From: shahafs <shahafs@mellanox.com>
To: yuanhan.liu@linux.intel.com, stable@dpdk.org
Cc: Shahaf Shuler <shahafs@mellanox.com>
Subject: [dpdk-stable] [PATCH] net/mlx5: fix inconsistent link status
Date: Tue, 21 Feb 2017 09:23:29 +0200 [thread overview]
Message-ID: <1487661809-33942-1-git-send-email-shahafs@mellanox.com> (raw)
From: Shahaf Shuler <shahafs@mellanox.com>
[ backported from upstream commit a9f2fbc42f0c58df614896a407fd694899315233 ]
Querying the link status can end up being in an inconsistent state,
like the port is reporting speed although it is down.
For this case another query is scheduled.
A race condition can occur between the scheduled query and link
status interrupt handlers.
When the scheduled query by-pass interrupt handlers, the link status
will be stuck in an inconsistent state.
This patch addresses the race condition by not blocking link status
queries in case delayed query is used.
Fixes: 198a3c339a8f ("mlx5: handle link status interrupts")
Cc: stable@dpdk.org
Signed-off-by: Shahaf Shuler <shahafs@mellanox.com>
Acked-by: Nelio Laranjeiro <nelio.laranjeiro@6wind.com>
---
drivers/net/mlx5/mlx5_ethdev.c | 26 +++++++++++---------------
1 file changed, 11 insertions(+), 15 deletions(-)
diff --git a/drivers/net/mlx5/mlx5_ethdev.c b/drivers/net/mlx5/mlx5_ethdev.c
index a709a3a..06cfd01 100644
--- a/drivers/net/mlx5/mlx5_ethdev.c
+++ b/drivers/net/mlx5/mlx5_ethdev.c
@@ -1183,7 +1183,7 @@ struct priv *
priv_dev_link_status_handler(struct priv *priv, struct rte_eth_dev *dev)
{
struct ibv_async_event event;
- int port_change = 0;
+ struct rte_eth_link *link = &dev->data->dev_link;
int ret = 0;
/* Read all message and acknowledge them. */
@@ -1191,29 +1191,24 @@ struct priv *
if (ibv_get_async_event(priv->ctx, &event))
break;
- if (event.event_type == IBV_EVENT_PORT_ACTIVE ||
- event.event_type == IBV_EVENT_PORT_ERR)
- port_change = 1;
- else
+ 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);
}
-
- 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)) {
+ mlx5_link_update(dev, 0);
+ if (((link->link_speed == 0) && link->link_status) ||
+ ((link->link_speed != 0) && !link->link_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);
- } else
- ret = 1;
+ }
+ } else {
+ ret = 1;
}
return ret;
}
@@ -1233,6 +1228,7 @@ struct priv *
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)
--
1.8.3.1
next reply other threads:[~2017-02-21 7:23 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-02-21 7:23 shahafs [this message]
2017-02-21 7:31 ` Yuanhan Liu
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1487661809-33942-1-git-send-email-shahafs@mellanox.com \
--to=shahafs@mellanox.com \
--cc=stable@dpdk.org \
--cc=yuanhan.liu@linux.intel.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).