* [PATCH 20.11 1/3] common/mlx5: add Netlink event helpers
2022-03-11 7:49 [PATCH 20.11 0/3] net/mlx5: fix link state detection Dmitry Kozlyuk
@ 2022-03-11 7:49 ` Dmitry Kozlyuk
2022-03-11 7:49 ` [PATCH 20.11 2/3] net/mlx5: fix link status change detection Dmitry Kozlyuk
` (2 subsequent siblings)
3 siblings, 0 replies; 5+ messages in thread
From: Dmitry Kozlyuk @ 2022-03-11 7:49 UTC (permalink / raw)
To: stable; +Cc: Luca Boccassi
[ upstream commit be66461cba371c3138ce942eb9fe5657f9e9a446 ]
Introduce mlx5_nl_read_events() to read Netlink events
(technically, messages) from a socket that was configured
to listen for them via a new mlx5_nl_init() parameter.
Add mlx5_nl_parse_link_status_update() helper
to extract information from link-related events.
This patch is a shared base for later fixes.
Signed-off-by: Dmitry Kozlyuk <dkozlyuk@nvidia.com>
---
drivers/common/mlx5/linux/mlx5_nl.c | 105 +++++++++++++++++++++++++-
drivers/common/mlx5/linux/mlx5_nl.h | 8 +-
drivers/common/mlx5/version.map | 3 +
drivers/net/mlx5/linux/mlx5_os.c | 8 +-
drivers/net/mlx5/linux/mlx5_vlan_os.c | 2 +-
drivers/vdpa/mlx5/mlx5_vdpa.c | 2 +-
6 files changed, 120 insertions(+), 8 deletions(-)
diff --git a/drivers/common/mlx5/linux/mlx5_nl.c b/drivers/common/mlx5/linux/mlx5_nl.c
index 145e354b2c..15a5f7b303 100644
--- a/drivers/common/mlx5/linux/mlx5_nl.c
+++ b/drivers/common/mlx5/linux/mlx5_nl.c
@@ -180,19 +180,22 @@ uint32_t atomic_sn;
*
* @param protocol
* Netlink protocol (e.g. NETLINK_ROUTE, NETLINK_RDMA).
+ * @param groups
+ * Groups to listen (e.g. RTMGRP_LINK), can be 0.
*
* @return
* A file descriptor on success, a negative errno value otherwise and
* rte_errno is set.
*/
int
-mlx5_nl_init(int protocol)
+mlx5_nl_init(int protocol, int groups)
{
int fd;
int buf_size;
socklen_t opt_size;
struct sockaddr_nl local = {
.nl_family = AF_NETLINK,
+ .nl_groups = groups,
};
int ret;
@@ -1798,3 +1801,103 @@ mlx5_nl_enable_roce_set(int nlsk_fd, int family_id, const char *pci_addr,
/* Now, need to reload the driver. */
return mlx5_nl_driver_reload(nlsk_fd, family_id, pci_addr);
}
+
+/**
+ * Try to parse a Netlink message as a link status update.
+ *
+ * @param hdr
+ * Netlink message header.
+ * @param[out] ifindex
+ * Index of the updated interface.
+ * @param[out] flags
+ * New interface flags.
+ *
+ * @return
+ * 0 on success, negative on failure.
+ */
+int
+mlx5_nl_parse_link_status_update(struct nlmsghdr *hdr, uint32_t *ifindex)
+{
+ struct ifinfomsg *info;
+
+ switch (hdr->nlmsg_type) {
+ case RTM_NEWLINK:
+ case RTM_DELLINK:
+ case RTM_GETLINK:
+ case RTM_SETLINK:
+ info = NLMSG_DATA(hdr);
+ *ifindex = info->ifi_index;
+ return 0;
+ }
+ return -1;
+}
+
+/**
+ * Read pending events from a Netlink socket.
+ *
+ * @param nlsk_fd
+ * Netlink socket.
+ * @param cb
+ * Callback invoked for each of the events.
+ * @param cb_arg
+ * User data for the callback.
+ *
+ * @return
+ * 0 on success, including the case when there are no events.
+ * Negative on failure and rte_errno is set.
