examples/l3fwd-power: add Rx interrupt timeout
[dpdk.git] / examples / qos_sched / app_thread.c
old mode 100755 (executable)
new mode 100644 (file)
index 68b621d..dbc878b
@@ -1,34 +1,5 @@
-/*-
- *   BSD LICENSE
- * 
- *   Copyright(c) 2010-2013 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.
+/* SPDX-License-Identifier: BSD-3-Clause
+ * Copyright(c) 2010-2014 Intel Corporation
  */
 
 #include <stdint.h>
  * QoS parameters are encoded as follows:
  *             Outer VLAN ID defines subport
  *             Inner VLAN ID defines pipe
- *             Destination IP 0.0.XXX.0 defines traffic class
  *             Destination IP host (0.0.0.XXX) defines queue
  * Values below define offset to each field from start of frame
  */
 #define SUBPORT_OFFSET 7
 #define PIPE_OFFSET            9
-#define TC_OFFSET              20
 #define QUEUE_OFFSET   20
 #define COLOR_OFFSET   19
 
@@ -64,16 +33,27 @@ get_pkt_sched(struct rte_mbuf *m, uint32_t *subport, uint32_t *pipe,
                        uint32_t *traffic_class, uint32_t *queue, uint32_t *color)
 {
        uint16_t *pdata = rte_pktmbuf_mtod(m, uint16_t *);
+       uint16_t pipe_queue;
 
+       /* Outer VLAN ID*/
        *subport = (rte_be_to_cpu_16(pdata[SUBPORT_OFFSET]) & 0x0FFF) &
-                       (port_params.n_subports_per_port - 1); /* Outer VLAN ID*/
+               (port_params.n_subports_per_port - 1);
+
+       /* Inner VLAN ID */
        *pipe = (rte_be_to_cpu_16(pdata[PIPE_OFFSET]) & 0x0FFF) &
-                       (port_params.n_pipes_per_subport - 1); /* Inner VLAN ID */
-       *traffic_class = (pdata[QUEUE_OFFSET] & 0x0F) &
-                       (RTE_SCHED_TRAFFIC_CLASSES_PER_PIPE - 1); /* Destination IP */
-       *queue = ((pdata[QUEUE_OFFSET] >> 8) & 0x0F) &
-                       (RTE_SCHED_QUEUES_PER_TRAFFIC_CLASS - 1) ; /* Destination IP */
-       *color = pdata[COLOR_OFFSET] & 0x03;    /* Destination IP */
+               (subport_params[*subport].n_pipes_per_subport_enabled - 1);
+
+       pipe_queue = active_queues[(pdata[QUEUE_OFFSET] >> 8) % n_active_queues];
+
+       /* Traffic class (Destination IP) */
+       *traffic_class = pipe_queue > RTE_SCHED_TRAFFIC_CLASS_BE ?
+                       RTE_SCHED_TRAFFIC_CLASS_BE : pipe_queue;
+
+       /* Traffic class queue (Destination IP) */
+       *queue = pipe_queue - *traffic_class;
+
+       /* Color (Destination IP) */
+       *color = pdata[COLOR_OFFSET] & 0x03;
 
        return 0;
 }
@@ -102,12 +82,15 @@ app_rx_thread(struct thread_conf **confs)
                        for(i = 0; i < nb_rx; i++) {
                                get_pkt_sched(rx_mbufs[i],
                                                &subport, &pipe, &traffic_class, &queue, &color);
-                               rte_sched_port_pkt_write(rx_mbufs[i], subport, pipe,
-                                               traffic_class, queue, (enum rte_meter_color) color);
+                               rte_sched_port_pkt_write(conf->sched_port,
+                                               rx_mbufs[i],
+                                               subport, pipe,
+                                               traffic_class, queue,
+                                               (enum rte_color) color);
                        }
 
                        if (unlikely(rte_ring_sp_enqueue_bulk(conf->rx_ring,
-                                                               (void **)rx_mbufs, nb_rx) != 0)) {
+                                       (void **)rx_mbufs, nb_rx, NULL) == 0)) {
                                for(i = 0; i < nb_rx; i++) {
                                        rte_pktmbuf_free(rx_mbufs[i]);
 
