[dpdk-dev,3/3] vhost: Add VHOST PMD

  The patch introduces a new PMD. This PMD is implemented as thin wrapper
of librte_vhost. It means librte_vhost is also needed to compile the PMD.
The vhost messages will be handled only when a port is started. So start
a port first, then invoke QEMU.

The PMD has 2 parameters.
 - iface:  The parameter is used to specify a path to connect to a
           virtio-net device.
 - queues: The parameter is used to specify the number of the queues
           virtio-net device has.
           (Default: 1)

Here is an example.
$ ./testpmd -c f -n 4 --vdev 'eth_vhost0,iface=/tmp/sock0,queues=1' -- -i

To connect above testpmd, here is qemu command example.

$ qemu-system-x86_64 \
        <snip>
        -chardev socket,id=chr0,path=/tmp/sock0 \
        -netdev vhost-user,id=net0,chardev=chr0,vhostforce,queues=1 \
        -device virtio-net-pci,netdev=net0

Signed-off-by: Tetsuya Mukawa <mukawa@igel.co.jp>
---
 config/common_linuxapp                      |   6 +
 doc/guides/nics/index.rst                   |   1 +
 doc/guides/nics/vhost.rst                   |  82 +++
 doc/guides/rel_notes/release_2_2.rst        |   2 +
 drivers/net/Makefile                        |   4 +
 drivers/net/vhost/Makefile                  |  62 +++
 drivers/net/vhost/rte_eth_vhost.c           | 765 ++++++++++++++++++++++++++++
 drivers/net/vhost/rte_eth_vhost.h           |  65 +++
 drivers/net/vhost/rte_pmd_vhost_version.map |   8 +
 mk/rte.app.mk                               |   8 +-
 10 files changed, 1002 insertions(+), 1 deletion(-)
 create mode 100644 doc/guides/nics/vhost.rst
 create mode 100644 drivers/net/vhost/Makefile
 create mode 100644 drivers/net/vhost/rte_eth_vhost.c
 create mode 100644 drivers/net/vhost/rte_eth_vhost.h
 create mode 100644 drivers/net/vhost/rte_pmd_vhost_version.map
  

The patch introduces a new PMD. This PMD is implemented as thin wrapper
of librte_vhost. The patch will work on below patch series.
 - [PATCH v7 00/28] remove pci driver from vdevs

* Known issue.
We may see issues while handling RESET_OWNER message.
These handlings are done in vhost library, so not a part of vhost PMD.
So far, we are waiting for QEMU fixing.

PATCH v2 changes:
 - Remove a below patch that fixes vhost library.
   The patch was applied as a separate patch.
   - vhost: fix crash with multiqueue enabled
 - Fix typos.
   (Thanks to Thomas, Monjalon)
 - Rebase on latest tree with above bernard's patches.

PATCH v1 changes:
 - Support vhost multiple queues.
 - Rebase on "remove pci driver from vdevs".
 - Optimize RX/TX functions.
 - Fix resource leaks.
 - Fix compile issue.
 - Add patch to fix vhost library.

RFC PATCH v3 changes:
 - Optimize performance.
   In RX/TX functions, change code to access only per core data.
 - Add below API to allow user to use vhost library APIs for a port managed
   by vhost PMD. There are a few limitations. See "rte_eth_vhost.h".
    - rte_eth_vhost_portid2vdev()
   To support this functionality, vhost library is also changed.
   Anyway, if users doesn't use vhost PMD, can fully use vhost library APIs.
 - Add code to support vhost multiple queues.
   Actually, multiple queues functionality is not enabled so far.

RFC PATCH v2 changes:
 - Fix issues reported by checkpatch.pl
   (Thanks to Stephen Hemminger)


Tetsuya Mukawa (2):
  vhost: Add callback and private data for vhost PMD
  vhost: Add VHOST PMD

 config/common_linuxapp                        |   6 +
 doc/guides/nics/index.rst                     |   1 +
 doc/guides/nics/vhost.rst                     |  82 +++
 doc/guides/rel_notes/release_2_2.rst          |   2 +
 drivers/net/Makefile                          |   4 +
 drivers/net/vhost/Makefile                    |  62 +++
 drivers/net/vhost/rte_eth_vhost.c             | 765 ++++++++++++++++++++++++++
 drivers/net/vhost/rte_eth_vhost.h             |  65 +++
 drivers/net/vhost/rte_pmd_vhost_version.map   |   8 +
 lib/librte_vhost/rte_vhost_version.map        |   6 +
 lib/librte_vhost/rte_virtio_net.h             |   3 +
 lib/librte_vhost/vhost_user/virtio-net-user.c |  13 +-
 lib/librte_vhost/virtio-net.c                 |  56 +-
 lib/librte_vhost/virtio-net.h                 |   4 +-
 mk/rte.app.mk                                 |   8 +-
 15 files changed, 1072 insertions(+), 13 deletions(-)
 create mode 100644 doc/guides/nics/vhost.rst
 create mode 100644 drivers/net/vhost/Makefile
 create mode 100644 drivers/net/vhost/rte_eth_vhost.c
 create mode 100644 drivers/net/vhost/rte_eth_vhost.h
 create mode 100644 drivers/net/vhost/rte_pmd_vhost_version.map
  

Hi,

Could someone please review below patch series?

