#pragma GCC diagnostic error "-Wpedantic"
#endif
-/* DPDK headers don't like -pedantic. */
-#ifdef PEDANTIC
-#pragma GCC diagnostic ignored "-Wpedantic"
-#endif
#include <rte_malloc.h>
#include <rte_ethdev.h>
#include <rte_ethdev_pci.h>
#include <rte_pci.h>
#include <rte_common.h>
#include <rte_kvargs.h>
-#ifdef PEDANTIC
-#pragma GCC diagnostic error "-Wpedantic"
-#endif
#include "mlx5.h"
#include "mlx5_utils.h"
/* Default PMD specific parameter value. */
#define MLX5_ARG_UNSET (-1)
+#ifndef HAVE_IBV_MLX5_MOD_MPW
+#define MLX5DV_CONTEXT_FLAGS_MPW_ALLOWED (1 << 2)
+#define MLX5DV_CONTEXT_FLAGS_ENHANCED_MPW (1 << 3)
+#endif
+
struct mlx5_args {
int cqe_comp;
int txq_inline;
return atoi(val);
}
+/**
+ * Verbs callback to allocate a memory. This function should allocate the space
+ * according to the size provided residing inside a huge page.
+ * Please note that all allocation must respect the alignment from libmlx5
+ * (i.e. currently sysconf(_SC_PAGESIZE)).
+ *
+ * @param[in] size
+ * The size in bytes of the memory to allocate.
+ * @param[in] data
+ * A pointer to the callback data.
+ *
+ * @return
+ * a pointer to the allocate space.
+ */
+static void *
+mlx5_alloc_verbs_buf(size_t size, void *data)
+{
+ struct priv *priv = data;
+ void *ret;
+ size_t alignment = sysconf(_SC_PAGESIZE);
+
+ assert(data != NULL);
+ assert(!mlx5_is_secondary());
+ ret = rte_malloc_socket(__func__, size, alignment,
+ priv->dev->device->numa_node);
+ DEBUG("Extern alloc size: %lu, align: %lu: %p", size, alignment, ret);
+ return ret;
+}
+
+/**
+ * Verbs callback to free a memory.
+ *
+ * @param[in] ptr
+ * A pointer to the memory to free.
+ * @param[in] data
+ * A pointer to the callback data.
+ */
+static void
+mlx5_free_verbs_buf(void *ptr, void *data __rte_unused)
+{
+ assert(data != NULL);
+ assert(!mlx5_is_secondary());
+ DEBUG("Extern free request: %p", ptr);
+ rte_free(ptr);
+}
+
/**
* DPDK callback to close the device.
*
{
struct priv *priv = mlx5_get_priv(dev);
unsigned int i;
+ int ret;
priv_lock(priv);
DEBUG("%p: closing device \"%s\"",
priv_mac_addrs_disable(priv);
priv_destroy_hash_rxqs(priv);
- /* Remove flow director elements. */
- priv_fdir_disable(priv);
- priv_fdir_delete_filters_list(priv);
-
/* Prevent crashes when queues are still in use. */
dev->rx_pkt_burst = removed_rx_burst;
dev->tx_pkt_burst = removed_tx_burst;
/* XXX race condition if mlx5_rx_burst() is still running. */
usleep(1000);
for (i = 0; (i != priv->rxqs_n); ++i) {
- struct rxq *rxq = (*priv->rxqs)[i];
- struct rxq_ctrl *rxq_ctrl;
+ struct mlx5_rxq_data *rxq = (*priv->rxqs)[i];
+ struct mlx5_rxq_ctrl *rxq_ctrl;
if (rxq == NULL)
continue;
- rxq_ctrl = container_of(rxq, struct rxq_ctrl, rxq);
+ rxq_ctrl = container_of(rxq, struct mlx5_rxq_ctrl, rxq);
(*priv->rxqs)[i] = NULL;
- rxq_cleanup(rxq_ctrl);
+ mlx5_rxq_cleanup(rxq_ctrl);
rte_free(rxq_ctrl);
}
priv->rxqs_n = 0;
/* XXX race condition if mlx5_tx_burst() is still running. */
usleep(1000);
for (i = 0; (i != priv->txqs_n); ++i) {
- struct txq *txq = (*priv->txqs)[i];
- struct txq_ctrl *txq_ctrl;