@@ -139,17 +122,11 @@ app_send_burst(struct thread_conf *qconf)
 
        do {
                ret = rte_eth_tx_burst(qconf->tx_port, qconf->tx_queue, mbufs, (uint16_t)n);
-               if (unlikely(ret < n)) { /* we cannot drop the packets, so re-send */
-                       /* update number of packets to be sent */
-                       n -= ret;
-                       mbufs = (struct rte_mbuf **)&mbufs[ret];
-                       /* limit number of retries to avoid endless loop */
-                       /* reset retry counter if some packets were sent */
-                       if (likely(ret != 0)) {
-                               continue;
-                       }
-               }
-       } while (ret != n);
+               /* we cannot drop the packets, so re-send */
+               /* update number of packets to be sent */
+               n -= ret;
+               mbufs = (struct rte_mbuf **)&mbufs[ret];
+       } while (n);
 }
 
 
@@ -185,8 +162,8 @@ app_tx_thread(struct thread_conf **confs)
 
        while ((conf = confs[conf_idx])) {
                retval = rte_ring_sc_dequeue_bulk(conf->tx_ring, (void **)mbufs,
-                                       burst_conf.qos_dequeue);
-               if (likely(retval == 0)) {
+                                       burst_conf.qos_dequeue, NULL);
+               if (likely(retval != 0)) {
                        app_send_packets(conf, mbufs, burst_conf.qos_dequeue);
 
                        conf->counter = 0; /* reset empty read loop counter */
@@ -221,23 +198,24 @@ app_worker_thread(struct thread_conf **confs)
 
        while ((conf = confs[conf_idx])) {
                uint32_t nb_pkt;
-               int retval;
 
                /* Read packet from the ring */
-               retval = rte_ring_sc_dequeue_bulk(conf->rx_ring, (void **)mbufs,
-                                       burst_conf.ring_burst);
-               if (likely(retval == 0)) {
+               nb_pkt = rte_ring_sc_dequeue_burst(conf->rx_ring, (void **)mbufs,
+                                       burst_conf.ring_burst, NULL);
+               if (likely(nb_pkt)) {
                        int nb_sent = rte_sched_port_enqueue(conf->sched_port, mbufs,
-                                       burst_conf.ring_burst);
+                                       nb_pkt);
 
-                       APP_STATS_ADD(conf->stat.nb_drop, burst_conf.ring_burst - nb_sent);
-                       APP_STATS_ADD(conf->stat.nb_rx, burst_conf.ring_burst);
+                       APP_STATS_ADD(conf->stat.nb_drop, nb_pkt - nb_sent);
+                       APP_STATS_ADD(conf->stat.nb_rx, nb_pkt);
                }
 
                nb_pkt = rte_sched_port_dequeue(conf->sched_port, mbufs,
                                        burst_conf.qos_dequeue);
                if (likely(nb_pkt > 0))
-                       while (rte_ring_sp_enqueue_bulk(conf->tx_ring, (void **)mbufs, nb_pkt) != 0);
+                       while (rte_ring_sp_enqueue_bulk(conf->tx_ring,
+                                       (void **)mbufs, nb_pkt, NULL) == 0)
+                               ; /* empty body */
 
                conf_idx++;
                if (confs[conf_idx] == NULL)
@@ -256,17 +234,16 @@ app_mixed_thread(struct thread_conf **confs)
 
        while ((conf = confs[conf_idx])) {
                uint32_t nb_pkt;
-               int retval;
 
                /* Read packet from the ring */
-               retval = rte_ring_sc_dequeue_bulk(conf->rx_ring, (void **)mbufs,
-                                       burst_conf.ring_burst);
-               if (likely(retval == 0)) {
+               nb_pkt = rte_ring_sc_dequeue_burst(conf->rx_ring, (void **)mbufs,
+                                       burst_conf.ring_burst, NULL);
+               if (likely(nb_pkt)) {
                        int nb_sent = rte_sched_port_enqueue(conf->sched_port, mbufs,
-                                       burst_conf.ring_burst);
+                                       nb_pkt);
 
-                       APP_STATS_ADD(conf->stat.nb_drop, burst_conf.ring_burst - nb_sent);
-                       APP_STATS_ADD(conf->stat.nb_rx, burst_conf.ring_burst);
+                       APP_STATS_ADD(conf->stat.nb_drop, nb_pkt - nb_sent);
+                       APP_STATS_ADD(conf->stat.nb_rx, nb_pkt);
                }
 
 
@@ -297,5 +274,3 @@ app_mixed_thread(struct thread_conf **confs)
                        conf_idx = 0;
        }
 }
-
-