Regards,
Tetsuya

On 2015/11/02 12:58, Tetsuya Mukawa wrote:
> The patch introduces a new PMD. This PMD is implemented as thin wrapper
> of librte_vhost. The patch will work on below patch series.
>  - [PATCH v7 00/28] remove pci driver from vdevs
>
> * Known issue.
> We may see issues while handling RESET_OWNER message.
> These handlings are done in vhost library, so not a part of vhost PMD.
> So far, we are waiting for QEMU fixing.
>
> PATCH v2 changes:
>  - Remove a below patch that fixes vhost library.
>    The patch was applied as a separate patch.
>    - vhost: fix crash with multiqueue enabled
>  - Fix typos.
>    (Thanks to Thomas, Monjalon)
>  - Rebase on latest tree with above bernard's patches.
>
> PATCH v1 changes:
>  - Support vhost multiple queues.
>  - Rebase on "remove pci driver from vdevs".
>  - Optimize RX/TX functions.
>  - Fix resource leaks.
>  - Fix compile issue.
>  - Add patch to fix vhost library.
>
> RFC PATCH v3 changes:
>  - Optimize performance.
>    In RX/TX functions, change code to access only per core data.
>  - Add below API to allow user to use vhost library APIs for a port managed
>    by vhost PMD. There are a few limitations. See "rte_eth_vhost.h".
>     - rte_eth_vhost_portid2vdev()
>    To support this functionality, vhost library is also changed.
>    Anyway, if users doesn't use vhost PMD, can fully use vhost library APIs.
>  - Add code to support vhost multiple queues.
>    Actually, multiple queues functionality is not enabled so far.
>
> RFC PATCH v2 changes:
>  - Fix issues reported by checkpatch.pl
>    (Thanks to Stephen Hemminger)
>
>
> Tetsuya Mukawa (2):
>   vhost: Add callback and private data for vhost PMD
>   vhost: Add VHOST PMD
>
>  config/common_linuxapp                        |   6 +
>  doc/guides/nics/index.rst                     |   1 +
>  doc/guides/nics/vhost.rst                     |  82 +++
>  doc/guides/rel_notes/release_2_2.rst          |   2 +
>  drivers/net/Makefile                          |   4 +
>  drivers/net/vhost/Makefile                    |  62 +++
>  drivers/net/vhost/rte_eth_vhost.c             | 765 ++++++++++++++++++++++++++
>  drivers/net/vhost/rte_eth_vhost.h             |  65 +++
>  drivers/net/vhost/rte_pmd_vhost_version.map   |   8 +
>  lib/librte_vhost/rte_vhost_version.map        |   6 +
>  lib/librte_vhost/rte_virtio_net.h             |   3 +
>  lib/librte_vhost/vhost_user/virtio-net-user.c |  13 +-
>  lib/librte_vhost/virtio-net.c                 |  56 +-
>  lib/librte_vhost/virtio-net.h                 |   4 +-
>  mk/rte.app.mk                                 |   8 +-
>  15 files changed, 1072 insertions(+), 13 deletions(-)
>  create mode 100644 doc/guides/nics/vhost.rst
>  create mode 100644 drivers/net/vhost/Makefile
>  create mode 100644 drivers/net/vhost/rte_eth_vhost.c
>  create mode 100644 drivers/net/vhost/rte_eth_vhost.h
>  create mode 100644 drivers/net/vhost/rte_pmd_vhost_version.map
>
  

On Tue, 27 Oct 2015 15:12:55 +0900
Tetsuya Mukawa <mukawa@igel.co.jp> wrote:

