app/testpmd: various updates
authorIntel <intel.com>
Wed, 19 Dec 2012 23:00:00 +0000 (00:00 +0100)
committerThomas Monjalon <thomas.monjalon@6wind.com>
Thu, 25 Jul 2013 13:54:18 +0000 (15:54 +0200)
Signed-off-by: Intel
app/test-pmd/config.c
app/test-pmd/csumonly.c
app/test-pmd/parameters.c
app/test-pmd/testpmd.c
app/test-pmd/testpmd.h
app/test-pmd/txonly.c

index 809d2b5..64d7855 100644 (file)
@@ -214,7 +214,7 @@ port_infos_display(portid_t port_id)
                return;
        }
        port = &ports[port_id];
-       rte_eth_link_get(port_id, &link);
+       rte_eth_link_get_nowait(port_id, &link);
        printf("\n%s Infos for port %-2d %s\n",
               info_border, port_id, info_border);
        print_ethaddr("MAC address: ", &port->eth_addr);
@@ -665,6 +665,9 @@ simple_fwd_config_setup(void)
        cur_fwd_config.nb_fwd_streams =
                (streamid_t) cur_fwd_config.nb_fwd_ports;
 
+       /* reinitialize forwarding streams */
+       init_fwd_streams();
+
        /*
         * In the simple forwarding test, the number of forwarding cores
         * must be lower or equal to the number of forwarding ports.
@@ -730,6 +733,10 @@ rss_fwd_config_setup(void)
        else
                cur_fwd_config.nb_fwd_lcores =
                        (lcoreid_t)cur_fwd_config.nb_fwd_streams;
+
+       /* reinitialize forwarding streams */
+       init_fwd_streams();
+
        setup_fwd_config_of_each_lcore(&cur_fwd_config);
        rxp = 0; rxq = 0;
        for (lc_id = 0; lc_id < cur_fwd_config.nb_fwd_lcores; lc_id++) {
@@ -931,7 +938,7 @@ fwd_config_display(void)
        pkt_fwd_config_display(&cur_fwd_config);
 }
 
-void
+int
 set_fwd_lcores_list(unsigned int *lcorelist, unsigned int nb_lc)
 {
        unsigned int i;
@@ -943,13 +950,15 @@ set_fwd_lcores_list(unsigned int *lcorelist, unsigned int nb_lc)
        for (i = 0; i < nb_lc; i++) {
                lcore_cpuid = lcorelist[i];
                if (! rte_lcore_is_enabled(lcore_cpuid)) {
-                       printf("Logical core %u not enabled\n", lcore_cpuid);
-                       return;
+                       printf("lcore %u not enabled\n", lcore_cpuid);
+                       return -1;
                }
                if (lcore_cpuid == rte_get_master_lcore()) {
-                       printf("Master core %u cannot forward packets\n",
+                       printf("lcore %u cannot be masked on for running "
+                              "packet forwarding, which is the master lcore "
+                              "and reserved for command line parsing only\n",
                               lcore_cpuid);
-                       return;
+                       return -1;
                }
                if (record_now)
                        fwd_lcores_cpuids[i] = lcore_cpuid;
@@ -965,9 +974,11 @@ set_fwd_lcores_list(unsigned int *lcorelist, unsigned int nb_lc)
                       (unsigned int) nb_fwd_lcores, nb_lc);
                nb_fwd_lcores = (lcoreid_t) nb_lc;
        }
+
+       return 0;
 }
 
-void
+int
 set_fwd_lcores_mask(uint64_t lcoremask)
 {
        unsigned int lcorelist[64];
@@ -976,7 +987,7 @@ set_fwd_lcores_mask(uint64_t lcoremask)
 
        if (lcoremask == 0) {
                printf("Invalid NULL mask of cores\n");
-               return;
+               return -1;
        }
        nb_lc = 0;
        for (i = 0; i < 64; i++) {
@@ -984,7 +995,7 @@ set_fwd_lcores_mask(uint64_t lcoremask)
                        continue;
                lcorelist[nb_lc++] = i;
        }
-       set_fwd_lcores_list(lcorelist, nb_lc);
+       return set_fwd_lcores_list(lcorelist, nb_lc);
 }
 
 void
@@ -1422,13 +1433,13 @@ fdir_get_infos(portid_t port_id)
 
        rte_eth_dev_fdir_get_infos(port_id, &fdir_infos);
 
-       printf("\n  %s FDIR infos for port %-2d %s\n",
+       printf("\n  %s FDIR infos for port %-2d     %s\n",
               fdir_stats_border, port_id, fdir_stats_border);
 
