-/*-
- * 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 Technologies, Ltd
*/
#include <stddef.h>
-#include <assert.h>
+#include <inttypes.h>
#include <unistd.h>
+#include <stdbool.h>
#include <stdint.h>
#include <stdio.h>
#include <string.h>
#include <linux/ethtool.h>
#include <linux/sockios.h>
#include <fcntl.h>
+#include <stdalign.h>
+#include <sys/un.h>
+#include <time.h>
-/* DPDK headers don't like -pedantic. */
-#ifdef PEDANTIC
-#pragma GCC diagnostic ignored "-Wpedantic"
-#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>
#include <rte_interrupts.h>
-#include <rte_alarm.h>
#include <rte_malloc.h>
-#ifdef PEDANTIC
-#pragma GCC diagnostic error "-Wpedantic"
-#endif
+#include <rte_string_fns.h>
+#include <rte_rwlock.h>
+#include <rte_cycles.h>
+
+#include <mlx5_glue.h>
+#include <mlx5_devx_cmds.h>
+#include <mlx5_common.h>
#include "mlx5.h"
#include "mlx5_rxtx.h"
#include "mlx5_utils.h"
-/**
- * Return private structure associated with an Ethernet device.
- *
- * @param dev
- * Pointer to Ethernet device structure.
- *
- * @return
- * Pointer to private structure.
- */
-struct priv *
-mlx5_get_priv(struct rte_eth_dev *dev)
-{
- struct mlx5_secondary_data *sd;
-
- if (!mlx5_is_secondary())
- return dev->data->dev_private;
- sd = &mlx5_secondary_data[dev->data->port_id];
- return sd->data.dev_private;
-}
+/* Supported speed values found in /usr/include/linux/ethtool.h */
+#ifndef HAVE_SUPPORTED_40000baseKR4_Full
+#define SUPPORTED_40000baseKR4_Full (1 << 23)
+#endif
+#ifndef HAVE_SUPPORTED_40000baseCR4_Full
+#define SUPPORTED_40000baseCR4_Full (1 << 24)
+#endif
+#ifndef HAVE_SUPPORTED_40000baseSR4_Full
+#define SUPPORTED_40000baseSR4_Full (1 << 25)
+#endif
+#ifndef HAVE_SUPPORTED_40000baseLR4_Full
+#define SUPPORTED_40000baseLR4_Full (1 << 26)
+#endif
+#ifndef HAVE_SUPPORTED_56000baseKR4_Full
+#define SUPPORTED_56000baseKR4_Full (1 << 27)
+#endif
+#ifndef HAVE_SUPPORTED_56000baseCR4_Full
+#define SUPPORTED_56000baseCR4_Full (1 << 28)
+#endif
+#ifndef HAVE_SUPPORTED_56000baseSR4_Full
+#define SUPPORTED_56000baseSR4_Full (1 << 29)
+#endif
+#ifndef HAVE_SUPPORTED_56000baseLR4_Full
+#define SUPPORTED_56000baseLR4_Full (1 << 30)
+#endif
-/**
- * Check if running as a secondary process.
- *
- * @return
- * Nonzero if running as a secondary process.
- */
-inline int
-mlx5_is_secondary(void)
-{
- return rte_eal_process_type() != RTE_PROC_PRIMARY;
-}
+/* 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.
+ * Get master interface name from private structure.
*
- * @param[in] priv
- * Pointer to private structure.
+ * @param[in] dev
+ * Pointer to Ethernet device.
* @param[out] ifname
* Interface name output buffer.
*
* @return
- * 0 on success, -1 on failure and errno is set.
+ * 0 on success, a negative errno value otherwise and rte_errno is set.
*/
int
-priv_get_ifname(const struct priv *priv, char (*ifname)[IF_NAMESIZE])
+mlx5_get_master_ifname(const char *ibdev_path, char (*ifname)[IF_NAMESIZE])
{
DIR *dir;
struct dirent *dent;
unsigned int dev_port_prev = ~0u;
char match[IF_NAMESIZE] = "";
+ MLX5_ASSERT(ibdev_path);
{
- MKSTR(path, "%s/device/net", priv->ctx->device->ibdev_path);
+ MKSTR(path, "%s/device/net", ibdev_path);
dir = opendir(path);
- if (dir == NULL)
- return -1;
+ if (dir == NULL) {
+ rte_errno = errno;
+ return -rte_errno;
+ }
}
while ((dent = readdir(dir)) != NULL) {
char *name = dent->d_name;
continue;
MKSTR(path, "%s/device/net/%s/%s",
- priv->ctx->device->ibdev_path, name,
+ ibdev_path, name,
(dev_type ? "dev_id" : "dev_port"));
file = fopen(path, "rb");
if (dev_port == dev_port_prev)
goto try_dev_id;
dev_port_prev = dev_port;
- if (dev_port == (priv->port - 1u))
- snprintf(match, sizeof(match), "%s", name);
+ if (dev_port == 0)
+ strlcpy(match, name, sizeof(match));
}
closedir(dir);
- if (match[0] == '\0')
- return -1;
+ if (match[0] == '\0') {
+ rte_errno = ENOENT;
+ return -rte_errno;
+ }
strncpy(*ifname, match, sizeof(*ifname));
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.
+ * Get interface name from private structure.
*
- * @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.
+ * This is a port representor-aware version of mlx5_get_master_ifname().
*
- * @param priv
- * Pointer to private structure.
- * @param[in] name
- * Entry name relative to sysfs path.
- * @param[out] value
- * Value output buffer.
+ * @param[in] dev
+ * Pointer to Ethernet device.
+ * @param[out] ifname
+ * Interface name output buffer.
*
* @return
- * 0 on success, -1 on failure and errno is set.
+ * 0 on success, a negative errno value otherwise and rte_errno is set.
*/
-static int
-priv_get_sysfs_ulong(struct priv *priv, const char *name, unsigned long *value)
+int
+mlx5_get_ifname(const struct rte_eth_dev *dev, char (*ifname)[IF_NAMESIZE])
{
- 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;
+ struct mlx5_priv *priv = dev->data->dev_private;
+ unsigned int ifindex;
+
+ MLX5_ASSERT(priv);
+ MLX5_ASSERT(priv->sh);
+ ifindex = mlx5_ifindex(dev);
+ if (!ifindex) {
+ if (!priv->representor)
+ return mlx5_get_master_ifname(priv->sh->ibdev_path,
+ ifname);
+ rte_errno = ENXIO;
+ return -rte_errno;
}
- *value = value_ret;
- return 0;
+ if (if_indextoname(ifindex, &(*ifname)[0]))
+ return 0;
+ rte_errno = errno;
+ return -rte_errno;
}
/**
- * Set unsigned long sysfs property.
+ * Get the interface index from device name.
*
- * @param priv
- * Pointer to private structure.
- * @param[in] name
- * Entry name relative to sysfs path.
- * @param value
- * Value to set.
+ * @param[in] dev
+ * Pointer to Ethernet device.
*
* @return
- * 0 on success, -1 on failure and errno is set.
+ * Nonzero interface index on success, zero otherwise and rte_errno is set.
*/
-static int
-priv_set_sysfs_ulong(struct priv *priv, const char *name, unsigned long value)
+unsigned int
+mlx5_ifindex(const struct rte_eth_dev *dev)
{
- 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;
+ struct mlx5_priv *priv = dev->data->dev_private;
+ unsigned int ifindex;
+
+ MLX5_ASSERT(priv);
+ MLX5_ASSERT(priv->if_index);
+ ifindex = priv->if_index;
+ if (!ifindex)
+ rte_errno = ENXIO;
+ return ifindex;
}
/**
* Perform ifreq ioctl() on associated Ethernet device.
*
- * @param[in] priv
- * Pointer to private structure.
+ * @param[in] dev
+ * Pointer to Ethernet device.
* @param req
* Request number to pass to ioctl().
* @param[out] ifr
* Interface request structure output buffer.
*
* @return
- * 0 on success, -1 on failure and errno is set.
+ * 0 on success, a negative errno value otherwise and rte_errno is set.
*/
int
-priv_ifreq(const struct priv *priv, int req, struct ifreq *ifr)
+mlx5_ifreq(const struct rte_eth_dev *dev, int req, struct ifreq *ifr)
{
int sock = socket(PF_INET, SOCK_DGRAM, IPPROTO_IP);
- int ret = -1;
+ int ret = 0;
- if (sock == -1)
- return ret;
- if (priv_get_ifname(priv, &ifr->ifr_name) == 0)
- ret = ioctl(sock, req, ifr);
+ if (sock == -1) {
+ rte_errno = errno;
+ return -rte_errno;
+ }
+ ret = mlx5_get_ifname(dev, &ifr->ifr_name);
+ if (ret)
+ goto error;
+ ret = ioctl(sock, req, ifr);
+ if (ret == -1) {
+ rte_errno = errno;
+ goto error;
+ }
close(sock);
- return ret;
-}
-
-/**
- * Return the number of active VFs for the current device.
- *
- * @param[in] priv
- * Pointer to private structure.
- * @param[out] num_vfs
- * Number of active VFs.
- *
- * @return
- * 0 on success, -1 on failure and errno is set.
- */
-int
-priv_get_num_vfs(struct priv *priv, uint16_t *num_vfs)
-{
- /* The sysfs entry name depends on the operating system. */
- const char **name = (const char *[]){
- "device/sriov_numvfs",
- "device/mlx5_num_vfs",
- NULL,
- };
- int ret;
-
- do {
- unsigned long ulong_num_vfs;
-
- ret = priv_get_sysfs_ulong(priv, *name, &ulong_num_vfs);
- if (!ret)
- *num_vfs = ulong_num_vfs;
- } while (*(++name) && ret);
- return ret;
+ return 0;
+error:
+ close(sock);
+ return -rte_errno;
}
/**
* Get device MTU.
*
- * @param priv
- * Pointer to private structure.
+ * @param dev
+ * Pointer to Ethernet device.
* @param[out] mtu
* MTU value output buffer.
*
* @return
- * 0 on success, -1 on failure and errno is set.
+ * 0 on success, a negative errno value otherwise and rte_errno is set.
*/
int
-priv_get_mtu(struct priv *priv, uint16_t *mtu)
+mlx5_get_mtu(struct rte_eth_dev *dev, uint16_t *mtu)
{
- unsigned long ulong_mtu;
+ struct ifreq request;
+ int ret = mlx5_ifreq(dev, 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;
}
/**
* Set device MTU.
*
- * @param priv
- * Pointer to private structure.
+ * @param dev
+ * Pointer to Ethernet device.
* @param mtu
* MTU value to set.
*
* @return
- * 0 on success, -1 on failure and errno is set.
+ * 0 on success, a negative errno value otherwise and rte_errno is set.
*/
static int
-priv_set_mtu(struct priv *priv, uint16_t mtu)
+mlx5_set_mtu(struct rte_eth_dev *dev, uint16_t mtu)
{
- uint16_t new_mtu;
+ struct ifreq request = { .ifr_mtu = mtu, };
- if (priv_set_sysfs_ulong(priv, "mtu", mtu) ||
- priv_get_mtu(priv, &new_mtu))
- return -1;
- if (new_mtu == mtu)
- return 0;
- errno = EINVAL;
- return -1;
+ return mlx5_ifreq(dev, SIOCSIFMTU, &request);
}
/**
* Set device flags.
