#include "sfc_rx.h"
#include "sfc_tx.h"
#include "sfc_flow.h"
+#include "sfc_flow_tunnel.h"
#include "sfc_dp.h"
#include "sfc_dp_rx.h"
#include "sfc_repr.h"
#include "sfc_sw_stats.h"
#include "sfc_switch.h"
+#include "sfc_nic_dma.h"
#define SFC_XSTAT_ID_INVALID_VAL UINT64_MAX
#define SFC_XSTAT_ID_INVALID_NAME '\0'
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");
dev_info->max_vfs = sa->sriov.num_vfs;
/* Autonegotiation may be disabled */
- dev_info->speed_capa = ETH_LINK_SPEED_FIXED;
+ dev_info->speed_capa = RTE_ETH_LINK_SPEED_FIXED;
if (sa->port.phy_adv_cap_mask & (1u << EFX_PHY_CAP_1000FDX))
- dev_info->speed_capa |= ETH_LINK_SPEED_1G;
+ dev_info->speed_capa |= RTE_ETH_LINK_SPEED_1G;
if (sa->port.phy_adv_cap_mask & (1u << EFX_PHY_CAP_10000FDX))
- dev_info->speed_capa |= ETH_LINK_SPEED_10G;
+ dev_info->speed_capa |= RTE_ETH_LINK_SPEED_10G;
if (sa->port.phy_adv_cap_mask & (1u << EFX_PHY_CAP_25000FDX))
- dev_info->speed_capa |= ETH_LINK_SPEED_25G;
+ dev_info->speed_capa |= RTE_ETH_LINK_SPEED_25G;
if (sa->port.phy_adv_cap_mask & (1u << EFX_PHY_CAP_40000FDX))
- dev_info->speed_capa |= ETH_LINK_SPEED_40G;
+ dev_info->speed_capa |= RTE_ETH_LINK_SPEED_40G;
if (sa->port.phy_adv_cap_mask & (1u << EFX_PHY_CAP_50000FDX))
- dev_info->speed_capa |= ETH_LINK_SPEED_50G;
+ dev_info->speed_capa |= RTE_ETH_LINK_SPEED_50G;
if (sa->port.phy_adv_cap_mask & (1u << EFX_PHY_CAP_100000FDX))
- dev_info->speed_capa |= ETH_LINK_SPEED_100G;
+ dev_info->speed_capa |= RTE_ETH_LINK_SPEED_100G;
dev_info->max_rx_queues = sa->rxq_max;
dev_info->max_tx_queues = sa->txq_max;
dev_info->tx_offload_capa = sfc_tx_get_dev_offload_caps(sa) |
dev_info->tx_queue_offload_capa;
- if (dev_info->tx_offload_capa & DEV_TX_OFFLOAD_MBUF_FAST_FREE)
- txq_offloads_def |= DEV_TX_OFFLOAD_MBUF_FAST_FREE;
-
- dev_info->default_txconf.offloads |= txq_offloads_def;
-
if (rss->context_type != EFX_RX_SCALE_UNAVAILABLE) {
uint64_t rte_hf = 0;
unsigned int i;
dev_info->dev_capa = RTE_ETH_DEV_CAPA_RUNTIME_RX_QUEUE_SETUP |
RTE_ETH_DEV_CAPA_RUNTIME_TX_QUEUE_SETUP;
+ dev_info->dev_capa &= ~RTE_ETH_DEV_CAPA_FLOW_RULE_KEEP;
- if (mae->status == SFC_MAE_STATUS_SUPPORTED) {
+ if (mae->status == SFC_MAE_STATUS_SUPPORTED ||
+ mae->status == SFC_MAE_STATUS_ADMIN) {
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;
sfc_eth_dev_clear_ops(dev);
+ sfc_nic_dma_detach(sa);
sfc_detach(sa);
sfc_unprobe(sa);
switch (link_fc) {
case 0:
- fc_conf->mode = RTE_FC_NONE;
+ fc_conf->mode = RTE_ETH_FC_NONE;
break;
case EFX_FCNTL_RESPOND:
- fc_conf->mode = RTE_FC_RX_PAUSE;
