-/*-
- * 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 <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 "-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 "mlx5.h"
+#include "mlx5_glue.h"
#include "mlx5_rxtx.h"
#include "mlx5_utils.h"
#define ETHTOOL_LINK_MODE_100000baseCR4_Full_BIT 38
#define ETHTOOL_LINK_MODE_100000baseLR4_ER4_Full_BIT 39
#endif
-#define ETHTOOL_LINK_MODE_MASK_MAX_KERNEL_NU32 (SCHAR_MAX)
-
-/**
- * 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;
-}
-
-/**
- * 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;
-}
/**
* 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;
}
-/**
- * Check if the counter is located on ib counters file.
- *
- * @param[in] cntr
- * Counter name.
- *
- * @return
- * 1 if counter is located on ib counters file , 0 otherwise.
- */
-int
-priv_is_ib_cntr(const char *cntr)
-{
- if (!strcmp(cntr, "out_of_buffer"))
- return 1;
- 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;
-
- if (priv_is_ib_cntr(entry)) {
- MKSTR(path, "%s/ports/1/hw_counters/%s",
- priv->ctx->device->ibdev_path, entry);
- file = fopen(path, "rb");
- } else {
- 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.
*
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;
-}
-
/**
* Get device MTU.
*
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;
- return 0;
-}
-
-/**
- * Read device counter from sysfs.
- *
- * @param priv
- * Pointer to private structure.
- * @param name
- * Counter name.
- * @param[out] cntr
- * Counter output buffer.
- *
- * @return
- * 0 on success, -1 on failure and errno is set.
- */
-int
-priv_get_cntr_sysfs(struct priv *priv, const char *name, uint64_t *cntr)
-{
- unsigned long ulong_ctr;
-
- if (priv_get_sysfs_ulong(priv, name, &ulong_ctr) == -1)
- return -1;
- *cntr = ulong_ctr;
+ if (ret)
+ return ret;
+ *mtu = request.ifr_mtu;
return 0;
}
static int
priv_set_mtu(struct priv *priv, 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 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 & (~keep));
- 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.
+ * 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 i;
unsigned int j;
unsigned int reta_idx_n;
-
- priv->rss_hf = dev->data->dev_conf.rx_adv_conf.rss_conf.rss_hf;
+ 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->ind_table_max_size) {
+ if (rxqs_n > priv->config.ind_table_max_size) {
ERROR("cannot handle this many RX queues (%u)", rxqs_n);
return EINVAL;
}
* 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 :
+ priv->config.ind_table_max_size :
rxqs_n));
if (priv_rss_reta_index_resize(priv, reta_idx_n))
return ENOMEM;
j = 0;
}
return 0;
-}
-
-/**
- * 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 (mlx5_is_secondary())
- return -E_RTE_SECONDARY;
-
- priv_lock(priv);
- ret = dev_configure(dev);
- assert(ret >= 0);
- priv_unlock(priv);
- return -ret;
}
/**
void
mlx5_dev_infos_get(struct rte_eth_dev *dev, struct rte_eth_dev_info *info)
{
- struct priv *priv = mlx5_get_priv(dev);
+ struct priv *priv = dev->data->dev_private;
+ struct mlx5_dev_config *config = &priv->config;
unsigned int max;
char ifname[IF_NAMESIZE];
info->pci_dev = RTE_ETH_DEV_TO_PCI(dev);
-
- priv_lock(priv);
/* 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;
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->tso)
- info->tx_offload_capa |= DEV_TX_OFFLOAD_TCP_TSO;
- if (priv->tunnel_en)
- info->tx_offload_capa |= (DEV_TX_OFFLOAD_OUTER_IPV4_CKSUM |
- DEV_TX_OFFLOAD_VXLAN_TNL_TSO |
- DEV_TX_OFFLOAD_GRE_TNL_TSO);
+ 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);
info->reta_size = priv->reta_idx_n ?
- priv->reta_idx_n : priv->ind_table_max_size;
- info->hash_key_size = ((*priv->rss_conf) ?
- (*priv->rss_conf)[0]->rss_key_len :
- 0);
+ 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;
- priv_unlock(priv);
+ info->flow_type_rss_offloads = ~MLX5_RSS_HF_MASK;
}
+/**
+ * Get supported packet types.
+ *
+ * @param dev
+ * Pointer to Ethernet device structure.
+ *
+ * @return
+ * A pointer to the supported Packet types array.
+ */
const uint32_t *
mlx5_dev_supported_ptypes_get(struct rte_eth_dev *dev)
{
*
* @param dev
* Pointer to Ethernet device structure.
- * @param wait_to_complete
- * Wait for request completion (ignored).
+ *
+ * @return
+ * 0 on success, -1 on error.
