Cisco VIC card - support for receive all mode, and work around for IPv6 issues

Signed-off-by: Ido Barnea <ibarnea@cisco.com>

3 files changed, 109 insertions, 72 deletions

diff --git a/src/dpdk/drivers/net/enic/base/vnic_dev.c b/src/dpdk/drivers/net/enic/base/vnic_dev.c
index 713b6089..e50b90e7 100644
--- a/src/dpdk/drivers/net/enic/base/vnic_dev.c
+++ b/src/dpdk/drivers/net/enic/base/vnic_dev.c
@@ -667,7 +667,12 @@ int vnic_dev_packet_filter(struct vnic_dev *vdev, int directed, int multicast,
 	     (promisc ? CMD_PFILTER_PROMISCUOUS : 0) |
 	     (allmulti ? CMD_PFILTER_ALL_MULTICAST : 0);
 
+#define TREX_PATCH
+#ifdef TREX_PATCH
+	err = vnic_dev_cmd(vdev, CMD_PACKET_FILTER_ALL, &a0, &a1, wait);
+#else
 	err = vnic_dev_cmd(vdev, CMD_PACKET_FILTER, &a0, &a1, wait);
+#endif
 	if (err)
 		pr_err("Can't set packet filter\n");
 
diff --git a/src/dpdk/drivers/net/enic/enic_clsf.c b/src/dpdk/drivers/net/enic/enic_clsf.c
index 23cb0124..8f68faab 100644
--- a/src/dpdk/drivers/net/enic/enic_clsf.c
+++ b/src/dpdk/drivers/net/enic/enic_clsf.c
@@ -107,6 +107,7 @@ enic_set_layer(struct filter_generic_1 *gp, unsigned int flag,
 	memcpy(gp->layer[layer].val, val, len);
 }
 
+
 /* Copy Flow Director filter to a VIC ipv4 filter (for Cisco VICs
  * without advanced filter support.
  */
@@ -132,6 +133,28 @@ copy_fltr_v1(struct filter_v2 *fltr, struct rte_eth_fdir_input *input,
 	fltr->u.ipv4.flags = FILTER_FIELDS_IPV4_5TUPLE;
 }
 
+#define TREX_PATCH
+#ifdef TREX_PATCH
+void
+copy_fltr_recv_all(struct filter_v2 *fltr, struct rte_eth_fdir_input *input,
+             struct rte_eth_fdir_masks *masks) {
+	struct filter_generic_1 *gp = &fltr->u.generic_1;
+	memset(gp, 0, sizeof(*gp));
+
+    struct ether_hdr eth_mask, eth_val;
+    memset(&eth_mask, 0, sizeof(eth_mask));
+    memset(&eth_val, 0, sizeof(eth_val));
+
+    eth_val.ether_type = 0x0806;
+    eth_mask.ether_type = 0;
+
+    gp->position = 0;
+    enic_set_layer(gp, 0, FILTER_GENERIC_1_L2,
+			       &eth_mask, &eth_val, sizeof(struct ether_hdr));
+
+}
+#endif
+
 /* Copy Flow Director filter to a VIC generic filter (requires advanced
  * filter support.
  */
@@ -146,6 +169,11 @@ copy_fltr_v2(struct filter_v2 *fltr, struct rte_eth_fdir_input *input,
 
 	fltr->type = FILTER_DPDK_1;
 	memset(gp, 0, sizeof(*gp));
+#ifdef TREX_PATCH
+    // important for this to be below 2.
+    // If added with position 2, IPv6 UDP and ICMP seems to be caught by some other rule
+    gp->position = 1;
+#endif
 
 	if (input->flow_type == RTE_ETH_FLOW_NONFRAG_IPV4_UDP) {
 		struct udp_hdr udp_mask, udp_val;
@@ -344,11 +372,23 @@ int enic_fdir_del_fltr(struct enic *enic, struct rte_eth_fdir_filter *params)
 	case -EINVAL:
 	case -ENOENT:
 		enic->fdir.stats.f_remove++;
+#ifdef TREX_PATCH
+		return pos;
+#else
 		return -EINVAL;
+#endif
 	default:
 		/* The entry is present in the table */
 		key = enic->fdir.nodes[pos];
 
