X-Git-Url: http://git.droids-corp.org/?a=blobdiff_plain;f=drivers%2Fnet%2Fsfc%2Fsfc_ethdev.c;h=0297e791ca3b18cc47e3f43cd472d78ff9f59d54;hb=96a46ead5b9c3f544703237eb69ba6e2df603f5a;hp=165776b6520be1bee97e4da6cdd14152889cd3ec;hpb=0cb551b690c1c52de3e35ea67c502ddf4477a62b;p=dpdk.git diff --git a/drivers/net/sfc/sfc_ethdev.c b/drivers/net/sfc/sfc_ethdev.c index 165776b652..0297e791ca 100644 --- a/drivers/net/sfc/sfc_ethdev.c +++ b/drivers/net/sfc/sfc_ethdev.c @@ -1,6 +1,6 @@ /* SPDX-License-Identifier: BSD-3-Clause * - * Copyright(c) 2019-2020 Xilinx, Inc. + * Copyright(c) 2019-2021 Xilinx, Inc. * Copyright(c) 2016-2019 Solarflare Communications Inc. * * This software was jointly developed between OKTET Labs (under contract @@ -8,8 +8,8 @@ */ #include -#include -#include +#include +#include #include #include #include @@ -28,6 +28,12 @@ #include "sfc_flow.h" #include "sfc_dp.h" #include "sfc_dp_rx.h" +#include "sfc_repr.h" +#include "sfc_sw_stats.h" +#include "sfc_switch.h" + +#define SFC_XSTAT_ID_INVALID_VAL UINT64_MAX +#define SFC_XSTAT_ID_INVALID_NAME '\0' uint32_t sfc_logtype_driver; @@ -46,14 +52,6 @@ sfc_fw_version_get(struct rte_eth_dev *dev, char *fw_version, size_t fw_size) int ret; int rc; - /* - * Return value of the callback is likely supposed to be - * equal to or greater than 0, nevertheless, if an error - * occurs, it will be desirable to pass it to the caller - */ - if ((fw_version == NULL) || (fw_size == 0)) - return -EINVAL; - rc = efx_nic_get_fw_version(sa->nic, &enfi); if (rc != 0) return -rc; @@ -93,6 +91,7 @@ sfc_dev_infos_get(struct rte_eth_dev *dev, struct rte_eth_dev_info *dev_info) struct sfc_adapter_shared *sas = sfc_adapter_shared_by_eth_dev(dev); struct sfc_adapter *sa = sfc_adapter_by_eth_dev(dev); struct sfc_rss *rss = &sas->rss; + struct sfc_mae *mae = &sa->mae; uint64_t txq_offloads_def = 0; sfc_log_init(sa, "entry"); @@ -187,6 +186,12 @@ sfc_dev_infos_get(struct rte_eth_dev *dev, struct rte_eth_dev_info *dev_info) dev_info->dev_capa = RTE_ETH_DEV_CAPA_RUNTIME_RX_QUEUE_SETUP | RTE_ETH_DEV_CAPA_RUNTIME_TX_QUEUE_SETUP; + if (mae->status == SFC_MAE_STATUS_SUPPORTED) { + dev_info->switch_info.name = dev->device->driver->name; + dev_info->switch_info.domain_id = mae->switch_domain_id; + dev_info->switch_info.port_id = mae->switch_port_id; + } + return 0; } @@ -210,9 +215,9 @@ sfc_dev_configure(struct rte_eth_dev *dev) sfc_adapter_lock(sa); switch (sa->state) { - case SFC_ADAPTER_CONFIGURED: + case SFC_ETHDEV_CONFIGURED: /* FALLTHROUGH */ - case SFC_ADAPTER_INITIALIZED: + case SFC_ETHDEV_INITIALIZED: rc = sfc_configure(sa); break; default: @@ -254,7 +259,7 @@ sfc_dev_link_update(struct rte_eth_dev *dev, int wait_to_complete) sfc_log_init(sa, "entry"); - if (sa->state != SFC_ADAPTER_STARTED) { + if (sa->state != SFC_ETHDEV_STARTED) { sfc_port_link_mode_to_info(EFX_LINK_UNKNOWN, ¤t_link); } else if (wait_to_complete) { efx_link_mode_t link_mode; @@ -276,7 +281,7 @@ sfc_dev_link_update(struct rte_eth_dev *dev, int wait_to_complete) return ret; } -static void +static int sfc_dev_stop(struct rte_eth_dev *dev) { struct sfc_adapter *sa = sfc_adapter_by_eth_dev(dev); @@ -288,6 +293,8 @@ sfc_dev_stop(struct rte_eth_dev *dev) sfc_adapter_unlock(sa); sfc_log_init(sa, "done"); + + return 0; } static int @@ -324,11 +331,7 @@ static void sfc_eth_dev_secondary_clear_ops(struct rte_eth_dev *dev) { free(dev->process_private); - dev->process_private = NULL; - dev->dev_ops = NULL; - dev->tx_pkt_prepare = NULL; - dev->tx_pkt_burst = NULL; - dev->rx_pkt_burst = NULL; + rte_eth_dev_release_port(dev); } static int @@ -343,17 +346,19 @@ sfc_dev_close(struct rte_eth_dev *dev) return 0; } + sfc_pre_detach(sa); + sfc_adapter_lock(sa); switch (sa->state) { - case SFC_ADAPTER_STARTED: + case SFC_ETHDEV_STARTED: sfc_stop(sa); - SFC_ASSERT(sa->state == SFC_ADAPTER_CONFIGURED); + SFC_ASSERT(sa->state == SFC_ETHDEV_CONFIGURED); /* FALLTHROUGH */ - case SFC_ADAPTER_CONFIGURED: + case SFC_ETHDEV_CONFIGURED: sfc_close(sa); - SFC_ASSERT(sa->state == SFC_ADAPTER_INITIALIZED); + SFC_ASSERT(sa->state == SFC_ETHDEV_INITIALIZED); /* FALLTHROUGH */ - case SFC_ADAPTER_INITIALIZED: + case SFC_ETHDEV_INITIALIZED: break; default: sfc_err(sa, "unexpected adapter state %u on close", sa->state); @@ -380,7 +385,6 @@ sfc_dev_close(struct rte_eth_dev *dev) /* Required for logging, so cleanup last */ sa->eth_dev = NULL; - dev->process_private = NULL; free(sa); return 0; @@ -410,7 +414,7 @@ sfc_dev_filter_set(struct rte_eth_dev *dev, enum sfc_dev_filter_mode mode, sfc_warn(sa, "the change is to be applied on the next " "start provided that isolated mode is " "disabled prior the next start"); - } else if ((sa->state == SFC_ADAPTER_STARTED) && + } else if ((sa->state == SFC_ETHDEV_STARTED) && ((rc = sfc_set_rx_mode(sa)) != 0)) { *toggle = !(enabled); sfc_warn(sa, "Failed to %s %s mode, rc = %d", @@ -467,26 +471,31 @@ sfc_dev_allmulti_disable(struct rte_eth_dev *dev) } static int -sfc_rx_queue_setup(struct rte_eth_dev *dev, uint16_t rx_queue_id, +sfc_rx_queue_setup(struct rte_eth_dev *dev, uint16_t ethdev_qid, uint16_t nb_rx_desc, unsigned int socket_id, const struct rte_eth_rxconf *rx_conf, struct rte_mempool *mb_pool) { struct sfc_adapter_shared *sas = sfc_adapter_shared_by_eth_dev(dev); struct sfc_adapter *sa = sfc_adapter_by_eth_dev(dev); + sfc_ethdev_qid_t sfc_ethdev_qid = ethdev_qid; + struct sfc_rxq_info *rxq_info; + sfc_sw_index_t sw_index; int rc; sfc_log_init(sa, "RxQ=%u nb_rx_desc=%u socket_id=%u", - rx_queue_id, nb_rx_desc, socket_id); + ethdev_qid, nb_rx_desc, socket_id); sfc_adapter_lock(sa); - rc = sfc_rx_qinit(sa, rx_queue_id, nb_rx_desc, socket_id, + sw_index = sfc_rxq_sw_index_by_ethdev_rx_qid(sas, sfc_ethdev_qid); + rc = sfc_rx_qinit(sa, sw_index, nb_rx_desc, socket_id, rx_conf, mb_pool); if (rc != 0) goto fail_rx_qinit; - dev->data->rx_queues[rx_queue_id] = sas->rxq_info[rx_queue_id].dp; + rxq_info = sfc_rxq_info_by_ethdev_qid(sas, sfc_ethdev_qid); + dev->data->rx_queues[ethdev_qid] = rxq_info->dp; sfc_adapter_unlock(sa); @@ -499,12 +508,12 @@ fail_rx_qinit: } static void -sfc_rx_queue_release(void *queue) +sfc_rx_queue_release(struct rte_eth_dev *dev, uint16_t qid) { - struct sfc_dp_rxq *dp_rxq = queue; + struct sfc_dp_rxq *dp_rxq = dev->data->rx_queues[qid]; struct sfc_rxq *rxq; struct sfc_adapter *sa; - unsigned int sw_index; + sfc_sw_index_t sw_index; if (dp_rxq == NULL) return; @@ -523,24 +532,28 @@ sfc_rx_queue_release(void *queue) } static int -sfc_tx_queue_setup(struct rte_eth_dev *dev, uint16_t tx_queue_id, +sfc_tx_queue_setup(struct rte_eth_dev *dev, uint16_t ethdev_qid, uint16_t nb_tx_desc, unsigned int socket_id, const struct rte_eth_txconf *tx_conf) { struct sfc_adapter_shared *sas = sfc_adapter_shared_by_eth_dev(dev); struct sfc_adapter *sa = sfc_adapter_by_eth_dev(dev); + struct sfc_txq_info *txq_info; + sfc_sw_index_t sw_index; int rc; sfc_log_init(sa, "TxQ = %u, nb_tx_desc = %u, socket_id = %u", - tx_queue_id, nb_tx_desc, socket_id); + ethdev_qid, nb_tx_desc, socket_id); sfc_adapter_lock(sa); - rc = sfc_tx_qinit(sa, tx_queue_id, nb_tx_desc, socket_id, tx_conf); + sw_index = sfc_txq_sw_index_by_ethdev_tx_qid(sas, ethdev_qid); + rc = sfc_tx_qinit(sa, sw_index, nb_tx_desc, socket_id, tx_conf); if (rc != 0) goto fail_tx_qinit; - dev->data->tx_queues[tx_queue_id] = sas->txq_info[tx_queue_id].dp; + txq_info = sfc_txq_info_by_ethdev_qid(sas, ethdev_qid); + dev->data->tx_queues[ethdev_qid] = txq_info->dp; sfc_adapter_unlock(sa); return 0; @@ -552,11 +565,11 @@ fail_tx_qinit: } static void -sfc_tx_queue_release(void *queue) +sfc_tx_queue_release(struct rte_eth_dev *dev, uint16_t qid) { - struct sfc_dp_txq *dp_txq = queue; + struct sfc_dp_txq *dp_txq = dev->data->tx_queues[qid]; struct sfc_txq *txq; - unsigned int sw_index; + sfc_sw_index_t sw_index; struct sfc_adapter *sa; if (dp_txq == NULL) @@ -577,6 +590,60 @@ sfc_tx_queue_release(void *queue) sfc_adapter_unlock(sa); } +static void +sfc_stats_get_dp_rx(struct sfc_adapter *sa, uint64_t *pkts, uint64_t *bytes) +{ + struct sfc_adapter_shared *sas = sfc_sa2shared(sa); + uint64_t pkts_sum = 0; + uint64_t bytes_sum = 0; + unsigned int i; + + for (i = 0; i < sas->ethdev_rxq_count; ++i) { + struct sfc_rxq_info *rxq_info; + + rxq_info = sfc_rxq_info_by_ethdev_qid(sas, i); + if (rxq_info->state & SFC_RXQ_INITIALIZED) { + union sfc_pkts_bytes qstats; + + sfc_pkts_bytes_get(&rxq_info->dp->dpq.stats, &qstats); + pkts_sum += qstats.pkts - + sa->sw_stats.reset_rx_pkts[i]; + bytes_sum += qstats.bytes - + sa->sw_stats.reset_rx_bytes[i]; + } + } + + *pkts = pkts_sum; + *bytes = bytes_sum; +} + +static void +sfc_stats_get_dp_tx(struct sfc_adapter *sa, uint64_t *pkts, uint64_t *bytes) +{ + struct sfc_adapter_shared *sas = sfc_sa2shared(sa); + uint64_t pkts_sum = 0; + uint64_t bytes_sum = 0; + unsigned int i; + + for (i = 0; i < sas->ethdev_txq_count; ++i) { + struct sfc_txq_info *txq_info; + + txq_info = sfc_txq_info_by_ethdev_qid(sas, i); + if (txq_info->state & SFC_TXQ_INITIALIZED) { + union sfc_pkts_bytes qstats; + + sfc_pkts_bytes_get(&txq_info->dp->dpq.stats, &qstats); + pkts_sum += qstats.pkts - + sa->sw_stats.reset_tx_pkts[i]; + bytes_sum += qstats.bytes - + sa->sw_stats.reset_tx_bytes[i]; + } + } + + *pkts = pkts_sum; + *bytes = bytes_sum; +} + /* * Some statistics are computed as A - B where A and B each increase * monotonically with some hardware counter(s) and the counters are read @@ -603,14 +670,22 @@ sfc_update_diff_stat(uint64_t *stat, uint64_t newval) static int sfc_stats_get(struct rte_eth_dev *dev, struct rte_eth_stats *stats) { + const struct sfc_adapter_priv *sap = sfc_adapter_priv_by_eth_dev(dev); + bool have_dp_rx_stats = sap->dp_rx->features & SFC_DP_RX_FEAT_STATS; + bool have_dp_tx_stats = sap->dp_tx->features & SFC_DP_TX_FEAT_STATS; struct sfc_adapter *sa = sfc_adapter_by_eth_dev(dev); struct sfc_port *port = &sa->port; uint64_t *mac_stats; int ret; - rte_spinlock_lock(&port->mac_stats_lock); + sfc_adapter_lock(sa); + + if (have_dp_rx_stats) + sfc_stats_get_dp_rx(sa, &stats->ipackets, &stats->ibytes); + if (have_dp_tx_stats) + sfc_stats_get_dp_tx(sa, &stats->opackets, &stats->obytes); - ret = sfc_port_update_mac_stats(sa); + ret = sfc_port_update_mac_stats(sa, B_FALSE); if (ret != 0) goto unlock; @@ -618,28 +693,41 @@ sfc_stats_get(struct rte_eth_dev *dev, struct rte_eth_stats *stats) if (EFX_MAC_STAT_SUPPORTED(port->mac_stats_mask, EFX_MAC_VADAPTER_RX_UNICAST_PACKETS)) { - stats->ipackets = - mac_stats[EFX_MAC_VADAPTER_RX_UNICAST_PACKETS] + - mac_stats[EFX_MAC_VADAPTER_RX_MULTICAST_PACKETS] + - mac_stats[EFX_MAC_VADAPTER_RX_BROADCAST_PACKETS]; - stats->opackets = - mac_stats[EFX_MAC_VADAPTER_TX_UNICAST_PACKETS] + - mac_stats[EFX_MAC_VADAPTER_TX_MULTICAST_PACKETS] + - mac_stats[EFX_MAC_VADAPTER_TX_BROADCAST_PACKETS]; - stats->ibytes = - mac_stats[EFX_MAC_VADAPTER_RX_UNICAST_BYTES] + - mac_stats[EFX_MAC_VADAPTER_RX_MULTICAST_BYTES] + - mac_stats[EFX_MAC_VADAPTER_RX_BROADCAST_BYTES]; - stats->obytes = - mac_stats[EFX_MAC_VADAPTER_TX_UNICAST_BYTES] + - mac_stats[EFX_MAC_VADAPTER_TX_MULTICAST_BYTES] + - mac_stats[EFX_MAC_VADAPTER_TX_BROADCAST_BYTES]; + if (!have_dp_rx_stats) { + stats->ipackets = + mac_stats[EFX_MAC_VADAPTER_RX_UNICAST_PACKETS] + + mac_stats[EFX_MAC_VADAPTER_RX_MULTICAST_PACKETS] + + mac_stats[EFX_MAC_VADAPTER_RX_BROADCAST_PACKETS]; + stats->ibytes = + mac_stats[EFX_MAC_VADAPTER_RX_UNICAST_BYTES] + + mac_stats[EFX_MAC_VADAPTER_RX_MULTICAST_BYTES] + + mac_stats[EFX_MAC_VADAPTER_RX_BROADCAST_BYTES]; + + /* CRC is included in these stats, but shouldn't be */ + stats->ibytes -= stats->ipackets * RTE_ETHER_CRC_LEN; + } + if (!have_dp_tx_stats) { + stats->opackets = + mac_stats[EFX_MAC_VADAPTER_TX_UNICAST_PACKETS] + + mac_stats[EFX_MAC_VADAPTER_TX_MULTICAST_PACKETS] + + mac_stats[EFX_MAC_VADAPTER_TX_BROADCAST_PACKETS]; + stats->obytes = + mac_stats[EFX_MAC_VADAPTER_TX_UNICAST_BYTES] + + mac_stats[EFX_MAC_VADAPTER_TX_MULTICAST_BYTES] + + mac_stats[EFX_MAC_VADAPTER_TX_BROADCAST_BYTES]; + + /* CRC is included in these stats, but shouldn't be */ + stats->obytes -= stats->opackets * RTE_ETHER_CRC_LEN; + } stats->imissed = mac_stats[EFX_MAC_VADAPTER_RX_BAD_PACKETS]; stats->oerrors = mac_stats[EFX_MAC_VADAPTER_TX_BAD_PACKETS]; } else { - stats->opackets = mac_stats[EFX_MAC_TX_PKTS]; - stats->ibytes = mac_stats[EFX_MAC_RX_OCTETS]; - stats->obytes = mac_stats[EFX_MAC_TX_OCTETS]; + if (!have_dp_tx_stats) { + stats->opackets = mac_stats[EFX_MAC_TX_PKTS]; + stats->obytes = mac_stats[EFX_MAC_TX_OCTETS] - + mac_stats[EFX_MAC_TX_PKTS] * RTE_ETHER_CRC_LEN; + } + /* * Take into account stats which are whenever supported * on EF10. If some stat is not supported by current @@ -663,16 +751,20 @@ sfc_stats_get(struct rte_eth_dev *dev, struct rte_eth_stats *stats) mac_stats[EFX_MAC_RX_JABBER_PKTS]; /* no oerrors counters supported on EF10 */ - /* Exclude missed, errors and pauses from Rx packets */ - sfc_update_diff_stat(&port->ipackets, - mac_stats[EFX_MAC_RX_PKTS] - - mac_stats[EFX_MAC_RX_PAUSE_PKTS] - - stats->imissed - stats->ierrors); - stats->ipackets = port->ipackets; + if (!have_dp_rx_stats) { + /* Exclude missed, errors and pauses from Rx packets */ + sfc_update_diff_stat(&port->ipackets, + mac_stats[EFX_MAC_RX_PKTS] - + mac_stats[EFX_MAC_RX_PAUSE_PKTS] - + stats->imissed - stats->ierrors); + stats->ipackets = port->ipackets; + stats->ibytes = mac_stats[EFX_MAC_RX_OCTETS] - + mac_stats[EFX_MAC_RX_PKTS] * RTE_ETHER_CRC_LEN; + } } unlock: - rte_spinlock_unlock(&port->mac_stats_lock); + sfc_adapter_unlock(sa); SFC_ASSERT(ret >= 0); return -ret; } @@ -684,12 +776,15 @@ sfc_stats_reset(struct rte_eth_dev *dev) struct sfc_port *port = &sa->port; int rc; - if (sa->state != SFC_ADAPTER_STARTED) { + sfc_adapter_lock(sa); + + if (sa->state != SFC_ETHDEV_STARTED) { /* * The operation cannot be done if port is not started; it * will be scheduled to be done during the next port start */ port->mac_stats_reset_pending = B_TRUE; + sfc_adapter_unlock(sa); return 0; } @@ -697,46 +792,49 @@ sfc_stats_reset(struct rte_eth_dev *dev) if (rc != 0) sfc_err(sa, "failed to reset statistics (rc = %d)", rc); + sfc_sw_xstats_reset(sa); + + sfc_adapter_unlock(sa); + SFC_ASSERT(rc >= 0); return -rc; } +static unsigned int +sfc_xstats_get_nb_supported(struct sfc_adapter *sa) +{ + struct sfc_port *port = &sa->port; + unsigned int nb_supported; + + sfc_adapter_lock(sa); + nb_supported = port->mac_stats_nb_supported + + sfc_sw_xstats_get_nb_supported(sa); + sfc_adapter_unlock(sa); + + return nb_supported; +} + static int sfc_xstats_get(struct rte_eth_dev *dev, struct rte_eth_xstat *xstats, unsigned int xstats_count) { struct sfc_adapter *sa = sfc_adapter_by_eth_dev(dev); - struct sfc_port *port = &sa->port; - uint64_t *mac_stats; + unsigned int nb_written = 0; + unsigned int nb_supported = 0; int rc; - unsigned int i; - int nstats = 0; - - rte_spinlock_lock(&port->mac_stats_lock); - - rc = sfc_port_update_mac_stats(sa); - if (rc != 0) { - SFC_ASSERT(rc > 0); - nstats = -rc; - goto unlock; - } - mac_stats = port->mac_stats_buf; + if (unlikely(xstats == NULL)) + return sfc_xstats_get_nb_supported(sa); - for (i = 0; i < EFX_MAC_NSTATS; ++i) { - if (EFX_MAC_STAT_SUPPORTED(port->mac_stats_mask, i)) { - if (xstats != NULL && nstats < (int)xstats_count) { - xstats[nstats].id = nstats; - xstats[nstats].value = mac_stats[i]; - } - nstats++; - } - } + rc = sfc_port_get_mac_stats(sa, xstats, xstats_count, &nb_written); + if (rc < 0) + return rc; -unlock: - rte_spinlock_unlock(&port->mac_stats_lock); + nb_supported = rc; + sfc_sw_xstats_get_vals(sa, xstats, xstats_count, &nb_written, + &nb_supported); - return nstats; + return nb_supported; } static int @@ -748,17 +846,31 @@ sfc_xstats_get_names(struct rte_eth_dev *dev, struct sfc_port *port = &sa->port; unsigned int i; unsigned int nstats = 0; + unsigned int nb_written = 0; + int ret; + + if (unlikely(xstats_names == NULL)) + return sfc_xstats_get_nb_supported(sa); for (i = 0; i < EFX_MAC_NSTATS; ++i) { if (EFX_MAC_STAT_SUPPORTED(port->mac_stats_mask, i)) { - if (xstats_names != NULL && nstats < xstats_count) + if (nstats < xstats_count) { strlcpy(xstats_names[nstats].name, efx_mac_stat_name(sa->nic, i), sizeof(xstats_names[0].name)); + nb_written++; + } nstats++; } } + ret = sfc_sw_xstats_get_names(sa, xstats_names, xstats_count, + &nb_written, &nstats); + if (ret != 0) { + SFC_ASSERT(ret < 0); + return ret; + } + return nstats; } @@ -768,76 +880,94 @@ sfc_xstats_get_by_id(struct rte_eth_dev *dev, const uint64_t *ids, { struct sfc_adapter *sa = sfc_adapter_by_eth_dev(dev); struct sfc_port *port = &sa->port; - uint64_t *mac_stats; - unsigned int nb_supported = 0; - unsigned int nb_written = 0; + unsigned int nb_supported; unsigned int i; - int ret; int rc; - if (unlikely(values == NULL) || - unlikely((ids == NULL) && (n < port->mac_stats_nb_supported))) - return port->mac_stats_nb_supported; - - rte_spinlock_lock(&port->mac_stats_lock); - - rc = sfc_port_update_mac_stats(sa); - if (rc != 0) { - SFC_ASSERT(rc > 0); - ret = -rc; - goto unlock; - } + if (unlikely(ids == NULL || values == NULL)) + return -EINVAL; - mac_stats = port->mac_stats_buf; + /* + * Values array could be filled in nonsequential order. Fill values with + * constant indicating invalid ID first. + */ + for (i = 0; i < n; i++) + values[i] = SFC_XSTAT_ID_INVALID_VAL; - for (i = 0; (i < EFX_MAC_NSTATS) && (nb_written < n); ++i) { - if (!EFX_MAC_STAT_SUPPORTED(port->mac_stats_mask, i)) - continue; + rc = sfc_port_get_mac_stats_by_id(sa, ids, values, n); + if (rc != 0) + return rc; - if ((ids == NULL) || (ids[nb_written] == nb_supported)) - values[nb_written++] = mac_stats[i]; + nb_supported = port->mac_stats_nb_supported; + sfc_sw_xstats_get_vals_by_id(sa, ids, values, n, &nb_supported); - ++nb_supported; + /* Return number of written stats before invalid ID is encountered. */ + for (i = 0; i < n; i++) { + if (values[i] == SFC_XSTAT_ID_INVALID_VAL) + return i; } - ret = nb_written; - -unlock: - rte_spinlock_unlock(&port->mac_stats_lock); - - return ret; + return n; } static int sfc_xstats_get_names_by_id(struct rte_eth_dev *dev, + const uint64_t *ids, struct rte_eth_xstat_name *xstats_names, - const uint64_t *ids, unsigned int size) + unsigned int size) { struct sfc_adapter *sa = sfc_adapter_by_eth_dev(dev); struct sfc_port *port = &sa->port; - unsigned int nb_supported = 0; - unsigned int nb_written = 0; + unsigned int nb_supported; unsigned int i; + int ret; - if (unlikely(xstats_names == NULL) || - unlikely((ids == NULL) && (size < port->mac_stats_nb_supported))) - return port->mac_stats_nb_supported; + if (unlikely(xstats_names == NULL && ids != NULL) || + unlikely(xstats_names != NULL && ids == NULL)) + return -EINVAL; - for (i = 0; (i < EFX_MAC_NSTATS) && (nb_written < size); ++i) { - if (!EFX_MAC_STAT_SUPPORTED(port->mac_stats_mask, i)) - continue; + if (unlikely(xstats_names == NULL && ids == NULL)) + return sfc_xstats_get_nb_supported(sa); - if ((ids == NULL) || (ids[nb_written] == nb_supported)) { - char *name = xstats_names[nb_written++].name; + /* + * Names array could be filled in nonsequential order. Fill names with + * string indicating invalid ID first. + */ + for (i = 0; i < size; i++) + xstats_names[i].name[0] = SFC_XSTAT_ID_INVALID_NAME; - strlcpy(name, efx_mac_stat_name(sa->nic, i), + sfc_adapter_lock(sa); + + SFC_ASSERT(port->mac_stats_nb_supported <= + RTE_DIM(port->mac_stats_by_id)); + + for (i = 0; i < size; i++) { + if (ids[i] < port->mac_stats_nb_supported) { + strlcpy(xstats_names[i].name, + efx_mac_stat_name(sa->nic, + port->mac_stats_by_id[ids[i]]), sizeof(xstats_names[0].name)); } + } + + nb_supported = port->mac_stats_nb_supported; + + sfc_adapter_unlock(sa); - ++nb_supported; + ret = sfc_sw_xstats_get_names_by_id(sa, ids, xstats_names, size, + &nb_supported); + if (ret != 0) { + SFC_ASSERT(ret < 0); + return ret; + } + + /* Return number of written names before invalid ID is encountered. */ + for (i = 0; i < size; i++) { + if (xstats_names[i].name[0] == SFC_XSTAT_ID_INVALID_NAME) + return i; } - return nb_written; + return size; } static int @@ -850,7 +980,7 @@ sfc_flow_ctrl_get(struct rte_eth_dev *dev, struct rte_eth_fc_conf *fc_conf) sfc_adapter_lock(sa); - if (sa->state == SFC_ADAPTER_STARTED) + if (sa->state == SFC_ETHDEV_STARTED) efx_mac_fcntl_get(sa->nic, &wanted_fc, &link_fc); else link_fc = sa->port.flow_ctrl; @@ -916,7 +1046,7 @@ sfc_flow_ctrl_set(struct rte_eth_dev *dev, struct rte_eth_fc_conf *fc_conf) sfc_adapter_lock(sa); - if (sa->state == SFC_ADAPTER_STARTED) { + if (sa->state == SFC_ETHDEV_STARTED) { rc = efx_mac_fcntl_set(sa->nic, fcntl, fc_conf->autoneg); if (rc != 0) goto fail_mac_fcntl_set; @@ -996,7 +1126,7 @@ sfc_dev_set_mtu(struct rte_eth_dev *dev, uint16_t mtu) goto fail_check_scatter; if (pdu != sa->port.pdu) { - if (sa->state == SFC_ADAPTER_STARTED) { + if (sa->state == SFC_ETHDEV_STARTED) { sfc_stop(sa); old_pdu = sa->port.pdu; @@ -1013,7 +1143,7 @@ sfc_dev_set_mtu(struct rte_eth_dev *dev, uint16_t mtu) * The driver does not use it, but other PMDs update jumbo frame * flag and max_rx_pkt_len when MTU is set. */ - if (mtu > RTE_ETHER_MAX_LEN) { + if (mtu > RTE_ETHER_MTU) { struct rte_eth_rxmode *rxmode = &dev->data->dev_conf.rxmode; rxmode->offloads |= DEV_RX_OFFLOAD_JUMBO_FRAME; } @@ -1073,7 +1203,7 @@ sfc_mac_addr_set(struct rte_eth_dev *dev, struct rte_ether_addr *mac_addr) goto unlock; } - if (sa->state != SFC_ADAPTER_STARTED) { + if (sa->state != SFC_ETHDEV_STARTED) { sfc_notice(sa, "the port is not started"); sfc_notice(sa, "the new MAC address will be set on port start"); @@ -1160,7 +1290,7 @@ sfc_set_mc_addr_list(struct rte_eth_dev *dev, port->nb_mcast_addrs = nb_mc_addr; - if (sa->state != SFC_ADAPTER_STARTED) + if (sa->state != SFC_ETHDEV_STARTED) return 0; rc = efx_mac_multicast_list_set(sa->nic, port->mcast_addrs, @@ -1177,15 +1307,14 @@ sfc_set_mc_addr_list(struct rte_eth_dev *dev, * use any process-local pointers from the adapter data. */ static void -sfc_rx_queue_info_get(struct rte_eth_dev *dev, uint16_t rx_queue_id, +sfc_rx_queue_info_get(struct rte_eth_dev *dev, uint16_t ethdev_qid, struct rte_eth_rxq_info *qinfo) { struct sfc_adapter_shared *sas = sfc_adapter_shared_by_eth_dev(dev); + sfc_ethdev_qid_t sfc_ethdev_qid = ethdev_qid; struct sfc_rxq_info *rxq_info; - SFC_ASSERT(rx_queue_id < sas->rxq_count); - - rxq_info = &sas->rxq_info[rx_queue_id]; + rxq_info = sfc_rxq_info_by_ethdev_qid(sas, sfc_ethdev_qid); qinfo->mp = rxq_info->refill_mb_pool; qinfo->conf.rx_free_thresh = rxq_info->refill_threshold; @@ -1204,15 +1333,15 @@ sfc_rx_queue_info_get(struct rte_eth_dev *dev, uint16_t rx_queue_id, * use any process-local pointers from the adapter data. */ static void -sfc_tx_queue_info_get(struct rte_eth_dev *dev, uint16_t tx_queue_id, +sfc_tx_queue_info_get(struct rte_eth_dev *dev, uint16_t ethdev_qid, struct rte_eth_txq_info *qinfo) { struct sfc_adapter_shared *sas = sfc_adapter_shared_by_eth_dev(dev); struct sfc_txq_info *txq_info; - SFC_ASSERT(tx_queue_id < sas->txq_count); + SFC_ASSERT(ethdev_qid < sas->ethdev_txq_count); - txq_info = &sas->txq_info[tx_queue_id]; + txq_info = sfc_txq_info_by_ethdev_qid(sas, ethdev_qid); memset(qinfo, 0, sizeof(*qinfo)); @@ -1227,14 +1356,14 @@ sfc_tx_queue_info_get(struct rte_eth_dev *dev, uint16_t tx_queue_id, * use any process-local pointers from the adapter data. */ static uint32_t -sfc_rx_queue_count(struct rte_eth_dev *dev, uint16_t rx_queue_id) +sfc_rx_queue_count(struct rte_eth_dev *dev, uint16_t ethdev_qid) { const struct sfc_adapter_priv *sap = sfc_adapter_priv_by_eth_dev(dev); struct sfc_adapter_shared *sas = sfc_adapter_shared_by_eth_dev(dev); + sfc_ethdev_qid_t sfc_ethdev_qid = ethdev_qid; struct sfc_rxq_info *rxq_info; - SFC_ASSERT(rx_queue_id < sas->rxq_count); - rxq_info = &sas->rxq_info[rx_queue_id]; + rxq_info = sfc_rxq_info_by_ethdev_qid(sas, sfc_ethdev_qid); if ((rxq_info->state & SFC_RXQ_STARTED) == 0) return 0; @@ -1242,21 +1371,6 @@ sfc_rx_queue_count(struct rte_eth_dev *dev, uint16_t rx_queue_id) return sap->dp_rx->qdesc_npending(rxq_info->dp); } -/* - * The function is used by the secondary process as well. It must not - * use any process-local pointers from the adapter data. - */ -static int -sfc_rx_descriptor_done(void *queue, uint16_t offset) -{ - struct sfc_dp_rxq *dp_rxq = queue; - const struct sfc_dp_rx *dp_rx; - - dp_rx = sfc_dp_rx_by_dp_rxq(dp_rxq); - - return offset < dp_rx->qdesc_npending(dp_rxq); -} - /* * The function is used by the secondary process as well. It must not * use any process-local pointers from the adapter data. @@ -1288,28 +1402,33 @@ sfc_tx_descriptor_status(void *queue, uint16_t offset) } static int -sfc_rx_queue_start(struct rte_eth_dev *dev, uint16_t rx_queue_id) +sfc_rx_queue_start(struct rte_eth_dev *dev, uint16_t ethdev_qid) { struct sfc_adapter_shared *sas = sfc_adapter_shared_by_eth_dev(dev); struct sfc_adapter *sa = sfc_adapter_by_eth_dev(dev); + sfc_ethdev_qid_t sfc_ethdev_qid = ethdev_qid; + struct sfc_rxq_info *rxq_info; + sfc_sw_index_t sw_index; int rc; - sfc_log_init(sa, "RxQ=%u", rx_queue_id); + sfc_log_init(sa, "RxQ=%u", ethdev_qid); sfc_adapter_lock(sa); rc = EINVAL; - if (sa->state != SFC_ADAPTER_STARTED) + if (sa->state != SFC_ETHDEV_STARTED) goto fail_not_started; - if (sas->rxq_info[rx_queue_id].state != SFC_RXQ_INITIALIZED) + rxq_info = sfc_rxq_info_by_ethdev_qid(sas, sfc_ethdev_qid); + if (rxq_info->state != SFC_RXQ_INITIALIZED) goto fail_not_setup; - rc = sfc_rx_qstart(sa, rx_queue_id); + sw_index = sfc_rxq_sw_index_by_ethdev_rx_qid(sas, sfc_ethdev_qid); + rc = sfc_rx_qstart(sa, sw_index); if (rc != 0) goto fail_rx_qstart; - sas->rxq_info[rx_queue_id].deferred_started = B_TRUE; + rxq_info->deferred_started = B_TRUE; sfc_adapter_unlock(sa); @@ -1324,17 +1443,23 @@ fail_not_started: } static int -sfc_rx_queue_stop(struct rte_eth_dev *dev, uint16_t rx_queue_id) +sfc_rx_queue_stop(struct rte_eth_dev *dev, uint16_t ethdev_qid) { struct sfc_adapter_shared *sas = sfc_adapter_shared_by_eth_dev(dev); struct sfc_adapter *sa = sfc_adapter_by_eth_dev(dev); + sfc_ethdev_qid_t sfc_ethdev_qid = ethdev_qid; + struct sfc_rxq_info *rxq_info; + sfc_sw_index_t sw_index; - sfc_log_init(sa, "RxQ=%u", rx_queue_id); + sfc_log_init(sa, "RxQ=%u", ethdev_qid); sfc_adapter_lock(sa); - sfc_rx_qstop(sa, rx_queue_id); - sas->rxq_info[rx_queue_id].deferred_started = B_FALSE; + sw_index = sfc_rxq_sw_index_by_ethdev_rx_qid(sas, sfc_ethdev_qid); + sfc_rx_qstop(sa, sw_index); + + rxq_info = sfc_rxq_info_by_ethdev_qid(sas, sfc_ethdev_qid); + rxq_info->deferred_started = B_FALSE; sfc_adapter_unlock(sa); @@ -1342,28 +1467,32 @@ sfc_rx_queue_stop(struct rte_eth_dev *dev, uint16_t rx_queue_id) } static int -sfc_tx_queue_start(struct rte_eth_dev *dev, uint16_t tx_queue_id) +sfc_tx_queue_start(struct rte_eth_dev *dev, uint16_t ethdev_qid) { struct sfc_adapter_shared *sas = sfc_adapter_shared_by_eth_dev(dev); struct sfc_adapter *sa = sfc_adapter_by_eth_dev(dev); + struct sfc_txq_info *txq_info; + sfc_sw_index_t sw_index; int rc; - sfc_log_init(sa, "TxQ = %u", tx_queue_id); + sfc_log_init(sa, "TxQ = %u", ethdev_qid); sfc_adapter_lock(sa); rc = EINVAL; - if (sa->state != SFC_ADAPTER_STARTED) + if (sa->state != SFC_ETHDEV_STARTED) goto fail_not_started; - if (sas->txq_info[tx_queue_id].state != SFC_TXQ_INITIALIZED) + txq_info = sfc_txq_info_by_ethdev_qid(sas, ethdev_qid); + if (txq_info->state != SFC_TXQ_INITIALIZED) goto fail_not_setup; - rc = sfc_tx_qstart(sa, tx_queue_id); + sw_index = sfc_txq_sw_index_by_ethdev_tx_qid(sas, ethdev_qid); + rc = sfc_tx_qstart(sa, sw_index); if (rc != 0) goto fail_tx_qstart; - sas->txq_info[tx_queue_id].deferred_started = B_TRUE; + txq_info->deferred_started = B_TRUE; sfc_adapter_unlock(sa); return 0; @@ -1378,18 +1507,22 @@ fail_not_started: } static int -sfc_tx_queue_stop(struct rte_eth_dev *dev, uint16_t tx_queue_id) +sfc_tx_queue_stop(struct rte_eth_dev *dev, uint16_t ethdev_qid) { struct sfc_adapter_shared *sas = sfc_adapter_shared_by_eth_dev(dev); struct sfc_adapter *sa = sfc_adapter_by_eth_dev(dev); + struct sfc_txq_info *txq_info; + sfc_sw_index_t sw_index; - sfc_log_init(sa, "TxQ = %u", tx_queue_id); + sfc_log_init(sa, "TxQ = %u", ethdev_qid); sfc_adapter_lock(sa); - sfc_tx_qstop(sa, tx_queue_id); + sw_index = sfc_txq_sw_index_by_ethdev_tx_qid(sas, ethdev_qid); + sfc_tx_qstop(sa, sw_index); - sas->txq_info[tx_queue_id].deferred_started = B_FALSE; + txq_info = sfc_txq_info_by_ethdev_qid(sas, ethdev_qid); + txq_info->deferred_started = B_FALSE; sfc_adapter_unlock(sa); return 0; @@ -1455,7 +1588,7 @@ sfc_dev_udp_tunnel_op(struct rte_eth_dev *dev, if (rc != 0) goto fail_op; - if (sa->state == SFC_ADAPTER_STARTED) { + if (sa->state == SFC_ETHDEV_STARTED) { rc = efx_tunnel_reconfigure(sa->nic); if (rc == EAGAIN) { /* @@ -1591,7 +1724,7 @@ sfc_dev_rss_hash_update(struct rte_eth_dev *dev, } if (rss_conf->rss_key != NULL) { - if (sa->state == SFC_ADAPTER_STARTED) { + if (sa->state == SFC_ETHDEV_STARTED) { for (key_i = 0; key_i < n_contexts; key_i++) { rc = efx_rx_scale_key_set(sa->nic, contexts[key_i], @@ -1718,7 +1851,7 @@ sfc_dev_rss_reta_update(struct rte_eth_dev *dev, } } - if (sa->state == SFC_ADAPTER_STARTED) { + if (sa->state == SFC_ETHDEV_STARTED) { rc = efx_rx_scale_tbl_set(sa->nic, EFX_RSS_CONTEXT_DEFAULT, rss_tbl_new, EFX_RSS_TBL_SIZE); if (rc != 0) @@ -1738,59 +1871,11 @@ bad_reta_entry: } static int -sfc_dev_filter_ctrl(struct rte_eth_dev *dev, enum rte_filter_type filter_type, - enum rte_filter_op filter_op, - void *arg) +sfc_dev_flow_ops_get(struct rte_eth_dev *dev __rte_unused, + const struct rte_flow_ops **ops) { - struct sfc_adapter *sa = sfc_adapter_by_eth_dev(dev); - int rc = ENOTSUP; - - sfc_log_init(sa, "entry"); - - switch (filter_type) { - case RTE_ETH_FILTER_NONE: - sfc_err(sa, "Global filters configuration not supported"); - break; - case RTE_ETH_FILTER_MACVLAN: - sfc_err(sa, "MACVLAN filters not supported"); - break; - case RTE_ETH_FILTER_ETHERTYPE: - sfc_err(sa, "EtherType filters not supported"); - break; - case RTE_ETH_FILTER_FLEXIBLE: - sfc_err(sa, "Flexible filters not supported"); - break; - case RTE_ETH_FILTER_SYN: - sfc_err(sa, "SYN filters not supported"); - break; - case RTE_ETH_FILTER_NTUPLE: - sfc_err(sa, "NTUPLE filters not supported"); - break; - case RTE_ETH_FILTER_TUNNEL: - sfc_err(sa, "Tunnel filters not supported"); - break; - case RTE_ETH_FILTER_FDIR: - sfc_err(sa, "Flow Director filters not supported"); - break; - case RTE_ETH_FILTER_HASH: - sfc_err(sa, "Hash filters not