+ struct mlx5_txq_data *txq = (*priv->txqs)[i];
+ struct mlx5_txq_ctrl *txq_ctrl;
if (txq == NULL)
continue;
- txq_ctrl = container_of(txq, struct txq_ctrl, txq);
+ txq_ctrl = container_of(txq, struct mlx5_txq_ctrl, txq);
(*priv->txqs)[i] = NULL;
- txq_cleanup(txq_ctrl);
+ mlx5_txq_cleanup(txq_ctrl);
rte_free(txq_ctrl);
}
priv->txqs_n = 0;
}
if (priv->reta_idx != NULL)
rte_free(priv->reta_idx);
+ priv_socket_uninit(priv);
+ ret = mlx5_priv_rxq_ibv_verify(priv);
+ if (ret)
+ WARN("%p: some Verbs Rx queue still remain", (void *)priv);
+ ret = mlx5_priv_txq_ibv_verify(priv);
+ if (ret)
+ WARN("%p: some Verbs Tx queue still remain", (void *)priv);
+ ret = priv_flow_verify(priv);
+ if (ret)
+ WARN("%p: some flows still remain", (void *)priv);
+ ret = priv_mr_verify(priv);
+ if (ret)
+ WARN("%p: some Memory Region still remain", (void *)priv);
priv_unlock(priv);
memset(priv, 0, sizeof(*priv));
}
.filter_ctrl = mlx5_dev_filter_ctrl,
.rx_descriptor_status = mlx5_rx_descriptor_status,
.tx_descriptor_status = mlx5_tx_descriptor_status,
-#ifdef HAVE_UPDATE_CQ_CI
.rx_queue_intr_enable = mlx5_rx_intr_enable,
.rx_queue_intr_disable = mlx5_rx_intr_disable,
-#endif
+};
+
+
+static const struct eth_dev_ops mlx5_dev_sec_ops = {
+ .stats_get = mlx5_stats_get,
+ .stats_reset = mlx5_stats_reset,
+ .xstats_get = mlx5_xstats_get,
+ .xstats_reset = mlx5_xstats_reset,
+ .xstats_get_names = mlx5_xstats_get_names,
+ .dev_infos_get = mlx5_dev_infos_get,
+ .rx_descriptor_status = mlx5_rx_descriptor_status,
+ .tx_descriptor_status = mlx5_tx_descriptor_status,
};
static struct {
struct ibv_device *ibv_dev;
int err = 0;
struct ibv_context *attr_ctx = NULL;
- struct ibv_device_attr device_attr;
+ struct ibv_device_attr_ex device_attr;
unsigned int sriov;
unsigned int mps;
- unsigned int tunnel_en;
+ unsigned int tunnel_en = 0;
int idx;
int i;
+ struct mlx5dv_context attrs_out;
(void)pci_drv;
assert(pci_drv == &mlx5_driver);
PCI_DEVICE_ID_MELLANOX_CONNECTX5VF) ||
(pci_dev->id.device_id ==
PCI_DEVICE_ID_MELLANOX_CONNECTX5EXVF));
- /*
- * Multi-packet send is supported by ConnectX-4 Lx PF as well
- * as all ConnectX-5 devices.
- */
switch (pci_dev->id.device_id) {
case PCI_DEVICE_ID_MELLANOX_CONNECTX4:
tunnel_en = 1;
- mps = MLX5_MPW_DISABLED;
break;
case PCI_DEVICE_ID_MELLANOX_CONNECTX4LX:
- mps = MLX5_MPW;
- break;
case PCI_DEVICE_ID_MELLANOX_CONNECTX5:
case PCI_DEVICE_ID_MELLANOX_CONNECTX5VF:
case PCI_DEVICE_ID_MELLANOX_CONNECTX5EX:
case PCI_DEVICE_ID_MELLANOX_CONNECTX5EXVF:
tunnel_en = 1;
- mps = MLX5_MPW_ENHANCED;
break;
default:
- mps = MLX5_MPW_DISABLED;
+ break;
}
INFO("PCI information matches, using device \"%s\""
- " (SR-IOV: %s, %sMPS: %s)",
+ " (SR-IOV: %s)",
list[i]->name,
- sriov ? "true" : "false",
- mps == MLX5_MPW_ENHANCED ? "Enhanced " : "",
- mps != MLX5_MPW_DISABLED ? "true" : "false");
+ sriov ? "true" : "false");
attr_ctx = ibv_open_device(list[i]);
err = errno;
break;
ibv_dev = list[i];
DEBUG("device opened");
- if (ibv_query_device(attr_ctx, &device_attr))
+ /*
+ * Multi-packet send is supported by ConnectX-4 Lx PF as well
+ * as all ConnectX-5 devices.