+#ifdef TREX_PATCH
+        switch (params->soft_id) {
+        case 100:
+            // remove promisc when we delete 'receive all' filter
+            vnic_dev_packet_filter(enic->vdev, 1, 1, 1, 0, 1);
+            break;
+        }
+#endif
 		/* Delete the filter */
 		vnic_dev_classifier(enic->vdev, CLSF_DEL,
 			&key->fltr_id, NULL);
@@ -455,8 +495,19 @@ int enic_fdir_add_fltr(struct enic *enic, struct rte_eth_fdir_filter *params)
 	key->filter = *params;
 	key->rq_index = queue;
 
-	enic->fdir.copy_fltr_fn(&fltr, &params->input,
-				&enic->rte_dev->data->dev_conf.fdir_conf.mask);
+#ifdef TREX_PATCH
+    switch (params->soft_id) {
+    case 100:
+        vnic_dev_packet_filter(enic->vdev, 1, 1, 1, 1, 1);
+        copy_fltr_recv_all(&fltr, &params->input, &enic->rte_dev->data->dev_conf.fdir_conf.mask);
+        break;
+    default:
+#endif
+        enic->fdir.copy_fltr_fn(&fltr, &params->input,
+                                &enic->rte_dev->data->dev_conf.fdir_conf.mask);
+#ifdef TREX_PATCH
+    }
+#endif
 
 	if (!vnic_dev_classifier(enic->vdev, CLSF_ADD, &queue, &fltr)) {
 		key->fltr_id = queue;
diff --git a/src/main_dpdk.cpp b/src/main_dpdk.cpp
index a88d4a7c..6d78719f 100644
--- a/src/main_dpdk.cpp
+++ b/src/main_dpdk.cpp
@@ -443,8 +443,8 @@ public:
 
 private:
 
-    virtual void add_del_rules(enum rte_filter_op op, uint8_t port_id, uint16_t type, uint8_t ttl
-                               , uint16_t ip_id, uint8_t l4_proto, int queue);
+    virtual void add_del_rules(enum rte_filter_op op, uint8_t port_id, uint16_t type, uint16_t id
+                               , uint8_t l4_proto, uint8_t tos, int queue);
     virtual int add_del_eth_type_rule(uint8_t port_id, enum rte_filter_op op, uint16_t eth_type);
     virtual int configure_rx_filter_rules_statefull(CPhyEthIF * _if);
 
@@ -6583,7 +6583,7 @@ void CTRexExtendedDriverBaseMlnx5G::add_del_rules(enum rte_filter_op op, uint8_t
     memset(&filter,0,sizeof(struct rte_eth_fdir_filter));
 
 #if 0
-    printf("40g::%s rules: port:%d type:%d ip_id:%x l4:%d q:%d\n"
+    printf("MLNX add_del_rules::%s rules: port:%d type:%d ip_id:%x l4:%d q:%d\n"
            , (op == RTE_ETH_FILTER_ADD) ?  "add" : "del"
            , port_id, type, ip_id, l4_proto, queue);
 #endif
@@ -6599,7 +6599,6 @@ void CTRexExtendedDriverBaseMlnx5G::add_del_rules(enum rte_filter_op op, uint8_t
     case RTE_ETH_FLOW_NONFRAG_IPV4_TCP:
     case RTE_ETH_FLOW_NONFRAG_IPV4_SCTP:
     case RTE_ETH_FLOW_NONFRAG_IPV4_OTHER:
-        //        filter.input.flow.ip4_flow.ttl = ttl;
         filter.input.flow.ip4_flow.ip_id = ip_id;
         if (l4_proto != 0)
             filter.input.flow.ip4_flow.proto = l4_proto;
@@ -6607,7 +6606,6 @@ void CTRexExtendedDriverBaseMlnx5G::add_del_rules(enum rte_filter_op op, uint8_t
     case RTE_ETH_FLOW_NONFRAG_IPV6_UDP:
     case RTE_ETH_FLOW_NONFRAG_IPV6_TCP:
     case RTE_ETH_FLOW_NONFRAG_IPV6_OTHER:
-        //        filter.input.flow.ipv6_flow.hop_limits=ttl;
         filter.input.flow.ipv6_flow.flow_label = ip_id;
         filter.input.flow.ipv6_flow.proto = l4_proto;
         break;
@@ -6745,30 +6743,13 @@ void CTRexExtendedDriverBaseVIC::update_configuration(port_cfg_t * cfg){
     cfg->m_tx_conf.tx_thresh.hthresh = TX_HTHRESH;
     cfg->m_tx_conf.tx_thresh.wthresh = TX_WTHRESH;
     cfg->m_port_conf.rxmode.max_rx_pkt_len =9*1000-10;
-
-    if (get_is_stateless()) {
-        /* work in TOS mode */
-        cfg->m_port_conf.fdir_conf.mask.ipv4_mask.tos = 0x01;
-        cfg->m_port_conf.fdir_conf.mask.ipv6_mask.tc  = 0x01;
-    }else{
-        #ifdef VIC_TTL_FILTER
-          cfg->m_port_conf.fdir_conf.mask.ipv4_mask.ttl = 0xff;
-          cfg->m_port_conf.fdir_conf.mask.ipv6_mask.hop_limits  = 0xff;
-        #else
-          cfg->m_port_conf.fdir_conf.mask.ipv4_mask.tos = 0x01;
-          cfg->m_port_conf.fdir_conf.mask.ipv6_mask.tc  = 0x01;
-        #endif
-    }
+    cfg->m_port_conf.fdir_conf.mask.ipv4_mask.tos = 0x0f;
+    cfg->m_port_conf.fdir_conf.mask.ipv6_mask.tc  = 0x0f;
 }
 
