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                                  void *accel_priv,
  99                                  select_queue_fallback_t fallback)
 100 {
 101         struct opa_vnic_adapter *adapter = opa_vnic_priv(netdev);
 102         struct opa_vnic_skb_mdata *mdata;
 103         int rc;
 104
 105         /* pass entropy and vl as metadata in skb */
 106         mdata = skb_push(skb, sizeof(*mdata));
 107         mdata->entropy =  opa_vnic_calc_entropy(adapter, skb);
 108         mdata->vl = opa_vnic_get_vl(adapter, skb);
 109         rc = adapter->rn_ops->ndo_select_queue(netdev, skb,
 110                                                accel_priv, fallback);
 111         skb_pull(skb, sizeof(*mdata));
 112         return rc;
 113 }
 114
 115 static void opa_vnic_update_state(struct opa_vnic_adapter *adapter, bool up)
 116 {
 117         struct __opa_veswport_info *info = &adapter->info;
 118
 119         mutex_lock(&adapter->lock);
 120         /* Operational state can only be DROP_ALL or FORWARDING */
 121         if ((info->vport.config_state == OPA_VNIC_STATE_FORWARDING) && up) {
 122                 info->vport.oper_state = OPA_VNIC_STATE_FORWARDING;
 123                 info->vport.eth_link_status = OPA_VNIC_ETH_LINK_UP;
 124         } else {
 125                 info->vport.oper_state = OPA_VNIC_STATE_DROP_ALL;
 126                 info->vport.eth_link_status = OPA_VNIC_ETH_LINK_DOWN;
 127         }
 128
 129         if (info->vport.config_state == OPA_VNIC_STATE_FORWARDING)
 130                 netif_dormant_off(adapter->netdev);
 131         else
 132                 netif_dormant_on(adapter->netdev);
 133         mutex_unlock(&adapter->lock);
 134 }
 135
 136 /* opa_vnic_process_vema_config - process vema configuration updates */
 137 void opa_vnic_process_vema_config(struct opa_vnic_adapter *adapter)
 138 {
 139         struct __opa_veswport_info *info = &adapter->info;
 140         struct rdma_netdev *rn = netdev_priv(adapter->netdev);
 141         u8 port_num[OPA_VESW_MAX_NUM_DEF_PORT] = { 0 };
 142         struct net_device *netdev = adapter->netdev;
 143         u8 i, port_count = 0;
 144         u16 port_mask;
 145
 146         /* If the base_mac_addr is changed, update the interface mac address */
 147         if (memcmp(info->vport.base_mac_addr, adapter->vema_mac_addr,
 148                    ARRAY_SIZE(info->vport.base_mac_addr))) {
 149                 struct sockaddr saddr;
 150
 151                 memcpy(saddr.sa_data, info->vport.base_mac_addr,
 152                        ARRAY_SIZE(info->vport.base_mac_addr));
 153                 mutex_lock(&adapter->lock);
 154                 eth_commit_mac_addr_change(netdev, &saddr);
 155                 memcpy(adapter->vema_mac_addr,
 156                        info->vport.base_mac_addr, ETH_ALEN);
 157                 mutex_unlock(&adapter->lock);
 158         }
 159
 160         rn->set_id(netdev, info->vesw.vesw_id);
 161
 162         /* Handle MTU limit change */
 163         rtnl_lock();
 164         netdev->max_mtu = max_t(unsigned int, info->vesw.eth_mtu,
 165                                 netdev->min_mtu);
 166         if (netdev->mtu > netdev->max_mtu)
 167                 dev_set_mtu(netdev, netdev->max_mtu);
 168         rtnl_unlock();
 169
 170         /* Update flow to default port redirection table */
 171         port_mask = info->vesw.def_port_mask;
 172         for (i = 0; i < OPA_VESW_MAX_NUM_DEF_PORT; i++) {
 173                 if (port_mask & 1)
 174                         port_num[port_count++] = i;
 175                 port_mask >>= 1;
 176         }
 177
 178         /*
 179          * Build the flow table. Flow table is required when destination LID
 180          * is not available. Up to OPA_VNIC_FLOW_TBL_SIZE flows supported.
 181          * Each flow need a default port number to get its dlid from the
 182          * u_ucast_dlid array.
 183          */
 184         for (i = 0; i < OPA_VNIC_FLOW_TBL_SIZE; i++)
 185                 adapter->flow_tbl[i] = port_count ? port_num[i % port_count] :
 186                                                     OPA_VNIC_INVALID_PORT;
 187
 188         /* update state */
 189         opa_vnic_update_state(adapter, !!(netdev->flags & IFF_UP));
 190 }
 191
 192 /*
 193  * Set the power on default values in adapter's vema interface structure.