> The patch introduces a new PMD. This PMD is implemented as thin wrapper
> of librte_vhost. It means librte_vhost is also needed to compile the PMD.
> The vhost messages will be handled only when a port is started. So start
> a port first, then invoke QEMU.
> 
> The PMD has 2 parameters.
>  - iface:  The parameter is used to specify a path to connect to a
>            virtio-net device.
>  - queues: The parameter is used to specify the number of the queues
>            virtio-net device has.
>            (Default: 1)
> 
> Here is an example.
> $ ./testpmd -c f -n 4 --vdev 'eth_vhost0,iface=/tmp/sock0,queues=1' -- -i
> 
> To connect above testpmd, here is qemu command example.
> 
> $ qemu-system-x86_64 \
>         <snip>
>         -chardev socket,id=chr0,path=/tmp/sock0 \
>         -netdev vhost-user,id=net0,chardev=chr0,vhostforce,queues=1 \
>         -device virtio-net-pci,netdev=net0
> 
> Signed-off-by: Tetsuya Mukawa <mukawa@igel.co.jp>

Brocade developed a much simpler vhost PMD, without all the atomics and
locking.


/*-
 *   BSD LICENSE
 *
 *   Copyright (C) Brocade Communications Systems, Inc.
 *   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 Brocade Communications Systems, Inc.
 *       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_mbuf.h>
#include <rte_ether.h>
#include <rte_ethdev.h>
#include <rte_malloc.h>
#include <rte_memcpy.h>
#include <rte_dev.h>
#include <rte_log.h>

#include "../librte_vhost/rte_virtio_net.h"
#include "../librte_vhost/virtio-net.h"

struct pmd_internals;

struct vhost_queue {
	struct pmd_internals *internals;

	struct rte_mempool *mb_pool;

	uint64_t	pkts;
	uint64_t	bytes;
};

struct pmd_internals {
	struct virtio_net *dev;
	unsigned numa_node;
	struct eth_driver *eth_drv;

	unsigned nb_rx_queues;
	unsigned nb_tx_queues;

	struct vhost_queue rx_queues[1];
	struct vhost_queue tx_queues[1];
	uint8_t port_id;
};


static const char *drivername = "Vhost PMD";

static struct rte_eth_link pmd_link = {
	.link_speed = 10000,
	.link_duplex = ETH_LINK_FULL_DUPLEX,
	.link_status = 0
};

static uint16_t
eth_vhost_rx(void *q, struct rte_mbuf **bufs, uint16_t nb_bufs)
{
	int ret, i;
	struct vhost_queue *h = q;

	ret = rte_vhost_dequeue_burst(h->internals->dev,
		VIRTIO_TXQ, h->mb_pool, bufs, nb_bufs);

	for (i = 0; i < ret ; i++) {
		struct rte_mbuf *m = bufs[i];

		m->port = h->internals->port_id;
		++h->pkts;
		h->bytes += rte_pktmbuf_pkt_len(m);
	}
	return ret;
}

static uint16_t
eth_vhost_tx(void *q, struct rte_mbuf **bufs, uint16_t nb_bufs)
{
	int ret, i;
	struct vhost_queue *h = q;

	ret = rte_vhost_enqueue_burst(h->internals->dev,
		VIRTIO_RXQ, bufs, nb_bufs);

	for (i = 0; i < ret; i++) {
		struct rte_mbuf *m = bufs[i];

		++h->pkts;
		h->bytes += rte_pktmbuf_pkt_len(m);
		rte_pktmbuf_free(m);
	}

	return ret;
}

static int
eth_dev_configure(struct rte_eth_dev *dev __rte_unused)
{
	return 0;
}

static int
eth_dev_start(struct rte_eth_dev *dev)
{
	struct pmd_internals *internals = dev->data->dev_private;

	dev->data->dev_link.link_status = 1;
	RTE_LOG(INFO, PMD, "vhost(%s): link up\n", internals->dev->ifname);
	return 0;
}

static void
eth_dev_stop(struct rte_eth_dev *dev)
{
	struct pmd_internals *internals = dev->data->dev_private;

	dev->data->dev_link.link_status = 0;
	RTE_LOG(INFO, PMD, "vhost(%s): link down\n", internals->dev->ifname);
}

static int
eth_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;

	internals->rx_queues[rx_queue_id].mb_pool = mb_pool;
	dev->data->rx_queues[rx_queue_id] =
		&internals->rx_queues[rx_queue_id];
	internals->rx_queues[rx_queue_id].internals = internals;

	return 0;
}

static int
eth_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;

	dev->data->tx_queues[tx_queue_id] =
		&internals->tx_queues[tx_queue_id];
	internals->tx_queues[tx_queue_id].internals = internals;

	return 0;
}


static void
eth_dev_info(struct rte_eth_dev *dev,
		struct rte_eth_dev_info *dev_info)
{
	struct pmd_internals *internals = dev->data->dev_private;