*
- * @param priv
- * Pointer to private structure.
+ * @param dev
+ * Pointer to Ethernet device.
* @param keep
* Bitmask for flags that must remain untouched.
* @param flags
* Bitmask for flags to modify.
*
* @return
- * 0 on success, -1 on failure and errno is set.
+ * 0 on success, a negative errno value otherwise and rte_errno is set.
*/
int
-priv_set_flags(struct priv *priv, unsigned int keep, unsigned int flags)
+mlx5_set_flags(struct rte_eth_dev *dev, unsigned int keep, unsigned int flags)
{
- unsigned long tmp;
+ struct ifreq request;
+ int ret = mlx5_ifreq(dev, SIOCGIFFLAGS, &request);
- if (priv_get_sysfs_ulong(priv, "flags", &tmp) == -1)
- return -1;
- tmp &= keep;
- tmp |= (flags & (~keep));
- return priv_set_sysfs_ulong(priv, "flags", tmp);
+ if (ret)
+ return ret;
+ request.ifr_flags &= keep;
+ request.ifr_flags |= flags & ~keep;
+ return mlx5_ifreq(dev, SIOCSIFFLAGS, &request);
}
/**
- * Ethernet device configuration.
- *
- * Prepare the driver for a given number of TX and RX queues.
+ * DPDK callback for Ethernet device configuration.
*
* @param dev
* Pointer to Ethernet device structure.
*
* @return
- * 0 on success, errno value on failure.
+ * 0 on success, a negative errno value otherwise and rte_errno is set.
*/
-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;
+ struct mlx5_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 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;
+ int ret = 0;
+
+ if (use_app_rss_key &&
+ (dev->data->dev_conf.rx_adv_conf.rss_conf.rss_key_len !=
+ MLX5_RSS_HASH_KEY_LEN)) {
+ DRV_LOG(ERR, "port %u RSS key len must be %s Bytes long",
+ dev->data->port_id, RTE_STR(MLX5_RSS_HASH_KEY_LEN));
+ rte_errno = EINVAL;
+ return -rte_errno;
+ }
+ priv->rss_conf.rss_key =
+ rte_realloc(priv->rss_conf.rss_key,
+ MLX5_RSS_HASH_KEY_LEN, 0);
+ if (!priv->rss_conf.rss_key) {
+ DRV_LOG(ERR, "port %u cannot allocate RSS hash key memory (%u)",
+ dev->data->port_id, rxqs_n);
+ rte_errno = ENOMEM;
+ return -rte_errno;
+ }
+
+ if (dev->data->dev_conf.rxmode.mq_mode & ETH_MQ_RX_RSS_FLAG)
+ dev->data->dev_conf.rxmode.offloads |= DEV_RX_OFFLOAD_RSS_HASH;
- priv->rss_hf = dev->data->dev_conf.rx_adv_conf.rss_conf.rss_hf;
+ 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,
+ MLX5_RSS_HASH_KEY_LEN);
+ priv->rss_conf.rss_key_len = MLX5_RSS_HASH_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) {
- INFO("%p: TX queues number update: %u -> %u",
- (void *)dev, priv->txqs_n, txqs_n);
+ DRV_LOG(INFO, "port %u Tx queues number update: %u -> %u",
+ dev->data->port_id, priv->txqs_n, txqs_n);
priv->txqs_n = txqs_n;
}
- if (rxqs_n > priv->ind_table_max_size) {
- ERROR("cannot handle this many RX queues (%u)", rxqs_n);
- return EINVAL;
+ if (rxqs_n > priv->config.ind_table_max_size) {
+ DRV_LOG(ERR, "port %u cannot handle this many Rx queues (%u)",
+ dev->data->port_id, rxqs_n);
+ rte_errno = EINVAL;
+ return -rte_errno;
}
- if (rxqs_n == priv->rxqs_n)
- return 0;
- INFO("%p: RX queues number update: %u -> %u",
- (void *)dev, priv->rxqs_n, rxqs_n);
- priv->rxqs_n = rxqs_n;
- /* 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->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;
+ if (rxqs_n != priv->rxqs_n) {
+ DRV_LOG(INFO, "port %u Rx queues number update: %u -> %u",
+ dev->data->port_id, priv->rxqs_n, rxqs_n);
+ priv->rxqs_n = rxqs_n;
}
+ priv->skip_default_rss_reta = 0;
+ ret = mlx5_proc_priv_init(dev);
+ if (ret)
+ return ret;
return 0;
}
/**
- * DPDK callback for Ethernet device configuration.
+ * Configure default RSS reta.
*
* @param dev
* Pointer to Ethernet device structure.
*
* @return
- * 0 on success, negative errno value on failure.
+ * 0 on success, a negative errno value otherwise and rte_errno is set.
*/
int
-mlx5_dev_configure(struct rte_eth_dev *dev)
+mlx5_dev_configure_rss_reta(struct rte_eth_dev *dev)
{
- struct priv *priv = dev->data->dev_private;
- int ret;
+ struct mlx5_priv *priv = dev->data->dev_private;
+ unsigned int rxqs_n = dev->data->nb_rx_queues;
+ unsigned int i;
+ unsigned int j;
+ unsigned int reta_idx_n;
+ int ret = 0;
+ unsigned int *rss_queue_arr = NULL;
+ unsigned int rss_queue_n = 0;
+
+ if (priv->skip_default_rss_reta)
+ return ret;
+ rss_queue_arr = rte_malloc("", rxqs_n * sizeof(unsigned int), 0);
+ if (!rss_queue_arr) {
+ DRV_LOG(ERR, "port %u cannot allocate RSS queue list (%u)",
+ dev->data->port_id, rxqs_n);
+ rte_errno = ENOMEM;
+ return -rte_errno;
+ }
+ for (i = 0, j = 0; i < rxqs_n; i++) {
+ struct mlx5_rxq_data *rxq_data;
+ struct mlx5_rxq_ctrl *rxq_ctrl;
+
+ rxq_data = (*priv->rxqs)[i];
+ rxq_ctrl = container_of(rxq_data, struct mlx5_rxq_ctrl, rxq);
+ if (rxq_ctrl && rxq_ctrl->type == MLX5_RXQ_TYPE_STANDARD)
+ rss_queue_arr[j++] = i;
+ }
+ rss_queue_n = j;
+ if (rss_queue_n > priv->config.ind_table_max_size) {
+ DRV_LOG(ERR, "port %u cannot handle this many Rx queues (%u)",
+ dev->data->port_id, rss_queue_n);
+ rte_errno = EINVAL;
+ rte_free(rss_queue_arr);
+ return -rte_errno;
+ }
+ DRV_LOG(INFO, "port %u Rx queues number update: %u -> %u",
+ dev->data->port_id, priv->rxqs_n, rxqs_n);
+ priv->rxqs_n = rxqs_n;
+ /*
+ * 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((rss_queue_n & (rss_queue_n - 1)) ?
+ priv->config.ind_table_max_size :
+ rss_queue_n));
+ ret = mlx5_rss_reta_index_resize(dev, reta_idx_n);
+ if (ret) {
+ rte_free(rss_queue_arr);
+ return ret;
+ }
+ /*
+ * 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] = rss_queue_arr[j];
+ if (++j == rss_queue_n)
+ j = 0;
+ }
+ rte_free(rss_queue_arr);
+ return ret;
+}
- if (mlx5_is_secondary())
- return -E_RTE_SECONDARY;
+/**
+ * Sets default tuning parameters.
+ *
+ * @param dev
+ * Pointer to Ethernet device.
+ * @param[out] info
+ * Info structure output buffer.
+ */
+static void
+mlx5_set_default_params(struct rte_eth_dev *dev, struct rte_eth_dev_info *info)
+{
+ struct mlx5_priv *priv = dev->data->dev_private;
+
+ /* Minimum CPU utilization. */
+ info->default_rxportconf.ring_size = 256;
+ info->default_txportconf.ring_size = 256;
+ info->default_rxportconf.burst_size = MLX5_RX_DEFAULT_BURST;
+ info->default_txportconf.burst_size = MLX5_TX_DEFAULT_BURST;
+ if (priv->link_speed_capa & ETH_LINK_SPEED_100G) {
+ info->default_rxportconf.nb_queues = 16;
+ info->default_txportconf.nb_queues = 16;
+ if (dev->data->nb_rx_queues > 2 ||
+ dev->data->nb_tx_queues > 2) {
+ /* Max Throughput. */
+ info->default_rxportconf.ring_size = 2048;
+ info->default_txportconf.ring_size = 2048;
+ }
+ } else {
+ info->default_rxportconf.nb_queues = 8;
+ info->default_txportconf.nb_queues = 8;
+ if (dev->data->nb_rx_queues > 2 ||
+ dev->data->nb_tx_queues > 2) {
+ /* Max Throughput. */
+ info->default_rxportconf.ring_size = 4096;
+ info->default_txportconf.ring_size = 4096;
+ }
+ }
+}
- priv_lock(priv);
- ret = dev_configure(dev);
- assert(ret >= 0);
- priv_unlock(priv);
- return -ret;
+/**
+ * Sets tx mbuf limiting parameters.
+ *
+ * @param dev
+ * Pointer to Ethernet device.
+ * @param[out] info
+ * Info structure output buffer.
+ */
+static void
+mlx5_set_txlimit_params(struct rte_eth_dev *dev, struct rte_eth_dev_info *info)
+{
+ struct mlx5_priv *priv = dev->data->dev_private;
+ struct mlx5_dev_config *config = &priv->config;
+ unsigned int inlen;
+ uint16_t nb_max;
+
+ inlen = (config->txq_inline_max == MLX5_ARG_UNSET) ?
+ MLX5_SEND_DEF_INLINE_LEN :
+ (unsigned int)config->txq_inline_max;
+ MLX5_ASSERT(config->txq_inline_min >= 0);
+ inlen = RTE_MAX(inlen, (unsigned int)config->txq_inline_min);
+ inlen = RTE_MIN(inlen, MLX5_WQE_SIZE_MAX +
+ MLX5_ESEG_MIN_INLINE_SIZE -
+ MLX5_WQE_CSEG_SIZE -
+ MLX5_WQE_ESEG_SIZE -
+ MLX5_WQE_DSEG_SIZE * 2);
+ nb_max = (MLX5_WQE_SIZE_MAX +
+ MLX5_ESEG_MIN_INLINE_SIZE -
+ MLX5_WQE_CSEG_SIZE -
+ MLX5_WQE_ESEG_SIZE -
+ MLX5_WQE_DSEG_SIZE -
+ inlen) / MLX5_WSEG_SIZE;
+ info->tx_desc_lim.nb_seg_max = nb_max;
+ info->tx_desc_lim.nb_mtu_seg_max = nb_max;
}
/**
* @param[out] info
* Info structure output buffer.
