netdev-dpdk: Enable tx-retries-max config.

vhost tx retries can provide some mitigation against
dropped packets due to a temporarily slow guest/limited queue
size for an interface, but on the other hand when a system
is fully loaded those extra cycles retrying could mean
packets are dropped elsewhere.

Up to now max vhost tx retries have been hardcoded, which meant
no tuning and no way to disable for debugging to see if extra
cycles spent retrying resulted in rx drops on some other
interface.

Add an option to change the max retries, with a value of
0 effectively disabling vhost tx retries.

Signed-off-by: Kevin Traynor <ktraynor@redhat.com>
Acked-by: Eelco Chaudron <echaudro@redhat.com>
Acked-by: Flavio Leitner <fbl@sysclose.org>
Acked-by: Ilya Maximets <i.maximets@samsung.com>
Signed-off-by: Ian Stokes <ian.stokes@intel.com>

Documentation/topics/dpdk/vhost-user.rst

@@ -350,6 +350,31 @@ The default value is ``false``.
 
 .. _dpdk-testpmd:
 
+vhost-user-client tx retries config
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+For vhost-user-client interfaces, the max amount of retries can be changed from
+the default 8 by setting ``tx-retries-max``.
+
+The minimum is 0 which means there will be no retries and if any packets in
+each batch cannot be sent immediately they will be dropped. The maximum is 32,
+which would mean that after the first packet(s) in the batch was sent there
+could be a maximum of 32 more retries.
+
+Retries can help with avoiding packet loss when temporarily unable to send to a
+vhost interface because the virtqueue is full. However, spending more time
+retrying to send to one interface, will reduce the time available for rx/tx and
+processing packets on other interfaces, so some tuning may be required for best
+performance.
+
+Tx retries max can be set for vhost-user-client ports::
+
+    $ ovs-vsctl set Interface vhost-client-1 options:tx-retries-max=0
+
+.. note::
+
+  Configurable vhost tx retries are not supported with vhost-user ports.
+
 DPDK in the Guest
 -----------------
 
@@ -495,6 +520,9 @@ there will be a retry, with a default maximum of 8 occurring. If at any time no
 packets can be sent, it may mean the guest is not accepting packets, so there
 are no (more) retries.
 
+For information about configuring the maximum amount of tx retries for
+vhost-user-client interfaces see `vhost-user-client tx retries config`_.
+
 .. note::
 
   Maximum vhost tx batch size is defined by NETDEV_MAX_BURST, and is currently

NEWS

@@ -7,6 +7,8 @@ Post-v2.11.0
      * OVS validated with DPDK 18.11.2 which is the new minimal supported
        version.
      * DPDK 18.11.1 and lower is no longer supported.
+     * New option 'tx-retries-max' to set the maximum amount of vhost tx
+       retries that can be made.
    - OpenFlow:
      * All features required by OpenFlow 1.5 are now implemented, so
        ovs-vswitchd now enables OpenFlow 1.5 by default (in addition to

lib/netdev-dpdk.c

@@ -164,7 +164,13 @@ BUILD_ASSERT_DECL((MAX_NB_MBUF / ROUND_DOWN_POW2(MAX_NB_MBUF / MIN_NB_MBUF))
 typedef uint16_t dpdk_port_t;
 #define DPDK_PORT_ID_FMT "%"PRIu16
 
-#define VHOST_ENQ_RETRY_NUM 8
+/* Minimum amount of vhost tx retries, effectively a disable. */
+#define VHOST_ENQ_RETRY_MIN 0
+/* Maximum amount of vhost tx retries. */
+#define VHOST_ENQ_RETRY_MAX 32
+/* Legacy default value for vhost tx retries. */
+#define VHOST_ENQ_RETRY_DEF 8
+
 #define IF_NAME_SZ (PATH_MAX > IFNAMSIZ ? PATH_MAX : IFNAMSIZ)
 
 static const struct rte_eth_conf port_conf = {
@@ -417,7 +423,9 @@ struct netdev_dpdk {
 
