* [PATCH net V2 1/7] net/mlx5: IFC updates for changing max EQs
2024-06-27 18:02 [PATCH net V2 0/7] mlx5 fixes 2024-06-27 Tariq Toukan
@ 2024-06-27 18:02 ` Tariq Toukan
2024-06-30 7:30 ` Tariq Toukan
2024-06-27 18:02 ` [PATCH net V2 2/7] net/mlx5: Use max_num_eqs_24b capability if set Tariq Toukan
` (6 subsequent siblings)
7 siblings, 1 reply; 10+ messages in thread
From: Tariq Toukan @ 2024-06-27 18:02 UTC (permalink / raw)
To: David S. Miller, Jakub Kicinski, Paolo Abeni, Eric Dumazet
Cc: netdev, Saeed Mahameed, Gal Pressman, Leon Romanovsky,
Daniel Jurgens, Parav Pandit, William Tu, Tariq Toukan
From: Daniel Jurgens <danielj@nvidia•com>
Expose new capability to support changing the number of EQs available
to other functions.
Fixes: 93197c7c509d ("mlx5/core: Support max_io_eqs for a function")
Signed-off-by: Daniel Jurgens <danielj@nvidia•com>
Reviewed-by: Parav Pandit <parav@nvidia•com>
Reviewed-by: William Tu <witu@nvidia•com>
Signed-off-by: Tariq Toukan <tariqt@nvidia•com>
---
include/linux/mlx5/mlx5_ifc.h | 6 +++++-
1 file changed, 5 insertions(+), 1 deletion(-)
diff --git a/include/linux/mlx5/mlx5_ifc.h b/include/linux/mlx5/mlx5_ifc.h
index 5df52e15f7d6..d45bfb7cf81d 100644
--- a/include/linux/mlx5/mlx5_ifc.h
+++ b/include/linux/mlx5/mlx5_ifc.h
@@ -2029,7 +2029,11 @@ struct mlx5_ifc_cmd_hca_cap_2_bits {
u8 pcc_ifa2[0x1];
u8 reserved_at_3f1[0xf];
- u8 reserved_at_400[0x400];
+ u8 reserved_at_400[0x40];
+
+ u8 reserved_at_440[0x8];
+ u8 max_num_eqs_24b[0x18];
+ u8 reserved_at_460[0x3a0];
};
enum mlx5_ifc_flow_destination_type {
--
2.31.1
^ permalink raw reply related [flat|nested] 10+ messages in thread* Re: [PATCH net V2 1/7] net/mlx5: IFC updates for changing max EQs
2024-06-27 18:02 ` [PATCH net V2 1/7] net/mlx5: IFC updates for changing max EQs Tariq Toukan
@ 2024-06-30 7:30 ` Tariq Toukan
0 siblings, 0 replies; 10+ messages in thread
From: Tariq Toukan @ 2024-06-30 7:30 UTC (permalink / raw)
To: Tariq Toukan, David S. Miller, Jakub Kicinski, Paolo Abeni,
Eric Dumazet
Cc: netdev, Saeed Mahameed, Gal Pressman, Leon Romanovsky,
Daniel Jurgens, Parav Pandit, William Tu
On 27/06/2024 21:02, Tariq Toukan wrote:
> From: Daniel Jurgens <danielj@nvidia•com>
>
> Expose new capability to support changing the number of EQs available
> to other functions.
>
> Fixes: 93197c7c509d ("mlx5/core: Support max_io_eqs for a function")
> Signed-off-by: Daniel Jurgens <danielj@nvidia•com>
> Reviewed-by: Parav Pandit <parav@nvidia•com>
> Reviewed-by: William Tu <witu@nvidia•com>
> Signed-off-by: Tariq Toukan <tariqt@nvidia•com>
> ---
> include/linux/mlx5/mlx5_ifc.h | 6 +++++-
> 1 file changed, 5 insertions(+), 1 deletion(-)
>
> diff --git a/include/linux/mlx5/mlx5_ifc.h b/include/linux/mlx5/mlx5_ifc.h
> index 5df52e15f7d6..d45bfb7cf81d 100644
> --- a/include/linux/mlx5/mlx5_ifc.h
> +++ b/include/linux/mlx5/mlx5_ifc.h
> @@ -2029,7 +2029,11 @@ struct mlx5_ifc_cmd_hca_cap_2_bits {
> u8 pcc_ifa2[0x1];
> u8 reserved_at_3f1[0xf];
>
> - u8 reserved_at_400[0x400];
> + u8 reserved_at_400[0x40];
> +
> + u8 reserved_at_440[0x8];
> + u8 max_num_eqs_24b[0x18];
> + u8 reserved_at_460[0x3a0];
> };
>
> enum mlx5_ifc_flow_destination_type {
Hi,
Please note that this is expected to conflict with net-next commit:
99be56171fa9 net/mlx5e: SHAMPO, Re-enable HW-GRO.