-       printf("  collision: %-10"PRIu64" free: %-10"PRIu64"\n"
-              "  maxhash: %-10"PRIu64" maxlen: %-10"PRIu64"\n"
-              "  add : %-10"PRIu64"   remove : %-10"PRIu64"\n"
-              "  f_add: %-10"PRIu64" f_remove: %-10"PRIu64"\n",
+       printf("  collision: %-10"PRIu64"  free:     %"PRIu64"\n"
+              "  maxhash:   %-10"PRIu64"  maxlen:   %"PRIu64"\n"
+              "  add:       %-10"PRIu64"  remove:   %"PRIu64"\n"
+              "  f_add:     %-10"PRIu64"  f_remove: %"PRIu64"\n",
               (uint64_t)(fdir_infos.collision), (uint64_t)(fdir_infos.free),
               (uint64_t)(fdir_infos.maxhash), (uint64_t)(fdir_infos.maxlen),
               fdir_infos.add, fdir_infos.remove,
index b5bc5eb..d1e1a4e 100644 (file)
@@ -132,8 +132,8 @@ get_ipv4_cksum(struct ipv4_hdr *ipv4_hdr)
 }
 
 
-static inline
-uint16_t get_ipv4_psd_sum (struct ipv4_hdr * ip_hdr)
+static inline uint16_t
+get_ipv4_psd_sum (struct ipv4_hdr * ip_hdr)
 {
        struct psd_header psd_hdr;
        psd_hdr.src_addr = ip_hdr->src_addr;
@@ -145,11 +145,11 @@ uint16_t get_ipv4_psd_sum (struct ipv4_hdr * ip_hdr)
        return get_16b_sum((uint16_t*)&psd_hdr, sizeof(struct psd_header));
 }
 
-static inline
-uint16_t get_ipv6_psd_sum (struct ipv6_hdr * ip_hdr)
+static inline uint16_t
+get_ipv6_psd_sum (struct ipv6_hdr * ip_hdr)
 {
        struct ipv6_psd_header psd_hdr;
-       rte_memcpy(psd_hdr.src_addr, ip_hdr->src_addr, sizeof(ip_hdr->src_addr)
+       rte_memcpy(&psd_hdr.src_addr, ip_hdr->src_addr, sizeof(ip_hdr->src_addr)
                        + sizeof(ip_hdr->dst_addr));
 
        psd_hdr.zero[0]   = 0;
@@ -224,6 +224,7 @@ pkt_burst_checksum_forward(struct fwd_stream *fs)
        uint16_t pkt_ol_flags;
        uint16_t tx_ol_flags;
        uint16_t l4_proto;
+       uint16_t eth_type;
        uint8_t  l2_len;
        uint8_t  l3_len;
 
@@ -266,15 +267,30 @@ pkt_burst_checksum_forward(struct fwd_stream *fs)
                ol_flags = (uint16_t) (pkt_ol_flags & (~PKT_TX_L4_MASK));
 
                eth_hdr = (struct ether_hdr *) mb->pkt.data;
-               if (rte_be_to_cpu_16(eth_hdr->ether_type) == ETHER_TYPE_VLAN) {
+               eth_type = rte_be_to_cpu_16(eth_hdr->ether_type);
+               if (eth_type == ETHER_TYPE_VLAN) {
                        /* Only allow single VLAN label here */
                        l2_len  += sizeof(struct vlan_hdr);
+                        eth_type = rte_be_to_cpu_16(*(uint16_t *)
+                               ((uintptr_t)&eth_hdr->ether_type +
+                               sizeof(struct vlan_hdr)));
                }
 
                /* Update the L3/L4 checksum error packet count  */
                rx_bad_ip_csum += (uint16_t) ((pkt_ol_flags & PKT_RX_IP_CKSUM_BAD) != 0);
                rx_bad_l4_csum += (uint16_t) ((pkt_ol_flags & PKT_RX_L4_CKSUM_BAD) != 0);
 
+               /*
+                * Try to figure out L3 packet type by SW.
+                */
+               if ((pkt_ol_flags & (PKT_RX_IPV4_HDR | PKT_RX_IPV4_HDR_EXT |
+                               PKT_RX_IPV6_HDR | PKT_RX_IPV6_HDR_EXT)) == 0) {
+                       if (eth_type == ETHER_TYPE_IPv4)
+                               pkt_ol_flags |= PKT_RX_IPV4_HDR;
+                       else if (eth_type == ETHER_TYPE_IPv6)
+                               pkt_ol_flags |= PKT_RX_IPV6_HDR;
+               }
+
                /*
                 * Simplify the protocol parsing
                 * Assuming the incoming packets format as
@@ -354,8 +370,8 @@ pkt_burst_checksum_forward(struct fwd_stream *fs)
                        }
                        /* End of L4 Handling*/
                }
+               else if (pkt_ol_flags & PKT_RX_IPV6_HDR) {
 
-               else {
                        ipv6_hdr = (struct ipv6_hdr *) (rte_pktmbuf_mtod(mb,
                                        unsigned char *) + l2_len);
                        l3_len = sizeof(struct ipv6_hdr) ;
@@ -409,7 +425,11 @@ pkt_burst_checksum_forward(struct fwd_stream *fs)
                        } else {
                                printf("Test flow control for 1G PMD \n");
                        }
-                       /* End of L4 Handling*/
+                       /* End of L6 Handling*/
+               }
+               else {
+                       l3_len = 0;
+                       printf("Unhandled packet type: %#hx\n", eth_type);
                }
 