*/
static int
-mlx5_link_update_unlocked_gset(struct rte_eth_dev *dev, int wait_to_complete)
+mlx5_link_update_unlocked_gset(struct rte_eth_dev *dev)
{
- struct priv *priv = mlx5_get_priv(dev);
+ struct priv *priv = dev->data->dev_private;
struct ethtool_cmd edata = {
.cmd = ETHTOOL_GSET /* Deprecated since Linux v4.5. */
};
struct rte_eth_link dev_link;
int link_speed = 0;
- /* priv_lock() is not taken to allow concurrent calls. */
-
- (void)wait_to_complete;
if (priv_ifreq(priv, SIOCGIFFLAGS, &ifr)) {
WARN("ioctl(SIOCGIFFLAGS) failed: %s", strerror(errno));
return -1;
*
* @param dev
* Pointer to Ethernet device structure.
- * @param wait_to_complete
- * Wait for request completion (ignored).
+ *
+ * @return
+ * 0 on success, -1 on error.
*/
static int
-mlx5_link_update_unlocked_gs(struct rte_eth_dev *dev, int wait_to_complete)
+mlx5_link_update_unlocked_gs(struct rte_eth_dev *dev)
{
- struct priv *priv = mlx5_get_priv(dev);
- __extension__ struct {
- struct ethtool_link_settings edata;
- uint32_t link_mode_data[3 *
- ETHTOOL_LINK_MODE_MASK_MAX_KERNEL_NU32];
- } ecmd;
-
+ 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;
- (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));
- memset(&ecmd, 0, sizeof(ecmd));
- ecmd.edata.cmd = ETHTOOL_GLINKSETTINGS;
- ifr.ifr_data = (void *)&ecmd;
+ ifr.ifr_data = (void *)&gcmd;
if (priv_ifreq(priv, SIOCETHTOOL, &ifr)) {
DEBUG("ioctl(SIOCETHTOOL, ETHTOOL_GLINKSETTINGS) failed: %s",
strerror(errno));
return -1;
}
- ecmd.edata.link_mode_masks_nwords = -ecmd.edata.link_mode_masks_nwords;
+ 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.edata.speed;
- sc = ecmd.edata.link_mode_masks[0] |
- ((uint64_t)ecmd.edata.link_mode_masks[1] << 32);
+ 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 & ETHTOOL_LINK_MODE_Autoneg_BIT)
+ if (sc & MLX5_BITSHIFT(ETHTOOL_LINK_MODE_Autoneg_BIT))
priv->link_speed_capa |= ETH_LINK_SPEED_AUTONEG;
- if (sc & (ETHTOOL_LINK_MODE_1000baseT_Full_BIT |
- ETHTOOL_LINK_MODE_1000baseKX_Full_BIT))
+ 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 & (ETHTOOL_LINK_MODE_10000baseKX4_Full_BIT |
- ETHTOOL_LINK_MODE_10000baseKR_Full_BIT |
- ETHTOOL_LINK_MODE_10000baseR_FEC_BIT))
+ 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 & (ETHTOOL_LINK_MODE_20000baseMLD2_Full_BIT |
- ETHTOOL_LINK_MODE_20000baseKR2_Full_BIT))
+ 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 & (ETHTOOL_LINK_MODE_40000baseKR4_Full_BIT |
- ETHTOOL_LINK_MODE_40000baseCR4_Full_BIT |
- ETHTOOL_LINK_MODE_40000baseSR4_Full_BIT |
- ETHTOOL_LINK_MODE_40000baseLR4_Full_BIT))
+ 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 & (ETHTOOL_LINK_MODE_56000baseKR4_Full_BIT |
- ETHTOOL_LINK_MODE_56000baseCR4_Full_BIT |
- ETHTOOL_LINK_MODE_56000baseSR4_Full_BIT |
- ETHTOOL_LINK_MODE_56000baseLR4_Full_BIT))
+ 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 & (ETHTOOL_LINK_MODE_25000baseCR_Full_BIT |
- ETHTOOL_LINK_MODE_25000baseKR_Full_BIT |
- ETHTOOL_LINK_MODE_25000baseSR_Full_BIT))
+ 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 & (ETHTOOL_LINK_MODE_50000baseCR2_Full_BIT |
- ETHTOOL_LINK_MODE_50000baseKR2_Full_BIT))
+ 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 & (ETHTOOL_LINK_MODE_100000baseKR4_Full_BIT |
- ETHTOOL_LINK_MODE_100000baseSR4_Full_BIT |
- ETHTOOL_LINK_MODE_100000baseCR4_Full_BIT |
- ETHTOOL_LINK_MODE_100000baseLR4_ER4_Full_BIT))
+ 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.edata.duplex == DUPLEX_HALF) ?
