// SPDX-License-Identifier: GPL-2.0-only /**************************************************************************** * Driver for Solarflare network controllers and boards * Copyright 2005-2006 Fen Systems Ltd. * Copyright 2005-2013 Solarflare Communications Inc. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "net_driver.h" #include #include #include "efx.h" #include "efx_common.h" #include "efx_channels.h" #include "rx_common.h" #include "tx_common.h" #include "nic.h" #include "io.h" #include "selftest.h" #include "sriov.h" #include "mcdi.h" #include "mcdi_pcol.h" #include "workarounds.h" /************************************************************************** * * Type name strings * ************************************************************************** */ /* UDP tunnel type names */ static const char *const efx_udp_tunnel_type_names[] = { [TUNNEL_ENCAP_UDP_PORT_ENTRY_VXLAN] = "vxlan", [TUNNEL_ENCAP_UDP_PORT_ENTRY_GENEVE] = "geneve", }; void efx_get_udp_tunnel_type_name(u16 type, char *buf, size_t buflen) { if (type < ARRAY_SIZE(efx_udp_tunnel_type_names) && efx_udp_tunnel_type_names[type] != NULL) snprintf(buf, buflen, "%s", efx_udp_tunnel_type_names[type]); else snprintf(buf, buflen, "type %d", type); } /************************************************************************** * * Configurable values * *************************************************************************/ /* * Use separate channels for TX and RX events * * Set this to 1 to use separate channels for TX and RX. It allows us * to control interrupt affinity separately for TX and RX. * * This is only used in MSI-X interrupt mode */ bool efx_separate_tx_channels; module_param(efx_separate_tx_channels, bool, 0444); MODULE_PARM_DESC(efx_separate_tx_channels, "Use separate channels for TX and RX"); /* Initial interrupt moderation settings. They can be modified after * module load with ethtool. * * The default for RX should strike a balance between increasing the * round-trip latency and reducing overhead. */ static unsigned int rx_irq_mod_usec = 60; /* Initial interrupt moderation settings. They can be modified after * module load with ethtool. * * This default is chosen to ensure that a 10G link does not go idle * while a TX queue is stopped after it has become full. A queue is * restarted when it drops below half full. The time this takes (assuming * worst case 3 descriptors per packet and 1024 descriptors) is * 512 / 3 * 1.2 = 205 usec. */ static unsigned int tx_irq_mod_usec = 150; static bool phy_flash_cfg; module_param(phy_flash_cfg, bool, 0644); MODULE_PARM_DESC(phy_flash_cfg, "Set PHYs into reflash mode initially"); static unsigned debug = (NETIF_MSG_DRV | NETIF_MSG_PROBE | NETIF_MSG_LINK | NETIF_MSG_IFDOWN | NETIF_MSG_IFUP | NETIF_MSG_RX_ERR | NETIF_MSG_TX_ERR | NETIF_MSG_HW); module_param(debug, uint, 0); MODULE_PARM_DESC(debug, "Bitmapped debugging message enable value"); /************************************************************************** * * Utility functions and prototypes * *************************************************************************/ static const struct efx_channel_type efx_default_channel_type; static void efx_remove_port(struct efx_nic *efx); static int efx_xdp_setup_prog(struct efx_nic *efx, struct bpf_prog *prog); static int efx_xdp(struct net_device *dev, struct netdev_bpf *xdp); static int efx_xdp_xmit(struct net_device *dev, int n, struct xdp_frame **xdpfs, u32 flags); #define EFX_ASSERT_RESET_SERIALISED(efx) \ do { \ if ((efx->state == STATE_READY) || \ (efx->state == STATE_RECOVERY) || \ (efx->state == STATE_DISABLED)) \ ASSERT_RTNL(); \ } while (0) /************************************************************************** * * Channel handling * *************************************************************************/ void efx_schedule_slow_fill(struct efx_rx_queue *rx_queue) { mod_timer(&rx_queue->slow_fill, jiffies + msecs_to_jiffies(10)); } /************************************************************************** * * Port handling * **************************************************************************/ void efx_link_set_advertising(struct efx_nic *efx, const unsigned long *advertising) { memcpy(efx->link_advertising, advertising, sizeof(__ETHTOOL_DECLARE_LINK_MODE_MASK())); efx->link_advertising[0] |= ADVERTISED_Autoneg; if (advertising[0] & ADVERTISED_Pause) efx->wanted_fc |= (EFX_FC_TX | EFX_FC_RX); else efx->wanted_fc &= ~(EFX_FC_TX | EFX_FC_RX); if (advertising[0] & ADVERTISED_Asym_Pause) efx->wanted_fc ^= EFX_FC_TX; } /* Equivalent to efx_link_set_advertising with all-zeroes, except does not * force the Autoneg bit on. */ void efx_link_clear_advertising(struct efx_nic *efx) { bitmap_zero(efx->link_advertising, __ETHTOOL_LINK_MODE_MASK_NBITS); efx->wanted_fc &= ~(EFX_FC_TX | EFX_FC_RX); } void efx_link_set_wanted_fc(struct efx_nic *efx, u8 wanted_fc) { efx->wanted_fc = wanted_fc; if (efx->link_advertising[0]) { if (wanted_fc & EFX_FC_RX) efx->link_advertising[0] |= (ADVERTISED_Pause | ADVERTISED_Asym_Pause); else efx->link_advertising[0] &= ~(ADVERTISED_Pause | ADVERTISED_Asym_Pause); if (wanted_fc & EFX_FC_TX) efx->link_advertising[0] ^= ADVERTISED_Asym_Pause; } } static void efx_fini_port(struct efx_nic *efx); static int efx_probe_port(struct efx_nic *efx) { int rc; netif_dbg(efx, probe, efx->net_dev, "create port\n"); if (phy_flash_cfg) efx->phy_mode = PHY_MODE_SPECIAL; /* Connect up MAC/PHY operations table */ rc = efx->type->probe_port(efx); if (rc) return rc; /* Initialise MAC address to permanent address */ ether_addr_copy(efx->net_dev->dev_addr, efx->net_dev->perm_addr); return 0; } static int efx_init_port(struct efx_nic *efx) { int rc; netif_dbg(efx, drv, efx->net_dev, "init port\n"); mutex_lock(&efx->mac_lock); rc = efx->phy_op->init(efx); if (rc) goto fail1; efx->port_initialized = true; /* Reconfigure the MAC before creating dma queues (required for * Falcon/A1 where RX_INGR_EN/TX_DRAIN_EN isn't supported) */ efx_mac_reconfigure(efx); /* Ensure the PHY advertises the correct flow control settings */ rc = efx->phy_op->reconfigure(efx); if (rc && rc != -EPERM) goto fail2; mutex_unlock(&efx->mac_lock); return 0; fail2: efx->phy_op->fini(efx); fail1: mutex_unlock(&efx->mac_lock); return