examples/kni: convert to new ethdev offloads API
authorShahaf Shuler <shahafs@mellanox.com>
Tue, 26 Dec 2017 09:23:10 +0000 (11:23 +0200)
committerFerruh Yigit <ferruh.yigit@intel.com>
Tue, 16 Jan 2018 17:47:49 +0000 (18:47 +0100)
Ethdev offloads API has changed since:

commit ce17eddefc20 ("ethdev: introduce Rx queue offloads API")
commit cba7f53b717d ("ethdev: introduce Tx queue offloads API")

This commit support the new API.

Signed-off-by: Shahaf Shuler <shahafs@mellanox.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
examples/kni/main.c

index e8b4840..5700740 100644 (file)
@@ -95,11 +95,8 @@ static struct kni_port_params *kni_port_params_array[RTE_MAX_ETHPORTS];
 /* Options for configuring ethernet port */
 static struct rte_eth_conf port_conf = {
        .rxmode = {
-               .header_split = 0,      /* Header Split disabled */
-               .hw_ip_checksum = 0,    /* IP checksum offload disabled */
-               .hw_vlan_filter = 0,    /* VLAN filtering disabled */
-               .jumbo_frame = 0,       /* Jumbo Frame Support disabled */
-               .hw_strip_crc = 1,      /* CRC stripped by hardware */
+               .ignore_offload_bitfield = 1,
+               .offloads = DEV_RX_OFFLOAD_CRC_STRIP,
        },
        .txmode = {
                .mq_mode = ETH_MQ_TX_NONE,
@@ -578,11 +575,19 @@ init_port(uint16_t port)
        int ret;
        uint16_t nb_rxd = NB_RXD;
        uint16_t nb_txd = NB_TXD;
+       struct rte_eth_dev_info dev_info;
+       struct rte_eth_rxconf rxq_conf;
+       struct rte_eth_txconf txq_conf;
+       struct rte_eth_conf local_port_conf = port_conf;
 
        /* Initialise device and RX/TX queues */
        RTE_LOG(INFO, APP, "Initialising port %u ...\n", (unsigned)port);
        fflush(stdout);
-       ret = rte_eth_dev_configure(port, 1, 1, &port_conf);
+       rte_eth_dev_info_get(port, &dev_info);
+       if (dev_info.tx_offload_capa & DEV_TX_OFFLOAD_MBUF_FAST_FREE)
+               local_port_conf.txmode.offloads |=
+                       DEV_TX_OFFLOAD_MBUF_FAST_FREE;
+       ret = rte_eth_dev_configure(port, 1, 1, &local_port_conf);
        if (ret < 0)
                rte_exit(EXIT_FAILURE, "Could not configure port%u (%d)\n",
                            (unsigned)port, ret);
@@ -592,14 +597,19 @@ init_port(uint16_t port)
                rte_exit(EXIT_FAILURE, "Could not adjust number of descriptors "
                                "for port%u (%d)\n", (unsigned)port, ret);
 
+       rxq_conf = dev_info.default_rxconf;
+       rxq_conf.offloads = local_port_conf.rxmode.offloads;
        ret = rte_eth_rx_queue_setup(port, 0, nb_rxd,
-               rte_eth_dev_socket_id(port), NULL, pktmbuf_pool);
+               rte_eth_dev_socket_id(port), &rxq_conf, pktmbuf_pool);
        if (ret < 0)
                rte_exit(EXIT_FAILURE, "Could not setup up RX queue for "
                                "port%u (%d)\n", (unsigned)port, ret);
 
+       txq_conf = dev_info.default_txconf;
+       txq_conf.txq_flags = ETH_TXQ_FLAGS_IGNORE;
+       txq_conf.offloads = local_port_conf.txmode.offloads;
        ret = rte_eth_tx_queue_setup(port, 0, nb_txd,
-               rte_eth_dev_socket_id(port), NULL);
+               rte_eth_dev_socket_id(port), &txq_conf);
        if (ret < 0)
                rte_exit(EXIT_FAILURE, "Could not setup up TX queue for "
                                "port%u (%d)\n", (unsigned)port, ret);
@@ -673,7 +683,10 @@ static int
 kni_change_mtu(uint16_t port_id, unsigned int new_mtu)
 {
        int ret;
+       uint16_t nb_rxd = NB_RXD;
        struct rte_eth_conf conf;
+       struct rte_eth_dev_info dev_info;
+       struct rte_eth_rxconf rxq_conf;
 
        if (port_id >= rte_eth_dev_count()) {
                RTE_LOG(ERR, APP, "Invalid port id %d\n", port_id);
@@ -688,9 +701,9 @@ kni_change_mtu(uint16_t port_id, unsigned int new_mtu)
        memcpy(&conf, &port_conf, sizeof(conf));
        /* Set new MTU */
        if (new_mtu > ETHER_MAX_LEN)
-               conf.rxmode.jumbo_frame = 1;
+               conf.rxmode.offloads |= DEV_RX_OFFLOAD_JUMBO_FRAME;
        else
-               conf.rxmode.jumbo_frame = 0;
+               conf.rxmode.offloads &= ~DEV_RX_OFFLOAD_JUMBO_FRAME;
 
        /* mtu + length of header + length of FCS = max pkt length */
        conf.rxmode.max_rx_pkt_len = new_mtu + KNI_ENET_HEADER_SIZE +
@@ -701,6 +714,23 @@ kni_change_mtu(uint16_t port_id, unsigned int new_mtu)
                return ret;
        }
 
+       ret = rte_eth_dev_adjust_nb_rx_tx_desc(port_id, &nb_rxd, NULL);
+       if (ret < 0)
+               rte_exit(EXIT_FAILURE, "Could not adjust number of descriptors "
+                               "for port%u (%d)\n", (unsigned int)port_id,
+                               ret);
+
+       rte_eth_dev_info_get(port_id, &dev_info);
+       rxq_conf = dev_info.default_rxconf;
+       rxq_conf.offloads = conf.rxmode.offloads;
+       ret = rte_eth_rx_queue_setup(port_id, 0, nb_rxd,
+               rte_eth_dev_socket_id(port_id), &rxq_conf, pktmbuf_pool);
+       if (ret < 0) {
+               RTE_LOG(ERR, APP, "Fail to setup Rx queue of port %d\n",
+                               port_id);
+               return ret;
+       }
+
        /* Restart specific port */
        ret = rte_eth_dev_start(port_id);
        if (ret < 0) {