net/mlx5: probe all port representors
authorAdrien Mazarguil <adrien.mazarguil@6wind.com>
Tue, 10 Jul 2018 16:04:54 +0000 (18:04 +0200)
committerShahaf Shuler <shahafs@mellanox.com>
Wed, 11 Jul 2018 13:37:19 +0000 (15:37 +0200)
Probe existing port representors in addition to their master device and
associate them automatically.

To avoid collision between Ethernet devices, they are named as follows:

- "{DBDF}" for master/switch devices.
- "{DBDF}_representor_{rep}" with "rep" starting from 0 for port
  representors.

(Patch based on prior work from Yuanhan Liu)

Signed-off-by: Adrien Mazarguil <adrien.mazarguil@6wind.com>
Signed-off-by: Nelio Laranjeiro <nelio.laranjeiro@6wind.com>
Reviewed-by: Xueming Li <xuemingl@mellanox.com>
drivers/net/mlx5/mlx5.c
drivers/net/mlx5/mlx5.h
drivers/net/mlx5/mlx5_ethdev.c
drivers/net/mlx5/mlx5_mac.c
drivers/net/mlx5/mlx5_stats.c

index 958d60a..b8d31f3 100644 (file)
@@ -307,7 +307,27 @@ mlx5_dev_close(struct rte_eth_dev *dev)
        if (ret)
                DRV_LOG(WARNING, "port %u some flows still remain",
                        dev->data->port_id);
+       if (priv->domain_id != RTE_ETH_DEV_SWITCH_DOMAIN_ID_INVALID) {
+               unsigned int c = 0;
+               unsigned int i = mlx5_dev_to_port_id(dev->device, NULL, 0);
+               uint16_t port_id[i];
+
+               i = RTE_MIN(mlx5_dev_to_port_id(dev->device, port_id, i), i);
+               while (i--) {
+                       struct priv *opriv =
+                               rte_eth_devices[port_id[i]].data->dev_private;
+
+                       if (!opriv ||
+                           opriv->domain_id != priv->domain_id ||
+                           &rte_eth_devices[port_id[i]] == dev)
+                               continue;
+                       ++c;
+               }
+               if (!c)
+                       claim_zero(rte_eth_switch_domain_free(priv->domain_id));
+       }
        memset(priv, 0, sizeof(*priv));
+       priv->domain_id = RTE_ETH_DEV_SWITCH_DOMAIN_ID_INVALID;
 }
 
 const struct eth_dev_ops mlx5_dev_ops = {
@@ -647,6 +667,8 @@ mlx5_uar_init_secondary(struct rte_eth_dev *dev)
  *   Verbs device.
  * @param vf
  *   If nonzero, enable VF-specific features.
+ * @param[in] switch_info
+ *   Switch properties of Ethernet device.
  *
  * @return
  *   A valid Ethernet device object on success, NULL otherwise and rte_errno
@@ -655,7 +677,8 @@ mlx5_uar_init_secondary(struct rte_eth_dev *dev)
 static struct rte_eth_dev *
 mlx5_dev_spawn(struct rte_device *dpdk_dev,
               struct ibv_device *ibv_dev,
-              int vf)
+              int vf,
+              const struct mlx5_switch_info *switch_info)
 {
        struct ibv_context *ctx;
        struct ibv_device_attr_ex attr;
@@ -697,6 +720,8 @@ mlx5_dev_spawn(struct rte_device *dpdk_dev,
 #endif
        struct ether_addr mac;
        char name[RTE_ETH_NAME_MAX_LEN];
+       int own_domain_id = 0;
+       unsigned int i;
 
