-/*-
- * BSD LICENSE
- *
- * Copyright 2015 6WIND S.A.
- * Copyright 2015 Mellanox.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions
- * are met:
- *
- * * Redistributions of source code must retain the above copyright
- * notice, this list of conditions and the following disclaimer.
- * * Redistributions in binary form must reproduce the above copyright
- * notice, this list of conditions and the following disclaimer in
- * the documentation and/or other materials provided with the
- * distribution.
- * * Neither the name of 6WIND S.A. nor the names of its
- * contributors may be used to endorse or promote products derived
- * from this software without specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
- * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
- * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
- * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
- * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
- * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
- * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
- * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
- * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
- * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
- * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+/* SPDX-License-Identifier: BSD-3-Clause
+ * Copyright 2015 6WIND S.A.
+ * Copyright 2015 Mellanox.
*/
+#define _GNU_SOURCE
+
#include <stddef.h>
#include <assert.h>
+#include <inttypes.h>
#include <unistd.h>
#include <stdint.h>
#include <stdio.h>
#include <net/if.h>
#include <sys/ioctl.h>
#include <sys/socket.h>
+#include <sys/utsname.h>
#include <netinet/in.h>
-#include <linux/if.h>
#include <linux/ethtool.h>
#include <linux/sockios.h>
+#include <linux/version.h>
+#include <fcntl.h>
+#include <stdalign.h>
+#include <sys/un.h>
-/* DPDK headers don't like -pedantic. */
-#ifdef PEDANTIC
-#pragma GCC diagnostic ignored "-pedantic"
-#endif
#include <rte_atomic.h>
-#include <rte_ethdev.h>
+#include <rte_ethdev_driver.h>
+#include <rte_bus_pci.h>
#include <rte_mbuf.h>
#include <rte_common.h>
-#ifdef PEDANTIC
-#pragma GCC diagnostic error "-pedantic"
-#endif
+#include <rte_interrupts.h>
+#include <rte_alarm.h>
+#include <rte_malloc.h>
#include "mlx5.h"
+#include "mlx5_glue.h"
#include "mlx5_rxtx.h"
#include "mlx5_utils.h"
+/* Add defines in case the running kernel is not the same as user headers. */
+#ifndef ETHTOOL_GLINKSETTINGS
+struct ethtool_link_settings {
+ uint32_t cmd;
+ uint32_t speed;
+ uint8_t duplex;
+ uint8_t port;
+ uint8_t phy_address;
+ uint8_t autoneg;
+ uint8_t mdio_support;
+ uint8_t eth_to_mdix;
+ uint8_t eth_tp_mdix_ctrl;
+ int8_t link_mode_masks_nwords;
+ uint32_t reserved[8];
+ uint32_t link_mode_masks[];
+};
+
+#define ETHTOOL_GLINKSETTINGS 0x0000004c
+#define ETHTOOL_LINK_MODE_1000baseT_Full_BIT 5
+#define ETHTOOL_LINK_MODE_Autoneg_BIT 6
+#define ETHTOOL_LINK_MODE_1000baseKX_Full_BIT 17
+#define ETHTOOL_LINK_MODE_10000baseKX4_Full_BIT 18
+#define ETHTOOL_LINK_MODE_10000baseKR_Full_BIT 19
+#define ETHTOOL_LINK_MODE_10000baseR_FEC_BIT 20
+#define ETHTOOL_LINK_MODE_20000baseMLD2_Full_BIT 21
+#define ETHTOOL_LINK_MODE_20000baseKR2_Full_BIT 22
+#define ETHTOOL_LINK_MODE_40000baseKR4_Full_BIT 23
+#define ETHTOOL_LINK_MODE_40000baseCR4_Full_BIT 24
+#define ETHTOOL_LINK_MODE_40000baseSR4_Full_BIT 25
+#define ETHTOOL_LINK_MODE_40000baseLR4_Full_BIT 26
+#define ETHTOOL_LINK_MODE_56000baseKR4_Full_BIT 27
+#define ETHTOOL_LINK_MODE_56000baseCR4_Full_BIT 28
+#define ETHTOOL_LINK_MODE_56000baseSR4_Full_BIT 29
+#define ETHTOOL_LINK_MODE_56000baseLR4_Full_BIT 30
+#endif
+#ifndef HAVE_ETHTOOL_LINK_MODE_25G
+#define ETHTOOL_LINK_MODE_25000baseCR_Full_BIT 31
+#define ETHTOOL_LINK_MODE_25000baseKR_Full_BIT 32
+#define ETHTOOL_LINK_MODE_25000baseSR_Full_BIT 33
+#endif
+#ifndef HAVE_ETHTOOL_LINK_MODE_50G
+#define ETHTOOL_LINK_MODE_50000baseCR2_Full_BIT 34
+#define ETHTOOL_LINK_MODE_50000baseKR2_Full_BIT 35
+#endif
+#ifndef HAVE_ETHTOOL_LINK_MODE_100G
+#define ETHTOOL_LINK_MODE_100000baseKR4_Full_BIT 36
+#define ETHTOOL_LINK_MODE_100000baseSR4_Full_BIT 37
+#define ETHTOOL_LINK_MODE_100000baseCR4_Full_BIT 38
+#define ETHTOOL_LINK_MODE_100000baseLR4_ER4_Full_BIT 39
+#endif
+
/**
* Get interface name from private structure.
*
char match[IF_NAMESIZE] = "";
{
- MKSTR(path, "%s/device/net", priv->ctx->device->ibdev_path);
+ MKSTR(path, "%s/device/net", priv->ibdev_path);
dir = opendir(path);
if (dir == NULL)
continue;
MKSTR(path, "%s/device/net/%s/%s",
- priv->ctx->device->ibdev_path, name,
+ priv->ibdev_path, name,
(dev_type ? "dev_id" : "dev_port"));
file = fopen(path, "rb");
return 0;
}
-/**
- * Read from sysfs entry.
- *
- * @param[in] priv
- * Pointer to private structure.
- * @param[in] entry
- * Entry name relative to sysfs path.
- * @param[out] buf
- * Data output buffer.
- * @param size
- * Buffer size.
- *
- * @return
- * 0 on success, -1 on failure and errno is set.
- */
-static int
-priv_sysfs_read(const struct priv *priv, const char *entry,
- char *buf, size_t size)
-{
- char ifname[IF_NAMESIZE];
- FILE *file;
- int ret;
- int err;
-
- if (priv_get_ifname(priv, &ifname))
- return -1;
-
- MKSTR(path, "%s/device/net/%s/%s", priv->ctx->device->ibdev_path,
- ifname, entry);
-
- file = fopen(path, "rb");
- if (file == NULL)
- return -1;
- ret = fread(buf, 1, size, file);
- err = errno;
- if (((size_t)ret < size) && (ferror(file)))
- ret = -1;
- else
- ret = size;
- fclose(file);
- errno = err;
- return ret;
-}
-
-/**
- * Write to sysfs entry.
