* [dpdk-stable] [PATCH v2] net/mlx5: fix link status is always inconsist
[not found] <485348339-43905-1-git-send-email-shahafs@mellanox.com>
@ 2017-01-29 8:22 ` Shahaf Shuler
2017-01-31 13:13 ` [dpdk-stable] [PATCH v3] " Shahaf Shuler
2017-01-29 8:28 ` [dpdk-stable] [PATCH v2] " Shahaf Shuler
1 sibling, 1 reply; 5+ messages in thread
From: Shahaf Shuler @ 2017-01-29 8:22 UTC (permalink / raw)
To: adrien.mazarguil, nelio.laranjeiro; +Cc: dev, stable
Query the link status can end up in an inconsist state where the port is
down but it is reporting speed. For that another query is scheduled for a
later time.
A race condition is possible between the scheduled call and other link
status interrupt handlers. When the scheduled query by-pass those handlers,
the link status will be stuck in an in-consist state.
This patch addresses the race condition by not blocking link status
queries in case delayed query is on the flight.
Fixes: 198a3c339a8f ("mlx5: handle link status interrupts")
CC: stable@dpdk.org
Signed-off-by: Shahaf Shuler <shahafs@mellanox.com>
---
on v2:
* change the commit log to better explain the race
* instead of blocking other interrupt handlers allow everyone to query the link status
---
drivers/net/mlx5/mlx5_ethdev.c | 35 ++++++++++++++---------------------
1 file changed, 14 insertions(+), 21 deletions(-)
diff --git a/drivers/net/mlx5/mlx5_ethdev.c b/drivers/net/mlx5/mlx5_ethdev.c
index e77238f..6219d64 100644
--- a/drivers/net/mlx5/mlx5_ethdev.c
+++ b/drivers/net/mlx5/mlx5_ethdev.c
@@ -1128,7 +1128,7 @@ 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. */
@@ -1136,30 +1136,22 @@ 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(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;
- }
+ mlx5_link_update(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;
return ret;
}
@@ -1178,6 +1170,7 @@ 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
^ permalink raw reply [flat|nested] 5+ messages in thread
* [dpdk-stable] [PATCH v3] net/mlx5: fix link status is always inconsist
2017-01-29 8:22 ` [dpdk-stable] [PATCH v2] net/mlx5: fix link status is always inconsist Shahaf Shuler
@ 2017-01-31 13:13 ` Shahaf Shuler
2017-01-31 15:40 ` Nélio Laranjeiro
0 siblings, 1 reply; 5+ messages in thread
From: Shahaf Shuler @ 2017-01-31 13:13 UTC (permalink / raw)
To: adrien.mazarguil, nelio.laranjeiro; +Cc: dev, stable
Query the link status can end up in an inconsist state where the port is
down but it is reporting speed. For that another query is scheduled for a
later time.
A race condition is possible between the scheduled call and other link
status interrupt handlers. When the scheduled query by-pass those handlers,
the link status will be stuck in an in-consist state.
This patch addresses the race condition by not blocking link status
queries in case delayed query is on the flight.
Fixes: 198a3c339a8f ("mlx5: handle link status interrupts")
CC: stable@dpdk.org
Signed-off-by: Shahaf Shuler <shahafs@mellanox.com>
---
on v3:
* allow only single alarm on the flight.
on v2:
* change the commit log to better explain the race
* instead of blocking other interrupt handlers allow everyone to query the link status
---
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 e77238f..fcbe273 100644
--- a/drivers/net/mlx5/mlx5_ethdev.c
+++ b/drivers/net/mlx5/mlx5_ethdev.c
@@ -1128,7 +1128,7 @@ 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. */
@@ -1136,29 +1136,24 @@ 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(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;
}
@@ -1178,6 +1173,7 @@ 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
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: [dpdk-stable] [PATCH v3] net/mlx5: fix link status is always inconsist
2017-01-31 13:13 ` [dpdk-stable] [PATCH v3] " Shahaf Shuler
@ 2017-01-31 15:40 ` Nélio Laranjeiro
2017-02-01 10:57 ` [dpdk-stable] [dpdk-dev] " Ferruh Yigit
0 siblings, 1 reply; 5+ messages in thread
From: Nélio Laranjeiro @ 2017-01-31 15:40 UTC (permalink / raw)
To: Shahaf Shuler; +Cc: dev, stable
On Tue, Jan 31, 2017 at 03:13:38PM +0200, Shahaf Shuler wrote:
> Query the link status can end up in an inconsist state where the port is
> down but it is reporting speed. For that another query is scheduled for a
> later time.
