Skip to contentSkip to conversation

flow_offload: follow-ups to HW stats type patchset

PATCH In review Jiri Pirko 4 participants 11 messages Mar 10, 2020 – Mar 12, 2020

Subsystems (4)
lore
Share & tools
lore

Jiri Pirko

author
#0 lore raw

cover flow_offload: follow-ups to HW stats type patchset

This patchset includes couple of patches in reaction to the discussions to the original HW stats patchset. The first patch is a fix, the other two patches are basically cosmetics.

Jiri Pirko (3):
flow_offload: fix allowed types check
flow_offload: turn hw_stats_type into dedicated enum
flow_offload: restrict driver to pass one allowed bit to
flow_action_hw_stats_types_check()

 .../net/ethernet/mellanox/mlx5/core/en_tc.c   |  4 +-
 include/net/flow_offload.h                    | 46 +++++++++++++------
 2 files changed, 35 insertions(+), 15 deletions(-)
Signature · 1 line
2.21.1
author
#1 lore raw

PATCH 1/3 flow_offload: fix allowed types check

Change the check to see if the passed allowed type bit is enabled.

Fixes 319a1d19471e ("flow_offload: check for basic action hw stats type")
---
 include/net/flow_offload.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
include/net/flow_offload.h +1 -1
index 891e15055708..2fda4178ba35 100644
@@ -306,7 +306,7 @@ flow_action_hw_stats_types_check(const struct flow_action *action,
 		NL_SET_ERR_MSG_MOD(extack, "Driver supports only default HW stats type \"any\"");
 		return false;
 	} else if (allowed_hw_stats_type != 0 &&
-		   action_entry->hw_stats_type != allowed_hw_stats_type) {
+		   !(action_entry->hw_stats_type & allowed_hw_stats_type)) {
 		NL_SET_ERR_MSG_MOD(extack, "Driver does not support selected HW stats type");
 		return false;
 	}
Signature · 1 line
2.21.1
author
#2 lore raw

PATCH 2/3 flow_offload: turn hw_stats_type into dedicated enum

In reply to Jiri Pirko

Put the values into enum and add an enum to define the bits.

---
 include/net/flow_offload.h | 22 ++++++++++++++++------
 1 file changed, 16 insertions(+), 6 deletions(-)
include/net/flow_offload.h +16 -6
index 2fda4178ba35..6849cb5d4883 100644
@@ -155,11 +155,21 @@ enum flow_action_mangle_base {
 	FLOW_ACT_MANGLE_HDR_TYPE_UDP,
 };
 
-#define FLOW_ACTION_HW_STATS_TYPE_IMMEDIATE BIT(0)
-#define FLOW_ACTION_HW_STATS_TYPE_DELAYED BIT(1)
-#define FLOW_ACTION_HW_STATS_TYPE_ANY (FLOW_ACTION_HW_STATS_TYPE_IMMEDIATE | \
-				       FLOW_ACTION_HW_STATS_TYPE_DELAYED)
-#define FLOW_ACTION_HW_STATS_TYPE_DISABLED 0
+enum flow_action_hw_stats_type_bit {
+	FLOW_ACTION_HW_STATS_TYPE_IMMEDIATE_BIT,
+	FLOW_ACTION_HW_STATS_TYPE_DELAYED_BIT,
+};
+
+enum flow_action_hw_stats_type {
+	FLOW_ACTION_HW_STATS_TYPE_DISABLED = 0,
+	FLOW_ACTION_HW_STATS_TYPE_IMMEDIATE =
+		BIT(FLOW_ACTION_HW_STATS_TYPE_IMMEDIATE_BIT),
+	FLOW_ACTION_HW_STATS_TYPE_DELAYED =
+		BIT(FLOW_ACTION_HW_STATS_TYPE_DELAYED_BIT),
+	FLOW_ACTION_HW_STATS_TYPE_ANY =
+		FLOW_ACTION_HW_STATS_TYPE_IMMEDIATE |
+		FLOW_ACTION_HW_STATS_TYPE_DELAYED,
+};
 
 typedef void (*action_destr)(void *priv);
 
@@ -175,7 +185,7 @@ void flow_action_cookie_destroy(struct flow_action_cookie *cookie);
 
 struct flow_action_entry {
 	enum flow_action_id		id;
-	u8				hw_stats_type;
+	enum flow_action_hw_stats_type	hw_stats_type;
 	action_destr			destructor;
 	void				*destructor_priv;
 	union {
Signature · 1 line
2.21.1
author
#3 lore raw

PATCH 3/3 flow_offload: restrict driver to pass one allowed bit to flow_action_hw_stats_types_check()

In reply to Jiri Pirko

The intention of this helper was to allow driver to specify one type that it supports, so not only "any" value would pass. So make the API more strict and allow driver to pass only 1 bit that is going to be checked.

