_bytes", offsetof(struct rte_eth_stats, q_obytes)},
+       {"tx_errors", offsetof(struct rte_eth_stats, q_errors)},
+};
+#define RTE_NB_TXQ_STATS (sizeof(rte_txq_stats_strings) /      \
+               sizeof(rte_txq_stats_strings[0]))
+
+
 /**
  * The user application callback description.
  *
@@ -124,6 +167,7 @@ struct rte_eth_dev_callback {
        rte_eth_dev_cb_fn cb_fn;                /**< Callback address */
        void *cb_arg;                           /**< Parameter for callback */
        enum rte_eth_event_type event;          /**< Interrupt event type */
+       uint32_t active;                        /**< Callback is executing */
 };
 
 enum {
@@ -152,21 +196,39 @@ rte_eth_dev_data_alloc(void)
                                RTE_MAX_ETHPORTS * sizeof(*rte_eth_dev_data));
 }
 
-static inline struct rte_eth_dev *
-rte_eth_dev_allocate(void)
+static struct rte_eth_dev *
+rte_eth_dev_allocated(const char *name)
+{
+       unsigned i;
+
+       for (i = 0; i < nb_ports; i++) {
+               if (strcmp(rte_eth_devices[i].data->name, name) == 0)
+                       return &rte_eth_devices[i];
+       }
+       return NULL;
+}
+
+struct rte_eth_dev *
+rte_eth_dev_allocate(const char *name)
 {
        struct rte_eth_dev *eth_dev;
 
        if (nb_ports == RTE_MAX_ETHPORTS) {
-               PMD_DEBUG_TRACE("Reached maximum number of ethernet ports\n");
+               PMD_DEBUG_TRACE("Reached maximum number of Ethernet ports\n");
                return NULL;
        }
 
        if (rte_eth_dev_data == NULL)
                rte_eth_dev_data_alloc();
 
+       if (rte_eth_dev_allocated(name) != NULL) {
+               PMD_DEBUG_TRACE("Ethernet Device with name %s already allocated!\n", name);
+               return NULL;
+       }
+
        eth_dev = &rte_eth_devices[nb_ports];
        eth_dev->data = &rte_eth_dev_data[nb_ports];
+       snprintf(eth_dev->data->name, sizeof(eth_dev->data->name), "%s", name);
        eth_dev->data->port_id = nb_ports++;
        return eth_dev;
 }
@@ -177,18 +239,24 @@ rte_eth_dev_init(struct rte_pci_driver *pci_drv,
 {
        struct eth_driver    *eth_drv;
        struct rte_eth_dev *eth_dev;
+       char ethdev_name[RTE_ETH_NAME_MAX_LEN];
+
        int diag;
 
        eth_drv = (struct eth_driver *)pci_drv;
 
-       eth_dev = rte_eth_dev_allocate();
+       /* Create unique Ethernet device name using PCI address */
+       snprintf(ethdev_name, RTE_ETH_NAME_MAX_LEN, "%d:%d.%d",
+                       pci_dev->addr.bus, pci_dev->addr.devid, pci_dev->addr.function);
+
+       eth_dev = rte_eth_dev_allocate(ethdev_name);
        if (eth_dev == NULL)
                return -ENOMEM;
 
        if (rte_eal_process_type() == RTE_PROC_PRIMARY){
                eth_dev->data->dev_private = rte_zmalloc("ethdev private structure",
                                  eth_drv->dev_private_size,
-                                 CACHE_LINE_SIZE);
+                                 RTE_CACHE_LINE_SIZE);
                if (eth_dev->data->dev_private == NULL)
                        rte_panic("Cannot allocate memzone for private port data\n");
        }
@@ -200,9 +268,9 @@ rte_eth_dev_init(struct rte_pci_driver *pci_drv,
        TAILQ_INIT(&(eth_dev->callbacks));
 
        /*
-        * Set the default maximum frame size.
+        * Set the default MTU.
         */
-       eth_dev->data->max_frame_size = ETHER_MAX_LEN;
+       eth_dev->data->mtu = ETHER_MTU;
 
        /* Invoke PMD device initialization function */
        diag = (*eth_drv->eth_dev_init)(eth_drv, eth_dev);
@@ -240,6 +308,14 @@ rte_eth_driver_register(struct eth_driver *eth_drv)
        rte_eal_pci_register(&eth_drv->pci_drv);
 }
 
+int
+rte_eth_dev_socket_id(uint8_t port_id)
+{
+       if (port_id >= nb_ports)
+               return -1;
+       return rte_eth_devices[port_id].pci_dev->numa_node;
+}
+
 uint8_t
 rte_eth_dev_count(void)
 {
@@ -253,23 +329,23 @@ rte_eth_dev_rx_queue_config(struct rte_eth_dev *dev, uint16_t nb_queues)
        void **rxq;
        unsigned i;
 
-       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->rx_queue_release, -ENOTSUP);
-
-       if (dev->data->rx_queues == NULL) {
+       if (dev->data->rx_queues == NULL) { /* first time configuration */
                dev->data->rx_queues = rte_zmalloc("ethdev->rx_queues",
                                sizeof(dev->data->rx_queues[0]) * nb_queues,
-                               CACHE_LINE_SIZE);
+                               RTE_CACHE_LINE_SIZE);
                if (dev->data->rx_queues == NULL) {
                        dev->data->nb_rx_queues = 0;
                        return -(ENOMEM);
                }
-       } else {
+       } else { /* re-configure */
+               FUNC_PTR_OR_ERR_RET(*dev->dev_ops->rx_queue_release, -ENOTSUP);
+
                rxq = dev->data->rx_queues;
 
                for (i = nb_queues; i < old_nb_queues; i++)
                        (*dev->dev_ops->rx_queue_release)(rxq[i]);
                rxq = rte_realloc(rxq, sizeof(rxq[0]) * nb_queues,
-                               CACHE_LINE_SIZE);
+                               RTE_CACHE_LINE_SIZE);
                if (rxq == NULL)
                        return -(ENOMEM);
 
@@ -284,6 +360,110 @@ rte_eth_dev_rx_queue_config(struct rte_eth_dev *dev, uint16_t nb_queues)
        return (0);
 }
 
+int
+rte_eth_dev_rx_queue_start(uint8_t port_id, uint16_t rx_queue_id)
+{
+       struct rte_eth_dev *dev;
+
+       /* This function is only safe when called from the primary process
+        * in a multi-process setup*/
+       PROC_PRIMARY_OR_ERR_RET(-E_RTE_SECONDARY);
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return -EINVAL;
+       }
+
+       dev = &rte_eth_devices[port_id];
+       if (rx_queue_id >= dev->data->nb_rx_queues) {
+               PMD_DEBUG_TRACE("Invalid RX queue_id=%d\n", rx_queue_id);
+               return -EINVAL;
+       }
+
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->rx_queue_start, -ENOTSUP);
+
+       return dev->dev_ops->rx_queue_start(dev, rx_queue_id);
+
+}
+
+int
+rte_eth_dev_rx_queue_stop(uint8_t port_id, uint16_t rx_queue_id)
+{
+       struct rte_eth_dev *dev;
+
+       /* This function is only safe when called from the primary process
+        * in a multi-process setup*/
+       PROC_PRIMARY_OR_ERR_RET(-E_RTE_SECONDARY);
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return -EINVAL;
+       }
+
+       dev = &rte_eth_devices[port_id];
+       if (rx_queue_id >= dev->data->nb_rx_queues) {
+               PMD_DEBUG_TRACE("Invalid RX queue_id=%d\n", rx_queue_id);
+               return -EINVAL;
+       }
+
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->rx_queue_stop, -ENOTSUP);
+
+       return dev->dev_ops->rx_queue_stop(dev, rx_queue_id);
+
+}
+
+int
+rte_eth_dev_tx_queue_start(uint8_t port_id, uint16_t tx_queue_id)
+{
+       struct rte_eth_dev *dev;
+
+       /* This function is only safe when called from the primary process
+        * in a multi-process setup*/
+       PROC_PRIMARY_OR_ERR_RET(-E_RTE_SECONDARY);
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return -EINVAL;
+       }
+
+       dev = &rte_eth_devices[port_id];
+       if (tx_queue_id >= dev->data->nb_tx_queues) {
+               PMD_DEBUG_TRACE("Invalid TX queue_id=%d\n", tx_queue_id);
+               return -EINVAL;
+       }
+
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->tx_queue_start, -ENOTSUP);
+
+       return dev->dev_ops->tx_queue_start(dev, tx_queue_id);
+
+}
+
+int
+rte_eth_dev_tx_queue_stop(uint8_t port_id, uint16_t tx_queue_id)
+{
+       struct rte_eth_dev *dev;
+
+       /* This function is only safe when called from the primary process
+        * in a multi-process setup*/
+       PROC_PRIMARY_OR_ERR_RET(-E_RTE_SECONDARY);
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return -EINVAL;
+       }
+
+       dev = &rte_eth_devices[port_id];
+       if (tx_queue_id >= dev->data->nb_tx_queues) {
+               PMD_DEBUG_TRACE("Invalid TX queue_id=%d\n", tx_queue_id);
+               return -EINVAL;
+       }
+
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->tx_queue_stop, -ENOTSUP);
+
+       return dev->dev_ops->tx_queue_stop(dev, tx_queue_id);
+
+}
+
 static int
 rte_eth_dev_tx_queue_config(struct rte_eth_dev *dev, uint16_t nb_queues)
 {
@@ -291,23 +471,23 @@ rte_eth_dev_tx_queue_config(struct rte_eth_dev *dev, uint16_t nb_queues)
        void **txq;
        unsigned i;
 
-       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->tx_queue_release, -ENOTSUP);
-
-       if (dev->data->tx_queues == NULL) {
+       if (dev->data->tx_queues == NULL) { /* first time configuration */
                dev->data->tx_queues = rte_zmalloc("ethdev->tx_queues",
                                sizeof(dev->data->tx_queues[0]) * nb_queues,
-                               CACHE_LINE_SIZE);
+                               RTE_CACHE_LINE_SIZE);
                if (dev->data->tx_queues == NULL) {
                        dev->data->nb_tx_queues = 0;
                        return -(ENOMEM);
                }
-       } else {
+       } else { /* re-configure */
+               FUNC_PTR_OR_ERR_RET(*dev->dev_ops->tx_queue_release, -ENOTSUP);
+
                txq = dev->data->tx_queues;
 
                for (i = nb_queues; i < old_nb_queues; i++)
                        (*dev->dev_ops->tx_queue_release)(txq[i]);
                txq = rte_realloc(txq, sizeof(txq[0]) * nb_queues,
-                               CACHE_LINE_SIZE);
+                               RTE_CACHE_LINE_SIZE);
                if (txq == NULL)
                        return -(ENOMEM);
 
@@ -322,6 +502,154 @@ rte_eth_dev_tx_queue_config(struct rte_eth_dev *dev, uint16_t nb_queues)
        return (0);
 }
 