         /* True if vHost device is 'up' and has been reconfigured at least once */
         bool vhost_reconfigured;
-        /* 3 pad bytes here. */
+
+        atomic_uint8_t vhost_tx_retries_max;
+        /* 2 pad bytes here. */
     );
 
     PADDED_MEMBERS(CACHE_LINE_SIZE,
@@ -1261,6 +1269,8 @@ vhost_common_construct(struct netdev *netdev)
         return ENOMEM;
     }
 
+    atomic_init(&dev->vhost_tx_retries_max, VHOST_ENQ_RETRY_DEF);
+
     return common_construct(netdev, DPDK_ETH_PORT_ID_INVALID,
                             DPDK_DEV_VHOST, socket_id);
 }
@@ -1921,6 +1931,7 @@ 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;
 
     ovs_mutex_lock(&dev->mutex);
     if (!(dev->vhost_driver_flags & RTE_VHOST_USER_CLIENT)) {
@@ -1937,6 +1948,19 @@ netdev_dpdk_vhost_client_set_config(struct netdev *netdev,
             netdev_request_reconfigure(netdev);
         }
     }
+
+    max_tx_retries = smap_get_int(args, "tx-retries-max",
+                                  VHOST_ENQ_RETRY_DEF);
+    if (max_tx_retries < VHOST_ENQ_RETRY_MIN
+        || max_tx_retries > VHOST_ENQ_RETRY_MAX) {
+        max_tx_retries = VHOST_ENQ_RETRY_DEF;
+    }
+    atomic_read_relaxed(&dev->vhost_tx_retries_max, &cur_max_tx_retries);
+    if (max_tx_retries != cur_max_tx_retries) {
+        atomic_store_relaxed(&dev->vhost_tx_retries_max, max_tx_retries);
+        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;
@@ -2350,6 +2374,7 @@ __netdev_dpdk_vhost_send(struct netdev *netdev, int qid,
     unsigned int total_pkts = cnt;
     unsigned int dropped = 0;
     int i, retries = 0;
+    int max_retries = VHOST_ENQ_RETRY_MIN;
     int vid = netdev_dpdk_get_vid(dev);
 
     qid = dev->tx_q[qid % netdev->n_txq].map;
@@ -2379,18 +2404,25 @@ __netdev_dpdk_vhost_send(struct netdev *netdev, int qid,
             cnt -= tx_pkts;
             /* Prepare for possible retry.*/
             cur_pkts = &cur_pkts[tx_pkts];
+            if (OVS_UNLIKELY(cnt && !retries)) {
+                /*
+                 * Read max retries as there are packets not sent
+                 * and no retries have already occurred.
+                 */
+                atomic_read_relaxed(&dev->vhost_tx_retries_max, &max_retries);
+            }
         } else {
             /* No packets sent - do not retry.*/
             break;
         }
-    } while (cnt && (retries++ < VHOST_ENQ_RETRY_NUM));
+    } while (cnt && (retries++ < max_retries));
 
     rte_spinlock_unlock(&dev->tx_q[qid].tx_lock);
 
     rte_spinlock_lock(&dev->stats_lock);
     netdev_dpdk_vhost_update_tx_counters(&dev->stats, pkts, total_pkts,
                                          cnt + dropped);
-    dev->tx_retries += MIN(retries, VHOST_ENQ_RETRY_NUM);
+    dev->tx_retries += MIN(retries, max_retries);
     rte_spinlock_unlock(&dev->stats_lock);
 
 out:

vswitchd/vswitch.xml

@@ -3127,6 +3127,18 @@ ovs-vsctl add-port br0 p0 -- set Interface p0 type=patch options:peer=p1 \
         </p>
       </column>
 
+      <column name="options" key="tx-retries-max"
+              type='{"type": "integer", "minInteger": 0, "maxInteger": 32}'>
+        <p>
+          The value specifies the maximum amount of vhost tx retries that can
+          be made while trying to send a batch of packets to an interface.
+          Only supported by dpdkvhostuserclient interfaces.
+        </p>
+        <p>
+          Default value is 8.
+        </p>
+      </column>
+
       <column name="options" key="n_rxq_desc"
               type='{"type": "integer", "minInteger": 1, "maxInteger": 4096}'>
         <p>