        /* Prepare shared data between primary and secondary process. */
        mlx5_prepare_shared_data();
@@ -805,7 +830,12 @@ mlx5_dev_spawn(struct rte_device *dpdk_dev,
                DEBUG("ibv_query_device_ex() failed");
                goto error;
        }
-       rte_strlcpy(name, dpdk_dev->name, sizeof(name));
+       if (!switch_info->representor)
+               rte_strlcpy(name, dpdk_dev->name, sizeof(name));
+       else
+               snprintf(name, sizeof(name), "%s_representor_%u",
+                        dpdk_dev->name, switch_info->port_name);
+       DRV_LOG(DEBUG, "naming Ethernet device \"%s\"", name);
        if (rte_eal_process_type() == RTE_PROC_SECONDARY) {
                eth_dev = rte_eth_dev_attach_secondary(name);
                if (eth_dev == NULL) {
@@ -874,6 +904,8 @@ mlx5_dev_spawn(struct rte_device *dpdk_dev,
                goto error;
        }
        priv->ctx = ctx;
+       strncpy(priv->ibdev_name, priv->ctx->device->name,
+               sizeof(priv->ibdev_name));
        strncpy(priv->ibdev_path, priv->ctx->device->ibdev_path,
                sizeof(priv->ibdev_path));
        priv->device_attr = attr;
@@ -883,6 +915,41 @@ mlx5_dev_spawn(struct rte_device *dpdk_dev,
        priv->nl_socket_rdma = mlx5_nl_init(0, NETLINK_RDMA);
        priv->nl_socket_route = mlx5_nl_init(RTMGRP_LINK, NETLINK_ROUTE);
        priv->nl_sn = 0;
+       priv->representor = !!switch_info->representor;
+       priv->domain_id = RTE_ETH_DEV_SWITCH_DOMAIN_ID_INVALID;
+       priv->representor_id =
+               switch_info->representor ? switch_info->port_name : -1;
+       /*
+        * Look for sibling devices in order to reuse their switch domain
+        * if any, otherwise allocate one.
+        */
+       i = mlx5_dev_to_port_id(dpdk_dev, NULL, 0);
+       if (i > 0) {
+               uint16_t port_id[i];
+
+               i = RTE_MIN(mlx5_dev_to_port_id(dpdk_dev, port_id, i), i);
+               while (i--) {
+                       const struct priv *opriv =
+                               rte_eth_devices[port_id[i]].data->dev_private;
+
+                       if (!opriv ||
+                           opriv->domain_id ==
+                           RTE_ETH_DEV_SWITCH_DOMAIN_ID_INVALID)
+                               continue;
+                       priv->domain_id = opriv->domain_id;
+                       break;
+               }
+       }
+       if (priv->domain_id == RTE_ETH_DEV_SWITCH_DOMAIN_ID_INVALID) {
+               err = rte_eth_switch_domain_alloc(&priv->domain_id);
+               if (err) {
+                       err = rte_errno;
+                       DRV_LOG(ERR, "unable to allocate switch domain: %s",
+                               strerror(rte_errno));
+                       goto error;
+               }
+               own_domain_id = 1;
+       }
        err = mlx5_args(&config, dpdk_dev->devargs);
        if (err) {
                err = rte_errno;
@@ -966,6 +1033,8 @@ mlx5_dev_spawn(struct rte_device *dpdk_dev,
                err = ENOMEM;
                goto error;
        }
+       if (priv->representor)
+               eth_dev->data->dev_flags |= RTE_ETH_DEV_REPRESENTOR;
        eth_dev->data->dev_private = priv;
        priv->dev_data = eth_dev->data;
        eth_dev->data->mac_addrs = priv->mac;
@@ -1084,6 +1153,8 @@ error:
                        close(priv->nl_socket_route);
                if (priv->nl_socket_rdma >= 0)
                        close(priv->nl_socket_rdma);
+               if (own_domain_id)
+                       claim_zero(rte_eth_switch_domain_free(priv->domain_id));
                rte_free(priv);
        }
        if (pd)
@@ -1100,7 +1171,7 @@ error:
 /**
  * DPDK callback to register a PCI device.
  *
- * This function spawns an Ethernet device out of a given PCI device.
+ * This function spawns Ethernet devices out of a given PCI device.
  *
  * @param[in] pci_drv
  *   PCI driver structure (mlx5_driver).
@@ -1115,7 +1186,6 @@ mlx5_pci_probe(struct rte_pci_driver *pci_drv __rte_unused,
               struct rte_pci_device *pci_dev)
 {
        struct ibv_device **ibv_list;
-       struct rte_eth_dev *eth_dev = NULL;
        unsigned int n = 0;
        int vf;
        int ret;
@@ -1150,9 +1220,11 @@ mlx5_pci_probe(struct rte_pci_driver *pci_drv __rte_unused,
 
        unsigned int ifindex[n];
        struct mlx5_switch_info info[n];
+       struct rte_eth_dev *eth_list[n];
        int nl_route = n ? mlx5_nl_init(0, NETLINK_ROUTE) : -1;
        int nl_rdma = n ? mlx5_nl_init(0, NETLINK_RDMA) : -1;
        unsigned int i;
+       unsigned int u;
 