	dev_info->driver_name = drivername;
	dev_info->max_mac_addrs = 1;
	dev_info->max_rx_pktlen = -1;
	dev_info->max_rx_queues = (uint16_t)internals->nb_rx_queues;
	dev_info->max_tx_queues = (uint16_t)internals->nb_tx_queues;
	dev_info->min_rx_bufsize = 0;
	dev_info->pci_dev = NULL;
}

static void
eth_stats_get(struct rte_eth_dev *dev, struct rte_eth_stats *stats)
{
	const struct pmd_internals *internal = dev->data->dev_private;
	unsigned i;

	for (i = 0; i < internal->nb_rx_queues; i++) {
		const struct vhost_queue *h = &internal->rx_queues[i];

		stats->ipackets += h->pkts;
		stats->ibytes += h->bytes;

		if (i < RTE_ETHDEV_QUEUE_STAT_CNTRS) {
			stats->q_ibytes[i] = h->bytes;
			stats->q_ipackets[i] = h->pkts;
		}
	}

	for (i = 0; i < internal->nb_tx_queues; i++) {
		const struct vhost_queue *h = &internal->tx_queues[i];

		stats->opackets += h->pkts;
		stats->obytes += h->bytes;

		if (i < RTE_ETHDEV_QUEUE_STAT_CNTRS) {
			stats->q_obytes[i] = h->bytes;
			stats->q_opackets[i] = h->pkts;
		}
	}
}

static void
eth_stats_reset(struct rte_eth_dev *dev)
{
	unsigned i;
	struct pmd_internals *internal = dev->data->dev_private;

	for (i = 0; i < internal->nb_rx_queues; i++) {
		internal->rx_queues[i].pkts = 0;
		internal->rx_queues[i].bytes = 0;
	}

	for (i = 0; i < internal->nb_tx_queues; i++) {
		internal->tx_queues[i].pkts = 0;
		internal->tx_queues[i].bytes = 0;
	}
}

static struct eth_driver rte_vhost_pmd = {
	.pci_drv = {
		.name = "rte_vhost_pmd",
		.drv_flags = RTE_PCI_DRV_DETACHABLE,
	},
};

static void
eth_queue_release(void *q __rte_unused)
{
}

static int
eth_link_update(struct rte_eth_dev *dev __rte_unused,
		int wait_to_complete __rte_unused)
{
	return 0;
}

static struct eth_dev_ops eth_ops = {
	.dev_start = eth_dev_start,
	.dev_stop = eth_dev_stop,
	.dev_configure = eth_dev_configure,
	.dev_infos_get = eth_dev_info,
	.rx_queue_setup = eth_rx_queue_setup,
	.tx_queue_setup = eth_tx_queue_setup,
	.rx_queue_release = eth_queue_release,
	.tx_queue_release = eth_queue_release,
	.link_update = eth_link_update,
	.stats_get = eth_stats_get,
	.stats_reset = eth_stats_reset,
};

static int
eth_dev_vhost_create(const char *name, const unsigned numa_node)
{
	const unsigned nb_rx_queues = 1;
	const unsigned nb_tx_queues = 1;
	struct rte_eth_dev_data *data = NULL;
	struct rte_pci_device *pci_dev = NULL;
	struct pmd_internals *internals = NULL;
	struct rte_eth_dev *eth_dev = NULL;
	struct virtio_net *vhost_dev = NULL;
	struct eth_driver *eth_drv = NULL;
	struct rte_pci_id *id_table = NULL;
	struct ether_addr *eth_addr = NULL;

	if (name == NULL)
		return -EINVAL;

	vhost_dev = get_device_by_name(name);

	if (vhost_dev == NULL)
		return -EINVAL;

	RTE_LOG(INFO, PMD, "Creating vhost ethdev on numa socket %u\n",
			numa_node);

	data = rte_zmalloc_socket(name, sizeof(*data), 0, numa_node);
	if (data == NULL)
		goto error;

	pci_dev = rte_zmalloc_socket(name, sizeof(*pci_dev), 0, numa_node);
	if (pci_dev == NULL)
		goto error;

	id_table = rte_zmalloc_socket(name, sizeof(*id_table), 0, numa_node);
	if (id_table == NULL)
		goto error;

	internals = rte_zmalloc_socket(name, sizeof(*internals), 0, numa_node);
	if (internals == NULL)
		goto error;

	eth_addr = rte_zmalloc_socket(name, sizeof(*eth_addr), 0, numa_node);
	if (internals == NULL)
		goto error;

	eth_dev = rte_eth_dev_allocate(name, RTE_ETH_DEV_VIRTUAL);
	if (eth_dev == NULL)
		goto error;

	eth_drv = rte_zmalloc_socket(name, sizeof(*eth_drv), 0, numa_node);
	if (eth_drv == NULL)
		goto error;