supported"); - break; - case RTE_ETH_FILTER_GENERIC: - if (filter_op != RTE_ETH_FILTER_GET) { - rc = EINVAL; - } else { - *(const void **)arg = &sfc_flow_ops; - rc = 0; - } - break; - default: - sfc_err(sa, "Unknown filter type %u", filter_type); - break; - } - - sfc_log_init(sa, "exit: %d", -rc); - SFC_ASSERT(rc >= 0); - return -rc; + *ops = &sfc_flow_ops; + return 0; } static int @@ -1809,31 +1894,202 @@ sfc_pool_ops_supported(struct rte_eth_dev *dev, const char *pool) } static int -sfc_rx_queue_intr_enable(struct rte_eth_dev *dev, uint16_t queue_id) +sfc_rx_queue_intr_enable(struct rte_eth_dev *dev, uint16_t ethdev_qid) { const struct sfc_adapter_priv *sap = sfc_adapter_priv_by_eth_dev(dev); struct sfc_adapter_shared *sas = sfc_adapter_shared_by_eth_dev(dev); + sfc_ethdev_qid_t sfc_ethdev_qid = ethdev_qid; struct sfc_rxq_info *rxq_info; - SFC_ASSERT(queue_id < sas->rxq_count); - rxq_info = &sas->rxq_info[queue_id]; + rxq_info = sfc_rxq_info_by_ethdev_qid(sas, sfc_ethdev_qid); return sap->dp_rx->intr_enable(rxq_info->dp); } static int -sfc_rx_queue_intr_disable(struct rte_eth_dev *dev, uint16_t queue_id) +sfc_rx_queue_intr_disable(struct rte_eth_dev *dev, uint16_t ethdev_qid) { const struct sfc_adapter_priv *sap = sfc_adapter_priv_by_eth_dev(dev); struct sfc_adapter_shared *sas = sfc_adapter_shared_by_eth_dev(dev); + sfc_ethdev_qid_t sfc_ethdev_qid = ethdev_qid; struct sfc_rxq_info *rxq_info; - SFC_ASSERT(queue_id < sas->rxq_count); - rxq_info = &sas->rxq_info[queue_id]; + rxq_info = sfc_rxq_info_by_ethdev_qid(sas, sfc_ethdev_qid); return sap->dp_rx->intr_disable(rxq_info->dp); } +struct sfc_mport_journal_ctx { + struct sfc_adapter *sa; + uint16_t switch_domain_id; + uint32_t mcdi_handle; + bool controllers_assigned; + efx_pcie_interface_t *controllers; + size_t nb_controllers; +}; + +static int +sfc_journal_ctx_add_controller(struct sfc_mport_journal_ctx *ctx, + efx_pcie_interface_t intf) +{ + efx_pcie_interface_t *new_controllers; + size_t i, target; + size_t new_size; + + if (ctx->controllers == NULL) { + ctx->controllers = rte_malloc("sfc_controller_mapping", + sizeof(ctx->controllers[0]), 0); + if (ctx->controllers == NULL) + return ENOMEM; + + ctx->controllers[0] = intf; + ctx->nb_controllers = 1; + + return 0; + } + + for (i = 0; i < ctx->nb_controllers; i++) { + if (ctx->controllers[i] == intf) + return 0; + if (ctx->controllers[i] > intf) + break; + } + target = i; + + ctx->nb_controllers += 1; + new_size = ctx->nb_controllers * sizeof(ctx->controllers[0]); + + new_controllers = rte_realloc(ctx->controllers, new_size, 0); + if (new_controllers == NULL) { + rte_free(ctx->controllers); + return ENOMEM; + } + ctx->controllers = new_controllers; + + for (i = target + 1; i < ctx->nb_controllers; i++) + ctx->controllers[i] = ctx->controllers[i - 1]; + + ctx->controllers[target] = intf; + + return 0; +} + +static efx_rc_t +sfc_process_mport_journal_entry(struct sfc_mport_journal_ctx *ctx, + efx_mport_desc_t *mport) +{ + efx_mport_sel_t ethdev_mport; + int rc; + + sfc_dbg(ctx->sa, + "processing mport id %u (controller %u pf %u vf %u)", + mport->emd_id.id, mport->emd_vnic.ev_intf, + mport->emd_vnic.ev_pf, mport->emd_vnic.ev_vf); + efx_mae_mport_invalid(ðdev_mport); + + if (!ctx->controllers_assigned) { + rc = sfc_journal_ctx_add_controller(ctx, + mport->emd_vnic.ev_intf); + if (rc != 0) + return rc; + } + + return 0; +} + +static efx_rc_t +sfc_process_mport_journal_cb(void *data, efx_mport_desc_t *mport, + size_t mport_len) +{ + struct sfc_mport_journal_ctx *ctx = data; + + if (ctx == NULL || ctx->sa == NULL) { + sfc_err(ctx->sa, "received NULL context or SFC adapter"); + return EINVAL; + } + + if (mport_len != sizeof(*mport)) { + sfc_err(ctx->sa, "actual and expected mport buffer sizes differ"); + return EINVAL; + } + + SFC_ASSERT(sfc_adapter_is_locked(ctx->sa)); + + /* + * If a zombie flag is set, it means the mport has been marked for + * deletion and cannot be used for any new operations. The mport will + * be destroyed completely once all references to it are released. + */ + if (mport->emd_zombie) { + sfc_dbg(ctx->sa, "mport is a zombie, skipping"); + return 0; + } + if (mport->emd_type != EFX_MPORT_TYPE_VNIC) { + sfc_dbg(ctx->sa, "mport is not a VNIC, skipping"); + return 0; + } + if (mport->emd_vnic.ev_client_type != EFX_MPORT_VNIC_CLIENT_FUNCTION) { + sfc_dbg(ctx->sa, "mport is not a function, skipping"); + return 0; + } + if (mport->emd_vnic.ev_handle == ctx->mcdi_handle) { + sfc_dbg(ctx->sa, "mport is this driver instance, skipping"); + return 0; + } + + return sfc_process_mport_journal_entry(ctx, mport); +} + +static int +sfc_process_mport_journal(struct sfc_adapter *sa) +{ + struct sfc_mport_journal_ctx ctx; + const efx_pcie_interface_t *controllers; + size_t nb_controllers; + efx_rc_t efx_rc; + int rc; + + memset(&ctx, 0, sizeof(ctx)); + ctx.sa = sa; + ctx.switch_domain_id = sa->mae.switch_domain_id; + + efx_rc = efx_mcdi_get_own_client_handle(sa->nic, &ctx.mcdi_handle); + if (efx_rc != 0) { + sfc_err(sa, "failed to get own MCDI handle"); + SFC_ASSERT(efx_rc > 0); + return efx_rc; + } + + rc = sfc_mae_switch_domain_controllers(ctx.switch_domain_id, + &controllers, &nb_controllers); + if (rc != 0) { + sfc_err(sa, "failed to get controller mapping"); + return rc; + } + + ctx.controllers_assigned = controllers != NULL; + ctx.controllers = NULL; + ctx.nb_controllers = 0; + + efx_rc = efx_mae_read_mport_journal(sa->nic, + sfc_process_mport_journal_cb, &ctx); + if (efx_rc != 0) { + sfc_err(sa, "failed to process MAE mport journal"); + SFC_ASSERT(efx_rc > 0); + return efx_rc; + } + + if (controllers == NULL) { + rc = sfc_mae_switch_domain_map_controllers(ctx.switch_domain_id, + ctx.controllers, + ctx.nb_controllers); + if (rc != 0) + return rc; + } + + return 0; +} + static const struct eth_dev_ops sfc_eth_dev_ops = { .dev_configure = sfc_dev_configure, .dev_start = sfc_dev_start, @@ -1873,7 +2129,7 @@ static const struct eth_dev_ops sfc_eth_dev_ops = { .reta_query = sfc_dev_rss_reta_query, .rss_hash_update = sfc_dev_rss_hash_update, .rss_hash_conf_get = sfc_dev_rss_hash_conf_get, - .filter_ctrl = sfc_dev_filter_ctrl, + .flow_ops_get = sfc_dev_flow_ops_get, .set_mc_addr_list = sfc_set_mc_addr_list, .rxq_info_get = sfc_rx_queue_info_get, .txq_info_get = sfc_tx_queue_info_get, @@ -1883,6 +2139,10 @@ static const struct eth_dev_ops sfc_eth_dev_ops = { .pool_ops_supported = sfc_pool_ops_supported, }; +struct sfc_ethdev_init_data { + uint16_t nb_representors; +}; + /** * Duplicate a string in potentially shared memory required for * multi-process support. @@ -2020,7 +2280,6 @@ sfc_eth_dev_set_ops(struct rte_eth_dev *dev) dev->tx_pkt_burst = dp_tx->pkt_burst; dev->rx_queue_count = sfc_rx_queue_count; - dev->rx_descriptor_done = sfc_rx_descriptor_done; dev->rx_descriptor_status = sfc_rx_descriptor_status; dev->tx_descriptor_status = sfc_tx_descriptor_status; dev->dev_ops = &sfc_eth_dev_ops; @@ -2128,7 +2387,6 @@ sfc_eth_dev_secondary_init(struct rte_eth_dev *dev, uint32_t logtype_main) dev->tx_pkt_prepare = dp_tx->pkt_prepare; dev->tx_pkt_burst = dp_tx->pkt_burst; dev->rx_queue_count = sfc_rx_queue_count; - dev->rx_descriptor_done = sfc_rx_descriptor_done; dev->rx_descriptor_status = sfc_rx_descriptor_status; dev->tx_descriptor_status = sfc_tx_descriptor_status; dev->dev_ops = &sfc_eth_dev_secondary_ops; @@ -2164,10 +2422,52 @@ sfc_register_dp(void) } static int -sfc_eth_dev_init(struct rte_eth_dev *dev) +sfc_parse_switch_mode(struct sfc_adapter *sa, bool has_representors) +{ + const efx_nic_cfg_t *encp = efx_nic_cfg_get(sa->nic); + const char *switch_mode = NULL; + int rc; + + sfc_log_init(sa, "entry"); + + rc = sfc_kvargs_process(sa, SFC_KVARG_SWITCH_MODE, + sfc_kvarg_string_handler, &switch_mode); + if (rc != 0) + goto fail_kvargs; + + if (switch_mode == NULL) { + sa->switchdev = encp->enc_mae_supported && + (!encp->enc_datapath_cap_evb || + has_representors); + } else if (strcasecmp(switch_mode, SFC_KVARG_SWITCH_MODE_LEGACY) == 0) { + sa->switchdev = false; + } else if (strcasecmp(switch_mode, + SFC_KVARG_SWITCH_MODE_SWITCHDEV) == 0) { + sa->switchdev = true; + } else { + sfc_err(sa, "invalid switch mode device argument '%s'", + switch_mode); + rc = EINVAL; + goto fail_mode; + } + + sfc_log_init(sa, "done"); + + return 0; + +fail_mode: +fail_kvargs: + sfc_log_init(sa, "failed: %s", rte_strerror(rc)); + + return rc; +} + +static int +sfc_eth_dev_init(struct rte_eth_dev *dev, void *init_params) { struct sfc_adapter_shared *sas = sfc_adapter_shared_by_eth_dev(dev); struct rte_pci_device *pci_dev = RTE_ETH_DEV_TO_PCI(dev); + struct sfc_ethdev_init_data *init_data = init_params; uint32_t logtype_main; struct sfc_adapter *sa; int rc; @@ -2175,6 +2475,17 @@ sfc_eth_dev_init(struct rte_eth_dev *dev) const struct rte_ether_addr *from; int ret; + if (sfc_efx_dev_class_get(pci_dev->device.devargs) != + SFC_EFX_DEV_CLASS_NET) { + SFC_GENERIC_LOG(DEBUG, + "Incompatible device class: skip probing, should be probed by other sfc driver."); + return 1; + } + + rc = sfc_dp_mport_register(); + if (rc != 0) + return rc; + sfc_register_dp(); logtype_main = sfc_register_logtype(&pci_dev->addr, @@ -2220,6 +2531,8 @@ sfc_eth_dev_init(struct rte_eth_dev *dev) /* Copy PCI device info to the dev->data */ rte_eth_copy_pci_info(dev, pci_dev); + dev->data->dev_flags |= RTE_ETH_DEV_AUTOFILL_QUEUE_XSTATS; + dev->data->dev_flags |= RTE_ETH_DEV_FLOW_OPS_THREAD_SAFE; rc = sfc_kvargs_parse(sa); if (rc != 0) @@ -2241,6 +2554,14 @@ sfc_eth_dev_init(struct rte_eth_dev *dev) if (rc != 0) goto fail_probe; + /* + * Selecting a default switch mode requires the NIC to be probed and + * to have its capabilities filled in. + */ + rc = sfc_parse_switch_mode(sa, init_data->nb_representors > 0); + if (rc != 0) + goto fail_switch_mode; + sfc_log_init(sa, "set device ops"); rc = sfc_eth_dev_set_ops(dev); if (rc != 0) @@ -2251,6 +2572,13 @@ sfc_eth_dev_init(struct rte_eth_dev *dev) if (rc != 0) goto fail_attach; + if (sa->switchdev && sa->mae.status != SFC_MAE_STATUS_SUPPORTED) { + sfc_err(sa, + "failed to enable switchdev mode without MAE support"); + rc = ENOTSUP; + goto fail_switchdev_no_mae; + } + encp = efx_nic_cfg_get(sa->nic); /* @@ -2265,10 +2593,14 @@ sfc_eth_dev_init(struct rte_eth_dev *dev) sfc_log_init(sa, "done"); return 0; +fail_switchdev_no_mae: + sfc_detach(sa); + fail_attach: sfc_eth_dev_clear_ops(dev); fail_set_ops: +fail_switch_mode: sfc_unprobe(sa); fail_probe: @@ -2311,11 +2643,297 @@ static const struct rte_pci_id pci_id_sfc_efx_map[] = { { .vendor_id = 0 /* sentinel */ } }; +static int +sfc_parse_rte_devargs(const char *args, struct rte_eth_devargs *devargs) +{ + struct rte_eth_devargs eth_da = { .nb_representor_ports = 0 }; + int rc; + + if (args != NULL) { + rc = rte_eth_devargs_parse(args, ð_da); + if (rc != 0) { + SFC_GENERIC_LOG(ERR, + "Failed to parse generic devargs '%s'", + args); + return rc; + } + } + + *devargs = eth_da; + + return 0; +} + +static int +sfc_eth_dev_find_or_create(struct rte_pci_device *pci_dev, + struct sfc_ethdev_init_data *init_data, + struct rte_eth_dev **devp, + bool *dev_created) +{ + struct rte_eth_dev *dev; + bool created = false; + int rc; + + dev = rte_eth_dev_allocated(pci_dev->device.name); + if (dev == NULL) { + rc = rte_eth_dev_create(&pci_dev->device, pci_dev->device.name, + sizeof(struct sfc_adapter_shared), + eth_dev_pci_specific_init, pci_dev, + sfc_eth_dev_init, init_data); + if (rc != 0) { + SFC_GENERIC_LOG(ERR, "Failed to create sfc ethdev '%s'", + pci_dev->device.name); + return rc; + } + + created = true; + + dev = rte_eth_dev_allocated(pci_dev->device.name); + if (dev == NULL) { + SFC_GENERIC_LOG(ERR, + "Failed to find allocated sfc ethdev '%s'", + pci_dev->device.name); + return -ENODEV; + } + } + + *devp = dev; + *dev_created = created; + + return 0; +} + +static int +sfc_eth_dev_create_repr(struct sfc_adapter *sa, + efx_pcie_interface_t controller, + uint16_t port, + uint16_t repr_port, + enum rte_eth_representor_type type) +{ + struct sfc_repr_entity_info entity; + efx_mport_sel_t mport_sel; + int rc; + + switch (type) { + case RTE_ETH_REPRESENTOR_NONE: + return 0; + case RTE_ETH_REPRESENTOR_VF: + case RTE_ETH_REPRESENTOR_PF: + break; + case