+ */
+int
+mlx5_nl_read_events(int nlsk_fd, mlx5_nl_event_cb *cb, void *cb_arg)
+{
+ char buf[8192];
+ struct sockaddr_nl addr;
+ struct iovec iov = {
+ .iov_base = buf,
+ .iov_len = sizeof(buf),
+ };
+ struct msghdr msg = {
+ .msg_name = &addr,
+ .msg_namelen = sizeof(addr),
+ .msg_iov = &iov,
+ .msg_iovlen = 1,
+ };
+ struct nlmsghdr *hdr;
+ ssize_t size;
+
+ while (1) {
+ size = recvmsg(nlsk_fd, &msg, MSG_DONTWAIT);
+ if (size < 0) {
+ if (errno == EAGAIN)
+ return 0;
+ if (errno == EINTR)
+ continue;
+ DRV_LOG(DEBUG, "Failed to receive netlink message: %s",
+ strerror(errno));
+ rte_errno = errno;
+ return -rte_errno;
+
+ }
+ hdr = (struct nlmsghdr *)buf;
+ while (size >= (ssize_t)sizeof(*hdr)) {
+ ssize_t msg_len = hdr->nlmsg_len;
+ ssize_t data_len = msg_len - sizeof(*hdr);
+ ssize_t aligned_len;
+
+ if (data_len < 0) {
+ DRV_LOG(DEBUG, "Netlink message too short");
+ rte_errno = EINVAL;
+ return -rte_errno;
+ }
+ aligned_len = NLMSG_ALIGN(msg_len);
+ if (aligned_len > size) {
+ DRV_LOG(DEBUG, "Netlink message too long");
+ rte_errno = EINVAL;
+ return -rte_errno;
+ }
+ cb(hdr, cb_arg);
+ hdr = RTE_PTR_ADD(hdr, aligned_len);
+ size -= aligned_len;
+ }
+ }
+ return 0;
+}
diff --git a/drivers/common/mlx5/linux/mlx5_nl.h b/drivers/common/mlx5/linux/mlx5_nl.h
index 15129ffdc8..05189e997a 100644
--- a/drivers/common/mlx5/linux/mlx5_nl.h
+++ b/drivers/common/mlx5/linux/mlx5_nl.h
@@ -11,6 +11,7 @@
#include "mlx5_common.h"
+typedef void (mlx5_nl_event_cb)(struct nlmsghdr *hdr, void *user_data);
/* VLAN netdev for VLAN workaround. */
struct mlx5_nl_vlan_dev {
@@ -30,7 +31,7 @@ struct mlx5_nl_vlan_vmwa_context {
};
__rte_internal
-int mlx5_nl_init(int protocol);
+int mlx5_nl_init(int protocol, int groups);
__rte_internal
int mlx5_nl_mac_addr_add(int nlsk_fd, unsigned int iface_idx, uint64_t *mac_own,
struct rte_ether_addr *mac, uint32_t index);
@@ -77,4 +78,9 @@ __rte_internal
int mlx5_nl_enable_roce_set(int nlsk_fd, int family_id, const char *pci_addr,
int enable);
+__rte_internal
+int mlx5_nl_read_events(int nlsk_fd, mlx5_nl_event_cb *cb, void *cb_arg);
+__rte_internal
+int mlx5_nl_parse_link_status_update(struct nlmsghdr *hdr, uint32_t *ifindex);
+
#endif /* RTE_PMD_MLX5_NL_H_ */
diff --git a/drivers/common/mlx5/version.map b/drivers/common/mlx5/version.map
index 983714a82c..6c9956c8f1 100644
--- a/drivers/common/mlx5/version.map
+++ b/drivers/common/mlx5/version.map
@@ -103,4 +103,7 @@ INTERNAL {
mlx5_free;
mlx5_pci_driver_register;
+
+ mlx5_nl_parse_link_status_update;
+ mlx5_nl_read_events;
};
diff --git a/drivers/net/mlx5/linux/mlx5_os.c b/drivers/net/mlx5/linux/mlx5_os.c
index aa66879d74..b3a5a8d3c7 100644
--- a/drivers/net/mlx5/linux/mlx5_os.c
+++ b/drivers/net/mlx5/linux/mlx5_os.c
@@ -1002,8 +1002,8 @@ mlx5_dev_spawn(struct rte_device *dpdk_dev,
priv->pci_dev = spawn->pci_dev;
priv->mtu = RTE_ETHER_MTU;
/* Some internal functions rely on Netlink sockets, open them now. */
- priv->nl_socket_rdma = mlx5_nl_init(NETLINK_RDMA);
- priv->nl_socket_route = mlx5_nl_init(NETLINK_ROUTE);
+ priv->nl_socket_rdma = mlx5_nl_init(NETLINK_RDMA, 0);
+ priv->nl_socket_route = mlx5_nl_init(NETLINK_ROUTE, 0);
priv->representor = !!switch_info->representor;
priv->master = !!switch_info->master;
priv->domain_id = RTE_ETH_DEV_SWITCH_DOMAIN_ID_INVALID;
@@ -1877,8 +1877,8 @@ mlx5_os_pci_probe(struct rte_pci_driver *pci_drv __rte_unused,
* matching ones, gathering into the list.