+static int
+rte_eth_dev_check_mq_mode(uint8_t port_id, uint16_t nb_rx_q, uint16_t nb_tx_q,
+                     const struct rte_eth_conf *dev_conf)
+{
+       struct rte_eth_dev *dev = &rte_eth_devices[port_id];
+
+       if (RTE_ETH_DEV_SRIOV(dev).active != 0) {
+               /* check multi-queue mode */
+               if ((dev_conf->rxmode.mq_mode == ETH_MQ_RX_RSS) ||
+                   (dev_conf->rxmode.mq_mode == ETH_MQ_RX_DCB) ||
+                   (dev_conf->rxmode.mq_mode == ETH_MQ_RX_DCB_RSS) ||
+                   (dev_conf->txmode.mq_mode == ETH_MQ_TX_DCB)) {
+                       /* SRIOV only works in VMDq enable mode */
+                       PMD_DEBUG_TRACE("ethdev port_id=%" PRIu8
+                                       " SRIOV active, "
+                                       "wrong VMDQ mq_mode rx %u tx %u\n",
+                                       port_id,
+                                       dev_conf->rxmode.mq_mode,
+                                       dev_conf->txmode.mq_mode);
+                       return (-EINVAL);
+               }
+
+               switch (dev_conf->rxmode.mq_mode) {
+               case ETH_MQ_RX_VMDQ_RSS:
+               case ETH_MQ_RX_VMDQ_DCB:
+               case ETH_MQ_RX_VMDQ_DCB_RSS:
+                       /* DCB/RSS VMDQ in SRIOV mode, not implement yet */
+                       PMD_DEBUG_TRACE("ethdev port_id=%" PRIu8
+                                       " SRIOV active, "
+                                       "unsupported VMDQ mq_mode rx %u\n",
+                                       port_id, dev_conf->rxmode.mq_mode);
+                       return (-EINVAL);
+               default: /* ETH_MQ_RX_VMDQ_ONLY or ETH_MQ_RX_NONE */
+                       /* if nothing mq mode configure, use default scheme */
+                       dev->data->dev_conf.rxmode.mq_mode = ETH_MQ_RX_VMDQ_ONLY;
+                       if (RTE_ETH_DEV_SRIOV(dev).nb_q_per_pool > 1)
+                               RTE_ETH_DEV_SRIOV(dev).nb_q_per_pool = 1;
+                       break;
+               }
+
+               switch (dev_conf->txmode.mq_mode) {
+               case ETH_MQ_TX_VMDQ_DCB:
+                       /* DCB VMDQ in SRIOV mode, not implement yet */
+                       PMD_DEBUG_TRACE("ethdev port_id=%" PRIu8
+                                       " SRIOV active, "
+                                       "unsupported VMDQ mq_mode tx %u\n",
+                                       port_id, dev_conf->txmode.mq_mode);
+                       return (-EINVAL);
+               default: /* ETH_MQ_TX_VMDQ_ONLY or ETH_MQ_TX_NONE */
+                       /* if nothing mq mode configure, use default scheme */
+                       dev->data->dev_conf.txmode.mq_mode = ETH_MQ_TX_VMDQ_ONLY;
+                       if (RTE_ETH_DEV_SRIOV(dev).nb_q_per_pool > 1)
+                               RTE_ETH_DEV_SRIOV(dev).nb_q_per_pool = 1;
+                       break;
+               }
+
+               /* check valid queue number */
+               if ((nb_rx_q > RTE_ETH_DEV_SRIOV(dev).nb_q_per_pool) ||
+                   (nb_tx_q > RTE_ETH_DEV_SRIOV(dev).nb_q_per_pool)) {
+                       PMD_DEBUG_TRACE("ethdev port_id=%d SRIOV active, "
+                                   "queue number must less equal to %d\n",
+                                       port_id, RTE_ETH_DEV_SRIOV(dev).nb_q_per_pool);
+                       return (-EINVAL);
+               }
+       } else {
+               /* For vmdb+dcb mode check our configuration before we go further */
+               if (dev_conf->rxmode.mq_mode == ETH_MQ_RX_VMDQ_DCB) {
+                       const struct rte_eth_vmdq_dcb_conf *conf;
+
+                       if (nb_rx_q != ETH_VMDQ_DCB_NUM_QUEUES) {
+                               PMD_DEBUG_TRACE("ethdev port_id=%d VMDQ+DCB, nb_rx_q "
+                                               "!= %d\n",
+                                               port_id, ETH_VMDQ_DCB_NUM_QUEUES);
+                               return (-EINVAL);
+                       }
+                       conf = &(dev_conf->rx_adv_conf.vmdq_dcb_conf);
+                       if (! (conf->nb_queue_pools == ETH_16_POOLS ||
+                              conf->nb_queue_pools == ETH_32_POOLS)) {
+                               PMD_DEBUG_TRACE("ethdev port_id=%d VMDQ+DCB selected, "
+                                               "nb_queue_pools must be %d or %d\n",
+                                               port_id, ETH_16_POOLS, ETH_32_POOLS);
+                               return (-EINVAL);
+                       }
+               }
+               if (dev_conf->txmode.mq_mode == ETH_MQ_TX_VMDQ_DCB) {
+                       const struct rte_eth_vmdq_dcb_tx_conf *conf;
+
+                       if (nb_tx_q != ETH_VMDQ_DCB_NUM_QUEUES) {
+                               PMD_DEBUG_TRACE("ethdev port_id=%d VMDQ+DCB, nb_tx_q "
+                                               "!= %d\n",
+                                               port_id, ETH_VMDQ_DCB_NUM_QUEUES);
+                               return (-EINVAL);
+                       }
+                       conf = &(dev_conf->tx_adv_conf.vmdq_dcb_tx_conf);
+                       if (! (conf->nb_queue_pools == ETH_16_POOLS ||
+                              conf->nb_queue_pools == ETH_32_POOLS)) {
+                               PMD_DEBUG_TRACE("ethdev port_id=%d VMDQ+DCB selected, "
+                                               "nb_queue_pools != %d or nb_queue_pools "
+                                               "!= %d\n",
+                                               port_id, ETH_16_POOLS, ETH_32_POOLS);
+                               return (-EINVAL);
+                       }
+               }
+
+               /* For DCB mode check our configuration before we go further */
+               if (dev_conf->rxmode.mq_mode == ETH_MQ_RX_DCB) {
+                       const struct rte_eth_dcb_rx_conf *conf;
+
+                       if (nb_rx_q != ETH_DCB_NUM_QUEUES) {
+                               PMD_DEBUG_TRACE("ethdev port_id=%d DCB, nb_rx_q "
+                                               "!= %d\n",
+                                               port_id, ETH_DCB_NUM_QUEUES);
+                               return (-EINVAL);
+                       }
+                       conf = &(dev_conf->rx_adv_conf.dcb_rx_conf);
+                       if (! (conf->nb_tcs == ETH_4_TCS ||
+                              conf->nb_tcs == ETH_8_TCS)) {
+                               PMD_DEBUG_TRACE("ethdev port_id=%d DCB selected, "
+                                               "nb_tcs != %d or nb_tcs "
+                                               "!= %d\n",
+                                               port_id, ETH_4_TCS, ETH_8_TCS);
+                               return (-EINVAL);
+                       }
+               }
+
+               if (dev_conf->txmode.mq_mode == ETH_MQ_TX_DCB) {
+                       const struct rte_eth_dcb_tx_conf *conf;
+
+                       if (nb_tx_q != ETH_DCB_NUM_QUEUES) {
+                               PMD_DEBUG_TRACE("ethdev port_id=%d DCB, nb_tx_q "
+                                               "!= %d\n",
+                                               port_id, ETH_DCB_NUM_QUEUES);
+                               return (-EINVAL);
+                       }
+                       conf = &(dev_conf->tx_adv_conf.dcb_tx_conf);
+                       if (! (conf->nb_tcs == ETH_4_TCS ||
+                              conf->nb_tcs == ETH_8_TCS)) {
+                               PMD_DEBUG_TRACE("ethdev port_id=%d DCB selected, "
+                                               "nb_tcs != %d or nb_tcs "
+                                               "!= %d\n",
+                                               port_id, ETH_4_TCS, ETH_8_TCS);
+                               return (-EINVAL);
+                       }
+               }
+       }
+       return 0;
+}
+
 int
 rte_eth_dev_configure(uint8_t port_id, uint16_t nb_rx_q, uint16_t nb_tx_q,
                      const struct rte_eth_conf *dev_conf)
@@ -378,6 +706,20 @@ rte_eth_dev_configure(uint8_t port_id, uint16_t nb_rx_q, uint16_t nb_tx_q,
        /* Copy the dev_conf parameter into the dev structure */
        memcpy(&dev->data->dev_conf, dev_conf, sizeof(dev->data->dev_conf));
 
+       /*
+        * If link state interrupt is enabled, check that the
+        * device supports it.
+        */
+       if (dev_conf->intr_conf.lsc == 1) {
+               const struct rte_pci_driver *pci_drv = &dev->driver->pci_drv;
+
+               if (!(pci_drv->drv_flags & RTE_PCI_DRV_INTR_LSC)) {
+                       PMD_DEBUG_TRACE("driver %s does not support lsc\n",
+                                       pci_drv->name);
+                       return (-EINVAL);
+               }
+       }
+
        /*
         * If jumbo frames are enabled, check that the maximum RX packet
         * length is supported by the configured device.
@@ -400,88 +742,20 @@ rte_eth_dev_configure(uint8_t port_id, uint16_t nb_rx_q, uint16_t nb_tx_q,
                                (unsigned)ETHER_MIN_LEN);
                        return (-EINVAL);
                }
-       } else
-               /* Use default value */
-               dev->data->dev_conf.rxmode.max_rx_pkt_len = ETHER_MAX_LEN;
-
-       /* For vmdb+dcb mode check our configuration before we go further */
-       if (dev_conf->rxmode.mq_mode == ETH_VMDQ_DCB) {
-               const struct rte_eth_vmdq_dcb_conf *conf;
-
-               if (nb_rx_q != ETH_VMDQ_DCB_NUM_QUEUES) {
-                       PMD_DEBUG_TRACE("ethdev port_id=%d VMDQ+DCB, nb_rx_q "
-                                       "!= %d\n",
-                                       port_id, ETH_VMDQ_DCB_NUM_QUEUES);
-                       return (-EINVAL);
-               }
-               conf = &(dev_conf->rx_adv_conf.vmdq_dcb_conf);
-               if (! (conf->nb_queue_pools == ETH_16_POOLS ||
-                      conf->nb_queue_pools == ETH_32_POOLS)) {
-                   PMD_DEBUG_TRACE("ethdev port_id=%d VMDQ+DCB selected, "
-                                   "nb_queue_pools must be %d or %d\n",
-                                   port_id, ETH_16_POOLS, ETH_32_POOLS);
-                   return (-EINVAL);
-               }
-       }
-       if (dev_conf->txmode.mq_mode == ETH_VMDQ_DCB_TX) {
-               const struct rte_eth_vmdq_dcb_tx_conf *conf;
-
-               if (nb_tx_q != ETH_VMDQ_DCB_NUM_QUEUES) {
-                       PMD_DEBUG_TRACE("ethdev port_id=%d VMDQ+DCB, nb_tx_q "
-                                       "!= %d\n",
-                                       port_id, ETH_VMDQ_DCB_NUM_QUEUES);
-                       return (-EINVAL);
-               }
-               conf = &(dev_conf->tx_adv_conf.vmdq_dcb_tx_conf);
-               if (! (conf->nb_queue_pools == ETH_16_POOLS ||
-                      conf->nb_queue_pools == ETH_32_POOLS)) {
-                       PMD_DEBUG_TRACE("ethdev port_id=%d VMDQ+DCB selected, "
-                                   "nb_queue_pools != %d or nb_queue_pools "
-                                   "!= %d\n",
-                                   port_id, ETH_16_POOLS, ETH_32_POOLS);
-                       return (-EINVAL);
-               }
-       }
-       
-       /* For DCB mode check our configuration before we go further */
-       if (dev_conf->rxmode.mq_mode == ETH_DCB_RX) {
-               const struct rte_eth_dcb_rx_conf *conf;
-
-               if (nb_rx_q != ETH_DCB_NUM_QUEUES) {
-                       PMD_DEBUG_TRACE("ethdev port_id=%d DCB, nb_rx_q "
-                                       "!= %d\n",
-                                       port_id, ETH_DCB_NUM_QUEUES);
-                       return (-EINVAL);
-               }
-               conf = &(dev_conf->rx_adv_conf.dcb_rx_conf);
-               if (! (conf->nb_tcs == ETH_4_TCS ||
-                      conf->nb_tcs == ETH_8_TCS)) {
-                       PMD_DEBUG_TRACE("ethdev port_id=%d DCB selected, "
-                                   "nb_tcs != %d or nb_tcs "
-                                   "!= %d\n",
-                                   port_id, ETH_4_TCS, ETH_8_TCS);
-                       return (-EINVAL);
-               }
+       } else {
+               if (dev_conf->rxmode.max_rx_pkt_len < ETHER_MIN_LEN ||
+                       dev_conf->rxmode.max_rx_pkt_len > ETHER_MAX_LEN)
+                       /* Use default value */
+                       dev->data->dev_conf.rxmode.max_rx_pkt_len =
+                                                       ETHER_MAX_LEN;
        }
 