*/
-void
+int
mlx5_dev_infos_get(struct rte_eth_dev *dev, struct rte_eth_dev_info *info)
{
- struct priv *priv = mlx5_get_priv(dev);
+ struct mlx5_priv *priv = dev->data->dev_private;
+ struct mlx5_dev_config *config = &priv->config;
unsigned int max;
- char ifname[IF_NAMESIZE];
- priv_lock(priv);
/* FIXME: we should ask the device for these values. */
info->min_rx_bufsize = 32;
info->max_rx_pktlen = 65536;
+ info->max_lro_pkt_size = MLX5_MAX_LRO_SIZE;
/*
* 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);
- /* If max >= 65535 then max = 0, max_rx_queues is uint16_t. */
- if (max >= 65535)
- max = 65535;
+ max = RTE_MIN(priv->sh->device_attr.orig_attr.max_cq,
+ priv->sh->device_attr.orig_attr.max_qp);
+ /* max_rx_queues is uint16_t. */
+ max = RTE_MIN(max, (unsigned int)UINT16_MAX);
info->max_rx_queues = max;
info->max_tx_queues = max;
- info->max_mac_addrs = RTE_DIM(priv->mac);
- info->rx_offload_capa =
- (priv->hw_csum ?
- (DEV_RX_OFFLOAD_IPV4_CKSUM |
- DEV_RX_OFFLOAD_UDP_CKSUM |
- DEV_RX_OFFLOAD_TCP_CKSUM) :
- 0) |
- (priv->hw_vlan_strip ? DEV_RX_OFFLOAD_VLAN_STRIP : 0);
- if (!priv->mps)
- info->tx_offload_capa = DEV_TX_OFFLOAD_VLAN_INSERT;
- if (priv->hw_csum)
- info->tx_offload_capa |=
- (DEV_TX_OFFLOAD_IPV4_CKSUM |
- DEV_TX_OFFLOAD_UDP_CKSUM |
- DEV_TX_OFFLOAD_TCP_CKSUM);
- if (priv_get_ifname(priv, &ifname) == 0)
- info->if_index = if_nametoindex(ifname);
- /* FIXME: RETA update/query API expects the callee to know the size of
- * the indirection table, for this PMD the size varies depending on
- * the number of RX queues, it becomes impossible to find the correct
- * size if it is not fixed.
- * The API should be updated to solve this problem. */
- info->reta_size = priv->ind_table_max_size;
- info->hash_key_size = ((*priv->rss_conf) ?
- (*priv->rss_conf)[0]->rss_key_len :
- 0);
- info->speed_capa =
- ETH_LINK_SPEED_1G |
- ETH_LINK_SPEED_10G |
- ETH_LINK_SPEED_20G |
- ETH_LINK_SPEED_25G |
- ETH_LINK_SPEED_40G |
- ETH_LINK_SPEED_50G |
- ETH_LINK_SPEED_56G |
- ETH_LINK_SPEED_100G;
- priv_unlock(priv);
-}
-
-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_L3_IPV4,
- RTE_PTYPE_L3_IPV6,
- RTE_PTYPE_INNER_L3_IPV4,
- RTE_PTYPE_INNER_L3_IPV6,
- RTE_PTYPE_UNKNOWN
-
- };
-
- if (dev->rx_pkt_burst == mlx5_rx_burst)
- return ptypes;
- return NULL;
+ info->max_mac_addrs = MLX5_MAX_UC_MAC_ADDRESSES;
+ info->rx_queue_offload_capa = mlx5_get_rx_queue_offloads(dev);
+ info->rx_offload_capa = (mlx5_get_rx_port_offloads() |
+ info->rx_queue_offload_capa);
+ info->tx_offload_capa = mlx5_get_tx_port_offloads(dev);
+ info->if_index = mlx5_ifindex(dev);
+ info->reta_size = priv->reta_idx_n ?
+ priv->reta_idx_n : config->ind_table_max_size;
+ info->hash_key_size = MLX5_RSS_HASH_KEY_LEN;
+ info->speed_capa = priv->link_speed_capa;
+ info->flow_type_rss_offloads = ~MLX5_RSS_HF_MASK;
+ mlx5_set_default_params(dev, info);
+ mlx5_set_txlimit_params(dev, info);
+ info->switch_info.name = dev->data->name;
+ info->switch_info.domain_id = priv->domain_id;
+ info->switch_info.port_id = priv->representor_id;
+ if (priv->representor) {
+ uint16_t port_id;
+
+ if (priv->pf_bond >= 0) {
+ /*
+ * Switch port ID is opaque value with driver defined
+ * format. Push the PF index in bonding configurations
+ * in upper four bits of port ID. If we get too many
+ * representors (more than 4K) or PFs (more than 15)
+ * this approach must be reconsidered.
+ */
+ if ((info->switch_info.port_id >>
+ MLX5_PORT_ID_BONDING_PF_SHIFT) ||
+ priv->pf_bond > MLX5_PORT_ID_BONDING_PF_MASK) {
+ DRV_LOG(ERR, "can't update switch port ID"
+ " for bonding device");
+ MLX5_ASSERT(false);
+ return -ENODEV;
+ }
+ info->switch_info.port_id |=
+ priv->pf_bond << MLX5_PORT_ID_BONDING_PF_SHIFT;
+ }
+ MLX5_ETH_FOREACH_DEV(port_id, priv->pci_dev) {
+ struct mlx5_priv *opriv =
+ rte_eth_devices[port_id].data->dev_private;
+
+ if (!opriv ||
+ opriv->representor ||
+ opriv->sh != priv->sh ||
+ opriv->domain_id != priv->domain_id)
+ continue;
+ /*
+ * Override switch name with that of the master
+ * device.
+ */
+ info->switch_info.name = opriv->dev_data->name;
+ break;
+ }
+ }
+ return 0;
}
/**
- * DPDK callback to retrieve physical link information (unlocked version).
+ * Get device current raw clock counter
*
* @param dev
* Pointer to Ethernet device structure.
- * @param wait_to_complete
- * Wait for request completion (ignored).
+ * @param[out] time
+ * Current raw clock counter of the device.
+ *
+ * @return
+ * 0 if the clock has correctly been read
+ * The value of errno in case of error
*/
int
-mlx5_link_update_unlocked(struct rte_eth_dev *dev, int wait_to_complete)
+mlx5_read_clock(struct rte_eth_dev *dev, uint64_t *clock)
{
- struct priv *priv = mlx5_get_priv(dev);
- struct ethtool_cmd edata = {
- .cmd = ETHTOOL_GSET
- };
- 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 = (void *)&edata;
- if (priv_ifreq(priv, SIOCETHTOOL, &ifr)) {
- WARN("ioctl(SIOCETHTOOL, ETHTOOL_GSET) failed: %s",
- strerror(errno));
- return -1;
- }
- link_speed = ethtool_cmd_speed(&edata);
- if (link_speed == -1)
- dev_link.link_speed = 0;
- else
- dev_link.link_speed = link_speed;
- 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;
+ struct mlx5_priv *priv = dev->data->dev_private;
+ struct ibv_context *ctx = priv->sh->ctx;
+ struct ibv_values_ex values;
+ int err = 0;
+
+ values.comp_mask = IBV_VALUES_MASK_RAW_CLOCK;
+ err = mlx5_glue->query_rt_values_ex(ctx, &values);
+ if (err != 0) {
+ DRV_LOG(WARNING, "Could not query the clock !");
+ return err;
}
- /* Link status is still the same. */
- return -1;
+ *clock = values.raw_clock.tv_nsec;
+ return 0;
}
/**
- * DPDK callback to retrieve physical link information.
+ * Get firmware version of a device.
*
* @param dev
- * Pointer to Ethernet device structure.
- * @param wait_to_complete
- * Wait for request completion (ignored).
+ * Ethernet device port.
+ * @param fw_ver
+ * String output allocated by caller.
+ * @param fw_size
+ * Size of the output string, including terminating null byte.
+ *
+ * @return
+ * 0 on success, or the size of the non truncated string if too big.
*/
-int
-mlx5_link_update(struct rte_eth_dev *dev, int wait_to_complete)
+int mlx5_fw_version_get(struct rte_eth_dev *dev, char *fw_ver, size_t fw_size)
{
- struct priv *priv = mlx5_get_priv(dev);
- int ret;
-
- priv_lock(priv);
- ret = mlx5_link_update_unlocked(dev, wait_to_complete);
- priv_unlock(priv);
- return ret;
+ struct mlx5_priv *priv = dev->data->dev_private;
+ struct ibv_device_attr *attr = &priv->sh->device_attr.orig_attr;
+ size_t size = strnlen(attr->fw_ver, sizeof(attr->fw_ver)) + 1;
+
+ if (fw_size < size)
+ return size;
+ if (fw_ver != NULL)
+ strlcpy(fw_ver, attr->fw_ver, fw_size);
+ return 0;
}
/**
- * 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.
+ * Get supported packet types.
*
* @param dev
* Pointer to Ethernet device structure.
- * @param in_mtu
- * New MTU.
*
* @return
- * 0 on success, negative errno value on failure.
+ * A pointer to the supported Packet types array.
*/
-int
-mlx5_dev_set_mtu(struct rte_eth_dev *dev, uint16_t mtu)
+const uint32_t *
+mlx5_dev_supported_ptypes_get(struct rte_eth_dev *dev)
{
- struct priv *priv = dev->data->dev_private;
- int ret = 0;
- unsigned int i;
- uint16_t (*rx_func)(void *, struct rte_mbuf **, uint16_t) =
- mlx5_rx_burst;
- unsigned int max_frame_len;
- int rehash;
- int restart = priv->started;
-
- if (mlx5_is_secondary())
- return -E_RTE_SECONDARY;
-
- priv_lock(priv);
- /* Set kernel interface MTU first. */
- if (priv_set_mtu(priv, mtu)) {
- ret = errno;
- WARN("cannot set port %u MTU to %u: %s", priv->port, mtu,
- strerror(ret));
- goto out;
- } else
- DEBUG("adapter port %u MTU set to %u", priv->port, 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);
- /* MTU does not include header and CRC. */
- max_frame_len = ETHER_HDR_LEN + mtu + ETHER_CRC_LEN;
- /* Check if at least one queue is going to need a SGE update. */
- for (i = 0; i != priv->rxqs_n; ++i) {
- struct rxq *rxq = (*priv->rxqs)[i];
- unsigned int mb_len;
- unsigned int size = RTE_PKTMBUF_HEADROOM + max_frame_len;
- unsigned int sges_n;
-
- if (rxq == NULL)
- continue;
- mb_len = rte_pktmbuf_data_room_size(rxq->mp);
- assert(mb_len >= RTE_PKTMBUF_HEADROOM);
- /*
- * Determine the number of SGEs needed for a full packet
- * and round it to the next power of two.
- */
- sges_n = log2above((size / mb_len) + !!(size % mb_len));
- if (sges_n != rxq->sges_n)
- break;
+ 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_mprq ||
+ dev->rx_pkt_burst == mlx5_rx_burst_vec)
+ return ptypes;
+ return NULL;
+}
+
+/**
+ * Retrieve the master device for representor in the same switch domain.
+ *
+ * @param dev
+ * Pointer to representor Ethernet device structure.
+ *
+ * @return
+ * Master device structure on success, NULL otherwise.
+ */
+
+static struct rte_eth_dev *
+mlx5_find_master_dev(struct rte_eth_dev *dev)
+{
+ struct mlx5_priv *priv;
+ uint16_t port_id;
+ uint16_t domain_id;
+
+ priv = dev->data->dev_private;
+ domain_id = priv->domain_id;
+ MLX5_ASSERT(priv->representor);
+ MLX5_ETH_FOREACH_DEV(port_id, priv->pci_dev) {
+ struct mlx5_priv *opriv =
+ rte_eth_devices[port_id].data->dev_private;
+ if (opriv &&
+ opriv->master &&
+ opriv->domain_id == domain_id &&
+ opriv->sh == priv->sh)
+ return &rte_eth_devices[port_id];
}
- /*
- * If all queues have the right number of SGEs, a simple rehash
- * of their buffers is enough, otherwise SGE information can only
- * be updated in a queue by recreating it. All resources that depend
- * on queues (flows, indirection tables) must be recreated as well in
- * that case.