+ fc_conf->mode = RTE_ETH_FC_RX_PAUSE;
break;
case EFX_FCNTL_GENERATE:
- fc_conf->mode = RTE_FC_TX_PAUSE;
+ fc_conf->mode = RTE_ETH_FC_TX_PAUSE;
break;
case (EFX_FCNTL_RESPOND | EFX_FCNTL_GENERATE):
- fc_conf->mode = RTE_FC_FULL;
+ fc_conf->mode = RTE_ETH_FC_FULL;
break;
default:
sfc_err(sa, "%s: unexpected flow control value %#x",
}
switch (fc_conf->mode) {
- case RTE_FC_NONE:
+ case RTE_ETH_FC_NONE:
fcntl = 0;
break;
- case RTE_FC_RX_PAUSE:
+ case RTE_ETH_FC_RX_PAUSE:
fcntl = EFX_FCNTL_RESPOND;
break;
- case RTE_FC_TX_PAUSE:
+ case RTE_ETH_FC_TX_PAUSE:
fcntl = EFX_FCNTL_GENERATE;
break;
- case RTE_FC_FULL:
+ case RTE_ETH_FC_FULL:
fcntl = EFX_FCNTL_RESPOND | EFX_FCNTL_GENERATE;
break;
default:
}
}
- /*
- * 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_MTU) {
- struct rte_eth_rxmode *rxmode = &dev->data->dev_conf.rxmode;
- rxmode->offloads |= DEV_RX_OFFLOAD_JUMBO_FRAME;
- }
-
- dev->data->dev_conf.rxmode.max_rx_pkt_len = sa->port.pdu;
-
sfc_adapter_unlock(sa);
sfc_log_init(sa, "done");
qinfo->conf.rx_deferred_start = rxq_info->deferred_start;
qinfo->conf.offloads = dev->data->dev_conf.rxmode.offloads;
if (rxq_info->type_flags & EFX_RXQ_FLAG_SCATTER) {
- qinfo->conf.offloads |= DEV_RX_OFFLOAD_SCATTER;
+ qinfo->conf.offloads |= RTE_ETH_RX_OFFLOAD_SCATTER;
qinfo->scattered_rx = 1;
}
qinfo->nb_desc = rxq_info->entries;
* use any process-local pointers from the adapter data.
*/
static uint32_t
-sfc_rx_queue_count(struct rte_eth_dev *dev, uint16_t ethdev_qid)
+sfc_rx_queue_count(void *rx_queue)
{
- 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_dp_rxq *dp_rxq = rx_queue;
+ const struct sfc_dp_rx *dp_rx;
struct sfc_rxq_info *rxq_info;
- rxq_info = sfc_rxq_info_by_ethdev_qid(sas, sfc_ethdev_qid);
+ dp_rx = sfc_dp_rx_by_dp_rxq(dp_rxq);
+ rxq_info = sfc_rxq_info_by_dp_rxq(dp_rxq);
if ((rxq_info->state & SFC_RXQ_STARTED) == 0)
return 0;
- return sap->dp_rx->qdesc_npending(rxq_info->dp);
+ return dp_rx->qdesc_npending(dp_rxq);
}
/*
sfc_tunnel_rte_type_to_efx_udp_proto(enum rte_eth_tunnel_type rte_type)
{
switch (rte_type) {
- case RTE_TUNNEL_TYPE_VXLAN:
+ case RTE_ETH_TUNNEL_TYPE_VXLAN:
return EFX_TUNNEL_PROTOCOL_VXLAN;
- case RTE_TUNNEL_TYPE_GENEVE:
+ case RTE_ETH_TUNNEL_TYPE_GENEVE:
return EFX_TUNNEL_PROTOCOL_GENEVE;
default:
return EFX_TUNNEL_NPROTOS;
/*
* Mapping of hash configuration between RTE and EFX is not one-to-one,
- * hence, conversion is done here to derive a correct set of ETH_RSS
+ * hence, conversion is done here to derive a correct set of RTE_ETH_RSS