Resolution is
u8 reserved_at_400[0x1];
u8 min_mkey_log_entity_size_fixed_buffer_valid[0x1];
u8 reserved_at_402[0x1e];
u8 reserved_at_420[0x20];
u8 reserved_at_440[0x8];
u8 max_num_eqs_24b[0x18];
u8 reserved_at_460[0x3a0];
Thanks,
Tariq
^ permalink raw reply [flat|nested] 10+ messages in thread
* [PATCH net V2 2/7] net/mlx5: Use max_num_eqs_24b capability if set
2024-06-27 18:02 [PATCH net V2 0/7] mlx5 fixes 2024-06-27 Tariq Toukan
2024-06-27 18:02 ` [PATCH net V2 1/7] net/mlx5: IFC updates for changing max EQs Tariq Toukan
@ 2024-06-27 18:02 ` Tariq Toukan
2024-06-27 18:02 ` [PATCH net V2 3/7] net/mlx5: Use max_num_eqs_24b when setting max_io_eqs Tariq Toukan
` (5 subsequent siblings)
7 siblings, 0 replies; 10+ messages in thread
From: Tariq Toukan @ 2024-06-27 18:02 UTC (permalink / raw)
To: David S. Miller, Jakub Kicinski, Paolo Abeni, Eric Dumazet
Cc: netdev, Saeed Mahameed, Gal Pressman, Leon Romanovsky,
Daniel Jurgens, Parav Pandit, William Tu, Tariq Toukan
From: Daniel Jurgens <danielj@nvidia•com>
A new capability with more bits is added. If it's set use that value as
the maximum number of EQs available.
This cap is also writable by the vhca_resource_manager to allow limiting
the number of EQs available to SFs and VFs.
Fixes: 93197c7c509d ("mlx5/core: Support max_io_eqs for a function")
Signed-off-by: Daniel Jurgens <danielj@nvidia•com>
Reviewed-by: Parav Pandit <parav@nvidia•com>
Reviewed-by: William Tu <witu@nvidia•com>
Signed-off-by: Tariq Toukan <tariqt@nvidia•com>
---
drivers/net/ethernet/mellanox/mlx5/core/eq.c | 4 +---
drivers/net/ethernet/mellanox/mlx5/core/mlx5_core.h | 10 ++++++++++
drivers/net/ethernet/mellanox/mlx5/core/pci_irq.c | 4 +---
3 files changed, 12 insertions(+), 6 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eq.c b/drivers/net/ethernet/mellanox/mlx5/core/eq.c
index 5693986ae656..ac1565c0c8af 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/eq.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eq.c
@@ -1197,9 +1197,7 @@ static int get_num_eqs(struct mlx5_core_dev *dev)
if (!mlx5_core_is_eth_enabled(dev) && mlx5_eth_supported(dev))
return 1;
- max_dev_eqs = MLX5_CAP_GEN(dev, max_num_eqs) ?