- *
- * @param[in] priv
- * Pointer to private structure.
- * @param[in] entry
- * Entry name relative to sysfs path.
- * @param[in] buf
- * Data buffer.
- * @param size
- * Buffer size.
- *
- * @return
- * 0 on success, -1 on failure and errno is set.
- */
-static int
-priv_sysfs_write(const struct priv *priv, const char *entry,
- char *buf, size_t size)
-{
- char ifname[IF_NAMESIZE];
- FILE *file;
- int ret;
- int err;
-
- if (priv_get_ifname(priv, &ifname))
- return -1;
-
- MKSTR(path, "%s/device/net/%s/%s", priv->ctx->device->ibdev_path,
- ifname, entry);
-
- file = fopen(path, "wb");
- if (file == NULL)
- return -1;
- ret = fwrite(buf, 1, size, file);
- err = errno;
- if (((size_t)ret < size) || (ferror(file)))
- ret = -1;
- else
- ret = size;
- fclose(file);
- errno = err;
- return ret;
-}
-
-/**
- * Get unsigned long sysfs property.
- *
- * @param priv
- * Pointer to private structure.
- * @param[in] name
- * Entry name relative to sysfs path.
- * @param[out] value
- * Value output buffer.
- *
- * @return
- * 0 on success, -1 on failure and errno is set.
- */
-static int
-priv_get_sysfs_ulong(struct priv *priv, const char *name, unsigned long *value)
-{
- int ret;
- unsigned long value_ret;
- char value_str[32];
-
- ret = priv_sysfs_read(priv, name, value_str, (sizeof(value_str) - 1));
- if (ret == -1) {
- DEBUG("cannot read %s value from sysfs: %s",
- name, strerror(errno));
- return -1;
- }
- value_str[ret] = '\0';
- errno = 0;
- value_ret = strtoul(value_str, NULL, 0);
- if (errno) {
- DEBUG("invalid %s value `%s': %s", name, value_str,
- strerror(errno));
- return -1;
- }
- *value = value_ret;
- return 0;
-}
-
-/**
- * Set unsigned long sysfs property.
- *
- * @param priv
- * Pointer to private structure.
- * @param[in] name
- * Entry name relative to sysfs path.
- * @param value
- * Value to set.
- *
- * @return
- * 0 on success, -1 on failure and errno is set.
- */
-static int
-priv_set_sysfs_ulong(struct priv *priv, const char *name, unsigned long value)
-{
- int ret;
- MKSTR(value_str, "%lu", value);
-
- ret = priv_sysfs_write(priv, name, value_str, (sizeof(value_str) - 1));
- if (ret == -1) {
- DEBUG("cannot write %s `%s' (%lu) to sysfs: %s",
- name, value_str, value, strerror(errno));
- return -1;
- }
- return 0;
-}
-
/**
* Perform ifreq ioctl() on associated Ethernet device.
*
int
priv_get_mtu(struct priv *priv, uint16_t *mtu)
{
- unsigned long ulong_mtu;
+ struct ifreq request;
+ int ret = priv_ifreq(priv, SIOCGIFMTU, &request);
- if (priv_get_sysfs_ulong(priv, "mtu", &ulong_mtu) == -1)
- return -1;
- *mtu = ulong_mtu;
+ if (ret)
+ return ret;
+ *mtu = request.ifr_mtu;
return 0;
}
static int
priv_set_mtu(struct priv *priv, uint16_t mtu)
{
- return priv_set_sysfs_ulong(priv, "mtu", mtu);
+ struct ifreq request = { .ifr_mtu = mtu, };
+
+ return priv_ifreq(priv, SIOCSIFMTU, &request);
}
/**
int
priv_set_flags(struct priv *priv, unsigned int keep, unsigned int flags)
{
- unsigned long tmp;
+ struct ifreq request;
+ int ret = priv_ifreq(priv, SIOCGIFFLAGS, &request);
- if (priv_get_sysfs_ulong(priv, "flags", &tmp) == -1)
- return -1;
- tmp &= keep;
- tmp |= flags;
- return priv_set_sysfs_ulong(priv, "flags", tmp);
+ if (ret)
+ return ret;
+ request.ifr_flags &= keep;
+ request.ifr_flags |= flags & ~keep;
+ return priv_ifreq(priv, SIOCSIFFLAGS, &request);
}
/**
- * Ethernet device configuration.
- *
- * Prepare the driver for a given number of TX and RX queues.
- * Allocate parent RSS queue when several RX queues are requested.
+ * DPDK callback for Ethernet device configuration.
*
* @param dev
* Pointer to Ethernet device structure.
*
* @return
- * 0 on success, errno value on failure.
+ * 0 on success, negative errno value on failure.
*/
-static int
-dev_configure(struct rte_eth_dev *dev)
+int
+mlx5_dev_configure(struct rte_eth_dev *dev)
{
struct priv *priv = dev->data->dev_private;
unsigned int rxqs_n = dev->data->nb_rx_queues;
unsigned int txqs_n = dev->data->nb_tx_queues;
- unsigned int tmp;
- int ret;
+ unsigned int i;
+ unsigned int j;
+ unsigned int reta_idx_n;
+ const uint8_t use_app_rss_key =
+ !!dev->data->dev_conf.rx_adv_conf.rss_conf.rss_key;
+ uint64_t supp_tx_offloads = mlx5_priv_get_tx_port_offloads(priv);
+ uint64_t tx_offloads = dev->data->dev_conf.txmode.offloads;
+ uint64_t supp_rx_offloads =
+ (mlx5_priv_get_rx_port_offloads(priv) |
+ mlx5_priv_get_rx_queue_offloads(priv));
+ uint64_t rx_offloads = dev->data->dev_conf.rxmode.offloads;
+ if ((tx_offloads & supp_tx_offloads) != tx_offloads) {
+ ERROR("Some Tx offloads are not supported "
+ "requested 0x%" PRIx64 " supported 0x%" PRIx64,
+ tx_offloads, supp_tx_offloads);
+ return ENOTSUP;
+ }
+ if ((rx_offloads & supp_rx_offloads) != rx_offloads) {
+ ERROR("Some Rx offloads are not supported "
+ "requested 0x%" PRIx64 " supported 0x%" PRIx64,
+ rx_offloads, supp_rx_offloads);
+ return ENOTSUP;
+ }
+ if (use_app_rss_key &&
+ (dev->data->dev_conf.rx_adv_conf.rss_conf.rss_key_len !=
+ rss_hash_default_key_len)) {
+ /* MLX5 RSS only support 40bytes key. */
+ return EINVAL;
+ }
+ priv->rss_conf.rss_key =
+ rte_realloc(priv->rss_conf.rss_key,
+ rss_hash_default_key_len, 0);
+ if (!priv->rss_conf.rss_key) {
+ ERROR("cannot allocate RSS hash key memory (%u)", rxqs_n);
+ return ENOMEM;
+ }
+ memcpy(priv->rss_conf.rss_key,
+ use_app_rss_key ?
