* [dpdk-dev] [PATCH v1] net/mlx5: fix flow director rule deletion crash
@ 2018-04-26 16:17 Adrien Mazarguil
2018-04-27 6:55 ` Nélio Laranjeiro
0 siblings, 1 reply; 3+ messages in thread
From: Adrien Mazarguil @ 2018-04-26 16:17 UTC (permalink / raw)
To: Nelio Laranjeiro; +Cc: Shahaf Shuler, dev, stable
Flow director rules matching traffic properties above layer 2 do not
target a fixed hash Rx queue (HASH_RXQ_ETH), it actually depends on the
highest protocol layer specified by each flow rule.
mlx5_fdir_filter_delete() makes this wrong assumption and causes a crash
when attempting to destroy flow rules with L3/L4 specifications.
Fixes: 4c3e9bcdd52e ("net/mlx5: support flow director")
Cc: Nelio Laranjeiro <nelio.laranjeiro@6wind.com>
Cc: stable@dpdk.org
Signed-off-by: Adrien Mazarguil <adrien.mazarguil@6wind.com>
---
drivers/net/mlx5/mlx5_flow.c | 10 +++++-----
1 file changed, 5 insertions(+), 5 deletions(-)
diff --git a/drivers/net/mlx5/mlx5_flow.c b/drivers/net/mlx5/mlx5_flow.c
index 8f5fcf4d6..05def2b14 100644
--- a/drivers/net/mlx5/mlx5_flow.c
+++ b/drivers/net/mlx5/mlx5_flow.c
@@ -3409,13 +3409,13 @@ mlx5_fdir_filter_delete(struct rte_eth_dev *dev,
if (parser.drop) {
struct ibv_flow_spec_action_drop *drop;
- drop = (void *)((uintptr_t)parser.queue[HASH_RXQ_ETH].ibv_attr +
- parser.queue[HASH_RXQ_ETH].offset);
+ drop = (void *)((uintptr_t)parser.queue[parser.layer].ibv_attr +
+ parser.queue[parser.layer].offset);
*drop = (struct ibv_flow_spec_action_drop){
.type = IBV_FLOW_SPEC_ACTION_DROP,
.size = sizeof(struct ibv_flow_spec_action_drop),
};
- parser.queue[HASH_RXQ_ETH].ibv_attr->num_of_specs++;
+ parser.queue[parser.layer].ibv_attr->num_of_specs++;
}
TAILQ_FOREACH(flow, &priv->flows, next) {
struct ibv_flow_attr *attr;
@@ -3426,8 +3426,8 @@ mlx5_fdir_filter_delete(struct rte_eth_dev *dev,
void *flow_spec;
unsigned int specs_n;
- attr = parser.queue[HASH_RXQ_ETH].ibv_attr;
- flow_attr = flow->frxq[HASH_RXQ_ETH].ibv_attr;
+ attr = parser.queue[parser.layer].ibv_attr;
+ flow_attr = flow->frxq[parser.layer].ibv_attr;
/* Compare first the attributes. */
if (memcmp(attr, flow_attr, sizeof(struct ibv_flow_attr)))
continue;
--
2.11.0
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [dpdk-dev] [PATCH v1] net/mlx5: fix flow director rule deletion crash
2018-04-26 16:17 [dpdk-dev] [PATCH v1] net/mlx5: fix flow director rule deletion crash Adrien Mazarguil
@ 2018-04-27 6:55 ` Nélio Laranjeiro
2018-04-29 5:51 ` Shahaf Shuler
0 siblings, 1 reply; 3+ messages in thread
From: Nélio Laranjeiro @ 2018-04-27 6:55 UTC (permalink / raw)
To: Adrien Mazarguil; +Cc: Shahaf Shuler, dev, stable
On Thu, Apr 26, 2018 at 06:17:46PM +0200, Adrien Mazarguil wrote:
> Flow director rules matching traffic properties above layer 2 do not
> target a fixed hash Rx queue (HASH_RXQ_ETH), it actually depends on the
> highest protocol layer specified by each flow rule.
>
> mlx5_fdir_filter_delete() makes this wrong assumption and causes a crash
> when attempting to destroy flow rules with L3/L4 specifications.