- MLX5_CAP_GEN(dev, max_num_eqs) :
- 1 << MLX5_CAP_GEN(dev, log_max_eq);
+ max_dev_eqs = mlx5_max_eq_cap_get(dev);
num_eqs = min_t(int, mlx5_irq_table_get_num_comp(eq_table->irq_table),
max_dev_eqs - MLX5_MAX_ASYNC_EQS);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/mlx5_core.h b/drivers/net/ethernet/mellanox/mlx5/core/mlx5_core.h
index c38342b9f320..a7fd18888b6e 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/mlx5_core.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/mlx5_core.h
@@ -383,4 +383,14 @@ static inline int mlx5_vport_to_func_id(const struct mlx5_core_dev *dev, u16 vpo
: vport;
}
+static inline int mlx5_max_eq_cap_get(const struct mlx5_core_dev *dev)
+{
+ if (MLX5_CAP_GEN_2(dev, max_num_eqs_24b))
+ return MLX5_CAP_GEN_2(dev, max_num_eqs_24b);
+
+ if (MLX5_CAP_GEN(dev, max_num_eqs))
+ return MLX5_CAP_GEN(dev, max_num_eqs);
+
+ return 1 << MLX5_CAP_GEN(dev, log_max_eq);
+}
#endif /* __MLX5_CORE_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/pci_irq.c b/drivers/net/ethernet/mellanox/mlx5/core/pci_irq.c
index fb8787e30d3f..401d39069680 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/pci_irq.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/pci_irq.c
@@ -711,9 +711,7 @@ int mlx5_irq_table_get_num_comp(struct mlx5_irq_table *table)
int mlx5_irq_table_create(struct mlx5_core_dev *dev)
{
- int num_eqs = MLX5_CAP_GEN(dev, max_num_eqs) ?
- MLX5_CAP_GEN(dev, max_num_eqs) :
- 1 << MLX5_CAP_GEN(dev, log_max_eq);
+ int num_eqs = mlx5_max_eq_cap_get(dev);
int total_vec;
int pcif_vec;
int req_vec;
--
2.31.1
^ permalink raw reply related [flat|nested] 10+ messages in thread* [PATCH net V2 3/7] net/mlx5: Use max_num_eqs_24b when setting max_io_eqs
2024-06-27 18:02 [PATCH net V2 0/7] mlx5 fixes 2024-06-27 Tariq Toukan
2024-06-27 18:02 ` [PATCH net V2 1/7] net/mlx5: IFC updates for changing max EQs Tariq Toukan
2024-06-27 18:02 ` [PATCH net V2 2/7] net/mlx5: Use max_num_eqs_24b capability if set Tariq Toukan
@ 2024-06-27 18:02 ` Tariq Toukan
2024-06-27 18:02 ` [PATCH net V2 4/7] net/mlx5: E-switch, Create ingress ACL when needed Tariq Toukan
` (4 subsequent siblings)
7 siblings, 0 replies; 10+ messages in thread
From: Tariq Toukan @ 2024-06-27 18:02 UTC (permalink / raw)
To: David S. Miller, Jakub Kicinski, Paolo Abeni, Eric Dumazet
Cc: netdev, Saeed Mahameed, Gal Pressman, Leon Romanovsky,
Daniel Jurgens, Parav Pandit, Tariq Toukan
From: Daniel Jurgens <danielj@nvidia•com>
Due a bug in the device max_num_eqs doesn't always reflect a written
value. As a result, setting max_io_eqs may not work but appear
successful. Instead write max_num_eqs_24b, which reflects correct
value.
Fixes: 93197c7c509d ("mlx5/core: Support max_io_eqs for a function")
Signed-off-by: Daniel Jurgens <danielj@nvidia•com>
Reviewed-by: Parav Pandit <parav@nvidia•com>
Signed-off-by: Tariq Toukan <tariqt@nvidia•com>
---
.../mellanox/mlx5/core/eswitch_offloads.c | 22 ++++++++++++++-----
1 file changed, 17 insertions(+), 5 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c b/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
index 592143d5e1da..72949cb85244 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
@@ -4600,20 +4600,26 @@ mlx5_devlink_port_fn_max_io_eqs_get(struct devlink_port *port, u32 *max_io_eqs,
return -EOPNOTSUPP;
}
+ if (!MLX5_CAP_GEN_2(esw->dev, max_num_eqs_24b)) {
+ NL_SET_ERR_MSG_MOD(extack,
