This patch complements [1], adding to MLX5 PMD the option to set
flow rule for egress traffic.
[1] "net/mlx5: support metadata as flow rule criteria"
http://mails.dpdk.org/archives/dev/2018-September/113275.html
Signed-off-by: Dekel Peled <dek...@mellanox.com>
---
drivers/net/mlx5/mlx5_flow.c | 54 ++++++++++++++++++++++++++++++++++++++
drivers/net/mlx5/mlx5_flow.h | 6 +++++
drivers/net/mlx5/mlx5_flow_dv.c | 24 ++++++++++-------
drivers/net/mlx5/mlx5_flow_verbs.c | 7 ++++-
4 files changed, 80 insertions(+), 11 deletions(-)
diff --git a/drivers/net/mlx5/mlx5_flow.c b/drivers/net/mlx5/mlx5_flow.c
index 9581691..79a06df 100644
--- a/drivers/net/mlx5/mlx5_flow.c
+++ b/drivers/net/mlx5/mlx5_flow.c
@@ -644,6 +644,8 @@ uint32_t mlx5_flow_adjust_priority(struct rte_eth_dev *dev,
int32_t priority,
*
* @param[in] action_flags
* Bit-fields that holds the actions detected until now.
+ * @param[in] attr
+ * Attributes of flow that includes this action.
* @param[out] error
* Pointer to error structure.
*
@@ -652,6 +654,7 @@ uint32_t mlx5_flow_adjust_priority(struct rte_eth_dev *dev,
int32_t priority,
*/
int
mlx5_flow_validate_action_flag(uint64_t action_flags,
+ const struct rte_flow_attr *attr,
struct rte_flow_error *error)
{
@@ -668,6 +671,12 @@ uint32_t mlx5_flow_adjust_priority(struct rte_eth_dev
*dev, int32_t priority,
RTE_FLOW_ERROR_TYPE_ACTION, NULL,
"can't have 2 flag"
" actions in same flow");
+ if (attr->egress)
+ return rte_flow_error_set(error, ENOTSUP,
+ RTE_FLOW_ERROR_TYPE_ATTR_EGRESS,
+ NULL,
+ "flag action not supported for "
+ "egress");
return 0;
}
@@ -678,6 +687,8 @@ uint32_t mlx5_flow_adjust_priority(struct rte_eth_dev *dev,
int32_t priority,
* Pointer to the queue action.
* @param[in] action_flags
* Bit-fields that holds the actions detected until now.
+ * @param[in] attr
+ * Attributes of flow that includes this action.
* @param[out] error
* Pointer to error structure.
*
@@ -687,6 +698,7 @@ uint32_t mlx5_flow_adjust_priority(struct rte_eth_dev *dev,
int32_t priority,
int
mlx5_flow_validate_action_mark(const struct rte_flow_action *action,
uint64_t action_flags,
+ const struct rte_flow_attr *attr,
struct rte_flow_error *error)
{
const struct rte_flow_action_mark *mark = action->conf;
@@ -715,6 +727,12 @@ uint32_t mlx5_flow_adjust_priority(struct rte_eth_dev
*dev, int32_t priority,
RTE_FLOW_ERROR_TYPE_ACTION, NULL,
"can't have 2 flag actions in same"
" flow");
+ if (attr->egress)
+ return rte_flow_error_set(error, ENOTSUP,
+ RTE_FLOW_ERROR_TYPE_ATTR_EGRESS,
+ NULL,
+ "mark action not supported for "
+ "egress");
return 0;
}
@@ -723,6 +741,8 @@ uint32_t mlx5_flow_adjust_priority(struct rte_eth_dev *dev,
int32_t priority,
*
* @param[in] action_flags
* Bit-fields that holds the actions detected until now.
+ * @param[in] attr
+ * Attributes of flow that includes this action.
* @param[out] error
* Pointer to error structure.
