diff mbox

[dpdk-dev,v5] net/kni: add KNI PMD

Message ID 20170130165739.16253-1-ferruh.yigit@intel.com (mailing list archive)
State Changes Requested, archived
Delegated to: Ferruh Yigit
Headers show

Checks

Context Check Description
ci/checkpatch success coding style OK
ci/Intel compilation fail Compilation issues

Commit Message

Ferruh Yigit Jan. 30, 2017, 4:57 p.m. UTC
Add KNI PMD which wraps librte_kni for ease of use.

KNI PMD can be used as any regular PMD to send / receive packets to the
Linux networking stack.

Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
---

v5:
* add kvargs "no_request_thread" to disable a specific pthread creation
to handle control requests.
* add documentation

v4:
* allow only single queue
* use driver.name as name

v3:
* rebase on top of latest master

v2:
* updated driver name eth_kni -> net_kni
---
 MAINTAINERS                             |   5 +
 config/common_base                      |   1 +
 config/common_linuxapp                  |   1 +
 doc/guides/nics/features/kni.ini        |   7 +
 doc/guides/nics/index.rst               |   1 +
 doc/guides/nics/kni.rst                 | 197 ++++++++++++
 drivers/net/Makefile                    |   1 +
 drivers/net/kni/Makefile                |  64 ++++
 drivers/net/kni/rte_eth_kni.c           | 515 ++++++++++++++++++++++++++++++++
 drivers/net/kni/rte_pmd_kni_version.map |   4 +
 mk/rte.app.mk                           |  10 +-
 11 files changed, 801 insertions(+), 5 deletions(-)
 create mode 100644 doc/guides/nics/features/kni.ini
 create mode 100644 doc/guides/nics/kni.rst
 create mode 100644 drivers/net/kni/Makefile
 create mode 100644 drivers/net/kni/rte_eth_kni.c
 create mode 100644 drivers/net/kni/rte_pmd_kni_version.map

Comments

Yong Wang Jan. 30, 2017, 7:05 p.m. UTC | #1
> -----Original Message-----

> From: Ferruh Yigit [mailto:ferruh.yigit@intel.com]

> Sent: Monday, January 30, 2017 8:58 AM

> To: Thomas Monjalon <thomas.monjalon@6wind.com>; Yong Wang

> <yongwang@vmware.com>

> Cc: dev@dpdk.org; Ferruh Yigit <ferruh.yigit@intel.com>

> Subject: [PATCH v5] net/kni: add KNI PMD

> 

> Add KNI PMD which wraps librte_kni for ease of use.

> 

> KNI PMD can be used as any regular PMD to send / receive packets to the

> Linux networking stack.

> 

> Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>

> ---


Looks good except a few typos in the documentation added.

Reviewed-by: Yong Wang <yongwang@vmware.com>


> 

> v5:

> * add kvargs "no_request_thread" to disable a specific pthread creation

> to handle control requests.

> * add documentation

> 

> v4:

> * allow only single queue

> * use driver.name as name

> 

> v3:

> * rebase on top of latest master

> 

> v2:

> * updated driver name eth_kni -> net_kni

> ---

>  MAINTAINERS                             |   5 +

>  config/common_base                      |   1 +

>  config/common_linuxapp                  |   1 +

>  doc/guides/nics/features/kni.ini        |   7 +

>  doc/guides/nics/index.rst               |   1 +

>  doc/guides/nics/kni.rst                 | 197 ++++++++++++

>  drivers/net/Makefile                    |   1 +

>  drivers/net/kni/Makefile                |  64 ++++

>  drivers/net/kni/rte_eth_kni.c           | 515

> ++++++++++++++++++++++++++++++++

>  drivers/net/kni/rte_pmd_kni_version.map |   4 +

>  mk/rte.app.mk                           |  10 +-

>  11 files changed, 801 insertions(+), 5 deletions(-)

>  create mode 100644 doc/guides/nics/features/kni.ini

>  create mode 100644 doc/guides/nics/kni.rst

>  create mode 100644 drivers/net/kni/Makefile

>  create mode 100644 drivers/net/kni/rte_eth_kni.c

>  create mode 100644 drivers/net/kni/rte_pmd_kni_version.map

> 

> diff --git a/MAINTAINERS b/MAINTAINERS

> index f071138..8eb83f5 100644

> --- a/MAINTAINERS

> +++ b/MAINTAINERS

> @@ -404,6 +404,11 @@ M: Keith Wiles <keith.wiles@intel.com>

>  F: drivers/net/tap/

>  F: doc/guides/nics/tap.rst

> 

> +KNI PMD

> +M: Ferruh Yigit <ferruh.yigit@intel.com>

> +F: drivers/net/kni/

> +F: doc/guides/nics/kni.rst

> +

>  Ring PMD

>  M: Bruce Richardson <bruce.richardson@intel.com>

>  F: drivers/net/ring/

> diff --git a/config/common_base b/config/common_base

> index 61efb87..2e1bbd5 100644

> --- a/config/common_base

> +++ b/config/common_base

> @@ -576,6 +576,7 @@ CONFIG_RTE_PIPELINE_STATS_COLLECT=n

>  # Compile librte_kni

>  #

>  CONFIG_RTE_LIBRTE_KNI=n

> +CONFIG_RTE_LIBRTE_PMD_KNI=n

>  CONFIG_RTE_KNI_KMOD=n

>  CONFIG_RTE_KNI_KMOD_ETHTOOL=n

>  CONFIG_RTE_KNI_PREEMPT_DEFAULT=y

> diff --git a/config/common_linuxapp b/config/common_linuxapp

> index 00ebaac..d03a60a 100644

> --- a/config/common_linuxapp

> +++ b/config/common_linuxapp

> @@ -39,6 +39,7 @@ CONFIG_RTE_EAL_IGB_UIO=y

>  CONFIG_RTE_EAL_VFIO=y

>  CONFIG_RTE_KNI_KMOD=y

>  CONFIG_RTE_LIBRTE_KNI=y

> +CONFIG_RTE_LIBRTE_PMD_KNI=y

>  CONFIG_RTE_LIBRTE_VHOST=y

>  CONFIG_RTE_LIBRTE_PMD_VHOST=y

>  CONFIG_RTE_LIBRTE_PMD_AF_PACKET=y

> diff --git a/doc/guides/nics/features/kni.ini

> b/doc/guides/nics/features/kni.ini

> new file mode 100644

> index 0000000..6deb66a

> --- /dev/null

> +++ b/doc/guides/nics/features/kni.ini

> @@ -0,0 +1,7 @@

> +;

> +; Supported features of the 'kni' network poll mode driver.

> +;

> +; Refer to default.ini for the full list of available PMD features.

> +;

> +[Features]

> +Usage doc            = Y

> diff --git a/doc/guides/nics/index.rst b/doc/guides/nics/index.rst

> index 87f9334..5248625 100644

> --- a/doc/guides/nics/index.rst

> +++ b/doc/guides/nics/index.rst

> @@ -46,6 +46,7 @@ Network Interface Controller Drivers

>      i40e

>      ixgbe

>      intel_vf

> +    kni

>      mlx4

>      mlx5

>      nfp

> diff --git a/doc/guides/nics/kni.rst b/doc/guides/nics/kni.rst

> new file mode 100644

> index 0000000..2e2032b

> --- /dev/null

> +++ b/doc/guides/nics/kni.rst

> @@ -0,0 +1,197 @@

> +..  BSD LICENSE

> +    Copyright(c) 2017 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

> +    are met:

> +

> +    * 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

> +    distribution.

> +    * 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

> +    OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH

> DAMAGE.

> +

> +KNI Poll Mode Driver

> +======================

> +

> +KNI PMD is wrapper to the :ref:`librte_kni <kni>` library.

> +

> +This PMD enables using KNI without having a KNI specific application,

> +any forwarding application can use PMD interface for KNI.

> +

> +Sending packets to the any DPDK controlled interface or sending to the


"to the any DPDK" should be "too any DPDK"
 
> +Linux networking stack will be transparent to the DPDK application.

> +

> +To create a KNI device ``net_kni#`` device name should be used, and this

> +will create ``kni#`` Linux virtual network interface.

> +

> +There is no physical device backend for the virtual KNI device.

> +

> +Packets sent to the KNI Linux interface will be received by the DPDK

> +application, and DPDK application may forward packets to a physical NIC

> +or to a virtual device (like another KNI interface or PCAP interface).

> +

> +To forward any traffic from physical NIC to the Linux networking stack,

> +an application should control a physical port and create one virtual KNI port,

> +and forward between two.

> +

> +Using this PMD requires KNI kernel module be inserted.