*/
struct ibv_device *ibv_match[ret + 1];
- int nl_route = mlx5_nl_init(NETLINK_ROUTE);
- int nl_rdma = mlx5_nl_init(NETLINK_RDMA);
+ int nl_route = mlx5_nl_init(NETLINK_ROUTE, 0);
+ int nl_rdma = mlx5_nl_init(NETLINK_RDMA, 0);
unsigned int i;
while (ret-- > 0) {
diff --git a/drivers/net/mlx5/linux/mlx5_vlan_os.c b/drivers/net/mlx5/linux/mlx5_vlan_os.c
index 40e895e080..598026c414 100644
--- a/drivers/net/mlx5/linux/mlx5_vlan_os.c
+++ b/drivers/net/mlx5/linux/mlx5_vlan_os.c
@@ -136,7 +136,7 @@ mlx5_vlan_vmwa_init(struct rte_eth_dev *dev, uint32_t ifindex)
return NULL;
}
rte_spinlock_init(&vmwa->sl);
- vmwa->nl_socket = mlx5_nl_init(NETLINK_ROUTE);
+ vmwa->nl_socket = mlx5_nl_init(NETLINK_ROUTE, 0);
if (vmwa->nl_socket < 0) {
DRV_LOG(WARNING,
"Can not create Netlink socket"
diff --git a/drivers/vdpa/mlx5/mlx5_vdpa.c b/drivers/vdpa/mlx5/mlx5_vdpa.c
index 06e602e72a..6519b9c9ac 100644
--- a/drivers/vdpa/mlx5/mlx5_vdpa.c
+++ b/drivers/vdpa/mlx5/mlx5_vdpa.c
@@ -504,7 +504,7 @@ mlx5_vdpa_get_ib_device_match(struct rte_pci_addr *addr)
static int
mlx5_vdpa_nl_roce_disable(const char *addr)
{
- int nlsk_fd = mlx5_nl_init(NETLINK_GENERIC);
+ int nlsk_fd = mlx5_nl_init(NETLINK_GENERIC, 0);
int devlink_id;
int enable;
int ret;
--
2.25.1
^ permalink raw reply [flat|nested] 5+ messages in thread* [PATCH 20.11 2/3] net/mlx5: fix link status change detection
2022-03-11 7:49 [PATCH 20.11 0/3] net/mlx5: fix link state detection Dmitry Kozlyuk
2022-03-11 7:49 ` [PATCH 20.11 1/3] common/mlx5: add Netlink event helpers Dmitry Kozlyuk
@ 2022-03-11 7:49 ` Dmitry Kozlyuk
2022-03-11 7:49 ` [PATCH 20.11 3/3] net/mlx5: fix initial link status detection Dmitry Kozlyuk
2022-03-11 12:19 ` [PATCH 20.11 0/3] net/mlx5: fix link state detection Luca Boccassi
3 siblings, 0 replies; 5+ messages in thread
From: Dmitry Kozlyuk @ 2022-03-11 7:49 UTC (permalink / raw)
To: stable; +Cc: Luca Boccassi
[ upstream commit 17f95513adca1f4ee5bba305b154ac984ee50cee ]
Sometimes net/mlx5 devices did not detect link status change to "up".
Each shared device was monitoring IBV_EVENT_PORT_{ACTIVE,ERR}
and queried the link status upon receiving the event.
IBV_EVENT_PORT_ACTIVE is delivered when the logical link status
(UP flag) is set, but the physical link status (RUNNING flag)
may be down at that time, in which case the new link status
would be errornously considered down.
IBV interface is insufficient for the task.
Monitor interface events using Netlink.