-
-/* Add rule to send packets with protocol 'type', and ttl 'ttl' to rx queue 1 */
-// ttl is used in statefull mode, and ip_id in stateless. We configure the driver registers so that only one of them applies.
-// So, the rule will apply if packet has either the correct ttl or IP ID, depending if we are in statfull or stateless.
-void CTRexExtendedDriverBaseVIC::add_del_rules(enum rte_filter_op op, uint8_t port_id, uint16_t type, uint8_t ttl
-                                               , uint16_t ip_id, uint8_t l4_proto, int queue) {
+void CTRexExtendedDriverBaseVIC::add_del_rules(enum rte_filter_op op, uint8_t port_id, uint16_t type
+                                               , uint16_t id, uint8_t l4_proto, uint8_t tos, int queue) {
     int ret=rte_eth_dev_filter_supported(port_id, RTE_ETH_FILTER_FDIR);
-    static int filter_soft_id = 0;
 
     if ( ret != 0 ){
         rte_exit(EXIT_FAILURE, "rte_eth_dev_filter_supported "
@@ -6781,15 +6762,15 @@ void CTRexExtendedDriverBaseVIC::add_del_rules(enum rte_filter_op op, uint8_t po
     memset(&filter,0,sizeof(struct rte_eth_fdir_filter));
 
 #if 0
-    printf("40g::%s rules: port:%d type:%d ttl:%d ip_id:%x l4:%d q:%d\n"
+    printf("VIC add_del_rules::%s rules: port:%d type:%d id:%d l4:%d tod:%d, q:%d\n"
            , (op == RTE_ETH_FILTER_ADD) ?  "add" : "del"
-           , port_id, type, ttl, ip_id, l4_proto, queue);
+           , port_id, type, id, l4_proto, tos, queue);
 #endif
 
     filter.action.rx_queue = queue;
-    filter.action.behavior =RTE_ETH_FDIR_ACCEPT;
-    filter.action.report_status =RTE_ETH_FDIR_NO_REPORT_STATUS;
-    filter.soft_id = filter_soft_id++;
+    filter.action.behavior = RTE_ETH_FDIR_ACCEPT;
+    filter.action.report_status = RTE_ETH_FDIR_NO_REPORT_STATUS;
+    filter.soft_id = id;
     filter.input.flow_type = type;
 
     switch (type) {
@@ -6797,22 +6778,22 @@ void CTRexExtendedDriverBaseVIC::add_del_rules(enum rte_filter_op op, uint8_t po
     case RTE_ETH_FLOW_NONFRAG_IPV4_TCP:
     case RTE_ETH_FLOW_NONFRAG_IPV4_SCTP:
     case RTE_ETH_FLOW_NONFRAG_IPV4_OTHER:
-        filter.input.flow.ip4_flow.tos=ttl;
-        filter.input.flow.ip4_flow.ip_id = ip_id;
-        if (l4_proto != 0)
-            filter.input.flow.ip4_flow.proto = l4_proto;
+        filter.input.flow.ip4_flow.tos = tos;
+        filter.input.flow.ip4_flow.proto = l4_proto;
         break;
     case RTE_ETH_FLOW_NONFRAG_IPV6_UDP:
     case RTE_ETH_FLOW_NONFRAG_IPV6_TCP:
     case RTE_ETH_FLOW_NONFRAG_IPV6_OTHER:
-        filter.input.flow.ipv6_flow.tc=ttl;
-        filter.input.flow.ipv6_flow.flow_label = ip_id;
+        filter.input.flow.ipv6_flow.tc = tos;
         filter.input.flow.ipv6_flow.proto = l4_proto;
         break;
     }
 