*
@@ -731,6 +751,7 @@ uint32_t mlx5_flow_adjust_priority(struct rte_eth_dev *dev,
int32_t priority,
*/
int
mlx5_flow_validate_action_drop(uint64_t action_flags,
+ const struct rte_flow_attr *attr,
struct rte_flow_error *error)
{
if (action_flags & MLX5_ACTION_FLAG)
@@ -747,6 +768,12 @@ uint32_t mlx5_flow_adjust_priority(struct rte_eth_dev
*dev, int32_t priority,
RTE_FLOW_ERROR_TYPE_ACTION, NULL,
"can't have 2 fate actions in"
" same flow");
+ if (attr->egress)
+ return rte_flow_error_set(error, ENOTSUP,
+ RTE_FLOW_ERROR_TYPE_ATTR_EGRESS,
+ NULL,
+ "drop action not supported for "
+ "egress");
return 0;
}
@@ -759,6 +786,8 @@ uint32_t mlx5_flow_adjust_priority(struct rte_eth_dev *dev,
int32_t priority,
* Bit-fields that holds the actions detected until now.
* @param[in] dev
* Pointer to the Ethernet device structure.
+ * @param[in] attr
+ * Attributes of flow that includes this action.
* @param[out] error
* Pointer to error structure.
*
@@ -769,6 +798,7 @@ uint32_t mlx5_flow_adjust_priority(struct rte_eth_dev *dev,
int32_t priority,
mlx5_flow_validate_action_queue(const struct rte_flow_action *action,
uint64_t action_flags,
struct rte_eth_dev *dev,
+ const struct rte_flow_attr *attr,
struct rte_flow_error *error)
{
struct priv *priv = dev->data->dev_private;
@@ -790,6 +820,12 @@ uint32_t mlx5_flow_adjust_priority(struct rte_eth_dev
*dev, int32_t priority,
RTE_FLOW_ERROR_TYPE_ACTION_CONF,
&queue->index,
"queue is not configured");
+ if (attr->egress)
+ return rte_flow_error_set(error, ENOTSUP,
+ RTE_FLOW_ERROR_TYPE_ATTR_EGRESS,
+ NULL,
+ "queue action not supported for "
+ "egress");
return 0;
}
@@ -802,6 +838,8 @@ uint32_t mlx5_flow_adjust_priority(struct rte_eth_dev *dev,
int32_t priority,
* Bit-fields that holds the actions detected until now.
* @param[in] dev
* Pointer to the Ethernet device structure.
+ * @param[in] attr
+ * Attributes of flow that includes this action.
* @param[out] error
* Pointer to error structure.
*
@@ -812,6 +850,7 @@ uint32_t mlx5_flow_adjust_priority(struct rte_eth_dev *dev,
int32_t priority,
mlx5_flow_validate_action_rss(const struct rte_flow_action *action,
uint64_t action_flags,
struct rte_eth_dev *dev,
+ const struct rte_flow_attr *attr,
struct rte_flow_error *error)
{
struct priv *priv = dev->data->dev_private;
@@ -866,6 +905,12 @@ uint32_t mlx5_flow_adjust_priority(struct rte_eth_dev
*dev, int32_t priority,
(error, EINVAL, RTE_FLOW_ERROR_TYPE_ACTION_CONF,
&rss->queue[i], "queue is not configured");
}
+ if (attr->egress)
+ return rte_flow_error_set(error, ENOTSUP,
+ RTE_FLOW_ERROR_TYPE_ATTR_EGRESS,
+ NULL,
+ "rss action not supported for "
+ "egress");
return 0;
}
@@ -874,6 +919,8 @@ uint32_t mlx5_flow_adjust_priority(struct rte_eth_dev *dev,
int32_t priority,
*
* @param[in] dev
* Pointer to the Ethernet device structure.
+ * @param[in] attr
+ * Attributes of flow that includes this action.
* @param[out] error
* Pointer to error structure.