                /* Combine the packet header write. VLAN is not consider here */
index df56e26..6325b39 100644 (file)
@@ -102,7 +102,9 @@ usage(char* progname)
        printf("  --nb-ports=N set the number of forwarding ports"
               " (1 <= N <= %d)\n", nb_ports);
        printf("  --coremask=COREMASK: hexadecimal bitmask of cores running "
-              "the packet forwarding test\n");
+              "the packet forwarding test. The master lcore is reserved for "
+              "command line parsing only, and can not be masked on for running"
+              "the packet forwarding\n");
        printf("  --portmask=PORTMASK: hexadecimal bitmask of ports used "
               "by the packet forwarding test\n");
        printf("  --numa: enable NUMA-aware allocation of RX/TX rings and of "
@@ -129,6 +131,7 @@ usage(char* progname)
        printf("  --crc-strip: enable CRC stripping by hardware\n");
        printf("  --enable-rx-cksum: enable rx hardware checksum offload\n");
        printf("  --disable-hw-vlan: disable hardware vlan\n");
+       printf("  --enable-drop-en: enable per queue packet drop\n");
        printf("  --disable-rss: disable rss\n");
        printf("  --port-topology=N: set port topology (N: paired (default) or "
               "chained)\n");
@@ -212,8 +215,8 @@ parse_fwd_coremask(const char *coremask)
        cm = strtoull(coremask, &end, 16);
        if ((coremask[0] == '\0') || (end == NULL) || (*end != '\0'))
                rte_exit(EXIT_FAILURE, "Invalid fwd core mask\n");
-       else
-               set_fwd_lcores_mask((uint64_t) cm);
+       else if (set_fwd_lcores_mask((uint64_t) cm) < 0)
+               rte_exit(EXIT_FAILURE, "coremask is not valid\n");
 }
 
 /*
@@ -322,6 +325,8 @@ launch_args_parse(int argc, char** argv)
        int n, opt;
        char **argvopt;
        int opt_idx;
+       enum { TX, RX };
+
        static struct option lgopts[] = {
                { "help",                       0, 0, 0 },
                { "interactive",                0, 0, 0 },
@@ -342,7 +347,9 @@ launch_args_parse(int argc, char** argv)
                { "pkt-filter-flexbytes-offset",1, 0, 0 },
                { "pkt-filter-drop-queue",      1, 0, 0 },
                { "crc-strip",                  0, 0, 0 },
+               { "enable-rx-cksum",            0, 0, 0 },
                { "disable-hw-vlan",            0, 0, 0 },
+               { "enable-drop-en",            0, 0, 0 },
                { "disable-rss",                0, 0, 0 },
                { "port-topology",              1, 0, 0 },
                { "rss-ip",                     0, 0, 0 },
@@ -562,8 +569,8 @@ launch_args_parse(int argc, char** argv)
                        if (!strcmp(lgopts[opt_idx].name, "rss-ip"))
                                rss_hf = ETH_RSS_IPV4 | ETH_RSS_IPV6;
                        if (!strcmp(lgopts[opt_idx].name, "rss-udp"))
-                               rss_hf = ETH_RSS_IPV4 | ETH_RSS_IPV6 |
-                                       ETH_RSS_IPV4_UDP;
+                               rss_hf = ETH_RSS_IPV4 |
+                                               ETH_RSS_IPV6 | ETH_RSS_IPV4_UDP;
                        if (!strcmp(lgopts[opt_idx].name, "rxq")) {
                                n = atoi(optarg);
                                if (n >= 1 && n <= (int) MAX_QUEUE_ID)
index f2df50b..d4226dd 100644 (file)
@@ -628,40 +628,83 @@ static void
 fwd_port_stats_display(portid_t port_id, struct rte_eth_stats *stats)
 {
        struct rte_port *port;
+       uint8_t i;
 
        static const char *fwd_stats_border = "----------------------";
 
        port = &ports[port_id];
        printf("\n  %s Forward statistics for port %-2d %s\n",
-               fwd_stats_border, port_id, fwd_stats_border);
-       printf("  RX-packets: %-14"PRIu64" RX-dropped: %-14"PRIu64"RX-total: "
-              "%-"PRIu64"\n",
-              stats->ipackets, stats->ierrors,
-              (uint64_t) (stats->ipackets + stats->ierrors));
+              fwd_stats_border, port_id, fwd_stats_border);
 
-       if (cur_fwd_eng == &csum_fwd_engine)
-               printf("  Bad-ipcsum: %-14"PRIu64" Bad-l4csum: %-14"PRIu64" \n",
-                               port->rx_bad_ip_csum, port->rx_bad_l4_csum);
+       if ((!port->rx_queue_stats_mapping_enabled) && (!port->tx_queue_stats_mapping_enabled)) {
+               printf("  RX-packets: %-14"PRIu64" RX-dropped: %-14"PRIu64"RX-total: "
+                      "%-"PRIu64"\n",
+                      stats->ipackets, stats->ierrors,
+                      (uint64_t) (stats->ipackets + stats->ierrors));
 