     ret = rte_eth_dev_filter_ctrl(port_id, RTE_ETH_FILTER_FDIR, op, (void*)&filter);
     if ( ret != 0 ) {
+        if (((op == RTE_ETH_FILTER_ADD) && (ret == -EEXIST)) || ((op == RTE_ETH_FILTER_DELETE) && (ret == -ENOENT)))
+            return;
+
         rte_exit(EXIT_FAILURE, "rte_eth_dev_filter_ctrl: err=%d, port=%u\n",
                  ret, port_id);
     }
@@ -6831,38 +6812,46 @@ int CTRexExtendedDriverBaseVIC::add_del_eth_type_rule(uint8_t port_id, enum rte_
     return ret;
 }
 
-extern "C" int rte_eth_fdir_stats_reset(uint8_t port_id, uint32_t *stats, uint32_t start, uint32_t len);
-
 int CTRexExtendedDriverBaseVIC::configure_rx_filter_rules_statefull(CPhyEthIF * _if) {
     uint32_t port_id = _if->get_port_id();
-#ifndef VIC_TTL_FILTER
-    add_del_rules(RTE_ETH_FILTER_ADD, port_id, RTE_ETH_FLOW_NONFRAG_IPV4_UDP, 0x1,  0, 17, MAIN_DPDK_RX_Q); /*TCP/UDP */
-    add_del_rules(RTE_ETH_FILTER_ADD, port_id, RTE_ETH_FLOW_NONFRAG_IPV4_TCP, 0x1,  0, 6, MAIN_DPDK_RX_Q);
-    add_del_rules(RTE_ETH_FILTER_ADD, port_id, RTE_ETH_FLOW_NONFRAG_IPV4_SCTP, 0x1,  0, 132, MAIN_DPDK_RX_Q); /*SCTP*/
-    add_del_rules(RTE_ETH_FILTER_ADD, port_id, RTE_ETH_FLOW_NONFRAG_IPV4_OTHER, 0x1,  0, 1, MAIN_DPDK_RX_Q);  /*ICMP*/
-
-    /* Ipv6*/
-    add_del_rules(RTE_ETH_FILTER_ADD, port_id, RTE_ETH_FLOW_NONFRAG_IPV6_OTHER, 0x1,  0, 0, MAIN_DPDK_RX_Q); /*Any protocol on Ipv6*/
-#else
-    uint16_t hops = get_rx_check_hops();
-    int i;
-    for (i = 0; i < 2; i++) {
-        uint8_t ttl = TTL_RESERVE_DUPLICATE - i - hops;
-        add_del_rules(RTE_ETH_FILTER_ADD, port_id, RTE_ETH_FLOW_NONFRAG_IPV4_UDP, ttl, 0, 17, MAIN_DPDK_RX_Q);
-        add_del_rules(RTE_ETH_FILTER_ADD, port_id, RTE_ETH_FLOW_NONFRAG_IPV4_TCP, ttl, 0, 6, MAIN_DPDK_RX_Q);
-        add_del_rules(RTE_ETH_FILTER_ADD, port_id, RTE_ETH_FLOW_NONFRAG_IPV6_UDP, ttl, 0, RX_CHECK_V6_OPT_TYPE, MAIN_DPDK_RX_Q);
-        add_del_rules(RTE_ETH_FILTER_ADD, port_id, RTE_ETH_FLOW_NONFRAG_IPV6_TCP, ttl, 0, RX_CHECK_V6_OPT_TYPE, MAIN_DPDK_RX_Q);
-        add_del_rules(RTE_ETH_FILTER_ADD, port_id, RTE_ETH_FLOW_NONFRAG_IPV6_OTHER, ttl, 0, RX_CHECK_V6_OPT_TYPE, MAIN_DPDK_RX_Q);
-        /* Rules for latency measurement packets */
-        add_del_rules(RTE_ETH_FILTER_ADD, port_id, RTE_ETH_FLOW_NONFRAG_IPV4_OTHER, ttl, 0, IPPROTO_ICMP, MAIN_DPDK_RX_Q);
-        add_del_rules(RTE_ETH_FILTER_ADD, port_id, RTE_ETH_FLOW_NONFRAG_IPV4_SCTP, ttl, 0, 138, MAIN_DPDK_RX_Q);
-    }
-#endif
+
+    set_rcv_all(_if, false);
+
+    // Rules to direct all IP packets with tos lsb bit 1 to RX Q.
+    // IPv4
+    add_del_rules(RTE_ETH_FILTER_ADD, port_id, RTE_ETH_FLOW_NONFRAG_IPV4_UDP, 1, 17, 0x1, MAIN_DPDK_RX_Q);
+    add_del_rules(RTE_ETH_FILTER_ADD, port_id, RTE_ETH_FLOW_NONFRAG_IPV4_TCP, 1, 6,  0x1, MAIN_DPDK_RX_Q);
+    add_del_rules(RTE_ETH_FILTER_ADD, port_id, RTE_ETH_FLOW_NONFRAG_IPV4_SCTP, 1, 132,  0x1, MAIN_DPDK_RX_Q); /*SCTP*/
+    add_del_rules(RTE_ETH_FILTER_ADD, port_id, RTE_ETH_FLOW_NONFRAG_IPV4_OTHER, 1, 1,  0x1, MAIN_DPDK_RX_Q);  /*ICMP*/
+    // Ipv6
+    add_del_rules(RTE_ETH_FILTER_ADD, port_id, RTE_ETH_FLOW_NONFRAG_IPV6_OTHER, 1, 6,  0x1, MAIN_DPDK_RX_Q);
+    add_del_rules(RTE_ETH_FILTER_ADD, port_id, RTE_ETH_FLOW_NONFRAG_IPV6_UDP, 1, 17,  0x1, MAIN_DPDK_RX_Q);
+
+    // Because of some issue with VIC firmware, IPv6 UDP and ICMP go by default to q 1, so we
+    // need these rules to make them go to q 0.
+    // rule appply to all packets with 0 on tos lsb.
+    add_del_rules(RTE_ETH_FILTER_ADD, port_id, RTE_ETH_FLOW_NONFRAG_IPV6_OTHER, 1, 6,  0, MAIN_DPDK_DATA_Q);
+    add_del_rules(RTE_ETH_FILTER_ADD, port_id, RTE_ETH_FLOW_NONFRAG_IPV6_UDP, 1, 17,  0, MAIN_DPDK_DATA_Q);
 