-       if (dev_conf->txmode.mq_mode == ETH_DCB_TX) {
-               const struct rte_eth_dcb_tx_conf *conf;
-
-               if (nb_tx_q != ETH_DCB_NUM_QUEUES) {
-                       PMD_DEBUG_TRACE("ethdev port_id=%d DCB, nb_tx_q "
-                                       "!= %d\n",
-                                       port_id, ETH_DCB_NUM_QUEUES);
-                       return (-EINVAL);
-               }
-               conf = &(dev_conf->tx_adv_conf.dcb_tx_conf);
-               if (! (conf->nb_tcs == ETH_4_TCS ||
-                      conf->nb_tcs == ETH_8_TCS)) {
-                       PMD_DEBUG_TRACE("ethdev port_id=%d DCB selected, "
-                                   "nb_tcs != %d or nb_tcs "
-                                   "!= %d\n",
-                                   port_id, ETH_4_TCS, ETH_8_TCS);
-                       return (-EINVAL);
-               }
+       /* multipe queue mode checking */
+       diag = rte_eth_dev_check_mq_mode(port_id, nb_rx_q, nb_tx_q, dev_conf);
+       if (diag != 0) {
+               PMD_DEBUG_TRACE("port%d rte_eth_dev_check_mq_mode = %d\n",
+                               port_id, diag);
+               return diag;
        }
 
        /*
@@ -521,11 +795,15 @@ rte_eth_dev_config_restore(uint8_t port_id)
        struct rte_eth_dev_info dev_info;
        struct ether_addr addr;
        uint16_t i;
+       uint32_t pool = 0;
 
        dev = &rte_eth_devices[port_id];
 
        rte_eth_dev_info_get(port_id, &dev_info);
 
+       if (RTE_ETH_DEV_SRIOV(dev).active)
+               pool = RTE_ETH_DEV_SRIOV(dev).def_vmdq_idx;
+
        /* replay MAC address configuration */
        for (i = 0; i < dev_info.max_mac_addrs; i++) {
                addr = dev->data->mac_addrs[i];
@@ -535,8 +813,9 @@ rte_eth_dev_config_restore(uint8_t port_id)
                        continue;
 
                /* add address to the hardware */
-               if  (*dev->dev_ops->mac_addr_add)
-                       (*dev->dev_ops->mac_addr_add)(dev, &addr, i, 0);
+               if  (*dev->dev_ops->mac_addr_add &&
+                       dev->data->mac_pool_sel[i] & (1ULL << pool))
+                       (*dev->dev_ops->mac_addr_add)(dev, &addr, i, pool);
                else {
                        PMD_DEBUG_TRACE("port %d: MAC address array not supported\n",
                                        port_id);
@@ -569,12 +848,20 @@ rte_eth_dev_start(uint8_t port_id)
        PROC_PRIMARY_OR_ERR_RET(-E_RTE_SECONDARY);
 
        if (port_id >= nb_ports) {
-               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               PMD_DEBUG_TRACE("Invalid port_id=%" PRIu8 "\n", port_id);
                return (-EINVAL);
        }
        dev = &rte_eth_devices[port_id];
 
        FUNC_PTR_OR_ERR_RET(*dev->dev_ops->dev_start, -ENOTSUP);
+
+       if (dev->data->dev_started != 0) {
+               PMD_DEBUG_TRACE("Device with port_id=%" PRIu8
+                       " already started\n",
+                       port_id);
+               return (0);
+       }
+
        diag = (*dev->dev_ops->dev_start)(dev);
        if (diag == 0)
                dev->data->dev_started = 1;
@@ -596,16 +883,62 @@ rte_eth_dev_stop(uint8_t port_id)
        PROC_PRIMARY_OR_RET();
 
        if (port_id >= nb_ports) {
-               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               PMD_DEBUG_TRACE("Invalid port_id=%" PRIu8 "\n", port_id);
                return;
        }
        dev = &rte_eth_devices[port_id];
 
        FUNC_PTR_OR_RET(*dev->dev_ops->dev_stop);
+
+       if (dev->data->dev_started == 0) {
+               PMD_DEBUG_TRACE("Device with port_id=%" PRIu8
+                       " already stopped\n",
+                       port_id);
+               return;
+       }
+
        dev->data->dev_started = 0;
        (*dev->dev_ops->dev_stop)(dev);
 }
 
+int
+rte_eth_dev_set_link_up(uint8_t port_id)
+{
+       struct rte_eth_dev *dev;
+
+       /* This function is only safe when called from the primary process
+        * in a multi-process setup*/
+       PROC_PRIMARY_OR_ERR_RET(-E_RTE_SECONDARY);
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return -EINVAL;
+       }
+       dev = &rte_eth_devices[port_id];
+
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->dev_set_link_up, -ENOTSUP);
+       return (*dev->dev_ops->dev_set_link_up)(dev);
+}
+
+int
+rte_eth_dev_set_link_down(uint8_t port_id)
+{
+       struct rte_eth_dev *dev;
+
+       /* This function is only safe when called from the primary process
+        * in a multi-process setup*/
+       PROC_PRIMARY_OR_ERR_RET(-E_RTE_SECONDARY);
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return -EINVAL;
+       }
+       dev = &rte_eth_devices[port_id];
+
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->dev_set_link_down, -ENOTSUP);
+       return (*dev->dev_ops->dev_set_link_down)(dev);
+}
+
 void
 rte_eth_dev_close(uint8_t port_id)
 {
@@ -633,6 +966,8 @@ rte_eth_rx_queue_setup(uint8_t port_id, uint16_t rx_queue_id,
                       const struct rte_eth_rxconf *rx_conf,
                       struct rte_mempool *mp)
 {
+       int ret;
+       uint32_t mbp_buf_size;
        struct rte_eth_dev *dev;
        struct rte_pktmbuf_pool_private *mbp_priv;
        struct rte_eth_dev_info dev_info;
@@ -665,22 +1000,22 @@ rte_eth_rx_queue_setup(uint8_t port_id, uint16_t rx_queue_id,
         * This value must be provided in the private data of the memory pool.
         * First check that the memory pool has a valid private data.
         */
-       (*dev->dev_ops->dev_infos_get)(dev, &dev_info);
+       rte_eth_dev_info_get(port_id, &dev_info);
        if (mp->private_data_size < sizeof(struct rte_pktmbuf_pool_private)) {
                PMD_DEBUG_TRACE("%s private_data_size %d < %d\n",
                                mp->name, (int) mp->private_data_size,
                                (int) sizeof(struct rte_pktmbuf_pool_private));
                return (-ENOSPC);
        }
-       mbp_priv = (struct rte_pktmbuf_pool_private *)
-               ((char *)mp + sizeof(struct rte_mempool));
-       if ((uint32_t) (mbp_priv->mbuf_data_room_size - RTE_PKTMBUF_HEADROOM) <
-           dev_info.min_rx_bufsize) {
+       mbp_priv = rte_mempool_get_priv(mp);
+       mbp_buf_size = mbp_priv->mbuf_data_room_size;
+
+       if ((mbp_buf_size - RTE_PKTMBUF_HEADROOM) < dev_info.min_rx_bufsize) {
                PMD_DEBUG_TRACE("%s mbuf_data_room_size %d < %d "
                                "(RTE_PKTMBUF_HEADROOM=%d + min_rx_bufsize(dev)"
                                "=%d)\n",
                                mp->name,
-                               (int)mbp_priv->mbuf_data_room_size,
+                               (int)mbp_buf_size,
                                (int)(RTE_PKTMBUF_HEADROOM +
                                      dev_info.min_rx_bufsize),
                                (int)RTE_PKTMBUF_HEADROOM,
@@ -688,8 +1023,18 @@ rte_eth_rx_queue_setup(uint8_t port_id, uint16_t rx_queue_id,
                return (-EINVAL);
        }
 
-       return (*dev->dev_ops->rx_queue_setup)(dev, rx_queue_id, nb_rx_desc,
-                                              socket_id, rx_conf, mp);
+       if (rx_conf == NULL)
+               rx_conf = &dev_info.default_rxconf;
+
+       ret = (*dev->dev_ops->rx_queue_setup)(dev, rx_queue_id, nb_rx_desc,
+                                             socket_id, rx_conf, mp);
+       if (!ret) {
+               if (!dev->data->min_rx_buf_size ||
+                   dev->data->min_rx_buf_size > mbp_buf_size)
+                       dev->data->min_rx_buf_size = mbp_buf_size;
+       }
+
+       return ret;
 }
 
 int
@@ -698,6 +1043,7 @@ rte_eth_tx_queue_setup(uint8_t port_id, uint16_t tx_queue_id,
                       const struct rte_eth_txconf *tx_conf)
 {
        struct rte_eth_dev *dev;
+       struct rte_eth_dev_info dev_info;
 
        /* This function is only safe when called from the primary process
         * in a multi-process setup*/
@@ -719,7 +1065,14 @@ rte_eth_tx_queue_setup(uint8_t port_id, uint16_t tx_queue_id,
                return -EBUSY;
        }
 