- */
- rehash = (i == priv->rxqs_n);
- if (!rehash) {
- /* Clean up everything as with mlx5_dev_stop(). */
- priv_special_flow_disable_all(priv);
- priv_mac_addrs_disable(priv);
- priv_destroy_hash_rxqs(priv);
- priv_fdir_disable(priv);
- priv_dev_interrupt_handler_uninstall(priv, dev);
+ return NULL;
+}
+
+/**
+ * DPDK callback to retrieve physical link information.
+ *
+ * @param dev
+ * Pointer to Ethernet device structure.
+ * @param[out] link
+ * Storage for current link status.
+ *
+ * @return
+ * 0 on success, a negative errno value otherwise and rte_errno is set.
+ */
+static int
+mlx5_link_update_unlocked_gset(struct rte_eth_dev *dev,
+ struct rte_eth_link *link)
+{
+ struct mlx5_priv *priv = dev->data->dev_private;
+ struct ethtool_cmd edata = {
+ .cmd = ETHTOOL_GSET /* Deprecated since Linux v4.5. */
+ };
+ struct ifreq ifr;
+ struct rte_eth_link dev_link;
+ int link_speed = 0;
+ int ret;
+
+ ret = mlx5_ifreq(dev, SIOCGIFFLAGS, &ifr);
+ if (ret) {
+ DRV_LOG(WARNING, "port %u ioctl(SIOCGIFFLAGS) failed: %s",
+ dev->data->port_id, strerror(rte_errno));
+ return ret;
}
-recover:
- /* Reconfigure each RX queue. */
- for (i = 0; (i != priv->rxqs_n); ++i) {
- struct rxq *rxq = (*priv->rxqs)[i];
- struct rxq_ctrl *rxq_ctrl =
- container_of(rxq, struct rxq_ctrl, rxq);
- int sp;
- unsigned int mb_len;
- unsigned int tmp;
-
- if (rxq == NULL)
- continue;
- mb_len = rte_pktmbuf_data_room_size(rxq->mp);
- assert(mb_len >= RTE_PKTMBUF_HEADROOM);
- /* Toggle scattered support (sp) if necessary. */
- sp = (max_frame_len > (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;
- if (rehash)
- ret = rxq_rehash(dev, rxq_ctrl);
- else
- ret = rxq_ctrl_setup(dev, rxq_ctrl, 1 << rxq->elts_n,
- rxq_ctrl->socket, NULL, rxq->mp);
- if (!ret)
- continue;
- /* Attempt to roll back in case of error. */
- tmp = (mb_len << rxq->sges_n) - RTE_PKTMBUF_HEADROOM;
- if (max_frame_len != tmp) {
- max_frame_len = tmp;
- goto recover;
+ dev_link = (struct rte_eth_link) {
+ .link_status = ((ifr.ifr_flags & IFF_UP) &&
+ (ifr.ifr_flags & IFF_RUNNING)),
+ };
+ ifr = (struct ifreq) {
+ .ifr_data = (void *)&edata,
+ };
+ ret = mlx5_ifreq(dev, SIOCETHTOOL, &ifr);
+ if (ret) {
+ if (ret == -ENOTSUP && priv->representor) {
+ struct rte_eth_dev *master;
+
+ /*
+ * For representors we can try to inherit link
+ * settings from the master device. Actually
+ * link settings do not make a lot of sense
+ * for representors due to missing physical
+ * link. The old kernel drivers supported
+ * emulated settings query for representors,
+ * the new ones do not, so we have to add
+ * this code for compatibility issues.
+ */
+ master = mlx5_find_master_dev(dev);
+ if (master) {
+ ifr = (struct ifreq) {
+ .ifr_data = (void *)&edata,
+ };
+ ret = mlx5_ifreq(master, SIOCETHTOOL, &ifr);
+ }
+ }
+ if (ret) {
+ DRV_LOG(WARNING,
+ "port %u ioctl(SIOCETHTOOL,"
+ " ETHTOOL_GSET) failed: %s",
+ dev->data->port_id, strerror(rte_errno));
+ return ret;
}
- /* Double fault, disable RX. */
- break;
}
- /*
- * Use a safe RX burst function in case of error, otherwise mimic
- * mlx5_dev_start().
- */
+ link_speed = ethtool_cmd_speed(&edata);
+ if (link_speed == -1)
+ dev_link.link_speed = ETH_SPEED_NUM_NONE;
+ 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 (((dev_link.link_speed && !dev_link.link_status) ||
+ (!dev_link.link_speed && dev_link.link_status))) {
+ rte_errno = EAGAIN;
+ return -rte_errno;
+ }
+ *link = dev_link;
+ return 0;
+}
+
+/**
+ * Retrieve physical link information (unlocked version using new ioctl).
+ *
+ * @param dev
+ * Pointer to Ethernet device structure.
+ * @param[out] link
+ * Storage for current link status.
+ *
+ * @return
+ * 0 on success, a negative errno value otherwise and rte_errno is set.
+ */
+static int
+mlx5_link_update_unlocked_gs(struct rte_eth_dev *dev,
+ struct rte_eth_link *link)
+
+{
+ struct mlx5_priv *priv = dev->data->dev_private;
+ struct ethtool_link_settings gcmd = { .cmd = ETHTOOL_GLINKSETTINGS };
+ struct ifreq ifr;
+ struct rte_eth_link dev_link;
+ struct rte_eth_dev *master = NULL;
+ uint64_t sc;
+ int ret;
+
+ ret = mlx5_ifreq(dev, SIOCGIFFLAGS, &ifr);
+ if (ret) {
+ DRV_LOG(WARNING, "port %u ioctl(SIOCGIFFLAGS) failed: %s",
+ dev->data->port_id, strerror(rte_errno));
+ return ret;
+ }
+ dev_link = (struct rte_eth_link) {
+ .link_status = ((ifr.ifr_flags & IFF_UP) &&
+ (ifr.ifr_flags & IFF_RUNNING)),
+ };
+ ifr = (struct ifreq) {
+ .ifr_data = (void *)&gcmd,
+ };
+ ret = mlx5_ifreq(dev, SIOCETHTOOL, &ifr);
+ if (ret) {
+ if (ret == -ENOTSUP && priv->representor) {
+ /*
+ * For representors we can try to inherit link
+ * settings from the master device. Actually
+ * link settings do not make a lot of sense
+ * for representors due to missing physical
+ * link. The old kernel drivers supported
+ * emulated settings query for representors,
+ * the new ones do not, so we have to add
+ * this code for compatibility issues.
+ */
+ master = mlx5_find_master_dev(dev);
+ if (master) {
+ ifr = (struct ifreq) {
+ .ifr_data = (void *)&gcmd,
+ };
+ ret = mlx5_ifreq(master, SIOCETHTOOL, &ifr);
+ }
+ }
+ if (ret) {
+ DRV_LOG(DEBUG,
+ "port %u ioctl(SIOCETHTOOL,"
+ " ETHTOOL_GLINKSETTINGS) failed: %s",
+ dev->data->port_id, strerror(rte_errno));
+ return ret;
+ }
+
+ }
+ 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;
+ ret = mlx5_ifreq(master ? master : dev, SIOCETHTOOL, &ifr);
if (ret) {
- ERROR("unable to reconfigure RX queues, RX disabled");
- rx_func = removed_rx_burst;
- } else if (restart &&
- !rehash &&
- !priv_create_hash_rxqs(priv) &&
- !priv_rehash_flows(priv)) {
- if (dev->data->dev_conf.fdir_conf.mode == RTE_FDIR_MODE_NONE)
- priv_fdir_enable(priv);
- priv_dev_interrupt_handler_install(priv, dev);
+ DRV_LOG(DEBUG,
+ "port %u ioctl(SIOCETHTOOL,"
+ "ETHTOOL_GLINKSETTINGS) failed: %s",
+ dev->data->port_id, strerror(rte_errno));
+ return ret;
+ }
+ dev_link.link_speed = (ecmd->speed == UINT32_MAX) ? ETH_SPEED_NUM_NONE :
+ 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 (((dev_link.link_speed && !dev_link.link_status) ||
+ (!dev_link.link_speed && dev_link.link_status))) {
+ rte_errno = EAGAIN;
+ return -rte_errno;
+ }
+ *link = dev_link;
+ return 0;
+}
+
+/**
+ * DPDK callback to retrieve physical link information.
+ *
+ * @param dev
+ * Pointer to Ethernet device structure.
+ * @param wait_to_complete
+ * Wait for request completion.
+ *
+ * @return
+ * 0 if link status was not updated, positive if it was, a negative errno
+ * value otherwise and rte_errno is set.
+ */
+int
+mlx5_link_update(struct rte_eth_dev *dev, int wait_to_complete)
+{
+ int ret;
+ struct rte_eth_link dev_link;
+ time_t start_time = time(NULL);
+ int retry = MLX5_GET_LINK_STATUS_RETRY_COUNT;
+
+ do {
+ ret = mlx5_link_update_unlocked_gs(dev, &dev_link);
+ if (ret == -ENOTSUP)
+ ret = mlx5_link_update_unlocked_gset(dev, &dev_link);
+ if (ret == 0)
+ break;
+ /* Handle wait to complete situation. */
+ if ((wait_to_complete || retry) && ret == -EAGAIN) {
+ if (abs((int)difftime(time(NULL), start_time)) <
+ MLX5_LINK_STATUS_TIMEOUT) {
+ usleep(0);
+ continue;
+ } else {
+ rte_errno = EBUSY;
+ return -rte_errno;
+ }
+ } else if (ret < 0) {
+ return ret;
+ }
+ } while (wait_to_complete || retry-- > 0);
+ ret = !!memcmp(&dev->data->dev_link, &dev_link,
+ sizeof(struct rte_eth_link));
+ dev->data->dev_link = dev_link;
+ return ret;
+}
+
+/**
+ * DPDK callback to change the MTU.
+ *
+ * @param dev
+ * Pointer to Ethernet device structure.
+ * @param in_mtu
+ * New MTU.
+ *
+ * @return
+ * 0 on success, a negative errno value otherwise and rte_errno is set.
+ */
+int
+mlx5_dev_set_mtu(struct rte_eth_dev *dev, uint16_t mtu)
+{
+ struct mlx5_priv *priv = dev->data->dev_private;
+ uint16_t kern_mtu = 0;
+ int ret;
+
+ ret = mlx5_get_mtu(dev, &kern_mtu);
+ if (ret)
+ return ret;
+ /* Set kernel interface MTU first. */
+ ret = mlx5_set_mtu(dev, mtu);
+ if (ret)
+ return ret;
+ ret = mlx5_get_mtu(dev, &kern_mtu);
+ if (ret)
+ return ret;
+ if (kern_mtu == mtu) {
+ priv->mtu = mtu;
+ DRV_LOG(DEBUG, "port %u adapter MTU set to %u",
+ dev->data->port_id, mtu);
+ return 0;
}
- priv->mtu = mtu;
- /* Burst functions can now be called again. */
- rte_wmb();
- dev->rx_pkt_burst = rx_func;
-out:
- priv_unlock(priv);
- assert(ret >= 0);
- return -ret;
+ rte_errno = EAGAIN;
+ return -rte_errno;
}
/**
* Flow control output buffer.