> +

> +

> +Usage

> +-----

> +

> +EAL ``--vdev`` argument can be used to create KNI device instance, like::

> +

> +        testpmd --vdev=net_kni0 --vdev=net_kn1 -- -i

> +

> +Above command will create ``kni0`` and ``kni1`` Linux network interfaces,

> +those interfaces can be controlled by standard Linux tools.

> +

> +When testpmd forwarding started, any packets send to ``kni0`` interface


s/started/starts
s/send/sent

> +forwarded to the ``kni1`` interface and vice versa.

> +

> +There is no hard limit on number of interfaces can be created.


Add "that" after "interfaces".

> +

> +

> +Default interface configuration

> +-------------------------------

> +

> +``librte_kni`` can create Linux network interfaces with different features,

> +feature set controlled by a configuration struct, and KNI PMD uses a fixed

> +configuration:

> +

> +    .. code-block:: console

> +

> +        Interface name: kni#

> +        force bind kernel thread to a core : NO

> +        mbuf size: MAX_PACKET_SZ

> +

> +KNI control path is not supported with the PMD, since there is no physical

> +backend device by default.

> +

> +

> +PMD arguments

> +-------------

> +

> +``no_request_thread``, by default PMD creates a phtread for each KNI

> interface

> +to handle Linux network interface control commands, like ``ifconfig kni0 up``

> +

> +With ``no_request_thread`` option, pthread is not created and control

> commands

> +not handled by PMD.

> +

> +By default request thread is enabled. And this argument should not be used

> +most of the time, unless this PMD used with customized DPDK application

> to handle

> +requests itself.

> +

> +Argument usage::

> +

> +        testpmd --vdev "net_kni0,no_request_thread=1" -- -i

> +

> +

> +PMD log messages

> +----------------

> +

> +If KNI kernel module (rte_kni.ko) not inserted, following error log printed::

> +

> +        "KNI: KNI subsystem has not been initialized. Invoke rte_kni_init() first"

> +

> +

> +PMD testing

> +-----------

> +

> +It is possible to test PMD quicly using KNI kernel module loopback feature:


s/quicly/quickly

> +

> +* Insert KNI kernel module with loopback support:

> +

> +    .. code-block:: console

> +

> +        insmod build/kmod/rte_kni.ko lo_mode=lo_mode_fifo_skb

> +

> +* Start testpmd with no physical device but two KNI virtual devices:

> +

> +    .. code-block:: console

> +

> +        ./testpmd --vdev net_kni0 --vdev net_kni1 -- -i

> +

> +    .. code-block:: console

> +

> +        ...

> +        Configuring Port 0 (socket 0)

> +        KNI: pci: 00:00:00       c580:b8

> +        Port 0: 1A:4A:5B:7C:A2:8C

> +        Configuring Port 1 (socket 0)

> +        KNI: pci: 00:00:00       600:b9

> +        Port 1: AE:95:21:07:93:DD

> +        Checking link statuses...

> +        Port 0 Link Up - speed 10000 Mbps - full-duplex

> +        Port 1 Link Up - speed 10000 Mbps - full-duplex

> +        Done

> +        testpmd>

> +

> +* Observe Linux interfaces

> +

> +    .. code-block:: console

> +

> +        $ ifconfig kni0 && ifconfig kni1

> +        kni0: flags=4098<BROADCAST,MULTICAST>  mtu 1500

> +                ether ae:8e:79:8e:9b:c8  txqueuelen 1000  (Ethernet)

> +                RX packets 0  bytes 0 (0.0 B)

> +                RX errors 0  dropped 0  overruns 0  frame 0

> +                TX packets 0  bytes 0 (0.0 B)

> +                TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

> +

> +        kni1: flags=4098<BROADCAST,MULTICAST>  mtu 1500

> +                ether 9e:76:43:53:3e:9b  txqueuelen 1000  (Ethernet)

> +                RX packets 0  bytes 0 (0.0 B)

> +                RX errors 0  dropped 0  overruns 0  frame 0

> +                TX packets 0  bytes 0 (0.0 B)

> +                TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

> +

> +

> +* Start forwarding with tx_first:

> +

> +    .. code-block:: console

> +

> +        testpmd> start tx_first

> +

> +* Quit and check forwarding stats:

> +

> +    .. code-block:: console

> +

> +        testpmd> quit

> +        Telling cores to stop...

> +        Waiting for lcores to finish...

> +

> +        ---------------------- Forward statistics for port 0  ----------------------

> +        RX-packets: 35637905       RX-dropped: 0             RX-total: 35637905

> +        TX-packets: 35637947       TX-dropped: 0             TX-total: 35637947

> +        ----------------------------------------------------------------------------

> +

> +        ---------------------- Forward statistics for port 1  ----------------------

> +        RX-packets: 35637915       RX-dropped: 0             RX-total: 35637915

> +        TX-packets: 35637937       TX-dropped: 0             TX-total: 35637937

> +        ----------------------------------------------------------------------------

> +

> +        +++++++++++++++ Accumulated forward statistics for all

> ports+++++++++++++++

> +        RX-packets: 71275820       RX-dropped: 0             RX-total: 71275820

> +        TX-packets: 71275884       TX-dropped: 0             TX-total: 71275884

> +

> ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

> ++++++++++++++++++

> +

> diff --git a/drivers/net/Makefile b/drivers/net/Makefile

> index 40fc333..8fd6745 100644

> --- a/drivers/net/Makefile

> +++ b/drivers/net/Makefile

> @@ -41,6 +41,7 @@ DIRS-$(CONFIG_RTE_LIBRTE_ENIC_PMD) += enic

>  DIRS-$(CONFIG_RTE_LIBRTE_FM10K_PMD) += fm10k

>  DIRS-$(CONFIG_RTE_LIBRTE_I40E_PMD) += i40e

>  DIRS-$(CONFIG_RTE_LIBRTE_IXGBE_PMD) += ixgbe

> +DIRS-$(CONFIG_RTE_LIBRTE_PMD_KNI) += kni

>  DIRS-$(CONFIG_RTE_LIBRTE_MLX4_PMD) += mlx4

>  DIRS-$(CONFIG_RTE_LIBRTE_MLX5_PMD) += mlx5

>  DIRS-$(CONFIG_RTE_LIBRTE_MPIPE_PMD) += mpipe

> diff --git a/drivers/net/kni/Makefile b/drivers/net/kni/Makefile

> new file mode 100644

> index 0000000..b3017b1

> --- /dev/null

> +++ b/drivers/net/kni/Makefile

> @@ -0,0 +1,64 @@

> +#   BSD LICENSE

> +#

> +#   Copyright(c) 2017 Intel Corporation. All rights reserved.

> +#

> +#   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

> +#       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

> +#       distribution.

> +#     * 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

> +#   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH

> DAMAGE.

> +

> +include $(RTE_SDK)/mk/rte.vars.mk

> +

> +#

> +# library name

> +#

> +LIB = librte_pmd_kni.a

> +

> +CFLAGS += -O3

> +CFLAGS += $(WERROR_FLAGS)

> +LDLIBS += -lpthread

> +

> +EXPORT_MAP := rte_pmd_kni_version.map

> +

> +LIBABIVER := 1

> +

> +#

> +# all source are stored in SRCS-y

> +#

> +SRCS-$(CONFIG_RTE_LIBRTE_PMD_KNI) += rte_eth_kni.c

> +

> +#

> +# Export include files

> +#

> +SYMLINK-y-include +=

> +

> +# this lib depends upon:

> +DEPDIRS-$(CONFIG_RTE_LIBRTE_PMD_KNI) += lib/librte_eal

> +DEPDIRS-$(CONFIG_RTE_LIBRTE_PMD_KNI) += lib/librte_ether

> +DEPDIRS-$(CONFIG_RTE_LIBRTE_PMD_KNI) += lib/librte_kni

> +DEPDIRS-$(CONFIG_RTE_LIBRTE_PMD_KNI) += lib/librte_kvargs

> +DEPDIRS-$(CONFIG_RTE_LIBRTE_PMD_KNI) += lib/librte_mbuf

> +DEPDIRS-$(CONFIG_RTE_LIBRTE_PMD_KNI) += lib/librte_mempool

> +

> +include $(RTE_SDK)/mk/rte.lib.mk

> diff --git a/drivers/net/kni/rte_eth_kni.c b/drivers/net/kni/rte_eth_kni.c

> new file mode 100644

> index 0000000..59205e7

> --- /dev/null

> +++ b/drivers/net/kni/rte_eth_kni.c

> @@ -0,0 +1,515 @@

> +/*-

> + *   BSD LICENSE

> + *

> + *   Copyright(c) 2017 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

> + *   are met:

> + *

> + *     * 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

> + *       distribution.

> + *     * 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

> + *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH

> DAMAGE.