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->dev_infos_get, -ENOTSUP);
        FUNC_PTR_OR_ERR_RET(*dev->dev_ops->tx_queue_setup, -ENOTSUP);
+
+       rte_eth_dev_info_get(port_id, &dev_info);
+
+       if (tx_conf == NULL)
+               tx_conf = &dev_info.default_txconf;
+
        return (*dev->dev_ops->tx_queue_setup)(dev, tx_queue_id, nb_tx_desc,
                                               socket_id, tx_conf);
 }
@@ -840,11 +1193,11 @@ rte_eth_link_get(uint8_t port_id, struct rte_eth_link *eth_link)
                return;
        }
        dev = &rte_eth_devices[port_id];
-       FUNC_PTR_OR_RET(*dev->dev_ops->link_update);
 
        if (dev->data->dev_conf.intr_conf.lsc != 0)
                rte_eth_dev_atomic_read_link_status(dev, eth_link);
        else {
+               FUNC_PTR_OR_RET(*dev->dev_ops->link_update);
                (*dev->dev_ops->link_update)(dev, 1);
                *eth_link = dev->data->dev_link;
        }
@@ -860,11 +1213,11 @@ rte_eth_link_get_nowait(uint8_t port_id, struct rte_eth_link *eth_link)
                return;
        }
        dev = &rte_eth_devices[port_id];
-       FUNC_PTR_OR_RET(*dev->dev_ops->link_update);
 
        if (dev->data->dev_conf.intr_conf.lsc != 0)
                rte_eth_dev_atomic_read_link_status(dev, eth_link);
        else {
+               FUNC_PTR_OR_RET(*dev->dev_ops->link_update);
                (*dev->dev_ops->link_update)(dev, 0);
                *eth_link = dev->data->dev_link;
        }
@@ -880,6 +1233,7 @@ rte_eth_stats_get(uint8_t port_id, struct rte_eth_stats *stats)
                return;
        }
        dev = &rte_eth_devices[port_id];
+       memset(stats, 0, sizeof(*stats));
 
        FUNC_PTR_OR_RET(*dev->dev_ops->stats_get);
        (*dev->dev_ops->stats_get)(dev, stats);
@@ -901,26 +1255,121 @@ rte_eth_stats_reset(uint8_t port_id)
        (*dev->dev_ops->stats_reset)(dev);
 }
 
-
-static int
-set_queue_stats_mapping(uint8_t port_id, uint16_t queue_id, uint8_t stat_idx,
-               uint8_t is_rx)
+/* retrieve ethdev extended statistics */
+int
+rte_eth_xstats_get(uint8_t port_id, struct rte_eth_xstats *xstats,
+       unsigned n)
 {
+       struct rte_eth_stats eth_stats;
        struct rte_eth_dev *dev;
+       unsigned count, i, q;
+       uint64_t val;
+       char *stats_ptr;
 
        if (port_id >= nb_ports) {
                PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
-               return -ENODEV;
+               return -1;
        }
        dev = &rte_eth_devices[port_id];
 
-       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->queue_stats_mapping_set, -ENOTSUP);
-       return (*dev->dev_ops->queue_stats_mapping_set)
-                       (dev, queue_id, stat_idx, is_rx);
+       /* implemented by the driver */
+       if (dev->dev_ops->xstats_get != NULL)
+               return (*dev->dev_ops->xstats_get)(dev, xstats, n);
+
+       /* else, return generic statistics */
+       count = RTE_NB_STATS;
+       count += dev->data->nb_rx_queues * RTE_NB_RXQ_STATS;
+       count += dev->data->nb_tx_queues * RTE_NB_TXQ_STATS;
+       if (n < count)
+               return count;
+
+       /* now fill the xstats structure */
+
+       count = 0;
+       memset(&eth_stats, 0, sizeof(eth_stats));
+       rte_eth_stats_get(port_id, &eth_stats);
+
+       /* global stats */
+       for (i = 0; i < RTE_NB_STATS; i++) {
+               stats_ptr = (char *)&eth_stats + rte_stats_strings[i].offset;
+               val = *(uint64_t *)stats_ptr;
+               snprintf(xstats[count].name, sizeof(xstats[count].name),
+                       "%s", rte_stats_strings[i].name);
+               xstats[count++].value = val;
+       }
+
+       /* per-rxq stats */
+       for (q = 0; q < dev->data->nb_rx_queues; q++) {
+               for (i = 0; i < RTE_NB_RXQ_STATS; i++) {
+                       stats_ptr = (char *)&eth_stats;
+                       stats_ptr += rte_rxq_stats_strings[i].offset;
+                       stats_ptr += q * sizeof(uint64_t);
+                       val = *(uint64_t *)stats_ptr;
+                       snprintf(xstats[count].name, sizeof(xstats[count].name),
+                               "rx_queue_%u_%s", q,
+                               rte_rxq_stats_strings[i].name);
+                       xstats[count++].value = val;
+               }
+       }
+
+       /* per-txq stats */
+       for (q = 0; q < dev->data->nb_tx_queues; q++) {
+               for (i = 0; i < RTE_NB_TXQ_STATS; i++) {
+                       stats_ptr = (char *)&eth_stats;
+                       stats_ptr += rte_txq_stats_strings[i].offset;
+                       stats_ptr += q * sizeof(uint64_t);
+                       val = *(uint64_t *)stats_ptr;
+                       snprintf(xstats[count].name, sizeof(xstats[count].name),
+                               "tx_queue_%u_%s", q,
+                               rte_txq_stats_strings[i].name);
+                       xstats[count++].value = val;
+               }
+       }
+
+       return count;
 }
 
+/* reset ethdev extended statistics */
+void
+rte_eth_xstats_reset(uint8_t port_id)
+{
+       struct rte_eth_dev *dev;
 
-int
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return;
+       }
+       dev = &rte_eth_devices[port_id];
+
+       /* implemented by the driver */
+       if (dev->dev_ops->xstats_reset != NULL) {
+               (*dev->dev_ops->xstats_reset)(dev);
+               return;
+       }
+
+       /* fallback to default */
+       rte_eth_stats_reset(port_id);
+}
+
+static int
+set_queue_stats_mapping(uint8_t port_id, uint16_t queue_id, uint8_t stat_idx,
+               uint8_t is_rx)
+{
+       struct rte_eth_dev *dev;
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return -ENODEV;
+       }
+       dev = &rte_eth_devices[port_id];
+
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->queue_stats_mapping_set, -ENOTSUP);
+       return (*dev->dev_ops->queue_stats_mapping_set)
+                       (dev, queue_id, stat_idx, is_rx);
+}
+
+
+int
 rte_eth_dev_set_tx_queue_stats_mapping(uint8_t port_id, uint16_t tx_queue_id,
                uint8_t stat_idx)
 {
@@ -949,10 +1398,13 @@ rte_eth_dev_info_get(uint8_t port_id, struct rte_eth_dev_info *dev_info)
        }
        dev = &rte_eth_devices[port_id];
 
+       memset(dev_info, 0, sizeof(struct rte_eth_dev_info));
+
        FUNC_PTR_OR_RET(*dev->dev_ops->dev_infos_get);
        (*dev->dev_ops->dev_infos_get)(dev, dev_info);
        dev_info->pci_dev = dev->pci_dev;
-       dev_info->driver_name = dev->driver->pci_drv.name;
+       if (dev->driver)
+               dev_info->driver_name = dev->driver->pci_drv.name;
 }
 
 void
@@ -968,6 +1420,43 @@ rte_eth_macaddr_get(uint8_t port_id, struct ether_addr *mac_addr)
        ether_addr_copy(&dev->data->mac_addrs[0], mac_addr);
 }
 
+
+int
+rte_eth_dev_get_mtu(uint8_t port_id, uint16_t *mtu)
+{
+       struct rte_eth_dev *dev;
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return (-ENODEV);
+       }
+
+       dev = &rte_eth_devices[port_id];
+       *mtu = dev->data->mtu;
+       return 0;
+}
+
+int
+rte_eth_dev_set_mtu(uint8_t port_id, uint16_t mtu)
+{
+       int ret;
+       struct rte_eth_dev *dev;
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return (-ENODEV);
+       }
+
+       dev = &rte_eth_devices[port_id];
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->mtu_set, -ENOTSUP);
+
+       ret = (*dev->dev_ops->mtu_set)(dev, mtu);
+       if (!ret)
+               dev->data->mtu = mtu;
+
+       return ret;
+}
+
 int
 rte_eth_dev_vlan_filter(uint8_t port_id, uint16_t vlan_id, int on)
 {
@@ -1039,7 +1528,7 @@ rte_eth_dev_set_vlan_offload(uint8_t port_id, int offload_mask)
        int ret = 0;
        int mask = 0;
        int cur, org = 0;
-       
+
        if (port_id >= nb_ports) {
                PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
                return (-ENODEV);
@@ -1054,7 +1543,7 @@ rte_eth_dev_set_vlan_offload(uint8_t port_id, int offload_mask)
                dev->data->dev_conf.rxmode.hw_vlan_strip = (uint8_t)cur;
                mask |= ETH_VLAN_STRIP_MASK;
        }
-       
+
        cur = !!(offload_mask & ETH_VLAN_FILTER_OFFLOAD);
        org = !!(dev->data->dev_conf.rxmode.hw_vlan_filter);
        if (cur != org){
@@ -1072,7 +1561,7 @@ rte_eth_dev_set_vlan_offload(uint8_t port_id, int offload_mask)
        /*no change*/
        if(mask == 0)
                return ret;
-       
+
        FUNC_PTR_OR_ERR_RET(*dev->dev_ops->vlan_offload_set, -ENOTSUP);
        (*dev->dev_ops->vlan_offload_set)(dev, mask);
 
@@ -1104,6 +1593,21 @@ rte_eth_dev_get_vlan_offload(uint8_t port_id)
        return ret;
 }
 
+int
+rte_eth_dev_set_vlan_pvid(uint8_t port_id, uint16_t pvid, int on)
+{
+       struct rte_eth_dev *dev;
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return (-ENODEV);
+       }
+       dev = &rte_eth_devices[port_id];
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->vlan_pvid_set, -ENOTSUP);
+       (*dev->dev_ops->vlan_pvid_set)(dev, pvid, on);
+
+       return 0;
+}
 
 int
 rte_eth_dev_fdir_add_signature_filter(uint8_t port_id,
@@ -1365,6 +1869,22 @@ rte_eth_dev_fdir_set_masks(uint8_t port_id, struct rte_fdir_masks *fdir_mask)
        return (*dev->dev_ops->fdir_set_masks)(dev, fdir_mask);
 }
 