-       printf("  TX-packets: %-14"PRIu64" TX-dropped: %-14"PRIu64"TX-total: "
-              "%-"PRIu64"\n",
-              stats->opackets, port->tx_dropped,
-              (uint64_t) (stats->opackets + port->tx_dropped));
+               if (cur_fwd_eng == &csum_fwd_engine)
+                       printf("  Bad-ipcsum: %-14"PRIu64" Bad-l4csum: %-14"PRIu64" \n",
+                              port->rx_bad_ip_csum, port->rx_bad_l4_csum);
 
-       if (stats->rx_nombuf > 0)
-               printf("  RX-nombufs: %-14"PRIu64"\n", stats->rx_nombuf);
+               printf("  TX-packets: %-14"PRIu64" TX-dropped: %-14"PRIu64"TX-total: "
+                      "%-"PRIu64"\n",
+                      stats->opackets, port->tx_dropped,
+                      (uint64_t) (stats->opackets + port->tx_dropped));
+
+               if (stats->rx_nombuf > 0)
+                       printf("  RX-nombufs: %-14"PRIu64"\n", stats->rx_nombuf);
+
+       }
+       else {
+               printf("  RX-packets:             %14"PRIu64"    RX-dropped:%14"PRIu64"    RX-total:"
+                      "%14"PRIu64"\n",
+                      stats->ipackets, stats->ierrors,
+                      (uint64_t) (stats->ipackets + stats->ierrors));
+
+               if (cur_fwd_eng == &csum_fwd_engine)
+                       printf("  Bad-ipcsum:%14"PRIu64"    Bad-l4csum:%14"PRIu64"\n",
+                              port->rx_bad_ip_csum, port->rx_bad_l4_csum);
+
+               printf("  TX-packets:             %14"PRIu64"    TX-dropped:%14"PRIu64"    TX-total:"
+                      "%14"PRIu64"\n",
+                      stats->opackets, port->tx_dropped,
+                      (uint64_t) (stats->opackets + port->tx_dropped));
+
+               if (stats->rx_nombuf > 0)
+                       printf("  RX-nombufs:%14"PRIu64"\n", stats->rx_nombuf);
+       }
 #ifdef RTE_TEST_PMD_RECORD_BURST_STATS
        if (port->rx_stream)
-               pkt_burst_stats_display("RX", &port->rx_stream->rx_burst_stats);
+               pkt_burst_stats_display("RX",
+                       &port->rx_stream->rx_burst_stats);
        if (port->tx_stream)
-               pkt_burst_stats_display("TX", &port->tx_stream->tx_burst_stats);
+               pkt_burst_stats_display("TX",
+                       &port->tx_stream->tx_burst_stats);
 #endif
        /* stats fdir */
        if (fdir_conf.mode != RTE_FDIR_MODE_NONE)
-               printf("  Fdirmiss: %-14"PRIu64"   Fdirmatch: %-14"PRIu64"\n",
+               printf("  Fdirmiss:%14"PRIu64"    Fdirmatch:%14"PRIu64"\n",
                       stats->fdirmiss,
                       stats->fdirmatch);
 
+       if (port->rx_queue_stats_mapping_enabled) {
+               printf("\n");
+               for (i = 0; i < RTE_ETHDEV_QUEUE_STAT_CNTRS; i++) {
+                       printf("  Stats reg %2d RX-packets:%14"PRIu64
+                              "     RX-errors:%14"PRIu64
+                              "    RX-bytes:%14"PRIu64"\n",
+                              i, stats->q_ipackets[i], stats->q_errors[i], stats->q_ibytes[i]);
+               }
+               printf("\n");
+       }
+       if (port->tx_queue_stats_mapping_enabled) {
+               for (i = 0; i < RTE_ETHDEV_QUEUE_STAT_CNTRS; i++) {
+                       printf("  Stats reg %2d TX-packets:%14"PRIu64
+                              "                                 TX-bytes:%14"PRIu64"\n",
+                              i, stats->q_opackets[i], stats->q_obytes[i]);
+               }
+       }
+
        printf("  %s--------------------------------%s\n",
               fwd_stats_border, fwd_stats_border);
 }
@@ -685,8 +728,8 @@ fwd_stream_stats_display(streamid_t stream_id)
 
        /* if checksum mode */
        if (cur_fwd_eng == &csum_fwd_engine) {
-              printf("  RX- bad IP checksum: %-14u  Rx- bad L4 checksum: %-14u\n",
-              fs->rx_bad_ip_csum, fs->rx_bad_l4_csum);
+              printf("  RX- bad IP checksum: %-14u  Rx- bad L4 checksum: "
+                       "%-14u\n", fs->rx_bad_ip_csum, fs->rx_bad_l4_csum);
        }
 