* flags which corresponds to the active EFX configuration stored
* locally in 'sfc_adapter' and kept up-to-date
*/
struct sfc_adapter *sa = sfc_adapter_by_eth_dev(dev);
struct sfc_rss *rss = &sfc_sa2shared(sa)->rss;
unsigned int efx_hash_types;
- uint32_t contexts[] = {EFX_RSS_CONTEXT_DEFAULT, rss->dummy_rss_context};
unsigned int n_contexts;
unsigned int mode_i = 0;
unsigned int key_i = 0;
+ uint32_t contexts[2];
unsigned int i = 0;
int rc = 0;
- n_contexts = rss->dummy_rss_context == EFX_RSS_CONTEXT_DEFAULT ? 1 : 2;
-
if (sfc_sa2shared(sa)->isolated)
return -ENOTSUP;
if (rc != 0)
goto fail_rx_hf_rte_to_efx;
+ contexts[0] = EFX_RSS_CONTEXT_DEFAULT;
+ contexts[1] = rss->dummy_ctx.nic_handle;
+ n_contexts = (rss->dummy_ctx.nic_handle_refcnt == 0) ? 1 : 2;
+
for (mode_i = 0; mode_i < n_contexts; mode_i++) {
rc = efx_rx_scale_mode_set(sa->nic, contexts[mode_i],
rss->hash_alg, efx_hash_types,
return -EINVAL;
for (entry = 0; entry < reta_size; entry++) {
- int grp = entry / RTE_RETA_GROUP_SIZE;
- int grp_idx = entry % RTE_RETA_GROUP_SIZE;
+ int grp = entry / RTE_ETH_RETA_GROUP_SIZE;
+ int grp_idx = entry % RTE_ETH_RETA_GROUP_SIZE;
if ((reta_conf[grp].mask >> grp_idx) & 1)
reta_conf[grp].reta[grp_idx] = rss->tbl[entry];
rte_memcpy(rss_tbl_new, rss->tbl, sizeof(rss->tbl));
for (entry = 0; entry < reta_size; entry++) {
- int grp_idx = entry % RTE_RETA_GROUP_SIZE;
+ int grp_idx = entry % RTE_ETH_RETA_GROUP_SIZE;
struct rte_eth_rss_reta_entry64 *grp;
- grp = &reta_conf[entry / RTE_RETA_GROUP_SIZE];
+ grp = &reta_conf[entry / RTE_ETH_RETA_GROUP_SIZE];
if (grp->mask & (1ull << grp_idx)) {
if (grp->reta[grp_idx] >= rss->channels) {
sfc_process_mport_journal_entry(struct sfc_mport_journal_ctx *ctx,
efx_mport_desc_t *mport)
{
+ struct sfc_mae_switch_port_request req;
+ efx_mport_sel_t entity_selector;
efx_mport_sel_t ethdev_mport;
+ uint16_t switch_port_id;
+ efx_rc_t efx_rc;
int rc;
sfc_dbg(ctx->sa,
return rc;
}
+ /* Build Mport selector */
+ efx_rc = efx_mae_mport_by_pcie_mh_function(mport->emd_vnic.ev_intf,
+ mport->emd_vnic.ev_pf,
+ mport->emd_vnic.ev_vf,
+ &entity_selector);
+ if (efx_rc != 0) {
+ sfc_err(ctx->sa, "failed to build entity mport selector for c%upf%uvf%u",
+ mport->emd_vnic.ev_intf,
+ mport->emd_vnic.ev_pf,
+ mport->emd_vnic.ev_vf);
+ return efx_rc;
+ }
+
+ rc = sfc_mae_switch_port_id_by_entity(ctx->switch_domain_id,
+ &entity_selector,
+ SFC_MAE_SWITCH_PORT_REPRESENTOR,
+ &switch_port_id);
+ switch (rc) {
+ case 0:
+ /* Already registered */
+ break;
+ case ENOENT:
+ /*
+ * No representor has been created for this entity.