+int
+rte_eth_dev_flow_ctrl_get(uint8_t port_id, struct rte_eth_fc_conf *fc_conf)
+{
+       struct rte_eth_dev *dev;
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return (-ENODEV);
+       }
+
+       dev = &rte_eth_devices[port_id];
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->flow_ctrl_get, -ENOTSUP);
+       memset(fc_conf, 0, sizeof(*fc_conf));
+       return (*dev->dev_ops->flow_ctrl_get)(dev, fc_conf);
+}
+
 int
 rte_eth_dev_flow_ctrl_set(uint8_t port_id, struct rte_eth_fc_conf *fc_conf)
 {
@@ -1407,6 +1927,202 @@ rte_eth_dev_priority_flow_ctrl_set(uint8_t port_id, struct rte_eth_pfc_conf *pfc
        return (-ENOTSUP);
 }
 
+static inline int
+rte_eth_check_reta_mask(struct rte_eth_rss_reta_entry64 *reta_conf,
+                       uint16_t reta_size)
+{
+       uint16_t i, num;
+
+       if (!reta_conf)
+               return -EINVAL;
+
+       if (reta_size != RTE_ALIGN(reta_size, RTE_RETA_GROUP_SIZE)) {
+               PMD_DEBUG_TRACE("Invalid reta size, should be %u aligned\n",
+                                                       RTE_RETA_GROUP_SIZE);
+               return -EINVAL;
+       }
+
+       num = reta_size / RTE_RETA_GROUP_SIZE;
+       for (i = 0; i < num; i++) {
+               if (reta_conf[i].mask)
+                       return 0;
+       }
+
+       return -EINVAL;
+}
+
+static inline int
+rte_eth_check_reta_entry(struct rte_eth_rss_reta_entry64 *reta_conf,
+                        uint16_t reta_size,
+                        uint8_t max_rxq)
+{
+       uint16_t i, idx, shift;
+
+       if (!reta_conf)
+               return -EINVAL;
+
+       if (max_rxq == 0) {
+               PMD_DEBUG_TRACE("No receive queue is available\n");
+               return -EINVAL;
+       }
+
+       for (i = 0; i < reta_size; i++) {
+               idx = i / RTE_RETA_GROUP_SIZE;
+               shift = i % RTE_RETA_GROUP_SIZE;
+               if ((reta_conf[idx].mask & (1ULL << shift)) &&
+                       (reta_conf[idx].reta[shift] >= max_rxq)) {
+                       PMD_DEBUG_TRACE("reta_conf[%u]->reta[%u]: %u exceeds "
+                               "the maximum rxq index: %u\n", idx, shift,
+                               reta_conf[idx].reta[shift], max_rxq);
+                       return -EINVAL;
+               }
+       }
+
+       return 0;
+}
+
+int
+rte_eth_dev_rss_reta_update(uint8_t port_id,
+                           struct rte_eth_rss_reta_entry64 *reta_conf,
+                           uint16_t reta_size)
+{
+       struct rte_eth_dev *dev;
+       int ret;
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return -ENODEV;
+       }
+
+       /* Check mask bits */
+       ret = rte_eth_check_reta_mask(reta_conf, reta_size);
+       if (ret < 0)
+               return ret;
+
+       dev = &rte_eth_devices[port_id];
+
+       /* Check entry value */
+       ret = rte_eth_check_reta_entry(reta_conf, reta_size,
+                               dev->data->nb_rx_queues);
+       if (ret < 0)
+               return ret;
+
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->reta_update, -ENOTSUP);
+       return (*dev->dev_ops->reta_update)(dev, reta_conf, reta_size);
+}
+
+int
+rte_eth_dev_rss_reta_query(uint8_t port_id,
+                          struct rte_eth_rss_reta_entry64 *reta_conf,
+                          uint16_t reta_size)
+{
+       struct rte_eth_dev *dev;
+       int ret;
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return -ENODEV;
+       }
+
+       /* Check mask bits */
+       ret = rte_eth_check_reta_mask(reta_conf, reta_size);
+       if (ret < 0)
+               return ret;
+
+       dev = &rte_eth_devices[port_id];
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->reta_query, -ENOTSUP);
+       return (*dev->dev_ops->reta_query)(dev, reta_conf, reta_size);
+}
+
+int
+rte_eth_dev_rss_hash_update(uint8_t port_id, struct rte_eth_rss_conf *rss_conf)
+{
+       struct rte_eth_dev *dev;
+       uint16_t rss_hash_protos;
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return (-ENODEV);
+       }
+       rss_hash_protos = rss_conf->rss_hf;
+       if ((rss_hash_protos != 0) &&
+           ((rss_hash_protos & ETH_RSS_PROTO_MASK) == 0)) {
+               PMD_DEBUG_TRACE("Invalid rss_hash_protos=0x%x\n",
+                               rss_hash_protos);
+               return (-EINVAL);
+       }
+       dev = &rte_eth_devices[port_id];
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->rss_hash_update, -ENOTSUP);
+       return (*dev->dev_ops->rss_hash_update)(dev, rss_conf);
+}
+
+int
+rte_eth_dev_rss_hash_conf_get(uint8_t port_id,
+                             struct rte_eth_rss_conf *rss_conf)
+{
+       struct rte_eth_dev *dev;
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return (-ENODEV);
+       }
+       dev = &rte_eth_devices[port_id];
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->rss_hash_conf_get, -ENOTSUP);
+       return (*dev->dev_ops->rss_hash_conf_get)(dev, rss_conf);
+}
+
+int
+rte_eth_dev_udp_tunnel_add(uint8_t port_id,
+                          struct rte_eth_udp_tunnel *udp_tunnel)
+{
+       struct rte_eth_dev *dev;
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return -ENODEV;
+       }
+
+       if (udp_tunnel == NULL) {
+               PMD_DEBUG_TRACE("Invalid udp_tunnel parameter\n");
+               return -EINVAL;
+       }
+
+       if (udp_tunnel->prot_type >= RTE_TUNNEL_TYPE_MAX) {
+               PMD_DEBUG_TRACE("Invalid tunnel type\n");
+               return -EINVAL;
+       }
+
+       dev = &rte_eth_devices[port_id];
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->udp_tunnel_add, -ENOTSUP);
+       return (*dev->dev_ops->udp_tunnel_add)(dev, udp_tunnel);
+}
+
+int
+rte_eth_dev_udp_tunnel_delete(uint8_t port_id,
+                             struct rte_eth_udp_tunnel *udp_tunnel)
+{
+       struct rte_eth_dev *dev;
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return -ENODEV;
+       }
+       dev = &rte_eth_devices[port_id];
+
+       if (udp_tunnel == NULL) {
+               PMD_DEBUG_TRACE("Invalid udp_tunnel parametr\n");
+               return -EINVAL;
+       }
+
+       if (udp_tunnel->prot_type >= RTE_TUNNEL_TYPE_MAX) {
+               PMD_DEBUG_TRACE("Invalid tunnel type\n");
+               return -EINVAL;
+       }
+
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->udp_tunnel_del, -ENOTSUP);
+       return (*dev->dev_ops->udp_tunnel_del)(dev, udp_tunnel);
+}
+
 int
 rte_eth_led_on(uint8_t port_id)
 {
@@ -1461,10 +2177,11 @@ static struct ether_addr null_mac_addr = {{0, 0, 0, 0, 0, 0}};
 
 int
 rte_eth_dev_mac_addr_add(uint8_t port_id, struct ether_addr *addr,
-               uint32_t pool)
+                       uint32_t pool)
 {
        struct rte_eth_dev *dev;
        int index;
+       uint64_t pool_mask;
 
        if (port_id >= nb_ports) {
                PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
@@ -1474,19 +2191,29 @@ rte_eth_dev_mac_addr_add(uint8_t port_id, struct ether_addr *addr,
        FUNC_PTR_OR_ERR_RET(*dev->dev_ops->mac_addr_add, -ENOTSUP);
 
        if (is_zero_ether_addr(addr)) {
-               PMD_DEBUG_TRACE("port %d: Cannot add NULL MAC address\n", port_id);
+               PMD_DEBUG_TRACE("port %d: Cannot add NULL MAC address\n",
+                       port_id);
+               return (-EINVAL);
+       }
+       if (pool >= ETH_64_POOLS) {
+               PMD_DEBUG_TRACE("pool id must be 0-%d\n",ETH_64_POOLS - 1);
                return (-EINVAL);
        }
 
-       /* Check if it's already there, and do nothing */
        index = get_mac_addr_index(port_id, addr);
-       if (index >= 0)
-               return 0;
-
-       index = get_mac_addr_index(port_id, &null_mac_addr);
        if (index < 0) {
-               PMD_DEBUG_TRACE("port %d: MAC address array full\n", port_id);
-               return (-ENOSPC);
+               index = get_mac_addr_index(port_id, &null_mac_addr);
+               if (index < 0) {
+                       PMD_DEBUG_TRACE("port %d: MAC address array full\n",
+                               port_id);
+                       return (-ENOSPC);
+               }
+       } else {
+               pool_mask = dev->data->mac_pool_sel[index];
+
+               /* Check if both MAC address and pool is alread there, and do nothing */
+               if (pool_mask & (1ULL << pool))
+                       return 0;
        }
 
        /* Update NIC */
@@ -1495,6 +2222,9 @@ rte_eth_dev_mac_addr_add(uint8_t port_id, struct ether_addr *addr,
        /* Update address in NIC data structure */
        ether_addr_copy(addr, &dev->data->mac_addrs[index]);
 
+       /* Update pool bitmap in NIC data structure */
+       dev->data->mac_pool_sel[index] |= (1ULL << pool);
+
        return 0;
 }
 
@@ -1524,91 +2254,468 @@ rte_eth_dev_mac_addr_remove(uint8_t port_id, struct ether_addr *addr)
        /* Update address in NIC data structure */
        ether_addr_copy(&null_mac_addr, &dev->data->mac_addrs[index]);
 
+       /* reset pool bitmap */
+       dev->data->mac_pool_sel[index] = 0;
+
        return 0;
 }
 
-#ifdef RTE_LIBRTE_ETHDEV_DEBUG
-uint16_t
-rte_eth_rx_burst(uint8_t port_id, uint16_t queue_id,
-                struct rte_mbuf **rx_pkts, uint16_t nb_pkts)
+int
+rte_eth_dev_set_vf_rxmode(uint8_t port_id,  uint16_t vf,
+                               uint16_t rx_mode, uint8_t on)
 {
+       uint16_t num_vfs;
        struct rte_eth_dev *dev;
+       struct rte_eth_dev_info dev_info;
 
        if (port_id >= nb_ports) {
-               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
-               return 0;
+               PMD_DEBUG_TRACE("set VF RX mode:Invalid port_id=%d\n",
+                               port_id);
+               return (-ENODEV);
        }
+
        dev = &rte_eth_devices[port_id];
-       FUNC_PTR_OR_ERR_RET(*dev->rx_pkt_burst, -ENOTSUP);
-       if (queue_id >= dev->data->nb_rx_queues) {
-               PMD_DEBUG_TRACE("Invalid RX queue_id=%d\n", queue_id);
-               return 0;
+       rte_eth_dev_info_get(port_id, &dev_info);
+
+       num_vfs = dev_info.max_vfs;
+       if (vf > num_vfs)
+       {
+               PMD_DEBUG_TRACE("set VF RX mode:invalid VF id %d\n", vf);
+               return (-EINVAL);
        }
-       return (*dev->rx_pkt_burst)(dev->data->rx_queues[queue_id],
-                                               rx_pkts, nb_pkts);
+       if (rx_mode == 0)
+       {
+               PMD_DEBUG_TRACE("set VF RX mode:mode mask ca not be zero\n");
+               return (-EINVAL);
+       }
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->set_vf_rx_mode, -ENOTSUP);
+       return (*dev->dev_ops->set_vf_rx_mode)(dev, vf, rx_mode, on);
 }
 