+ 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);
}
/**
- * DPDK callback to retrieve physical link information.
+ * Enable receiving and transmitting traffic.
*
- * @param dev
- * Pointer to Ethernet device structure.
+ * @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
-mlx5_link_update(struct rte_eth_dev *dev, int wait_to_complete)
+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))
- return mlx5_link_update_unlocked_gset(dev, wait_to_complete);
- return mlx5_link_update_unlocked_gs(dev, wait_to_complete);
+ 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;
}
/**
- * DPDK callback to change the MTU.
+ * 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.
*
- * 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 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 __rte_unused)
+{
+ struct priv *priv = dev->data->dev_private;
+ int ret;
+
+ ret = priv_link_update(priv, wait_to_complete);
+ return ret;
+}
+
+/**
+ * DPDK callback to change the MTU.
*
* @param dev
* Pointer to Ethernet device structure.
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;
- unsigned int max_frame_len;
- int rehash;
- int restart = priv->started;
-
- if (mlx5_is_secondary())
- return -E_RTE_SECONDARY;
- 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 = errno;
- WARN("cannot set port %u MTU to %u: %s", priv->port, mtu,
- strerror(ret));
+ ret = priv_set_mtu(priv, mtu);
+ if (ret)
+ goto out;
+ ret = priv_get_mtu(priv, &kern_mtu);
+ if (ret)
goto out;
- } else
+ if (kern_mtu == mtu) {
+ priv->mtu = mtu;
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 dev->rx_pkt_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;
- }
- /*
- * 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);
}
-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);
- 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);
- /* Provide new values to rxq_setup(). */
- dev->data->dev_conf.rxmode.jumbo_frame =
- (max_frame_len > ETHER_MAX_LEN);
- 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;
- }
- /* Double fault, disable RX. */
- break;
- }
- /* Mimic mlx5_dev_start(). */
- if (ret) {
- ERROR("unable to reconfigure RX queues, RX disabled");
- } 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);
- }
- priv->mtu = mtu;
- /* Burst functions can now be called again. */
- rte_wmb();
- /*
- * Use a safe RX burst function in case of error, otherwise select RX
- * burst function again.
- */
- if (!ret)
- priv_select_rx_function(priv);
+ return 0;
out:
- priv_unlock(priv);
+ ret = errno;
+ WARN("cannot set port %u MTU to %u: %s", priv->port, mtu,
+ strerror(ret));
assert(ret >= 0);
return -ret;
}
};
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)"
strerror(ret));
goto out;
}
-
fc_conf->autoneg = ethpause.autoneg;
if (ethpause.rx_pause && ethpause.tx_pause)
fc_conf->mode = RTE_FC_FULL;
else
fc_conf->mode = RTE_FC_NONE;
ret = 0;
-
out:
- priv_unlock(priv);
assert(ret >= 0);
return -ret;
}
};
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)"
goto out;
}
ret = 0;
-
out:
- priv_unlock(priv);
assert(ret >= 0);
return -ret;
}
}
/**
- * Link status handler.
+ * Update the link status.
*
* @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.
+ * Zero if the callback process can be called immediately.
*/
static int
-priv_dev_link_status_handler(struct priv *priv, struct rte_eth_dev *dev)
+priv_link_status_update(struct priv *priv)
{
- struct ibv_async_event event;
- struct rte_eth_link *link = &dev->data->dev_link;
- int ret = 0;
-
- /* Read all message and acknowledge them. */
- for (;;) {
- if (ibv_get_async_event(priv->ctx, &event))
- break;
+ struct rte_eth_link *link = &priv->dev->data->dev_link;
- if (event.event_type != IBV_EVENT_PORT_ACTIVE &&
- event.event_type != IBV_EVENT_PORT_ERR)
- DEBUG("event type %d on port %d not handled",
- event.event_type, event.element.port_num);
- ibv_ack_async_event(&event);
- }
- mlx5_link_update(dev, 0);
+ priv_link_update(priv, 0);
if (((link->link_speed == 0) && link->link_status) ||
- ((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) {
- /* Inconsistent status, check again later. */
priv->pending_alarm = 1;
rte_eal_alarm_set(MLX5_ALARM_TIMEOUT_US,
mlx5_dev_link_status_handler,
- dev);
+ priv->dev);
}
- } else {
- ret = 1;
+ 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;
}
struct priv *priv = dev->data->dev_private;
int ret;
- priv_lock(priv);
- assert(priv->pending_alarm == 1);
priv->pending_alarm = 0;
- 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,
- NULL);
+ ret = priv_link_status_update(priv);
+ if (!ret)
+ _rte_eth_dev_callback_process(dev, RTE_ETH_EVENT_INTR_LSC, NULL);
}
/**
{
struct rte_eth_dev *dev = cb_arg;
struct priv *priv = dev->data->dev_private;
- int ret;
+ uint32_t events;
- 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,
- NULL);
+ 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);
}
/**
void
priv_dev_interrupt_handler_uninstall(struct priv *priv, struct rte_eth_dev *dev)
{
- if (!dev->data->dev_conf.intr_conf.lsc)
- return;
- rte_intr_callback_unregister(&priv->intr_handle,
- mlx5_dev_interrupt_handler,
- dev);
- if (priv->pending_alarm)
+ 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->pending_alarm = 0;
+ }
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;
}
/**
{
int rc, flags;
- if (!dev->data->dev_conf.intr_conf.lsc)
- 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");
dev->data->dev_conf.intr_conf.lsc = 0;
- } else {
+ 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);
+ 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 Ethernet device structure.