+ dev->data->dev_conf.rx_adv_conf.rss_conf.rss_key :
+ rss_hash_default_key,
+ rss_hash_default_key_len);
+ priv->rss_conf.rss_key_len = rss_hash_default_key_len;
+ priv->rss_conf.rss_hf = dev->data->dev_conf.rx_adv_conf.rss_conf.rss_hf;
priv->rxqs = (void *)dev->data->rx_queues;
priv->txqs = (void *)dev->data->tx_queues;
if (txqs_n != priv->txqs_n) {
(void *)dev, priv->txqs_n, txqs_n);
priv->txqs_n = txqs_n;
}
+ if (rxqs_n > priv->config.ind_table_max_size) {
+ ERROR("cannot handle this many RX queues (%u)", rxqs_n);
+ return EINVAL;
+ }
if (rxqs_n == priv->rxqs_n)
return 0;
INFO("%p: RX queues number update: %u -> %u",
(void *)dev, priv->rxqs_n, rxqs_n);
- /* If RSS is enabled, disable it first. */
- if (priv->rss) {
- unsigned int i;
-
- /* Only if there are no remaining child RX queues. */
- for (i = 0; (i != priv->rxqs_n); ++i)
- if ((*priv->rxqs)[i] != NULL)
- return EINVAL;
- rxq_cleanup(&priv->rxq_parent);
- priv->rss = 0;
- priv->rxqs_n = 0;
- }
- if (rxqs_n <= 1) {
- /* Nothing else to do. */
- priv->rxqs_n = rxqs_n;
- return 0;
- }
- /* Allocate a new RSS parent queue if supported by hardware. */
- if (!priv->hw_rss) {
- ERROR("%p: only a single RX queue can be configured when"
- " hardware doesn't support RSS",
- (void *)dev);
- return EINVAL;
- }
- /* Fail if hardware doesn't support that many RSS queues. */
- if (rxqs_n >= priv->max_rss_tbl_sz) {
- ERROR("%p: only %u RX queues can be configured for RSS",
- (void *)dev, priv->max_rss_tbl_sz);
- return EINVAL;
- }
- priv->rss = 1;
- tmp = priv->rxqs_n;
priv->rxqs_n = rxqs_n;
- ret = rxq_setup(dev, &priv->rxq_parent, 0, 0, NULL, NULL);
- if (!ret)
- return 0;
- /* Failure, rollback. */
- priv->rss = 0;
- priv->rxqs_n = tmp;
- assert(ret > 0);
- return ret;
-}
-
-/**
- * DPDK callback for Ethernet device configuration.
- *
- * @param dev
- * Pointer to Ethernet device structure.
- *
- * @return
- * 0 on success, negative errno value on failure.
- */
-int
-mlx5_dev_configure(struct rte_eth_dev *dev)
-{
- struct priv *priv = dev->data->dev_private;
- int ret;
+ /* If the requested number of RX queues is not a power of two, use the
+ * maximum indirection table size for better balancing.
+ * The result is always rounded to the next power of two. */
+ reta_idx_n = (1 << log2above((rxqs_n & (rxqs_n - 1)) ?
+ priv->config.ind_table_max_size :
+ rxqs_n));
+ if (priv_rss_reta_index_resize(priv, reta_idx_n))
+ return ENOMEM;
+ /* When the number of RX queues is not a power of two, the remaining
+ * table entries are padded with reused WQs and hashes are not spread
+ * uniformly. */
+ for (i = 0, j = 0; (i != reta_idx_n); ++i) {
+ (*priv->reta_idx)[i] = j;
+ if (++j == rxqs_n)
+ j = 0;
+ }
+ return 0;
- priv_lock(priv);
- ret = dev_configure(dev);
- assert(ret >= 0);
- priv_unlock(priv);
- return -ret;
}
/**
mlx5_dev_infos_get(struct rte_eth_dev *dev, struct rte_eth_dev_info *info)
{
struct priv *priv = dev->data->dev_private;
+ struct mlx5_dev_config *config = &priv->config;
unsigned int max;
char ifname[IF_NAMESIZE];
- priv_lock(priv);
+ info->pci_dev = RTE_ETH_DEV_TO_PCI(dev);
/* FIXME: we should ask the device for these values. */
info->min_rx_bufsize = 32;
info->max_rx_pktlen = 65536;
* Since we need one CQ per QP, the limit is the minimum number
* between the two values.
*/
- max = ((priv->device_attr.max_cq > priv->device_attr.max_qp) ?
- priv->device_attr.max_qp : priv->device_attr.max_cq);
+ max = RTE_MIN(priv->device_attr.orig_attr.max_cq,
+ priv->device_attr.orig_attr.max_qp);
/* If max >= 65535 then max = 0, max_rx_queues is uint16_t. */
if (max >= 65535)
max = 65535;
info->max_rx_queues = max;
info->max_tx_queues = max;
- /* Last array entry is reserved for broadcast. */
- info->max_mac_addrs = (RTE_DIM(priv->mac) - 1);
- info->rx_offload_capa =
- (priv->hw_csum ?
- (DEV_RX_OFFLOAD_IPV4_CKSUM |
- DEV_RX_OFFLOAD_UDP_CKSUM |
- DEV_RX_OFFLOAD_TCP_CKSUM) :
- 0);
- info->tx_offload_capa =
- (priv->hw_csum ?
- (DEV_TX_OFFLOAD_IPV4_CKSUM |
- DEV_TX_OFFLOAD_UDP_CKSUM |
- DEV_TX_OFFLOAD_TCP_CKSUM) :
- 0);
+ info->max_mac_addrs = RTE_DIM(priv->mac);
+ info->rx_queue_offload_capa =
+ mlx5_priv_get_rx_queue_offloads(priv);
+ info->rx_offload_capa = (mlx5_priv_get_rx_port_offloads(priv) |
+ info->rx_queue_offload_capa);
+ info->tx_offload_capa = mlx5_priv_get_tx_port_offloads(priv);
if (priv_get_ifname(priv, &ifname) == 0)
info->if_index = if_nametoindex(ifname);
- priv_unlock(priv);
+ info->reta_size = priv->reta_idx_n ?
+ priv->reta_idx_n : config->ind_table_max_size;
+ info->hash_key_size = priv->rss_conf.rss_key_len;
+ info->speed_capa = priv->link_speed_capa;
+ info->flow_type_rss_offloads = ~MLX5_RSS_HF_MASK;
}
/**
- * DPDK callback to retrieve physical link information (unlocked version).
+ * Get supported packet types.
*
* @param dev
* Pointer to Ethernet device structure.
- * @param wait_to_complete
- * Wait for request completion (ignored).
+ *
+ * @return
+ * A pointer to the supported Packet types array.