rc; } static void efx_fini_port(struct efx_nic *efx) { netif_dbg(efx, drv, efx->net_dev, "shut down port\n"); if (!efx->port_initialized) return; efx->phy_op->fini(efx); efx->port_initialized = false; efx->link_state.up = false; efx_link_status_changed(efx); } static void efx_remove_port(struct efx_nic *efx) { netif_dbg(efx, drv, efx->net_dev, "destroying port\n"); efx->type->remove_port(efx); } /************************************************************************** * * NIC handling * **************************************************************************/ static LIST_HEAD(efx_primary_list); static LIST_HEAD(efx_unassociated_list); static bool efx_same_controller(struct efx_nic *left, struct efx_nic *right) { return left->type == right->type && left->vpd_sn && right->vpd_sn && !strcmp(left->vpd_sn, right->vpd_sn); } static void efx_associate(struct efx_nic *efx) { struct efx_nic *other, *next; if (efx->primary == efx) { /* Adding primary function; look for secondaries */ netif_dbg(efx, probe, efx->net_dev, "adding to primary list\n"); list_add_tail(&efx->node, &efx_primary_list); list_for_each_entry_safe(other, next, &efx_unassociated_list, node) { if (efx_same_controller(efx, other)) { list_del(&other->node); netif_dbg(other, probe, other->net_dev, "moving to secondary list of %s %s\n", pci_name(efx->pci_dev), efx->net_dev->name); list_add_tail(&other->node, &efx->secondary_list); other->primary = efx; } } } else { /* Adding secondary function; look for primary */ list_for_each_entry(other, &efx_primary_list, node) { if (efx_same_controller(efx, other)) { netif_dbg(efx, probe, efx->net_dev, "adding to secondary list of %s %s\n", pci_name(other->pci_dev), other->net_dev->name); list_add_tail(&efx->node, &other->secondary_list); efx->primary = other; return; } } netif_dbg(efx, probe, efx->net_dev, "adding to unassociated list\n"); list_add_tail(&efx->node, &efx_unassociated_list); } } static void efx_dissociate(struct efx_nic *efx) { struct efx_nic *other, *next; list_del(&efx->node); efx->primary = NULL; list_for_each_entry_safe(other, next, &efx->secondary_list, node) { list_del(&other->node); netif_dbg(other, probe, other->net_dev, "moving to unassociated list\n"); list_add_tail(&other->node, &efx_unassociated_list); other->primary = NULL; } } void efx_set_default_rx_indir_table(struct efx_nic *efx, struct efx_rss_context *ctx) { size_t i; for (i = 0; i < ARRAY_SIZE(ctx->rx_indir_table); i++) ctx->rx_indir_table[i] = ethtool_rxfh_indir_default(i, efx->rss_spread); } static int efx_probe_nic(struct efx_nic *efx) { int rc; netif_dbg(efx, probe, efx->net_dev, "creating NIC\n"); /* Carry out hardware-type specific initialisation */ rc = efx->type->probe(efx); if (rc) return rc; do { if (!efx->max_channels || !efx->max_tx_channels) { netif_err(efx, drv, efx->net_dev, "Insufficient resources to allocate" " any channels\n"); rc = -ENOSPC; goto fail1; } /* Determine the number of channels and queues by trying * to hook in MSI-X interrupts. */ rc = efx_probe_interrupts(efx); if (rc) goto fail1; rc = efx_set_channels(efx); if (rc) goto fail1; /* dimension_resources can fail with EAGAIN */ rc = efx->type->dimension_resources(efx); if (rc != 0 && rc != -EAGAIN) goto fail2; if (rc == -EAGAIN) /* try again with new max_channels */ efx_remove_interrupts(efx); } while (rc == -EAGAIN); if (efx->n_channels > 1) netdev_rss_key_fill(efx->rss_context.rx_hash_key, sizeof(efx->rss_context.rx_hash_key)); efx_set_default_rx_indir_table(efx, &efx->rss_context); netif_set_real_num_tx_queues(efx->net_dev, efx->n_tx_channels); netif_set_real_num_rx_queues(efx->net_dev, efx->n_rx_channels); /* Initialise the interrupt moderation settings */ efx->irq_mod_step_us = DIV_ROUND_UP(efx->timer_quantum_ns, 1000); efx_init_irq_moderation(efx, tx_irq_mod_usec, rx_irq_mod_usec, true, true); return 0; fail2: efx_remove_interrupts(efx); fail1: efx->type->remove(efx); return rc; } static void efx_remove_nic(struct efx_nic *efx) { netif_dbg(efx, drv, efx->net_dev, "destroying NIC\n"); efx_remove_interrupts(efx); efx->type->remove(efx); } static int efx_probe_filters(struct efx_nic *efx) { int rc; init_rwsem(&efx->filter_sem); mutex_lock(&efx->mac_lock); down_write(&efx->filter_sem); rc = efx->type->filter_table_probe(efx); if (rc) goto out_unlock; #ifdef CONFIG_RFS_ACCEL if (efx->type->offload_features & NETIF_F_NTUPLE) { struct efx_channel *channel; int i, success = 1; efx_for_each_channel(channel, efx) { channel->rps_flow_id = kcalloc(efx->type->max_rx_ip_filters, sizeof(*channel->rps_flow_id), GFP_KERNEL); if (!channel->rps_flow_id) success = 0; else for (i = 0; i < efx->type->max_rx_ip_filters; ++i) channel->rps_flow_id[i] = RPS_FLOW_ID_INVALID; channel->rfs_expire_index = 0; channel->rfs_filter_count = 0; } if (!success) { efx_for_each_channel(channel, efx) kfree(channel->rps_flow_id); efx->type->filter_table_remove(efx); rc = -ENOMEM; goto out_unlock; } } #endif out_unlock: up_write(&efx->filter_sem); mutex_unlock(&efx->mac_lock); return rc; } static void efx_remove_filters(struct efx_nic *efx) { #ifdef CONFIG_RFS_ACCEL struct efx_channel *channel; efx_for_each_channel(channel, efx) { cancel_delayed_work_sync(&channel->filter_work); kfree(channel->rps_flow_id); } #endif down_write(&efx->filter_sem); efx->type->filter_table_remove(efx); up_write(&efx->filter_sem); } /************************************************************************** * * NIC startup/shutdown * *************************************************************************/ static int efx_probe_all(struct efx_nic *efx) { int rc; rc = efx_probe_nic(efx); if (rc) { netif_err(efx, probe, efx->net_dev, "failed to create NIC\n"); goto fail1; } rc = efx_probe_port(efx); if (rc) { netif_err(efx, probe, efx->net_dev, "failed to create port\n"); goto fail2; } BUILD_BUG_ON(EFX_DEFAULT_DMAQ_SIZE < EFX_RXQ_MIN_ENT); if (WARN_ON(EFX_DEFAULT_DMAQ_SIZE < EFX_TXQ_MIN_ENT(efx))) { rc = -EINVAL; goto fail3; } efx->rxq_entries = efx->txq_entries = EFX_DEFAULT_DMAQ_SIZE; #ifdef CONFIG_SFC_SRIOV rc = efx->type->vswitching_probe(efx); if (rc) /* not fatal; the PF will still work fine */ netif_warn(efx, probe, efx->net_dev, "failed to setup vswitching rc=%d;" " VFs may not function\n", rc); #endif rc = efx_probe_filters(efx); if (rc) { netif_err(efx, probe, efx->net_dev, "failed to create filter tables\n"); goto fail4; } rc = efx_probe_channels(efx); if (rc) goto fail5; return 0; fail5: efx_remove_filters(efx); fail4: #ifdef CONFIG_SFC_SRIOV efx->type->vswitching_remove(efx); #endif fail3: efx_remove_port(efx); fail2: efx_remove_nic(efx); fail1: return rc; } static void efx_remove_all(struct efx_nic *efx) { rtnl_lock(); efx_xdp_setup_prog(efx, NULL); rtnl_unlock(); efx_remove_channels(efx); efx_remove_filters(efx); #ifdef CONFIG_SFC_SRIOV efx->type->vswitching_remove(efx); #endif efx_remove_port(efx); efx_remove_nic(efx); } /************************************************************************** * * Interrupt moderation * **************************************************************************/ unsigned int efx_usecs_to_ticks(struct efx_nic *efx, unsigned int usecs) { if (usecs == 0) return 0; if (usecs * 1000 < efx->timer_quantum_ns) return 1; /* never round down to 0 */ return usecs * 1000 / efx->timer_quantum_ns; } unsigned int efx_ticks_to_usecs(struct efx_nic *efx, unsigned int ticks) { /* We must round up when converting ticks to microseconds * because we round down when converting the other way. */ return DIV_ROUND_UP(ticks * efx->timer_quantum_ns, 1000); } /* Set interrupt moderation parameters */ int efx_init_irq_moderation(struct efx_nic *efx, unsigned int tx_usecs, unsigned int rx_usecs, bool rx_adaptive, bool rx_may_override_tx) { struct efx_channel *channel; unsigned int timer_max_us; EFX_ASSERT_RESET_SERIALISED(efx); timer_max_us = efx->timer_max_ns / 1000; if (tx_usecs > timer_max_us || rx_usecs > timer_max_us) return -EINVAL; if (tx_usecs != rx_usecs && efx->tx_channel_offset == 0 && !rx_may_override_tx) { netif_err(efx, drv, efx->net_dev, "Channels are shared. " "RX and TX IRQ moderation must be equal\n"); return -EINVAL; } efx->irq_rx_adaptive = rx_adaptive; efx->irq_rx_moderation_us = rx_usecs; efx_for_each_channel(channel, efx) { if (efx_channel_has_rx_queue(channel)) channel->irq_moderation_us = rx_usecs; else if (efx_channel_has_tx_queues(channel)) channel->irq_moderation_us = tx_usecs; else if (efx_channel_is_xdp_tx(channel)) channel->irq_moderation_us = tx_usecs; } return 0; } void efx_get_irq_moderation(struct efx_nic *efx, unsigned int *tx_usecs, unsigned int *rx_usecs, bool *rx_adaptive) { *rx_adaptive = efx->irq_rx_adaptive; *rx_usecs = efx->irq_rx_moderation_us; /* If channels are shared between RX and TX, so is IRQ * moderation. Otherwise, IRQ moderation is the same for all * TX channels and is not adaptive. */ if (efx->tx_channel_offset == 0) { *tx_usecs = *rx_usecs; } else { struct efx_channel *tx_channel; tx_channel = efx->channel[efx->tx_channel_offset]; *tx_usecs = tx_channel->irq_moderation_us; } } /************************************************************************** * * ioctls * *************************************************************************/ /* Net device ioctl * Context: process, rtnl_lock() held. */ static int efx_ioctl(struct net_device *net_dev, struct ifreq *ifr, int cmd) { struct efx_nic *efx = netdev_priv(net_dev); struct mii_ioctl_data *data = if_mii(ifr); if (cmd == SIOCSHWTSTAMP) return efx_ptp_set_ts_config(efx, ifr); if (cmd == SIOCGHWTSTAMP) return efx_ptp_get_ts_config(efx, ifr); /* Convert phy_id from older PRTAD/DEVAD format */ if ((cmd == SIOCGMIIREG || cmd == SIOCSMIIREG) && (data->phy_id & 0xfc00) == 0x0400) data->phy_id ^= MDIO_PHY_ID_C45 | 0x0400; return mdio_mii_ioctl(&efx->mdio, data, cmd); } /************************************************************************** * * Kernel net device interface * *************************************************************************/ /* Context: process, rtnl_lock() held. */ int efx_net_open(struct net_device *net_dev) { struct efx_nic *efx = netdev_priv(net_dev); int rc; netif_dbg(efx, ifup, efx->net_dev, "opening device on CPU %d\n", raw_smp_processor_id()); rc = efx_check_disabled(efx); if (rc) return rc; if (efx->phy_mode & PHY_MODE_SPECIAL) return -EBUSY; if (efx_mcdi_poll_reboot(efx) && efx_reset(efx, RESET_TYPE_ALL)) return -EIO; /* Notify the kernel of the link state polled during driver load, * before the monitor starts running */ efx_link_status_changed(efx); efx_start_all(efx); if (efx->state == STATE_DISABLED || efx->reset_pending) netif_device_detach(efx->net_dev); efx_selftest_async_start(efx); return 0; } /* Context: process, rtnl_lock() held. * Note that the kernel will ignore our return code; this method * should really be a void. */ int efx_net_stop(struct net_device *net_dev) { struct efx_nic *efx = netdev_priv(net_dev); netif_dbg(efx, ifdown, efx->net_dev, "closing on CPU %d\n", raw_smp_processor_id()); /* Stop the device and flush all the channels */ efx_stop_all(efx); return 0; } /* Context: process, dev_base_lock or RTNL held, non-blocking. */ static void efx_net_stats(struct net_device *net_dev, struct rtnl_link_stats64 *stats) { struct efx_nic *efx = netdev_priv(net_dev); spin_lock_bh(&efx->stats_lock); efx->type->update_stats(efx, NULL, stats); spin_unlock_bh(&efx->stats_lock); } /* Context: netif_tx_lock held, BHs disabled. */ static void efx_watchdog(struct net_device *net_dev, unsigned int txqueue) { struct efx_nic *efx = netdev_priv(net_dev); netif_err(efx, tx_err, efx->net_dev, "TX stuck with port_enabled=%d: resetting channels\n", efx->port_enabled); efx_schedule_reset(efx, RESET_TYPE_TX_WATCHDOG); } static unsigned int efx_xdp_max_mtu(struct efx_nic *efx) { /* The maximum MTU that we can fit in a single page, allowing for * framing, overhead and XDP headroom. */ int overhead = EFX_MAX_FRAME_LEN(0) + sizeof(struct efx_rx_page_state) + efx->rx_prefix_size + efx->type->rx_buffer_padding + efx->rx_ip_align + XDP_PACKET_HEADROOM; return PAGE_SIZE - overhead; } /* Context: process, rtnl_lock() held. */ static int efx_change_mtu(struct net_device *net_dev, int new_mtu) { struct efx_nic *efx = netdev_priv(net_dev); int rc; rc = efx_check_disabled(efx); if (rc) return rc; if (rtnl_dereference(efx->xdp_prog) && new_mtu > efx_xdp_max_mtu(efx)) { netif_err(efx, drv, efx->net_dev, "Requested MTU of %d too big for XDP (max: %d)\n", new_mtu, efx_xdp_max_mtu(efx)); return -EINVAL; } netif_dbg(efx, drv, efx->net_dev, "changing