*
* @return
- * 0 on success, negative errno value on failure.
+ * 0 on success, a negative errno value otherwise and rte_errno is set.
*/
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;
- if (mlx5_is_secondary())
- return -E_RTE_SECONDARY;
-
ifr.ifr_data = (void *)ðpause;
- priv_lock(priv);
- if (priv_ifreq(priv, SIOCETHTOOL, &ifr)) {
- ret = errno;
- WARN("ioctl(SIOCETHTOOL, ETHTOOL_GPAUSEPARAM)"
- " failed: %s",
- strerror(ret));
- goto out;
+ ret = mlx5_ifreq(dev, SIOCETHTOOL, &ifr);
+ if (ret) {
+ DRV_LOG(WARNING,
+ "port %u ioctl(SIOCETHTOOL, ETHTOOL_GPAUSEPARAM) failed:"
+ " %s",
+ dev->data->port_id, strerror(rte_errno));
+ return ret;
}
-
fc_conf->autoneg = ethpause.autoneg;
if (ethpause.rx_pause && ethpause.tx_pause)
fc_conf->mode = RTE_FC_FULL;
fc_conf->mode = RTE_FC_TX_PAUSE;
else
fc_conf->mode = RTE_FC_NONE;
- ret = 0;
-
-out:
- priv_unlock(priv);
- assert(ret >= 0);
- return -ret;
+ return 0;
}
/**
* Flow control parameters.
*
* @return
- * 0 on success, negative errno value on failure.
+ * 0 on success, a negative errno value otherwise and rte_errno is set.
*/
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;
- if (mlx5_is_secondary())
- return -E_RTE_SECONDARY;
-
ifr.ifr_data = (void *)ðpause;
ethpause.autoneg = fc_conf->autoneg;
if (((fc_conf->mode & RTE_FC_FULL) == RTE_FC_FULL) ||
ethpause.tx_pause = 1;
else
ethpause.tx_pause = 0;
-
- priv_lock(priv);
- if (priv_ifreq(priv, SIOCETHTOOL, &ifr)) {
- ret = errno;
- WARN("ioctl(SIOCETHTOOL, ETHTOOL_SPAUSEPARAM)"
- " failed: %s",
- strerror(ret));
- goto out;
+ ret = mlx5_ifreq(dev, SIOCETHTOOL, &ifr);
+ if (ret) {
+ DRV_LOG(WARNING,
+ "port %u ioctl(SIOCETHTOOL, ETHTOOL_SPAUSEPARAM)"
+ " failed: %s",
+ dev->data->port_id, strerror(rte_errno));
+ return ret;
}
- ret = 0;
-
-out:
- priv_unlock(priv);
- assert(ret >= 0);
- return -ret;
+ return 0;
}
/**
- * Get PCI information from struct ibv_device.
+ * Handle asynchronous removal event for entire multiport device.
*
- * @param device
- * Pointer to Ethernet device structure.
- * @param[out] pci_addr
- * PCI bus address output buffer.
- *
- * @return
- * 0 on success, -1 on failure and errno is set.
+ * @param sh
+ * Infiniband device shared context.
*/
-int
-mlx5_ibv_device_to_pci_addr(const struct ibv_device *device,
- struct rte_pci_addr *pci_addr)
+static void
+mlx5_dev_interrupt_device_fatal(struct mlx5_ibv_shared *sh)
{
- FILE *file;
- char line[32];
- MKSTR(path, "%s/device/uevent", device->ibdev_path);
-
- file = fopen(path, "rb");
- if (file == NULL)
- return -1;
- while (fgets(line, sizeof(line), file) == line) {
- size_t len = strlen(line);
- int ret;
+ uint32_t i;
- /* Truncate long lines. */
- if (len == (sizeof(line) - 1))
- while (line[(len - 1)] != '\n') {
- ret = fgetc(file);
- if (ret == EOF)
- break;
- line[(len - 1)] = ret;
- }
- /* Extract information. */
- if (sscanf(line,
- "PCI_SLOT_NAME="
- "%" SCNx16 ":%" SCNx8 ":%" SCNx8 ".%" SCNx8 "\n",
- &pci_addr->domain,
- &pci_addr->bus,
- &pci_addr->devid,
- &pci_addr->function) == 4) {
- ret = 0;
- break;
+ for (i = 0; i < sh->max_port; ++i) {
+ struct rte_eth_dev *dev;
+
+ if (sh->port[i].ih_port_id >= RTE_MAX_ETHPORTS) {
+ /*
+ * Or not existing port either no
+ * handler installed for this port.
+ */
+ continue;
}
+ dev = &rte_eth_devices[sh->port[i].ih_port_id];
+ MLX5_ASSERT(dev);
+ if (dev->data->dev_conf.intr_conf.rmv)
+ _rte_eth_dev_callback_process
+ (dev, RTE_ETH_EVENT_INTR_RMV, NULL);
}
- fclose(file);
- return 0;
}
/**
- * Link status handler.
+ * Handle shared asynchronous events the NIC (removal event
+ * and link status change). Supports multiport IB device.
*
- * @param priv
- * Pointer to private structure.
- * @param dev
- * Pointer to the rte_eth_dev structure.
- *
- * @return
- * Nonzero if the callback process can be called immediately.
+ * @param cb_arg
+ * Callback argument.
*/
-static int
-priv_dev_link_status_handler(struct priv *priv, struct rte_eth_dev *dev)
+void
+mlx5_dev_interrupt_handler(void *cb_arg)
{
+ struct mlx5_ibv_shared *sh = cb_arg;
struct ibv_async_event event;
- int port_change = 0;
- int ret = 0;
- /* Read all message and acknowledge them. */
+ /* Read all message from the IB device and acknowledge them. */
for (;;) {
- if (ibv_get_async_event(priv->ctx, &event))
- break;
+ struct rte_eth_dev *dev;
+ uint32_t tmp;
- if (event.event_type == IBV_EVENT_PORT_ACTIVE ||
- event.event_type == IBV_EVENT_PORT_ERR)
- port_change = 1;
- else
- DEBUG("event type %d on port %d not handled",
- event.event_type, event.element.port_num);
- ibv_ack_async_event(&event);
- }
-
- if (port_change ^ priv->pending_alarm) {
- struct rte_eth_link *link = &dev->data->dev_link;
-
- priv->pending_alarm = 0;
- mlx5_link_update_unlocked(dev, 0);
- if (((link->link_speed == 0) && link->link_status) ||
- ((link->link_speed != 0) && !link->link_status)) {
- /* Inconsistent status, check again later. */
- priv->pending_alarm = 1;
- rte_eal_alarm_set(MLX5_ALARM_TIMEOUT_US,
- mlx5_dev_link_status_handler,
- dev);
- } else
- ret = 1;
+ if (mlx5_glue->get_async_event(sh->ctx, &event))
+ break;
+ /* Retrieve and check IB port index. */
+ tmp = (uint32_t)event.element.port_num;
+ if (!tmp && event.event_type == IBV_EVENT_DEVICE_FATAL) {
+ /*
+ * The DEVICE_FATAL event is called once for
+ * entire device without port specifying.
+ * We should notify all existing ports.
+ */
+ mlx5_glue->ack_async_event(&event);
+ mlx5_dev_interrupt_device_fatal(sh);
+ continue;
+ }
+ MLX5_ASSERT(tmp && (tmp <= sh->max_port));
+ if (!tmp) {
+ /* Unsupported devive level event. */
+ mlx5_glue->ack_async_event(&event);
+ DRV_LOG(DEBUG,
+ "unsupported common event (type %d)",
+ event.event_type);
+ continue;
+ }
+ if (tmp > sh->max_port) {
+ /* Invalid IB port index. */
+ mlx5_glue->ack_async_event(&event);
+ DRV_LOG(DEBUG,
+ "cannot handle an event (type %d)"
+ "due to invalid IB port index (%u)",
+ event.event_type, tmp);
+ continue;
+ }
+ if (sh->port[tmp - 1].ih_port_id >= RTE_MAX_ETHPORTS) {
+ /* No handler installed. */
+ mlx5_glue->ack_async_event(&event);
+ DRV_LOG(DEBUG,
+ "cannot handle an event (type %d)"
+ "due to no handler installed for port %u",
+ event.event_type, tmp);
+ continue;
+ }
+ /* Retrieve ethernet device descriptor. */
+ tmp = sh->port[tmp - 1].ih_port_id;
+ dev = &rte_eth_devices[tmp];
+ MLX5_ASSERT(dev);
+ if ((event.event_type == IBV_EVENT_PORT_ACTIVE ||
+ event.event_type == IBV_EVENT_PORT_ERR) &&
+ dev->data->dev_conf.intr_conf.lsc) {
+ mlx5_glue->ack_async_event(&event);
+ if (mlx5_link_update(dev, 0) == -EAGAIN) {
+ usleep(0);
+ continue;
+ }
+ _rte_eth_dev_callback_process
+ (dev, RTE_ETH_EVENT_INTR_LSC, NULL);
+ continue;
+ }
+ DRV_LOG(DEBUG,
+ "port %u cannot handle an unknown event (type %d)",
+ dev->data->port_id, event.event_type);
+ mlx5_glue->ack_async_event(&event);
}
- return ret;
}
-/**
- * Handle delayed link status event.
+/*
+ * Unregister callback handler safely. The handler may be active
+ * while we are trying to unregister it, in this case code -EAGAIN
+ * is returned by rte_intr_callback_unregister(). This routine checks
+ * the return code and tries to unregister handler again.
*
- * @param arg
- * Registered argument.
+ * @param handle
+ * interrupt handle
+ * @param cb_fn
+ * pointer to callback routine
+ * @cb_arg
+ * opaque callback parameter
*/
void
-mlx5_dev_link_status_handler(void *arg)
+mlx5_intr_callback_unregister(const struct rte_intr_handle *handle,
+ rte_intr_callback_fn cb_fn, void *cb_arg)
{
- struct rte_eth_dev *dev = arg;
- struct priv *priv = dev->data->dev_private;
- int ret;
+ /*
+ * Try to reduce timeout management overhead by not calling
+ * the timer related routines on the first iteration. If the
+ * unregistering succeeds on first call there will be no
+ * timer calls at all.
+ */
+ uint64_t twait = 0;
+ uint64_t start = 0;
- priv_lock(priv);
- assert(priv->pending_alarm == 1);
- ret = priv_dev_link_status_handler(priv, dev);
- priv_unlock(priv);
- if (ret)
- _rte_eth_dev_callback_process(dev, RTE_ETH_EVENT_INTR_LSC, NULL);
+ do {
+ int ret;
+
+ ret = rte_intr_callback_unregister(handle, cb_fn, cb_arg);
+ if (ret >= 0)
+ return;
+ if (ret != -EAGAIN) {
+ DRV_LOG(INFO, "failed to unregister interrupt"
+ " handler (error: %d)", ret);
+ MLX5_ASSERT(false);
+ return;
+ }
+ if (twait) {
+ struct timespec onems;
+
+ /* Wait one millisecond and try again. */
+ onems.tv_sec = 0;
+ onems.tv_nsec = NS_PER_S / MS_PER_S;
+ nanosleep(&onems, 0);
+ /* Check whether one second elapsed. */
+ if ((rte_get_timer_cycles() - start) <= twait)
+ continue;
+ } else {
+ /*
+ * We get the amount of timer ticks for one second.