> + */

> +

> +#include <fcntl.h>

> +#include <pthread.h>

> +#include <unistd.h>

> +

> +#include <rte_ethdev.h>

> +#include <rte_kni.h>

> +#include <rte_kvargs.h>

> +#include <rte_malloc.h>

> +#include <rte_vdev.h>

> +

> +/* Only single queue supported */

> +#define KNI_MAX_QUEUE_PER_PORT 1

> +

> +#define MAX_PACKET_SZ 2048

> +#define MAX_KNI_PORTS 8

> +

> +#define ETH_KNI_NO_REQUEST_THREAD_ARG	"no_request_thread"

> +static const char * const valid_arguments[] = {

> +	ETH_KNI_NO_REQUEST_THREAD_ARG,

> +	NULL

> +};

> +

> +struct eth_kni_args {

> +	int no_request_thread;

> +};

> +

> +struct pmd_queue_stats {

> +	uint64_t pkts;

> +	uint64_t bytes;

> +	uint64_t err_pkts;

> +};

> +

> +struct pmd_queue {

> +	struct pmd_internals *internals;

> +	struct rte_mempool *mb_pool;

> +

> +	struct pmd_queue_stats rx;

> +	struct pmd_queue_stats tx;

> +};

> +

> +struct pmd_internals {

> +	struct rte_kni *kni;

> +	int is_kni_started;

> +

> +	pthread_t thread;

> +	int stop_thread;

> +	int no_request_thread;

> +

> +	struct ether_addr eth_addr;

> +

> +	struct pmd_queue rx_queues[KNI_MAX_QUEUE_PER_PORT];

> +	struct pmd_queue tx_queues[KNI_MAX_QUEUE_PER_PORT];

> +};

> +

> +static const struct rte_eth_link pmd_link = {

> +		.link_speed = ETH_SPEED_NUM_10G,

> +		.link_duplex = ETH_LINK_FULL_DUPLEX,

> +		.link_status = ETH_LINK_DOWN,

> +		.link_autoneg = ETH_LINK_SPEED_AUTONEG,

> +};

> +static int is_kni_initialized;

> +

> +static uint16_t

> +eth_kni_rx(void *q, struct rte_mbuf **bufs, uint16_t nb_bufs)

> +{

> +	struct pmd_queue *kni_q = q;

> +	struct rte_kni *kni = kni_q->internals->kni;

> +	uint16_t nb_pkts;

> +

> +	nb_pkts = rte_kni_rx_burst(kni, bufs, nb_bufs);

> +

> +	kni_q->rx.pkts += nb_pkts;

> +	kni_q->rx.err_pkts += nb_bufs - nb_pkts;

> +

> +	return nb_pkts;

> +}

> +

> +static uint16_t

> +eth_kni_tx(void *q, struct rte_mbuf **bufs, uint16_t nb_bufs)

> +{

> +	struct pmd_queue *kni_q = q;

> +	struct rte_kni *kni = kni_q->internals->kni;

> +	uint16_t nb_pkts;

> +

> +	nb_pkts =  rte_kni_tx_burst(kni, bufs, nb_bufs);

> +

> +	kni_q->tx.pkts += nb_pkts;

> +	kni_q->tx.err_pkts += nb_bufs - nb_pkts;

> +

> +	return nb_pkts;

> +}

> +

> +static void *

> +kni_handle_request(void *param)

> +{

> +	struct pmd_internals *internals = param;

> +#define MS 1000

> +

> +	while (!internals->stop_thread) {

> +		rte_kni_handle_request(internals->kni);

> +		usleep(500 * MS);

> +	}

> +

> +	return param;

> +}

> +

> +static int

> +eth_kni_start(struct rte_eth_dev *dev)

> +{

> +	struct pmd_internals *internals = dev->data->dev_private;

> +	uint16_t port_id = dev->data->port_id;

> +	struct rte_mempool *mb_pool;

> +	struct rte_kni_conf conf;

> +	const char *name = dev->data->name + 4; /* remove net_ */

> +

> +	snprintf(conf.name, RTE_KNI_NAMESIZE, "%s", name);

> +	conf.force_bind = 0;

> +	conf.group_id = port_id;

> +	conf.mbuf_size = MAX_PACKET_SZ;

> +	mb_pool = internals->rx_queues[0].mb_pool;

> +

> +	internals->kni = rte_kni_alloc(mb_pool, &conf, NULL);

> +	if (internals->kni == NULL) {

> +		RTE_LOG(ERR, PMD,

> +			"Fail to create kni interface for port: %d\n",

> +			port_id);

> +		return -1;

> +	}

> +

> +	return 0;

> +}

> +

> +static int

> +eth_kni_dev_start(struct rte_eth_dev *dev)

> +{

> +	struct pmd_internals *internals = dev->data->dev_private;

> +	int ret;

> +

> +	if (internals->is_kni_started == 0) {

> +		ret = eth_kni_start(dev);

> +		if (ret)

> +			return -1;

> +		internals->is_kni_started = 1;

> +	}

> +

> +	if (internals->no_request_thread == 0) {

> +		ret = pthread_create(&internals->thread, NULL,

> +			kni_handle_request, internals);

> +		if (ret) {

> +			RTE_LOG(ERR, PMD,

> +				"Fail to create kni request thread\n");

> +			return -1;

> +		}

> +	}

> +

> +	dev->data->dev_link.link_status = 1;

> +

> +	return 0;

> +}

> +

> +static void

> +eth_kni_dev_stop(struct rte_eth_dev *dev)

> +{

> +	struct pmd_internals *internals = dev->data->dev_private;

> +	int ret;

> +

> +	if (internals->no_request_thread == 0) {

> +		internals->stop_thread = 1;

> +

> +		ret = pthread_cancel(internals->thread);

> +		if (ret)

> +			RTE_LOG(ERR, PMD, "Can't cancel the thread\n");

> +

> +		ret = pthread_join(internals->thread, NULL);

> +		if (ret)

> +			RTE_LOG(ERR, PMD, "Can't join the thread\n");

> +

> +		internals->stop_thread = 0;

> +	}

> +

> +	dev->data->dev_link.link_status = 0;

> +}

> +

> +static int

> +eth_kni_dev_configure(struct rte_eth_dev *dev __rte_unused)

> +{

> +	return 0;

> +}

> +

> +static void

> +eth_kni_dev_info(struct rte_eth_dev *dev __rte_unused,

> +		struct rte_eth_dev_info *dev_info)

> +{

> +	dev_info->max_mac_addrs = 1;

> +	dev_info->max_rx_pktlen = UINT32_MAX;

> +	dev_info->max_rx_queues = KNI_MAX_QUEUE_PER_PORT;

> +	dev_info->max_tx_queues = KNI_MAX_QUEUE_PER_PORT;

> +	dev_info->min_rx_bufsize = 0;

> +	dev_info->pci_dev = NULL;

> +}

> +

> +static int

> +eth_kni_rx_queue_setup(struct rte_eth_dev *dev,

> +		uint16_t rx_queue_id,

> +		uint16_t nb_rx_desc __rte_unused,

> +		unsigned int socket_id __rte_unused,

> +		const struct rte_eth_rxconf *rx_conf __rte_unused,

> +		struct rte_mempool *mb_pool)

> +{

> +	struct pmd_internals *internals = dev->data->dev_private;

> +	struct pmd_queue *q;

> +

> +	q = &internals->rx_queues[rx_queue_id];

> +	q->internals = internals;

> +	q->mb_pool = mb_pool;

> +

> +	dev->data->rx_queues[rx_queue_id] = q;

> +

> +	return 0;

> +}

> +

> +static int

> +eth_kni_tx_queue_setup(struct rte_eth_dev *dev,

> +		uint16_t tx_queue_id,

> +		uint16_t nb_tx_desc __rte_unused,

> +		unsigned int socket_id __rte_unused,

> +		const struct rte_eth_txconf *tx_conf __rte_unused)

> +{

> +	struct pmd_internals *internals = dev->data->dev_private;

> +	struct pmd_queue *q;

> +

> +	q = &internals->tx_queues[tx_queue_id];

> +	q->internals = internals;

> +

> +	dev->data->tx_queues[tx_queue_id] = q;

> +

> +	return 0;

> +}

> +

> +static void

> +eth_kni_queue_release(void *q __rte_unused)

> +{

> +}

> +

> +static int

> +eth_kni_link_update(struct rte_eth_dev *dev __rte_unused,

> +		int wait_to_complete __rte_unused)

> +{

> +	return 0;

> +}

> +

> +static void

> +eth_kni_stats_get(struct rte_eth_dev *dev, struct rte_eth_stats *stats)

> +{

> +	unsigned long rx_packets_total = 0, rx_bytes_total = 0;

> +	unsigned long tx_packets_total = 0, tx_bytes_total = 0;

> +	struct rte_eth_dev_data *data = dev->data;

