* [PATCH] net/mlx5: fix the E-Switch mirror flow rule validation
@ 2023-10-11 6:43 Jiawei Wang
2023-10-29 14:30 ` Ori Kam
2023-10-31 8:03 ` Raslan Darawsheh
0 siblings, 2 replies; 3+ messages in thread
From: Jiawei Wang @ 2023-10-11 6:43 UTC (permalink / raw)
To: suanmingm, viacheslavo; +Cc: dev, rasland, stable
The port action and jump to flow table action are not supported
in the mirror flows (RTE_FLOW_ACTION_TYPE_SAMPLE with sample ratio=1)
in E-Switch domain (transfer attribute set) without presented encap
action.
The encap action is supported for uplink port only. So, if flow with
mirroring contains encap action application should provide encap
and uplink port actions in the mirror action list and PMD validates
this condition (to make sure we cover the hardware limitation).
This patch adds the validation for E-Switch mirror flow rule checking
and rejects as invalid.
Fixes: 6a951567c159 ("net/mlx5: support E-Switch mirroring and jump in one flow")
Cc: stable@dpdk.org
Signed-off-by: Jiawei Wang <jiaweiw@nvidia.com>
---
doc/guides/nics/mlx5.rst | 5 ++
drivers/net/mlx5/mlx5_flow_dv.c | 93 +++++++++++++++++++++++++--------
2 files changed, 76 insertions(+), 22 deletions(-)
diff --git a/doc/guides/nics/mlx5.rst b/doc/guides/nics/mlx5.rst
index 7bee57d9dd..16c322578a 100644
--- a/doc/guides/nics/mlx5.rst
+++ b/doc/guides/nics/mlx5.rst
@@ -545,6 +545,11 @@ Limitations
sample actions list.
- For E-Switch mirroring flow, supports ``RAW ENCAP``, ``Port ID``,
``VXLAN ENCAP``, ``NVGRE ENCAP`` in the sample actions list.
+ - For E-Switch mirroring flow with sample ratio = 1, the ``ENCAP`` action
+ support for uplink port only.
+ - For E-Switch mirroring flow with sample ratio = 1, the ``PORT`` action and
+ ``JUMP`` action are not supported without presented ``ENCAP`` action in the
+ sample actions list.
- For ConnectX-5 trusted device, the application metadata with SET_TAG index 0
is not supported before ``RTE_FLOW_ACTION_TYPE_SAMPLE`` action.
diff --git a/drivers/net/mlx5/mlx5_flow_dv.c b/drivers/net/mlx5/mlx5_flow_dv.c
index 3dc2fe5c71..8c9a00ef85 100644
--- a/drivers/net/mlx5/mlx5_flow_dv.c
+++ b/drivers/net/mlx5/mlx5_flow_dv.c
@@ -6029,6 +6029,7 @@ flow_dv_modify_clone_free_cb(void *tool_ctx, struct mlx5_list_entry *entry)
*/
static int
flow_dv_validate_action_sample(uint64_t *action_flags,
+ uint64_t *sub_action_flags,
const struct rte_flow_action *action,
struct rte_eth_dev *dev,
const struct rte_flow_attr *attr,
@@ -6037,14 +6038,15 @@ flow_dv_validate_action_sample(uint64_t *action_flags,
const struct rte_flow_action_rss **sample_rss,
const struct rte_flow_action_count **count,
int *fdb_mirror,
+ uint16_t *sample_port_id,
bool root,
struct rte_flow_error *error)
{
struct mlx5_priv *priv = dev->data->dev_private;
struct mlx5_sh_config *dev_conf = &priv->sh->config;
const struct rte_flow_action_sample *sample = action->conf;
+ const struct rte_flow_action_port_id *port = NULL;
const struct rte_flow_action *act;
- uint64_t sub_action_flags = 0;
uint16_t queue_index = 0xFFFF;
int actions_n = 0;
int ret;
@@ -6091,20 +6093,20 @@ flow_dv_validate_action_sample(uint64_t *action_flags,
switch (act->type) {
case RTE_FLOW_ACTION_TYPE_QUEUE:
ret = mlx5_flow_validate_action_queue(act,
- sub_action_flags,
+ *sub_action_flags,
dev,
attr, error);
if (ret < 0)
return ret;
queue_index = ((const struct rte_flow_action_queue *)
(act->conf))->index;
- sub_action_flags |= MLX5_FLOW_ACTION_QUEUE;
+ *sub_action_flags |= MLX5_FLOW_ACTION_QUEUE;