Fixes: 198a3c339a8f ("mlx5: handle link status interrupts")
Signed-off-by: Dmitry Kozlyuk <dkozlyuk@nvidia.com>
---
drivers/net/mlx5/linux/mlx5_ethdev_os.c | 64 ++++++++++++++++++++-----
drivers/net/mlx5/linux/mlx5_os.c | 48 +++++++++++++++++++
drivers/net/mlx5/mlx5.c | 1 +
drivers/net/mlx5/mlx5.h | 3 ++
drivers/net/mlx5/mlx5_trigger.c | 12 ++++-
5 files changed, 114 insertions(+), 14 deletions(-)
diff --git a/drivers/net/mlx5/linux/mlx5_ethdev_os.c b/drivers/net/mlx5/linux/mlx5_ethdev_os.c
index 08b51b7dc8..e89d16aace 100644
--- a/drivers/net/mlx5/linux/mlx5_ethdev_os.c
+++ b/drivers/net/mlx5/linux/mlx5_ethdev_os.c
@@ -38,6 +38,7 @@
#include <mlx5_devx_cmds.h>
#include <mlx5_common.h>
#include <mlx5_malloc.h>
+#include <mlx5_nl.h>
#include "mlx5.h"
#include "mlx5_rxtx.h"
@@ -730,6 +731,57 @@ mlx5_dev_interrupt_device_fatal(struct mlx5_dev_ctx_shared *sh)
}
}
+static void
+mlx5_dev_interrupt_nl_cb(struct nlmsghdr *hdr, void *cb_arg)
+{
+ struct mlx5_dev_ctx_shared *sh = cb_arg;
+ uint32_t i;
+ uint32_t if_index;
+
+ if (mlx5_nl_parse_link_status_update(hdr, &if_index) < 0)
+ return;
+ for (i = 0; i < sh->max_port; i++) {
+ struct mlx5_dev_shared_port *port = &sh->port[i];
+ struct rte_eth_dev *dev;
+ struct mlx5_priv *priv;
+ bool configured;
+
+ if (port->nl_ih_port_id >= RTE_MAX_ETHPORTS)
+ continue;
+ dev = &rte_eth_devices[port->nl_ih_port_id];
+ configured = dev->process_private != NULL;
+ /* Probing may initiate an LSC before configuration is done. */
+ if (configured && !dev->data->dev_conf.intr_conf.lsc)
+ break;
+ priv = dev->data->dev_private;
+ if (priv->if_index == if_index) {
+ /* Block logical LSC events. */
+ uint16_t prev_status = dev->data->dev_link.link_status;
+
+ if (mlx5_link_update(dev, 0) < 0)
+ DRV_LOG(ERR, "Failed to update link status: %s",
+ rte_strerror(rte_errno));
+ else if (prev_status != dev->data->dev_link.link_status)
+ rte_eth_dev_callback_process
+ (dev, RTE_ETH_EVENT_INTR_LSC, NULL);
+ break;
+ }
+ }
+}
+
+void
+mlx5_dev_interrupt_handler_nl(void *arg)
+{
+ struct mlx5_dev_ctx_shared *sh = arg;
+ int nlsk_fd = sh->intr_handle_nl.fd;
+
+ if (nlsk_fd < 0)
+ return;
+ if (mlx5_nl_read_events(nlsk_fd, mlx5_dev_interrupt_nl_cb, sh) < 0)
+ DRV_LOG(ERR, "Failed to process Netlink events: %s",
+ rte_strerror(rte_errno));
+}
+
/**
* Handle shared asynchronous events the NIC (removal event
* and link status change). Supports multiport IB device.