+ * If this amount elapsed it means we spent one
+ * second in waiting. This branch is executed once
+ * on first iteration.
+ */
+ twait = rte_get_timer_hz();
+ MLX5_ASSERT(twait);
+ }
+ /*
+ * Timeout elapsed, show message (once a second) and retry.
+ * We have no other acceptable option here, if we ignore
+ * the unregistering return code the handler will not
+ * be unregistered, fd will be closed and we may get the
+ * crush. Hanging and messaging in the loop seems not to be
+ * the worst choice.
+ */
+ DRV_LOG(INFO, "Retrying to unregister interrupt handler");
+ start = rte_get_timer_cycles();
+ } while (true);
}
/**
- * Handle interrupts from the NIC.
+ * Handle DEVX interrupts from the NIC.
+ * This function is probably called from the DPDK host thread.
*
- * @param[in] intr_handle
- * Interrupt handler.
* @param cb_arg
* Callback argument.
*/
void
-mlx5_dev_interrupt_handler(struct rte_intr_handle *intr_handle, void *cb_arg)
+mlx5_dev_interrupt_handler_devx(void *cb_arg)
{
- struct rte_eth_dev *dev = cb_arg;
- struct priv *priv = dev->data->dev_private;
- int ret;
+#ifndef HAVE_IBV_DEVX_ASYNC
+ (void)cb_arg;
+ return;
+#else
+ struct mlx5_ibv_shared *sh = cb_arg;
+ union {
+ struct mlx5dv_devx_async_cmd_hdr cmd_resp;
+ uint8_t buf[MLX5_ST_SZ_BYTES(query_flow_counter_out) +
+ MLX5_ST_SZ_BYTES(traffic_counter) +
+ sizeof(struct mlx5dv_devx_async_cmd_hdr)];
+ } out;
+ uint8_t *buf = out.buf + sizeof(out.cmd_resp);
+
+ while (!mlx5_glue->devx_get_async_cmd_comp(sh->devx_comp,
+ &out.cmd_resp,
+ sizeof(out.buf)))
+ mlx5_flow_async_pool_query_handle
+ (sh, (uint64_t)out.cmd_resp.wr_id,
+ mlx5_devx_get_out_command_status(buf));
+#endif /* HAVE_IBV_DEVX_ASYNC */
+}
- (void)intr_handle;
- priv_lock(priv);
- ret = priv_dev_link_status_handler(priv, dev);
- priv_unlock(priv);
- if (ret)
- _rte_eth_dev_callback_process(dev, RTE_ETH_EVENT_INTR_LSC, NULL);
+/**
+ * Uninstall shared asynchronous device events handler.
+ * This function is implemented to support event sharing
+ * between multiple ports of single IB device.
+ *
+ * @param dev
+ * Pointer to Ethernet device.
+ */
+static void
+mlx5_dev_shared_handler_uninstall(struct rte_eth_dev *dev)
+{
+ struct mlx5_priv *priv = dev->data->dev_private;
+ struct mlx5_ibv_shared *sh = priv->sh;
+
+ if (rte_eal_process_type() != RTE_PROC_PRIMARY)
+ return;
+ pthread_mutex_lock(&sh->intr_mutex);
+ MLX5_ASSERT(priv->ibv_port);
+ MLX5_ASSERT(priv->ibv_port <= sh->max_port);
+ MLX5_ASSERT(dev->data->port_id < RTE_MAX_ETHPORTS);
+ if (sh->port[priv->ibv_port - 1].ih_port_id >= RTE_MAX_ETHPORTS)
+ goto exit;
+ MLX5_ASSERT(sh->port[priv->ibv_port - 1].ih_port_id ==
+ (uint32_t)dev->data->port_id);
+ MLX5_ASSERT(sh->intr_cnt);
+ sh->port[priv->ibv_port - 1].ih_port_id = RTE_MAX_ETHPORTS;
+ if (!sh->intr_cnt || --sh->intr_cnt)
+ goto exit;
+ mlx5_intr_callback_unregister(&sh->intr_handle,
+ mlx5_dev_interrupt_handler, sh);
+ sh->intr_handle.fd = 0;
+ sh->intr_handle.type = RTE_INTR_HANDLE_UNKNOWN;
+exit:
+ pthread_mutex_unlock(&sh->intr_mutex);
}
/**
- * Uninstall interrupt handler.
+ * Uninstall devx shared asynchronous device events handler.
+ * This function is implemeted to support event sharing
+ * between multiple ports of single IB device.
*
- * @param priv
- * Pointer to private structure.
* @param dev
- * Pointer to the rte_eth_dev structure.
+ * Pointer to Ethernet device.
*/
-void
-priv_dev_interrupt_handler_uninstall(struct priv *priv, struct rte_eth_dev *dev)
+static void
+mlx5_dev_shared_handler_devx_uninstall(struct rte_eth_dev *dev)
{
- if (!dev->data->dev_conf.intr_conf.lsc)
+ struct mlx5_priv *priv = dev->data->dev_private;
+ struct mlx5_ibv_shared *sh = priv->sh;
+
+ if (rte_eal_process_type() != RTE_PROC_PRIMARY)
return;
- rte_intr_callback_unregister(&priv->intr_handle,
- mlx5_dev_interrupt_handler,
- dev);
- if (priv->pending_alarm)
- rte_eal_alarm_cancel(mlx5_dev_link_status_handler, dev);
- priv->pending_alarm = 0;
- priv->intr_handle.fd = 0;
- priv->intr_handle.type = RTE_INTR_HANDLE_UNKNOWN;
+ pthread_mutex_lock(&sh->intr_mutex);
+ MLX5_ASSERT(priv->ibv_port);
+ MLX5_ASSERT(priv->ibv_port <= sh->max_port);
+ MLX5_ASSERT(dev->data->port_id < RTE_MAX_ETHPORTS);
+ if (sh->port[priv->ibv_port - 1].devx_ih_port_id >= RTE_MAX_ETHPORTS)
+ goto exit;
+ MLX5_ASSERT(sh->port[priv->ibv_port - 1].devx_ih_port_id ==
+ (uint32_t)dev->data->port_id);
+ sh->port[priv->ibv_port - 1].devx_ih_port_id = RTE_MAX_ETHPORTS;
+ if (!sh->devx_intr_cnt || --sh->devx_intr_cnt)
+ goto exit;
+ if (sh->intr_handle_devx.fd) {
+ rte_intr_callback_unregister(&sh->intr_handle_devx,
+ mlx5_dev_interrupt_handler_devx,
+ sh);
+ sh->intr_handle_devx.fd = 0;
+ sh->intr_handle_devx.type = RTE_INTR_HANDLE_UNKNOWN;
+ }
+ if (sh->devx_comp) {
+ mlx5_glue->devx_destroy_cmd_comp(sh->devx_comp);
+ sh->devx_comp = NULL;
+ }
+exit:
+ pthread_mutex_unlock(&sh->intr_mutex);
}
/**
- * Install interrupt handler.
+ * Install shared asynchronous device events handler.
+ * This function is implemented to support event sharing
+ * between multiple ports of single IB device.
*
- * @param priv
- * Pointer to private structure.
* @param dev
- * Pointer to the rte_eth_dev structure.
+ * Pointer to Ethernet device.
*/
-void
-priv_dev_interrupt_handler_install(struct priv *priv, struct rte_eth_dev *dev)
+static void
+mlx5_dev_shared_handler_install(struct rte_eth_dev *dev)
{
- int rc, flags;
+ struct mlx5_priv *priv = dev->data->dev_private;
+ struct mlx5_ibv_shared *sh = priv->sh;
+ int ret;
+ int flags;
- if (!dev->data->dev_conf.intr_conf.lsc)
+ if (rte_eal_process_type() != RTE_PROC_PRIMARY)
return;
- 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");
+ pthread_mutex_lock(&sh->intr_mutex);
+ MLX5_ASSERT(priv->ibv_port);
+ MLX5_ASSERT(priv->ibv_port <= sh->max_port);
+ MLX5_ASSERT(dev->data->port_id < RTE_MAX_ETHPORTS);
+ if (sh->port[priv->ibv_port - 1].ih_port_id < RTE_MAX_ETHPORTS) {
+ /* The handler is already installed for this port. */
+ MLX5_ASSERT(sh->intr_cnt);
+ goto exit;
+ }
+ if (sh->intr_cnt) {
+ sh->port[priv->ibv_port - 1].ih_port_id =
+ (uint32_t)dev->data->port_id;
+ sh->intr_cnt++;
+ goto exit;
+ }
+ /* No shared handler installed. */
+ MLX5_ASSERT(sh->ctx->async_fd > 0);
+ flags = fcntl(sh->ctx->async_fd, F_GETFL);
+ ret = fcntl(sh->ctx->async_fd, F_SETFL, flags | O_NONBLOCK);
+ if (ret) {
+ DRV_LOG(INFO, "failed to change file descriptor async event"
+ " queue");
+ /* Indicate there will be no interrupts. */
dev->data->dev_conf.intr_conf.lsc = 0;
+ dev->data->dev_conf.intr_conf.rmv = 0;
} else {
- 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);
+ sh->intr_handle.fd = sh->ctx->async_fd;
+ sh->intr_handle.type = RTE_INTR_HANDLE_EXT;
+ rte_intr_callback_register(&sh->intr_handle,
+ mlx5_dev_interrupt_handler, sh);
+ sh->intr_cnt++;
+ sh->port[priv->ibv_port - 1].ih_port_id =
+ (uint32_t)dev->data->port_id;
}
+exit:
+ pthread_mutex_unlock(&sh->intr_mutex);
}
/**
- * Change the link state (UP / DOWN).
+ * Install devx shared asyncronous device events handler.
+ * This function is implemeted to support event sharing
+ * between multiple ports of single IB device.
*
- * @param priv
- * Pointer to Ethernet device structure.
- * @param up
- * Nonzero for link up, otherwise link down.
- *
- * @return
- * 0 on success, errno value on failure.
+ * @param dev
+ * Pointer to Ethernet device.