*
@@ -882,6 +929,7 @@ uint32_t mlx5_flow_adjust_priority(struct rte_eth_dev *dev,
int32_t priority,
*/
int
mlx5_flow_validate_action_count(struct rte_eth_dev *dev,
+ const struct rte_flow_attr *attr,
struct rte_flow_error *error)
{
struct priv *priv = dev->data->dev_private;
@@ -890,6 +938,12 @@ uint32_t mlx5_flow_adjust_priority(struct rte_eth_dev
*dev, int32_t priority,
return rte_flow_error_set(error, ENOTSUP,
RTE_FLOW_ERROR_TYPE_ACTION, NULL,
"flow counters are not supported.");
+ if (attr->egress)
+ return rte_flow_error_set(error, ENOTSUP,
+ RTE_FLOW_ERROR_TYPE_ATTR_EGRESS,
+ NULL,
+ "count action not supported for "
+ "egress");
return 0;
}
diff --git a/drivers/net/mlx5/mlx5_flow.h b/drivers/net/mlx5/mlx5_flow.h
index d91ae17..a353d07 100644
--- a/drivers/net/mlx5/mlx5_flow.h
+++ b/drivers/net/mlx5/mlx5_flow.h
@@ -283,21 +283,27 @@ uint64_t mlx5_flow_hashfields_adjust(struct mlx5_flow
*dev_flow, int tunnel,
uint32_t mlx5_flow_adjust_priority(struct rte_eth_dev *dev, int32_t priority,
uint32_t subpriority);
int mlx5_flow_validate_action_count(struct rte_eth_dev *dev,
+ const struct rte_flow_attr *attr,
struct rte_flow_error *error);
int mlx5_flow_validate_action_drop(uint64_t action_flags,
+ const struct rte_flow_attr *attr,
struct rte_flow_error *error);
int mlx5_flow_validate_action_flag(uint64_t action_flags,
+ const struct rte_flow_attr *attr,
struct rte_flow_error *error);
int mlx5_flow_validate_action_mark(const struct rte_flow_action *action,
uint64_t action_flags,
+ const struct rte_flow_attr *attr,
struct rte_flow_error *error);
int mlx5_flow_validate_action_queue(const struct rte_flow_action *action,
uint64_t action_flags,
struct rte_eth_dev *dev,
+ const struct rte_flow_attr *attr,
struct rte_flow_error *error);
int mlx5_flow_validate_action_rss(const struct rte_flow_action *action,
uint64_t action_flags,
struct rte_eth_dev *dev,
+ const struct rte_flow_attr *attr,
struct rte_flow_error *error);
int mlx5_flow_validate_attributes(struct rte_eth_dev *dev,
const struct rte_flow_attr *attributes,
diff --git a/drivers/net/mlx5/mlx5_flow_dv.c b/drivers/net/mlx5/mlx5_flow_dv.c
index 2439f5e..1f3fcb8 100644
--- a/drivers/net/mlx5/mlx5_flow_dv.c
+++ b/drivers/net/mlx5/mlx5_flow_dv.c
@@ -117,21 +117,17 @@
RTE_FLOW_ERROR_TYPE_ATTR_PRIORITY,
NULL,
"priority out of range");
- if (attributes->egress)
- return rte_flow_error_set(error, ENOTSUP,
- RTE_FLOW_ERROR_TYPE_ATTR_EGRESS,
- NULL,
- "egress is not supported");
if (attributes->transfer)
return rte_flow_error_set(error, ENOTSUP,
RTE_FLOW_ERROR_TYPE_ATTR_TRANSFER,
NULL,
"transfer is not supported");
- if (!attributes->ingress)
+ if (!(attributes->egress ^ attributes->ingress))
return rte_flow_error_set(error, ENOTSUP,
- RTE_FLOW_ERROR_TYPE_ATTR_INGRESS,
+ RTE_FLOW_ERROR_TYPE_ATTR,
NULL,
- "ingress attribute is mandatory");
+ "must specify exactly one of "
+ "ingress or egress");
return 0;
}
@@ -288,6 +284,7 @@