+ "Device doesn't support getting the max number of EQs");
+ return -EOPNOTSUPP;
+ }
+
query_ctx = kzalloc(query_out_sz, GFP_KERNEL);
if (!query_ctx)
return -ENOMEM;
mutex_lock(&esw->state_lock);
err = mlx5_vport_get_other_func_cap(esw->dev, vport_num, query_ctx,
- MLX5_CAP_GENERAL);
+ MLX5_CAP_GENERAL_2);
if (err) {
NL_SET_ERR_MSG_MOD(extack, "Failed getting HCA caps");
goto out;
}
hca_caps = MLX5_ADDR_OF(query_hca_cap_out, query_ctx, capability);
- max_eqs = MLX5_GET(cmd_hca_cap, hca_caps, max_num_eqs);
+ max_eqs = MLX5_GET(cmd_hca_cap_2, hca_caps, max_num_eqs_24b);
if (max_eqs < MLX5_ESW_MAX_CTRL_EQS)
*max_io_eqs = 0;
else
@@ -4644,6 +4650,12 @@ mlx5_devlink_port_fn_max_io_eqs_set(struct devlink_port *port, u32 max_io_eqs,
return -EOPNOTSUPP;
}
+ if (!MLX5_CAP_GEN_2(esw->dev, max_num_eqs_24b)) {
+ NL_SET_ERR_MSG_MOD(extack,
+ "Device doesn't support changing the max number of EQs");
+ return -EOPNOTSUPP;
+ }
+
if (check_add_overflow(max_io_eqs, MLX5_ESW_MAX_CTRL_EQS, &max_eqs)) {
NL_SET_ERR_MSG_MOD(extack, "Supplied value out of range");
return -EINVAL;
@@ -4655,17 +4667,17 @@ mlx5_devlink_port_fn_max_io_eqs_set(struct devlink_port *port, u32 max_io_eqs,
mutex_lock(&esw->state_lock);
err = mlx5_vport_get_other_func_cap(esw->dev, vport_num, query_ctx,
- MLX5_CAP_GENERAL);
+ MLX5_CAP_GENERAL_2);
if (err) {
NL_SET_ERR_MSG_MOD(extack, "Failed getting HCA caps");
goto out;
}
hca_caps = MLX5_ADDR_OF(query_hca_cap_out, query_ctx, capability);
- MLX5_SET(cmd_hca_cap, hca_caps, max_num_eqs, max_eqs);
+ MLX5_SET(cmd_hca_cap_2, hca_caps, max_num_eqs_24b, max_eqs);
err = mlx5_vport_set_other_func_cap(esw->dev, hca_caps, vport_num,
- MLX5_SET_HCA_CAP_OP_MOD_GENERAL_DEVICE);
+ MLX5_SET_HCA_CAP_OP_MOD_GENERAL_DEVICE2);
if (err)
NL_SET_ERR_MSG_MOD(extack, "Failed setting HCA caps");
--
2.31.1
^ permalink raw reply related [flat|nested] 10+ messages in thread* [PATCH net V2 4/7] net/mlx5: E-switch, Create ingress ACL when needed
2024-06-27 18:02 [PATCH net V2 0/7] mlx5 fixes 2024-06-27 Tariq Toukan
` (2 preceding siblings ...)
2024-06-27 18:02 ` [PATCH net V2 3/7] net/mlx5: Use max_num_eqs_24b when setting max_io_eqs Tariq Toukan
@ 2024-06-27 18:02 ` Tariq Toukan
2024-06-27 18:02 ` [PATCH net V2 5/7] net/mlx5e: Add mqprio_rl cleanup and free in mlx5e_priv_cleanup() Tariq Toukan
` (3 subsequent siblings)
7 siblings, 0 replies; 10+ messages in thread
From: Tariq Toukan @ 2024-06-27 18:02 UTC (permalink / raw)
To: David S. Miller, Jakub Kicinski, Paolo Abeni, Eric Dumazet
Cc: netdev, Saeed Mahameed, Gal Pressman, Leon Romanovsky, Chris Mi,
Tariq Toukan
From: Chris Mi <cmi@nvidia•com>
Currently, ingress acl is used for three features. It is created only
when vport metadata match and prio tag are enabled. But active-backup
lag mode also uses it. It is independent of vport metadata match and
prio tag. And vport metadata match can be disabled using the
following devlink command:
# devlink dev param set pci/0000:08:00.0 name esw_port_metadata \
value false cmode runtime
If ingress acl is not created, will hit panic when creating drop rule
for active-backup lag mode. If always create it, there will be about
5% performance degradation.
Fix it by creating ingress acl when needed. If esw_port_metadata is
true, ingress acl exists, then create drop rule using existing
ingress acl. If esw_port_metadata is false, create ingress acl and
then create drop rule.