        /*
         * The existence of several matching entries (n > 1) means port
@@ -1187,28 +1259,14 @@ mlx5_pci_probe(struct rte_pci_driver *pci_drv __rte_unused,
                close(nl_rdma);
        if (nl_route >= 0)
                close(nl_route);
-       /* Look for master device. */
-       for (i = 0; i != n; ++i) {
-               if (!info[i].master)
-                       continue;
-               /* Make it the first entry. */
-               if (i == 0)
-                       break;
-               ibv_match[n] = ibv_match[0];
-               ibv_match[0] = ibv_match[i];
-               ibv_match[n] = NULL;
-               break;
-       }
-       if (n && i == n) {
-               if (n == 1 && !info[0].representor) {
+       /* Count unidentified devices. */
+       for (u = 0, i = 0; i != n; ++i)
+               if (!info[i].master && !info[i].representor)
+                       ++u;
+       if (u) {
+               if (n == 1 && u == 1) {
                        /* Case #2. */
                        DRV_LOG(INFO, "no switch support detected");
-               } else if (n == 1) {
-                       /* Case #3. */
-                       DRV_LOG(ERR,
-                               "device looks like a port representor, this is"
-                               " not supported yet");
-                       n = 0;
                } else {
                        /* Case #3. */
                        DRV_LOG(ERR,
@@ -1227,8 +1285,19 @@ mlx5_pci_probe(struct rte_pci_driver *pci_drv __rte_unused,
        default:
                vf = 0;
        }
-       if (n)
-               eth_dev = mlx5_dev_spawn(&pci_dev->device, ibv_match[0], vf);
+       for (i = 0; i != n; ++i) {
+               uint32_t restore;
+
+               eth_list[i] = mlx5_dev_spawn(&pci_dev->device, ibv_match[i],
+                                            vf, &info[i]);
+               if (!eth_list[i])
+                       break;
+               restore = eth_list[i]->data->dev_flags;
+               rte_eth_copy_pci_info(eth_list[i], pci_dev);
+               /* Restore non-PCI flags cleared by the above call. */
+               eth_list[i]->data->dev_flags |= restore;
+               rte_eth_dev_probing_finish(eth_list[i]);
+       }
        mlx5_glue->free_device_list(ibv_list);
        if (!n) {
                DRV_LOG(WARNING,
@@ -1238,7 +1307,7 @@ mlx5_pci_probe(struct rte_pci_driver *pci_drv __rte_unused,
                        pci_dev->addr.devid, pci_dev->addr.function);
                rte_errno = ENOENT;
                ret = -rte_errno;
-       } else if (!eth_dev) {
+       } else if (i != n) {
                DRV_LOG(ERR,
                        "probe of PCI device " PCI_PRI_FMT " aborted after"
                        " encountering an error: %s",
@@ -1246,9 +1315,16 @@ mlx5_pci_probe(struct rte_pci_driver *pci_drv __rte_unused,
                        pci_dev->addr.devid, pci_dev->addr.function,
                        strerror(rte_errno));
                ret = -rte_errno;
+               /* Roll back. */
+               while (i--) {
+                       mlx5_dev_close(eth_list[i]);
+                       if (rte_eal_process_type() == RTE_PROC_PRIMARY)
+                               rte_free(eth_list[i]->data->dev_private);
+                       claim_zero(rte_eth_dev_release_port(eth_list[i]));
+               }
+               /* Restore original error. */
+               rte_errno = -ret;
        } else {
-               rte_eth_copy_pci_info(eth_dev, pci_dev);
-               rte_eth_dev_probing_finish(eth_dev);
                ret = 0;
        }
        return ret;
index 7040462..cc01310 100644 (file)
@@ -159,6 +159,7 @@ struct priv {
        struct ibv_context *ctx; /* Verbs context. */
        struct ibv_device_attr_ex device_attr; /* Device properties. */
        struct ibv_pd *pd; /* Protection Domain. */
+       char ibdev_name[IBV_SYSFS_NAME_MAX]; /* IB device name. */
        char ibdev_path[IBV_SYSFS_PATH_MAX]; /* IB device path for secondary */
        struct ether_addr mac[MLX5_MAX_MAC_ADDRESSES]; /* MAC addresses. */
        BITFIELD_DECLARE(mac_own, uint64_t, MLX5_MAX_MAC_ADDRESSES);
@@ -168,6 +169,9 @@ struct priv {
        /* Device properties. */
        uint16_t mtu; /* Configured MTU. */
        unsigned int isolated:1; /* Whether isolated mode is enabled. */
+       unsigned int representor:1; /* Device is a port representor. */
+       uint16_t domain_id; /* Switch domain identifier. */
+       int32_t representor_id; /* Port representor identifier. */
        /* RX/TX queues. */
        unsigned int rxqs_n; /* RX queues array size. */
        unsigned int txqs_n; /* TX queues array size. */
@@ -217,9 +221,12 @@ int mlx5_getenv_int(const char *);
 