+ */
+ mlx5dv_query_device(attr_ctx, &attrs_out);
+ if (attrs_out.flags & (MLX5DV_CONTEXT_FLAGS_ENHANCED_MPW |
+ MLX5DV_CONTEXT_FLAGS_MPW_ALLOWED)) {
+ INFO("Enhanced MPW is detected\n");
+ mps = MLX5_MPW_ENHANCED;
+ } else if (attrs_out.flags & MLX5DV_CONTEXT_FLAGS_MPW_ALLOWED) {
+ INFO("MPW is detected\n");
+ mps = MLX5_MPW;
+ } else {
+ INFO("MPW is disabled\n");
+ mps = MLX5_MPW_DISABLED;
+ }
+ if (ibv_query_device_ex(attr_ctx, NULL, &device_attr))
goto error;
- INFO("%u port(s) detected", device_attr.phys_port_cnt);
+ INFO("%u port(s) detected", device_attr.orig_attr.phys_port_cnt);
- for (i = 0; i < device_attr.phys_port_cnt; i++) {
+ for (i = 0; i < device_attr.orig_attr.phys_port_cnt; i++) {
uint32_t port = i + 1; /* ports are indexed from one */
uint32_t test = (1 << i);
struct ibv_context *ctx = NULL;
struct ibv_pd *pd = NULL;
struct priv *priv = NULL;
struct rte_eth_dev *eth_dev;
- struct ibv_exp_device_attr exp_device_attr;
+ struct ibv_device_attr_ex device_attr_ex;
struct ether_addr mac;
uint16_t num_vfs = 0;
struct mlx5_args args = {
.rx_vec_en = MLX5_ARG_UNSET,
};
- exp_device_attr.comp_mask =
- IBV_EXP_DEVICE_ATTR_EXP_CAP_FLAGS |
- IBV_EXP_DEVICE_ATTR_RX_HASH |
- IBV_EXP_DEVICE_ATTR_VLAN_OFFLOADS |
- IBV_EXP_DEVICE_ATTR_RX_PAD_END_ALIGN |
- IBV_EXP_DEVICE_ATTR_TSO_CAPS |
- 0;
+ mlx5_dev[idx].ports |= test;
+
+ if (mlx5_is_secondary()) {
+ /* from rte_ethdev.c */
+ char name[RTE_ETH_NAME_MAX_LEN];
+
+ snprintf(name, sizeof(name), "%s port %u",
+ ibv_get_device_name(ibv_dev), port);
+ eth_dev = rte_eth_dev_attach_secondary(name);
+ if (eth_dev == NULL) {
+ ERROR("can not attach rte ethdev");
+ err = ENOMEM;
+ goto error;
+ }
+ eth_dev->device = &pci_dev->device;
+ eth_dev->dev_ops = &mlx5_dev_sec_ops;
+ priv = eth_dev->data->dev_private;
+ /* Receive command fd from primary process */
+ err = priv_socket_connect(priv);
+ if (err < 0) {
+ err = -err;
+ goto error;
+ }
+ /* Remap UAR for Tx queues. */
+ err = priv_tx_uar_remap(priv, err);
+ if (err < 0) {
+ err = -err;
+ goto error;
+ }
+ priv_dev_select_rx_function(priv, eth_dev);
+ priv_dev_select_tx_function(priv, eth_dev);
+ continue;
+ }
DEBUG("using port %u (%08" PRIx32 ")", port, test);
ctx = ibv_open_device(ibv_dev);
- if (ctx == NULL)
+ if (ctx == NULL) {
+ err = ENODEV;
goto port_error;
+ }
/* Check port status. */
err = ibv_query_port(ctx, port, &port_attr);
if (port_attr.link_layer != IBV_LINK_LAYER_ETHERNET) {
ERROR("port %d is not configured in Ethernet mode",
port);
+ err = EINVAL;
goto port_error;
}
}
priv->ctx = ctx;
+ strncpy(priv->ibdev_path, priv->ctx->device->ibdev_path,
+ sizeof(priv->ibdev_path));
priv->device_attr = device_attr;
priv->port = port;
priv->pd = pd;
goto port_error;
}