Fixes: 1749c4c51c16 ("net/mlx5: E-switch, add drop rule support to ingress ACL")
Signed-off-by: Chris Mi <cmi@nvidia•com>
Signed-off-by: Tariq Toukan <tariqt@nvidia•com>
---
.../mellanox/mlx5/core/esw/acl/ingress_ofld.c | 37 +++++++++++++++----
1 file changed, 29 insertions(+), 8 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/esw/acl/ingress_ofld.c b/drivers/net/ethernet/mellanox/mlx5/core/esw/acl/ingress_ofld.c
index 50d2ea323979..a436ce895e45 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/esw/acl/ingress_ofld.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/esw/acl/ingress_ofld.c
@@ -6,6 +6,9 @@
#include "helper.h"
#include "ofld.h"
+static int
+acl_ingress_ofld_setup(struct mlx5_eswitch *esw, struct mlx5_vport *vport);
+
static bool
esw_acl_ingress_prio_tag_enabled(struct mlx5_eswitch *esw,
const struct mlx5_vport *vport)
@@ -123,18 +126,31 @@ static int esw_acl_ingress_src_port_drop_create(struct mlx5_eswitch *esw,
{
struct mlx5_flow_act flow_act = {};
struct mlx5_flow_handle *flow_rule;
+ bool created = false;
int err = 0;
+ if (!vport->ingress.acl) {
+ err = acl_ingress_ofld_setup(esw, vport);
+ if (err)
+ return err;
+ created = true;
+ }
+
flow_act.action = MLX5_FLOW_CONTEXT_ACTION_DROP;
flow_act.fg = vport->ingress.offloads.drop_grp;
flow_rule = mlx5_add_flow_rules(vport->ingress.acl, NULL, &flow_act, NULL, 0);
if (IS_ERR(flow_rule)) {
err = PTR_ERR(flow_rule);
- goto out;
+ goto err_out;
}
vport->ingress.offloads.drop_rule = flow_rule;
-out:
+
+ return 0;
+err_out:
+ /* Only destroy ingress acl created in this function. */
+ if (created)
+ esw_acl_ingress_ofld_cleanup(esw, vport);
return err;
}
@@ -299,16 +315,12 @@ static void esw_acl_ingress_ofld_groups_destroy(struct mlx5_vport *vport)
}
}
-int esw_acl_ingress_ofld_setup(struct mlx5_eswitch *esw,
- struct mlx5_vport *vport)
+static int
+acl_ingress_ofld_setup(struct mlx5_eswitch *esw, struct mlx5_vport *vport)
{
int num_ftes = 0;
int err;
- if (!mlx5_eswitch_vport_match_metadata_enabled(esw) &&
- !esw_acl_ingress_prio_tag_enabled(esw, vport))
- return 0;
-
esw_acl_ingress_allow_rule_destroy(vport);
if (mlx5_eswitch_vport_match_metadata_enabled(esw))
@@ -347,6 +359,15 @@ int esw_acl_ingress_ofld_setup(struct mlx5_eswitch *esw,
return err;
}
+int esw_acl_ingress_ofld_setup(struct mlx5_eswitch *esw, struct mlx5_vport *vport)
+{
+ if (!mlx5_eswitch_vport_match_metadata_enabled(esw) &&
+ !esw_acl_ingress_prio_tag_enabled(esw, vport))
+ return 0;
+
+ return acl_ingress_ofld_setup(esw, vport);
+}
+
void esw_acl_ingress_ofld_cleanup(struct mlx5_eswitch *esw,
struct mlx5_vport *vport)
{
--
2.31.1
^ permalink raw reply related [flat|nested] 10+ messages in thread* [PATCH net V2 5/7] net/mlx5e: Add mqprio_rl cleanup and free in mlx5e_priv_cleanup()
2024-06-27 18:02 [PATCH net V2 0/7] mlx5 fixes 2024-06-27 Tariq Toukan
` (3 preceding siblings ...)