>
> Fixes: 4c3e9bcdd52e ("net/mlx5: support flow director")
> Cc: Nelio Laranjeiro <nelio.laranjeiro@6wind.com>
> Cc: stable@dpdk.org
>
> Signed-off-by: Adrien Mazarguil <adrien.mazarguil@6wind.com>
Acked-by: Nelio Laranjeiro <nelio.laranjeiro@6wind.com>
> ---
> drivers/net/mlx5/mlx5_flow.c | 10 +++++-----
> 1 file changed, 5 insertions(+), 5 deletions(-)
>
> diff --git a/drivers/net/mlx5/mlx5_flow.c b/drivers/net/mlx5/mlx5_flow.c
> index 8f5fcf4d6..05def2b14 100644
> --- a/drivers/net/mlx5/mlx5_flow.c
> +++ b/drivers/net/mlx5/mlx5_flow.c
> @@ -3409,13 +3409,13 @@ mlx5_fdir_filter_delete(struct rte_eth_dev *dev,
> if (parser.drop) {
> struct ibv_flow_spec_action_drop *drop;
>
> - drop = (void *)((uintptr_t)parser.queue[HASH_RXQ_ETH].ibv_attr +
> - parser.queue[HASH_RXQ_ETH].offset);
> + drop = (void *)((uintptr_t)parser.queue[parser.layer].ibv_attr +
> + parser.queue[parser.layer].offset);
> *drop = (struct ibv_flow_spec_action_drop){
> .type = IBV_FLOW_SPEC_ACTION_DROP,
> .size = sizeof(struct ibv_flow_spec_action_drop),
> };
> - parser.queue[HASH_RXQ_ETH].ibv_attr->num_of_specs++;
> + parser.queue[parser.layer].ibv_attr->num_of_specs++;
> }
> TAILQ_FOREACH(flow, &priv->flows, next) {
> struct ibv_flow_attr *attr;
> @@ -3426,8 +3426,8 @@ mlx5_fdir_filter_delete(struct rte_eth_dev *dev,
> void *flow_spec;
> unsigned int specs_n;
>
> - attr = parser.queue[HASH_RXQ_ETH].ibv_attr;
> - flow_attr = flow->frxq[HASH_RXQ_ETH].ibv_attr;
> + attr = parser.queue[parser.layer].ibv_attr;
> + flow_attr = flow->frxq[parser.layer].ibv_attr;
> /* Compare first the attributes. */
> if (memcmp(attr, flow_attr, sizeof(struct ibv_flow_attr)))
> continue;
> --
> 2.11.0
--
Nélio Laranjeiro
6WIND
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [dpdk-dev] [PATCH v1] net/mlx5: fix flow director rule deletion crash
2018-04-27 6:55 ` Nélio Laranjeiro
@ 2018-04-29 5:51 ` Shahaf Shuler
0 siblings, 0 replies; 3+ messages in thread
From: Shahaf Shuler @ 2018-04-29 5:51 UTC (permalink / raw)
To: Nélio Laranjeiro, Adrien Mazarguil; +Cc: dev, stable
Friday, April 27, 2018 9:55 AM, Nélio Laranjeiro:
> Subject: Re: [PATCH v1] net/mlx5: fix flow director rule deletion crash
>
> On Thu, Apr 26, 2018 at 06:17:46PM +0200, Adrien Mazarguil wrote:
> > Flow director rules matching traffic properties above layer 2 do not
> > target a fixed hash Rx queue (HASH_RXQ_ETH), it actually depends on
> > the highest protocol layer specified by each flow rule.
> >
> > mlx5_fdir_filter_delete() makes this wrong assumption and causes a
> > crash when attempting to destroy flow rules with L3/L4 specifications.
> >
> > Fixes: 4c3e9bcdd52e ("net/mlx5: support flow director")
> > Cc: Nelio Laranjeiro <nelio.laranjeiro@6wind.com>
> > Cc: stable@dpdk.org
> >
> > Signed-off-by: Adrien Mazarguil <adrien.mazarguil@6wind.com>
>
> Acked-by: Nelio Laranjeiro <nelio.laranjeiro@6wind.com>
>
> > ---
> > drivers/net/mlx5/mlx5_flow.c | 10 +++++-----
> > 1 file changed, 5 insertions(+), 5 deletions(-)
> >
> > diff --git a/drivers/net/mlx5/mlx5_flow.c
> > b/drivers/net/mlx5/mlx5_flow.c index 8f5fcf4d6..05def2b14 100644
> > --- a/drivers/net/mlx5/mlx5_flow.c
> > +++ b/drivers/net/mlx5/mlx5_flow.c
> > @@ -3409,13 +3409,13 @@ mlx5_fdir_filter_delete(struct rte_eth_dev
> *dev,
> > if (parser.drop) {
> > struct ibv_flow_spec_action_drop *drop;
> >
> > - drop = (void
> *)((uintptr_t)parser.queue[HASH_RXQ_ETH].ibv_attr +
> > - parser.queue[HASH_RXQ_ETH].offset);
> > + drop = (void
> *)((uintptr_t)parser.queue[parser.layer].ibv_attr +
> > + parser.queue[parser.layer].offset);
> > *drop = (struct ibv_flow_spec_action_drop){
> > .type = IBV_FLOW_SPEC_ACTION_DROP,
> > .size = sizeof(struct ibv_flow_spec_action_drop),
> > };
> > - parser.queue[HASH_RXQ_ETH].ibv_attr->num_of_specs++;
> > + parser.queue[parser.layer].ibv_attr->num_of_specs++;
> > }
> > TAILQ_FOREACH(flow, &priv->flows, next) {
> > struct ibv_flow_attr *attr;
> > @@ -3426,8 +3426,8 @@ mlx5_fdir_filter_delete(struct rte_eth_dev *dev,
> > void *flow_spec;
> > unsigned int specs_n;
> >
> > - attr = parser.queue[HASH_RXQ_ETH].ibv_attr;
> > - flow_attr = flow->frxq[HASH_RXQ_ETH].ibv_attr;
> > + attr = parser.queue[parser.layer].ibv_attr;
> > + flow_attr = flow->frxq[parser.layer].ibv_attr;
> > /* Compare first the attributes. */
> > if (memcmp(attr, flow_attr, sizeof(struct ibv_flow_attr)))
> > continue;
> > --
> > 2.11.0
Applied to next-net-mlx, thanks.
>
> --
> Nélio Laranjeiro
> 6WIND
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2018-04-29 5:51 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2018-04-26 16:17 [dpdk-dev] [PATCH v1] net/mlx5: fix flow director rule deletion crash Adrien Mazarguil
2018-04-27 6:55 ` Nélio Laranjeiro
2018-04-29 5:51 ` Shahaf Shuler
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).