drivers/infiniband/ulp/opa_vnic/opa_vnic_netdev.c

   1 /*
   2  * Copyright(c) 2017 Intel Corporation.
   3  *
   4  * This file is provided under a dual BSD/GPLv2 license.  When using or
   5  * redistributing this file, you may do so under either license.
   6  *
   7  * GPL LICENSE SUMMARY
   8  *
   9  * This program is free software; you can redistribute it and/or modify
  10  * it under the terms of version 2 of the GNU General Public License as
  11  * published by the Free Software Foundation.
  12  *
  13  * This program is distributed in the hope that it will be useful, but
  14  * WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  16  * General Public License for more details.
  17  *
  18  * BSD LICENSE
  19  *
  20  * Redistribution and use in source and binary forms, with or without
  21  * modification, are permitted provided that the following conditions
  22  * are met:
  23  *
  24  *  - Redistributions of source code must retain the above copyright
  25  *    notice, this list of conditions and the following disclaimer.
  26  *  - Redistributions in binary form must reproduce the above copyright
  27  *    notice, this list of conditions and the following disclaimer in
  28  *    the documentation and/or other materials provided with the
  29  *    distribution.
  30  *  - Neither the name of Intel Corporation nor the names of its
  31  *    contributors may be used to endorse or promote products derived
  32  *    from this software without specific prior written permission.
  33  *
  34  * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
  35  * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
  36  * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
  37  * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
  38  * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
  39  * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
  40  * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
  41  * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
  42  * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
  43  * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
  44  * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
  45  *
  46  */
  47
  48 /*
  49  * This file contains OPA Virtual Network Interface Controller (VNIC) driver
  50  * netdev functionality.
  51  */
  52
  53 #include <linux/module.h>
  54 #include <linux/if_vlan.h>
  55 #include <linux/crc32.h>
  56
  57 #include "opa_vnic_internal.h"
  58
  59 #define OPA_TX_TIMEOUT_MS 1000
  60
  61 #define OPA_VNIC_SKB_HEADROOM  \
  62                         ALIGN((OPA_VNIC_HDR_LEN + OPA_VNIC_SKB_MDATA_LEN), 8)
  63
  64 /* This function is overloaded for opa_vnic specific implementation */
  65 static void opa_vnic_get_stats64(struct net_device *netdev,
  66                                  struct rtnl_link_stats64 *stats)
  67 {
  68         struct opa_vnic_adapter *adapter = opa_vnic_priv(netdev);
  69         struct opa_vnic_stats vstats;
  70
  71         memset(&vstats, 0, sizeof(vstats));
  72         spin_lock(&adapter->stats_lock);
  73         adapter->rn_ops->ndo_get_stats64(netdev, &vstats.netstats);
  74         spin_unlock(&adapter->stats_lock);
  75         memcpy(stats, &vstats.netstats, sizeof(*stats));
  76 }
  77
  78 /* opa_netdev_start_xmit - transmit function */
  79 static netdev_tx_t opa_netdev_start_xmit(struct sk_buff *skb,
  80                                          struct net_device *netdev)
  81 {
  82         struct opa_vnic_adapter *adapter = opa_vnic_priv(netdev);
  83
  84         v_dbg("xmit: queue %d skb len %d\n", skb->queue_mapping, skb->len);
  85         /* pad to ensure mininum ethernet packet length */
  86         if (unlikely(skb->len < ETH_ZLEN)) {
  87                 if (skb_padto(skb, ETH_ZLEN))
  88                         return NETDEV_TX_OK;
  89
  90                 skb_put(skb, ETH_ZLEN - skb->len);
  91         }
  92
  93         opa_vnic_encap_skb(adapter, skb);
  94         return adapter->rn_ops->ndo_start_xmit(skb, netdev);
  95 }
  96
  97 static u16 opa_vnic_select_queue(struct net_device *netdev, struct sk_buff *skb,