 #ifdef RTE_TEST_PMD_RECORD_BURST_STATS
@@ -710,8 +753,7 @@ flush_all_rx_queues(void)
                        for (rxq = 0; rxq < nb_rxq; rxq++) {
                                do {
                                        nb_rx = rte_eth_rx_burst(rxp, rxq,
-                                                                pkts_burst,
-                                                                MAX_PKT_BURST);
+                                               pkts_burst, MAX_PKT_BURST);
                                        for (i = 0; i < nb_rx; i++)
                                                rte_pktmbuf_free(pkts_burst[i]);
                                } while (nb_rx > 0);
@@ -813,6 +855,10 @@ start_packet_forwarding(int with_tx_first)
                printf("Packet forwarding already started\n");
                return;
        }
+       if((dcb_test) && (nb_fwd_lcores == 1)) {
+               printf("In DCB mode,the nb forwarding cores should be larger than 1.\n");
+               return;
+       }
        test_done = 0;
        flush_all_rx_queues();
        fwd_config_setup();
@@ -823,6 +869,8 @@ start_packet_forwarding(int with_tx_first)
                port = &ports[pt_id];
                rte_eth_stats_get(pt_id, &port->stats);
                port->tx_dropped = 0;
+
+               map_port_queue_stats_mapping_registers(pt_id, port);
        }
        for (sm_id = 0; sm_id < cur_fwd_config.nb_fwd_streams; sm_id++) {
                fwd_streams[sm_id]->rx_packets = 0;
@@ -921,15 +969,19 @@ stop_packet_forwarding(void)
                                         fwd_streams[sm_id]->fwd_dropped);
                ports[fwd_streams[sm_id]->tx_port].tx_dropped = tx_dropped;
 
-               rx_bad_ip_csum = ports[fwd_streams[sm_id]->rx_port].rx_bad_ip_csum;
+               rx_bad_ip_csum =
+                       ports[fwd_streams[sm_id]->rx_port].rx_bad_ip_csum;
                rx_bad_ip_csum = (uint64_t) (rx_bad_ip_csum +
                                         fwd_streams[sm_id]->rx_bad_ip_csum);
-               ports[fwd_streams[sm_id]->rx_port].rx_bad_ip_csum = rx_bad_ip_csum;
+               ports[fwd_streams[sm_id]->rx_port].rx_bad_ip_csum =
+                                                       rx_bad_ip_csum;
 
-               rx_bad_l4_csum = ports[fwd_streams[sm_id]->rx_port].rx_bad_l4_csum;
+               rx_bad_l4_csum =
+                       ports[fwd_streams[sm_id]->rx_port].rx_bad_l4_csum;
                rx_bad_l4_csum = (uint64_t) (rx_bad_l4_csum +
                                         fwd_streams[sm_id]->rx_bad_l4_csum);
-               ports[fwd_streams[sm_id]->rx_port].rx_bad_l4_csum = rx_bad_l4_csum;
+               ports[fwd_streams[sm_id]->rx_port].rx_bad_l4_csum =
+                                                       rx_bad_l4_csum;
 
 #ifdef RTE_TEST_PMD_RECORD_CORE_CYCLES
                fwd_cycles = (uint64_t) (fwd_cycles +
@@ -1298,97 +1350,241 @@ check_all_ports_link_status(uint8_t port_num, uint32_t port_mask)
        }
 }
 
-static void
-init_ports(void)
+static int
+set_tx_queue_stats_mapping_registers(uint8_t port_id, struct rte_port *port)
 {
-       struct rte_eth_link   link;
-       struct rte_eth_conf   port_conf = {
-               .intr_conf = {
-                       .lsc = 0,
-               },
-       };
-       struct rte_eth_rxconf rx_conf;
-       struct rte_eth_txconf tx_conf;
-       struct rte_port *port;
-       unsigned int sock_id;
-       portid_t  pi;
-       queueid_t qi;
+       uint16_t i;
        int diag;
+       uint8_t mapping_found = 0;
+
+       for (i = 0; i < nb_tx_queue_stats_mappings; i++) {
+               if ((tx_queue_stats_mappings[i].port_id == port_id) &&
+                               (tx_queue_stats_mappings[i].queue_id < nb_txq )) {
+                       diag = rte_eth_dev_set_tx_queue_stats_mapping(port_id,
+                                       tx_queue_stats_mappings[i].queue_id,
+                                       tx_queue_stats_mappings[i].stats_counter_id);
+                       if (diag != 0)
+                               return diag;
+                       mapping_found = 1;
+               }
+       }
+       if (mapping_found)
+               port->tx_queue_stats_mapping_enabled = 1;
+       return 0;
+}
 
-       port_conf.rxmode = rx_mode;
-       port_conf.fdir_conf = fdir_conf;
-
-       if (nb_rxq > 0) { /* configure RSS */
-               port_conf.rx_adv_conf.rss_conf.rss_key = NULL;
-               /* use default hash key */
-               port_conf.rx_adv_conf.rss_conf.rss_hf = rss_hf;
-       } else
-               port_conf.rx_adv_conf.rss_conf.rss_hf = 0;
-       rx_conf.rx_thresh = rx_thresh;
-       rx_conf.rx_free_thresh = rx_free_thresh;
-       tx_conf.tx_thresh = tx_thresh;
-       tx_conf.tx_rs_thresh = tx_rs_thresh;
-       tx_conf.tx_free_thresh = tx_free_thresh;
+static int
+set_rx_queue_stats_mapping_registers(uint8_t port_id, struct rte_port *port)
+{
+       uint16_t i;
+       int diag;
+       uint8_t mapping_found = 0;
+
+       for (i = 0; i < nb_rx_queue_stats_mappings; i++) {
+               if ((rx_queue_stats_mappings[i].port_id == port_id) &&
+                               (rx_queue_stats_mappings[i].queue_id < nb_rxq )) {
+                       diag = rte_eth_dev_set_rx_queue_stats_mapping(port_id,
+                                       rx_queue_stats_mappings[i].queue_id,
+                                       rx_queue_stats_mappings[i].stats_counter_id);
+                       if (diag != 0)
+                               return diag;
+                       mapping_found = 1;
+               }
+       }
+       if (mapping_found)
+               port->rx_queue_stats_mapping_enabled = 1;
+       return 0;
+}
 