 194  */
 195 static inline void opa_vnic_set_pod_values(struct opa_vnic_adapter *adapter)
 196 {
 197         adapter->info.vport.max_mac_tbl_ent = OPA_VNIC_MAC_TBL_MAX_ENTRIES;
 198         adapter->info.vport.max_smac_ent = OPA_VNIC_MAX_SMAC_LIMIT;
 199         adapter->info.vport.config_state = OPA_VNIC_STATE_DROP_ALL;
 200         adapter->info.vport.eth_link_status = OPA_VNIC_ETH_LINK_DOWN;
 201         adapter->info.vesw.eth_mtu = ETH_DATA_LEN;
 202 }
 203
 204 /* opa_vnic_set_mac_addr - change mac address */
 205 static int opa_vnic_set_mac_addr(struct net_device *netdev, void *addr)
 206 {
 207         struct opa_vnic_adapter *adapter = opa_vnic_priv(netdev);
 208         struct sockaddr *sa = addr;
 209         int rc;
 210
 211         if (!memcmp(netdev->dev_addr, sa->sa_data, ETH_ALEN))
 212                 return 0;
 213
 214         mutex_lock(&adapter->lock);
 215         rc = eth_mac_addr(netdev, addr);
 216         mutex_unlock(&adapter->lock);
 217         if (rc)
 218                 return rc;
 219
 220         adapter->info.vport.uc_macs_gen_count++;
 221         opa_vnic_vema_report_event(adapter,
 222                                    OPA_VESWPORT_TRAP_IFACE_UCAST_MAC_CHANGE);
 223         return 0;
 224 }
 225
 226 /*
 227  * opa_vnic_mac_send_event - post event on possible mac list exchange
 228  *  Send trap when digest from uc/mc mac list differs from previous run.
 229  *  Digest is evaluated similar to how cksum does.
 230  */
 231 static void opa_vnic_mac_send_event(struct net_device *netdev, u8 event)
 232 {
 233         struct opa_vnic_adapter *adapter = opa_vnic_priv(netdev);
 234         struct netdev_hw_addr *ha;
 235         struct netdev_hw_addr_list *hw_list;
 236         u32 *ref_crc;
 237         u32 l, crc = 0;
 238
 239         switch (event) {
 240         case OPA_VESWPORT_TRAP_IFACE_UCAST_MAC_CHANGE:
 241                 hw_list = &netdev->uc;
 242                 adapter->info.vport.uc_macs_gen_count++;
 243                 ref_crc = &adapter->umac_hash;
 244                 break;
 245         case OPA_VESWPORT_TRAP_IFACE_MCAST_MAC_CHANGE:
 246                 hw_list = &netdev->mc;
 247                 adapter->info.vport.mc_macs_gen_count++;
 248                 ref_crc = &adapter->mmac_hash;
 249                 break;
 250         default:
 251                 return;
 252         }
 253         netdev_hw_addr_list_for_each(ha, hw_list) {
 254                 crc = crc32_le(crc, ha->addr, ETH_ALEN);
 255         }
 256         l = netdev_hw_addr_list_count(hw_list) * ETH_ALEN;
 257         crc = ~crc32_le(crc, (void *)&l, sizeof(l));
 258
 259         if (crc != *ref_crc) {
 260                 *ref_crc = crc;
 261                 opa_vnic_vema_report_event(adapter, event);
 262         }
 263 }
 264
 265 /* opa_vnic_set_rx_mode - handle uc/mc mac list change */
 266 static void opa_vnic_set_rx_mode(struct net_device *netdev)
 267 {
 268         opa_vnic_mac_send_event(netdev,
 269                                 OPA_VESWPORT_TRAP_IFACE_UCAST_MAC_CHANGE);
 270
 271         opa_vnic_mac_send_event(netdev,
 272                                 OPA_VESWPORT_TRAP_IFACE_MCAST_MAC_CHANGE);
 273 }
 274
 275 /* opa_netdev_open - activate network interface */
 276 static int opa_netdev_open(struct net_device *netdev)
 277 {
 278         struct opa_vnic_adapter *adapter = opa_vnic_priv(netdev);
 279         int rc;
 280
 281         rc = adapter->rn_ops->ndo_open(adapter->netdev);
 282         if (rc) {
 283                 v_dbg("open failed %d\n", rc);
 284                 return rc;
 285         }
 286
 287         /* Update status and send trap */
 288         opa_vnic_update_state(adapter, true);
 289         opa_vnic_vema_report_event(adapter,
 290                                    OPA_VESWPORT_TRAP_ETH_LINK_STATUS_CHANGE);
 291         return 0;
 292 }
 293
 294 /* opa_netdev_close - disable network interface */
