[dpdk-dev] [PATCH v7 6/6] virtio-user: add a new vdev named virtio-user

[Jianfeng Tan <jianfeng.tan@intel.com>]

Add a new virtual device named vhost-user, which can be used just like
eth_ring, eth_null, etc. To reuse the code of original virtio, we do
some adjustment in virtio_ethdev.c, such as remove key _static_ of
eth_virtio_dev_init() so that it can be reused in virtual device; and
we add some check to make sure it will not crash.

Configured parameters include:
  - queues (optional, 1 by default), number of queue pairs, multi-queue
    not supported for now.
  - cq (optional, 0 by default), not supported for now.
  - mac (optional), random value will be given if not specified.
  - queue_size (optional, 256 by default), size of virtqueues.
  - path (madatory), path of vhost, depends on the file type, vhost
    user if the given path points to a unix socket; vhost-net if the
    given path points to a char device.
  - ifname (optional), specify the name of backend tap device; only
    valid when backend is vhost-net.

When enable CONFIG_RTE_VIRTIO_USER (enabled by default), the compiled
library can be used in both VM and container environment.

Examples:
path_vhost=/dev/vhost-net # use vhost-net as a backend
path_vhost=<path_to_vhost_user> # use vhost-user as a backend

sudo ./examples/l2fwd/build/l2fwd -c 0x100000 -n 4 \
    --socket-mem 0,1024 --no-pci --file-prefix=l2fwd \
    --vdev=virtio-user0,mac=00:01:02:03:04:05,path=$path_vhost -- -p 0x1

Known issues:
 - Control queue and multi-queue are not supported yet.
 - Cannot work with --huge-unlink.
 - Cannot work with no-huge.
 - Cannot work when there are more than VHOST_MEMORY_MAX_NREGIONS(8)
   hugepages.
 - Root privilege is a must (mainly becase of sorting hugepages according
   to physical address).
 - Applications should not use file name like HUGEFILE_FMT ("%smap_%d").

Signed-off-by: Huawei Xie <huawei.xie@intel.com>
Signed-off-by: Jianfeng Tan <jianfeng.tan@intel.com>
Acked-by: Neil Horman <nhorman@tuxdriver.com>
---
 doc/guides/rel_notes/release_16_07.rst  |  11 ++
 doc/guides/sample_app_ug/vhost.rst      |  17 +++
 drivers/net/virtio/virtio_ethdev.c      |  19 ++-
 drivers/net/virtio/virtio_ethdev.h      |   2 +
 drivers/net/virtio/virtio_user_ethdev.c | 228 +++++++++++++++++++++++++++++++-
 5 files changed, 265 insertions(+), 12 deletions(-)

diff --git a/doc/guides/rel_notes/release_16_07.rst b/doc/guides/rel_notes/release_16_07.rst
index 30e78d4..4ecca7e 100644
--- a/doc/guides/rel_notes/release_16_07.rst
+++ b/doc/guides/rel_notes/release_16_07.rst
@@ -47,6 +47,17 @@ New Features
   * Dropped specific Xen Dom0 code.
   * Dropped specific anonymous mempool code in testpmd.
 
+* **Virtio support for containers.**
+
+  Add a new virtual device, named virtio-user, to support virtio for containers.
+
+  Known limitations:
+
+  * Control queue and multi-queue are not supported yet.
+  * Cannot work with --huge-unlink.
+  * Cannot work with --no-huge.
+  * Cannot work when there are more than VHOST_MEMORY_MAX_NREGIONS(8) hugepages.
+  * Root privilege is a must for sorting hugepages by physical address.
 