MTU to %d\n", new_mtu); efx_device_detach_sync(efx); efx_stop_all(efx); mutex_lock(&efx->mac_lock); net_dev->mtu = new_mtu; efx_mac_reconfigure(efx); mutex_unlock(&efx->mac_lock); efx_start_all(efx); efx_device_attach_if_not_resetting(efx); return 0; } static int efx_set_mac_address(struct net_device *net_dev, void *data) { struct efx_nic *efx = netdev_priv(net_dev); struct sockaddr *addr = data; u8 *new_addr = addr->sa_data; u8 old_addr[6]; int rc; if (!is_valid_ether_addr(new_addr)) { netif_err(efx, drv, efx->net_dev, "invalid ethernet MAC address requested: %pM\n", new_addr); return -EADDRNOTAVAIL; } /* save old address */ ether_addr_copy(old_addr, net_dev->dev_addr); ether_addr_copy(net_dev->dev_addr, new_addr); if (efx->type->set_mac_address) { rc = efx->type->set_mac_address(efx); if (rc) { ether_addr_copy(net_dev->dev_addr, old_addr); return rc; } } /* Reconfigure the MAC */ mutex_lock(&efx->mac_lock); efx_mac_reconfigure(efx); mutex_unlock(&efx->mac_lock); return 0; } /* Context: netif_addr_lock held, BHs disabled. */ static void efx_set_rx_mode(struct net_device *net_dev) { struct efx_nic *efx = netdev_priv(net_dev); if (efx->port_enabled) queue_work(efx->workqueue, &efx->mac_work); /* Otherwise efx_start_port() will do this */ } static int efx_set_features(struct net_device *net_dev, netdev_features_t data) { struct efx_nic *efx = netdev_priv(net_dev); int rc; /* If disabling RX n-tuple filtering, clear existing filters */ if (net_dev->features & ~data & NETIF_F_NTUPLE) { rc = efx->type->filter_clear_rx(efx, EFX_FILTER_PRI_MANUAL); if (rc) return rc; } /* If Rx VLAN filter is changed, update filters via mac_reconfigure. * If rx-fcs is changed, mac_reconfigure updates that too. */ if ((net_dev->features ^ data) & (NETIF_F_HW_VLAN_CTAG_FILTER | NETIF_F_RXFCS)) { /* efx_set_rx_mode() will schedule MAC work to update filters * when a new features are finally set in net_dev. */ efx_set_rx_mode(net_dev); } return 0; } static int efx_get_phys_port_id(struct net_device *net_dev, struct netdev_phys_item_id *ppid) { struct efx_nic *efx = netdev_priv(net_dev); if (efx->type->get_phys_port_id) return efx->type->get_phys_port_id(efx, ppid); else return -EOPNOTSUPP; } static int efx_get_phys_port_name(struct net_device *net_dev, char *name, size_t len) { struct efx_nic *efx = netdev_priv(net_dev); if (snprintf(name, len, "p%u", efx->port_num) >= len) return -EINVAL; return 0; } static int efx_vlan_rx_add_vid(struct net_device *net_dev, __be16 proto, u16 vid) { struct efx_nic *efx = netdev_priv(net_dev); if (efx->type->vlan_rx_add_vid) return efx->type->vlan_rx_add_vid(efx, proto, vid); else return -EOPNOTSUPP; } static int efx_vlan_rx_kill_vid(struct net_device *net_dev, __be16 proto, u16 vid) { struct efx_nic *efx = netdev_priv(net_dev); if (efx->type->vlan_rx_kill_vid) return efx->type->vlan_rx_kill_vid(efx, proto, vid); else return -EOPNOTSUPP; } static int efx_udp_tunnel_type_map(enum udp_parsable_tunnel_type in) { switch (in) { case UDP_TUNNEL_TYPE_VXLAN: return TUNNEL_ENCAP_UDP_PORT_ENTRY_VXLAN; case UDP_TUNNEL_TYPE_GENEVE: return TUNNEL_ENCAP_UDP_PORT_ENTRY_GENEVE; default: return -1; } } static void efx_udp_tunnel_add(struct net_device *dev, struct udp_tunnel_info *ti) { struct efx_nic *efx = netdev_priv(dev); struct efx_udp_tunnel tnl; int efx_tunnel_type; efx_tunnel_type = efx_udp_tunnel_type_map(ti->type); if (efx_tunnel_type < 0) return; tnl.type = (u16)efx_tunnel_type; tnl.port = ti->port; if (efx->type->udp_tnl_add_port) (void)efx->type->udp_tnl_add_port(efx, tnl); } static void efx_udp_tunnel_del(struct net_device *dev, struct udp_tunnel_info *ti) { struct efx_nic *efx = netdev_priv(dev); struct efx_udp_tunnel tnl; int efx_tunnel_type; efx_tunnel_type = efx_udp_tunnel_type_map(ti->type); if (efx_tunnel_type < 0) return; tnl.type = (u16)efx_tunnel_type; tnl.port = ti->port; if (efx->type->udp_tnl_del_port) (void)efx->type->udp_tnl_del_port(efx, tnl); } static const struct net_device_ops efx_netdev_ops = { .ndo_open = efx_net_open, .ndo_stop = efx_net_stop, .ndo_get_stats64 = efx_net_stats, .ndo_tx_timeout = efx_watchdog, .ndo_start_xmit = efx_hard_start_xmit, .ndo_validate_addr = eth_validate_addr, .ndo_do_ioctl = efx_ioctl, .ndo_change_mtu = efx_change_mtu, .ndo_set_mac_address = efx_set_mac_address, .ndo_set_rx_mode = efx_set_rx_mode, .ndo_set_features = efx_set_features, .ndo_vlan_rx_add_vid = efx_vlan_rx_add_vid, .ndo_vlan_rx_kill_vid = efx_vlan_rx_kill_vid, #ifdef CONFIG_SFC_SRIOV .ndo_set_vf_mac = efx_sriov_set_vf_mac, .ndo_set_vf_vlan = efx_sriov_set_vf_vlan, .ndo_set_vf_spoofchk = efx_sriov_set_vf_spoofchk, .ndo_get_vf_config = efx_sriov_get_vf_config, .ndo_set_vf_link_state = efx_sriov_set_vf_link_state, #endif .ndo_get_phys_port_id = efx_get_phys_port_id, .ndo_get_phys_port_name = efx_get_phys_port_name, .ndo_setup_tc = efx_setup_tc, #ifdef CONFIG_RFS_ACCEL .ndo_rx_flow_steer = efx_filter_rfs, #endif .ndo_udp_tunnel_add = efx_udp_tunnel_add, .ndo_udp_tunnel_del = efx_udp_tunnel_del, .ndo_xdp_xmit = efx_xdp_xmit, .ndo_bpf = efx_xdp }; static int efx_xdp_setup_prog(struct efx_nic *efx, struct bpf_prog *prog) { struct bpf_prog *old_prog; if (efx->xdp_rxq_info_failed) { netif_err(efx, drv, efx->net_dev, "Unable to bind XDP program due to previous failure of rxq_info\n"); return -EINVAL; } if (prog && efx->net_dev->mtu > efx_xdp_max_mtu(efx)) { netif_err(efx, drv, efx->net_dev, "Unable to configure XDP with MTU of %d (max: %d)\n", efx->net_dev->mtu, efx_xdp_max_mtu(efx)); return -EINVAL; } old_prog = rtnl_dereference(efx->xdp_prog); rcu_assign_pointer(efx->xdp_prog, prog); /* Release the reference that was originally passed by the caller. */ if (old_prog) bpf_prog_put(old_prog); return 0; } /* Context: process, rtnl_lock() held. */ static int efx_xdp(struct net_device *dev, struct netdev_bpf *xdp) { struct efx_nic *efx = netdev_priv(dev); struct bpf_prog *xdp_prog; switch (xdp->command) { case XDP_SETUP_PROG: return efx_xdp_setup_prog(efx, xdp->prog); case XDP_QUERY_PROG: xdp_prog = rtnl_dereference(efx->xdp_prog); xdp->prog_id = xdp_prog ? xdp_prog->aux->id : 0; return 0; default: return -EINVAL; } } static int efx_xdp_xmit(struct net_device *dev, int n, struct xdp_frame **xdpfs, u32 flags) { struct efx_nic *efx = netdev_priv(dev); if (!netif_running(dev)) return -EINVAL; return efx_xdp_tx_buffers(efx, n, xdpfs, flags & XDP_XMIT_FLUSH); } static void efx_update_name(struct efx_nic *efx) { strcpy(efx->name, efx->net_dev->name); efx_mtd_rename(efx); efx_set_channel_names(efx); } static int efx_netdev_event(struct notifier_block *this, unsigned long event, void *ptr) { struct net_device *net_dev = netdev_notifier_info_to_dev(ptr); if ((net_dev->netdev_ops == &efx_netdev_ops) && event == NETDEV_CHANGENAME) efx_update_name(netdev_priv(net_dev)); return NOTIFY_DONE; } static struct notifier_block efx_netdev_notifier = { .notifier_call = efx_netdev_event, }; static ssize_t show_phy_type(struct device *dev, struct device_attribute *attr, char *buf) { struct efx_nic *efx = dev_get_drvdata(dev); return sprintf(buf, "%d\n", efx->phy_type); } static DEVICE_ATTR(phy_type, 0444, show_phy_type, NULL); #ifdef CONFIG_SFC_MCDI_LOGGING static ssize_t show_mcdi_log(struct device *dev, struct device_attribute *attr, char *buf) { struct efx_nic *efx = dev_get_drvdata(dev); struct efx_mcdi_iface *mcdi = efx_mcdi(efx); return scnprintf(buf, PAGE_SIZE, "%d\n", mcdi->logging_enabled); } static ssize_t set_mcdi_log(struct device *dev, struct device_attribute *attr, const char *buf, size_t count) { struct efx_nic *efx = dev_get_drvdata(dev); struct efx_mcdi_iface *mcdi = efx_mcdi(efx); bool enable = count > 0 && *buf != '0'; mcdi->logging_enabled = enable; return count; } static DEVICE_ATTR(mcdi_logging, 0644, show_mcdi_log, set_mcdi_log); #endif static int efx_register_netdev(struct efx_nic *efx) { struct net_device *net_dev = efx->net_dev; struct efx_channel *channel; int rc; net_dev->watchdog_timeo = 5 * HZ; net_dev->irq = efx->pci_dev->irq; net_dev->netdev_ops = &efx_netdev_ops; if (efx_nic_rev(efx) >= EFX_REV_HUNT_A0) net_dev->priv_flags |= IFF_UNICAST_FLT; net_dev->ethtool_ops = &efx_ethtool_ops; net_dev->gso_max_segs = EFX_TSO_MAX_SEGS; net_dev->min_mtu = EFX_MIN_MTU; net_dev->max_mtu = EFX_MAX_MTU; rtnl_lock(); /* Enable resets to be scheduled and check whether any were * already requested. If so, the NIC is probably hosed so we * abort. */ efx->state = STATE_READY; smp_mb(); /* ensure we change state before checking reset_pending */ if (efx->reset_pending) { netif_err(efx, probe, efx->net_dev, "aborting probe due to scheduled reset\n"); rc = -EIO; goto fail_locked; } rc = dev_alloc_name(net_dev, net_dev->name); if (rc < 0) goto fail_locked; efx_update_name(efx); /* Always start with carrier off; PHY events will detect the link */ netif_carrier_off(net_dev); rc = register_netdevice(net_dev); if (rc) goto fail_locked; efx_for_each_channel(channel, efx) { struct efx_tx_queue *tx_queue; efx_for_each_channel_tx_queue(tx_queue, channel) efx_init_tx_queue_core_txq(tx_queue); } efx_associate(efx); rtnl_unlock(); rc = device_create_file(&efx->pci_dev->dev, &dev_attr_phy_type); if (rc) { netif_err(efx, drv, efx->net_dev, "failed to init net dev attributes\n"); goto fail_registered; } #ifdef CONFIG_SFC_MCDI_LOGGING rc = device_create_file(&efx->pci_dev->dev, &dev_attr_mcdi_logging); if (rc) { netif_err(efx, drv, efx->net_dev, "failed to init net dev attributes\n"); goto fail_attr_mcdi_logging; } #endif return 0; #ifdef CONFIG_SFC_MCDI_LOGGING fail_attr_mcdi_logging: device_remove_file(&efx->pci_dev->dev, &dev_attr_phy_type); #endif fail_registered: rtnl_lock(); efx_dissociate(efx); unregister_netdevice(net_dev); fail_locked: efx->state = STATE_UNINIT; rtnl_unlock(); netif_err(efx, drv, efx->net_dev, "could not register net dev\n"); return rc; } static void efx_unregister_netdev(struct efx_nic *efx) { if (!efx->net_dev) return; BUG_ON(netdev_priv(efx->net_dev) != efx); if (efx_dev_registered(efx)) { strlcpy(efx->name, pci_name(efx->pci_dev), sizeof(efx->name)); #ifdef CONFIG_SFC_MCDI_LOGGING device_remove_file(&efx->pci_dev->dev, &dev_attr_mcdi_logging); #endif device_remove_file(&efx->pci_dev->dev, &dev_attr_phy_type); unregister_netdev(efx->net_dev); } } /************************************************************************** * * List of NICs we support * **************************************************************************/ /* PCI device ID table */ static const struct pci_device_id efx_pci_table[] = { {PCI_DEVICE(PCI_VENDOR_ID_SOLARFLARE, 0x0803), /* SFC9020 */ .driver_data = (unsigned long) &siena_a0_nic_type}, {PCI_DEVICE(PCI_VENDOR_ID_SOLARFLARE, 0x0813), /* SFL9021 */ .driver_data = (unsigned long) &siena_a0_nic_type}, {PCI_DEVICE(PCI_VENDOR_ID_SOLARFLARE, 0x0903), /* SFC9120 PF */ .driver_data = (unsigned long) &efx_hunt_a0_nic_type}, {PCI_DEVICE(PCI_VENDOR_ID_SOLARFLARE, 0x1903), /* SFC9120 VF */ .driver_data = (unsigned long) &efx_hunt_a0_vf_nic_type}, {PCI_DEVICE(PCI_VENDOR_ID_SOLARFLARE, 0x0923), /* SFC9140 PF */ .driver_data = (unsigned long) &efx_hunt_a0_nic_type}, {PCI_DEVICE(PCI_VENDOR_ID_SOLARFLARE, 0x1923), /* SFC9140 VF */ .driver_data = (unsigned long) &efx_hunt_a0_vf_nic_type}, {PCI_DEVICE(PCI_VENDOR_ID_SOLARFLARE, 0x0a03), /* SFC9220 PF */ .driver_data = (unsigned long) &efx_hunt_a0_nic_type}, {PCI_DEVICE(PCI_VENDOR_ID_SOLARFLARE, 0x1a03), /* SFC9220 VF */ .driver_data = (unsigned long) &efx_hunt_a0_vf_nic_type}, {PCI_DEVICE(PCI_VENDOR_ID_SOLARFLARE, 0x0b03), /* SFC9250 PF */ .driver_data = (unsigned long) &efx_hunt_a0_nic_type}, {PCI_DEVICE(PCI_VENDOR_ID_SOLARFLARE, 0x1b03), /* SFC9250 VF */ .driver_data = (unsigned long) &efx_hunt_a0_vf_nic_type}, {0} /* end of list */ }; /************************************************************************** * * Data housekeeping * **************************************************************************/ void efx_update_sw_stats(struct efx_nic *efx, u64 *stats) { u64 n_rx_nodesc_trunc = 0; struct efx_channel *channel; efx_for_each_channel(channel, efx) n_rx_nodesc_trunc += channel->n_rx_nodesc_trunc; stats[GENERIC_STAT_rx_nodesc_trunc] = n_rx_nodesc_trunc; stats[GENERIC_STAT_rx_noskb_drops] = atomic_read(&efx->n_rx_noskb_drops); } bool efx_filter_spec_equal(const struct efx_filter_spec *left, const struct efx_filter_spec *right) { if ((left->match_flags ^ right->match_flags) | ((left->flags ^ right->flags) & (EFX_FILTER_FLAG_RX | EFX_FILTER_FLAG_TX))) return false; return