 /* mlx5_ethdev.c */
 
+int mlx5_get_master_ifname(const struct rte_eth_dev *dev,
+                          char (*ifname)[IF_NAMESIZE]);
 int mlx5_get_ifname(const struct rte_eth_dev *dev, char (*ifname)[IF_NAMESIZE]);
 int mlx5_ifindex(const struct rte_eth_dev *dev);
-int mlx5_ifreq(const struct rte_eth_dev *dev, int req, struct ifreq *ifr);
+int mlx5_ifreq(const struct rte_eth_dev *dev, int req, struct ifreq *ifr,
+              int master);
 int mlx5_get_mtu(struct rte_eth_dev *dev, uint16_t *mtu);
 int mlx5_set_flags(struct rte_eth_dev *dev, unsigned int keep,
                   unsigned int flags);
@@ -244,6 +251,9 @@ int mlx5_set_link_up(struct rte_eth_dev *dev);
 int mlx5_is_removed(struct rte_eth_dev *dev);
 eth_tx_burst_t mlx5_select_tx_function(struct rte_eth_dev *dev);
 eth_rx_burst_t mlx5_select_rx_function(struct rte_eth_dev *dev);
+unsigned int mlx5_dev_to_port_id(const struct rte_device *dev,
+                                uint16_t *port_list,
+                                unsigned int port_list_n);
 
 /* mlx5_mac.c */
 
index 819f5ba..ea01b39 100644 (file)
@@ -93,7 +93,7 @@ struct ethtool_link_settings {
 #endif
 
 /**
- * Get interface name from private structure.
+ * Get master interface name from private structure.
  *
  * @param[in] dev
  *   Pointer to Ethernet device.
@@ -104,7 +104,8 @@ struct ethtool_link_settings {
  *   0 on success, a negative errno value otherwise and rte_errno is set.
  */
 int
-mlx5_get_ifname(const struct rte_eth_dev *dev, char (*ifname)[IF_NAMESIZE])
+mlx5_get_master_ifname(const struct rte_eth_dev *dev,
+                      char (*ifname)[IF_NAMESIZE])
 {
        struct priv *priv = dev->data->dev_private;
        DIR *dir;
@@ -178,6 +179,39 @@ try_dev_id:
        return 0;
 }
 