 Resolved Issues
 ---------------
diff --git a/doc/guides/sample_app_ug/vhost.rst b/doc/guides/sample_app_ug/vhost.rst
index 5f81802..a93e54d 100644
--- a/doc/guides/sample_app_ug/vhost.rst
+++ b/doc/guides/sample_app_ug/vhost.rst
@@ -833,3 +833,20 @@ For example:
 The above message indicates that device 0 has been registered with MAC address cc:bb:bb:bb:bb:bb and VLAN tag 1000.
 Any packets received on the NIC with these values is placed on the devices receive queue.
 When a virtio-net device transmits packets, the VLAN tag is added to the packet by the DPDK vhost sample code.
+
+Running virtio-user with vhost-switch
+-------------------------------------
+
+We can also use virtio-user with vhost-switch now.
+Virtio-user is a virtual device that can be run in a application (container) parallelly with vhost in the same OS,
+aka, there is no need to start a VM. We just run it with a different --file-prefix to avoid startup failure.
+
+.. code-block:: console
+
+    cd ${RTE_SDK}/x86_64-native-linuxapp-gcc/app
+    ./testpmd -c 0x3 -n 4 --socket-mem 1024 --no-pci --file-prefix=virtio-user-testpmd \
+    --vdev=virtio-user0,mac=00:01:02:03:04:05,path=$path_vhost \
+    -- -i --txqflags=0xf01 --disable-hw-vlan
+
+There is no difference on the vhost side.
+Pleae note that there are some limitations (see release note for more information) in the usage of virtio-user.
diff --git a/drivers/net/virtio/virtio_ethdev.c b/drivers/net/virtio/virtio_ethdev.c
index 9ccce79..4523ceb 100644
--- a/drivers/net/virtio/virtio_ethdev.c
+++ b/drivers/net/virtio/virtio_ethdev.c
@@ -59,7 +59,6 @@
 #include "virtqueue.h"
 #include "virtio_rxtx.h"
 
-static int eth_virtio_dev_init(struct rte_eth_dev *eth_dev);
 static int eth_virtio_dev_uninit(struct rte_eth_dev *eth_dev);
 static int  virtio_dev_configure(struct rte_eth_dev *dev);
 static int  virtio_dev_start(struct rte_eth_dev *dev);
@@ -1081,7 +1080,7 @@ rx_func_get(struct rte_eth_dev *eth_dev)
  * This function is based on probe() function in virtio_pci.c
  * It returns 0 on success.
  */
-static int
+int
 eth_virtio_dev_init(struct rte_eth_dev *eth_dev)
 {
 	struct virtio_hw *hw = eth_dev->data->dev_private;
@@ -1112,9 +1111,11 @@ eth_virtio_dev_init(struct rte_eth_dev *eth_dev)
 
 	pci_dev = eth_dev->pci_dev;
 
-	ret = vtpci_init(pci_dev, hw, &dev_flags);
-	if (ret)
-		return ret;
+	if (pci_dev) {
+		ret = vtpci_init(pci_dev, hw, &dev_flags);
+		if (ret)
+			return ret;
+	}
 
 	/* Reset the device although not necessary at startup */
 	vtpci_reset(hw);
@@ -1206,7 +1207,8 @@ eth_virtio_dev_init(struct rte_eth_dev *eth_dev)
 
 	PMD_INIT_LOG(DEBUG, "hw->max_rx_queues=%d   hw->max_tx_queues=%d",
 			hw->max_rx_queues, hw->max_tx_queues);
-	PMD_INIT_LOG(DEBUG, "port %d vendorID=0x%x deviceID=0x%x",
+	if (pci_dev)
+		PMD_INIT_LOG(DEBUG, "port %d vendorID=0x%x deviceID=0x%x",
 			eth_dev->data->port_id, pci_dev->id.vendor_id,
 			pci_dev->id.device_id);
 
@@ -1494,7 +1496,10 @@ virtio_dev_info_get(struct rte_eth_dev *dev, struct rte_eth_dev_info *dev_info)
 {
 	struct virtio_hw *hw = dev->data->dev_private;
 