memcmp(&left->outer_vid, &right->outer_vid, sizeof(struct efx_filter_spec) - offsetof(struct efx_filter_spec, outer_vid)) == 0; } u32 efx_filter_spec_hash(const struct efx_filter_spec *spec) { BUILD_BUG_ON(offsetof(struct efx_filter_spec, outer_vid) & 3); return jhash2((const u32 *)&spec->outer_vid, (sizeof(struct efx_filter_spec) - offsetof(struct efx_filter_spec, outer_vid)) / 4, 0); } #ifdef CONFIG_RFS_ACCEL bool efx_rps_check_rule(struct efx_arfs_rule *rule, unsigned int filter_idx, bool *force) { if (rule->filter_id == EFX_ARFS_FILTER_ID_PENDING) { /* ARFS is currently updating this entry, leave it */ return false; } if (rule->filter_id == EFX_ARFS_FILTER_ID_ERROR) { /* ARFS tried and failed to update this, so it's probably out * of date. Remove the filter and the ARFS rule entry. */ rule->filter_id = EFX_ARFS_FILTER_ID_REMOVING; *force = true; return true; } else if (WARN_ON(rule->filter_id != filter_idx)) { /* can't happen */ /* ARFS has moved on, so old filter is not needed. Since we did * not mark the rule with EFX_ARFS_FILTER_ID_REMOVING, it will * not be removed by efx_rps_hash_del() subsequently. */ *force = true; return true; } /* Remove it iff ARFS wants to. */ return true; } static struct hlist_head *efx_rps_hash_bucket(struct efx_nic *efx, const struct efx_filter_spec *spec) { u32 hash = efx_filter_spec_hash(spec); lockdep_assert_held(&efx->rps_hash_lock); if (!efx->rps_hash_table) return NULL; return &efx->rps_hash_table[hash % EFX_ARFS_HASH_TABLE_SIZE]; } struct efx_arfs_rule *efx_rps_hash_find(struct efx_nic *efx, const struct efx_filter_spec *spec) { struct efx_arfs_rule *rule; struct hlist_head *head; struct hlist_node *node; head = efx_rps_hash_bucket(efx, spec); if (!head) return NULL; hlist_for_each(node, head) { rule = container_of(node, struct efx_arfs_rule, node); if (efx_filter_spec_equal(spec, &rule->spec)) return rule; } return NULL; } struct efx_arfs_rule *efx_rps_hash_add(struct efx_nic *efx, const struct efx_filter_spec *spec, bool *new) { struct efx_arfs_rule *rule; struct hlist_head *head; struct hlist_node *node; head = efx_rps_hash_bucket(efx, spec); if (!head) return NULL; hlist_for_each(node, head) { rule = container_of(node, struct efx_arfs_rule, node); if (efx_filter_spec_equal(spec, &rule->spec)) { *new = false; return rule; } } rule = kmalloc(sizeof(*rule), GFP_ATOMIC); *new = true; if (rule) { memcpy(&rule->spec, spec, sizeof(rule->spec)); hlist_add_head(&rule->node, head); } return rule; } void efx_rps_hash_del(struct efx_nic *efx, const struct efx_filter_spec *spec) { struct efx_arfs_rule *rule; struct hlist_head *head; struct hlist_node *node; head = efx_rps_hash_bucket(efx, spec); if (WARN_ON(!head)) return; hlist_for_each(node, head) { rule = container_of(node, struct efx_arfs_rule, node); if (efx_filter_spec_equal(spec, &rule->spec)) { /* Someone already reused the entry. We know that if * this check doesn't fire (i.e. filter_id == REMOVING) * then the REMOVING mark was put there by our caller, * because caller is holding a lock on filter table and * only holders of that lock set REMOVING. */ if (rule->filter_id != EFX_ARFS_FILTER_ID_REMOVING) return; hlist_del(node); kfree(rule); return; } } /* We didn't find it. */ WARN_ON(1); } #endif /* RSS contexts. We're using linked lists and crappy O(n) algorithms, because * (a) this is an infrequent control-plane operation and (b) n is small (max 64) */ struct efx_rss_context *efx_alloc_rss_context_entry(struct efx_nic *efx) { struct list_head *head = &efx->rss_context.list; struct efx_rss_context *ctx, *new; u32 id = 1; /* Don't use zero, that refers to the master RSS context */ WARN_ON(!mutex_is_locked(&efx->rss_lock)); /* Search for first gap in the numbering */ list_for_each_entry(ctx, head, list) { if (ctx->user_id != id) break; id++; /* Check for wrap. If this happens, we have nearly 2^32 * allocated RSS contexts, which seems unlikely. */ if (WARN_ON_ONCE(!id)) return NULL; } /* Create the new entry */ new = kmalloc(sizeof(struct efx_rss_context), GFP_KERNEL); if (!new) return NULL; new->context_id = EFX_EF10_RSS_CONTEXT_INVALID; new->rx_hash_udp_4tuple = false; /* Insert the new entry into the gap */ new->user_id = id; list_add_tail(&new->list, &ctx->list); return new; } struct efx_rss_context *efx_find_rss_context_entry(struct efx_nic *efx, u32 id) { struct list_head *head = &efx->rss_context.list; struct efx_rss_context *ctx; WARN_ON(!mutex_is_locked(&efx->rss_lock)); list_for_each_entry(ctx, head, list) if (ctx->user_id == id) return ctx; return NULL; } void efx_free_rss_context_entry(struct efx_rss_context *ctx) { list_del(&ctx->list); kfree(ctx); } /************************************************************************** * * PCI interface * **************************************************************************/ /* Main body of final NIC shutdown code * This is called only at module unload (or hotplug removal). */ static void efx_pci_remove_main(struct efx_nic *efx) { /* Flush reset_work. It can no longer be scheduled since we * are not READY. */ BUG_ON(efx->state == STATE_READY); efx_flush_reset_workqueue(efx); efx_disable_interrupts(efx); efx_clear_interrupt_affinity(efx); efx_nic_fini_interrupt(efx); efx_fini_port(efx); efx->type->fini(efx); efx_fini_napi(efx); efx_remove_all(efx); } /* Final NIC shutdown * This is called only at module unload (or hotplug removal). A PF can call * this on its VFs to ensure they are unbound first. */ static void efx_pci_remove(struct pci_dev *pci_dev) { struct efx_nic *efx; efx = pci_get_drvdata(pci_dev); if (!efx) return; /* Mark the NIC as fini, then stop the interface */ rtnl_lock(); efx_dissociate(efx); dev_close(efx->net_dev); efx_disable_interrupts(efx); efx->state = STATE_UNINIT; rtnl_unlock(); if (efx->type->sriov_fini) efx->type->sriov_fini(efx); efx_unregister_netdev(efx); efx_mtd_remove(efx); efx_pci_remove_main(efx); efx_fini_io(efx, efx->type->mem_bar(efx)); netif_dbg(efx, drv, efx->net_dev, "shutdown successful\n"); efx_fini_struct(efx); free_netdev(efx->net_dev); pci_disable_pcie_error_reporting(pci_dev); }; /* NIC VPD information * Called during probe to display the part number of the * installed NIC. VPD is potentially very large but this should * always appear within the first 512 bytes. */ #define SFC_VPD_LEN 512 static void efx_probe_vpd_strings(struct efx_nic *efx) { struct pci_dev *dev = efx->pci_dev; char vpd_data[SFC_VPD_LEN]; ssize_t vpd_size; int ro_start, ro_size, i, j; /* Get the vpd data from the device */ vpd_size = pci_read_vpd(dev, 0, sizeof(vpd_data), vpd_data); if (vpd_size <= 0) { netif_err(efx, drv, efx->net_dev, "Unable to read VPD\n"); return; } /* Get the Read only section */ ro_start = pci_vpd_find_tag(vpd_data, 0, vpd_size, PCI_VPD_LRDT_RO_DATA); if (ro_start < 0) { netif_err(efx, drv, efx->net_dev, "VPD Read-only not found\n"); return; } ro_size = pci_vpd_lrdt_size(&vpd_data[ro_start]); j = ro_size; i = ro_start + PCI_VPD_LRDT_TAG_SIZE; if (i + j > vpd_size) j = vpd_size - i; /* Get the Part number */ i = pci_vpd_find_info_keyword(vpd_data, i, j, "PN"); if (i < 0) { netif_err(efx, drv, efx->net_dev, "Part number not found\n"); return; } j = pci_vpd_info_field_size(&vpd_data[i]); i += PCI_VPD_INFO_FLD_HDR_SIZE; if (i + j > vpd_size) { netif_err(efx, drv, efx->net_dev, "Incomplete part number\n"); return; } netif_info(efx, drv, efx->net_dev, "Part Number : %.*s\n", j, &vpd_data[i]); i = ro_start + PCI_VPD_LRDT_TAG_SIZE; j = ro_size; i = pci_vpd_find_info_keyword(vpd_data, i, j, "SN"); if (i < 0) { netif_err(efx, drv, efx->net_dev, "Serial number not found\n"); return; } j = pci_vpd_info_field_size(&vpd_data[i]); i += PCI_VPD_INFO_FLD_HDR_SIZE; if (i + j > vpd_size) { netif_err(efx, drv, efx->net_dev, "Incomplete serial number\n"); return; } efx->vpd_sn = kmalloc(j + 1, GFP_KERNEL); if (!efx->vpd_sn) return; snprintf(efx->vpd_sn, j + 1, "%s", &vpd_data[i]); } /* Main body of NIC initialisation * This is called at module load (or hotplug insertion, theoretically). */ static int efx_pci_probe_main(struct efx_nic *efx) { int rc; /* Do start-of-day initialisation */ rc = efx_probe_all(efx); if (rc) goto fail1; efx_init_napi(efx); down_write(&efx->filter_sem); rc = efx->type->init(efx); up_write(&efx->filter_sem); if (rc) { netif_err(efx, probe, efx->net_dev, "failed to initialise NIC\n"); goto fail3; } rc = efx_init_port(efx); if (rc) { netif_err(efx, probe, efx->net_dev, "failed to initialise port\n"); goto fail4; } rc = efx_nic_init_interrupt(efx); if (rc) goto fail5; efx_set_interrupt_affinity(efx); rc = efx_enable_interrupts(efx); if (rc) goto fail6; return 0; fail6: efx_clear_interrupt_affinity(efx); efx_nic_fini_interrupt(efx); fail5: efx_fini_port(efx); fail4: efx->type->fini(efx); fail3: efx_fini_napi(efx); efx_remove_all(efx); fail1: return rc; } static int efx_pci_probe_post_io(struct efx_nic *efx) { struct net_device *net_dev = efx->net_dev; int rc = efx_pci_probe_main(efx); if (rc) return rc; if (efx->type->sriov_init) { rc = efx->type->sriov_init(efx); if (rc) netif_err(efx, probe, efx->net_dev, "SR-IOV can't be enabled rc %d\n", rc); } /* Determine netdevice features */ net_dev->features |= (efx->type->offload_features | NETIF_F_SG | NETIF_F_TSO | NETIF_F_RXCSUM | NETIF_F_RXALL); if (efx->type->offload_features & (NETIF_F_IPV6_CSUM | NETIF_F_HW_CSUM)) net_dev->features |= NETIF_F_TSO6; /* Check whether device supports TSO */ if (!efx->type->tso_versions || !efx->type->tso_versions(efx)) net_dev->features &= ~NETIF_F_ALL_TSO; /* Mask for features that also apply to VLAN devices */ net_dev->vlan_features |= (NETIF_F_HW_CSUM | NETIF_F_SG | NETIF_F_HIGHDMA | NETIF_F_ALL_TSO | NETIF_F_RXCSUM); net_dev->hw_features |= net_dev->features & ~efx->fixed_features; /* Disable receiving frames with bad FCS, by default. */ net_dev->features &= ~NETIF_F_RXALL; /* Disable VLAN filtering by default. It may be enforced if * the feature is fixed (i.e. VLAN filters are required to * receive VLAN tagged packets due to vPort restrictions). */ net_dev->features &= ~NETIF_F_HW_VLAN_CTAG_FILTER; net_dev->features |= efx->fixed_features; rc = efx_register_netdev(efx); if (!rc) return 0; efx_pci_remove_main(efx); return rc; } /* NIC initialisation * * This is called at module load (or hotplug insertion, * theoretically). It sets up PCI mappings, resets the NIC, * sets up and registers the network devices with the kernel and hooks * the interrupt service routine. It does not prepare the device for * transmission; this is left to the first time one of the network * interfaces is brought up (i.e. efx_net_open). */ static int efx_pci_probe(struct pci_dev *pci_dev, const struct pci_device_id *entry) { struct net_device *net_dev; struct efx_nic *efx; int rc; /* Allocate and initialise a struct net_device and struct efx_nic */ net_dev = alloc_etherdev_mqs(sizeof(*efx), EFX_MAX_CORE_TX_QUEUES, EFX_MAX_RX_QUEUES); if (!net_dev) return -ENOMEM; efx = netdev_priv(net_dev); efx->type = (const struct efx_nic_type *) entry->driver_data; efx->fixed_features |= NETIF_F_HIGHDMA; pci_set_drvdata(pci_dev, efx); SET_NETDEV_DEV(net_dev, &pci_dev->dev); rc = efx_init_struct(efx, pci_dev, net_dev); if (rc) goto fail1; netif_info(efx, probe, efx->net_dev, "Solarflare NIC detected\n"); if (!efx->type->is_vf) efx_probe_vpd_strings(efx); /* Set up basic I/O (BAR mappings etc) */ rc = efx_init_io(efx, efx->type->mem_bar(efx), efx->type->max_dma_mask, efx->type->mem_map_size(efx)); if (rc) goto fail2; rc = efx_pci_probe_post_io(efx); if (rc) { /* On failure, retry once immediately. * If we aborted probe due to a scheduled reset, dismiss it. */ efx->reset_pending = 0; rc = efx_pci_probe_post_io(efx); if (rc) { /* On another failure, retry once more * after a 50-305ms delay. */ unsigned char r; get_random_bytes(&r, 1); msleep((unsigned int)r + 50); efx->reset_pending = 0; rc = efx_pci_probe_post_io(efx); } } if (rc) goto fail3; netif_dbg(efx, probe, efx->net_dev, "initialisation successful\n"); /* Try to create MTDs, but allow this to fail */ rtnl_lock(); rc = efx_mtd_probe(efx); rtnl_unlock(); if (rc && rc != -EPERM) netif_warn(efx, probe, efx->net_dev, "failed to create MTDs (%d)\n", rc); (void)pci_enable_pcie_error_reporting(pci_dev); if (efx->type->udp_tnl_push_ports) efx->type->udp_tnl_push_ports(efx); return 0; fail3: efx_fini_io(efx, efx->type->mem_bar(efx)); fail2: efx_fini_struct(efx); fail1: WARN_ON(rc > 0); netif_dbg(efx, drv, efx->net_dev, "initialisation failed. rc=%d\n", rc); free_netdev(net_dev); return rc; } /* efx_pci_sriov_configure