+ */
+const uint32_t *
+mlx5_dev_supported_ptypes_get(struct rte_eth_dev *dev)
+{
+ static const uint32_t ptypes[] = {
+ /* refers to rxq_cq_to_pkt_type() */
+ RTE_PTYPE_L2_ETHER,
+ RTE_PTYPE_L3_IPV4_EXT_UNKNOWN,
+ RTE_PTYPE_L3_IPV6_EXT_UNKNOWN,
+ RTE_PTYPE_L4_NONFRAG,
+ RTE_PTYPE_L4_FRAG,
+ RTE_PTYPE_L4_TCP,
+ RTE_PTYPE_L4_UDP,
+ RTE_PTYPE_INNER_L3_IPV4_EXT_UNKNOWN,
+ RTE_PTYPE_INNER_L3_IPV6_EXT_UNKNOWN,
+ RTE_PTYPE_INNER_L4_NONFRAG,
+ RTE_PTYPE_INNER_L4_FRAG,
+ RTE_PTYPE_INNER_L4_TCP,
+ RTE_PTYPE_INNER_L4_UDP,
+ RTE_PTYPE_UNKNOWN
+ };
+
+ if (dev->rx_pkt_burst == mlx5_rx_burst ||
+ dev->rx_pkt_burst == mlx5_rx_burst_vec)
+ return ptypes;
+ return NULL;
+}
+
+/**
+ * DPDK callback to retrieve physical link information.
+ *
+ * @param dev
+ * Pointer to Ethernet device structure.
+ *
+ * @return
+ * 0 on success, -1 on error.
*/
static int
-mlx5_link_update_unlocked(struct rte_eth_dev *dev, int wait_to_complete)
+mlx5_link_update_unlocked_gset(struct rte_eth_dev *dev)
{
struct priv *priv = dev->data->dev_private;
struct ethtool_cmd edata = {
- .cmd = ETHTOOL_GSET
+ .cmd = ETHTOOL_GSET /* Deprecated since Linux v4.5. */
};
struct ifreq ifr;
struct rte_eth_link dev_link;
int link_speed = 0;
- (void)wait_to_complete;
if (priv_ifreq(priv, SIOCGIFFLAGS, &ifr)) {
WARN("ioctl(SIOCGIFFLAGS) failed: %s", strerror(errno));
return -1;
memset(&dev_link, 0, sizeof(dev_link));
dev_link.link_status = ((ifr.ifr_flags & IFF_UP) &&
(ifr.ifr_flags & IFF_RUNNING));
- ifr.ifr_data = &edata;
+ ifr.ifr_data = (void *)&edata;
if (priv_ifreq(priv, SIOCETHTOOL, &ifr)) {
WARN("ioctl(SIOCETHTOOL, ETHTOOL_GSET) failed: %s",
strerror(errno));
dev_link.link_speed = 0;
else
dev_link.link_speed = link_speed;
+ priv->link_speed_capa = 0;
+ if (edata.supported & SUPPORTED_Autoneg)
+ priv->link_speed_capa |= ETH_LINK_SPEED_AUTONEG;
+ if (edata.supported & (SUPPORTED_1000baseT_Full |
+ SUPPORTED_1000baseKX_Full))
+ priv->link_speed_capa |= ETH_LINK_SPEED_1G;
+ if (edata.supported & SUPPORTED_10000baseKR_Full)
+ priv->link_speed_capa |= ETH_LINK_SPEED_10G;
+ if (edata.supported & (SUPPORTED_40000baseKR4_Full |
+ SUPPORTED_40000baseCR4_Full |
+ SUPPORTED_40000baseSR4_Full |
+ SUPPORTED_40000baseLR4_Full))
+ priv->link_speed_capa |= ETH_LINK_SPEED_40G;
dev_link.link_duplex = ((edata.duplex == DUPLEX_HALF) ?
ETH_LINK_HALF_DUPLEX : ETH_LINK_FULL_DUPLEX);
+ dev_link.link_autoneg = !(dev->data->dev_conf.link_speeds &
+ ETH_LINK_SPEED_FIXED);
+ if (memcmp(&dev_link, &dev->data->dev_link, sizeof(dev_link))) {
+ /* Link status changed. */
+ dev->data->dev_link = dev_link;
+ return 0;
+ }
+ /* Link status is still the same. */
+ return -1;
+}
+
+/**
+ * Retrieve physical link information (unlocked version using new ioctl).
+ *
+ * @param dev
+ * Pointer to Ethernet device structure.
+ *
+ * @return
+ * 0 on success, -1 on error.
+ */
+static int
+mlx5_link_update_unlocked_gs(struct rte_eth_dev *dev)
+{
+ struct priv *priv = dev->data->dev_private;
+ struct ethtool_link_settings gcmd = { .cmd = ETHTOOL_GLINKSETTINGS };
+ struct ifreq ifr;
+ struct rte_eth_link dev_link;
+ uint64_t sc;
+
+ if (priv_ifreq(priv, SIOCGIFFLAGS, &ifr)) {
+ WARN("ioctl(SIOCGIFFLAGS) failed: %s", strerror(errno));
+ return -1;
+ }
+ memset(&dev_link, 0, sizeof(dev_link));
+ dev_link.link_status = ((ifr.ifr_flags & IFF_UP) &&
+ (ifr.ifr_flags & IFF_RUNNING));
+ ifr.ifr_data = (void *)&gcmd;
+ if (priv_ifreq(priv, SIOCETHTOOL, &ifr)) {
+ DEBUG("ioctl(SIOCETHTOOL, ETHTOOL_GLINKSETTINGS) failed: %s",
+ strerror(errno));
+ return -1;
+ }
+ gcmd.link_mode_masks_nwords = -gcmd.link_mode_masks_nwords;
+
+ alignas(struct ethtool_link_settings)
+ uint8_t data[offsetof(struct ethtool_link_settings, link_mode_masks) +
+ sizeof(uint32_t) * gcmd.link_mode_masks_nwords * 3];
+ struct ethtool_link_settings *ecmd = (void *)data;
+
+ *ecmd = gcmd;
+ ifr.ifr_data = (void *)ecmd;
+ if (priv_ifreq(priv, SIOCETHTOOL, &ifr)) {
+ DEBUG("ioctl(SIOCETHTOOL, ETHTOOL_GLINKSETTINGS) failed: %s",
+ strerror(errno));
+ return -1;
+ }
+ dev_link.link_speed = ecmd->speed;
+ sc = ecmd->link_mode_masks[0] |
+ ((uint64_t)ecmd->link_mode_masks[1] << 32);
+ priv->link_speed_capa = 0;
+ if (sc & MLX5_BITSHIFT(ETHTOOL_LINK_MODE_Autoneg_BIT))
+ priv->link_speed_capa |= ETH_LINK_SPEED_AUTONEG;
+ if (sc & (MLX5_BITSHIFT(ETHTOOL_LINK_MODE_1000baseT_Full_BIT) |
+ MLX5_BITSHIFT(ETHTOOL_LINK_MODE_1000baseKX_Full_BIT)))