+ * Pointer to private data structure.
* @param up
* Nonzero for link up, otherwise link down.
*
* 0 on success, errno value on failure.
*/
static int
-priv_set_link(struct priv *priv, int up)
+priv_dev_set_link(struct priv *priv, int up)
{
- 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;
- }
- return 0;
+ return priv_set_flags(priv, ~IFF_UP, up ? IFF_UP : ~IFF_UP);
}
/**
struct priv *priv = dev->data->dev_private;
int err;
- priv_lock(priv);
- err = priv_set_link(priv, 0);
- priv_unlock(priv);
+ err = priv_dev_set_link(priv, 0);
return err;
}
struct priv *priv = dev->data->dev_private;
int err;
- priv_lock(priv);
- err = priv_set_link(priv, 1);
- priv_unlock(priv);
+ err = priv_dev_set_link(priv, 1);
return err;
}
/**
- * Configure secondary process queues from a private data pointer (primary
- * or secondary) and update burst callbacks. Can take place only once.
- *
- * All queues must have been previously created by the primary process to
- * avoid undefined behavior.
+ * Configure the TX function to use.
*
* @param priv
- * Private data pointer from either primary or secondary process.
+ * Pointer to private data structure.
+ * @param dev
+ * Pointer to rte_eth_dev structure.
*
* @return
- * Private data pointer from secondary process, NULL in case of error.
+ * Pointer to selected Tx burst function.
*/
-struct priv *
-mlx5_secondary_data_setup(struct priv *priv)
+eth_tx_burst_t
+priv_select_tx_function(struct priv *priv, struct rte_eth_dev *dev)
{
- 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;
+ 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);
- /* 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;
- }
- /* 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;
- }
- /* 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);
-
- if (primary_rxq == NULL)
- continue;
- /* Not supported yet. */
- rx_queues[i] = 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.
- *
- * @param priv
- * Pointer to private structure.
- */
-void
-priv_select_tx_function(struct priv *priv)
-{
- priv->dev->tx_pkt_burst = mlx5_tx_burst;
/* Select appropriate TX function. */
- if (priv->mps == MLX5_MPW_ENHANCED) {
- if (priv_check_vec_tx_support(priv) > 0) {
- if (priv_check_raw_vec_tx_support(priv) > 0)
- priv->dev->tx_pkt_burst = mlx5_tx_burst_raw_vec;
+ 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
- priv->dev->tx_pkt_burst = mlx5_tx_burst_vec;
+ tx_pkt_burst = mlx5_tx_burst_vec;
DEBUG("selected Enhanced MPW TX vectorized function");
} else {
- priv->dev->tx_pkt_burst = mlx5_tx_burst_empw;
+ tx_pkt_burst = mlx5_tx_burst_empw;
DEBUG("selected Enhanced MPW TX function");
}
- } else if (priv->mps && priv->txq_inline) {
- priv->dev->tx_pkt_burst = mlx5_tx_burst_mpw_inline;
+ } else if (config->mps && (config->txq_inline > 0)) {
+ tx_pkt_burst = mlx5_tx_burst_mpw_inline;
DEBUG("selected MPW inline TX function");
- } else if (priv->mps) {
- priv->dev->tx_pkt_burst = mlx5_tx_burst_mpw;
+ } 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 structure.
+ * Pointer to private data structure.
+ * @param dev
+ * Pointer to rte_eth_dev structure.
+ *
+ * @return
+ * Pointer to selected Rx burst function.
*/
-void
-priv_select_rx_function(struct priv *priv)
+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) {
- priv_prep_vec_rx_function(priv);
- priv->dev->rx_pkt_burst = mlx5_rx_burst_vec;
+ rx_pkt_burst = mlx5_rx_burst_vec;
DEBUG("selected RX vectorized function");
- } else {
- priv->dev->rx_pkt_burst = mlx5_rx_burst;
}
+ 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;
}