+/**
+ * Get interface name from private structure.
+ *
+ * This is a port representor-aware version of mlx5_get_master_ifname().
+ *
+ * @param[in] dev
+ *   Pointer to Ethernet device.
+ * @param[out] ifname
+ *   Interface name output buffer.
+ *
+ * @return
+ *   0 on success, a negative errno value otherwise and rte_errno is set.
+ */
+int
+mlx5_get_ifname(const struct rte_eth_dev *dev, char (*ifname)[IF_NAMESIZE])
+{
+       struct priv *priv = dev->data->dev_private;
+       unsigned int ifindex =
+               priv->nl_socket_rdma >= 0 ?
+               mlx5_nl_ifindex(priv->nl_socket_rdma, priv->ibdev_name) : 0;
+
+       if (!ifindex) {
+               if (!priv->representor)
+                       return mlx5_get_master_ifname(dev, ifname);
+               rte_errno = ENXIO;
+               return -rte_errno;
+       }
+       if (if_indextoname(ifindex, &(*ifname)[0]))
+               return 0;
+       rte_errno = errno;
+       return -rte_errno;
+}
+
 /**
  * Get the interface index from device name.
  *
@@ -214,12 +248,16 @@ mlx5_ifindex(const struct rte_eth_dev *dev)
  *   Request number to pass to ioctl().
  * @param[out] ifr
  *   Interface request structure output buffer.
+ * @param master
+ *   When device is a port representor, perform request on master device
+ *   instead.
  *
  * @return
  *   0 on success, a negative errno value otherwise and rte_errno is set.
  */
 int
-mlx5_ifreq(const struct rte_eth_dev *dev, int req, struct ifreq *ifr)
+mlx5_ifreq(const struct rte_eth_dev *dev, int req, struct ifreq *ifr,
+          int master)
 {
        int sock = socket(PF_INET, SOCK_DGRAM, IPPROTO_IP);
        int ret = 0;
@@ -228,7 +266,10 @@ mlx5_ifreq(const struct rte_eth_dev *dev, int req, struct ifreq *ifr)
                rte_errno = errno;
                return -rte_errno;
        }
-       ret = mlx5_get_ifname(dev, &ifr->ifr_name);
+       if (master)
+               ret = mlx5_get_master_ifname(dev, &ifr->ifr_name);
+       else
+               ret = mlx5_get_ifname(dev, &ifr->ifr_name);
        if (ret)
                goto error;
        ret = ioctl(sock, req, ifr);
@@ -258,7 +299,7 @@ int
 mlx5_get_mtu(struct rte_eth_dev *dev, uint16_t *mtu)
 {
        struct ifreq request;
-       int ret = mlx5_ifreq(dev, SIOCGIFMTU, &request);
+       int ret = mlx5_ifreq(dev, SIOCGIFMTU, &request, 0);
 
        if (ret)
                return ret;
@@ -282,7 +323,7 @@ mlx5_set_mtu(struct rte_eth_dev *dev, uint16_t mtu)
 {
        struct ifreq request = { .ifr_mtu = mtu, };
 
-       return mlx5_ifreq(dev, SIOCSIFMTU, &request);
+       return mlx5_ifreq(dev, SIOCSIFMTU, &request, 0);
 }
 
 /**
@@ -302,13 +343,13 @@ int
 mlx5_set_flags(struct rte_eth_dev *dev, unsigned int keep, unsigned int flags)
 {
        struct ifreq request;
-       int ret = mlx5_ifreq(dev, SIOCGIFFLAGS, &request);
+       int ret = mlx5_ifreq(dev, SIOCGIFFLAGS, &request, 0);
 
        if (ret)
                return ret;
        request.ifr_flags &= keep;
        request.ifr_flags |= flags & ~keep;
-       return mlx5_ifreq(dev, SIOCSIFFLAGS, &request);
+       return mlx5_ifreq(dev, SIOCSIFFLAGS, &request, 0);
 }
 
 /**
@@ -477,6 +518,30 @@ mlx5_dev_infos_get(struct rte_eth_dev *dev, struct rte_eth_dev_info *info)
        info->speed_capa = priv->link_speed_capa;
        info->flow_type_rss_offloads = ~MLX5_RSS_HF_MASK;
        mlx5_set_default_params(dev, info);
+       info->switch_info.name = dev->data->name;
+       info->switch_info.domain_id = priv->domain_id;
+       info->switch_info.port_id = priv->representor_id;
+       if (priv->representor) {
+               unsigned int i = mlx5_dev_to_port_id(dev->device, NULL, 0);
+               uint16_t port_id[i];
+
+               i = RTE_MIN(mlx5_dev_to_port_id(dev->device, port_id, i), i);
+               while (i--) {
+                       struct priv *opriv =
+                               rte_eth_devices[port_id[i]].data->dev_private;
+
+                       if (!opriv ||
+                           opriv->representor ||
+                           opriv->domain_id != priv->domain_id)
+                               continue;
+                       /*
+                        * Override switch name with that of the master
+                        * device.
+                        */
+                       info->switch_info.name = opriv->dev_data->name;
+                       break;
+               }
+       }
 }
 