++actions_n;
break;
case RTE_FLOW_ACTION_TYPE_RSS:
*sample_rss = act->conf;
ret = mlx5_flow_validate_action_rss(act,
- sub_action_flags,
+ *sub_action_flags,
dev, attr,
item_flags,
error);
@@ -6120,48 +6122,57 @@ flow_dv_validate_action_sample(uint64_t *action_flags,
"or level in the same flow");
if (*sample_rss != NULL && (*sample_rss)->queue_num)
queue_index = (*sample_rss)->queue[0];
- sub_action_flags |= MLX5_FLOW_ACTION_RSS;
+ *sub_action_flags |= MLX5_FLOW_ACTION_RSS;
++actions_n;
break;
case RTE_FLOW_ACTION_TYPE_MARK:
ret = flow_dv_validate_action_mark(dev, act,
- sub_action_flags,
+ *sub_action_flags,
attr, error);
if (ret < 0)
return ret;
if (dev_conf->dv_xmeta_en != MLX5_XMETA_MODE_LEGACY)
- sub_action_flags |= MLX5_FLOW_ACTION_MARK |
+ *sub_action_flags |= MLX5_FLOW_ACTION_MARK |
MLX5_FLOW_ACTION_MARK_EXT;
else
- sub_action_flags |= MLX5_FLOW_ACTION_MARK;
+ *sub_action_flags |= MLX5_FLOW_ACTION_MARK;
++actions_n;
break;
case RTE_FLOW_ACTION_TYPE_COUNT:
ret = flow_dv_validate_action_count
- (dev, false, *action_flags | sub_action_flags,
+ (dev, false, *action_flags | *sub_action_flags,
root, error);
if (ret < 0)
return ret;
*count = act->conf;
- sub_action_flags |= MLX5_FLOW_ACTION_COUNT;
+ *sub_action_flags |= MLX5_FLOW_ACTION_COUNT;
*action_flags |= MLX5_FLOW_ACTION_COUNT;
++actions_n;
break;
case RTE_FLOW_ACTION_TYPE_PORT_ID:
case RTE_FLOW_ACTION_TYPE_REPRESENTED_PORT:
ret = flow_dv_validate_action_port_id(dev,
- sub_action_flags,
+ *sub_action_flags,
act,
attr,
error);
if (ret)
return ret;
- sub_action_flags |= MLX5_FLOW_ACTION_PORT_ID;
+ if (act->type == RTE_FLOW_ACTION_TYPE_PORT_ID) {
+ port = (const struct rte_flow_action_port_id *)
+ act->conf;
+ *sample_port_id = port->original ?
+ dev->data->port_id : port->id;
+ } else {
+ *sample_port_id = ((const struct rte_flow_action_ethdev *)
+ act->conf)->port_id;
+ }
+ *sub_action_flags |= MLX5_FLOW_ACTION_PORT_ID;
++actions_n;
break;
case RTE_FLOW_ACTION_TYPE_RAW_ENCAP:
ret = flow_dv_validate_action_raw_encap_decap
- (dev, NULL, act->conf, attr, &sub_action_flags,
+ (dev, NULL, act->conf, attr, sub_action_flags,
&actions_n, action, item_flags, error);
if (ret < 0)
return ret;
@@ -6170,12 +6181,12 @@ flow_dv_validate_action_sample(uint64_t *action_flags,
case RTE_FLOW_ACTION_TYPE_VXLAN_ENCAP:
case RTE_FLOW_ACTION_TYPE_NVGRE_ENCAP:
ret = flow_dv_validate_action_l2_encap(dev,
- sub_action_flags,
+ *sub_action_flags,
act, attr,
error);
if (ret < 0)
return ret;
- sub_action_flags |= MLX5_FLOW_ACTION_ENCAP;
+ *sub_action_flags |= MLX5_FLOW_ACTION_ENCAP;
++actions_n;
break;
default:
@@ -6187,7 +6198,7 @@ flow_dv_validate_action_sample(uint64_t *action_flags,
}
}
if (attr->ingress) {
- if (!(sub_action_flags & (MLX5_FLOW_ACTION_QUEUE |
+ if (!(*sub_action_flags & (MLX5_FLOW_ACTION_QUEUE |
MLX5_FLOW_ACTION_RSS)))
return rte_flow_error_set(error, EINVAL,
RTE_FLOW_ERROR_TYPE_ACTION,
@@ -6209,17 +6220,17 @@ flow_dv_validate_action_sample(uint64_t *action_flags,
"E-Switch doesn't support "
"any optional action "
"for sampling");
- if (sub_action_flags & MLX5_FLOW_ACTION_QUEUE)
+ if (*sub_action_flags & MLX5_FLOW_ACTION_QUEUE)
return rte_flow_error_set(error, ENOTSUP,
RTE_FLOW_ERROR_TYPE_ACTION,
NULL,
"unsupported action QUEUE");
- if (sub_action_flags & MLX5_FLOW_ACTION_RSS)
+ if (*sub_action_flags & MLX5_FLOW_ACTION_RSS)
return rte_flow_error_set(error, ENOTSUP,
RTE_FLOW_ERROR_TYPE_ACTION,
NULL,
"unsupported action QUEUE");
- if (!(sub_action_flags & MLX5_FLOW_ACTION_PORT_ID))