-uint16_t
-rte_eth_tx_burst(uint8_t port_id, uint16_t queue_id,
-                struct rte_mbuf **tx_pkts, uint16_t nb_pkts)
+/*
+ * Returns index into MAC address array of addr. Use 00:00:00:00:00:00 to find
+ * an empty spot.
+ */
+static inline int
+get_hash_mac_addr_index(uint8_t port_id, struct ether_addr *addr)
 {
-       struct rte_eth_dev *dev;
+       struct rte_eth_dev_info dev_info;
+       struct rte_eth_dev *dev = &rte_eth_devices[port_id];
+       unsigned i;
 
-       if (port_id >= nb_ports) {
-               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
-               return 0;
-       }
-       dev = &rte_eth_devices[port_id];
+       rte_eth_dev_info_get(port_id, &dev_info);
+       if (!dev->data->hash_mac_addrs)
+               return -1;
 
-       FUNC_PTR_OR_ERR_RET(*dev->tx_pkt_burst, -ENOTSUP);
-       if (queue_id >= dev->data->nb_tx_queues) {
-               PMD_DEBUG_TRACE("Invalid TX queue_id=%d\n", queue_id);
-               return 0;
-       }
-       return (*dev->tx_pkt_burst)(dev->data->tx_queues[queue_id],
-                                               tx_pkts, nb_pkts);
+       for (i = 0; i < dev_info.max_hash_mac_addrs; i++)
+               if (memcmp(addr, &dev->data->hash_mac_addrs[i],
+                       ETHER_ADDR_LEN) == 0)
+                       return i;
+
+       return -1;
 }
-#endif
+
+int
+rte_eth_dev_uc_hash_table_set(uint8_t port_id, struct ether_addr *addr,
+                               uint8_t on)
+{
+       int index;
+       int ret;
+       struct rte_eth_dev *dev;
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("unicast hash setting:Invalid port_id=%d\n",
+                       port_id);
+               return (-ENODEV);
+       }
+
+       dev = &rte_eth_devices[port_id];
+       if (is_zero_ether_addr(addr)) {
+               PMD_DEBUG_TRACE("port %d: Cannot add NULL MAC address\n",
+                       port_id);
+               return (-EINVAL);
+       }
+
+       index = get_hash_mac_addr_index(port_id, addr);
+       /* Check if it's already there, and do nothing */
+       if ((index >= 0) && (on))
+               return 0;
+
+       if (index < 0) {
+               if (!on) {
+                       PMD_DEBUG_TRACE("port %d: the MAC address was not"
+                               "set in UTA\n", port_id);
+                       return (-EINVAL);
+               }
+
+               index = get_hash_mac_addr_index(port_id, &null_mac_addr);
+               if (index < 0) {
+                       PMD_DEBUG_TRACE("port %d: MAC address array full\n",
+                                       port_id);
+                       return (-ENOSPC);
+               }
+       }
+
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->uc_hash_table_set, -ENOTSUP);
+       ret = (*dev->dev_ops->uc_hash_table_set)(dev, addr, on);
+       if (ret == 0) {
+               /* Update address in NIC data structure */
+               if (on)
+                       ether_addr_copy(addr,
+                                       &dev->data->hash_mac_addrs[index]);
+               else
+                       ether_addr_copy(&null_mac_addr,
+                                       &dev->data->hash_mac_addrs[index]);
+       }
+
+       return ret;
+}
+
+int
+rte_eth_dev_uc_all_hash_table_set(uint8_t port_id, uint8_t on)
+{
+       struct rte_eth_dev *dev;
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("unicast hash setting:Invalid port_id=%d\n",
+                       port_id);
+               return (-ENODEV);
+       }
+
+       dev = &rte_eth_devices[port_id];
+
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->uc_all_hash_table_set, -ENOTSUP);
+       return (*dev->dev_ops->uc_all_hash_table_set)(dev, on);
+}
+
+int
+rte_eth_dev_set_vf_rx(uint8_t port_id,uint16_t vf, uint8_t on)
+{
+       uint16_t num_vfs;
+       struct rte_eth_dev *dev;
+       struct rte_eth_dev_info dev_info;
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return (-ENODEV);
+       }
+
+       dev = &rte_eth_devices[port_id];
+       rte_eth_dev_info_get(port_id, &dev_info);
+
+       num_vfs = dev_info.max_vfs;
+       if (vf > num_vfs)
+       {
+               PMD_DEBUG_TRACE("port %d: invalid vf id\n", port_id);
+               return (-EINVAL);
+       }
+
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->set_vf_rx, -ENOTSUP);
+       return (*dev->dev_ops->set_vf_rx)(dev, vf,on);
+}
+
+int
+rte_eth_dev_set_vf_tx(uint8_t port_id,uint16_t vf, uint8_t on)
+{
+       uint16_t num_vfs;
+       struct rte_eth_dev *dev;
+       struct rte_eth_dev_info dev_info;
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("set pool tx:Invalid port_id=%d\n", port_id);
+               return (-ENODEV);
+       }
+
+       dev = &rte_eth_devices[port_id];
+       rte_eth_dev_info_get(port_id, &dev_info);
+
+       num_vfs = dev_info.max_vfs;
+       if (vf > num_vfs)
+       {
+               PMD_DEBUG_TRACE("set pool tx:invalid pool id=%d\n", vf);
+               return (-EINVAL);
+       }
+
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->set_vf_tx, -ENOTSUP);
+       return (*dev->dev_ops->set_vf_tx)(dev, vf,on);
+}
+
+int
+rte_eth_dev_set_vf_vlan_filter(uint8_t port_id, uint16_t vlan_id,
+                                uint64_t vf_mask,uint8_t vlan_on)
+{
+       struct rte_eth_dev *dev;
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("VF VLAN filter:invalid port id=%d\n",
+                               port_id);
+               return (-ENODEV);
+       }
+       dev = &rte_eth_devices[port_id];
+
+       if(vlan_id > ETHER_MAX_VLAN_ID)
+       {
+               PMD_DEBUG_TRACE("VF VLAN filter:invalid VLAN id=%d\n",
+                       vlan_id);
+               return (-EINVAL);
+       }
+       if (vf_mask == 0)
+       {
+               PMD_DEBUG_TRACE("VF VLAN filter:pool_mask can not be 0\n");
+               return (-EINVAL);
+       }
+
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->set_vf_vlan_filter, -ENOTSUP);
+       return (*dev->dev_ops->set_vf_vlan_filter)(dev, vlan_id,
+                                               vf_mask,vlan_on);
+}
+
+int rte_eth_set_queue_rate_limit(uint8_t port_id, uint16_t queue_idx,
+                                       uint16_t tx_rate)
+{
+       struct rte_eth_dev *dev;
+       struct rte_eth_dev_info dev_info;
+       struct rte_eth_link link;
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("set queue rate limit:invalid port id=%d\n",
+                               port_id);
+               return -ENODEV;
+       }
+
+       dev = &rte_eth_devices[port_id];
+       rte_eth_dev_info_get(port_id, &dev_info);
+       link = dev->data->dev_link;
+
+       if (queue_idx > dev_info.max_tx_queues) {
+               PMD_DEBUG_TRACE("set queue rate limit:port %d: "
+                               "invalid queue id=%d\n", port_id, queue_idx);
+               return -EINVAL;
+       }
+
+       if (tx_rate > link.link_speed) {
+               PMD_DEBUG_TRACE("set queue rate limit:invalid tx_rate=%d, "
+                               "bigger than link speed= %d\n",
+                       tx_rate, link.link_speed);
+               return -EINVAL;
+       }
+
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->set_queue_rate_limit, -ENOTSUP);
+       return (*dev->dev_ops->set_queue_rate_limit)(dev, queue_idx, tx_rate);
+}
+
+int rte_eth_set_vf_rate_limit(uint8_t port_id, uint16_t vf, uint16_t tx_rate,
+                               uint64_t q_msk)
+{
+       struct rte_eth_dev *dev;
+       struct rte_eth_dev_info dev_info;
+       struct rte_eth_link link;
+
+       if (q_msk == 0)
+               return 0;
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("set VF rate limit:invalid port id=%d\n",
+                               port_id);
+               return -ENODEV;
+       }
+
+       dev = &rte_eth_devices[port_id];
+       rte_eth_dev_info_get(port_id, &dev_info);
+       link = dev->data->dev_link;
+
+       if (vf > dev_info.max_vfs) {
+               PMD_DEBUG_TRACE("set VF rate limit:port %d: "
+                               "invalid vf id=%d\n", port_id, vf);
+               return -EINVAL;
+       }
+
+       if (tx_rate > link.link_speed) {
+               PMD_DEBUG_TRACE("set VF rate limit:invalid tx_rate=%d, "
+                               "bigger than link speed= %d\n",
+                               tx_rate, link.link_speed);
+               return -EINVAL;
+       }
+
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->set_vf_rate_limit, -ENOTSUP);
+       return (*dev->dev_ops->set_vf_rate_limit)(dev, vf, tx_rate, q_msk);
+}
+
+int
+rte_eth_mirror_rule_set(uint8_t port_id,
+                       struct rte_eth_vmdq_mirror_conf *mirror_conf,
+                       uint8_t rule_id, uint8_t on)
+{
+       struct rte_eth_dev *dev = &rte_eth_devices[port_id];
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return (-ENODEV);
+       }
+
+       if (mirror_conf->rule_type_mask == 0) {
+               PMD_DEBUG_TRACE("mirror rule type can not be 0.\n");
+               return (-EINVAL);
+       }
+
+       if (mirror_conf->dst_pool >= ETH_64_POOLS) {
+               PMD_DEBUG_TRACE("Invalid dst pool, pool id must"
+                       "be 0-%d\n",ETH_64_POOLS - 1);
+               return (-EINVAL);
+       }
+
+       if ((mirror_conf->rule_type_mask & ETH_VMDQ_POOL_MIRROR) &&
+               (mirror_conf->pool_mask == 0)) {
+               PMD_DEBUG_TRACE("Invalid mirror pool, pool mask can not"
+                               "be 0.\n");
+               return (-EINVAL);
+       }
+
+       if(rule_id >= ETH_VMDQ_NUM_MIRROR_RULE)
+       {
+               PMD_DEBUG_TRACE("Invalid rule_id, rule_id must be 0-%d\n",
+                       ETH_VMDQ_NUM_MIRROR_RULE - 1);
+               return (-EINVAL);
+       }
+
+       dev = &rte_eth_devices[port_id];
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->mirror_rule_set, -ENOTSUP);
+
+       return (*dev->dev_ops->mirror_rule_set)(dev, mirror_conf, rule_id, on);
+}
+
+int
+rte_eth_mirror_rule_reset(uint8_t port_id, uint8_t rule_id)
+{
+       struct rte_eth_dev *dev = &rte_eth_devices[port_id];
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return (-ENODEV);
+       }
+
+       if(rule_id >= ETH_VMDQ_NUM_MIRROR_RULE)
+       {
+               PMD_DEBUG_TRACE("Invalid rule_id, rule_id must be 0-%d\n",
+                       ETH_VMDQ_NUM_MIRROR_RULE-1);
+               return (-EINVAL);
+       }
+
+       dev = &rte_eth_devices[port_id];
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->mirror_rule_reset, -ENOTSUP);
+
+       return (*dev->dev_ops->mirror_rule_reset)(dev, rule_id);
+}
+
+#ifdef RTE_LIBRTE_ETHDEV_DEBUG
+uint16_t
+rte_eth_rx_burst(uint8_t port_id, uint16_t queue_id,
+                struct rte_mbuf **rx_pkts, uint16_t nb_pkts)
+{
+       struct rte_eth_dev *dev;
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return 0;
+       }
+       dev = &rte_eth_devices[port_id];
+       FUNC_PTR_OR_ERR_RET(*dev->rx_pkt_burst, 0);
+       if (queue_id >= dev->data->nb_rx_queues) {
+               PMD_DEBUG_TRACE("Invalid RX queue_id=%d\n", queue_id);
+               return 0;
+       }
+       return (*dev->rx_pkt_burst)(dev->data->rx_queues[queue_id],
+                                               rx_pkts, nb_pkts);
+}
+
+uint16_t
+rte_eth_tx_burst(uint8_t port_id, uint16_t queue_id,
+                struct rte_mbuf **tx_pkts, uint16_t nb_pkts)
+{
+       struct rte_eth_dev *dev;
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return 0;
+       }
+       dev = &rte_eth_devices[port_id];
+
+       FUNC_PTR_OR_ERR_RET(*dev->tx_pkt_burst, 0);
+       if (queue_id >= dev->data->nb_tx_queues) {
+               PMD_DEBUG_TRACE("Invalid TX queue_id=%d\n", queue_id);
+               return 0;
+       }
+       return (*dev->tx_pkt_burst)(dev->data->tx_queues[queue_id],
+                                               tx_pkts, nb_pkts);
+}
+
+uint32_t
+rte_eth_rx_queue_count(uint8_t port_id, uint16_t queue_id)
+{
+       struct rte_eth_dev *dev;
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return 0;
+       }
+       dev = &rte_eth_devices[port_id];
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->rx_queue_count, 0);
+       return (*dev->dev_ops->rx_queue_count)(dev, queue_id);
+}
+
+int
+rte_eth_rx_descriptor_done(uint8_t port_id, uint16_t queue_id, uint16_t offset)
+{
+       struct rte_eth_dev *dev;
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return (-ENODEV);
+       }
+       dev = &rte_eth_devices[port_id];
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->rx_descriptor_done, -ENOTSUP);
+       return (*dev->dev_ops->rx_descriptor_done)( \
+               dev->data->rx_queues[queue_id], offset);
+}
+#endif
 