 /**
@@ -540,7 +605,7 @@ mlx5_link_update_unlocked_gset(struct rte_eth_dev *dev,
        int link_speed = 0;
        int ret;
 
-       ret = mlx5_ifreq(dev, SIOCGIFFLAGS, &ifr);
+       ret = mlx5_ifreq(dev, SIOCGIFFLAGS, &ifr, 1);
        if (ret) {
                DRV_LOG(WARNING, "port %u ioctl(SIOCGIFFLAGS) failed: %s",
                        dev->data->port_id, strerror(rte_errno));
@@ -550,7 +615,7 @@ mlx5_link_update_unlocked_gset(struct rte_eth_dev *dev,
        dev_link.link_status = ((ifr.ifr_flags & IFF_UP) &&
                                (ifr.ifr_flags & IFF_RUNNING));
        ifr.ifr_data = (void *)&edata;
-       ret = mlx5_ifreq(dev, SIOCETHTOOL, &ifr);
+       ret = mlx5_ifreq(dev, SIOCETHTOOL, &ifr, 1);
        if (ret) {
                DRV_LOG(WARNING,
                        "port %u ioctl(SIOCETHTOOL, ETHTOOL_GSET) failed: %s",
@@ -611,7 +676,7 @@ mlx5_link_update_unlocked_gs(struct rte_eth_dev *dev,
        uint64_t sc;
        int ret;
 
-       ret = mlx5_ifreq(dev, SIOCGIFFLAGS, &ifr);
+       ret = mlx5_ifreq(dev, SIOCGIFFLAGS, &ifr, 1);
        if (ret) {
                DRV_LOG(WARNING, "port %u ioctl(SIOCGIFFLAGS) failed: %s",
                        dev->data->port_id, strerror(rte_errno));
@@ -621,7 +686,7 @@ mlx5_link_update_unlocked_gs(struct rte_eth_dev *dev,
        dev_link.link_status = ((ifr.ifr_flags & IFF_UP) &&
                                (ifr.ifr_flags & IFF_RUNNING));
        ifr.ifr_data = (void *)&gcmd;
-       ret = mlx5_ifreq(dev, SIOCETHTOOL, &ifr);
+       ret = mlx5_ifreq(dev, SIOCETHTOOL, &ifr, 1);
        if (ret) {
                DRV_LOG(DEBUG,
                        "port %u ioctl(SIOCETHTOOL, ETHTOOL_GLINKSETTINGS)"
@@ -638,7 +703,7 @@ mlx5_link_update_unlocked_gs(struct rte_eth_dev *dev,
 
        *ecmd = gcmd;
        ifr.ifr_data = (void *)ecmd;
-       ret = mlx5_ifreq(dev, SIOCETHTOOL, &ifr);
+       ret = mlx5_ifreq(dev, SIOCETHTOOL, &ifr, 1);
        if (ret) {
                DRV_LOG(DEBUG,
                        "port %u ioctl(SIOCETHTOOL, ETHTOOL_GLINKSETTINGS)"
@@ -801,7 +866,7 @@ mlx5_dev_get_flow_ctrl(struct rte_eth_dev *dev, struct rte_eth_fc_conf *fc_conf)
        int ret;
 