  98                                  struct net_device *sb_dev)
  99 {
 100         struct opa_vnic_adapter *adapter = opa_vnic_priv(netdev);
 101         struct opa_vnic_skb_mdata *mdata;
 102         int rc;
 103
 104         /* pass entropy and vl as metadata in skb */
 105         mdata = skb_push(skb, sizeof(*mdata));
 106         mdata->entropy = opa_vnic_calc_entropy(skb);
 107         mdata->vl = opa_vnic_get_vl(adapter, skb);
 108         rc = adapter->rn_ops->ndo_select_queue(netdev, skb, sb_dev);
 109         skb_pull(skb, sizeof(*mdata));
 110         return rc;
 111 }
 112
 113 static void opa_vnic_update_state(struct opa_vnic_adapter *adapter, bool up)
 114 {
 115         struct __opa_veswport_info *info = &adapter->info;
 116
 117         mutex_lock(&adapter->lock);
 118         /* Operational state can only be DROP_ALL or FORWARDING */
 119         if ((info->vport.config_state == OPA_VNIC_STATE_FORWARDING) && up) {
 120                 info->vport.oper_state = OPA_VNIC_STATE_FORWARDING;
 121                 info->vport.eth_link_status = OPA_VNIC_ETH_LINK_UP;
 122         } else {
 123                 info->vport.oper_state = OPA_VNIC_STATE_DROP_ALL;
 124                 info->vport.eth_link_status = OPA_VNIC_ETH_LINK_DOWN;
 125         }
 126
 127         if (info->vport.config_state == OPA_VNIC_STATE_FORWARDING)
 128                 netif_dormant_off(adapter->netdev);
 129         else
 130                 netif_dormant_on(adapter->netdev);
 131         mutex_unlock(&adapter->lock);
 132 }
 133
 134 /* opa_vnic_process_vema_config - process vema configuration updates */
 135 void opa_vnic_process_vema_config(struct opa_vnic_adapter *adapter)
 136 {
 137         struct __opa_veswport_info *info = &adapter->info;
 138         struct rdma_netdev *rn = netdev_priv(adapter->netdev);
 139         u8 port_num[OPA_VESW_MAX_NUM_DEF_PORT] = { 0 };
 140         struct net_device *netdev = adapter->netdev;
 141         u8 i, port_count = 0;
 142         u16 port_mask;
 143
 144         /* If the base_mac_addr is changed, update the interface mac address */
 145         if (memcmp(info->vport.base_mac_addr, adapter->vema_mac_addr,
 146                    ARRAY_SIZE(info->vport.base_mac_addr))) {
 147                 struct sockaddr saddr;
 148
 149                 memcpy(saddr.sa_data, info->vport.base_mac_addr,
 150                        ARRAY_SIZE(info->vport.base_mac_addr));
 151                 mutex_lock(&adapter->lock);
 152                 eth_commit_mac_addr_change(netdev, &saddr);
 153                 memcpy(adapter->vema_mac_addr,
 154                        info->vport.base_mac_addr, ETH_ALEN);
 155                 mutex_unlock(&adapter->lock);
 156         }
 157
 158         rn->set_id(netdev, info->vesw.vesw_id);
 159
 160         /* Handle MTU limit change */
 161         rtnl_lock();
 162         netdev->max_mtu = max_t(unsigned int, info->vesw.eth_mtu,
 163                                 netdev->min_mtu);
 164         if (netdev->mtu > netdev->max_mtu)
 165                 dev_set_mtu(netdev, netdev->max_mtu);
 166         rtnl_unlock();
 167
 168         /* Update flow to default port redirection table */
 169         port_mask = info->vesw.def_port_mask;
 170         for (i = 0; i < OPA_VESW_MAX_NUM_DEF_PORT; i++) {
 171                 if (port_mask & 1)
 172                         port_num[port_count++] = i;
 173                 port_mask >>= 1;
 174         }
 175
 176         /*
 177          * Build the flow table. Flow table is required when destination LID
 178          * is not available. Up to OPA_VNIC_FLOW_TBL_SIZE flows supported.
 179          * Each flow need a default port number to get its dlid from the
 180          * u_ucast_dlid array.
 181          */
 182         for (i = 0; i < OPA_VNIC_FLOW_TBL_SIZE; i++)
 183                 adapter->flow_tbl[i] = port_count ? port_num[i % port_count] :
 184                                                     OPA_VNIC_INVALID_PORT;
 185
 186         /* update state */
 187         opa_vnic_update_state(adapter, !!(netdev->flags & IFF_UP));
 188 }
 189
 190 /*
 191  * Set the power on default values in adapter's vema interface structure.