break;
case RTE_FLOW_ACTION_TYPE_FLAG:
ret = mlx5_flow_validate_action_flag(action_flags,
+ attr,
error);
if (ret < 0)
return ret;
@@ -297,6 +294,7 @@
case RTE_FLOW_ACTION_TYPE_MARK:
ret = mlx5_flow_validate_action_mark(actions,
action_flags,
+ attr,
error);
if (ret < 0)
return ret;
@@ -305,6 +303,7 @@
break;
case RTE_FLOW_ACTION_TYPE_DROP:
ret = mlx5_flow_validate_action_drop(action_flags,
+ attr,
error);
if (ret < 0)
return ret;
@@ -314,6 +313,7 @@
case RTE_FLOW_ACTION_TYPE_QUEUE:
ret = mlx5_flow_validate_action_queue(actions,
action_flags, dev,
+ attr,
error);
if (ret < 0)
return ret;
@@ -323,6 +323,7 @@
case RTE_FLOW_ACTION_TYPE_RSS:
ret = mlx5_flow_validate_action_rss(actions,
action_flags, dev,
+ attr,
error);
if (ret < 0)
return ret;
@@ -330,7 +331,7 @@
++actions_n;
break;
case RTE_FLOW_ACTION_TYPE_COUNT:
- ret = mlx5_flow_validate_action_count(dev, error);
+ ret = mlx5_flow_validate_action_count(dev, attr, error);
if (ret < 0)
return ret;
action_flags |= MLX5_ACTION_COUNT;
@@ -1080,6 +1081,7 @@
queue = action->conf;
flow->rss.queue_num = 1;
(*flow->queue)[0] = queue->index;
+ flow->actions |= MLX5_ACTION_QUEUE;
break;
case RTE_FLOW_ACTION_TYPE_RSS:
rss = action->conf;
@@ -1091,6 +1093,7 @@
flow->rss.types = rss->types;
flow->rss.level = rss->level;
/* Added to array only in apply since we need the QP */
+ flow->actions |= MLX5_ACTION_RSS;
break;
default:
break;
@@ -1301,7 +1304,8 @@
dv->actions[n].type = MLX5DV_FLOW_ACTION_DEST_IBV_QP;
dv->actions[n].qp = dv->hrxq->qp;
n++;
- } else {
+ } else if (flow->actions &
+ (MLX5_ACTION_QUEUE | MLX5_ACTION_RSS)) {
struct mlx5_hrxq *hrxq;
hrxq = mlx5_hrxq_get(dev, flow->key,
MLX5_RSS_HASH_KEY_LEN,
diff --git a/drivers/net/mlx5/mlx5_flow_verbs.c
b/drivers/net/mlx5/mlx5_flow_verbs.c
index 05ab5fd..b1aa704 100644
--- a/drivers/net/mlx5/mlx5_flow_verbs.c
+++ b/drivers/net/mlx5/mlx5_flow_verbs.c
@@ -1112,6 +1112,7 @@
break;
case RTE_FLOW_ACTION_TYPE_FLAG:
ret = mlx5_flow_validate_action_flag(action_flags,
+ attr,
error);
if (ret < 0)
return ret;
@@ -1120,6 +1121,7 @@
case RTE_FLOW_ACTION_TYPE_MARK:
ret = mlx5_flow_validate_action_mark(actions,
action_flags,
+ attr,
error);
if (ret < 0)
return ret;
@@ -1127,6 +1129,7 @@
break;
case RTE_FLOW_ACTION_TYPE_DROP:
ret = mlx5_flow_validate_action_drop(action_flags,
+ attr,
error);
if (ret < 0)
return ret;
@@ -1135,6 +1138,7 @@
case RTE_FLOW_ACTION_TYPE_QUEUE:
ret = mlx5_flow_validate_action_queue(actions,
action_flags, dev,
+ attr,
error);
if (ret < 0)
return ret;
@@ -1143,13 +1147,14 @@
case RTE_FLOW_ACTION_TYPE_RSS:
ret = mlx5_flow_validate_action_rss(actions,
action_flags, dev,
+ attr,
error);
if (ret < 0)
return ret;
action_flags |= MLX5_ACTION_RSS;
break;
case RTE_FLOW_ACTION_TYPE_COUNT:
- ret = mlx5_flow_validate_action_count(dev, error);
+ ret = mlx5_flow_validate_action_count(dev, attr, error);
if (ret < 0)
return ret;
action_flags |= MLX5_ACTION_COUNT;
--
1.8.3.1