*/
-static int
-priv_set_link(struct priv *priv, int up)
+static void
+mlx5_dev_shared_handler_devx_install(struct rte_eth_dev *dev)
{
- struct rte_eth_dev *dev = priv->dev;
- int err;
-
- if (up) {
- err = priv_set_flags(priv, ~IFF_UP, IFF_UP);
- if (err)
- return err;
- priv_select_tx_function(priv);
- priv_select_rx_function(priv);
- } else {
- err = priv_set_flags(priv, ~IFF_UP, ~IFF_UP);
- if (err)
- return err;
- dev->rx_pkt_burst = removed_rx_burst;
- dev->tx_pkt_burst = removed_tx_burst;
+ struct mlx5_priv *priv = dev->data->dev_private;
+ struct mlx5_ibv_shared *sh = priv->sh;
+
+ if (rte_eal_process_type() != RTE_PROC_PRIMARY)
+ return;
+ pthread_mutex_lock(&sh->intr_mutex);
+ MLX5_ASSERT(priv->ibv_port);
+ MLX5_ASSERT(priv->ibv_port <= sh->max_port);
+ MLX5_ASSERT(dev->data->port_id < RTE_MAX_ETHPORTS);
+ if (sh->port[priv->ibv_port - 1].devx_ih_port_id < RTE_MAX_ETHPORTS) {
+ /* The handler is already installed for this port. */
+ MLX5_ASSERT(sh->devx_intr_cnt);
+ goto exit;
}
- return 0;
+ if (sh->devx_intr_cnt) {
+ sh->devx_intr_cnt++;
+ sh->port[priv->ibv_port - 1].devx_ih_port_id =
+ (uint32_t)dev->data->port_id;
+ goto exit;
+ }
+ if (priv->config.devx) {
+#ifndef HAVE_IBV_DEVX_ASYNC
+ goto exit;
+#else
+ sh->devx_comp = mlx5_glue->devx_create_cmd_comp(sh->ctx);
+ if (sh->devx_comp) {
+ int flags = fcntl(sh->devx_comp->fd, F_GETFL);
+ int ret = fcntl(sh->devx_comp->fd, F_SETFL,
+ flags | O_NONBLOCK);
+
+ if (ret) {
+ DRV_LOG(INFO, "failed to change file descriptor"
+ " devx async event queue");
+ } else {
+ sh->intr_handle_devx.fd = sh->devx_comp->fd;
+ sh->intr_handle_devx.type = RTE_INTR_HANDLE_EXT;
+ rte_intr_callback_register
+ (&sh->intr_handle_devx,
+ mlx5_dev_interrupt_handler_devx, sh);
+ sh->devx_intr_cnt++;
+ sh->port[priv->ibv_port - 1].devx_ih_port_id =
+ (uint32_t)dev->data->port_id;
+ }
+ }
+#endif /* HAVE_IBV_DEVX_ASYNC */
+ }
+exit:
+ pthread_mutex_unlock(&sh->intr_mutex);
+}
+
+/**
+ * Uninstall interrupt handler.
+ *
+ * @param dev
+ * Pointer to Ethernet device.
+ */
+void
+mlx5_dev_interrupt_handler_uninstall(struct rte_eth_dev *dev)
+{
+ mlx5_dev_shared_handler_uninstall(dev);
+}
+
+/**
+ * Install interrupt handler.
+ *
+ * @param dev
+ * Pointer to Ethernet device.
+ */
+void
+mlx5_dev_interrupt_handler_install(struct rte_eth_dev *dev)
+{
+ mlx5_dev_shared_handler_install(dev);
+}
+
+/**
+ * Devx uninstall interrupt handler.
+ *
+ * @param dev
+ * Pointer to Ethernet device.
+ */
+void
+mlx5_dev_interrupt_handler_devx_uninstall(struct rte_eth_dev *dev)
+{
+ mlx5_dev_shared_handler_devx_uninstall(dev);
+}
+
+/**
+ * Devx install interrupt handler.
+ *
+ * @param dev
+ * Pointer to Ethernet device.
+ */
+void
+mlx5_dev_interrupt_handler_devx_install(struct rte_eth_dev *dev)
+{
+ mlx5_dev_shared_handler_devx_install(dev);
}
/**
* Pointer to Ethernet device structure.
*
* @return
- * 0 on success, errno value on failure.
+ * 0 on success, a negative errno value otherwise and rte_errno is set.
*/
int
mlx5_set_link_down(struct rte_eth_dev *dev)
{
- struct priv *priv = dev->data->dev_private;
- int err;
-
- priv_lock(priv);
- err = priv_set_link(priv, 0);
- priv_unlock(priv);
- return err;
+ return mlx5_set_flags(dev, ~IFF_UP, ~IFF_UP);
}
/**
* Pointer to Ethernet device structure.
*
* @return
- * 0 on success, errno value on failure.
+ * 0 on success, a negative errno value otherwise and rte_errno is set.
*/
int
mlx5_set_link_up(struct rte_eth_dev *dev)
{
- struct priv *priv = dev->data->dev_private;
- int err;
+ return mlx5_set_flags(dev, ~IFF_UP, IFF_UP);
+}
- priv_lock(priv);
- err = priv_set_link(priv, 1);
- priv_unlock(priv);
- return err;
+/**
+ * Configure the RX function to use.
+ *
+ * @param dev
+ * Pointer to private data structure.
+ *
+ * @return
+ * Pointer to selected Rx burst function.
+ */
+eth_rx_burst_t
+mlx5_select_rx_function(struct rte_eth_dev *dev)
+{
+ eth_rx_burst_t rx_pkt_burst = mlx5_rx_burst;
+
+ MLX5_ASSERT(dev != NULL);
+ if (mlx5_check_vec_rx_support(dev) > 0) {
+ rx_pkt_burst = mlx5_rx_burst_vec;
+ DRV_LOG(DEBUG, "port %u selected Rx vectorized function",
+ dev->data->port_id);
+ } else if (mlx5_mprq_enabled(dev)) {
+ rx_pkt_burst = mlx5_rx_burst_mprq;
+ }
+ return rx_pkt_burst;
}
/**
- * Configure secondary process queues from a private data pointer (primary
- * or secondary) and update burst callbacks. Can take place only once.
+ * Check if mlx5 device was removed.
+ *
+ * @param dev
+ * Pointer to Ethernet device structure.
*
- * All queues must have been previously created by the primary process to
- * avoid undefined behavior.
+ * @return
+ * 1 when device is removed, otherwise 0.
+ */
+int
+mlx5_is_removed(struct rte_eth_dev *dev)
+{
+ struct ibv_device_attr device_attr;
+ struct mlx5_priv *priv = dev->data->dev_private;
+
+ if (mlx5_glue->query_device(priv->sh->ctx, &device_attr) == EIO)
+ return 1;
+ return 0;
+}
+
+/**
+ * Get the E-Switch parameters by port id.
*
- * @param priv
- * Private data pointer from either primary or secondary process.
+ * @param[in] port
+ * Device port id.
+ * @param[in] valid
+ * Device port id is valid, skip check. This flag is useful
+ * when trials are performed from probing and device is not
+ * flagged as valid yet (in attaching process).
+ * @param[out] es_domain_id
+ * E-Switch domain id.
+ * @param[out] es_port_id
+ * The port id of the port in the E-Switch.
*
* @return
- * Private data pointer from secondary process, NULL in case of error.
+ * pointer to device private data structure containing data needed
+ * on success, NULL otherwise and rte_errno is set.
*/
-struct priv *
-mlx5_secondary_data_setup(struct priv *priv)
+struct mlx5_priv *
+mlx5_port_to_eswitch_info(uint16_t port, bool valid)
{
- unsigned int port_id = 0;
- struct mlx5_secondary_data *sd;
- void **tx_queues;
- void **rx_queues;
- unsigned int nb_tx_queues;
- unsigned int nb_rx_queues;
- unsigned int i;
+ struct rte_eth_dev *dev;
+ struct mlx5_priv *priv;
- /* priv must be valid at this point. */
- assert(priv != NULL);
- /* priv->dev must also be valid but may point to local memory from
- * another process, possibly with the same address and must not
- * be dereferenced yet. */
- assert(priv->dev != NULL);
- /* Determine port ID by finding out where priv comes from. */
- while (1) {
- sd = &mlx5_secondary_data[port_id];
- rte_spinlock_lock(&sd->lock);
- /* Primary process? */
- if (sd->primary_priv == priv)
- break;
- /* Secondary process? */
- if (sd->data.dev_private == priv)
- break;
- rte_spinlock_unlock(&sd->lock);
- if (++port_id == RTE_DIM(mlx5_secondary_data))
- port_id = 0;
+ if (port >= RTE_MAX_ETHPORTS) {
+ rte_errno = EINVAL;
+ return NULL;
}
- /* Switch to secondary private structure. If private data has already
- * been updated by another thread, there is nothing else to do. */
- priv = sd->data.dev_private;
- if (priv->dev->data == &sd->data)
- goto end;
- /* Sanity checks. Secondary private structure is supposed to point
- * to local eth_dev, itself still pointing to the shared device data
- * structure allocated by the primary process. */
- assert(sd->shared_dev_data != &sd->data);
- assert(sd->data.nb_tx_queues == 0);
- assert(sd->data.tx_queues == NULL);
- assert(sd->data.nb_rx_queues == 0);
- assert(sd->data.rx_queues == NULL);
- assert(priv != sd->primary_priv);
- assert(priv->dev->data == sd->shared_dev_data);
- assert(priv->txqs_n == 0);
- assert(priv->txqs == NULL);
- assert(priv->rxqs_n == 0);
- assert(priv->rxqs == NULL);
- nb_tx_queues = sd->shared_dev_data->nb_tx_queues;
- nb_rx_queues = sd->shared_dev_data->nb_rx_queues;
- /* Allocate local storage for queues. */
- tx_queues = rte_zmalloc("secondary ethdev->tx_queues",
- sizeof(sd->data.tx_queues[0]) * nb_tx_queues,
- RTE_CACHE_LINE_SIZE);
- rx_queues = rte_zmalloc("secondary ethdev->rx_queues",
- sizeof(sd->data.rx_queues[0]) * nb_rx_queues,
- RTE_CACHE_LINE_SIZE);
- if (tx_queues == NULL || rx_queues == NULL)
- goto error;
- /* Lock to prevent control operations during setup. */
- priv_lock(priv);
- /* TX queues. */
- for (i = 0; i != nb_tx_queues; ++i) {
- struct txq *primary_txq = (*sd->primary_priv->txqs)[i];
- struct txq_ctrl *primary_txq_ctrl;
- struct txq_ctrl *txq_ctrl;
-
- if (primary_txq == NULL)
- continue;
- primary_txq_ctrl = container_of(primary_txq,
- struct txq_ctrl, txq);
- txq_ctrl = rte_calloc_socket("TXQ", 1, sizeof(*txq_ctrl) +
- (1 << primary_txq->elts_n) *
- sizeof(struct rte_mbuf *), 0,
- primary_txq_ctrl->socket);
- if (txq_ctrl != NULL) {
- if (txq_ctrl_setup(priv->dev,
- txq_ctrl,
- 1 << primary_txq->elts_n,
- primary_txq_ctrl->socket,
- NULL) == 0) {
- txq_ctrl->txq.stats.idx =
- primary_txq->stats.idx;
- tx_queues[i] = &txq_ctrl->txq;
- continue;
- }
- rte_free(txq_ctrl);
- }
- while (i) {
- txq_ctrl = tx_queues[--i];
- txq_cleanup(txq_ctrl);
- rte_free(txq_ctrl);
- }
- goto error;
+ if (!valid && !rte_eth_dev_is_valid_port(port)) {
+ rte_errno = ENODEV;
+ return NULL;
+ }
+ dev = &rte_eth_devices[port];
+ priv = dev->data->dev_private;
+ if (!(priv->representor || priv->master)) {
+ rte_errno = EINVAL;
+ return NULL;
}
- /* RX queues. */
- for (i = 0; i != nb_rx_queues; ++i) {
- struct rxq_ctrl *primary_rxq =
- container_of((*sd->primary_priv->rxqs)[i],
- struct rxq_ctrl, rxq);
+ return priv;
+}
- if (primary_rxq == NULL)
- continue;
- /* Not supported yet. */
- rx_queues[i] = NULL;
+/**
+ * Get the E-Switch parameters by device instance.