 192  */
 193 static inline void opa_vnic_set_pod_values(struct opa_vnic_adapter *adapter)
 194 {
 195         adapter->info.vport.max_mac_tbl_ent = OPA_VNIC_MAC_TBL_MAX_ENTRIES;
 196         adapter->info.vport.max_smac_ent = OPA_VNIC_MAX_SMAC_LIMIT;
 197         adapter->info.vport.config_state = OPA_VNIC_STATE_DROP_ALL;
 198         adapter->info.vport.eth_link_status = OPA_VNIC_ETH_LINK_DOWN;
 199         adapter->info.vesw.eth_mtu = ETH_DATA_LEN;
 200 }
 201
 202 /* opa_vnic_set_mac_addr - change mac address */
 203 static int opa_vnic_set_mac_addr(struct net_device *netdev, void *addr)
 204 {
 205         struct opa_vnic_adapter *adapter = opa_vnic_priv(netdev);
 206         struct sockaddr *sa = addr;
 207         int rc;
 208
 209         if (!memcmp(netdev->dev_addr, sa->sa_data, ETH_ALEN))
 210                 return 0;
 211
 212         mutex_lock(&adapter->lock);
 213         rc = eth_mac_addr(netdev, addr);
 214         mutex_unlock(&adapter->lock);
 215         if (rc)
 216                 return rc;
 217
 218         adapter->info.vport.uc_macs_gen_count++;
 219         opa_vnic_vema_report_event(adapter,
 220                                    OPA_VESWPORT_TRAP_IFACE_UCAST_MAC_CHANGE);
 221         return 0;
 222 }
 223
 224 /*
 225  * opa_vnic_mac_send_event - post event on possible mac list exchange
 226  *  Send trap when digest from uc/mc mac list differs from previous run.
 227  *  Digest is evaluated similar to how cksum does.
 228  */
 229 static void opa_vnic_mac_send_event(struct net_device *netdev, u8 event)
 230 {
 231         struct opa_vnic_adapter *adapter = opa_vnic_priv(netdev);
 232         struct netdev_hw_addr *ha;
 233         struct netdev_hw_addr_list *hw_list;
 234         u32 *ref_crc;
 235         u32 l, crc = 0;
 236
 237         switch (event) {
 238         case OPA_VESWPORT_TRAP_IFACE_UCAST_MAC_CHANGE:
 239                 hw_list = &netdev->uc;
 240                 adapter->info.vport.uc_macs_gen_count++;
 241                 ref_crc = &adapter->umac_hash;
 242                 break;
 243         case OPA_VESWPORT_TRAP_IFACE_MCAST_MAC_CHANGE:
 244                 hw_list = &netdev->mc;
 245                 adapter->info.vport.mc_macs_gen_count++;
 246                 ref_crc = &adapter->mmac_hash;
 247                 break;
 248         default:
 249                 return;
 250         }
 251         netdev_hw_addr_list_for_each(ha, hw_list) {
 252                 crc = crc32_le(crc, ha->addr, ETH_ALEN);
 253         }
 254         l = netdev_hw_addr_list_count(hw_list) * ETH_ALEN;
 255         crc = ~crc32_le(crc, (void *)&l, sizeof(l));
 256
 257         if (crc != *ref_crc) {
 258                 *ref_crc = crc;
 259                 opa_vnic_vema_report_event(adapter, event);
 260         }
 261 }
 262
 263 /* opa_vnic_set_rx_mode - handle uc/mc mac list change */
 264 static void opa_vnic_set_rx_mode(struct net_device *netdev)
 265 {
 266         opa_vnic_mac_send_event(netdev,
 267                                 OPA_VESWPORT_TRAP_IFACE_UCAST_MAC_CHANGE);
 268
 269         opa_vnic_mac_send_event(netdev,
 270                                 OPA_VESWPORT_TRAP_IFACE_MCAST_MAC_CHANGE);
 271 }
 272
 273 /* opa_netdev_open - activate network interface */
 274 static int opa_netdev_open(struct net_device *netdev)
 275 {
 276         struct opa_vnic_adapter *adapter = opa_vnic_priv(netdev);
 277         int rc;
 278
 279         rc = adapter->rn_ops->ndo_open(adapter->netdev);
 280         if (rc) {
 281                 v_dbg("open failed %d\n", rc);
 282                 return rc;
 283         }
 284
 285         /* Update status and send trap */
 286         opa_vnic_update_state(adapter, true);
 287         opa_vnic_vema_report_event(adapter,
 288                                    OPA_VESWPORT_TRAP_ETH_LINK_STATUS_CHANGE);
 289         return 0;
 290 }
 291
 292 /* opa_netdev_close - disable network interface */