> +	unsigned long tx_packets_err_total = 0;

> +	unsigned int i, num_stats;

> +	struct pmd_queue *q;

> +

> +	num_stats = RTE_MIN((unsigned

> int)RTE_ETHDEV_QUEUE_STAT_CNTRS,

> +			data->nb_rx_queues);

> +	for (i = 0; i < num_stats; i++) {

> +		q = data->rx_queues[i];

> +		stats->q_ipackets[i] = q->rx.pkts;

> +		stats->q_ibytes[i] = q->rx.bytes;

> +		rx_packets_total += stats->q_ipackets[i];

> +		rx_bytes_total += stats->q_ibytes[i];

> +	}

> +

> +	num_stats = RTE_MIN((unsigned

> int)RTE_ETHDEV_QUEUE_STAT_CNTRS,

> +			data->nb_tx_queues);

> +	for (i = 0; i < num_stats; i++) {

> +		q = data->tx_queues[i];

> +		stats->q_opackets[i] = q->tx.pkts;

> +		stats->q_obytes[i] = q->tx.bytes;

> +		stats->q_errors[i] = q->tx.err_pkts;

> +		tx_packets_total += stats->q_opackets[i];

> +		tx_bytes_total += stats->q_obytes[i];

> +		tx_packets_err_total += stats->q_errors[i];

> +	}

> +

> +	stats->ipackets = rx_packets_total;

> +	stats->ibytes = rx_bytes_total;

> +	stats->opackets = tx_packets_total;

> +	stats->obytes = tx_bytes_total;

> +	stats->oerrors = tx_packets_err_total;

> +}

> +

> +static void

> +eth_kni_stats_reset(struct rte_eth_dev *dev)

> +{

> +	struct rte_eth_dev_data *data = dev->data;

> +	struct pmd_queue *q;

> +	unsigned int i;

> +

> +	for (i = 0; i < data->nb_rx_queues; i++) {

> +		q = data->rx_queues[i];

> +		q->rx.pkts = 0;

> +		q->rx.bytes = 0;

> +	}

> +	for (i = 0; i < data->nb_tx_queues; i++) {

> +		q = data->tx_queues[i];

> +		q->tx.pkts = 0;

> +		q->tx.bytes = 0;

> +		q->tx.err_pkts = 0;

> +	}

> +}

> +

> +static const struct eth_dev_ops eth_kni_ops = {

> +	.dev_start = eth_kni_dev_start,

> +	.dev_stop = eth_kni_dev_stop,

> +	.dev_configure = eth_kni_dev_configure,

> +	.dev_infos_get = eth_kni_dev_info,

> +	.rx_queue_setup = eth_kni_rx_queue_setup,

> +	.tx_queue_setup = eth_kni_tx_queue_setup,

> +	.rx_queue_release = eth_kni_queue_release,

> +	.tx_queue_release = eth_kni_queue_release,

> +	.link_update = eth_kni_link_update,

> +	.stats_get = eth_kni_stats_get,

> +	.stats_reset = eth_kni_stats_reset,

> +};

> +

> +static struct rte_vdev_driver eth_kni_drv;

> +

> +static struct rte_eth_dev *

> +eth_kni_create(const char *name, struct eth_kni_args *args,

> +		unsigned int numa_node)

> +{

> +	struct pmd_internals *internals = NULL;

> +	struct rte_eth_dev_data *data;

> +	struct rte_eth_dev *eth_dev;

> +

> +	RTE_LOG(INFO, PMD, "Creating kni ethdev on numa socket %u\n",

> +			numa_node);

> +

> +	data = rte_zmalloc_socket(name, sizeof(*data), 0, numa_node);

> +	if (data == NULL)

> +		goto error;

> +

> +	internals = rte_zmalloc_socket(name, sizeof(*internals), 0,

> numa_node);

> +	if (internals == NULL)

> +		goto error;

> +

> +	/* reserve an ethdev entry */

> +	eth_dev = rte_eth_dev_allocate(name);

> +	if (eth_dev == NULL)

> +		goto error;

> +

> +	data->dev_private = internals;

> +	data->port_id = eth_dev->data->port_id;

> +	memmove(data->name, eth_dev->data->name, sizeof(data-

> >name));

> +	data->nb_rx_queues = 1;

> +	data->nb_tx_queues = 1;

> +	data->dev_link = pmd_link;

> +	data->mac_addrs = &internals->eth_addr;

> +

> +	eth_random_addr(internals->eth_addr.addr_bytes);

> +

> +	eth_dev->data = data;

> +	eth_dev->dev_ops = &eth_kni_ops;

> +	eth_dev->driver = NULL;

> +

> +	data->dev_flags = RTE_ETH_DEV_DETACHABLE;

> +	data->kdrv = RTE_KDRV_NONE;

> +	data->drv_name = eth_kni_drv.driver.name;

> +	data->numa_node = numa_node;

> +

> +	internals->no_request_thread = args->no_request_thread;

> +

> +	return eth_dev;

> +

> +error:

> +	rte_free(data);

> +	rte_free(internals);

> +

> +	return NULL;

> +}

> +

> +static int

> +kni_init(void)

> +{

> +	if (is_kni_initialized == 0)

> +		rte_kni_init(MAX_KNI_PORTS);

> +

> +	is_kni_initialized++;

> +

> +	return 0;

> +}

> +

> +static int

> +eth_kni_kvargs_process(struct eth_kni_args *args, const char *params)

> +{

> +	struct rte_kvargs *kvlist;

> +

> +	kvlist = rte_kvargs_parse(params, valid_arguments);

> +	if (kvlist == NULL)

> +		return -1;

> +

> +	memset(args, 0, sizeof(struct eth_kni_args));

> +

> +	if (rte_kvargs_count(kvlist, ETH_KNI_NO_REQUEST_THREAD_ARG)

> == 1)

> +		args->no_request_thread = 1;

> +

> +	rte_kvargs_free(kvlist);

> +

> +	return 0;

> +}

> +

> +static int

> +eth_kni_probe(const char *name, const char *params)

> +{

> +	struct rte_eth_dev *eth_dev;

> +	struct eth_kni_args args;

> +	int ret;

> +

> +	RTE_LOG(INFO, PMD, "Initializing eth_kni for %s\n", name);

> +

> +	ret = eth_kni_kvargs_process(&args, params);

> +	if (ret < 0)

> +		return ret;

> +

> +	ret = kni_init();

> +	if (ret < 0)

> +		return ret;

> +

> +	eth_dev = eth_kni_create(name, &args, rte_socket_id());

> +	if (eth_dev == NULL)

> +		goto kni_uninit;

> +

> +	eth_dev->rx_pkt_burst = eth_kni_rx;

> +	eth_dev->tx_pkt_burst = eth_kni_tx;

> +

> +	return 0;

> +

> +kni_uninit:

> +	is_kni_initialized--;

> +	if (is_kni_initialized == 0)

> +		rte_kni_close();

> +	return -1;

> +}

> +

> +static int

> +eth_kni_remove(const char *name)

> +{

> +	struct rte_eth_dev *eth_dev;

> +	struct pmd_internals *internals;

> +

> +	RTE_LOG(INFO, PMD, "Un-Initializing eth_kni for %s\n", name);

> +

> +	/* find the ethdev entry */

> +	eth_dev = rte_eth_dev_allocated(name);

> +	if (eth_dev == NULL)

> +		return -1;

> +

> +	eth_kni_dev_stop(eth_dev);

> +

> +	if (eth_dev->data) {

> +		internals = eth_dev->data->dev_private;

> +		rte_kni_release(internals->kni);

> +

> +		rte_free(internals);

> +	}

> +	rte_free(eth_dev->data);

> +

> +	rte_eth_dev_release_port(eth_dev);

> +

> +	is_kni_initialized--;

> +	if (is_kni_initialized == 0)

> +		rte_kni_close();

> +

> +	return 0;

> +}

> +

> +static struct rte_vdev_driver eth_kni_drv = {

> +	.probe = eth_kni_probe,

> +	.remove = eth_kni_remove,

> +};

> +

> +RTE_PMD_REGISTER_VDEV(net_kni, eth_kni_drv);

> +RTE_PMD_REGISTER_PARAM_STRING(net_kni,

> ETH_KNI_NO_REQUEST_THREAD_ARG "=<int>");

> diff --git a/drivers/net/kni/rte_pmd_kni_version.map

> b/drivers/net/kni/rte_pmd_kni_version.map

> new file mode 100644

> index 0000000..31eca32

> --- /dev/null

> +++ b/drivers/net/kni/rte_pmd_kni_version.map

> @@ -0,0 +1,4 @@

> +DPDK_17.02 {

> +

> +	local: *;

> +};