        ifr.ifr_data = (void *)&ethpause;
-       ret = mlx5_ifreq(dev, SIOCETHTOOL, &ifr);
+       ret = mlx5_ifreq(dev, SIOCETHTOOL, &ifr, 1);
        if (ret) {
                DRV_LOG(WARNING,
                        "port %u ioctl(SIOCETHTOOL, ETHTOOL_GPAUSEPARAM) failed:"
@@ -854,7 +919,7 @@ mlx5_dev_set_flow_ctrl(struct rte_eth_dev *dev, struct rte_eth_fc_conf *fc_conf)
                ethpause.tx_pause = 1;
        else
                ethpause.tx_pause = 0;
-       ret = mlx5_ifreq(dev, SIOCETHTOOL, &ifr);
+       ret = mlx5_ifreq(dev, SIOCETHTOOL, &ifr, 0);
        if (ret) {
                DRV_LOG(WARNING,
                        "port %u ioctl(SIOCETHTOOL, ETHTOOL_SPAUSEPARAM)"
@@ -1193,3 +1258,40 @@ mlx5_is_removed(struct rte_eth_dev *dev)
                return 1;
        return 0;
 }
+
+/**
+ * Get port ID list of mlx5 instances sharing a common device.
+ *
+ * @param[in] dev
+ *   Device to look for.
+ * @param[out] port_list
+ *   Result buffer for collected port IDs.
+ * @param port_list_n
+ *   Maximum number of entries in result buffer. If 0, @p port_list can be
+ *   NULL.
+ *
+ * @return
+ *   Number of matching instances regardless of the @p port_list_n
+ *   parameter, 0 if none were found.
+ */
+unsigned int
+mlx5_dev_to_port_id(const struct rte_device *dev, uint16_t *port_list,
+                   unsigned int port_list_n)
+{
+       uint16_t id;
+       unsigned int n = 0;
+
+       RTE_ETH_FOREACH_DEV(id) {
+               struct rte_eth_dev *ldev = &rte_eth_devices[id];
+
+               if (!ldev->device ||
+                   !ldev->device->driver ||
+                   strcmp(ldev->device->driver->name, MLX5_DRIVER_NAME) ||
+                   ldev->device != dev)
+                       continue;
+               if (n < port_list_n)
+                       port_list[n] = id;
+               n++;
+       }
+       return n;
+}
index 672a476..12ee37f 100644 (file)
@@ -49,7 +49,7 @@ mlx5_get_mac(struct rte_eth_dev *dev, uint8_t (*mac)[ETHER_ADDR_LEN])
        struct ifreq request;
        int ret;
 
-       ret = mlx5_ifreq(dev, SIOCGIFHWADDR, &request);
+       ret = mlx5_ifreq(dev, SIOCGIFHWADDR, &request, 0);
        if (ret)
                return ret;
        memcpy(mac, request.ifr_hwaddr.sa_data, ETHER_ADDR_LEN);
index 875dd10..91f3d47 100644 (file)
@@ -146,7 +146,7 @@ mlx5_read_dev_counters(struct rte_eth_dev *dev, uint64_t *stats)
        et_stats->cmd = ETHTOOL_GSTATS;
        et_stats->n_stats = xstats_ctrl->stats_n;
        ifr.ifr_data = (caddr_t)et_stats;
-       ret = mlx5_ifreq(dev, SIOCETHTOOL, &ifr);
+       ret = mlx5_ifreq(dev, SIOCETHTOOL, &ifr, 1);
        if (ret) {
                DRV_LOG(WARNING,
                        "port %u unable to read statistic values from device",
@@ -194,7 +194,7 @@ mlx5_ethtool_get_stats_n(struct rte_eth_dev *dev) {
 
        drvinfo.cmd = ETHTOOL_GDRVINFO;
        ifr.ifr_data = (caddr_t)&drvinfo;
-       ret = mlx5_ifreq(dev, SIOCETHTOOL, &ifr);
+       ret = mlx5_ifreq(dev, SIOCETHTOOL, &ifr, 1);
        if (ret) {
                DRV_LOG(WARNING, "port %u unable to query number of statistics",
                        dev->data->port_id);
@@ -244,7 +244,7 @@ mlx5_xstats_init(struct rte_eth_dev *dev)
        strings->string_set = ETH_SS_STATS;
        strings->len = dev_stats_n;
        ifr.ifr_data = (caddr_t)strings;
-       ret = mlx5_ifreq(dev, SIOCETHTOOL, &ifr);
+       ret = mlx5_ifreq(dev, SIOCETHTOOL, &ifr, 1);
        if (ret) {
                DRV_LOG(WARNING, "port %u unable to get statistic names",
                        dev->data->port_id);