	internals->nb_rx_queues = nb_rx_queues;
	internals->nb_tx_queues = nb_tx_queues;
	internals->numa_node = numa_node;
	internals->dev = vhost_dev;

	internals->port_id = eth_dev->data->port_id;

	eth_drv->pci_drv.name = drivername;
	eth_drv->pci_drv.id_table = id_table;
	internals->eth_drv = eth_drv;

	pci_dev->numa_node = numa_node;
	pci_dev->driver = &eth_drv->pci_drv;

	data->dev_private = internals;
	data->port_id = eth_dev->data->port_id;
	data->nb_rx_queues = (uint16_t)nb_rx_queues;
	data->nb_tx_queues = (uint16_t)nb_tx_queues;
	data->dev_link = pmd_link;
	eth_random_addr(&eth_addr->addr_bytes[0]);
	data->mac_addrs = eth_addr;
	strncpy(data->name, eth_dev->data->name, strlen(eth_dev->data->name));

	eth_dev->data = data;
	eth_dev->dev_ops = &eth_ops;
	eth_dev->pci_dev = pci_dev;
	eth_dev->driver = &rte_vhost_pmd;
	eth_dev->rx_pkt_burst = eth_vhost_rx;
	eth_dev->tx_pkt_burst = eth_vhost_tx;
	TAILQ_INIT(&(eth_dev->link_intr_cbs));

	return 0;

error:
	rte_free(data);
	rte_free(pci_dev);
	rte_free(id_table);
	rte_free(eth_drv);
	rte_free(eth_addr);
	rte_free(internals);

	return -1;
}

static int
rte_pmd_vhost_devinit(const char *name,
		      const char *params __attribute__((unused)))
{
	unsigned numa_node;

	if (name == NULL)
		return -EINVAL;

	RTE_LOG(DEBUG, PMD, "Initializing pmd_vhost for %s\n", name);

	numa_node = rte_socket_id();

	return eth_dev_vhost_create(name, numa_node);
}

static int
rte_pmd_vhost_devuninit(const char *name)
{
	struct rte_eth_dev *eth_dev = NULL;
	struct pmd_internals *internals = NULL;

	if (name == NULL)
		return -EINVAL;

	RTE_LOG(DEBUG, PMD, "Closing vhost ethdev on numa socket %u\n",
			rte_socket_id());

	/* reserve an ethdev entry */
	eth_dev = rte_eth_dev_allocated(name);
	if (eth_dev == NULL)
		return -1;

	internals = (struct pmd_internals *)eth_dev->data->dev_private;
	rte_free(internals->eth_drv->pci_drv.id_table);
	rte_free(internals->eth_drv);
	rte_free(eth_dev->data->dev_private);
	rte_free(eth_dev->data->mac_addrs);
	rte_free(eth_dev->data);
	rte_free(eth_dev->pci_dev);

	rte_eth_dev_release_port(eth_dev);

	return 0;
}

static struct rte_driver pmd_vhost_drv = {
	.name = "vhost",
	.type = PMD_VDEV,
	.init = rte_pmd_vhost_devinit,
	.uninit = rte_pmd_vhost_devuninit,
};

PMD_REGISTER_DRIVER(pmd_vhost_drv);
  

On 2015/11/10 7:25, Stephen Hemminger wrote:
> On Tue, 27 Oct 2015 15:12:55 +0900
> Tetsuya Mukawa <mukawa@igel.co.jp> wrote:
>
>> The patch introduces a new PMD. This PMD is implemented as thin wrapper
>> of librte_vhost. It means librte_vhost is also needed to compile the PMD.
>> The vhost messages will be handled only when a port is started. So start
>> a port first, then invoke QEMU.
>>
>> The PMD has 2 parameters.
>>  - iface:  The parameter is used to specify a path to connect to a
>>            virtio-net device.
>>  - queues: The parameter is used to specify the number of the queues
>>            virtio-net device has.
>>            (Default: 1)
>>
>> Here is an example.
>> $ ./testpmd -c f -n 4 --vdev 'eth_vhost0,iface=/tmp/sock0,queues=1' -- -i
>>
>> To connect above testpmd, here is qemu command example.
>>
>> $ qemu-system-x86_64 \
>>         <snip>
>>         -chardev socket,id=chr0,path=/tmp/sock0 \
>>         -netdev vhost-user,id=net0,chardev=chr0,vhostforce,queues=1 \
>>         -device virtio-net-pci,netdev=net0
>>
>> Signed-off-by: Tetsuya Mukawa <mukawa@igel.co.jp>
> Brocade developed a much simpler vhost PMD, without all the atomics and
> locking.
>

Hi Stephen,

With your PMD, it seems we need to call some vhost library APIs before
start sending and receiving.
It means we need to manage virtio-net device connections in DPDK
application anyway.