> diff --git a/mk/rte.app.mk b/mk/rte.app.mk

> index a5daa84..2b3a53d 100644

> --- a/mk/rte.app.mk

> +++ b/mk/rte.app.mk

> @@ -59,11 +59,6 @@ _LDLIBS-y += -L$(RTE_SDK_BIN)/lib

>  #

>  # Order is important: from higher level to lower level

>  #

> -

> -ifeq ($(CONFIG_RTE_EXEC_ENV_LINUXAPP),y)

> -_LDLIBS-$(CONFIG_RTE_LIBRTE_KNI)            += -lrte_kni

> -endif

> -

>  _LDLIBS-$(CONFIG_RTE_LIBRTE_PIPELINE)       += -lrte_pipeline

>  _LDLIBS-$(CONFIG_RTE_LIBRTE_TABLE)          += -lrte_table

>  _LDLIBS-$(CONFIG_RTE_LIBRTE_PORT)           += -lrte_port

> @@ -84,6 +79,10 @@ _LDLIBS-$(CONFIG_RTE_LIBRTE_POWER)          += -

> lrte_power

> 

>  _LDLIBS-y += --whole-archive

> 

> +ifeq ($(CONFIG_RTE_EXEC_ENV_LINUXAPP),y)

> +_LDLIBS-$(CONFIG_RTE_LIBRTE_KNI)            += -lrte_kni

> +endif

> +

>  _LDLIBS-$(CONFIG_RTE_LIBRTE_TIMER)          += -lrte_timer

>  _LDLIBS-$(CONFIG_RTE_LIBRTE_HASH)           += -lrte_hash

>  _LDLIBS-$(CONFIG_RTE_LIBRTE_EFD)            += -lrte_efd

> @@ -116,6 +115,7 @@ _LDLIBS-$(CONFIG_RTE_LIBRTE_ENIC_PMD)       += -

> lrte_pmd_enic

>  _LDLIBS-$(CONFIG_RTE_LIBRTE_FM10K_PMD)      += -lrte_pmd_fm10k

>  _LDLIBS-$(CONFIG_RTE_LIBRTE_I40E_PMD)       += -lrte_pmd_i40e

>  _LDLIBS-$(CONFIG_RTE_LIBRTE_IXGBE_PMD)      += -lrte_pmd_ixgbe

> +_LDLIBS-$(CONFIG_RTE_LIBRTE_PMD_KNI)        += -lrte_pmd_kni

>  _LDLIBS-$(CONFIG_RTE_LIBRTE_MLX4_PMD)       += -lrte_pmd_mlx4 -

> libverbs

>  _LDLIBS-$(CONFIG_RTE_LIBRTE_MLX5_PMD)       += -lrte_pmd_mlx5 -

> libverbs

>  _LDLIBS-$(CONFIG_RTE_LIBRTE_MPIPE_PMD)      += -lrte_pmd_mpipe -lgxio

> --

> 2.9.3
Ferruh Yigit Jan. 30, 2017, 7:43 p.m. UTC | #2
On 1/30/2017 7:05 PM, Yong Wang wrote:
>> -----Original Message-----
>> From: Ferruh Yigit [mailto:ferruh.yigit@intel.com]
>> Sent: Monday, January 30, 2017 8:58 AM
>> To: Thomas Monjalon <thomas.monjalon@6wind.com>; Yong Wang
>> <yongwang@vmware.com>
>> Cc: dev@dpdk.org; Ferruh Yigit <ferruh.yigit@intel.com>
>> Subject: [PATCH v5] net/kni: add KNI PMD
>>
>> Add KNI PMD which wraps librte_kni for ease of use.
>>
>> KNI PMD can be used as any regular PMD to send / receive packets to the
>> Linux networking stack.
>>
>> Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
>> ---
> 
> Looks good except a few typos in the documentation added.

Thank you for the review, I will send a new version with them fixed.

> 
> Reviewed-by: Yong Wang <yongwang@vmware.com>
> 
>>

<...>
diff mbox

Patch

diff --git a/MAINTAINERS b/MAINTAINERS
index f071138..8eb83f5 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -404,6 +404,11 @@  M: Keith Wiles <keith.wiles@intel.com>
 F: drivers/net/tap/
 F: doc/guides/nics/tap.rst
 