@@ -793,18 +845,6 @@ mlx5_dev_interrupt_handler(void *cb_arg)
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);
diff --git a/drivers/net/mlx5/linux/mlx5_os.c b/drivers/net/mlx5/linux/mlx5_os.c
index b3a5a8d3c7..0e738d4a59 100644
--- a/drivers/net/mlx5/linux/mlx5_os.c
+++ b/drivers/net/mlx5/linux/mlx5_os.c
@@ -2438,6 +2438,40 @@ mlx5_os_net_cleanup(void)
mlx5_pmd_socket_uninit();
}
+static int
+mlx5_os_dev_shared_handler_install_lsc(struct mlx5_dev_ctx_shared *sh)
+{
+ int nlsk_fd, flags, ret;
+
+ nlsk_fd = mlx5_nl_init(NETLINK_ROUTE, RTMGRP_LINK);
+ if (nlsk_fd < 0) {
+ DRV_LOG(ERR, "Failed to create a socket for Netlink events: %s",
+ rte_strerror(rte_errno));
+ return -1;
+ }
+ flags = fcntl(nlsk_fd, F_GETFL);
+ ret = fcntl(nlsk_fd, F_SETFL, flags | O_NONBLOCK);
+ if (ret != 0) {
+ DRV_LOG(ERR, "Failed to make Netlink event socket non-blocking: %s",
+ strerror(errno));
+ rte_errno = errno;
+ goto error;
+ }
+ sh->intr_handle_nl.type = RTE_INTR_HANDLE_EXT;
+ sh->intr_handle_nl.fd = nlsk_fd;
+ if (rte_intr_callback_register(&sh->intr_handle_nl,
+ mlx5_dev_interrupt_handler_nl,
+ sh) != 0) {
+ DRV_LOG(ERR, "Failed to register Netlink events interrupt");
+ sh->intr_handle_nl.fd = -1;
+ goto error;
+ }
+ return 0;
+error:
+ close(nlsk_fd);
+ return -1;
+}
+
/**
* Install shared asynchronous device events handler.
* This function is implemented to support event sharing
@@ -2468,6 +2502,11 @@ mlx5_os_dev_shared_handler_install(struct mlx5_dev_ctx_shared *sh)
sh->intr_handle.fd = -1;
}
}
+ sh->intr_handle_nl.fd = -1;
+ if (mlx5_os_dev_shared_handler_install_lsc(sh) < 0) {
+ DRV_LOG(INFO, "Fail to install the shared Netlink event handler.");
+ sh->intr_handle_nl.fd = -1;
+ }
if (sh->devx) {
#ifdef HAVE_IBV_DEVX_ASYNC
sh->intr_handle_devx.fd = -1;
@@ -2508,9 +2547,18 @@ mlx5_os_dev_shared_handler_install(struct mlx5_dev_ctx_shared *sh)
void
mlx5_os_dev_shared_handler_uninstall(struct mlx5_dev_ctx_shared *sh)
{
+ int nlsk_fd;
+
if (sh->intr_handle.fd >= 0)
mlx5_intr_callback_unregister(&sh->intr_handle,
mlx5_dev_interrupt_handler, sh);
+ nlsk_fd = sh->intr_handle_nl.fd;
+ if (nlsk_fd >= 0) {
+ mlx5_intr_callback_unregister(&sh->intr_handle_nl,
+ mlx5_dev_interrupt_handler_nl,
+ sh);
+ close(nlsk_fd);
+ }
#ifdef HAVE_IBV_DEVX_ASYNC
if (sh->intr_handle_devx.fd >= 0)
rte_intr_callback_unregister(&sh->intr_handle_devx,
diff --git a/drivers/net/mlx5/mlx5.c b/drivers/net/mlx5/mlx5.c
index cff1188213..0b82969b4d 100644
--- a/drivers/net/mlx5/mlx5.c
+++ b/drivers/net/mlx5/mlx5.c
@@ -928,6 +928,7 @@ mlx5_alloc_shared_dev_ctx(const struct mlx5_dev_spawn_data *spawn,
for (i = 0; i < sh->max_port; i++) {
sh->port[i].ih_port_id = RTE_MAX_ETHPORTS;
sh->port[i].devx_ih_port_id = RTE_MAX_ETHPORTS;
+ sh->port[i].nl_ih_port_id = RTE_MAX_ETHPORTS;
}
sh->pd = mlx5_glue->alloc_pd(sh->ctx);
if (sh->pd == NULL) {
diff --git a/drivers/net/mlx5/mlx5.h b/drivers/net/mlx5/mlx5.h
index b18fffea52..6c85c4e2db 100644
--- a/drivers/net/mlx5/mlx5.h
+++ b/drivers/net/mlx5/mlx5.h
@@ -559,6 +559,7 @@ struct mlx5_age_info {
struct mlx5_dev_shared_port {
uint32_t ih_port_id;
uint32_t devx_ih_port_id;
+ uint32_t nl_ih_port_id;