Also, I guess all PMDs should provide feature to be replaced by one of
other PMD without heavy modification for DPDK application.
This is because I tried to manage virtio-net device connections in vhost
PMD.

Thanks,
Tetsuya

> /*-
>  *   BSD LICENSE
>  *
>  *   Copyright (C) Brocade Communications Systems, Inc.
>  *   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 Brocade Communications Systems, Inc.
>  *       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_mbuf.h>
> #include <rte_ether.h>
> #include <rte_ethdev.h>
> #include <rte_malloc.h>
> #include <rte_memcpy.h>
> #include <rte_dev.h>
> #include <rte_log.h>
>
> #include "../librte_vhost/rte_virtio_net.h"
> #include "../librte_vhost/virtio-net.h"
>
> struct pmd_internals;
>
> struct vhost_queue {
> 	struct pmd_internals *internals;
>
> 	struct rte_mempool *mb_pool;
>
> 	uint64_t	pkts;
> 	uint64_t	bytes;
> };
>
> struct pmd_internals {
> 	struct virtio_net *dev;
> 	unsigned numa_node;
> 	struct eth_driver *eth_drv;
>
> 	unsigned nb_rx_queues;
> 	unsigned nb_tx_queues;
>
> 	struct vhost_queue rx_queues[1];
> 	struct vhost_queue tx_queues[1];
> 	uint8_t port_id;
> };
>
>
> static const char *drivername = "Vhost PMD";
>
> static struct rte_eth_link pmd_link = {
> 	.link_speed = 10000,
> 	.link_duplex = ETH_LINK_FULL_DUPLEX,
> 	.link_status = 0
> };
>
> static uint16_t
> eth_vhost_rx(void *q, struct rte_mbuf **bufs, uint16_t nb_bufs)
> {
> 	int ret, i;
> 	struct vhost_queue *h = q;
>
> 	ret = rte_vhost_dequeue_burst(h->internals->dev,
> 		VIRTIO_TXQ, h->mb_pool, bufs, nb_bufs);
>
> 	for (i = 0; i < ret ; i++) {
> 		struct rte_mbuf *m = bufs[i];
>
> 		m->port = h->internals->port_id;
> 		++h->pkts;
> 		h->bytes += rte_pktmbuf_pkt_len(m);
> 	}
> 	return ret;
> }
>
> static uint16_t
> eth_vhost_tx(void *q, struct rte_mbuf **bufs, uint16_t nb_bufs)
> {
> 	int ret, i;
> 	struct vhost_queue *h = q;
>
> 	ret = rte_vhost_enqueue_burst(h->internals->dev,
> 		VIRTIO_RXQ, bufs, nb_bufs);
>
> 	for (i = 0; i < ret; i++) {
> 		struct rte_mbuf *m = bufs[i];
>
> 		++h->pkts;
> 		h->bytes += rte_pktmbuf_pkt_len(m);
> 		rte_pktmbuf_free(m);
> 	}
>
> 	return ret;
> }
>
> static int
> eth_dev_configure(struct rte_eth_dev *dev __rte_unused)
> {
> 	return 0;
> }
>
> static int
> eth_dev_start(struct rte_eth_dev *dev)
> {
> 	struct pmd_internals *internals = dev->data->dev_private;
>
> 	dev->data->dev_link.link_status = 1;
> 	RTE_LOG(INFO, PMD, "vhost(%s): link up\n", internals->dev->ifname);
> 	return 0;
> }
>
> static void
> eth_dev_stop(struct rte_eth_dev *dev)
> {
> 	struct pmd_internals *internals = dev->data->dev_private;
>
> 	dev->data->dev_link.link_status = 0;
> 	RTE_LOG(INFO, PMD, "vhost(%s): link down\n", internals->dev->ifname);
> }
>
> static int
> eth_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;
>
> 	internals->rx_queues[rx_queue_id].mb_pool = mb_pool;
> 	dev->data->rx_queues[rx_queue_id] =
> 		&internals->rx_queues[rx_queue_id];
> 	internals->rx_queues[rx_queue_id].internals = internals;
>
> 	return 0;
> }
>
> static int
> eth_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;
>
> 	dev->data->tx_queues[tx_queue_id] =
> 		&internals->tx_queues[tx_queue_id];
> 	internals->tx_queues[tx_queue_id].internals = internals;