mlx5_args_assign(priv, &args);
- if (ibv_exp_query_device(ctx, &exp_device_attr)) {
- ERROR("ibv_exp_query_device() failed");
+ if (ibv_query_device_ex(ctx, NULL, &device_attr_ex)) {
+ ERROR("ibv_query_device_ex() failed");
goto port_error;
}
priv->hw_csum =
- ((exp_device_attr.exp_device_cap_flags &
- IBV_EXP_DEVICE_RX_CSUM_TCP_UDP_PKT) &&
- (exp_device_attr.exp_device_cap_flags &
- IBV_EXP_DEVICE_RX_CSUM_IP_PKT));
+ !!(device_attr_ex.device_cap_flags_ex &
+ IBV_DEVICE_RAW_IP_CSUM);
DEBUG("checksum offloading is %ssupported",
(priv->hw_csum ? "" : "not "));
+#ifdef HAVE_IBV_DEVICE_VXLAN_SUPPORT
priv->hw_csum_l2tun = !!(exp_device_attr.exp_device_cap_flags &
- IBV_EXP_DEVICE_VXLAN_SUPPORT);
+ IBV_DEVICE_VXLAN_SUPPORT);
+#endif
DEBUG("L2 tunnel checksum offloads are %ssupported",
(priv->hw_csum_l2tun ? "" : "not "));
- priv->ind_table_max_size = exp_device_attr.rx_hash_caps.max_rwq_indirection_table_size;
+ priv->ind_table_max_size =
+ device_attr_ex.rss_caps.max_rwq_indirection_table_size;
/* Remove this check once DPDK supports larger/variable
* indirection tables. */
if (priv->ind_table_max_size >
priv->ind_table_max_size = ETH_RSS_RETA_SIZE_512;
DEBUG("maximum RX indirection table size is %u",
priv->ind_table_max_size);
- priv->hw_vlan_strip = !!(exp_device_attr.wq_vlan_offloads_cap &
- IBV_EXP_RECEIVE_WQ_CVLAN_STRIP);
+ priv->hw_vlan_strip = !!(device_attr_ex.raw_packet_caps &
+ IBV_RAW_PACKET_CAP_CVLAN_STRIPPING);
DEBUG("VLAN stripping is %ssupported",
(priv->hw_vlan_strip ? "" : "not "));
- priv->hw_fcs_strip = !!(exp_device_attr.exp_device_cap_flags &
- IBV_EXP_DEVICE_SCATTER_FCS);
+ priv->hw_fcs_strip =
+ !!(device_attr_ex.orig_attr.device_cap_flags &
+ IBV_WQ_FLAGS_SCATTER_FCS);
DEBUG("FCS stripping configuration is %ssupported",
(priv->hw_fcs_strip ? "" : "not "));
- priv->hw_padding = !!exp_device_attr.rx_pad_end_addr_align;
+#ifdef HAVE_IBV_WQ_FLAG_RX_END_PADDING
+ priv->hw_padding = !!device_attr_ex.rx_pad_end_addr_align;
+#endif
DEBUG("hardware RX end alignment padding is %ssupported",
(priv->hw_padding ? "" : "not "));
priv_get_num_vfs(priv, &num_vfs);
priv->sriov = (num_vfs || sriov);
priv->tso = ((priv->tso) &&
- (exp_device_attr.tso_caps.max_tso > 0) &&
- (exp_device_attr.tso_caps.supported_qpts &
- (1 << IBV_QPT_RAW_ETH)));
+ (device_attr_ex.tso_caps.max_tso > 0) &&
+ (device_attr_ex.tso_caps.supported_qpts &
+ (1 << IBV_QPT_RAW_PACKET)));
if (priv->tso)
priv->max_tso_payload_sz =
- exp_device_attr.tso_caps.max_tso;
+ device_attr_ex.tso_caps.max_tso;
if (priv->mps && !mps) {
ERROR("multi-packet send not supported on this device"
" (" MLX5_TXQ_MPW_EN ")");
if (priv_get_mac(priv, &mac.addr_bytes)) {
ERROR("cannot get MAC address, is mlx5_en loaded?"