-	dev_info->driver_name = dev->driver->pci_drv.name;
+	if (dev->pci_dev)
+		dev_info->driver_name = dev->driver->pci_drv.name;
+	else
+		dev_info->driver_name = "virtio-user PMD";
 	dev_info->max_rx_queues = (uint16_t)hw->max_rx_queues;
 	dev_info->max_tx_queues = (uint16_t)hw->max_tx_queues;
 	dev_info->min_rx_bufsize = VIRTIO_MIN_RX_BUFSIZE;
diff --git a/drivers/net/virtio/virtio_ethdev.h b/drivers/net/virtio/virtio_ethdev.h
index 7e77259..2ecec6e 100644
--- a/drivers/net/virtio/virtio_ethdev.h
+++ b/drivers/net/virtio/virtio_ethdev.h
@@ -113,6 +113,8 @@ uint16_t virtio_recv_pkts_vec(void *rx_queue, struct rte_mbuf **rx_pkts,
 uint16_t virtio_xmit_pkts_simple(void *tx_queue, struct rte_mbuf **tx_pkts,
 		uint16_t nb_pkts);
 
+int eth_virtio_dev_init(struct rte_eth_dev *eth_dev);
+
 /*
  * The VIRTIO_NET_F_GUEST_TSO[46] features permit the host to send us
  * frames larger than 1514 bytes. We do not yet support software LRO
diff --git a/drivers/net/virtio/virtio_user_ethdev.c b/drivers/net/virtio/virtio_user_ethdev.c
index e390242..e1e77f0 100644
--- a/drivers/net/virtio/virtio_user_ethdev.c
+++ b/drivers/net/virtio/virtio_user_ethdev.c
@@ -35,6 +35,10 @@
 #include <sys/types.h>
 #include <unistd.h>
 
+#include <rte_malloc.h>
+#include <rte_kvargs.h>
+
+#include "virtio_ethdev.h"
 #include "virtio_logs.h"
 #include "virtio_pci.h"
 #include "virtqueue.h"
@@ -44,7 +48,7 @@
 	((struct virtio_user_dev *)(hw)->virtio_user_dev);
 
 static void
-virtio_user_read_dev_config(struct virtio_hw *hw, uint64_t offset,
+virtio_user_read_dev_config(struct virtio_hw *hw, size_t offset,
 		     void *dst, int length)
 {
 	int i;
@@ -65,7 +69,7 @@ virtio_user_read_dev_config(struct virtio_hw *hw, uint64_t offset,
 }
 
 static void
-virtio_user_write_dev_config(struct virtio_hw *hw, uint64_t offset,
+virtio_user_write_dev_config(struct virtio_hw *hw, size_t offset,
 		      const void *src, int length)
 {
 	int i;
@@ -165,9 +169,9 @@ virtio_user_setup_queue(struct virtio_hw *hw, struct virtqueue *vq)
 				   VIRTIO_PCI_VRING_ALIGN);
 
 	dev->vrings[queue_idx].num = vq->vq_nentries;
-	dev->vrings[queue_idx].desc = (void *)desc_addr;
-	dev->vrings[queue_idx].avail = (void *)avail_addr;
-	dev->vrings[queue_idx].used = (void *)used_addr;
+	dev->vrings[queue_idx].desc = (void *)(uintptr_t)desc_addr;
+	dev->vrings[queue_idx].avail = (void *)(uintptr_t)avail_addr;
+	dev->vrings[queue_idx].used = (void *)(uintptr_t)used_addr;
 