+ priv->link_speed_capa |= ETH_LINK_SPEED_1G;
+ if (sc & (MLX5_BITSHIFT(ETHTOOL_LINK_MODE_10000baseKX4_Full_BIT) |
+ MLX5_BITSHIFT(ETHTOOL_LINK_MODE_10000baseKR_Full_BIT) |
+ MLX5_BITSHIFT(ETHTOOL_LINK_MODE_10000baseR_FEC_BIT)))
+ priv->link_speed_capa |= ETH_LINK_SPEED_10G;
+ if (sc & (MLX5_BITSHIFT(ETHTOOL_LINK_MODE_20000baseMLD2_Full_BIT) |
+ MLX5_BITSHIFT(ETHTOOL_LINK_MODE_20000baseKR2_Full_BIT)))
+ priv->link_speed_capa |= ETH_LINK_SPEED_20G;
+ if (sc & (MLX5_BITSHIFT(ETHTOOL_LINK_MODE_40000baseKR4_Full_BIT) |
+ MLX5_BITSHIFT(ETHTOOL_LINK_MODE_40000baseCR4_Full_BIT) |
+ MLX5_BITSHIFT(ETHTOOL_LINK_MODE_40000baseSR4_Full_BIT) |
+ MLX5_BITSHIFT(ETHTOOL_LINK_MODE_40000baseLR4_Full_BIT)))
+ priv->link_speed_capa |= ETH_LINK_SPEED_40G;
+ if (sc & (MLX5_BITSHIFT(ETHTOOL_LINK_MODE_56000baseKR4_Full_BIT) |
+ MLX5_BITSHIFT(ETHTOOL_LINK_MODE_56000baseCR4_Full_BIT) |
+ MLX5_BITSHIFT(ETHTOOL_LINK_MODE_56000baseSR4_Full_BIT) |
+ MLX5_BITSHIFT(ETHTOOL_LINK_MODE_56000baseLR4_Full_BIT)))
+ priv->link_speed_capa |= ETH_LINK_SPEED_56G;
+ if (sc & (MLX5_BITSHIFT(ETHTOOL_LINK_MODE_25000baseCR_Full_BIT) |
+ MLX5_BITSHIFT(ETHTOOL_LINK_MODE_25000baseKR_Full_BIT) |
+ MLX5_BITSHIFT(ETHTOOL_LINK_MODE_25000baseSR_Full_BIT)))
+ priv->link_speed_capa |= ETH_LINK_SPEED_25G;
+ if (sc & (MLX5_BITSHIFT(ETHTOOL_LINK_MODE_50000baseCR2_Full_BIT) |
+ MLX5_BITSHIFT(ETHTOOL_LINK_MODE_50000baseKR2_Full_BIT)))
+ priv->link_speed_capa |= ETH_LINK_SPEED_50G;
+ if (sc & (MLX5_BITSHIFT(ETHTOOL_LINK_MODE_100000baseKR4_Full_BIT) |
+ MLX5_BITSHIFT(ETHTOOL_LINK_MODE_100000baseSR4_Full_BIT) |
+ MLX5_BITSHIFT(ETHTOOL_LINK_MODE_100000baseCR4_Full_BIT) |
+ MLX5_BITSHIFT(ETHTOOL_LINK_MODE_100000baseLR4_ER4_Full_BIT)))
+ priv->link_speed_capa |= ETH_LINK_SPEED_100G;
+ dev_link.link_duplex = ((ecmd->duplex == DUPLEX_HALF) ?
+ ETH_LINK_HALF_DUPLEX : ETH_LINK_FULL_DUPLEX);
+ dev_link.link_autoneg = !(dev->data->dev_conf.link_speeds &
+ ETH_LINK_SPEED_FIXED);
if (memcmp(&dev_link, &dev->data->dev_link, sizeof(dev_link))) {
/* Link status changed. */
dev->data->dev_link = dev_link;
return -1;
}
+/**
+ * Enable receiving and transmitting traffic.
+ *
+ * @param priv
+ * Pointer to private structure.
+ */
+static void
+priv_link_start(struct priv *priv)
+{
+ struct rte_eth_dev *dev = priv->dev;
+ int err;
+
+ dev->tx_pkt_burst = priv_select_tx_function(priv, dev);
+ dev->rx_pkt_burst = priv_select_rx_function(priv, dev);
+ err = priv_dev_traffic_enable(priv, dev);
+ if (err)
+ ERROR("%p: error occurred while configuring control flows: %s",
+ (void *)priv, strerror(err));
+ err = priv_flow_start(priv, &priv->flows);
+ if (err)
+ ERROR("%p: error occurred while configuring flows: %s",
+ (void *)priv, strerror(err));
+}
+
+/**
+ * Disable receiving and transmitting traffic.
+ *
+ * @param priv
+ * Pointer to private structure.
+ */
+static void
+priv_link_stop(struct priv *priv)
+{
+ struct rte_eth_dev *dev = priv->dev;
+
+ priv_flow_stop(priv, &priv->flows);
+ priv_dev_traffic_disable(priv, dev);
+ dev->rx_pkt_burst = removed_rx_burst;
+ dev->tx_pkt_burst = removed_tx_burst;
+}
+
+/**
+ * Retrieve physical link information and update rx/tx_pkt_burst callbacks
+ * accordingly.
+ *
+ * @param priv
+ * Pointer to private structure.
+ * @param wait_to_complete
+ * Wait for request completion (ignored).
+ */
+int
+priv_link_update(struct priv *priv, int wait_to_complete __rte_unused)
+{
+ struct rte_eth_dev *dev = priv->dev;
+ struct utsname utsname;
+ int ver[3];
+ int ret;
+ struct rte_eth_link dev_link = dev->data->dev_link;
+
+ if (uname(&utsname) == -1 ||
+ sscanf(utsname.release, "%d.%d.%d",
+ &ver[0], &ver[1], &ver[2]) != 3 ||
+ KERNEL_VERSION(ver[0], ver[1], ver[2]) < KERNEL_VERSION(4, 9, 0))
+ ret = mlx5_link_update_unlocked_gset(dev);
+ else
+ ret = mlx5_link_update_unlocked_gs(dev);
+ /* If lsc interrupt is disabled, should always be ready for traffic. */
+ if (!dev->data->dev_conf.intr_conf.lsc) {
+ priv_link_start(priv);
+ return ret;
+ }
+ /* Re-select burst callbacks only if link status has been changed. */
+ if (!ret && dev_link.link_status != dev->data->dev_link.link_status) {
+ if (dev->data->dev_link.link_status == ETH_LINK_UP)
+ priv_link_start(priv);
+ else
+ priv_link_stop(priv);
+ }
+ return ret;
+}
+
+/**
+ * Querying the link status till it changes to the desired state.
+ * Number of query attempts is bounded by MLX5_MAX_LINK_QUERY_ATTEMPTS.
