On 4 Dec 2024, at 10:29, Maxime Coquelin wrote:
> This patch uses the new rte_vhost_driver_set_max_queue_num
> API to set the maximum number of queue pairs supported by
> the Vhost-user port.
>
> This is required for VDUSE which needs to specify the
> maximum number of queue pairs at creation time. Without it
> 128 queue pairs metadata would be allocated.
>
> To configure it, a new 'vhost-queue-pairs-max' option is
> introduced.
>
> Signed-off-by: Maxime Coquelin <[email protected]>
> Acked-by: Eelco Chaudron <[email protected]>
> ---
I only code-reviewed this patch (no actual testing) and I have some comments
below.
> Changes in v3:
> ==============
> - Introduce a new option to set the number of max queue pairs (Kevin)
> - Add documentation for new option
>
> Changes in v2:
> ==============
> - Address checkpatch warnings.
> ---
> Documentation/topics/dpdk/vhost-user.rst | 15 +++++++++++++
> lib/netdev-dpdk.c | 27 ++++++++++++++++++++++++
> 2 files changed, 42 insertions(+)
>
> diff --git a/Documentation/topics/dpdk/vhost-user.rst
> b/Documentation/topics/dpdk/vhost-user.rst
> index 7bba08ac2..5c3f1046f 100644
> --- a/Documentation/topics/dpdk/vhost-user.rst
> +++ b/Documentation/topics/dpdk/vhost-user.rst
We also need to add the definition to the vswitch.xml file.
> @@ -375,6 +375,21 @@ Tx retries max can be set for vhost-user-client ports::
>
> Configurable vhost tx retries are not supported with vhost-user ports.
>
> +vhost-user-client max queue pairs config
> +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> +
> +For vhost-user-client interfaces using the VDUSE backend, the maximum umber
> of
> +queue pairs the Virtio device will support can be set at port creation time.
> If
> +not set, the default value is 1 queue pair. As of DPDK v24.11, this value is
> +ignored for Vhost-user backends.
I think we should not reference a DPDK version here. We should just state that
this value is ignored for non-VDUSE backends.
> +
> +Maximum number of queue pairs can be set for vhost-user-client-ports::
> +
> + $ ovs-vsctl add-port br0 vduse0 \
> + -- set Interface vduse0 type=dpdkvhostuserclient \
> + options:vhost-server-path=/dev/vduse/vduse0 \
> + options:vhost-queue-pairs-max=4
The order in the naming looks odd, what about vhost-max-queue-pairs? If you
agree also all defines and code variables should change.
> +
> .. _dpdk-testpmd:
>
> DPDK in the Guest
> diff --git a/lib/netdev-dpdk.c b/lib/netdev-dpdk.c
> index dc52a2b56..b2d14b678 100644
> --- a/lib/netdev-dpdk.c
> +++ b/lib/netdev-dpdk.c
> @@ -19,6 +19,7 @@
>
> #include <errno.h>
> #include <signal.h>
> +#include <stdint.h>
> #include <stdlib.h>
> #include <string.h>
> #include <unistd.h>
> @@ -153,6 +154,9 @@ typedef uint16_t dpdk_port_t;
> /* Legacy default value for vhost tx retries. */
> #define VHOST_ENQ_RETRY_DEF 8
>
> +/* VDUSE-only, ignore for Vhost-user */
> +#define VHOST_QUEUE_PAIRS_MAX_DEF 1
> +
> #define IF_NAME_SZ (PATH_MAX > IFNAMSIZ ? PATH_MAX : IFNAMSIZ)
>
> /* List of required flags advertised by the hardware that will be used
> @@ -497,6 +501,9 @@ struct netdev_dpdk {
>
> atomic_uint8_t vhost_tx_retries_max;
>
> + /* Ignored by DPDK for Vhost-user backends, only for VDUSE */
> + uint32_t vhost_queue_pairs_max;
> +
> /* Flags for virtio features recovery mechanism. */
> uint8_t virtio_features_state;
>
> @@ -1609,6 +1616,8 @@ vhost_common_construct(struct netdev *netdev)
>
> atomic_init(&dev->vhost_tx_retries_max, VHOST_ENQ_RETRY_DEF);
>
> + dev->vhost_queue_pairs_max = VHOST_QUEUE_PAIRS_MAX_DEF;
> +
> return common_construct(netdev, DPDK_ETH_PORT_ID_INVALID,
> DPDK_DEV_VHOST, socket_id);
> }
> @@ -2491,6 +2500,7 @@ netdev_dpdk_vhost_client_set_config(struct netdev
> *netdev,
> struct netdev_dpdk *dev = netdev_dpdk_cast(netdev);
> const char *path;
> int max_tx_retries, cur_max_tx_retries;
> + uint32_t queue_pairs_max;
>
> ovs_mutex_lock(&dev->mutex);
> if (!(dev->vhost_driver_flags & RTE_VHOST_USER_CLIENT)) {
> @@ -2498,6 +2508,14 @@ netdev_dpdk_vhost_client_set_config(struct netdev
> *netdev,
> if (!nullable_string_is_equal(path, dev->vhost_id)) {
> free(dev->vhost_id);
> dev->vhost_id = nullable_xstrdup(path);
> +
> + queue_pairs_max = smap_get_int(args, "vhost-queue-pairs-max",
> + VHOST_QUEUE_PAIRS_MAX_DEF);
> + if (queue_pairs_max < 1) {
> + queue_pairs_max = VHOST_QUEUE_PAIRS_MAX_DEF;
> + }
> + dev->vhost_queue_pairs_max = queue_pairs_max;
> +
> netdev_request_reconfigure(netdev);
> }
> }
> @@ -2514,6 +2532,7 @@ netdev_dpdk_vhost_client_set_config(struct netdev
> *netdev,
> VLOG_INFO("Max Tx retries for vhost device '%s' set to %d",
> netdev_get_name(netdev), max_tx_retries);
> }
> +
> ovs_mutex_unlock(&dev->mutex);
>
> return 0;
> @@ -6400,6 +6419,14 @@ netdev_dpdk_vhost_client_reconfigure(struct netdev
> *netdev)
> goto unlock;
> }
>
> + err = rte_vhost_driver_set_max_queue_num(dev->vhost_id,
> + dev->vhost_queue_pairs_max);
> + if (err) {
> + VLOG_ERR("rte_vhost_driver_set_max_queue_num failed for "
> + "vhost-user client port: %s\n", dev->up.name);
> + goto unlock;
> + }
> +
> err = rte_vhost_driver_start(dev->vhost_id);
> if (err) {
> VLOG_ERR("rte_vhost_driver_start failed for vhost user "
> --
> 2.47.0
_______________________________________________
dev mailing list
[email protected]
https://mail.openvswitch.org/mailman/listinfo/ovs-dev