 	return 0;
 }
@@ -216,3 +220,217 @@ static const struct virtio_pci_ops virtio_user_ops = {
 	.del_queue	= virtio_user_del_queue,
 	.notify_queue	= virtio_user_notify_queue,
 };
+
+static const char *valid_args[] = {
+#define VIRTIO_USER_ARG_QUEUES_NUM     "queues"
+	VIRTIO_USER_ARG_QUEUES_NUM,
+#define VIRTIO_USER_ARG_CQ_NUM         "cq"
+	VIRTIO_USER_ARG_CQ_NUM,
+#define VIRTIO_USER_ARG_MAC            "mac"
+	VIRTIO_USER_ARG_MAC,
+#define VIRTIO_USER_ARG_PATH           "path"
+	VIRTIO_USER_ARG_PATH,
+#define VIRTIO_USER_ARG_QUEUE_SIZE     "queue_size"
+	VIRTIO_USER_ARG_QUEUE_SIZE,
+#define VIRTIO_USER_ARG_IFNAME         "ifname"
+	VIRTIO_USER_ARG_IFNAME,
+	NULL
+};
+
+#define VIRTIO_USER_DEF_CQ_EN	0
+#define VIRTIO_USER_DEF_Q_NUM	1
+#define VIRTIO_USER_DEF_Q_SZ	256
+
+static int
+get_string_arg(const char *key __rte_unused,
+	       const char *value, void *extra_args)
+{
+	if (!value || !extra_args)
+		return -EINVAL;
+
+	*(char **)extra_args = strdup(value);
+
+	return 0;
+}
+
+static int
+get_integer_arg(const char *key __rte_unused,
+		const char *value, void *extra_args)
+{
+	if (!value || !extra_args)
+		return -EINVAL;
+
+	*(uint64_t *)extra_args = strtoull(value, NULL, 0);
+
+	return 0;
+}
+
+static struct rte_eth_dev *
+virtio_user_eth_dev_alloc(const char *name)
+{
+	struct rte_eth_dev *eth_dev;
+	struct rte_eth_dev_data *data;
+	struct virtio_hw *hw;
+	struct virtio_user_dev *dev;
+
+	eth_dev = rte_eth_dev_allocate(name, RTE_ETH_DEV_VIRTUAL);
+	if (!eth_dev) {
+		PMD_INIT_LOG(ERR, "cannot alloc rte_eth_dev");
+		return NULL;
+	}
+
+	data = eth_dev->data;
+
+	hw = rte_zmalloc(NULL, sizeof(*hw), 0);
+	if (!hw) {
+		PMD_INIT_LOG(ERR, "malloc virtio_hw failed");
+		rte_eth_dev_release_port(eth_dev);
+		return NULL;
+	}
+
+	dev = rte_zmalloc(NULL, sizeof(*dev), 0);
+	if (!dev) {
+		PMD_INIT_LOG(ERR, "malloc virtio_user_dev failed");
+		rte_eth_dev_release_port(eth_dev);
+		rte_free(hw);
+		return NULL;
+	}
+
+	hw->vtpci_ops = &virtio_user_ops;
+	hw->use_msix = 0;
+	hw->modern   = 0;
+	hw->virtio_user_dev = dev;
+	data->dev_private = hw;
+	data->numa_node = SOCKET_ID_ANY;
+	data->kdrv = RTE_KDRV_NONE;
+	data->dev_flags = RTE_ETH_DEV_DETACHABLE;
+	eth_dev->pci_dev = NULL;
+	eth_dev->driver = NULL;
+	return eth_dev;
+}
+
+/* Dev initialization routine. Invoked once for each virtio vdev at
+ * EAL init time, see rte_eal_dev_init().
+ * Returns 0 on success.
+ */
+static int
+virtio_user_pmd_devinit(const char *name, const char *params)
+{
+	struct rte_kvargs *kvlist;
+	struct rte_eth_dev *eth_dev;
+	struct virtio_hw *hw;
+	uint64_t queues = VIRTIO_USER_DEF_Q_NUM;
+	uint64_t cq = VIRTIO_USER_DEF_CQ_EN;
+	uint64_t queue_size = VIRTIO_USER_DEF_Q_SZ;
+	char *path = NULL;
+	char *mac_addr = NULL;
+	char *ifname = NULL;
+	int ret = -1;
+
+	if (!params || params[0] == '\0') {