+KNI PMD
+M: Ferruh Yigit <ferruh.yigit@intel.com>
+F: drivers/net/kni/
+F: doc/guides/nics/kni.rst
+
 Ring PMD
 M: Bruce Richardson <bruce.richardson@intel.com>
 F: drivers/net/ring/
diff --git a/config/common_base b/config/common_base
index 61efb87..2e1bbd5 100644
--- a/config/common_base
+++ b/config/common_base
@@ -576,6 +576,7 @@  CONFIG_RTE_PIPELINE_STATS_COLLECT=n
 # Compile librte_kni
 #
 CONFIG_RTE_LIBRTE_KNI=n
+CONFIG_RTE_LIBRTE_PMD_KNI=n
 CONFIG_RTE_KNI_KMOD=n
 CONFIG_RTE_KNI_KMOD_ETHTOOL=n
 CONFIG_RTE_KNI_PREEMPT_DEFAULT=y
diff --git a/config/common_linuxapp b/config/common_linuxapp
index 00ebaac..d03a60a 100644
--- a/config/common_linuxapp
+++ b/config/common_linuxapp
@@ -39,6 +39,7 @@  CONFIG_RTE_EAL_IGB_UIO=y
 CONFIG_RTE_EAL_VFIO=y
 CONFIG_RTE_KNI_KMOD=y
 CONFIG_RTE_LIBRTE_KNI=y
+CONFIG_RTE_LIBRTE_PMD_KNI=y
 CONFIG_RTE_LIBRTE_VHOST=y
 CONFIG_RTE_LIBRTE_PMD_VHOST=y
 CONFIG_RTE_LIBRTE_PMD_AF_PACKET=y
diff --git a/doc/guides/nics/features/kni.ini b/doc/guides/nics/features/kni.ini
new file mode 100644
index 0000000..6deb66a
--- /dev/null
+++ b/doc/guides/nics/features/kni.ini
@@ -0,0 +1,7 @@ 
+;
+; Supported features of the 'kni' network poll mode driver.
+;
+; Refer to default.ini for the full list of available PMD features.
+;
+[Features]
+Usage doc            = Y
diff --git a/doc/guides/nics/index.rst b/doc/guides/nics/index.rst
index 87f9334..5248625 100644
--- a/doc/guides/nics/index.rst
+++ b/doc/guides/nics/index.rst
@@ -46,6 +46,7 @@  Network Interface Controller Drivers
     i40e
     ixgbe
     intel_vf
+    kni
     mlx4
     mlx5
     nfp
diff --git a/doc/guides/nics/kni.rst b/doc/guides/nics/kni.rst
new file mode 100644
index 0000000..2e2032b
--- /dev/null
+++ b/doc/guides/nics/kni.rst
@@ -0,0 +1,197 @@ 
+..  BSD LICENSE
+    Copyright(c) 2017 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
+    are met:
+
+    * 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
+    distribution.
+    * 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
+    OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+KNI Poll Mode Driver
+======================
+
+KNI PMD is wrapper to the :ref:`librte_kni <kni>` library.
+
+This PMD enables using KNI without having a KNI specific application,
+any forwarding application can use PMD interface for KNI.
+
+Sending packets to the any DPDK controlled interface or sending to the
+Linux networking stack will be transparent to the DPDK application.
+
+To create a KNI device ``net_kni#`` device name should be used, and this
+will create ``kni#`` Linux virtual network interface.
+
+There is no physical device backend for the virtual KNI device.
+
+Packets sent to the KNI Linux interface will be received by the DPDK
+application, and DPDK application may forward packets to a physical NIC
+or to a virtual device (like another KNI interface or PCAP interface).
+
+To forward any traffic from physical NIC to the Linux networking stack,
+an application should control a physical port and create one virtual KNI port,
+and forward between two.
+
+Using this PMD requires KNI kernel module be inserted.
+
+
+Usage
+-----
+
+EAL ``--vdev`` argument can be used to create KNI device instance, like::
+
+        testpmd --vdev=net_kni0 --vdev=net_kn1 -- -i
+
+Above command will create ``kni0`` and ``kni1`` Linux network interfaces,
+those interfaces can be controlled by standard Linux tools.
+
+When testpmd forwarding started, any packets send to ``kni0`` interface
+forwarded to the ``kni1`` interface and vice versa.
+
+There is no hard limit on number of interfaces can be created.
+
+
+Default interface configuration
+-------------------------------
+
+``librte_kni`` can create Linux network interfaces with different features,
+feature set controlled by a configuration struct, and KNI PMD uses a fixed
+configuration:
+
+    .. code-block:: console
+
+        Interface name: kni#
+        force bind kernel thread to a core : NO
+        mbuf size: MAX_PACKET_SZ
+
+KNI control path is not supported with the PMD, since there is no physical
+backend device by default.
+
+
+PMD arguments
+-------------
+
+``no_request_thread``, by default PMD creates a phtread for each KNI interface
+to handle Linux network interface control commands, like ``ifconfig kni0 up``
+
+With ``no_request_thread`` option, pthread is not created and control commands
+not handled by PMD.
+
+By default request thread is enabled. And this argument should not be used
+most of the time, unless this PMD used with customized DPDK application to handle
+requests itself.
+
+Argument usage::
+
+        testpmd --vdev "net_kni0,no_request_thread=1" -- -i
+
+
+PMD log messages
+----------------
+
+If KNI kernel module (rte_kni.ko) not inserted, following error log printed::
+
+        "KNI: KNI subsystem has not been initialized. Invoke rte_kni_init() first"
+
+
+PMD testing
+-----------
+
+It is possible to test PMD quicly using KNI kernel module loopback feature:
+
+* Insert KNI kernel module with loopback support:
+
+    .. code-block:: console
+
+        insmod build/kmod/rte_kni.ko lo_mode=lo_mode_fifo_skb
+
+* Start testpmd with no physical device but two KNI virtual devices:
+
+    .. code-block:: console
+
+        ./testpmd --vdev net_kni0 --vdev net_kni1 -- -i
+
+    .. code-block:: console
+
+        ...
+        Configuring Port 0 (socket 0)
+        KNI: pci: 00:00:00       c580:b8
+        Port 0: 1A:4A:5B:7C:A2:8C
+        Configuring Port 1 (socket 0)
+        KNI: pci: 00:00:00       600:b9
+        Port 1: AE:95:21:07:93:DD
+        Checking link statuses...
+        Port 0 Link Up - speed 10000 Mbps - full-duplex
+        Port 1 Link Up - speed 10000 Mbps - full-duplex
+        Done
+        testpmd>
+
+* Observe Linux interfaces
+
+    .. code-block:: console
+
+        $ ifconfig kni0 && ifconfig kni1
+        kni0: flags=4098<BROADCAST,MULTICAST>  mtu 1500
+                ether ae:8e:79:8e:9b:c8  txqueuelen 1000  (Ethernet)
+                RX packets 0  bytes 0 (0.0 B)
+                RX errors 0  dropped 0  overruns 0  frame 0
+                TX packets 0  bytes 0 (0.0 B)
+                TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0
+
+        kni1: flags=4098<BROADCAST,MULTICAST>  mtu 1500
+                ether 9e:76:43:53:3e:9b  txqueuelen 1000  (Ethernet)
+                RX packets 0  bytes 0 (0.0 B)
+                RX errors 0  dropped 0  overruns 0  frame 0
+                TX packets 0  bytes 0 (0.0 B)
+                TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0
+
+
+* Start forwarding with tx_first:
+
+    .. code-block:: console
+
+        testpmd> start tx_first
+
+* Quit and check forwarding stats:
+
+    .. code-block:: console
+
+        testpmd> quit
+        Telling cores to stop...
+        Waiting for lcores to finish...
+
+        ---------------------- Forward statistics for port 0  ----------------------
+        RX-packets: 35637905       RX-dropped: 0             RX-total: 35637905
+        TX-packets: 35637947       TX-dropped: 0             TX-total: 35637947
+        ----------------------------------------------------------------------------
+
+        ---------------------- Forward statistics for port 1  ----------------------
+        RX-packets: 35637915       RX-dropped: 0             RX-total: 35637915
+        TX-packets: 35637937       TX-dropped: 0             TX-total: 35637937
+        ----------------------------------------------------------------------------
+
+        +++++++++++++++ Accumulated forward statistics for all ports+++++++++++++++
+        RX-packets: 71275820       RX-dropped: 0             RX-total: 71275820
+        TX-packets: 71275884       TX-dropped: 0             TX-total: 71275884
+        ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
+
diff --git a/drivers/net/Makefile b/drivers/net/Makefile
index 40fc333..8fd6745 100644
--- a/drivers/net/Makefile
+++ b/drivers/net/Makefile
@@ -41,6 +41,7 @@  DIRS-$(CONFIG_RTE_LIBRTE_ENIC_PMD) += enic
 DIRS-$(CONFIG_RTE_LIBRTE_FM10K_PMD) += fm10k
 DIRS-$(CONFIG_RTE_LIBRTE_I40E_PMD) += i40e
 DIRS-$(CONFIG_RTE_LIBRTE_IXGBE_PMD) += ixgbe
+DIRS-$(CONFIG_RTE_LIBRTE_PMD_KNI) += kni
 DIRS-$(CONFIG_RTE_LIBRTE_MLX4_PMD) += mlx4
 DIRS-$(CONFIG_RTE_LIBRTE_MLX5_PMD) += mlx5
 DIRS-$(CONFIG_RTE_LIBRTE_MPIPE_PMD) += mpipe
diff --git a/drivers/net/kni/Makefile b/drivers/net/kni/Makefile
new file mode 100644
index 0000000..b3017b1
--- /dev/null
+++ b/drivers/net/kni/Makefile
@@ -0,0 +1,64 @@ 
+#   BSD LICENSE
+#
+#   Copyright(c) 2017 Intel Corporation. All rights reserved.
+#
+#   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
+#       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
+#       distribution.
+#     * 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
+#   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+include $(RTE_SDK)/mk/rte.vars.mk
+
+#
+# library name
+#
+LIB = librte_pmd_kni.a
+
+CFLAGS += -O3
+CFLAGS += $(WERROR_FLAGS)
+LDLIBS += -lpthread
+
+EXPORT_MAP := rte_pmd_kni_version.map
+
+LIBABIVER := 1
+
+#
+# all source are stored in SRCS-y
+#
+SRCS-$(CONFIG_RTE_LIBRTE_PMD_KNI) += rte_eth_kni.c
+
+#
+# Export include files
+#
+SYMLINK-y-include +=
+
+# this lib depends upon:
+DEPDIRS-$(CONFIG_RTE_LIBRTE_PMD_KNI) += lib/librte_eal
+DEPDIRS-$(CONFIG_RTE_LIBRTE_PMD_KNI) += lib/librte_ether
+DEPDIRS-$(CONFIG_RTE_LIBRTE_PMD_KNI) += lib/librte_kni
+DEPDIRS-$(CONFIG_RTE_LIBRTE_PMD_KNI) += lib/librte_kvargs
+DEPDIRS-$(CONFIG_RTE_LIBRTE_PMD_KNI) += lib/librte_mbuf
+DEPDIRS-$(CONFIG_RTE_LIBRTE_PMD_KNI) += lib/librte_mempool
+
+include $(RTE_SDK)/mk/rte.lib.mk
diff --git a/drivers/net/kni/rte_eth_kni.c b/drivers/net/kni/rte_eth_kni.c
new file mode 100644
index 0000000..59205e7
--- /dev/null
+++ b/drivers/net/kni/rte_eth_kni.c
@@ -0,0 +1,515 @@ 
+/*-
+ *   BSD LICENSE
+ *
+ *   Copyright(c) 2017 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
+ *   are met:
+ *
+ *     * 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
+ *       distribution.
+ *     * 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
+ *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <fcntl.h>
+#include <pthread.h>
+#include <unistd.h>
+
+#include <rte_ethdev.h>
+#include <rte_kni.h>
+#include <rte_kvargs.h>
+#include <rte_malloc.h>
+#include <rte_vdev.h>
+
+/* Only single queue supported */
+#define KNI_MAX_QUEUE_PER_PORT 1
+
+#define MAX_PACKET_SZ 2048
+#define MAX_KNI_PORTS 8
+
+#define ETH_KNI_NO_REQUEST_THREAD_ARG	"no_request_thread"
+static const char * const valid_arguments[] = {
+	ETH_KNI_NO_REQUEST_THREAD_ARG,
+	NULL
+};
+
+struct eth_kni_args {
+	int no_request_thread;
+};
+
+struct pmd_queue_stats {
+	uint64_t pkts;
+	uint64_t bytes;
+	uint64_t err_pkts;
+};
+
+struct pmd_queue {
+	struct pmd_internals *internals;
+	struct rte_mempool *mb_pool;
+
+	struct pmd_queue_stats rx;
+	struct pmd_queue_stats tx;
+};
+
+struct pmd_internals {
+	struct rte_kni *kni;
+	int is_kni_started;
+
+	pthread_t thread;
+	int stop_thread;
+	int no_request_thread;
+