 int
 rte_eth_dev_callback_register(uint8_t port_id,
                        enum rte_eth_event_type event,
                        rte_eth_dev_cb_fn cb_fn, void *cb_arg)
 {
-       int ret = -1;
        struct rte_eth_dev *dev;
-       struct rte_eth_dev_callback *user_cb = NULL;
+       struct rte_eth_dev_callback *user_cb;
 
        if (!cb_fn)
-               return -1;
+               return (-EINVAL);
        if (port_id >= nb_ports) {
                PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
-               return -1;
+               return (-EINVAL);
        }
+
        dev = &rte_eth_devices[port_id];
        rte_spinlock_lock(&rte_eth_dev_cb_lock);
+
        TAILQ_FOREACH(user_cb, &(dev->callbacks), next) {
                if (user_cb->cb_fn == cb_fn &&
                        user_cb->cb_arg == cb_arg &&
                        user_cb->event == event) {
-                       ret = 0;
-                       goto out;
+                       break;
                }
        }
-       user_cb = rte_malloc("INTR_USER_CALLBACK",
-               sizeof(struct rte_eth_dev_callback), 0);
-       if (!user_cb)
-               goto out;
-       user_cb->cb_fn = cb_fn;
-       user_cb->cb_arg = cb_arg;
-       user_cb->event = event;
-       TAILQ_INSERT_TAIL(&(dev->callbacks), user_cb, next);
-       ret = 0;
 
-out:
-       rte_spinlock_unlock(&rte_eth_dev_cb_lock);
+       /* create a new callback. */
+       if (user_cb == NULL && (user_cb = rte_zmalloc("INTR_USER_CALLBACK",
+                       sizeof(struct rte_eth_dev_callback), 0)) != NULL) {
+               user_cb->cb_fn = cb_fn;
+               user_cb->cb_arg = cb_arg;
+               user_cb->event = event;
+               TAILQ_INSERT_TAIL(&(dev->callbacks), user_cb, next);
+       }
 
-       return ret;
+       rte_spinlock_unlock(&rte_eth_dev_cb_lock);
+       return ((user_cb == NULL) ? -ENOMEM : 0);
 }
 
 int
@@ -1616,38 +2723,51 @@ rte_eth_dev_callback_unregister(uint8_t port_id,
                        enum rte_eth_event_type event,
                        rte_eth_dev_cb_fn cb_fn, void *cb_arg)
 {
-       int ret = -1;
+       int ret;
        struct rte_eth_dev *dev;
-       struct rte_eth_dev_callback *cb_lst = NULL;
+       struct rte_eth_dev_callback *cb, *next;
 
        if (!cb_fn)
-               return -1;
+               return (-EINVAL);
        if (port_id >= nb_ports) {
                PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
-               return -1;
+               return (-EINVAL);
        }
+
        dev = &rte_eth_devices[port_id];
        rte_spinlock_lock(&rte_eth_dev_cb_lock);
-       TAILQ_FOREACH(cb_lst, &(dev->callbacks), next) {
-               if (cb_lst->cb_fn != cb_fn || cb_lst->event != event)
+
+       ret = 0;
+       for (cb = TAILQ_FIRST(&dev->callbacks); cb != NULL; cb = next) {
+
+               next = TAILQ_NEXT(cb, next);
+
+               if (cb->cb_fn != cb_fn || cb->event != event ||
+                               (cb->cb_arg != (void *)-1 &&
+                               cb->cb_arg != cb_arg))
                        continue;
-               if (cb_lst->cb_arg == (void *)-1 ||
-                               cb_lst->cb_arg == cb_arg) {
-                       TAILQ_REMOVE(&(dev->callbacks), cb_lst, next);
-                       rte_free(cb_lst);
-                       ret = 0;
+
+               /*
+                * if this callback is not executing right now,
+                * then remove it.
+                */
+               if (cb->active == 0) {
+                       TAILQ_REMOVE(&(dev->callbacks), cb, next);
+                       rte_free(cb);
+               } else {
+                       ret = -EAGAIN;
                }
        }
 
        rte_spinlock_unlock(&rte_eth_dev_cb_lock);
-
-       return ret;
-}
+       return (ret);
+}
 
 void
-_rte_eth_dev_callback_process(struct rte_eth_dev *dev, enum rte_eth_event_type event)
+_rte_eth_dev_callback_process(struct rte_eth_dev *dev,
+       enum rte_eth_event_type event)
 {
-       struct rte_eth_dev_callback *cb_lst = NULL;
+       struct rte_eth_dev_callback *cb_lst;
        struct rte_eth_dev_callback dev_cb;
 
        rte_spinlock_lock(&rte_eth_dev_cb_lock);
@@ -1655,11 +2775,497 @@ _rte_eth_dev_callback_process(struct rte_eth_dev *dev, enum rte_eth_event_type e
                if (cb_lst->cb_fn == NULL || cb_lst->event != event)
                        continue;
                dev_cb = *cb_lst;
+               cb_lst->active = 1;
                rte_spinlock_unlock(&rte_eth_dev_cb_lock);
                dev_cb.cb_fn(dev->data->port_id, dev_cb.event,
                                                dev_cb.cb_arg);
                rte_spinlock_lock(&rte_eth_dev_cb_lock);
+               cb_lst->active = 0;
        }
        rte_spinlock_unlock(&rte_eth_dev_cb_lock);
 }
+#ifdef RTE_NIC_BYPASS
+int rte_eth_dev_bypass_init(uint8_t port_id)
+{
+       struct rte_eth_dev *dev;
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return (-ENODEV);
+       }
+
+       if ((dev= &rte_eth_devices[port_id]) == NULL) {
+               PMD_DEBUG_TRACE("Invalid port device\n");
+               return (-ENODEV);
+       }
+
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->bypass_init, -ENOTSUP);
+       (*dev->dev_ops->bypass_init)(dev);
+       return 0;
+}
+
+int
+rte_eth_dev_bypass_state_show(uint8_t port_id, uint32_t *state)
+{
+       struct rte_eth_dev *dev;
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return (-ENODEV);
+       }
+
+       if ((dev= &rte_eth_devices[port_id]) == NULL) {
+               PMD_DEBUG_TRACE("Invalid port device\n");
+               return (-ENODEV);
+       }
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->bypass_state_show, -ENOTSUP);
+       (*dev->dev_ops->bypass_state_show)(dev, state);
+       return 0;
+}
+
+int
+rte_eth_dev_bypass_state_set(uint8_t port_id, uint32_t *new_state)
+{
+       struct rte_eth_dev *dev;
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return (-ENODEV);
+       }
+
+       if ((dev= &rte_eth_devices[port_id]) == NULL) {
+               PMD_DEBUG_TRACE("Invalid port device\n");
+               return (-ENODEV);
+       }
+
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->bypass_state_set, -ENOTSUP);
+       (*dev->dev_ops->bypass_state_set)(dev, new_state);
+       return 0;
+}
+
+int
+rte_eth_dev_bypass_event_show(uint8_t port_id, uint32_t event, uint32_t *state)
+{
+       struct rte_eth_dev *dev;
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return (-ENODEV);
+       }
+
+       if ((dev= &rte_eth_devices[port_id]) == NULL) {
+               PMD_DEBUG_TRACE("Invalid port device\n");
+               return (-ENODEV);
+       }
+
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->bypass_state_show, -ENOTSUP);
+       (*dev->dev_ops->bypass_event_show)(dev, event, state);
+       return 0;
+}
+
+int
+rte_eth_dev_bypass_event_store(uint8_t port_id, uint32_t event, uint32_t state)
+{
+       struct rte_eth_dev *dev;
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return (-ENODEV);
+       }
+
+       if ((dev= &rte_eth_devices[port_id]) == NULL) {
+               PMD_DEBUG_TRACE("Invalid port device\n");
+               return (-ENODEV);
+       }
+
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->bypass_event_set, -ENOTSUP);
+       (*dev->dev_ops->bypass_event_set)(dev, event, state);
+       return 0;
+}
+
+int
+rte_eth_dev_wd_timeout_store(uint8_t port_id, uint32_t timeout)
+{
+       struct rte_eth_dev *dev;
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return (-ENODEV);
+       }
+
+       if ((dev= &rte_eth_devices[port_id]) == NULL) {
+               PMD_DEBUG_TRACE("Invalid port device\n");
+               return (-ENODEV);
+       }
+
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->bypass_wd_timeout_set, -ENOTSUP);
+       (*dev->dev_ops->bypass_wd_timeout_set)(dev, timeout);
+       return 0;
+}
+
+int
+rte_eth_dev_bypass_ver_show(uint8_t port_id, uint32_t *ver)
+{
+       struct rte_eth_dev *dev;
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return (-ENODEV);
+       }
+
+       if ((dev= &rte_eth_devices[port_id]) == NULL) {
+               PMD_DEBUG_TRACE("Invalid port device\n");
+               return (-ENODEV);
+       }
+
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->bypass_ver_show, -ENOTSUP);
+       (*dev->dev_ops->bypass_ver_show)(dev, ver);
+       return 0;
+}
+
+int
+rte_eth_dev_bypass_wd_timeout_show(uint8_t port_id, uint32_t *wd_timeout)
+{
+       struct rte_eth_dev *dev;
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return (-ENODEV);
+       }
+
+       if ((dev= &rte_eth_devices[port_id]) == NULL) {
+               PMD_DEBUG_TRACE("Invalid port device\n");
+               return (-ENODEV);
+       }
+
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->bypass_wd_timeout_show, -ENOTSUP);
+       (*dev->dev_ops->bypass_wd_timeout_show)(dev, wd_timeout);
+       return 0;
+}
 