-       for (pi = 0; pi < nb_ports; pi++) {
-               port = &ports[pi];
-               memcpy(&port->dev_conf, &port_conf, sizeof(port_conf));
-               sock_id = port->socket_id;
-               printf("Initializing port %d... ", pi);
-               fflush(stdout);
-               diag = rte_eth_dev_configure(pi, nb_rxq, nb_txq, &port_conf);
-               if (diag != 0) {
-                       fatal_init_error("rte_eth_dev_configure", pi, diag);
-                       /* NOT REACHED */
+static void
+map_port_queue_stats_mapping_registers(uint8_t pi, struct rte_port *port)
+{
+       int diag = 0;
+
+       diag = set_tx_queue_stats_mapping_registers(pi, port);
+       if (diag != 0) {
+               if (diag == -ENOTSUP) {
+                       port->tx_queue_stats_mapping_enabled = 0;
+                       printf("TX queue stats mapping not supported port id=%d\n", pi);
                }
-               rte_eth_macaddr_get(pi, &port->eth_addr);
-               for (qi = 0; qi < nb_txq; qi++) {
-                       diag = rte_eth_tx_queue_setup(pi, qi, nb_txd,
-                                                     sock_id,
-                                                     &tx_conf);
-                       if (diag != 0) {
-                               fatal_init_error("rte_eth_tx_queue_setup",
-                                                pi, diag);
-                               /* NOT REACHED */
-                       }
+               else
+                       rte_exit(EXIT_FAILURE,
+                                       "set_tx_queue_stats_mapping_registers "
+                                       "failed for port id=%d diag=%d\n",
+                                       pi, diag);
+       }
+
+       diag = set_rx_queue_stats_mapping_registers(pi, port);
+       if (diag != 0) {
+               if (diag == -ENOTSUP) {
+                       port->rx_queue_stats_mapping_enabled = 0;
+                       printf("RX queue stats mapping not supported port id=%d\n", pi);
                }
-               for (qi = 0; qi < nb_rxq; qi++) {
-                       diag = rte_eth_rx_queue_setup(pi, qi, nb_rxd, sock_id,
-                                                     &rx_conf,
-                                                     mbuf_pool_find(sock_id));
-                       if (diag != 0) {
-                               fatal_init_error("rte_eth_rx_queue_setup",
-                                                pi , diag);
-                               /* NOT REACHED */
-                       }
+               else
+                       rte_exit(EXIT_FAILURE,
+                                       "set_rx_queue_stats_mapping_registers "
+                                       "failed for port id=%d diag=%d\n",
+                                       pi, diag);
+       }
+}
+
+void
+init_port_config(void)
+{
+       portid_t pid;
+       struct rte_port *port;
+
+       for (pid = 0; pid < nb_ports; pid++) {
+               port = &ports[pid];
+               port->dev_conf.rxmode = rx_mode;
+               port->dev_conf.fdir_conf = fdir_conf;
+               if (nb_rxq > 0) {
+                       port->dev_conf.rx_adv_conf.rss_conf.rss_key = NULL;
+                       port->dev_conf.rx_adv_conf.rss_conf.rss_hf = rss_hf;
+               } else {
+                       port->dev_conf.rx_adv_conf.rss_conf.rss_key = NULL;
+                       port->dev_conf.rx_adv_conf.rss_conf.rss_hf = 0;
                }
+               port->rx_conf.rx_thresh = rx_thresh;
+               port->rx_conf.rx_free_thresh = rx_free_thresh;
+               port->rx_conf.rx_drop_en = rx_drop_en;
+               port->tx_conf.tx_thresh = tx_thresh;
+               port->tx_conf.tx_rs_thresh = tx_rs_thresh;
+               port->tx_conf.tx_free_thresh = tx_free_thresh;
+               port->tx_conf.txq_flags = txq_flags;
+
+               rte_eth_macaddr_get(pid, &port->eth_addr);
 