>
> 	return 0;
> }
>
>
> static void
> eth_dev_info(struct rte_eth_dev *dev,
> 		struct rte_eth_dev_info *dev_info)
> {
> 	struct pmd_internals *internals = dev->data->dev_private;
>
> 	dev_info->driver_name = drivername;
> 	dev_info->max_mac_addrs = 1;
> 	dev_info->max_rx_pktlen = -1;
> 	dev_info->max_rx_queues = (uint16_t)internals->nb_rx_queues;
> 	dev_info->max_tx_queues = (uint16_t)internals->nb_tx_queues;
> 	dev_info->min_rx_bufsize = 0;
> 	dev_info->pci_dev = NULL;
> }
>
> static void
> eth_stats_get(struct rte_eth_dev *dev, struct rte_eth_stats *stats)
> {
> 	const struct pmd_internals *internal = dev->data->dev_private;
> 	unsigned i;
>
> 	for (i = 0; i < internal->nb_rx_queues; i++) {
> 		const struct vhost_queue *h = &internal->rx_queues[i];
>
> 		stats->ipackets += h->pkts;
> 		stats->ibytes += h->bytes;
>
> 		if (i < RTE_ETHDEV_QUEUE_STAT_CNTRS) {
> 			stats->q_ibytes[i] = h->bytes;
> 			stats->q_ipackets[i] = h->pkts;
> 		}
> 	}
>
> 	for (i = 0; i < internal->nb_tx_queues; i++) {
> 		const struct vhost_queue *h = &internal->tx_queues[i];
>
> 		stats->opackets += h->pkts;
> 		stats->obytes += h->bytes;
>
> 		if (i < RTE_ETHDEV_QUEUE_STAT_CNTRS) {
> 			stats->q_obytes[i] = h->bytes;
> 			stats->q_opackets[i] = h->pkts;
> 		}
> 	}
> }
>
> static void
> eth_stats_reset(struct rte_eth_dev *dev)
> {
> 	unsigned i;
> 	struct pmd_internals *internal = dev->data->dev_private;
>
> 	for (i = 0; i < internal->nb_rx_queues; i++) {
> 		internal->rx_queues[i].pkts = 0;
> 		internal->rx_queues[i].bytes = 0;
> 	}
>
> 	for (i = 0; i < internal->nb_tx_queues; i++) {
> 		internal->tx_queues[i].pkts = 0;
> 		internal->tx_queues[i].bytes = 0;
> 	}
> }
>
> static struct eth_driver rte_vhost_pmd = {
> 	.pci_drv = {
> 		.name = "rte_vhost_pmd",
> 		.drv_flags = RTE_PCI_DRV_DETACHABLE,
> 	},
> };
>
> static void
> eth_queue_release(void *q __rte_unused)
> {
> }
>
> static int
> eth_link_update(struct rte_eth_dev *dev __rte_unused,
> 		int wait_to_complete __rte_unused)
> {
> 	return 0;
> }
>
> static struct eth_dev_ops eth_ops = {
> 	.dev_start = eth_dev_start,
> 	.dev_stop = eth_dev_stop,
> 	.dev_configure = eth_dev_configure,
> 	.dev_infos_get = eth_dev_info,
> 	.rx_queue_setup = eth_rx_queue_setup,
> 	.tx_queue_setup = eth_tx_queue_setup,
> 	.rx_queue_release = eth_queue_release,
> 	.tx_queue_release = eth_queue_release,
> 	.link_update = eth_link_update,
> 	.stats_get = eth_stats_get,
> 	.stats_reset = eth_stats_reset,
> };
>
> static int
> eth_dev_vhost_create(const char *name, const unsigned numa_node)
> {
> 	const unsigned nb_rx_queues = 1;
> 	const unsigned nb_tx_queues = 1;
> 	struct rte_eth_dev_data *data = NULL;
> 	struct rte_pci_device *pci_dev = NULL;
> 	struct pmd_internals *internals = NULL;
> 	struct rte_eth_dev *eth_dev = NULL;
> 	struct virtio_net *vhost_dev = NULL;
> 	struct eth_driver *eth_drv = NULL;
> 	struct rte_pci_id *id_table = NULL;
> 	struct ether_addr *eth_addr = NULL;
>
> 	if (name == NULL)
> 		return -EINVAL;
>
> 	vhost_dev = get_device_by_name(name);
>
> 	if (vhost_dev == NULL)
> 		return -EINVAL;
>
> 	RTE_LOG(INFO, PMD, "Creating vhost ethdev on numa socket %u\n",
> 			numa_node);
>
> 	data = rte_zmalloc_socket(name, sizeof(*data), 0, numa_node);