     return 0;
 }
 
 
+int CTRexExtendedDriverBaseVIC::set_rcv_all(CPhyEthIF * _if, bool set_on) {
+    uint8_t port_id = _if->get_rte_port_id();
+
+    // soft ID 100 tells VIC driver to add rule for all ether types.
+    // Added with highest priority (implicitly in the driver), so if it exists, it applies before all other rules
+    if (set_on) {
+        add_del_rules(RTE_ETH_FILTER_ADD, port_id, RTE_ETH_FLOW_NONFRAG_IPV4_UDP, 100, 30, 0, MAIN_DPDK_RX_Q);
+    } else {
+        add_del_rules(RTE_ETH_FILTER_DELETE, port_id, RTE_ETH_FLOW_NONFRAG_IPV4_UDP, 100, 30, 0, MAIN_DPDK_RX_Q);
+    }
+
+    return 0;
+
+}
+
 void CTRexExtendedDriverBaseVIC::clear_extended_stats(CPhyEthIF * _if){
     rte_eth_stats_reset(_if->get_port_id());
 }
@@ -6895,7 +6884,6 @@ void CTRexExtendedDriverBaseVIC::get_extended_stats(CPhyEthIF * _if,CPhyEthIFSta
     prev_stats->rx_nombuf = stats1.rx_nombuf;
 }
 
-
 int CTRexExtendedDriverBaseVIC::verify_fw_ver(int port_id) {
 
     struct rte_eth_fdir_info fdir_info;
@@ -6916,7 +6904,6 @@ int CTRexExtendedDriverBaseVIC::verify_fw_ver(int port_id) {
     exit(1);
 }
 
-
 int CTRexExtendedDriverBaseVIC::configure_rx_filter_rules(CPhyEthIF * _if) {
 
     if (get_is_stateless()) {
@@ -6944,18 +6931,12 @@ int CTRexExtendedDriverBaseVIC::dump_fdir_global_stats(CPhyEthIF * _if, FILE *fd
  return (0);
 }
 
-
 CFlowStatParser *CTRexExtendedDriverBaseVIC::get_flow_stat_parser() {
     CFlowStatParser *parser = new CFlowStatParser();
     assert (parser);
     return parser;
 }
 
-int CTRexExtendedDriverBaseVIC::set_rcv_all(CPhyEthIF * _if, bool set_on) {
-    //printf(" NOT supported yet \n");
-    return 0;
-}
-
 
 /////////////////////////////////////////////////////////////////////////////////////