+ if (!(*sub_action_flags & MLX5_FLOW_ACTION_PORT_ID))
return rte_flow_error_set(error, EINVAL,
RTE_FLOW_ERROR_TYPE_ACTION,
NULL,
@@ -6228,16 +6239,16 @@ flow_dv_validate_action_sample(uint64_t *action_flags,
*fdb_mirror = 1;
}
/* Continue validation for Xcap actions.*/
- if ((sub_action_flags & MLX5_FLOW_XCAP_ACTIONS) &&
+ if ((*sub_action_flags & MLX5_FLOW_XCAP_ACTIONS) &&
(queue_index == 0xFFFF || !mlx5_rxq_is_hairpin(dev, queue_index))) {
- if ((sub_action_flags & MLX5_FLOW_XCAP_ACTIONS) ==
+ if ((*sub_action_flags & MLX5_FLOW_XCAP_ACTIONS) ==
MLX5_FLOW_XCAP_ACTIONS)
return rte_flow_error_set(error, ENOTSUP,
RTE_FLOW_ERROR_TYPE_ACTION,
NULL, "encap and decap "
"combination aren't "
"supported");
- if (attr->ingress && (sub_action_flags & MLX5_FLOW_ACTION_ENCAP))
+ if (attr->ingress && (*sub_action_flags & MLX5_FLOW_ACTION_ENCAP))
return rte_flow_error_set(error, ENOTSUP,
RTE_FLOW_ERROR_TYPE_ACTION,
NULL, "encap is not supported"
@@ -7392,9 +7403,14 @@ flow_dv_validate(struct rte_eth_dev *dev, const struct rte_flow_attr *attr,
uint32_t tag_id = 0, tag_bitmap = 0;
const struct rte_flow_action_age *non_shared_age = NULL;
const struct rte_flow_action_count *count = NULL;
+ const struct rte_flow_action_port_id *port = NULL;
const struct mlx5_rte_flow_item_tag *mlx5_tag;
struct mlx5_priv *act_priv = NULL;
int aso_after_sample = 0;
+ struct mlx5_priv *port_priv = NULL;
+ uint64_t sub_action_flags = 0;
+ uint16_t sample_port_id = 0;
+ uint16_t port_id = 0;
if (items == NULL)
return -1;
@@ -7864,6 +7880,14 @@ flow_dv_validate(struct rte_eth_dev *dev, const struct rte_flow_attr *attr,
error);
if (ret)
return ret;
+ if (type == RTE_FLOW_ACTION_TYPE_PORT_ID) {
+ port = (const struct rte_flow_action_port_id *)
+ actions->conf;
+ port_id = port->original ? dev->data->port_id : port->id;
+ } else {
+ port_id = ((const struct rte_flow_action_ethdev *)
+ actions->conf)->port_id;
+ }
action_flags |= MLX5_FLOW_ACTION_PORT_ID;
++actions_n;
break;
@@ -8363,11 +8387,13 @@ flow_dv_validate(struct rte_eth_dev *dev, const struct rte_flow_attr *attr,
break;
case RTE_FLOW_ACTION_TYPE_SAMPLE:
ret = flow_dv_validate_action_sample(&action_flags,
+ &sub_action_flags,
actions, dev,
attr, item_flags,
rss, &sample_rss,
&sample_count,
&fdb_mirror,
+ &sample_port_id,
is_root,
error);
if (ret < 0)
@@ -8679,6 +8705,29 @@ flow_dv_validate(struct rte_eth_dev *dev, const struct rte_flow_attr *attr,
return rte_flow_error_set(error, ENOTSUP,
RTE_FLOW_ERROR_TYPE_ACTION, NULL,
"sample before ASO action is not supported");
+ if (sub_action_flags & MLX5_FLOW_ACTION_PORT_ID) {
+ port_priv = mlx5_port_to_eswitch_info(sample_port_id, false);
+ if (flow_source_vport_representor(priv, port_priv)) {
+ if (sub_action_flags & MLX5_FLOW_ACTION_ENCAP)
+ return rte_flow_error_set(error, ENOTSUP,
+ RTE_FLOW_ERROR_TYPE_ACTION, NULL,
+ "mirror to rep port with encap is not supported");
+ } else {
+ if ((sub_action_flags & ~MLX5_FLOW_ACTION_ENCAP) &&
+ (action_flags & MLX5_FLOW_ACTION_JUMP))
+ return rte_flow_error_set(error, ENOTSUP,
+ RTE_FLOW_ERROR_TYPE_ACTION, NULL,
+ "mirror to wire port without encap is not supported");
+ }
+ }
+ if ((action_flags & MLX5_FLOW_ACTION_PORT_ID) &&
+ (action_flags & MLX5_FLOW_ACTION_ENCAP)) {
+ port_priv = mlx5_port_to_eswitch_info(port_id, false);
+ if (flow_source_vport_representor(priv, port_priv))
+ return rte_flow_error_set(error, ENOTSUP,
+ RTE_FLOW_ERROR_TYPE_ACTION, NULL,
+ "mirror to rep port with encap is not supported");