2024-06-27 18:02 ` [PATCH net V2 4/7] net/mlx5: E-switch, Create ingress ACL when needed Tariq Toukan
@ 2024-06-27 18:02 ` Tariq Toukan
2024-06-27 18:02 ` [PATCH net V2 6/7] net/mlx5e: Present succeeded IPsec SA bytes and packet Tariq Toukan
` (2 subsequent siblings)
7 siblings, 0 replies; 10+ messages in thread
From: Tariq Toukan @ 2024-06-27 18:02 UTC (permalink / raw)
To: David S. Miller, Jakub Kicinski, Paolo Abeni, Eric Dumazet
Cc: netdev, Saeed Mahameed, Gal Pressman, Leon Romanovsky, Jianbo Liu,
Dragos Tatulea, Tariq Toukan
From: Jianbo Liu <jianbol@nvidia•com>
In the cited commit, mqprio_rl cleanup and free are mistakenly removed
in mlx5e_priv_cleanup(), and it causes the leakage of host memory and
firmware SCHEDULING_ELEMENT objects while changing eswitch mode. So,
add them back.
Fixes: 0bb7228f7096 ("net/mlx5e: Fix mqprio_rl handling on devlink reload")
Signed-off-by: Jianbo Liu <jianbol@nvidia•com>
Reviewed-by: Dragos Tatulea <dtatulea@nvidia•com>
Signed-off-by: Tariq Toukan <tariqt@nvidia•com>
---
drivers/net/ethernet/mellanox/mlx5/core/en_main.c | 5 +++++
1 file changed, 5 insertions(+)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
index a605eae56685..eedbcba22689 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
@@ -5868,6 +5868,11 @@ void mlx5e_priv_cleanup(struct mlx5e_priv *priv)
kfree(priv->htb_qos_sq_stats[i]);
kvfree(priv->htb_qos_sq_stats);
+ if (priv->mqprio_rl) {
+ mlx5e_mqprio_rl_cleanup(priv->mqprio_rl);
+ mlx5e_mqprio_rl_free(priv->mqprio_rl);
+ }
+
memset(priv, 0, sizeof(*priv));
}
--
2.31.1
^ permalink raw reply related [flat|nested] 10+ messages in thread* [PATCH net V2 6/7] net/mlx5e: Present succeeded IPsec SA bytes and packet
2024-06-27 18:02 [PATCH net V2 0/7] mlx5 fixes 2024-06-27 Tariq Toukan
` (4 preceding siblings ...)
2024-06-27 18:02 ` [PATCH net V2 5/7] net/mlx5e: Add mqprio_rl cleanup and free in mlx5e_priv_cleanup() Tariq Toukan
@ 2024-06-27 18:02 ` Tariq Toukan
2024-06-27 18:02 ` [PATCH net V2 7/7] net/mlx5e: Approximate IPsec per-SA payload data bytes count Tariq Toukan
2024-06-28 12:00 ` [PATCH net V2 0/7] mlx5 fixes 2024-06-27 patchwork-bot+netdevbpf
7 siblings, 0 replies; 10+ messages in thread
From: Tariq Toukan @ 2024-06-27 18:02 UTC (permalink / raw)
To: David S. Miller, Jakub Kicinski, Paolo Abeni, Eric Dumazet
Cc: netdev, Saeed Mahameed, Gal Pressman, Leon Romanovsky,
Tariq Toukan
From: Leon Romanovsky <leonro@nvidia•com>
IPsec SA statistics presents successfully decrypted and encrypted
packet and bytes, and not total handled by this SA. So update the
calculation logic to take into account failures.