+int
+rte_eth_dev_bypass_wd_reset(uint8_t port_id)
+{
+       struct rte_eth_dev *dev;
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return (-ENODEV);
+       }
+
+       if ((dev= &rte_eth_devices[port_id]) == NULL) {
+               PMD_DEBUG_TRACE("Invalid port device\n");
+               return (-ENODEV);
+       }
+
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->bypass_wd_reset, -ENOTSUP);
+       (*dev->dev_ops->bypass_wd_reset)(dev);
+       return 0;
+}
+#endif
+
+int
+rte_eth_dev_add_syn_filter(uint8_t port_id,
+                       struct rte_syn_filter *filter, uint16_t rx_queue)
+{
+       struct rte_eth_dev *dev;
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return -ENODEV;
+       }
+
+       dev = &rte_eth_devices[port_id];
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->add_syn_filter, -ENOTSUP);
+       return (*dev->dev_ops->add_syn_filter)(dev, filter, rx_queue);
+}
+
+int
+rte_eth_dev_remove_syn_filter(uint8_t port_id)
+{
+       struct rte_eth_dev *dev;
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return -ENODEV;
+       }
+
+       dev = &rte_eth_devices[port_id];
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->remove_syn_filter, -ENOTSUP);
+       return (*dev->dev_ops->remove_syn_filter)(dev);
+}
+
+int
+rte_eth_dev_get_syn_filter(uint8_t port_id,
+                       struct rte_syn_filter *filter, uint16_t *rx_queue)
+{
+       struct rte_eth_dev *dev;
+
+       if (filter == NULL || rx_queue == NULL)
+               return -EINVAL;
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return -ENODEV;
+       }
+
+       dev = &rte_eth_devices[port_id];
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->get_syn_filter, -ENOTSUP);
+       return (*dev->dev_ops->get_syn_filter)(dev, filter, rx_queue);
+}
+
+int
+rte_eth_dev_add_ethertype_filter(uint8_t port_id, uint16_t index,
+                       struct rte_ethertype_filter *filter, uint16_t rx_queue)
+{
+       struct rte_eth_dev *dev;
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return -ENODEV;
+       }
+       if (filter->ethertype == ETHER_TYPE_IPv4 ||
+               filter->ethertype == ETHER_TYPE_IPv6){
+               PMD_DEBUG_TRACE("IP and IPv6 are not supported"
+                       " in ethertype filter\n");
+               return -EINVAL;
+       }
+       dev = &rte_eth_devices[port_id];
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->add_ethertype_filter, -ENOTSUP);
+       return (*dev->dev_ops->add_ethertype_filter)(dev, index,
+                                       filter, rx_queue);
+}
+
+int
+rte_eth_dev_remove_ethertype_filter(uint8_t port_id,  uint16_t index)
+{
+       struct rte_eth_dev *dev;
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return -ENODEV;
+       }
+
+       dev = &rte_eth_devices[port_id];
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->remove_ethertype_filter, -ENOTSUP);
+       return (*dev->dev_ops->remove_ethertype_filter)(dev, index);
+}
+
+int
+rte_eth_dev_get_ethertype_filter(uint8_t port_id, uint16_t index,
+                       struct rte_ethertype_filter *filter, uint16_t *rx_queue)
+{
+       struct rte_eth_dev *dev;
+
+       if (filter == NULL || rx_queue == NULL)
+               return -EINVAL;
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return -ENODEV;
+       }
+
+       dev = &rte_eth_devices[port_id];
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->get_ethertype_filter, -ENOTSUP);
+       return (*dev->dev_ops->get_ethertype_filter)(dev, index,
+                                               filter, rx_queue);
+}
+
+int
+rte_eth_dev_add_2tuple_filter(uint8_t port_id, uint16_t index,
+                       struct rte_2tuple_filter *filter, uint16_t rx_queue)
+{
+       struct rte_eth_dev *dev;
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return -ENODEV;
+       }
+       if (filter->protocol != IPPROTO_TCP &&
+               filter->tcp_flags != 0){
+               PMD_DEBUG_TRACE("tcp flags is 0x%x, but the protocol value"
+                       " is not TCP\n",
+                       filter->tcp_flags);
+               return -EINVAL;
+       }
+
+       dev = &rte_eth_devices[port_id];
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->add_2tuple_filter, -ENOTSUP);
+       return (*dev->dev_ops->add_2tuple_filter)(dev, index, filter, rx_queue);
+}
+
+int
+rte_eth_dev_remove_2tuple_filter(uint8_t port_id, uint16_t index)
+{
+       struct rte_eth_dev *dev;
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return -ENODEV;
+       }
+
+       dev = &rte_eth_devices[port_id];
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->remove_2tuple_filter, -ENOTSUP);
+       return (*dev->dev_ops->remove_2tuple_filter)(dev, index);
+}
+
+int
+rte_eth_dev_get_2tuple_filter(uint8_t port_id, uint16_t index,
+                       struct rte_2tuple_filter *filter, uint16_t *rx_queue)
+{
+       struct rte_eth_dev *dev;
+
+       if (filter == NULL || rx_queue == NULL)
+               return -EINVAL;
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return -ENODEV;
+       }
+
+       dev = &rte_eth_devices[port_id];
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->get_2tuple_filter, -ENOTSUP);
+       return (*dev->dev_ops->get_2tuple_filter)(dev, index, filter, rx_queue);
+}
+
+int
+rte_eth_dev_add_5tuple_filter(uint8_t port_id, uint16_t index,
+                       struct rte_5tuple_filter *filter, uint16_t rx_queue)
+{
+       struct rte_eth_dev *dev;
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return -ENODEV;
+       }
+
+       if (filter->protocol != IPPROTO_TCP &&
+               filter->tcp_flags != 0){
+               PMD_DEBUG_TRACE("tcp flags is 0x%x, but the protocol value"
+                       " is not TCP\n",
+                       filter->tcp_flags);
+               return -EINVAL;
+       }
+
+       dev = &rte_eth_devices[port_id];
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->add_5tuple_filter, -ENOTSUP);
+       return (*dev->dev_ops->add_5tuple_filter)(dev, index, filter, rx_queue);
+}
+
+int
+rte_eth_dev_remove_5tuple_filter(uint8_t port_id, uint16_t index)
+{
+       struct rte_eth_dev *dev;
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return -ENODEV;
+       }
+
+       dev = &rte_eth_devices[port_id];
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->remove_5tuple_filter, -ENOTSUP);
+       return (*dev->dev_ops->remove_5tuple_filter)(dev, index);
+}
+
+int
+rte_eth_dev_get_5tuple_filter(uint8_t port_id, uint16_t index,
+                       struct rte_5tuple_filter *filter, uint16_t *rx_queue)
+{
+       struct rte_eth_dev *dev;
+
+       if (filter == NULL || rx_queue == NULL)
+               return -EINVAL;
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return -ENODEV;
+       }
+
+       dev = &rte_eth_devices[port_id];
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->get_5tuple_filter, -ENOTSUP);
+       return (*dev->dev_ops->get_5tuple_filter)(dev, index, filter,
+                                               rx_queue);
+}
+
+int
+rte_eth_dev_add_flex_filter(uint8_t port_id, uint16_t index,
+                       struct rte_flex_filter *filter, uint16_t rx_queue)
+{
+       struct rte_eth_dev *dev;
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return -ENODEV;
+       }
+
+       dev = &rte_eth_devices[port_id];
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->add_flex_filter, -ENOTSUP);
+       return (*dev->dev_ops->add_flex_filter)(dev, index, filter, rx_queue);
+}
+
+int
+rte_eth_dev_remove_flex_filter(uint8_t port_id, uint16_t index)
+{
+       struct rte_eth_dev *dev;
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return -ENODEV;
+       }
+
+       dev = &rte_eth_devices[port_id];
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->remove_flex_filter, -ENOTSUP);
+       return (*dev->dev_ops->remove_flex_filter)(dev, index);
+}
+
+int
+rte_eth_dev_get_flex_filter(uint8_t port_id, uint16_t index,
+                       struct rte_flex_filter *filter, uint16_t *rx_queue)
+{
+       struct rte_eth_dev *dev;
+
+       if (filter == NULL || rx_queue == NULL)
+               return -EINVAL;
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return -ENODEV;
+       }
+
+       dev = &rte_eth_devices[port_id];
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->get_flex_filter, -ENOTSUP);
+       return (*dev->dev_ops->get_flex_filter)(dev, index, filter,
+                                               rx_queue);
+}
+
+int
+rte_eth_dev_filter_supported(uint8_t port_id, enum rte_filter_type filter_type)
+{
+       struct rte_eth_dev *dev;
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return -ENODEV;
+       }
+
+       dev = &rte_eth_devices[port_id];
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->filter_ctrl, -ENOTSUP);
+       return (*dev->dev_ops->filter_ctrl)(dev, filter_type,
+                               RTE_ETH_FILTER_NOP, NULL);
+}
+
+int
+rte_eth_dev_filter_ctrl(uint8_t port_id, enum rte_filter_type filter_type,
+                      enum rte_filter_op filter_op, void *arg)
+{
+       struct rte_eth_dev *dev;
+
+       if (port_id >= nb_ports) {
+               PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
+               return -ENODEV;
+       }
+
+       dev = &rte_eth_devices[port_id];
+       FUNC_PTR_OR_ERR_RET(*dev->dev_ops->filter_ctrl, -ENOTSUP);
+       return (*dev->dev_ops->filter_ctrl)(dev, filter_type, filter_op, arg);
+}