> 	if (data == NULL)
> 		goto error;
>
> 	pci_dev = rte_zmalloc_socket(name, sizeof(*pci_dev), 0, numa_node);
> 	if (pci_dev == NULL)
> 		goto error;
>
> 	id_table = rte_zmalloc_socket(name, sizeof(*id_table), 0, numa_node);
> 	if (id_table == NULL)
> 		goto error;
>
> 	internals = rte_zmalloc_socket(name, sizeof(*internals), 0, numa_node);
> 	if (internals == NULL)
> 		goto error;
>
> 	eth_addr = rte_zmalloc_socket(name, sizeof(*eth_addr), 0, numa_node);
> 	if (internals == NULL)
> 		goto error;
>
> 	eth_dev = rte_eth_dev_allocate(name, RTE_ETH_DEV_VIRTUAL);
> 	if (eth_dev == NULL)
> 		goto error;
>
> 	eth_drv = rte_zmalloc_socket(name, sizeof(*eth_drv), 0, numa_node);
> 	if (eth_drv == NULL)
> 		goto error;
>
> 	internals->nb_rx_queues = nb_rx_queues;
> 	internals->nb_tx_queues = nb_tx_queues;
> 	internals->numa_node = numa_node;
> 	internals->dev = vhost_dev;
>
> 	internals->port_id = eth_dev->data->port_id;
>
> 	eth_drv->pci_drv.name = drivername;
> 	eth_drv->pci_drv.id_table = id_table;
> 	internals->eth_drv = eth_drv;
>
> 	pci_dev->numa_node = numa_node;
> 	pci_dev->driver = &eth_drv->pci_drv;
>
> 	data->dev_private = internals;
> 	data->port_id = eth_dev->data->port_id;
> 	data->nb_rx_queues = (uint16_t)nb_rx_queues;
> 	data->nb_tx_queues = (uint16_t)nb_tx_queues;
> 	data->dev_link = pmd_link;
> 	eth_random_addr(&eth_addr->addr_bytes[0]);
> 	data->mac_addrs = eth_addr;
> 	strncpy(data->name, eth_dev->data->name, strlen(eth_dev->data->name));
>
> 	eth_dev->data = data;
> 	eth_dev->dev_ops = &eth_ops;
> 	eth_dev->pci_dev = pci_dev;
> 	eth_dev->driver = &rte_vhost_pmd;
> 	eth_dev->rx_pkt_burst = eth_vhost_rx;
> 	eth_dev->tx_pkt_burst = eth_vhost_tx;
> 	TAILQ_INIT(&(eth_dev->link_intr_cbs));
>
> 	return 0;
>
> error:
> 	rte_free(data);
> 	rte_free(pci_dev);
> 	rte_free(id_table);
> 	rte_free(eth_drv);
> 	rte_free(eth_addr);
> 	rte_free(internals);
>
> 	return -1;
> }
>
> static int
> rte_pmd_vhost_devinit(const char *name,
> 		      const char *params __attribute__((unused)))
> {
> 	unsigned numa_node;
>
> 	if (name == NULL)
> 		return -EINVAL;
>
> 	RTE_LOG(DEBUG, PMD, "Initializing pmd_vhost for %s\n", name);
>
> 	numa_node = rte_socket_id();
>
> 	return eth_dev_vhost_create(name, numa_node);
> }
>
> static int
> rte_pmd_vhost_devuninit(const char *name)
> {
> 	struct rte_eth_dev *eth_dev = NULL;
> 	struct pmd_internals *internals = NULL;
>
> 	if (name == NULL)
> 		return -EINVAL;
>
> 	RTE_LOG(DEBUG, PMD, "Closing vhost ethdev on numa socket %u\n",
> 			rte_socket_id());
>
> 	/* reserve an ethdev entry */
> 	eth_dev = rte_eth_dev_allocated(name);
> 	if (eth_dev == NULL)
> 		return -1;
>
> 	internals = (struct pmd_internals *)eth_dev->data->dev_private;
> 	rte_free(internals->eth_drv->pci_drv.id_table);
> 	rte_free(internals->eth_drv);
> 	rte_free(eth_dev->data->dev_private);
> 	rte_free(eth_dev->data->mac_addrs);
> 	rte_free(eth_dev->data);
> 	rte_free(eth_dev->pci_dev);
>
> 	rte_eth_dev_release_port(eth_dev);
>
> 	return 0;
> }
>
> static struct rte_driver pmd_vhost_drv = {
> 	.name = "vhost",
> 	.type = PMD_VDEV,
> 	.init = rte_pmd_vhost_devinit,
> 	.uninit = rte_pmd_vhost_devuninit,
> };
>
> PMD_REGISTER_DRIVER(pmd_vhost_drv);