Fixes: 6fb7f9408779 ("net/mlx5e: Connect mlx5 IPsec statistics with XFRM core")
Signed-off-by: Leon Romanovsky <leonro@nvidia•com>
Signed-off-by: Tariq Toukan <tariqt@nvidia•com>
---
.../mellanox/mlx5/core/en_accel/ipsec.c | 36 ++++++++++++-------
1 file changed, 23 insertions(+), 13 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec.c b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec.c
index c54fd01ea635..2a10428d820a 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec.c
@@ -989,6 +989,10 @@ static void mlx5e_xfrm_update_stats(struct xfrm_state *x)
struct mlx5e_ipsec_sa_entry *sa_entry = to_ipsec_sa_entry(x);
struct mlx5e_ipsec_rule *ipsec_rule = &sa_entry->ipsec_rule;
struct net *net = dev_net(x->xso.dev);
+ u64 trailer_packets = 0, trailer_bytes = 0;
+ u64 replay_packets = 0, replay_bytes = 0;
+ u64 auth_packets = 0, auth_bytes = 0;
+ u64 success_packets, success_bytes;
u64 packets, bytes, lastuse;
lockdep_assert(lockdep_is_held(&x->lock) ||
@@ -999,26 +1003,32 @@ static void mlx5e_xfrm_update_stats(struct xfrm_state *x)
return;
if (sa_entry->attrs.dir == XFRM_DEV_OFFLOAD_IN) {
- mlx5_fc_query_cached(ipsec_rule->auth.fc, &bytes, &packets, &lastuse);
- x->stats.integrity_failed += packets;
- XFRM_ADD_STATS(net, LINUX_MIB_XFRMINSTATEPROTOERROR, packets);
-
- mlx5_fc_query_cached(ipsec_rule->trailer.fc, &bytes, &packets, &lastuse);
- XFRM_ADD_STATS(net, LINUX_MIB_XFRMINHDRERROR, packets);
+ mlx5_fc_query_cached(ipsec_rule->auth.fc, &auth_bytes,
+ &auth_packets, &lastuse);
+ x->stats.integrity_failed += auth_packets;
+ XFRM_ADD_STATS(net, LINUX_MIB_XFRMINSTATEPROTOERROR, auth_packets);
+
+ mlx5_fc_query_cached(ipsec_rule->trailer.fc, &trailer_bytes,
+ &trailer_packets, &lastuse);
+ XFRM_ADD_STATS(net, LINUX_MIB_XFRMINHDRERROR, trailer_packets);
}
if (x->xso.type != XFRM_DEV_OFFLOAD_PACKET)
return;
- mlx5_fc_query_cached(ipsec_rule->fc, &bytes, &packets, &lastuse);
- x->curlft.packets += packets;
- x->curlft.bytes += bytes;
-
if (sa_entry->attrs.dir == XFRM_DEV_OFFLOAD_IN) {
- mlx5_fc_query_cached(ipsec_rule->replay.fc, &bytes, &packets, &lastuse);
- x->stats.replay += packets;
- XFRM_ADD_STATS(net, LINUX_MIB_XFRMINSTATESEQERROR, packets);
+ mlx5_fc_query_cached(ipsec_rule->replay.fc, &replay_bytes,
+ &replay_packets, &lastuse);
+ x->stats.replay += replay_packets;
+ XFRM_ADD_STATS(net, LINUX_MIB_XFRMINSTATESEQERROR, replay_packets);
}
+
+ mlx5_fc_query_cached(ipsec_rule->fc, &bytes, &packets, &lastuse);
+ success_packets = packets - auth_packets - trailer_packets - replay_packets;
+ x->curlft.packets += success_packets;
+
+ success_bytes = bytes - auth_bytes - trailer_bytes - replay_bytes;
+ x->curlft.bytes += success_bytes;
}
static int mlx5e_xfrm_validate_policy(struct mlx5_core_dev *mdev,
--
2.31.1
^ permalink raw reply related [flat|nested] 10+ messages in thread* [PATCH net V2 7/7] net/mlx5e: Approximate IPsec per-SA payload data bytes count
2024-06-27 18:02 [PATCH net V2 0/7] mlx5 fixes 2024-06-27 Tariq Toukan
` (5 preceding siblings ...)
2024-06-27 18:02 ` [PATCH net V2 6/7] net/mlx5e: Present succeeded IPsec SA bytes and packet Tariq Toukan
@ 2024-06-27 18:02 ` Tariq Toukan
2024-06-28 12:00 ` [PATCH net V2 0/7] mlx5 fixes 2024-06-27 patchwork-bot+netdevbpf
7 siblings, 0 replies; 10+ messages in thread
From: Tariq Toukan @ 2024-06-27 18:02 UTC (permalink / raw)
To: David S. Miller, Jakub Kicinski, Paolo Abeni, Eric Dumazet
Cc: netdev, Saeed Mahameed, Gal Pressman, Leon Romanovsky,
Tariq Toukan
From: Leon Romanovsky <leonro@nvidia•com>
ConnectX devices lack ability to count payload data byte size which is
needed for SA to return to libreswan for rekeying.
As a solution let's approximate that by decreasing headers size from
total size counted by flow steering. The calculation doesn't take into
account any other headers which can be in the packet (e.g. IP extensions).