+ *
+ * @param priv
+ * Pointer to private structure.
+ * @param status
+ * Link desired status.
+ *
+ * @return
+ * 0 on success, negative errno value on failure.
+ */
+int
+priv_force_link_status_change(struct priv *priv, int status)
+{
+ int try = 0;
+
+ while (try < MLX5_MAX_LINK_QUERY_ATTEMPTS) {
+ priv_link_update(priv, 0);
+ if (priv->dev->data->dev_link.link_status == status)
+ return 0;
+ try++;
+ sleep(1);
+ }
+ return -EAGAIN;
+}
+
/**
* DPDK callback to retrieve physical link information.
*
* Pointer to Ethernet device structure.
* @param wait_to_complete
* Wait for request completion (ignored).
+ *
+ * @return
+ * 0 on success, -1 on error.
*/
int
-mlx5_link_update(struct rte_eth_dev *dev, int wait_to_complete)
+mlx5_link_update(struct rte_eth_dev *dev, int wait_to_complete __rte_unused)
{
struct priv *priv = dev->data->dev_private;
int ret;
- priv_lock(priv);
- ret = mlx5_link_update_unlocked(dev, wait_to_complete);
- priv_unlock(priv);
+ ret = priv_link_update(priv, wait_to_complete);
return ret;
}
/**
* DPDK callback to change the MTU.
*
- * Setting the MTU affects hardware MRU (packets larger than the MTU cannot be
- * received). Use this as a hint to enable/disable scattered packets support
- * and improve performance when not needed.
- * Since failure is not an option, reconfiguring queues on the fly is not
- * recommended.
- *
* @param dev
* Pointer to Ethernet device structure.
* @param in_mtu
mlx5_dev_set_mtu(struct rte_eth_dev *dev, uint16_t mtu)
{
struct priv *priv = dev->data->dev_private;
+ uint16_t kern_mtu;
int ret = 0;
- unsigned int i;
- uint16_t (*rx_func)(void *, struct rte_mbuf **, uint16_t) =
- mlx5_rx_burst;
- priv_lock(priv);
+ ret = priv_get_mtu(priv, &kern_mtu);
+ if (ret)
+ goto out;
/* Set kernel interface MTU first. */
- if (priv_set_mtu(priv, mtu)) {
+ ret = priv_set_mtu(priv, mtu);
+ if (ret)
+ goto out;
+ ret = priv_get_mtu(priv, &kern_mtu);
+ if (ret)
+ goto out;
+ if (kern_mtu == mtu) {
+ priv->mtu = mtu;
+ DEBUG("adapter port %u MTU set to %u", priv->port, mtu);
+ }
+ return 0;
+out:
+ ret = errno;
+ WARN("cannot set port %u MTU to %u: %s", priv->port, mtu,
+ strerror(ret));
+ assert(ret >= 0);
+ return -ret;
+}
+
+/**
+ * DPDK callback to get flow control status.
+ *
+ * @param dev
+ * Pointer to Ethernet device structure.
+ * @param[out] fc_conf
+ * Flow control output buffer.
+ *
+ * @return
+ * 0 on success, negative errno value on failure.
+ */
+int
+mlx5_dev_get_flow_ctrl(struct rte_eth_dev *dev, struct rte_eth_fc_conf *fc_conf)
+{
+ struct priv *priv = dev->data->dev_private;
+ struct ifreq ifr;
+ struct ethtool_pauseparam ethpause = {
+ .cmd = ETHTOOL_GPAUSEPARAM
+ };
+ int ret;
+
+ ifr.ifr_data = (void *)ðpause;
+ if (priv_ifreq(priv, SIOCETHTOOL, &ifr)) {
ret = errno;
- WARN("cannot set port %u MTU to %u: %s", priv->port, mtu,
+ WARN("ioctl(SIOCETHTOOL, ETHTOOL_GPAUSEPARAM)"
+ " failed: %s",
+ strerror(ret));
+ goto out;
+ }
+ fc_conf->autoneg = ethpause.autoneg;
+ if (ethpause.rx_pause && ethpause.tx_pause)
+ fc_conf->mode = RTE_FC_FULL;
+ else if (ethpause.rx_pause)
+ fc_conf->mode = RTE_FC_RX_PAUSE;
+ else if (ethpause.tx_pause)
+ fc_conf->mode = RTE_FC_TX_PAUSE;
+ else
+ fc_conf->mode = RTE_FC_NONE;
+ ret = 0;
+out:
+ assert(ret >= 0);
+ return -ret;
+}
+
+/**
+ * DPDK callback to modify flow control parameters.
+ *
+ * @param dev
+ * Pointer to Ethernet device structure.
+ * @param[in] fc_conf
+ * Flow control parameters.
+ *
+ * @return
+ * 0 on success, negative errno value on failure.
+ */
+int
+mlx5_dev_set_flow_ctrl(struct rte_eth_dev *dev, struct rte_eth_fc_conf *fc_conf)
+{
+ struct priv *priv = dev->data->dev_private;
+ struct ifreq ifr;
+ struct ethtool_pauseparam ethpause = {
+ .cmd = ETHTOOL_SPAUSEPARAM
+ };
+ int ret;
+
+ ifr.ifr_data = (void *)ðpause;
+ ethpause.autoneg = fc_conf->autoneg;
+ if (((fc_conf->mode & RTE_FC_FULL) == RTE_FC_FULL) ||
+ (fc_conf->mode & RTE_FC_RX_PAUSE))
+ ethpause.rx_pause = 1;
+ else
+ ethpause.rx_pause = 0;
+
+ if (((fc_conf->mode & RTE_FC_FULL) == RTE_FC_FULL) ||
+ (fc_conf->mode & RTE_FC_TX_PAUSE))
+ ethpause.tx_pause = 1;
+ else
+ ethpause.tx_pause = 0;
+ if (priv_ifreq(priv, SIOCETHTOOL, &ifr)) {
+ ret = errno;
+ WARN("ioctl(SIOCETHTOOL, ETHTOOL_SPAUSEPARAM)"
+ " failed: %s",
strerror(ret));
goto out;
- } else
- DEBUG("adapter port %u MTU set to %u", priv->port, mtu);
- priv->mtu = mtu;
- /* Temporarily replace RX handler with a fake one, assuming it has not
- * been copied elsewhere. */
- dev->rx_pkt_burst = removed_rx_burst;
- /* Make sure everyone has left mlx5_rx_burst() and uses
- * removed_rx_burst() instead. */
- rte_wmb();
- usleep(1000);
- /* Reconfigure each RX queue. */
- for (i = 0; (i != priv->rxqs_n); ++i) {
- struct rxq *rxq = (*priv->rxqs)[i];
- unsigned int max_frame_len;
- int sp;
-
- if (rxq == NULL)
- continue;
- /* Calculate new maximum frame length according to MTU and
- * toggle scattered support (sp) if necessary. */
- max_frame_len = (priv->mtu + ETHER_HDR_LEN +
- (ETHER_MAX_VLAN_FRAME_LEN - ETHER_MAX_LEN));
- sp = (max_frame_len > (rxq->mb_len - RTE_PKTMBUF_HEADROOM));
- /* Provide new values to rxq_setup(). */
- dev->data->dev_conf.rxmode.jumbo_frame = sp;
- dev->data->dev_conf.rxmode.max_rx_pkt_len = max_frame_len;
- ret = rxq_rehash(dev, rxq);
- if (ret) {
- /* Force SP RX if that queue requires it and abort. */
- if (rxq->sp)
- rx_func = mlx5_rx_burst_sp;
- break;
- }
- /* Reenable non-RSS queue attributes. No need to check
- * for errors at this stage. */
- if (!priv->rss) {
- if (priv->started)
- rxq_mac_addrs_add(rxq);
- if (priv->started && priv->promisc_req)
- rxq_promiscuous_enable(rxq);
- if (priv->started && priv->allmulti_req)
- rxq_allmulticast_enable(rxq);
- }
- /* Scattered burst function takes priority. */
- if (rxq->sp)
- rx_func = mlx5_rx_burst_sp;
}
- /* Burst functions can now be called again. */
- rte_wmb();
- dev->rx_pkt_burst = rx_func;
+ ret = 0;
out:
- priv_unlock(priv);
assert(ret >= 0);
return -ret;
}
/* Extract information. */
if (sscanf(line,
"PCI_SLOT_NAME="
- "%" SCNx16 ":%" SCNx8 ":%" SCNx8 ".%" SCNx8 "\n",
+ "%" SCNx32 ":%" SCNx8 ":%" SCNx8 ".%" SCNx8 "\n",
&pci_addr->domain,
&pci_addr->bus,
&pci_addr->devid,
fclose(file);
return 0;
}
+
+/**
+ * Update the link status.