RTE_ETH_REPRESENTOR_SF: + sfc_err(sa, "SF representors are not supported"); + return ENOTSUP; + default: + sfc_err(sa, "unknown representor type: %d", type); + return ENOTSUP; + } + + rc = efx_mae_mport_by_pcie_mh_function(controller, + port, + repr_port, + &mport_sel); + if (rc != 0) { + sfc_err(sa, + "failed to get m-port selector for controller %u port %u repr_port %u: %s", + controller, port, repr_port, rte_strerror(-rc)); + return rc; + } + + memset(&entity, 0, sizeof(entity)); + entity.type = type; + entity.intf = controller; + entity.pf = port; + entity.vf = repr_port; + + rc = sfc_repr_create(sa->eth_dev, &entity, sa->mae.switch_domain_id, + &mport_sel); + if (rc != 0) { + sfc_err(sa, + "failed to create representor for controller %u port %u repr_port %u: %s", + controller, port, repr_port, rte_strerror(-rc)); + return rc; + } + + return 0; +} + +static int +sfc_eth_dev_create_repr_port(struct sfc_adapter *sa, + const struct rte_eth_devargs *eth_da, + efx_pcie_interface_t controller, + uint16_t port) +{ + int first_error = 0; + uint16_t i; + int rc; + + if (eth_da->type == RTE_ETH_REPRESENTOR_PF) { + return sfc_eth_dev_create_repr(sa, controller, port, + EFX_PCI_VF_INVALID, + eth_da->type); + } + + for (i = 0; i < eth_da->nb_representor_ports; i++) { + rc = sfc_eth_dev_create_repr(sa, controller, port, + eth_da->representor_ports[i], + eth_da->type); + if (rc != 0 && first_error == 0) + first_error = rc; + } + + return first_error; +} + +static int +sfc_eth_dev_create_repr_controller(struct sfc_adapter *sa, + const struct rte_eth_devargs *eth_da, + efx_pcie_interface_t controller) +{ + const efx_nic_cfg_t *encp; + int first_error = 0; + uint16_t default_port; + uint16_t i; + int rc; + + if (eth_da->nb_ports == 0) { + encp = efx_nic_cfg_get(sa->nic); + default_port = encp->enc_intf == controller ? encp->enc_pf : 0; + return sfc_eth_dev_create_repr_port(sa, eth_da, controller, + default_port); + } + + for (i = 0; i < eth_da->nb_ports; i++) { + rc = sfc_eth_dev_create_repr_port(sa, eth_da, controller, + eth_da->ports[i]); + if (rc != 0 && first_error == 0) + first_error = rc; + } + + return first_error; +} + +static int +sfc_eth_dev_create_representors(struct rte_eth_dev *dev, + const struct rte_eth_devargs *eth_da) +{ + efx_pcie_interface_t intf; + const efx_nic_cfg_t *encp; + struct sfc_adapter *sa; + uint16_t switch_domain_id; + uint16_t i; + int rc; + + sa = sfc_adapter_by_eth_dev(dev); + switch_domain_id = sa->mae.switch_domain_id; + + switch (eth_da->type) { + case RTE_ETH_REPRESENTOR_NONE: + return 0; + case RTE_ETH_REPRESENTOR_PF: + case RTE_ETH_REPRESENTOR_VF: + break; + case RTE_ETH_REPRESENTOR_SF: + sfc_err(sa, "SF representors are not supported"); + return -ENOTSUP; + default: + sfc_err(sa, "unknown representor type: %d", + eth_da->type); + return -ENOTSUP; + } + + if (!sa->switchdev) { + sfc_err(sa, "cannot create representors in non-switchdev mode"); + return -EINVAL; + } + + if (!sfc_repr_available(sfc_sa2shared(sa))) { + sfc_err(sa, "cannot create representors: unsupported"); + + return -ENOTSUP; + } + + /* + * This is needed to construct the DPDK controller -> EFX interface + * mapping. + */ + sfc_adapter_lock(sa); + rc = sfc_process_mport_journal(sa); + sfc_adapter_unlock(sa); + if (rc != 0) { + SFC_ASSERT(rc > 0); + return -rc; + } + + if (eth_da->nb_mh_controllers > 0) { + for (i = 0; i < eth_da->nb_mh_controllers; i++) { + rc = sfc_mae_switch_domain_get_intf(switch_domain_id, + eth_da->mh_controllers[i], + &intf); + if (rc != 0) { + sfc_err(sa, "failed to get representor"); + continue; + } + sfc_eth_dev_create_repr_controller(sa, eth_da, intf); + } + } else { + encp = efx_nic_cfg_get(sa->nic); + sfc_eth_dev_create_repr_controller(sa, eth_da, encp->enc_intf); + } + + return 0; +} + static int sfc_eth_dev_pci_probe(struct rte_pci_driver *pci_drv __rte_unused, struct rte_pci_device *pci_dev) { - return rte_eth_dev_pci_generic_probe(pci_dev, - sizeof(struct sfc_adapter_shared), sfc_eth_dev_init); + struct sfc_ethdev_init_data init_data; + struct rte_eth_devargs eth_da; + struct rte_eth_dev *dev; + bool dev_created; + int rc; + + if (pci_dev->device.devargs != NULL) { + rc = sfc_parse_rte_devargs(pci_dev->device.devargs->args, + ð_da); + if (rc != 0) + return rc; + } else { + memset(ð_da, 0, sizeof(eth_da)); + } + + /* If no VF representors specified, check for PF ones */ + if (eth_da.nb_representor_ports > 0) + init_data.nb_representors = eth_da.nb_representor_ports; + else + init_data.nb_representors = eth_da.nb_ports; + + if (init_data.nb_representors > 0 && + rte_eal_process_type() != RTE_PROC_PRIMARY) { + SFC_GENERIC_LOG(ERR, + "Create representors from secondary process not supported, dev '%s'", + pci_dev->device.name); + return -ENOTSUP; + } + + /* + * Driver supports RTE_PCI_DRV_PROBE_AGAIN. Hence create device only + * if it does not already exist. Re-probing an existing device is + * expected to allow additional representors to be configured. + */ + rc = sfc_eth_dev_find_or_create(pci_dev, &init_data, &dev, + &dev_created); + if (rc != 0) + return rc; + + rc = sfc_eth_dev_create_representors(dev, ð_da); + if (rc != 0) { + if (dev_created) + (void)rte_eth_dev_destroy(dev, sfc_eth_dev_uninit); + + return rc; + } + + return 0; } static int sfc_eth_dev_pci_remove(struct rte_pci_device *pci_dev) @@ -2327,7 +2945,8 @@ static struct rte_pci_driver sfc_efx_pmd = { .id_table = pci_id_sfc_efx_map, .drv_flags = RTE_PCI_DRV_INTR_LSC | - RTE_PCI_DRV_NEED_MAPPING, + RTE_PCI_DRV_NEED_MAPPING | + RTE_PCI_DRV_PROBE_AGAIN, .probe = sfc_eth_dev_pci_probe, .remove = sfc_eth_dev_pci_remove, }; @@ -2336,6 +2955,7 @@ RTE_PMD_REGISTER_PCI(net_sfc_efx, sfc_efx_pmd); RTE_PMD_REGISTER_PCI_TABLE(net_sfc_efx, pci_id_sfc_efx_map); RTE_PMD_REGISTER_KMOD_DEP(net_sfc_efx, "* igb_uio | uio_pci_generic | vfio-pci"); RTE_PMD_REGISTER_PARAM_STRING(net_sfc_efx, + SFC_KVARG_SWITCH_MODE "=" SFC_KVARG_VALUES_SWITCH_MODE " " SFC_KVARG_RX_DATAPATH "=" SFC_KVARG_VALUES_RX_DATAPATH " " SFC_KVARG_TX_DATAPATH "=" SFC_KVARG_VALUES_TX_DATAPATH " " SFC_KVARG_PERF_PROFILE "=" SFC_KVARG_VALUES_PERF_PROFILE " "