-               /* Start device */
-               diag = rte_eth_dev_start(pi);
-               if (diag != 0) {
-                       fatal_init_error("rte_eth_dev_start", pi, diag);
-                       /* NOT REACHED */
+               map_port_queue_stats_mapping_registers(pid, port);
+       }
+}
+
+const uint16_t vlan_tags[] = {
+               0,  1,  2,  3,  4,  5,  6,  7,
+               8,  9, 10, 11,  12, 13, 14, 15,
+               16, 17, 18, 19, 20, 21, 22, 23,
+               24, 25, 26, 27, 28, 29, 30, 31
+};
+
+static  int
+get_eth_dcb_conf(struct rte_eth_conf *eth_conf, struct dcb_config *dcb_conf)
+{
+        uint8_t i;
+       /*
+        * Builds up the correct configuration for dcb+vt based on the vlan tags array
+        * given above, and the number of traffic classes available for use.
+        */
+       if (dcb_conf->dcb_mode == DCB_VT_ENABLED) {
+               struct rte_eth_vmdq_dcb_conf vmdq_rx_conf;
+               struct rte_eth_vmdq_dcb_tx_conf vmdq_tx_conf;
+               /* VMDQ+DCB RX and TX configrations */
+               vmdq_rx_conf.enable_default_pool = 0;
+               vmdq_rx_conf.default_pool = 0;
+               vmdq_rx_conf.nb_queue_pools =
+                       (dcb_conf->num_tcs ==  ETH_4_TCS ? ETH_32_POOLS : ETH_16_POOLS);
+               vmdq_tx_conf.nb_queue_pools =
+                       (dcb_conf->num_tcs ==  ETH_4_TCS ? ETH_32_POOLS : ETH_16_POOLS);
+               vmdq_rx_conf.nb_pool_maps = sizeof( vlan_tags )/sizeof( vlan_tags[ 0 ]);
+               for (i = 0; i < vmdq_rx_conf.nb_pool_maps; i++) {
+                       vmdq_rx_conf.pool_map[i].vlan_id = vlan_tags[ i ];
+                       vmdq_rx_conf.pool_map[i].pools = 1 << (i % vmdq_rx_conf.nb_queue_pools);
                }
-               printf("done: ");
-               rte_eth_link_get(pi, &link);
-               if (link.link_status) {
-                       printf(" Link Up - speed %u Mbps - %s\n",
-                              (unsigned) link.link_speed,
-                              (link.link_duplex == ETH_LINK_FULL_DUPLEX) ?
-                              ("full-duplex") : ("half-duplex\n"));
-               } else {
-                       printf(" Link Down\n");
+               for (i = 0; i < ETH_DCB_NUM_USER_PRIORITIES; i++) {
+                       vmdq_rx_conf.dcb_queue[i] = i;
+                       vmdq_tx_conf.dcb_queue[i] = i;
                }
+               /*set DCB mode of RX and TX of multiple queues*/
+               eth_conf->rxmode.mq_mode = ETH_VMDQ_DCB;
+               eth_conf->txmode.mq_mode = ETH_VMDQ_DCB_TX;
+               if (dcb_conf->pfc_en)
+                       eth_conf->dcb_capability_en = ETH_DCB_PG_SUPPORT|ETH_DCB_PFC_SUPPORT;
+               else
+                       eth_conf->dcb_capability_en = ETH_DCB_PG_SUPPORT;
+               (void)(rte_memcpy(&eth_conf->rx_adv_conf.vmdq_dcb_conf, &vmdq_rx_conf,
+                                sizeof(struct rte_eth_vmdq_dcb_conf)));
+               (void)(rte_memcpy(&eth_conf->tx_adv_conf.vmdq_dcb_tx_conf, &vmdq_tx_conf,
+                                sizeof(struct rte_eth_vmdq_dcb_tx_conf)));
+       }
+       else {
+               struct rte_eth_dcb_rx_conf rx_conf;
+               struct rte_eth_dcb_tx_conf tx_conf;
+               /* queue mapping configuration of DCB RX and TX */
+               if (dcb_conf->num_tcs == ETH_4_TCS)
+                       dcb_q_mapping = DCB_4_TCS_Q_MAPPING;
+               else
+                       dcb_q_mapping = DCB_8_TCS_Q_MAPPING;
+               rx_conf.nb_tcs = dcb_conf->num_tcs;
+               tx_conf.nb_tcs = dcb_conf->num_tcs;
+               for (i = 0; i < ETH_DCB_NUM_USER_PRIORITIES; i++){
+                       rx_conf.dcb_queue[i] = i;
+                       tx_conf.dcb_queue[i] = i;
+               }
+               eth_conf->rxmode.mq_mode = ETH_DCB_RX;
+               eth_conf->txmode.mq_mode = ETH_DCB_TX;
+               if (dcb_conf->pfc_en)
+                       eth_conf->dcb_capability_en = ETH_DCB_PG_SUPPORT|ETH_DCB_PFC_SUPPORT;
+               else
+                       eth_conf->dcb_capability_en = ETH_DCB_PG_SUPPORT;
+                
+               (void)(rte_memcpy(&eth_conf->rx_adv_conf.dcb_rx_conf, &rx_conf,
+                                sizeof(struct rte_eth_dcb_rx_conf)));
+               (void)(rte_memcpy(&eth_conf->tx_adv_conf.dcb_tx_conf, &tx_conf,
+                                sizeof(struct rte_eth_dcb_tx_conf)));
+       }
 