" (errno: %s)", strerror(errno));
+ err = ENODEV;
goto port_error;
}
INFO("port %u MAC address is %02x:%02x:%02x:%02x:%02x:%02x",
claim_zero(priv_mac_addr_add(priv, 0,
(const uint8_t (*)[ETHER_ADDR_LEN])
mac.addr_bytes));
- /* Initialize FD filters list. */
- err = fdir_init_filters_list(priv);
- if (err)
- goto port_error;
#ifndef NDEBUG
{
char ifname[IF_NAMESIZE];
err = ENOMEM;
goto port_error;
}
-
- /* Secondary processes have to use local storage for their
- * private data as well as a copy of eth_dev->data, but this
- * pointer must not be modified before burst functions are
- * actually called. */
- if (mlx5_is_secondary()) {
- struct mlx5_secondary_data *sd =
- &mlx5_secondary_data[eth_dev->data->port_id];
- sd->primary_priv = eth_dev->data->dev_private;
- if (sd->primary_priv == NULL) {
- ERROR("no private data for port %u",
- eth_dev->data->port_id);
- err = EINVAL;
- goto port_error;
- }
- sd->shared_dev_data = eth_dev->data;
- rte_spinlock_init(&sd->lock);
- memcpy(sd->data.name, sd->shared_dev_data->name,
- sizeof(sd->data.name));
- sd->data.dev_private = priv;
- sd->data.rx_mbuf_alloc_failed = 0;
- sd->data.mtu = ETHER_MTU;
- sd->data.port_id = sd->shared_dev_data->port_id;
- sd->data.mac_addrs = priv->mac;
- eth_dev->tx_pkt_burst = mlx5_tx_burst_secondary_setup;
- eth_dev->rx_pkt_burst = mlx5_rx_burst_secondary_setup;
- } else {
- eth_dev->data->dev_private = priv;
- eth_dev->data->mac_addrs = priv->mac;
- }
-
+ eth_dev->data->dev_private = priv;
+ eth_dev->data->mac_addrs = priv->mac;
eth_dev->device = &pci_dev->device;
rte_eth_copy_pci_info(eth_dev, pci_dev);
eth_dev->data->dev_flags |= RTE_ETH_DEV_DETACHABLE;
eth_dev->dev_ops = &mlx5_dev_ops;
TAILQ_INIT(&priv->flows);
+ /* Hint libmlx5 to use PMD allocator for data plane resources */
+ struct mlx5dv_ctx_allocators alctr = {
+ .alloc = &mlx5_alloc_verbs_buf,
+ .free = &mlx5_free_verbs_buf,
+ .data = priv,
+ };
+ mlx5dv_set_context_attr(ctx, MLX5DV_CTX_ATTR_BUF_ALLOCATORS,
+ (void *)((uintptr_t)&alctr));
+
/* Bring Ethernet device up. */
DEBUG("forcing Ethernet interface up");
priv_set_flags(priv, ~IFF_UP, IFF_UP);
},
.id_table = mlx5_pci_id_map,
.probe = mlx5_pci_probe,
- .drv_flags = RTE_PCI_DRV_INTR_LSC,
+ .drv_flags = RTE_PCI_DRV_INTR_LSC | RTE_PCI_DRV_INTR_RMV,
};
/**
* using this PMD, which is not supported in forked processes.
*/
setenv("RDMAV_HUGEPAGES_SAFE", "1", 1);
+ /* Don't map UAR to WC if BlueFlame is not used.*/
+ setenv("MLX5_SHUT_UP_BF", "1", 1);
ibv_fork_init();
rte_pci_register(&mlx5_driver);
}
git.droids-corp.org / dpdk.git / blobdiff