+	struct ether_addr eth_addr;
+
+	struct pmd_queue rx_queues[KNI_MAX_QUEUE_PER_PORT];
+	struct pmd_queue tx_queues[KNI_MAX_QUEUE_PER_PORT];
+};
+
+static const struct rte_eth_link pmd_link = {
+		.link_speed = ETH_SPEED_NUM_10G,
+		.link_duplex = ETH_LINK_FULL_DUPLEX,
+		.link_status = ETH_LINK_DOWN,
+		.link_autoneg = ETH_LINK_SPEED_AUTONEG,
+};
+static int is_kni_initialized;
+
+static uint16_t
+eth_kni_rx(void *q, struct rte_mbuf **bufs, uint16_t nb_bufs)
+{
+	struct pmd_queue *kni_q = q;
+	struct rte_kni *kni = kni_q->internals->kni;
+	uint16_t nb_pkts;
+
+	nb_pkts = rte_kni_rx_burst(kni, bufs, nb_bufs);
+
+	kni_q->rx.pkts += nb_pkts;
+	kni_q->rx.err_pkts += nb_bufs - nb_pkts;
+
+	return nb_pkts;
+}
+
+static uint16_t
+eth_kni_tx(void *q, struct rte_mbuf **bufs, uint16_t nb_bufs)
+{
+	struct pmd_queue *kni_q = q;
+	struct rte_kni *kni = kni_q->internals->kni;
+	uint16_t nb_pkts;
+
+	nb_pkts =  rte_kni_tx_burst(kni, bufs, nb_bufs);
+
+	kni_q->tx.pkts += nb_pkts;
+	kni_q->tx.err_pkts += nb_bufs - nb_pkts;
+
+	return nb_pkts;
+}
+
+static void *
+kni_handle_request(void *param)
+{
+	struct pmd_internals *internals = param;
+#define MS 1000
+
+	while (!internals->stop_thread) {
+		rte_kni_handle_request(internals->kni);
+		usleep(500 * MS);
+	}
+
+	return param;
+}
+
+static int
+eth_kni_start(struct rte_eth_dev *dev)
+{
+	struct pmd_internals *internals = dev->data->dev_private;
+	uint16_t port_id = dev->data->port_id;
+	struct rte_mempool *mb_pool;
+	struct rte_kni_conf conf;
+	const char *name = dev->data->name + 4; /* remove net_ */
+
+	snprintf(conf.name, RTE_KNI_NAMESIZE, "%s", name);
+	conf.force_bind = 0;
+	conf.group_id = port_id;
+	conf.mbuf_size = MAX_PACKET_SZ;
+	mb_pool = internals->rx_queues[0].mb_pool;
+
+	internals->kni = rte_kni_alloc(mb_pool, &conf, NULL);
+	if (internals->kni == NULL) {
+		RTE_LOG(ERR, PMD,
+			"Fail to create kni interface for port: %d\n",
+			port_id);
+		return -1;
+	}
+
+	return 0;
+}
+
+static int
+eth_kni_dev_start(struct rte_eth_dev *dev)
+{
+	struct pmd_internals *internals = dev->data->dev_private;
+	int ret;
+
+	if (internals->is_kni_started == 0) {
+		ret = eth_kni_start(dev);
+		if (ret)
+			return -1;
+		internals->is_kni_started = 1;
+	}
+
+	if (internals->no_request_thread == 0) {
+		ret = pthread_create(&internals->thread, NULL,
+			kni_handle_request, internals);
+		if (ret) {
+			RTE_LOG(ERR, PMD,
+				"Fail to create kni request thread\n");
+			return -1;
+		}
+	}
+
+	dev->data->dev_link.link_status = 1;
+
+	return 0;
+}
+
+static void
+eth_kni_dev_stop(struct rte_eth_dev *dev)
+{
+	struct pmd_internals *internals = dev->data->dev_private;
+	int ret;
+
+	if (internals->no_request_thread == 0) {
+		internals->stop_thread = 1;
+
+		ret = pthread_cancel(internals->thread);
+		if (ret)
+			RTE_LOG(ERR, PMD, "Can't cancel the thread\n");
+
+		ret = pthread_join(internals->thread, NULL);
+		if (ret)
+			RTE_LOG(ERR, PMD, "Can't join the thread\n");
+
+		internals->stop_thread = 0;
+	}
+
+	dev->data->dev_link.link_status = 0;
+}
+
+static int
+eth_kni_dev_configure(struct rte_eth_dev *dev __rte_unused)
+{
+	return 0;
+}
+
+static void
+eth_kni_dev_info(struct rte_eth_dev *dev __rte_unused,
+		struct rte_eth_dev_info *dev_info)
+{
+	dev_info->max_mac_addrs = 1;
+	dev_info->max_rx_pktlen = UINT32_MAX;
+	dev_info->max_rx_queues = KNI_MAX_QUEUE_PER_PORT;
+	dev_info->max_tx_queues = KNI_MAX_QUEUE_PER_PORT;
+	dev_info->min_rx_bufsize = 0;
+	dev_info->pci_dev = NULL;
+}
+
+static int
+eth_kni_rx_queue_setup(struct rte_eth_dev *dev,
+		uint16_t rx_queue_id,
+		uint16_t nb_rx_desc __rte_unused,
+		unsigned int socket_id __rte_unused,
+		const struct rte_eth_rxconf *rx_conf __rte_unused,
+		struct rte_mempool *mb_pool)
+{
+	struct pmd_internals *internals = dev->data->dev_private;
+	struct pmd_queue *q;
+
+	q = &internals->rx_queues[rx_queue_id];
+	q->internals = internals;
+	q->mb_pool = mb_pool;
+
+	dev->data->rx_queues[rx_queue_id] = q;
+
+	return 0;
+}
+
+static int
+eth_kni_tx_queue_setup(struct rte_eth_dev *dev,
+		uint16_t tx_queue_id,
+		uint16_t nb_tx_desc __rte_unused,
+		unsigned int socket_id __rte_unused,
+		const struct rte_eth_txconf *tx_conf __rte_unused)
+{
+	struct pmd_internals *internals = dev->data->dev_private;
+	struct pmd_queue *q;
+
+	q = &internals->tx_queues[tx_queue_id];
+	q->internals = internals;
+
+	dev->data->tx_queues[tx_queue_id] = q;
+
+	return 0;
+}
+
+static void
+eth_kni_queue_release(void *q __rte_unused)
+{
+}
+
+static int
+eth_kni_link_update(struct rte_eth_dev *dev __rte_unused,
+		int wait_to_complete __rte_unused)
+{
+	return 0;
+}
+
+static void
+eth_kni_stats_get(struct rte_eth_dev *dev, struct rte_eth_stats *stats)
+{
+	unsigned long rx_packets_total = 0, rx_bytes_total = 0;
+	unsigned long tx_packets_total = 0, tx_bytes_total = 0;
+	struct rte_eth_dev_data *data = dev->data;
+	unsigned long tx_packets_err_total = 0;
+	unsigned int i, num_stats;
+	struct pmd_queue *q;
+
+	num_stats = RTE_MIN((unsigned int)RTE_ETHDEV_QUEUE_STAT_CNTRS,
+			data->nb_rx_queues);
+	for (i = 0; i < num_stats; i++) {
+		q = data->rx_queues[i];
+		stats->q_ipackets[i] = q->rx.pkts;
+		stats->q_ibytes[i] = q->rx.bytes;
+		rx_packets_total += stats->q_ipackets[i];
+		rx_bytes_total += stats->q_ibytes[i];
+	}
+
+	num_stats = RTE_MIN((unsigned int)RTE_ETHDEV_QUEUE_STAT_CNTRS,
+			data->nb_tx_queues);
+	for (i = 0; i < num_stats; i++) {
+		q = data->tx_queues[i];
+		stats->q_opackets[i] = q->tx.pkts;
+		stats->q_obytes[i] = q->tx.bytes;
+		stats->q_errors[i] = q->tx.err_pkts;
+		tx_packets_total += stats->q_opackets[i];
+		tx_bytes_total += stats->q_obytes[i];
+		tx_packets_err_total += stats->q_errors[i];
+	}
+
+	stats->ipackets = rx_packets_total;
+	stats->ibytes = rx_bytes_total;
+	stats->opackets = tx_packets_total;
+	stats->obytes = tx_bytes_total;
+	stats->oerrors = tx_packets_err_total;
+}
+
+static void
+eth_kni_stats_reset(struct rte_eth_dev *dev)
+{
+	struct rte_eth_dev_data *data = dev->data;
+	struct pmd_queue *q;
+	unsigned int i;
+
+	for (i = 0; i < data->nb_rx_queues; i++) {
+		q = data->rx_queues[i];
+		q->rx.pkts = 0;
+		q->rx.bytes = 0;
+	}
+	for (i = 0; i < data->nb_tx_queues; i++) {
+		q = data->tx_queues[i];
+		q->tx.pkts = 0;
+		q->tx.bytes = 0;
+		q->tx.err_pkts = 0;
+	}
+}
+
+static const struct eth_dev_ops eth_kni_ops = {
+	.dev_start = eth_kni_dev_start,
+	.dev_stop = eth_kni_dev_stop,
+	.dev_configure = eth_kni_dev_configure,
+	.dev_infos_get = eth_kni_dev_info,
+	.rx_queue_setup = eth_kni_rx_queue_setup,
+	.tx_queue_setup = eth_kni_tx_queue_setup,
+	.rx_queue_release = eth_kni_queue_release,
+	.tx_queue_release = eth_kni_queue_release,
+	.link_update = eth_kni_link_update,
+	.stats_get = eth_kni_stats_get,
+	.stats_reset = eth_kni_stats_reset,
+};
+
+static struct rte_vdev_driver eth_kni_drv;
+
+static struct rte_eth_dev *
+eth_kni_create(const char *name, struct eth_kni_args *args,
+		unsigned int numa_node)
+{
+	struct pmd_internals *internals = NULL;
+	struct rte_eth_dev_data *data;
+	struct rte_eth_dev *eth_dev;
+
+	RTE_LOG(INFO, PMD, "Creating kni ethdev on numa socket %u\n",
+			numa_node);
+
+	data = rte_zmalloc_socket(name, sizeof(*data), 0, numa_node);
+	if (data == NULL)
+		goto error;
+
+	internals = rte_zmalloc_socket(name, sizeof(*internals), 0, numa_node);
+	if (internals == NULL)
+		goto error;
+
+	/* reserve an ethdev entry */
+	eth_dev = rte_eth_dev_allocate(name);
+	if (eth_dev == NULL)
+		goto error;
+
+	data->dev_private = internals;
+	data->port_id = eth_dev->data->port_id;
+	memmove(data->name, eth_dev->data->name, sizeof(data->name));
+	data->nb_rx_queues = 1;
+	data->nb_tx_queues = 1;
+	data->dev_link = pmd_link;
+	data->mac_addrs = &internals->eth_addr;
+
+	eth_random_addr(internals->eth_addr.addr_bytes);
+
+	eth_dev->data = data;
+	eth_dev->dev_ops = &eth_kni_ops;
+	eth_dev->driver = NULL;
+
+	data->dev_flags = RTE_ETH_DEV_DETACHABLE;
+	data->kdrv = RTE_KDRV_NONE;
+	data->drv_name = eth_kni_drv.driver.name;
+	data->numa_node = numa_node;
+
+	internals->no_request_thread = args->no_request_thread;
+
+	return eth_dev;
+
+error:
+	rte_free(data);
+	rte_free(internals);
+
+	return NULL;
+}
+
+static int
+kni_init(void)
+{
+	if (is_kni_initialized == 0)
+		rte_kni_init(MAX_KNI_PORTS);
+
+	is_kni_initialized++;
+
+	return 0;
+}
+
+static int
+eth_kni_kvargs_process(struct eth_kni_args *args, const char *params)
+{
+	struct rte_kvargs *kvlist;
+
+	kvlist = rte_kvargs_parse(params, valid_arguments);
+	if (kvlist == NULL)
+		return -1;
+
+	memset(args, 0, sizeof(struct eth_kni_args));
+
+	if (rte_kvargs_count(kvlist, ETH_KNI_NO_REQUEST_THREAD_ARG) == 1)
+		args->no_request_thread = 1;
+
+	rte_kvargs_free(kvlist);
+
+	return 0;
+}
+
+static int
+eth_kni_probe(const char *name, const char *params)
+{
+	struct rte_eth_dev *eth_dev;
+	struct eth_kni_args args;
+	int ret;
+
+	RTE_LOG(INFO, PMD, "Initializing eth_kni for %s\n", name);
+
+	ret = eth_kni_kvargs_process(&args, params);
+	if (ret < 0)
+		return ret;
+
+	ret = kni_init();
+	if (ret < 0)
+		return ret;
+
+	eth_dev = eth_kni_create(name, &args, rte_socket_id());
+	if (eth_dev == NULL)
+		goto kni_uninit;
+
+	eth_dev->rx_pkt_burst = eth_kni_rx;
+	eth_dev->tx_pkt_burst = eth_kni_tx;
+
+	return 0;
+
+kni_uninit:
+	is_kni_initialized--;
+	if (is_kni_initialized == 0)
+		rte_kni_close();
+	return -1;
+}
+
+static int
+eth_kni_remove(const char *name)
+{
+	struct rte_eth_dev *eth_dev;
+	struct pmd_internals *internals;
+
+	RTE_LOG(INFO, PMD, "Un-Initializing eth_kni for %s\n", name);
+
+	/* find the ethdev entry */
+	eth_dev = rte_eth_dev_allocated(name);
+	if (eth_dev == NULL)
+		return -1;
+
+	eth_kni_dev_stop(eth_dev);
+
+	if (eth_dev->data) {
+		internals = eth_dev->data->dev_private;
+		rte_kni_release(internals->kni);
+
+		rte_free(internals);
+	}
+	rte_free(eth_dev->data);
+
+	rte_eth_dev_release_port(eth_dev);
+
+	is_kni_initialized--;
+	if (is_kni_initialized == 0)
+		rte_kni_close();
+
+	return 0;
+}
+
+static struct rte_vdev_driver eth_kni_drv = {
+	.probe = eth_kni_probe,
+	.remove = eth_kni_remove,
+};
+
+RTE_PMD_REGISTER_VDEV(net_kni, eth_kni_drv);
+RTE_PMD_REGISTER_PARAM_STRING(net_kni, ETH_KNI_NO_REQUEST_THREAD_ARG "=<int>");
diff --git a/drivers/net/kni/rte_pmd_kni_version.map b/drivers/net/kni/rte_pmd_kni_version.map
new file mode 100644
index 0000000..31eca32
--- /dev/null
+++ b/drivers/net/kni/rte_pmd_kni_version.map
@@ -0,0 +1,4 @@ 
+DPDK_17.02 {
+
+	local: *;
+};
diff --git a/mk/rte.app.mk b/mk/rte.app.mk
index a5daa84..2b3a53d 100644
--- a/mk/rte.app.mk
+++ b/mk/rte.app.mk
@@ -59,11 +59,6 @@  _LDLIBS-y += -L$(RTE_SDK_BIN)/lib
 #
 # Order is important: from higher level to lower level
 #
-
-ifeq ($(CONFIG_RTE_EXEC_ENV_LINUXAPP),y)
-_LDLIBS-$(CONFIG_RTE_LIBRTE_KNI)            += -lrte_kni
-endif
-
 _LDLIBS-$(CONFIG_RTE_LIBRTE_PIPELINE)       += -lrte_pipeline
 _LDLIBS-$(CONFIG_RTE_LIBRTE_TABLE)          += -lrte_table
 _LDLIBS-$(CONFIG_RTE_LIBRTE_PORT)           += -lrte_port
@@ -84,6 +79,10 @@  _LDLIBS-$(CONFIG_RTE_LIBRTE_POWER)          += -lrte_power
 