/*
* Interrupt handler port_id. Used by shared interrupt
* handler to find the corresponding rte_eth device
@@ -741,6 +742,7 @@ struct mlx5_dev_ctx_shared {
/* Shared interrupt handler section. */
struct rte_intr_handle intr_handle; /* Interrupt handler for device. */
struct rte_intr_handle intr_handle_devx; /* DEVX interrupt handler. */
+ struct rte_intr_handle intr_handle_nl; /* Netlink interrupt handler. */
void *devx_comp; /* DEVX async comp obj. */
struct mlx5_devx_obj *tis; /* TIS object. */
struct mlx5_devx_obj *td; /* Transport domain. */
@@ -1085,6 +1087,7 @@ int mlx5_dev_set_flow_ctrl(struct rte_eth_dev *dev,
struct rte_eth_fc_conf *fc_conf);
void mlx5_dev_interrupt_handler(void *arg);
void mlx5_dev_interrupt_handler_devx(void *arg);
+void mlx5_dev_interrupt_handler_nl(void *arg);
int mlx5_set_link_down(struct rte_eth_dev *dev);
int mlx5_set_link_up(struct rte_eth_dev *dev);
int mlx5_is_removed(struct rte_eth_dev *dev);
diff --git a/drivers/net/mlx5/mlx5_trigger.c b/drivers/net/mlx5/mlx5_trigger.c
index 92eaef87d1..6273122e40 100644
--- a/drivers/net/mlx5/mlx5_trigger.c
+++ b/drivers/net/mlx5/mlx5_trigger.c
@@ -1134,11 +1134,18 @@ mlx5_dev_start(struct rte_eth_dev *dev)
priv->sh->port[priv->dev_port - 1].ih_port_id =
(uint32_t)dev->data->port_id;
} else {
- DRV_LOG(INFO, "port %u starts without LSC and RMV interrupts.",
+ DRV_LOG(INFO, "port %u starts without RMV interrupts.",
dev->data->port_id);
- dev->data->dev_conf.intr_conf.lsc = 0;
dev->data->dev_conf.intr_conf.rmv = 0;
}
+ if (priv->sh->intr_handle_nl.fd >= 0) {
+ priv->sh->port[priv->dev_port - 1].nl_ih_port_id =
+ (uint32_t)dev->data->port_id;
+ } else {
+ DRV_LOG(INFO, "port %u starts without LSC interrupts.",
+ dev->data->port_id);
+ dev->data->dev_conf.intr_conf.lsc = 0;
+ }
if (priv->sh->intr_handle_devx.fd >= 0)
priv->sh->port[priv->dev_port - 1].devx_ih_port_id =
(uint32_t)dev->data->port_id;
@@ -1188,6 +1195,7 @@ mlx5_dev_stop(struct rte_eth_dev *dev)
mlx5_rx_intr_vec_disable(dev);
priv->sh->port[priv->dev_port - 1].ih_port_id = RTE_MAX_ETHPORTS;
priv->sh->port[priv->dev_port - 1].devx_ih_port_id = RTE_MAX_ETHPORTS;
+ priv->sh->port[priv->dev_port - 1].nl_ih_port_id = RTE_MAX_ETHPORTS;
mlx5_txq_stop(dev);
mlx5_rxq_stop(dev);
if (priv->obj_ops.lb_dummy_queue_release)
--
2.25.1
^ permalink raw reply [flat|nested] 5+ messages in thread* [PATCH 20.11 3/3] net/mlx5: fix initial link status detection
2022-03-11 7:49 [PATCH 20.11 0/3] net/mlx5: fix link state detection Dmitry Kozlyuk
2022-03-11 7:49 ` [PATCH 20.11 1/3] common/mlx5: add Netlink event helpers Dmitry Kozlyuk
2022-03-11 7:49 ` [PATCH 20.11 2/3] net/mlx5: fix link status change detection Dmitry Kozlyuk
@ 2022-03-11 7:49 ` Dmitry Kozlyuk
2022-03-11 12:19 ` [PATCH 20.11 0/3] net/mlx5: fix link state detection Luca Boccassi
3 siblings, 0 replies; 5+ messages in thread
From: Dmitry Kozlyuk @ 2022-03-11 7:49 UTC (permalink / raw)
To: stable; +Cc: Luca Boccassi
[ upstream commit 655c3c26c11e7a6f1b5421afe40b06d580b8f2dd ]
Link status change takes time that depends on the HW and the kernel.
It was checked immediately after the change was issued at probing.