 295 static int opa_netdev_close(struct net_device *netdev)
 296 {
 297         struct opa_vnic_adapter *adapter = opa_vnic_priv(netdev);
 298         int rc;
 299
 300         rc = adapter->rn_ops->ndo_stop(adapter->netdev);
 301         if (rc) {
 302                 v_dbg("close failed %d\n", rc);
 303                 return rc;
 304         }
 305
 306         /* Update status and send trap */
 307         opa_vnic_update_state(adapter, false);
 308         opa_vnic_vema_report_event(adapter,
 309                                    OPA_VESWPORT_TRAP_ETH_LINK_STATUS_CHANGE);
 310         return 0;
 311 }
 312
 313 /* netdev ops */
 314 static const struct net_device_ops opa_netdev_ops = {
 315         .ndo_open = opa_netdev_open,
 316         .ndo_stop = opa_netdev_close,
 317         .ndo_start_xmit = opa_netdev_start_xmit,
 318         .ndo_get_stats64 = opa_vnic_get_stats64,
 319         .ndo_set_rx_mode = opa_vnic_set_rx_mode,
 320         .ndo_select_queue = opa_vnic_select_queue,
 321         .ndo_set_mac_address = opa_vnic_set_mac_addr,
 322 };
 323
 324 /* opa_vnic_add_netdev - create vnic netdev interface */
 325 struct opa_vnic_adapter *opa_vnic_add_netdev(struct ib_device *ibdev,
 326                                              u8 port_num, u8 vport_num)
 327 {
 328         struct opa_vnic_adapter *adapter;
 329         struct net_device *netdev;
 330         struct rdma_netdev *rn;
 331         int rc;
 332
 333         netdev = ibdev->alloc_rdma_netdev(ibdev, port_num,
 334                                           RDMA_NETDEV_OPA_VNIC,
 335                                           "veth%d", NET_NAME_UNKNOWN,
 336                                           ether_setup);
 337         if (!netdev)
 338                 return ERR_PTR(-ENOMEM);
 339         else if (IS_ERR(netdev))
 340                 return ERR_CAST(netdev);
 341
 342         rn = netdev_priv(netdev);
 343         adapter = kzalloc(sizeof(*adapter), GFP_KERNEL);
 344         if (!adapter) {
 345                 rc = -ENOMEM;
 346                 goto adapter_err;
 347         }
 348
 349         rn->clnt_priv = adapter;
 350         rn->hca = ibdev;
 351         rn->port_num = port_num;
 352         adapter->netdev = netdev;
 353         adapter->ibdev = ibdev;
 354         adapter->port_num = port_num;
 355         adapter->vport_num = vport_num;
 356         adapter->rn_ops = netdev->netdev_ops;
 357
 358         netdev->netdev_ops = &opa_netdev_ops;
 359         netdev->priv_flags |= IFF_LIVE_ADDR_CHANGE;
 360         netdev->hard_header_len += OPA_VNIC_SKB_HEADROOM;
 361         mutex_init(&adapter->lock);
 362         mutex_init(&adapter->mactbl_lock);
 363         spin_lock_init(&adapter->stats_lock);
 364
 365         SET_NETDEV_DEV(netdev, ibdev->dev.parent);
 366
 367         opa_vnic_set_ethtool_ops(netdev);
 368
 369         opa_vnic_set_pod_values(adapter);
 370
 371         rc = register_netdev(netdev);
 372         if (rc)
 373                 goto netdev_err;
 374
 375         netif_carrier_off(netdev);
 376         netif_dormant_on(netdev);
 377         v_info("initialized\n");
 378
 379         return adapter;
 380 netdev_err:
 381         mutex_destroy(&adapter->lock);
 382         mutex_destroy(&adapter->mactbl_lock);
 383         kfree(adapter);
 384 adapter_err:
 385         rn->free_rdma_netdev(netdev);
 386
 387         return ERR_PTR(rc);
 388 }
 389
 390 /* opa_vnic_rem_netdev - remove vnic netdev interface */
 391 void opa_vnic_rem_netdev(struct opa_vnic_adapter *adapter)
 392 {
 393         struct net_device *netdev = adapter->netdev;
 394         struct rdma_netdev *rn = netdev_priv(netdev);
 395
 396         v_info("removing\n");
 397         unregister_netdev(netdev);
 398         opa_vnic_release_mac_tbl(adapter);
 399         mutex_destroy(&adapter->lock);
 400         mutex_destroy(&adapter->mactbl_lock);
 401         kfree(adapter);
 402         rn->free_rdma_netdev(netdev);
 403 }