+ *
+ * @param[in] port
+ * Device port id.
+ * @param[out] es_domain_id
+ * E-Switch domain id.
+ * @param[out] es_port_id
+ * The port id of the port in the E-Switch.
+ *
+ * @return
+ * pointer to device private data structure containing data needed
+ * on success, NULL otherwise and rte_errno is set.
+ */
+struct mlx5_priv *
+mlx5_dev_to_eswitch_info(struct rte_eth_dev *dev)
+{
+ struct mlx5_priv *priv;
+
+ priv = dev->data->dev_private;
+ if (!(priv->representor || priv->master)) {
+ rte_errno = EINVAL;
+ return NULL;
}
- /* Update everything. */
- priv->txqs = (void *)tx_queues;
- priv->txqs_n = nb_tx_queues;
- priv->rxqs = (void *)rx_queues;
- priv->rxqs_n = nb_rx_queues;
- sd->data.rx_queues = rx_queues;
- sd->data.tx_queues = tx_queues;
- sd->data.nb_rx_queues = nb_rx_queues;
- sd->data.nb_tx_queues = nb_tx_queues;
- sd->data.dev_link = sd->shared_dev_data->dev_link;
- sd->data.mtu = sd->shared_dev_data->mtu;
- memcpy(sd->data.rx_queue_state, sd->shared_dev_data->rx_queue_state,
- sizeof(sd->data.rx_queue_state));
- memcpy(sd->data.tx_queue_state, sd->shared_dev_data->tx_queue_state,
- sizeof(sd->data.tx_queue_state));
- sd->data.dev_flags = sd->shared_dev_data->dev_flags;
- /* Use local data from now on. */
- rte_mb();
- priv->dev->data = &sd->data;
- rte_mb();
- priv_select_tx_function(priv);
- priv_select_rx_function(priv);
- priv_unlock(priv);
-end:
- /* More sanity checks. */
- assert(priv->dev->data == &sd->data);
- rte_spinlock_unlock(&sd->lock);
return priv;
-error:
- priv_unlock(priv);
- rte_free(tx_queues);
- rte_free(rx_queues);
- rte_spinlock_unlock(&sd->lock);
- return NULL;
}
/**
- * Configure the TX function to use.
+ * Get switch information associated with network interface.
*
- * @param priv
- * Pointer to private structure.
+ * @param ifindex
+ * Network interface index.
+ * @param[out] info
+ * Switch information object, populated in case of success.
+ *
+ * @return
+ * 0 on success, a negative errno value otherwise and rte_errno is set.
*/
-void
-priv_select_tx_function(struct priv *priv)
+int
+mlx5_sysfs_switch_info(unsigned int ifindex, struct mlx5_switch_info *info)
{
- priv->dev->tx_pkt_burst = mlx5_tx_burst;
- /* Display warning for unsupported configurations. */
- if (priv->sriov && priv->mps)
- WARN("multi-packet send WQE cannot be used on a SR-IOV setup");
- /* Select appropriate TX function. */
- if ((priv->sriov == 0) && priv->mps && priv->txq_inline) {
- priv->dev->tx_pkt_burst = mlx5_tx_burst_mpw_inline;
- DEBUG("selected MPW inline TX function");
- } else if ((priv->sriov == 0) && priv->mps) {
- priv->dev->tx_pkt_burst = mlx5_tx_burst_mpw;
- DEBUG("selected MPW TX function");
+ char ifname[IF_NAMESIZE];
+ char port_name[IF_NAMESIZE];
+ FILE *file;
+ struct mlx5_switch_info data = {
+ .master = 0,
+ .representor = 0,
+ .name_type = MLX5_PHYS_PORT_NAME_TYPE_NOTSET,
+ .port_name = 0,
+ .switch_id = 0,
+ };
+ DIR *dir;
+ bool port_switch_id_set = false;
+ bool device_dir = false;
+ char c;
+ int ret;
+
+ if (!if_indextoname(ifindex, ifname)) {
+ rte_errno = errno;
+ return -rte_errno;
+ }
+
+ MKSTR(phys_port_name, "/sys/class/net/%s/phys_port_name",
+ ifname);
+ MKSTR(phys_switch_id, "/sys/class/net/%s/phys_switch_id",
+ ifname);
+ MKSTR(pci_device, "/sys/class/net/%s/device",
+ ifname);
+
+ file = fopen(phys_port_name, "rb");
+ if (file != NULL) {
+ ret = fscanf(file, "%s", port_name);
+ fclose(file);
+ if (ret == 1)
+ mlx5_translate_port_name(port_name, &data);
+ }
+ file = fopen(phys_switch_id, "rb");
+ if (file == NULL) {
+ rte_errno = errno;
+ return -rte_errno;
}
+ port_switch_id_set =
+ fscanf(file, "%" SCNx64 "%c", &data.switch_id, &c) == 2 &&
+ c == '\n';
+ fclose(file);
+ dir = opendir(pci_device);
+ if (dir != NULL) {
+ closedir(dir);
+ device_dir = true;
+ }
+ if (port_switch_id_set) {
+ /* We have some E-Switch configuration. */
+ mlx5_sysfs_check_switch_info(device_dir, &data);
+ }
+ *info = data;
+ MLX5_ASSERT(!(data.master && data.representor));
+ if (data.master && data.representor) {
+ DRV_LOG(ERR, "ifindex %u device is recognized as master"
+ " and as representor", ifindex);
+ rte_errno = ENODEV;
+ return -rte_errno;
+ }
+ return 0;
}
/**
- * Configure the RX function to use.
+ * Analyze gathered port parameters via sysfs to recognize master
+ * and representor devices for E-Switch configuration.
*
- * @param priv
- * Pointer to private structure.
+ * @param[in] device_dir
+ * flag of presence of "device" directory under port device key.
+ * @param[inout] switch_info
+ * Port information, including port name as a number and port name
+ * type if recognized
+ *
+ * @return
+ * master and representor flags are set in switch_info according to
+ * recognized parameters (if any).
*/
void
-priv_select_rx_function(struct priv *priv)
+mlx5_sysfs_check_switch_info(bool device_dir,
+ struct mlx5_switch_info *switch_info)
{
- priv->dev->rx_pkt_burst = mlx5_rx_burst;
+ switch (switch_info->name_type) {
+ case MLX5_PHYS_PORT_NAME_TYPE_UNKNOWN:
+ /*
+ * Name is not recognized, assume the master,
+ * check the device directory presence.
+ */
+ switch_info->master = device_dir;
+ break;
+ case MLX5_PHYS_PORT_NAME_TYPE_NOTSET:
+ /*
+ * Name is not set, this assumes the legacy naming
+ * schema for master, just check if there is
+ * a device directory.
+ */
+ switch_info->master = device_dir;
+ break;
+ case MLX5_PHYS_PORT_NAME_TYPE_UPLINK:
+ /* New uplink naming schema recognized. */
+ switch_info->master = 1;
+ break;
+ case MLX5_PHYS_PORT_NAME_TYPE_LEGACY:
+ /* Legacy representors naming schema. */
+ switch_info->representor = !device_dir;
+ break;
+ case MLX5_PHYS_PORT_NAME_TYPE_PFVF:
+ /* New representors naming schema. */
+ switch_info->representor = 1;
+ break;
+ }
+}
+
+/**
+ * DPDK callback to retrieve plug-in module EEPROM information (type and size).
+ *
+ * @param dev
+ * Pointer to Ethernet device structure.
+ * @param[out] modinfo
+ * Storage for plug-in module EEPROM information.
+ *
+ * @return
+ * 0 on success, a negative errno value otherwise and rte_errno is set.
+ */
+int
+mlx5_get_module_info(struct rte_eth_dev *dev,
+ struct rte_eth_dev_module_info *modinfo)
+{
+ struct ethtool_modinfo info = {
+ .cmd = ETHTOOL_GMODULEINFO,
+ };
+ struct ifreq ifr = (struct ifreq) {
+ .ifr_data = (void *)&info,
+ };
+ int ret = 0;
+
+ if (!dev || !modinfo) {
+ DRV_LOG(WARNING, "missing argument, cannot get module info");
+ rte_errno = EINVAL;
+ return -rte_errno;
+ }
+ ret = mlx5_ifreq(dev, SIOCETHTOOL, &ifr);
+ if (ret) {
+ DRV_LOG(WARNING, "port %u ioctl(SIOCETHTOOL) failed: %s",
+ dev->data->port_id, strerror(rte_errno));
+ return ret;
+ }
+ modinfo->type = info.type;
+ modinfo->eeprom_len = info.eeprom_len;
+ return ret;
+}
+
+/**
+ * DPDK callback to retrieve plug-in module EEPROM data.
+ *
+ * @param dev
+ * Pointer to Ethernet device structure.
+ * @param[out] info
+ * Storage for plug-in module EEPROM data.
+ *
+ * @return
+ * 0 on success, a negative errno value otherwise and rte_errno is set.
+ */
+int mlx5_get_module_eeprom(struct rte_eth_dev *dev,
+ struct rte_dev_eeprom_info *info)
+{
+ struct ethtool_eeprom *eeprom;
+ struct ifreq ifr;
+ int ret = 0;
+
+ if (!dev || !info) {
+ DRV_LOG(WARNING, "missing argument, cannot get module eeprom");
+ rte_errno = EINVAL;
+ return -rte_errno;
+ }
+ eeprom = rte_calloc(__func__, 1,
+ (sizeof(struct ethtool_eeprom) + info->length), 0);
+ if (!eeprom) {
+ DRV_LOG(WARNING, "port %u cannot allocate memory for "
+ "eeprom data", dev->data->port_id);
+ rte_errno = ENOMEM;
+ return -rte_errno;
+ }
+ eeprom->cmd = ETHTOOL_GMODULEEEPROM;
+ eeprom->offset = info->offset;
+ eeprom->len = info->length;
+ ifr = (struct ifreq) {
+ .ifr_data = (void *)eeprom,
+ };
+ ret = mlx5_ifreq(dev, SIOCETHTOOL, &ifr);
+ if (ret)
+ DRV_LOG(WARNING, "port %u ioctl(SIOCETHTOOL) failed: %s",
+ dev->data->port_id, strerror(rte_errno));
+ else
+ rte_memcpy(info->data, eeprom->data, info->length);
+ rte_free(eeprom);
+ return ret;
+}
+
+/**
+ * DPDK callback to retrieve hairpin capabilities.
+ *
+ * @param dev
+ * Pointer to Ethernet device structure.
+ * @param[out] cap
+ * Storage for hairpin capability data.
+ *
+ * @return
+ * 0 on success, a negative errno value otherwise and rte_errno is set.
+ */
+int mlx5_hairpin_cap_get(struct rte_eth_dev *dev,
+ struct rte_eth_hairpin_cap *cap)
+{
+ struct mlx5_priv *priv = dev->data->dev_private;
+
+ if (priv->sh->devx == 0) {
+ rte_errno = ENOTSUP;
+ return -rte_errno;
+ }
+ cap->max_nb_queues = UINT16_MAX;
+ cap->max_rx_2_tx = 1;
+ cap->max_tx_2_rx = 1;
+ cap->max_nb_desc = 8192;
+ return 0;
}