+ * Create a dummy switch registry entry with an invalid ethdev
+ * mport selector. When a corresponding representor is created,
+ * this entry will be updated.
+ */
+ req.type = SFC_MAE_SWITCH_PORT_REPRESENTOR;
+ req.entity_mportp = &entity_selector;
+ req.ethdev_mportp = ðdev_mport;
+ req.ethdev_port_id = RTE_MAX_ETHPORTS;
+ req.port_data.repr.intf = mport->emd_vnic.ev_intf;
+ req.port_data.repr.pf = mport->emd_vnic.ev_pf;
+ req.port_data.repr.vf = mport->emd_vnic.ev_vf;
+
+ rc = sfc_mae_assign_switch_port(ctx->switch_domain_id,
+ &req, &switch_port_id);
+ if (rc != 0) {
+ sfc_err(ctx->sa,
+ "failed to assign MAE switch port for c%upf%uvf%u: %s",
+ mport->emd_vnic.ev_intf,
+ mport->emd_vnic.ev_pf,
+ mport->emd_vnic.ev_vf,
+ rte_strerror(rc));
+ return rc;
+ }
+ break;
+ default:
+ sfc_err(ctx->sa, "failed to find MAE switch port for c%upf%uvf%u: %s",
+ mport->emd_vnic.ev_intf,
+ mport->emd_vnic.ev_pf,
+ mport->emd_vnic.ev_vf,
+ rte_strerror(rc));
+ return rc;
+ }
+
return 0;
}
return 0;
}
+static void
+sfc_count_representors_cb(enum sfc_mae_switch_port_type type,
+ const efx_mport_sel_t *ethdev_mportp __rte_unused,
+ uint16_t ethdev_port_id __rte_unused,
+ const efx_mport_sel_t *entity_mportp __rte_unused,
+ uint16_t switch_port_id __rte_unused,
+ union sfc_mae_switch_port_data *port_datap
+ __rte_unused,
+ void *user_datap)
+{
+ int *counter = user_datap;
+
+ SFC_ASSERT(counter != NULL);
+
+ if (type == SFC_MAE_SWITCH_PORT_REPRESENTOR)
+ (*counter)++;
+}
+
+struct sfc_get_representors_ctx {
+ struct rte_eth_representor_info *info;
+ struct sfc_adapter *sa;
+ uint16_t switch_domain_id;
+ const efx_pcie_interface_t *controllers;
+ size_t nb_controllers;
+};
+
+static void
+sfc_get_representors_cb(enum sfc_mae_switch_port_type type,
+ const efx_mport_sel_t *ethdev_mportp __rte_unused,
+ uint16_t ethdev_port_id __rte_unused,
+ const efx_mport_sel_t *entity_mportp __rte_unused,
+ uint16_t switch_port_id,
+ union sfc_mae_switch_port_data *port_datap,
+ void *user_datap)
+{
+ struct sfc_get_representors_ctx *ctx = user_datap;
+ struct rte_eth_representor_range *range;
+ int ret;
+ int rc;
+
+ SFC_ASSERT(ctx != NULL);
+ SFC_ASSERT(ctx->info != NULL);
+ SFC_ASSERT(ctx->sa != NULL);
+
+ if (type != SFC_MAE_SWITCH_PORT_REPRESENTOR) {
+ sfc_dbg(ctx->sa, "not a representor, skipping");
+ return;
+ }
+ if (ctx->info->nb_ranges >= ctx->info->nb_ranges_alloc) {
+ sfc_dbg(ctx->sa, "info structure is full already");
+ return;
+ }
+
+ range = &ctx->info->ranges[ctx->info->nb_ranges];
+ rc = sfc_mae_switch_controller_from_mapping(ctx->controllers,
+ ctx->nb_controllers,
+ port_datap->repr.intf,
+ &range->controller);
+ if (rc != 0) {