 _LDLIBS-y += --whole-archive
 
+ifeq ($(CONFIG_RTE_EXEC_ENV_LINUXAPP),y)
+_LDLIBS-$(CONFIG_RTE_LIBRTE_KNI)            += -lrte_kni
+endif
+
 _LDLIBS-$(CONFIG_RTE_LIBRTE_TIMER)          += -lrte_timer
 _LDLIBS-$(CONFIG_RTE_LIBRTE_HASH)           += -lrte_hash
 _LDLIBS-$(CONFIG_RTE_LIBRTE_EFD)            += -lrte_efd
@@ -116,6 +115,7 @@  _LDLIBS-$(CONFIG_RTE_LIBRTE_ENIC_PMD)       += -lrte_pmd_enic
 _LDLIBS-$(CONFIG_RTE_LIBRTE_FM10K_PMD)      += -lrte_pmd_fm10k
 _LDLIBS-$(CONFIG_RTE_LIBRTE_I40E_PMD)       += -lrte_pmd_i40e
 _LDLIBS-$(CONFIG_RTE_LIBRTE_IXGBE_PMD)      += -lrte_pmd_ixgbe
+_LDLIBS-$(CONFIG_RTE_LIBRTE_PMD_KNI)        += -lrte_pmd_kni
 _LDLIBS-$(CONFIG_RTE_LIBRTE_MLX4_PMD)       += -lrte_pmd_mlx4 -libverbs
 _LDLIBS-$(CONFIG_RTE_LIBRTE_MLX5_PMD)       += -lrte_pmd_mlx5 -libverbs
 _LDLIBS-$(CONFIG_RTE_LIBRTE_MPIPE_PMD)      += -lrte_pmd_mpipe -lgxio