+ *
+ * @param priv
+ * Pointer to private structure.
+ *
+ * @return
+ * Zero if the callback process can be called immediately.
+ */
+static int
+priv_link_status_update(struct priv *priv)
+{
+ struct rte_eth_link *link = &priv->dev->data->dev_link;
+
+ priv_link_update(priv, 0);
+ if (((link->link_speed == 0) && link->link_status) ||
+ ((link->link_speed != 0) && !link->link_status)) {
+ /*
+ * Inconsistent status. Event likely occurred before the
+ * kernel netdevice exposes the new status.
+ */
+ if (!priv->pending_alarm) {
+ priv->pending_alarm = 1;
+ rte_eal_alarm_set(MLX5_ALARM_TIMEOUT_US,
+ mlx5_dev_link_status_handler,
+ priv->dev);
+ }
+ return 1;
+ } else if (unlikely(priv->pending_alarm)) {
+ /* Link interrupt occurred while alarm is already scheduled. */
+ priv->pending_alarm = 0;
+ rte_eal_alarm_cancel(mlx5_dev_link_status_handler, priv->dev);
+ }
+ return 0;
+}
+
+/**
+ * Device status handler.
+ *
+ * @param priv
+ * Pointer to private structure.
+ * @param events
+ * Pointer to event flags holder.
+ *
+ * @return
+ * Events bitmap of callback process which can be called immediately.
+ */
+static uint32_t
+priv_dev_status_handler(struct priv *priv)
+{
+ struct ibv_async_event event;
+ uint32_t ret = 0;
+
+ /* Read all message and acknowledge them. */
+ for (;;) {
+ if (mlx5_glue->get_async_event(priv->ctx, &event))
+ break;
+ if ((event.event_type == IBV_EVENT_PORT_ACTIVE ||
+ event.event_type == IBV_EVENT_PORT_ERR) &&
+ (priv->dev->data->dev_conf.intr_conf.lsc == 1))
+ ret |= (1 << RTE_ETH_EVENT_INTR_LSC);
+ else if (event.event_type == IBV_EVENT_DEVICE_FATAL &&
+ priv->dev->data->dev_conf.intr_conf.rmv == 1)
+ ret |= (1 << RTE_ETH_EVENT_INTR_RMV);
+ else
+ DEBUG("event type %d on port %d not handled",
+ event.event_type, event.element.port_num);
+ mlx5_glue->ack_async_event(&event);
+ }
+ if (ret & (1 << RTE_ETH_EVENT_INTR_LSC))
+ if (priv_link_status_update(priv))
+ ret &= ~(1 << RTE_ETH_EVENT_INTR_LSC);
+ return ret;
+}
+
+/**
+ * Handle delayed link status event.
+ *
+ * @param arg
+ * Registered argument.
+ */
+void
+mlx5_dev_link_status_handler(void *arg)
+{
+ struct rte_eth_dev *dev = arg;
+ struct priv *priv = dev->data->dev_private;
+ int ret;
+
+ priv->pending_alarm = 0;
+ ret = priv_link_status_update(priv);
+ if (!ret)
+ _rte_eth_dev_callback_process(dev, RTE_ETH_EVENT_INTR_LSC, NULL);
+}
+
+/**
+ * Handle interrupts from the NIC.
+ *
+ * @param[in] intr_handle
+ * Interrupt handler.
+ * @param cb_arg
+ * Callback argument.
+ */
+void
+mlx5_dev_interrupt_handler(void *cb_arg)
+{
+ struct rte_eth_dev *dev = cb_arg;
+ struct priv *priv = dev->data->dev_private;
+ uint32_t events;
+
+ events = priv_dev_status_handler(priv);
+ if (events & (1 << RTE_ETH_EVENT_INTR_LSC))
+ _rte_eth_dev_callback_process(dev, RTE_ETH_EVENT_INTR_LSC, NULL);
+ if (events & (1 << RTE_ETH_EVENT_INTR_RMV))
+ _rte_eth_dev_callback_process(dev, RTE_ETH_EVENT_INTR_RMV, NULL);
+}
+
+/**
+ * Handle interrupts from the socket.
+ *
+ * @param cb_arg
+ * Callback argument.
+ */
+static void
+mlx5_dev_handler_socket(void *cb_arg)
+{
+ struct rte_eth_dev *dev = cb_arg;
+ struct priv *priv = dev->data->dev_private;
+
+ priv_socket_handle(priv);
+}
+
+/**
+ * Uninstall interrupt handler.
+ *
+ * @param priv
+ * Pointer to private structure.
+ * @param dev
+ * Pointer to the rte_eth_dev structure.