+ sfc_err(ctx->sa, "invalid representor controller: %d",
+ port_datap->repr.intf);
+ range->controller = -1;
+ }
+ range->pf = port_datap->repr.pf;
+ range->id_base = switch_port_id;
+ range->id_end = switch_port_id;
+
+ if (port_datap->repr.vf != EFX_PCI_VF_INVALID) {
+ range->type = RTE_ETH_REPRESENTOR_VF;
+ range->vf = port_datap->repr.vf;
+ ret = snprintf(range->name, RTE_DEV_NAME_MAX_LEN,
+ "c%dpf%dvf%d", range->controller, range->pf,
+ range->vf);
+ } else {
+ range->type = RTE_ETH_REPRESENTOR_PF;
+ ret = snprintf(range->name, RTE_DEV_NAME_MAX_LEN,
+ "c%dpf%d", range->controller, range->pf);
+ }
+ if (ret >= RTE_DEV_NAME_MAX_LEN) {
+ sfc_err(ctx->sa, "representor name has been truncated: %s",
+ range->name);
+ }
+
+ ctx->info->nb_ranges++;
+}
+
+static int
+sfc_representor_info_get(struct rte_eth_dev *dev,
+ struct rte_eth_representor_info *info)
+{
+ struct sfc_adapter *sa = sfc_adapter_by_eth_dev(dev);
+ struct sfc_get_representors_ctx get_repr_ctx;
+ const efx_nic_cfg_t *nic_cfg;
+ uint16_t switch_domain_id;
+ uint32_t nb_repr;
+ int controller;
+ int rc;
+
+ sfc_adapter_lock(sa);
+
+ if (sa->mae.status != SFC_MAE_STATUS_ADMIN) {
+ sfc_adapter_unlock(sa);
+ return -ENOTSUP;
+ }
+
+ rc = sfc_process_mport_journal(sa);
+ if (rc != 0) {
+ sfc_adapter_unlock(sa);
+ SFC_ASSERT(rc > 0);
+ return -rc;
+ }
+
+ switch_domain_id = sa->mae.switch_domain_id;
+
+ nb_repr = 0;
+ rc = sfc_mae_switch_ports_iterate(switch_domain_id,
+ sfc_count_representors_cb,
+ &nb_repr);
+ if (rc != 0) {
+ sfc_adapter_unlock(sa);
+ SFC_ASSERT(rc > 0);
+ return -rc;
+ }
+
+ if (info == NULL) {
+ sfc_adapter_unlock(sa);
+ return nb_repr;
+ }
+
+ rc = sfc_mae_switch_domain_controllers(switch_domain_id,
+ &get_repr_ctx.controllers,
+ &get_repr_ctx.nb_controllers);
+ if (rc != 0) {
+ sfc_adapter_unlock(sa);
+ SFC_ASSERT(rc > 0);
+ return -rc;
+ }
+
+ nic_cfg = efx_nic_cfg_get(sa->nic);
+
+ rc = sfc_mae_switch_domain_get_controller(switch_domain_id,
+ nic_cfg->enc_intf,
+ &controller);
+ if (rc != 0) {
+ sfc_err(sa, "invalid controller: %d", nic_cfg->enc_intf);
+ controller = -1;
+ }
+
+ info->controller = controller;
+ info->pf = nic_cfg->enc_pf;
+
+ get_repr_ctx.info = info;
+ get_repr_ctx.sa = sa;
+ get_repr_ctx.switch_domain_id = switch_domain_id;
+ rc = sfc_mae_switch_ports_iterate(switch_domain_id,
+ sfc_get_representors_cb,
+ &get_repr_ctx);
+ if (rc != 0) {
+ sfc_adapter_unlock(sa);
+ SFC_ASSERT(rc > 0);
+ return -rc;
+ }
+
+ sfc_adapter_unlock(sa);
+ return nb_repr;
+}
+
+static int
+sfc_rx_metadata_negotiate(struct rte_eth_dev *dev, uint64_t *features)
+{
+ struct sfc_adapter *sa = sfc_adapter_by_eth_dev(dev);
+ uint64_t supported = 0;
+
+ sfc_adapter_lock(sa);
+