+		PMD_INIT_LOG(ERR, "arg %s is mandatory for virtio-user",
+			  VIRTIO_USER_ARG_QUEUE_SIZE);
+		goto end;
+	}
+
+	kvlist = rte_kvargs_parse(params, valid_args);
+	if (!kvlist) {
+		PMD_INIT_LOG(ERR, "error when parsing param");
+		goto end;
+	}
+
+	if (rte_kvargs_count(kvlist, VIRTIO_USER_ARG_PATH) == 1)
+		rte_kvargs_process(kvlist, VIRTIO_USER_ARG_PATH,
+				   &get_string_arg, &path);
+	else {
+		PMD_INIT_LOG(ERR, "arg %s is mandatory for virtio-user\n",
+			  VIRTIO_USER_ARG_QUEUE_SIZE);
+		goto end;
+	}
+
+	if (rte_kvargs_count(kvlist, VIRTIO_USER_ARG_MAC) == 1)
+		rte_kvargs_process(kvlist, VIRTIO_USER_ARG_MAC,
+				   &get_string_arg, &mac_addr);
+
+	if (rte_kvargs_count(kvlist, VIRTIO_USER_ARG_IFNAME) == 1)
+		rte_kvargs_process(kvlist, VIRTIO_USER_ARG_IFNAME,
+				   &get_string_arg, &ifname);
+
+	if (rte_kvargs_count(kvlist, VIRTIO_USER_ARG_QUEUE_SIZE) == 1)
+		rte_kvargs_process(kvlist, VIRTIO_USER_ARG_QUEUE_SIZE,
+				   &get_integer_arg, &queue_size);
+
+	if (rte_kvargs_count(kvlist, VIRTIO_USER_ARG_QUEUES_NUM) == 1)
+		rte_kvargs_process(kvlist, VIRTIO_USER_ARG_QUEUES_NUM,
+				   &get_integer_arg, &queues);
+
+	if (rte_kvargs_count(kvlist, VIRTIO_USER_ARG_CQ_NUM) == 1)
+		rte_kvargs_process(kvlist, VIRTIO_USER_ARG_CQ_NUM,
+				   &get_integer_arg, &cq);
+
+	eth_dev = virtio_user_eth_dev_alloc(name);
+	if (!eth_dev) {
+		PMD_INIT_LOG(ERR, "virtio-user fails to alloc device");
+		goto end;
+	}
+
+	hw = eth_dev->data->dev_private;
+	if (virtio_user_dev_init(hw->virtio_user_dev, path, queues, cq,
+				 queue_size, mac_addr, ifname) < 0)
+		goto end;
+
+	/* previously called by rte_eal_pci_probe() for physical dev */
+	if (eth_virtio_dev_init(eth_dev) < 0) {
+		PMD_INIT_LOG(ERR, "eth_virtio_dev_init fails");
+		goto end;
+	}
+	ret = 0;
+
+end:
+	if (path)
+		free(path);
+	if (mac_addr)
+		free(mac_addr);
+	if (ifname)
+		free(ifname);
+	return ret;
+}
+
+/** Called by rte_eth_dev_detach() */
+static int
+virtio_user_pmd_devuninit(const char *name)
+{
+	struct rte_eth_dev *eth_dev;
+	struct virtio_hw *hw;
+	struct virtio_user_dev *dev;
+
+	if (!name)
+		return -EINVAL;
+
+	PMD_DRV_LOG(INFO, "Un-Initializing %s\n", name);
+	eth_dev = rte_eth_dev_allocated(name);
+	if (!eth_dev)
+		return -ENODEV;
+
+	/* make sure the device is stopped, queues freed */
+	rte_eth_dev_close(eth_dev->data->port_id);
+
+	hw = eth_dev->data->dev_private;
+	dev = hw->virtio_user_dev;
+	virtio_user_dev_uninit(dev);
+
+	rte_free(eth_dev->data->dev_private);
+	rte_free(eth_dev->data);
+	rte_eth_dev_release_port(eth_dev);
+
+	return 0;
+}
+
+static struct rte_driver virtio_user_driver = {
+	.name   = "virtio-user",
+	.type   = PMD_VDEV,
+	.init   = virtio_user_pmd_devinit,
+	.uninit = virtio_user_pmd_devuninit,
+};
+
+PMD_REGISTER_DRIVER(virtio_user_driver);
-- 
2.1.4