+ */
+void
+priv_dev_interrupt_handler_uninstall(struct priv *priv, struct rte_eth_dev *dev)
+{
+ if (dev->data->dev_conf.intr_conf.lsc ||
+ dev->data->dev_conf.intr_conf.rmv)
+ rte_intr_callback_unregister(&priv->intr_handle,
+ mlx5_dev_interrupt_handler, dev);
+ if (priv->primary_socket)
+ rte_intr_callback_unregister(&priv->intr_handle_socket,
+ mlx5_dev_handler_socket, dev);
+ if (priv->pending_alarm) {
+ priv->pending_alarm = 0;
+ rte_eal_alarm_cancel(mlx5_dev_link_status_handler, dev);
+ }
+ priv->intr_handle.fd = 0;
+ priv->intr_handle.type = RTE_INTR_HANDLE_UNKNOWN;
+ priv->intr_handle_socket.fd = 0;
+ priv->intr_handle_socket.type = RTE_INTR_HANDLE_UNKNOWN;
+}
+
+/**
+ * Install interrupt handler.
+ *
+ * @param priv
+ * Pointer to private structure.
+ * @param dev
+ * Pointer to the rte_eth_dev structure.
+ */
+void
+priv_dev_interrupt_handler_install(struct priv *priv, struct rte_eth_dev *dev)
+{
+ int rc, flags;
+
+ assert(priv->ctx->async_fd > 0);
+ flags = fcntl(priv->ctx->async_fd, F_GETFL);
+ rc = fcntl(priv->ctx->async_fd, F_SETFL, flags | O_NONBLOCK);
+ if (rc < 0) {
+ INFO("failed to change file descriptor async event queue");
+ dev->data->dev_conf.intr_conf.lsc = 0;
+ dev->data->dev_conf.intr_conf.rmv = 0;
+ }
+ if (dev->data->dev_conf.intr_conf.lsc ||
+ dev->data->dev_conf.intr_conf.rmv) {
+ priv->intr_handle.fd = priv->ctx->async_fd;
+ priv->intr_handle.type = RTE_INTR_HANDLE_EXT;
+ rte_intr_callback_register(&priv->intr_handle,
+ mlx5_dev_interrupt_handler, dev);
+ }
+ rc = priv_socket_init(priv);
+ if (!rc && priv->primary_socket) {
+ priv->intr_handle_socket.fd = priv->primary_socket;
+ priv->intr_handle_socket.type = RTE_INTR_HANDLE_EXT;
+ rte_intr_callback_register(&priv->intr_handle_socket,
+ mlx5_dev_handler_socket, dev);
+ }
+}
+
+/**
+ * Change the link state (UP / DOWN).
+ *
+ * @param priv
+ * Pointer to private data structure.
+ * @param up
+ * Nonzero for link up, otherwise link down.
+ *
+ * @return
+ * 0 on success, errno value on failure.
+ */
+static int
+priv_dev_set_link(struct priv *priv, int up)
+{
+ return priv_set_flags(priv, ~IFF_UP, up ? IFF_UP : ~IFF_UP);
+}
+
+/**
+ * DPDK callback to bring the link DOWN.
+ *
+ * @param dev
+ * Pointer to Ethernet device structure.
+ *
+ * @return
+ * 0 on success, errno value on failure.
+ */
+int
+mlx5_set_link_down(struct rte_eth_dev *dev)
+{
+ struct priv *priv = dev->data->dev_private;
+ int err;
+
+ err = priv_dev_set_link(priv, 0);
+ return err;
+}
+
+/**
+ * DPDK callback to bring the link UP.
+ *
+ * @param dev
+ * Pointer to Ethernet device structure.
+ *
+ * @return
+ * 0 on success, errno value on failure.
+ */
+int
+mlx5_set_link_up(struct rte_eth_dev *dev)
+{
+ struct priv *priv = dev->data->dev_private;
+ int err;
+
+ err = priv_dev_set_link(priv, 1);
+ return err;
+}
+
+/**
+ * Configure the TX function to use.
+ *
+ * @param priv
+ * Pointer to private data structure.
+ * @param dev
+ * Pointer to rte_eth_dev structure.
+ *
+ * @return
+ * Pointer to selected Tx burst function.
+ */
+eth_tx_burst_t
+priv_select_tx_function(struct priv *priv, struct rte_eth_dev *dev)
+{
+ eth_tx_burst_t tx_pkt_burst = mlx5_tx_burst;
+ struct mlx5_dev_config *config = &priv->config;
+ uint64_t tx_offloads = dev->data->dev_conf.txmode.offloads;
+ int tso = !!(tx_offloads & (DEV_TX_OFFLOAD_TCP_TSO |
+ DEV_TX_OFFLOAD_VXLAN_TNL_TSO |
+ DEV_TX_OFFLOAD_GRE_TNL_TSO));
+ int vlan_insert = !!(tx_offloads & DEV_TX_OFFLOAD_VLAN_INSERT);
+
+ assert(priv != NULL);
+ /* Select appropriate TX function. */
+ if (vlan_insert || tso)
+ return tx_pkt_burst;
+ if (config->mps == MLX5_MPW_ENHANCED) {
+ if (priv_check_vec_tx_support(priv, dev) > 0) {
+ if (priv_check_raw_vec_tx_support(priv, dev) > 0)
+ tx_pkt_burst = mlx5_tx_burst_raw_vec;
+ else
+ tx_pkt_burst = mlx5_tx_burst_vec;
+ DEBUG("selected Enhanced MPW TX vectorized function");
+ } else {
+ tx_pkt_burst = mlx5_tx_burst_empw;
+ DEBUG("selected Enhanced MPW TX function");
+ }
+ } else if (config->mps && (config->txq_inline > 0)) {
+ tx_pkt_burst = mlx5_tx_burst_mpw_inline;
+ DEBUG("selected MPW inline TX function");
+ } else if (config->mps) {
+ tx_pkt_burst = mlx5_tx_burst_mpw;
+ DEBUG("selected MPW TX function");
+ }
+ return tx_pkt_burst;
+}
+
+/**
+ * Configure the RX function to use.
+ *
+ * @param priv
+ * Pointer to private data structure.
+ * @param dev
+ * Pointer to rte_eth_dev structure.
+ *
+ * @return
+ * Pointer to selected Rx burst function.
+ */
+eth_rx_burst_t
+priv_select_rx_function(struct priv *priv, __rte_unused struct rte_eth_dev *dev)
+{
+ eth_rx_burst_t rx_pkt_burst = mlx5_rx_burst;
+
+ assert(priv != NULL);
+ if (priv_check_vec_rx_support(priv) > 0) {
+ rx_pkt_burst = mlx5_rx_burst_vec;
+ DEBUG("selected RX vectorized function");
+ }
+ return rx_pkt_burst;
+}
+
+/**
+ * Check if mlx5 device was removed.
+ *
+ * @param dev
+ * Pointer to Ethernet device structure.
+ *
+ * @return
+ * 1 when device is removed, otherwise 0.
+ */
+int
+mlx5_is_removed(struct rte_eth_dev *dev)
+{
+ struct ibv_device_attr device_attr;
+ struct priv *priv = dev->data->dev_private;
+
+ if (mlx5_glue->query_device(priv->ctx, &device_attr) == EIO)
+ return 1;
+ return 0;
+}