+ if ((sa->priv.dp_rx->features & SFC_DP_RX_FEAT_FLOW_FLAG) != 0)
+ supported |= RTE_ETH_RX_METADATA_USER_FLAG;
+
+ if ((sa->priv.dp_rx->features & SFC_DP_RX_FEAT_FLOW_MARK) != 0)
+ supported |= RTE_ETH_RX_METADATA_USER_MARK;
+
+ if (sfc_ft_is_supported(sa))
+ supported |= RTE_ETH_RX_METADATA_TUNNEL_ID;
+
+ sa->negotiated_rx_metadata = supported & *features;
+ *features = sa->negotiated_rx_metadata;
+
+ sfc_adapter_unlock(sa);
+
+ return 0;
+}
+
static const struct eth_dev_ops sfc_eth_dev_ops = {
.dev_configure = sfc_dev_configure,
.dev_start = sfc_dev_start,
.xstats_get_by_id = sfc_xstats_get_by_id,
.xstats_get_names_by_id = sfc_xstats_get_names_by_id,
.pool_ops_supported = sfc_pool_ops_supported,
+ .representor_info_get = sfc_representor_info_get,
+ .rx_metadata_negotiate = sfc_rx_metadata_negotiate,
};
struct sfc_ethdev_init_data {
goto fail_dp_rx_name;
}
+ if (strcmp(dp_rx->dp.name, SFC_KVARG_DATAPATH_EF10_ESSB) == 0) {
+ /* FLAG and MARK are always available from Rx prefix. */
+ sa->negotiated_rx_metadata |= RTE_ETH_RX_METADATA_USER_FLAG;
+ sa->negotiated_rx_metadata |= RTE_ETH_RX_METADATA_USER_MARK;
+ }
+
sfc_notice(sa, "use %s Rx datapath", sas->dp_rx_name);
rc = sfc_kvargs_process(sa, SFC_KVARG_TX_DATAPATH,
goto fail_kvargs;
if (switch_mode == NULL) {
- sa->switchdev = encp->enc_mae_supported &&
+ sa->switchdev = encp->enc_mae_admin &&
(!encp->enc_datapath_cap_evb ||
has_representors);
} else if (strcasecmp(switch_mode, SFC_KVARG_SWITCH_MODE_LEGACY) == 0) {
/* 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)
goto fail_attach;
- if (sa->switchdev && sa->mae.status != SFC_MAE_STATUS_SUPPORTED) {
+ if (sa->switchdev && sa->mae.status != SFC_MAE_STATUS_ADMIN) {
sfc_err(sa,
- "failed to enable switchdev mode without MAE support");
+ "failed to enable switchdev mode without admin MAE privilege");
rc = ENOTSUP;
goto fail_switchdev_no_mae;
}
from = (const struct rte_ether_addr *)(encp->enc_mac_addr);
rte_ether_addr_copy(from, &dev->data->mac_addrs[0]);
+ /*
+ * Setup the NIC DMA mapping handler. All internal mempools
+ * MUST be created on attach before this point, and the
+ * adapter MUST NOT create mempools with the adapter lock
+ * held after this point.
+ */
+ rc = sfc_nic_dma_attach(sa);
+ if (rc != 0)
+ goto fail_nic_dma_attach;
+
sfc_adapter_unlock(sa);
sfc_log_init(sa, "done");
return 0;
+fail_nic_dma_attach:
fail_switchdev_no_mae:
sfc_detach(sa);
{ RTE_PCI_DEVICE(EFX_PCI_VENID_SFC, EFX_PCI_DEVID_MEDFORD2) },
{ RTE_PCI_DEVICE(EFX_PCI_VENID_SFC, EFX_PCI_DEVID_MEDFORD2_VF) },
{ RTE_PCI_DEVICE(EFX_PCI_VENID_XILINX, EFX_PCI_DEVID_RIVERHEAD) },
+ { RTE_PCI_DEVICE(EFX_PCI_VENID_XILINX, EFX_PCI_DEVID_RIVERHEAD_VF) },
{ .vendor_id = 0 /* sentinel */ }
};