-               /*
-                * If enabled, put device in promiscuous mode.
-                * This allows the PMD test in IO forwarding mode to forward
-                * packets to itself through 2 cross-connected  ports of the
-                * target machine.
-                */
-               if (promiscuous_on)
-                       rte_eth_promiscuous_enable(pi);
+       return 0;
+}
+
+int
+init_port_dcb_config(portid_t pid,struct dcb_config *dcb_conf)
+{
+       struct rte_eth_conf port_conf;
+       struct rte_port *rte_port;
+       int retval;
+       uint16_t nb_vlan;
+       uint16_t i;
+       /* rxq and txq configuration in dcb mode */
+       nb_rxq = 128;
+       nb_txq = 128;
+       rx_free_thresh = 64;
+       memset(&port_conf,0,sizeof(struct rte_eth_conf));
+       /* Enter DCB configuration status */
+       dcb_config = 1;
+       nb_vlan = sizeof( vlan_tags )/sizeof( vlan_tags[ 0 ]);
+       /*set configuration of DCB in vt mode and DCB in non-vt mode*/
+       retval = get_eth_dcb_conf(&port_conf, dcb_conf);
+       if (retval < 0)
+               return retval;
+       rte_port = &ports[pid];
+       memcpy(&rte_port->dev_conf, &port_conf,sizeof(struct rte_eth_conf));
+       rte_port->rx_conf.rx_thresh = rx_thresh;
+       rte_port->rx_conf.rx_free_thresh = rx_free_thresh;
+       rte_port->tx_conf.tx_thresh = tx_thresh;
+       rte_port->tx_conf.tx_rs_thresh = tx_rs_thresh;
+       rte_port->tx_conf.tx_free_thresh = tx_free_thresh;
+       /* VLAN filter */
+       rte_port->dev_conf.rxmode.hw_vlan_filter = 1;
+       for (i = 0; i < nb_vlan; i++){
+               rx_vft_set(pid, vlan_tags[i], 1);
        }
+       rte_eth_macaddr_get(pid, &rte_port->eth_addr);
+       map_port_queue_stats_mapping_registers(pid, rte_port);
+       return 0;
 }
 
 #ifdef RTE_EXEC_ENV_BAREMETAL
@@ -1399,6 +1595,7 @@ int
 main(int argc, char** argv)
 {
        int  diag;
+       uint8_t port_id;
 
        diag = rte_eal_init(argc, argv);
        if (diag < 0)
@@ -1412,7 +1609,8 @@ main(int argc, char** argv)
 
        nb_ports = (portid_t) rte_eth_dev_count();
        if (nb_ports == 0)
-               rte_exit(EXIT_FAILURE, "No probed ethernet devices - check that "
+               rte_exit(EXIT_FAILURE, "No probed ethernet devices - "
+                                                       "check that "
                          "CONFIG_RTE_LIBRTE_IGB_PMD=y and that "
                          "CONFIG_RTE_LIBRTE_EM_PMD=y and that "
                          "CONFIG_RTE_LIBRTE_IXGBE_PMD=y in your "
index 0b4b42b..dbaa8f5 100644 (file)
@@ -121,7 +121,7 @@ struct fwd_stream {
 /**
  * The data structure associated with each port.
  * tx_ol_flags is slightly different from ol_flags of rte_mbuf.
- *      Bit  0: Insert IP checksum
+ *   Bit  0: Insert IP checksum
  *   Bit  1: Insert UDP checksum
  *   Bit  2: Insert TCP checksum
  *   Bit  3: Insert SCTP checksum
@@ -409,6 +409,8 @@ void fwd_config_display(void);
 void rxtx_config_display(void);
 void fwd_config_setup(void);
 void set_def_fwd_config(void);
+int init_fwd_streams(void);
+
 
 void port_reg_bit_display(portid_t port_id, uint32_t reg_off, uint8_t bit_pos);
 void port_reg_bit_set(portid_t port_id, uint32_t reg_off, uint8_t bit_pos,
@@ -423,8 +425,8 @@ void port_reg_set(portid_t port_id, uint32_t reg_off, uint32_t value);
 void rx_ring_desc_display(portid_t port_id, queueid_t rxq_id, uint16_t rxd_id);
 void tx_ring_desc_display(portid_t port_id, queueid_t txq_id, uint16_t txd_id);
 
-void set_fwd_lcores_list(unsigned int *lcorelist, unsigned int nb_lc);
-void set_fwd_lcores_mask(uint64_t lcoremask);
+int set_fwd_lcores_list(unsigned int *lcorelist, unsigned int nb_lc);
+int set_fwd_lcores_mask(uint64_t lcoremask);
 void set_fwd_lcores_number(uint16_t nb_lc);
 
 void set_fwd_ports_list(unsigned int *portlist, unsigned int nb_pt);
index 63cc47a..b943ca1 100644 (file)
@@ -94,8 +94,8 @@ tx_mbuf_alloc(struct rte_mempool *mp)
        struct rte_mbuf *m;
 
        m = __rte_mbuf_raw_alloc(mp);
-       __rte_mbuf_sanity_check(m, RTE_MBUF_PKT, 1);
-       return m;
+       __rte_mbuf_sanity_check_raw(m, RTE_MBUF_PKT, 0);
+       return (m);
 }
 
 static void