+ }
}
/*
* Validation the NIC Egress flow on representor, except implicit
--
2.18.1
^ permalink raw reply [flat|nested] 3+ messages in thread
* RE: [PATCH] net/mlx5: fix the E-Switch mirror flow rule validation
2023-10-11 6:43 [PATCH] net/mlx5: fix the E-Switch mirror flow rule validation Jiawei Wang
@ 2023-10-29 14:30 ` Ori Kam
2023-10-31 8:03 ` Raslan Darawsheh
1 sibling, 0 replies; 3+ messages in thread
From: Ori Kam @ 2023-10-29 14:30 UTC (permalink / raw)
To: Jiawei(Jonny) Wang, Suanming Mou, Slava Ovsiienko
Cc: dev, Raslan Darawsheh, stable
Hi Jiawei,
> -----Original Message-----
> From: Jiawei Wang <jiaweiw@nvidia.com>
> Sent: Wednesday, October 11, 2023 9:43 AM
>
> The port action and jump to flow table action are not supported
> in the mirror flows (RTE_FLOW_ACTION_TYPE_SAMPLE with sample ratio=1)
> in E-Switch domain (transfer attribute set) without presented encap
> action.
>
> The encap action is supported for uplink port only. So, if flow with
> mirroring contains encap action application should provide encap
> and uplink port actions in the mirror action list and PMD validates
> this condition (to make sure we cover the hardware limitation).
>
> This patch adds the validation for E-Switch mirror flow rule checking
> and rejects as invalid.
>
> Fixes: 6a951567c159 ("net/mlx5: support E-Switch mirroring and jump in one
> flow")
> Cc: stable@dpdk.org
>
> Signed-off-by: Jiawei Wang <jiaweiw@nvidia.com>
> ---
Acked-by: Ori Kam <orika@nvidia.com>
Best,
Ori
^ permalink raw reply [flat|nested] 3+ messages in thread
* RE: [PATCH] net/mlx5: fix the E-Switch mirror flow rule validation
2023-10-11 6:43 [PATCH] net/mlx5: fix the E-Switch mirror flow rule validation Jiawei Wang
2023-10-29 14:30 ` Ori Kam
@ 2023-10-31 8:03 ` Raslan Darawsheh
1 sibling, 0 replies; 3+ messages in thread
From: Raslan Darawsheh @ 2023-10-31 8:03 UTC (permalink / raw)
To: Jiawei(Jonny) Wang, Suanming Mou, Slava Ovsiienko; +Cc: dev, stable
Hi,
> -----Original Message-----
> From: Jiawei(Jonny) Wang <jiaweiw@nvidia.com>
> Sent: Wednesday, October 11, 2023 9:43 AM
> To: Suanming Mou <suanmingm@nvidia.com>; Slava Ovsiienko
> <viacheslavo@nvidia.com>
> Cc: dev@dpdk.org; Raslan Darawsheh <rasland@nvidia.com>;
> stable@dpdk.org
> Subject: [PATCH] net/mlx5: fix the E-Switch mirror flow rule validation
>
> The port action and jump to flow table action are not supported in the mirror
> flows (RTE_FLOW_ACTION_TYPE_SAMPLE with sample ratio=1) in E-Switch
> domain (transfer attribute set) without presented encap action.
>
> The encap action is supported for uplink port only. So, if flow with mirroring
> contains encap action application should provide encap and uplink port
> actions in the mirror action list and PMD validates this condition (to make sure
> we cover the hardware limitation).
>
> This patch adds the validation for E-Switch mirror flow rule checking and
> rejects as invalid.
>
> Fixes: 6a951567c159 ("net/mlx5: support E-Switch mirroring and jump in one
> flow")
> Cc: stable@dpdk.org
>
> Signed-off-by: Jiawei Wang <jiaweiw@nvidia.com>
Patch applied to next-net-mlx,
Kindest regards,
Raslan Darawsheh
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2023-10-31 8:03 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2023-10-11 6:43 [PATCH] net/mlx5: fix the E-Switch mirror flow rule validation Jiawei Wang
2023-10-29 14:30 ` Ori Kam
2023-10-31 8:03 ` Raslan Darawsheh
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).