Fixes: 5a6cddb89b51 ("net/mlx5e: Update IPsec per SA packets/bytes count")
Signed-off-by: Leon Romanovsky <leonro@nvidia•com>
Signed-off-by: Tariq Toukan <tariqt@nvidia•com>
---
.../ethernet/mellanox/mlx5/core/en_accel/ipsec.c | 14 +++++++++++++-
1 file changed, 13 insertions(+), 1 deletion(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec.c b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec.c
index 2a10428d820a..3d274599015b 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec.c
@@ -994,6 +994,7 @@ static void mlx5e_xfrm_update_stats(struct xfrm_state *x)
u64 auth_packets = 0, auth_bytes = 0;
u64 success_packets, success_bytes;
u64 packets, bytes, lastuse;
+ size_t headers;
lockdep_assert(lockdep_is_held(&x->lock) ||
lockdep_is_held(&dev_net(x->xso.real_dev)->xfrm.xfrm_cfg_mutex) ||
@@ -1026,9 +1027,20 @@ static void mlx5e_xfrm_update_stats(struct xfrm_state *x)
mlx5_fc_query_cached(ipsec_rule->fc, &bytes, &packets, &lastuse);
success_packets = packets - auth_packets - trailer_packets - replay_packets;
x->curlft.packets += success_packets;
+ /* NIC counts all bytes passed through flow steering and doesn't have
+ * an ability to count payload data size which is needed for SA.
+ *
+ * To overcome HW limitestion, let's approximate the payload size
+ * by removing always available headers.
+ */
+ headers = sizeof(struct ethhdr);
+ if (sa_entry->attrs.family == AF_INET)
+ headers += sizeof(struct iphdr);
+ else
+ headers += sizeof(struct ipv6hdr);
success_bytes = bytes - auth_bytes - trailer_bytes - replay_bytes;
- x->curlft.bytes += success_bytes;
+ x->curlft.bytes += success_bytes - headers * success_packets;
}
static int mlx5e_xfrm_validate_policy(struct mlx5_core_dev *mdev,
--
2.31.1
^ permalink raw reply related [flat|nested] 10+ messages in thread* Re: [PATCH net V2 0/7] mlx5 fixes 2024-06-27
2024-06-27 18:02 [PATCH net V2 0/7] mlx5 fixes 2024-06-27 Tariq Toukan
` (6 preceding siblings ...)
2024-06-27 18:02 ` [PATCH net V2 7/7] net/mlx5e: Approximate IPsec per-SA payload data bytes count Tariq Toukan
@ 2024-06-28 12:00 ` patchwork-bot+netdevbpf
7 siblings, 0 replies; 10+ messages in thread
From: patchwork-bot+netdevbpf @ 2024-06-28 12:00 UTC (permalink / raw)
To: Tariq Toukan; +Cc: davem, kuba, pabeni, edumazet, netdev, saeedm, gal, leonro
Hello:
This series was applied to netdev/net.git (main)
by David S. Miller <davem@davemloft•net>:
On Thu, 27 Jun 2024 21:02:33 +0300 you wrote:
> Hi,
>
> This patchset provides fixes from the team to the mlx5 core and EN
> drivers.
>
> The first 3 patches by Daniel replace a buggy cap field with a newly
> introduced one.
>
> [...]
Here is the summary with links:
- [net,V2,1/7] net/mlx5: IFC updates for changing max EQs
https://git.kernel.org/netdev/net/c/048a403648fc
- [net,V2,2/7] net/mlx5: Use max_num_eqs_24b capability if set
https://git.kernel.org/netdev/net/c/29c6a562bf53
- [net,V2,3/7] net/mlx5: Use max_num_eqs_24b when setting max_io_eqs
https://git.kernel.org/netdev/net/c/5dbf647367e8
- [net,V2,4/7] net/mlx5: E-switch, Create ingress ACL when needed
https://git.kernel.org/netdev/net/c/b20c2fb45470
- [net,V2,5/7] net/mlx5e: Add mqprio_rl cleanup and free in mlx5e_priv_cleanup()
https://git.kernel.org/netdev/net/c/1da839eab6db
- [net,V2,6/7] net/mlx5e: Present succeeded IPsec SA bytes and packet
https://git.kernel.org/netdev/net/c/2d9dac5559f8
- [net,V2,7/7] net/mlx5e: Approximate IPsec per-SA payload data bytes count
https://git.kernel.org/netdev/net/c/e562f2d46d27
You are awesome, thank you!
--
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html
^ permalink raw reply [flat|nested] 10+ messages in thread