> A race condition is possible between the scheduled call and other link
> status interrupt handlers. When the scheduled query by-pass those handlers,
> the link status will be stuck in an in-consist state.
> This patch addresses the race condition by not blocking link status
> queries in case delayed query is on the flight.
>
> Fixes: 198a3c339a8f ("mlx5: handle link status interrupts")
> CC: stable@dpdk.org
>
> Signed-off-by: Shahaf Shuler <shahafs@mellanox.com>
> ---
> on v3:
> * allow only single alarm on the flight.
>
> on v2:
> * change the commit log to better explain the race
> * instead of blocking other interrupt handlers allow everyone to query the link status
> ---
> 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 e77238f..fcbe273 100644
> --- a/drivers/net/mlx5/mlx5_ethdev.c
> +++ b/drivers/net/mlx5/mlx5_ethdev.c
> @@ -1128,7 +1128,7 @@ 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. */
> @@ -1136,29 +1136,24 @@ 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(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;
> }
> @@ -1178,6 +1173,7 @@ 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
Acked-by: Nelio Laranjeiro <nelio.laranjeiro@6wind.com>
--
Nélio Laranjeiro
6WIND
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: [dpdk-stable] [dpdk-dev] [PATCH v3] net/mlx5: fix link status is always inconsist
2017-01-31 15:40 ` Nélio Laranjeiro
@ 2017-02-01 10:57 ` Ferruh Yigit
0 siblings, 0 replies; 5+ messages in thread
From: Ferruh Yigit @ 2017-02-01 10:57 UTC (permalink / raw)
To: Nélio Laranjeiro, Shahaf Shuler; +Cc: dev, stable
On 1/31/2017 3:40 PM, Nélio Laranjeiro wrote:
> On Tue, Jan 31, 2017 at 03:13:38PM +0200, Shahaf Shuler wrote:
>> Query the link status can end up in an inconsist state where the port is
>> down but it is reporting speed. For that another query is scheduled for a
>> later time.
>> A race condition is possible between the scheduled call and other link
>> status interrupt handlers. When the scheduled query by-pass those handlers,
>> the link status will be stuck in an in-consist state.
>> This patch addresses the race condition by not blocking link status
>> queries in case delayed query is on the flight.
>>
>> 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>
Applied to dpdk-next-net/master, thanks.
^ permalink raw reply [flat|nested] 5+ messages in thread
* [dpdk-stable] [PATCH v2] net/mlx5: fix link status is always inconsist
[not found] <485348339-43905-1-git-send-email-shahafs@mellanox.com>
2017-01-29 8:22 ` [dpdk-stable] [PATCH v2] net/mlx5: fix link status is always inconsist Shahaf Shuler
@ 2017-01-29 8:28 ` Shahaf Shuler
1 sibling, 0 replies; 5+ messages in thread
From: Shahaf Shuler @ 2017-01-29 8:28 UTC (permalink / raw)
To: adrien.mazarguil, nelio.laranjeiro; +Cc: dev, stable
Query the link status can end up in an inconsist state where the port is
down but it is reporting speed. For that another query is scheduled for a
later time.
A race condition is possible between the scheduled call and other link
status interrupt handlers. When the scheduled query by-pass those handlers,
the link status will be stuck in an in-consist state.
This patch addresses the race condition by not blocking link status
queries in case delayed query is on the flight.
Fixes: 198a3c339a8f ("mlx5: handle link status interrupts")
CC: stable@dpdk.org
Signed-off-by: Shahaf Shuler <shahafs@mellanox.com>
---
on v2:
* change the commit log to better explain the race
* instead of blocking other interrupt handlers allow everyone to query the link status
---
drivers/net/mlx5/mlx5_ethdev.c | 35 ++++++++++++++---------------------
1 file changed, 14 insertions(+), 21 deletions(-)
diff --git a/drivers/net/mlx5/mlx5_ethdev.c b/drivers/net/mlx5/mlx5_ethdev.c
index e77238f..6219d64 100644
--- a/drivers/net/mlx5/mlx5_ethdev.c
+++ b/drivers/net/mlx5/mlx5_ethdev.c
@@ -1128,7 +1128,7 @@ 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. */
@@ -1136,30 +1136,22 @@ 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(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;
- }
+ mlx5_link_update(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;
return ret;
}
@@ -1178,6 +1170,7 @@ 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
^ permalink raw reply [flat|nested] 5+ messages in thread