returns the actual number of Virtual Functions * enabled on success */ #ifdef CONFIG_SFC_SRIOV static int efx_pci_sriov_configure(struct pci_dev *dev, int num_vfs) { int rc; struct efx_nic *efx = pci_get_drvdata(dev); if (efx->type->sriov_configure) { rc = efx->type->sriov_configure(efx, num_vfs); if (rc) return rc; else return num_vfs; } else return -EOPNOTSUPP; } #endif static int efx_pm_freeze(struct device *dev) { struct efx_nic *efx = dev_get_drvdata(dev); rtnl_lock(); if (efx->state != STATE_DISABLED) { efx->state = STATE_UNINIT; efx_device_detach_sync(efx); efx_stop_all(efx); efx_disable_interrupts(efx); } rtnl_unlock(); return 0; } static int efx_pm_thaw(struct device *dev) { int rc; struct efx_nic *efx = dev_get_drvdata(dev); rtnl_lock(); if (efx->state != STATE_DISABLED) { rc = efx_enable_interrupts(efx); if (rc) goto fail; mutex_lock(&efx->mac_lock); efx->phy_op->reconfigure(efx); mutex_unlock(&efx->mac_lock); efx_start_all(efx); efx_device_attach_if_not_resetting(efx); efx->state = STATE_READY; efx->type->resume_wol(efx); } rtnl_unlock(); /* Reschedule any quenched resets scheduled during efx_pm_freeze() */ efx_queue_reset_work(efx); return 0; fail: rtnl_unlock(); return rc; } static int efx_pm_poweroff(struct device *dev) { struct pci_dev *pci_dev = to_pci_dev(dev); struct efx_nic *efx = pci_get_drvdata(pci_dev); efx->type->fini(efx); efx->reset_pending = 0; pci_save_state(pci_dev); return pci_set_power_state(pci_dev, PCI_D3hot); } /* Used for both resume and restore */ static int efx_pm_resume(struct device *dev) { struct pci_dev *pci_dev = to_pci_dev(dev); struct efx_nic *efx = pci_get_drvdata(pci_dev); int rc; rc = pci_set_power_state(pci_dev, PCI_D0); if (rc) return rc; pci_restore_state(pci_dev); rc = pci_enable_device(pci_dev); if (rc) return rc; pci_set_master(efx->pci_dev); rc = efx->type->reset(efx, RESET_TYPE_ALL); if (rc) return rc; down_write(&efx->filter_sem); rc = efx->type->init(efx); up_write(&efx->filter_sem); if (rc) return rc; rc = efx_pm_thaw(dev); return rc; } static int efx_pm_suspend(struct device *dev) { int rc; efx_pm_freeze(dev); rc = efx_pm_poweroff(dev); if (rc) efx_pm_resume(dev); return rc; } static const struct dev_pm_ops efx_pm_ops = { .suspend = efx_pm_suspend, .resume = efx_pm_resume, .freeze = efx_pm_freeze, .thaw = efx_pm_thaw, .poweroff = efx_pm_poweroff, .restore = efx_pm_resume, }; /* A PCI error affecting this device was detected. * At this point MMIO and DMA may be disabled. * Stop the software path and request a slot reset. */ static pci_ers_result_t efx_io_error_detected(struct pci_dev *pdev, enum pci_channel_state state) { pci_ers_result_t status = PCI_ERS_RESULT_RECOVERED; struct efx_nic *efx = pci_get_drvdata(pdev); if (state == pci_channel_io_perm_failure) return PCI_ERS_RESULT_DISCONNECT; rtnl_lock(); if (efx->state != STATE_DISABLED) { efx->state = STATE_RECOVERY; efx->reset_pending = 0; efx_device_detach_sync(efx); efx_stop_all(efx); efx_disable_interrupts(efx); status = PCI_ERS_RESULT_NEED_RESET; } else { /* If the interface is disabled we don't want to do anything * with it. */ status = PCI_ERS_RESULT_RECOVERED; } rtnl_unlock(); pci_disable_device(pdev); return status; } /* Fake a successful reset, which will be performed later in efx_io_resume. */ static pci_ers_result_t efx_io_slot_reset(struct pci_dev *pdev) { struct efx_nic *efx = pci_get_drvdata(pdev); pci_ers_result_t status = PCI_ERS_RESULT_RECOVERED; if (pci_enable_device(pdev)) { netif_err(efx, hw, efx->net_dev, "Cannot re-enable PCI device after reset.\n"); status = PCI_ERS_RESULT_DISCONNECT; } return status; } /* Perform the actual reset and resume I/O operations. */ static void efx_io_resume(struct pci_dev *pdev) { struct efx_nic *efx = pci_get_drvdata(pdev); int rc; rtnl_lock(); if (efx->state == STATE_DISABLED) goto out; rc = efx_reset(efx, RESET_TYPE_ALL); if (rc) { netif_err(efx, hw, efx->net_dev, "efx_reset failed after PCI error (%d)\n", rc); } else { efx->state = STATE_READY; netif_dbg(efx, hw, efx->net_dev, "Done resetting and resuming IO after PCI error.\n"); } out: rtnl_unlock(); } /* For simplicity and reliability, we always require a slot reset and try to * reset the hardware when a pci error affecting the device is detected. * We leave both the link_reset and mmio_enabled callback unimplemented: * with our request for slot reset the mmio_enabled callback will never be * called, and the link_reset callback is not used by AER or EEH mechanisms. */ static const struct pci_error_handlers efx_err_handlers = { .error_detected = efx_io_error_detected, .slot_reset = efx_io_slot_reset, .resume = efx_io_resume, }; static struct pci_driver efx_pci_driver = { .name = KBUILD_MODNAME, .id_table = efx_pci_table, .probe = efx_pci_probe, .remove = efx_pci_remove, .driver.pm = &efx_pm_ops, .err_handler = &efx_err_handlers, #ifdef CONFIG_SFC_SRIOV .sriov_configure = efx_pci_sriov_configure, #endif }; /************************************************************************** * * Kernel module interface * *************************************************************************/ static int __init efx_init_module(void) { int rc; printk(KERN_INFO "Solarflare NET driver v" EFX_DRIVER_VERSION "\n"); rc = register_netdevice_notifier(&efx_netdev_notifier); if (rc) goto err_notifier; #ifdef CONFIG_SFC_SRIOV rc = efx_init_sriov(); if (rc) goto err_sriov; #endif rc = efx_create_reset_workqueue(); if (rc) goto err_reset; rc = pci_register_driver(&efx_pci_driver); if (rc < 0) goto err_pci; return 0; err_pci: efx_destroy_reset_workqueue(); err_reset: #ifdef CONFIG_SFC_SRIOV efx_fini_sriov(); err_sriov: #endif unregister_netdevice_notifier(&efx_netdev_notifier); err_notifier: return rc; } static void __exit efx_exit_module(void) { printk(KERN_INFO "Solarflare NET driver unloading\n"); pci_unregister_driver(&efx_pci_driver); efx_destroy_reset_workqueue(); #ifdef CONFIG_SFC_SRIOV efx_fini_sriov(); #endif unregister_netdevice_notifier(&efx_netdev_notifier); } module_init(efx_init_module); module_exit(efx_exit_module); MODULE_AUTHOR("Solarflare Communications and " "Michael Brown "); MODULE_DESCRIPTION("Solarflare network driver"); MODULE_LICENSE("GPL"); MODULE_DEVICE_TABLE(pci, efx_pci_table); MODULE_VERSION(EFX_DRIVER_VERSION);