If the port had beed down before probing, a "down" state may be read,
while the port would be "up" imminently.
After that, DPDK reported the port as "down" mistakenly
and "ifconfig $DEV up" did not trigger an LSC event,
because from the system's perspective the port was "up" already.
Install Netlink event handler at port probe before requesting the port
to come up in order to receive LSC event even if it comes up
between probe and start.
Fixes: b6499434b83e ("net/mlx5: fix link status initialization")
Signed-off-by: Dmitry Kozlyuk <dkozlyuk@nvidia.com>
---
drivers/net/mlx5/linux/mlx5_os.c | 10 ++++------
1 file changed, 4 insertions(+), 6 deletions(-)
diff --git a/drivers/net/mlx5/linux/mlx5_os.c b/drivers/net/mlx5/linux/mlx5_os.c
index 0e738d4a59..f98cf90dd8 100644
--- a/drivers/net/mlx5/linux/mlx5_os.c
+++ b/drivers/net/mlx5/linux/mlx5_os.c
@@ -1519,13 +1519,11 @@ mlx5_dev_spawn(struct rte_device *dpdk_dev,
/* Bring Ethernet device up. */
DRV_LOG(DEBUG, "port %u forcing Ethernet interface up",
eth_dev->data->port_id);
- mlx5_set_link_up(eth_dev);
- /*
- * Even though the interrupt handler is not installed yet,
- * interrupts will still trigger on the async_fd from
- * Verbs context returned by ibv_open_device().
- */
mlx5_link_update(eth_dev, 0);
+ /* Watch LSC interrupts between port probe and port start. */
+ priv->sh->port[priv->dev_port - 1].nl_ih_port_id =
+ eth_dev->data->port_id;
+ mlx5_set_link_up(eth_dev);
/* Detect minimal data bytes to inline. */
mlx5_set_min_inline(spawn, config);
/* Store device configuration on private structure. */
--
2.25.1
^ permalink raw reply [flat|nested] 5+ messages in thread* Re: [PATCH 20.11 0/3] net/mlx5: fix link state detection
2022-03-11 7:49 [PATCH 20.11 0/3] net/mlx5: fix link state detection Dmitry Kozlyuk
` (2 preceding siblings ...)
2022-03-11 7:49 ` [PATCH 20.11 3/3] net/mlx5: fix initial link status detection Dmitry Kozlyuk
@ 2022-03-11 12:19 ` Luca Boccassi
3 siblings, 0 replies; 5+ messages in thread
From: Luca Boccassi @ 2022-03-11 12:19 UTC (permalink / raw)
To: Dmitry Kozlyuk, stable
[-- Attachment #1: Type: text/plain, Size: 1291 bytes --]
On Fri, 2022-03-11 at 09:49 +0200, Dmitry Kozlyuk wrote:
> [ backport of https://inbox.dpdk.org/dev/20220301121514.41497-1-dkozlyuk@nvidia.com ]
>
> This patchset fixes two related issues:
> * In rare occasions with any HW link state change to UP was missed.
> * If a port was DOWN before startup, its netdev would come UP,
> but appear DOWN in DPDK (especially probable with ConnectX-4).
>
> Dmitry Kozlyuk (3):
> common/mlx5: add Netlink event helpers
> net/mlx5: fix link status change detection
> net/mlx5: fix initial link status detection
>
> drivers/common/mlx5/linux/mlx5_nl.c | 105 +++++++++++++++++++++++-
> drivers/common/mlx5/linux/mlx5_nl.h | 8 +-
> drivers/common/mlx5/version.map | 3 +
> drivers/net/mlx5/linux/mlx5_ethdev_os.c | 64 ++++++++++++---
> drivers/net/mlx5/linux/mlx5_os.c | 66 ++++++++++++---
> drivers/net/mlx5/linux/mlx5_vlan_os.c | 2 +-
> drivers/net/mlx5/mlx5.c | 1 +
> drivers/net/mlx5/mlx5.h | 3 +
> drivers/net/mlx5/mlx5_trigger.c | 12 ++-
> drivers/vdpa/mlx5/mlx5_vdpa.c | 2 +-
> 10 files changed, 238 insertions(+), 28 deletions(-)
Thanks, applied and pushed.
--
Kind regards,
Luca Boccassi
[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 833 bytes --]
^ permalink raw reply [flat|nested] 5+ messages in thread