 293 static int opa_netdev_close(struct net_device *netdev)
 294 {
 295         struct opa_vnic_adapter *adapter = opa_vnic_priv(netdev);
 296         int rc;
 297
 298         rc = adapter->rn_ops->ndo_stop(adapter->netdev);
 299         if (rc) {
 300                 v_dbg("close failed %d\n", rc);
 301                 return rc;
 302         }
 303
 304         /* Update status and send trap */
 305         opa_vnic_update_state(adapter, false);
 306         opa_vnic_vema_report_event(adapter,
 307                                    OPA_VESWPORT_TRAP_ETH_LINK_STATUS_CHANGE);
 308         return 0;
 309 }
 310
 311 /* netdev ops */
 312 static const struct net_device_ops opa_netdev_ops = {
 313         .ndo_open = opa_netdev_open,
 314         .ndo_stop = opa_netdev_close,
 315         .ndo_start_xmit = opa_netdev_start_xmit,
 316         .ndo_get_stats64 = opa_vnic_get_stats64,
 317         .ndo_set_rx_mode = opa_vnic_set_rx_mode,
 318         .ndo_select_queue = opa_vnic_select_queue,
 319         .ndo_set_mac_address = opa_vnic_set_mac_addr,
 320 };
 321
 322 /* opa_vnic_add_netdev - create vnic netdev interface */
 323 struct opa_vnic_adapter *opa_vnic_add_netdev(struct ib_device *ibdev,
 324                                              u8 port_num, u8 vport_num)
 325 {
 326         struct opa_vnic_adapter *adapter;
 327         struct net_device *netdev;
 328         struct rdma_netdev *rn;
 329         int rc;
 330
 331         netdev = ibdev->ops.alloc_rdma_netdev(ibdev, port_num,
 332                                               RDMA_NETDEV_OPA_VNIC,
 333                                               "veth%d", NET_NAME_UNKNOWN,
 334                                               ether_setup);
 335         if (!netdev)
 336                 return ERR_PTR(-ENOMEM);
 337         else if (IS_ERR(netdev))
 338                 return ERR_CAST(netdev);
 339
 340         rn = netdev_priv(netdev);
 341         adapter = kzalloc(sizeof(*adapter), GFP_KERNEL);
 342         if (!adapter) {
 343                 rc = -ENOMEM;
 344                 goto adapter_err;
 345         }
 346
 347         rn->clnt_priv = adapter;
 348         rn->hca = ibdev;
 349         rn->port_num = port_num;
 350         adapter->netdev = netdev;
 351         adapter->ibdev = ibdev;
 352         adapter->port_num = port_num;
 353         adapter->vport_num = vport_num;
 354         adapter->rn_ops = netdev->netdev_ops;
 355
 356         netdev->netdev_ops = &opa_netdev_ops;
 357         netdev->priv_flags |= IFF_LIVE_ADDR_CHANGE;
 358         netdev->hard_header_len += OPA_VNIC_SKB_HEADROOM;
 359         mutex_init(&adapter->lock);
 360         mutex_init(&adapter->mactbl_lock);
 361         spin_lock_init(&adapter->stats_lock);
 362
 363         SET_NETDEV_DEV(netdev, ibdev->dev.parent);
 364
 365         opa_vnic_set_ethtool_ops(netdev);
 366
 367         opa_vnic_set_pod_values(adapter);
 368
 369         rc = register_netdev(netdev);
 370         if (rc)
 371                 goto netdev_err;
 372
 373         netif_carrier_off(netdev);
 374         netif_dormant_on(netdev);
 375         v_info("initialized\n");
 376
 377         return adapter;
 378 netdev_err:
 379         mutex_destroy(&adapter->lock);
 380         mutex_destroy(&adapter->mactbl_lock);
 381         kfree(adapter);
 382 adapter_err:
 383         rn->free_rdma_netdev(netdev);
 384
 385         return ERR_PTR(rc);
 386 }
 387
 388 /* opa_vnic_rem_netdev - remove vnic netdev interface */
 389 void opa_vnic_rem_netdev(struct opa_vnic_adapter *adapter)
 390 {
 391         struct net_device *netdev = adapter->netdev;
 392         struct rdma_netdev *rn = netdev_priv(netdev);
 393
 394         v_info("removing\n");
 395         unregister_netdev(netdev);
 396         opa_vnic_release_mac_tbl(adapter);
 397         mutex_destroy(&adapter->lock);
 398         mutex_destroy(&adapter->mactbl_lock);
 399         kfree(adapter);
 400         rn->free_rdma_netdev(netdev);
 401 }