---
 .../net/ethernet/mellanox/mlx5/core/en_tc.c   |  4 ++--
 include/net/flow_offload.h                    | 24 +++++++++++++------
 2 files changed, 19 insertions(+), 9 deletions(-)
drivers/net/ethernet/mellanox/mlx5/core/en_tc.c +2 -2
index 33d3e70418fb..f285713def77 100644
@@ -2879,7 +2879,7 @@ static int parse_tc_nic_actions(struct mlx5e_priv *priv,
 		return -EINVAL;
 
 	if (!flow_action_hw_stats_types_check(flow_action, extack,
-					      FLOW_ACTION_HW_STATS_TYPE_DELAYED))
+					      FLOW_ACTION_HW_STATS_TYPE_DELAYED_BIT))
 		return -EOPNOTSUPP;
 
 	attr->flow_tag = MLX5_FS_DEFAULT_FLOW_TAG;
@@ -3374,7 +3374,7 @@ static int parse_tc_fdb_actions(struct mlx5e_priv *priv,
 		return -EINVAL;
 
 	if (!flow_action_hw_stats_types_check(flow_action, extack,
-					      FLOW_ACTION_HW_STATS_TYPE_DELAYED))
+					      FLOW_ACTION_HW_STATS_TYPE_DELAYED_BIT))
 		return -EOPNOTSUPP;
 
 	flow_action_for_each(i, act, flow_action) {
include/net/flow_offload.h +17 -7
index 6849cb5d4883..d1b1e4aa310a 100644
@@ -300,9 +300,10 @@ flow_action_first_entry_get(const struct flow_action *action)
 }
 
 static inline bool
-flow_action_hw_stats_types_check(const struct flow_action *action,
-				 struct netlink_ext_ack *extack,
-				 u8 allowed_hw_stats_type)
+__flow_action_hw_stats_types_check(const struct flow_action *action,
+				   struct netlink_ext_ack *extack,
+				   bool check_allow_bit,
+				   enum flow_action_hw_stats_type_bit allow_bit)
 {
 	const struct flow_action_entry *action_entry;
 
@@ -311,23 +312,32 @@ flow_action_hw_stats_types_check(const struct flow_action *action,
 	if (!flow_action_mixed_hw_stats_types_check(action, extack))
 		return false;
 	action_entry = flow_action_first_entry_get(action);
-	if (allowed_hw_stats_type == 0 &&
+	if (!check_allow_bit &&
 	    action_entry->hw_stats_type != FLOW_ACTION_HW_STATS_TYPE_ANY) {
 		NL_SET_ERR_MSG_MOD(extack, "Driver supports only default HW stats type \"any\"");
 		return false;
-	} else if (allowed_hw_stats_type != 0 &&
-		   !(action_entry->hw_stats_type & allowed_hw_stats_type)) {
+	} else if (check_allow_bit &&
+		   !(action_entry->hw_stats_type & BIT(allow_bit))) {
 		NL_SET_ERR_MSG_MOD(extack, "Driver does not support selected HW stats type");
 		return false;
 	}
 	return true;
 }
 
+static inline bool
+flow_action_hw_stats_types_check(const struct flow_action *action,
+				 struct netlink_ext_ack *extack,
+				 enum flow_action_hw_stats_type_bit allow_bit)
+{
+	return __flow_action_hw_stats_types_check(action, extack,
+						  true, allow_bit);
+}
+
 static inline bool
 flow_action_basic_hw_stats_types_check(const struct flow_action *action,
 				       struct netlink_ext_ack *extack)
 {
-	return flow_action_hw_stats_types_check(action, extack, 0);
+	return __flow_action_hw_stats_types_check(action, extack, false, 0);
 }
 
 struct flow_rule {
Signature · 1 line
2.21.1
Acked-by on the v1 series
#4 lore raw
In reply to Jiri Pirko
Quoted 10 lines · Jiri Pirko

This patchset includes couple of patches in reaction to the discussions to the original HW stats patchset. The first patch is a fix, the other two patches are basically cosmetics.

Jiri Pirko (3):
flow_offload: fix allowed types check
flow_offload: turn hw_stats_type into dedicated enum
flow_offload: restrict driver to pass one allowed bit to
flow_action_hw_stats_types_check()

Quoted 4 lines
 .../net/ethernet/mellanox/mlx5/core/en_tc.c   |  4 +-
 include/net/flow_offload.h                    | 46 +++++++++++++------
 2 files changed, 35 insertions(+), 15 deletions(-)
maintainer Reviewed-by on the v1 series
#5 lore raw
In reply to Jiri Pirko
Quoted 3 lines · On Tue, 10 Mar 2020 16:49:06 +0100 Jiri Pirko wrote:

This patchset includes couple of patches in reaction to the discussions to the original HW stats patchset. The first patch is a fix, the other two patches are basically cosmetics.

This problem already exists, but writing a patch for nfp I noticed that there is no way for this:

	if (!flow_action_hw_stats_types_check(flow_action, extack,
					      FLOW_ACTION_HW_STATS_TYPE_DELAYED_BIT))
		return -EOPNOTSUPP;

to fit on a line for either bit, which kind of sucks.

I may send a rename...

author
#6 lore raw
Quoted 7 lines · Tue, Mar 10, 2020 at 08:05:19PM CET, [email protected] wrote:
Quoted 3 lines · On Tue, 10 Mar 2020 16:49:06 +0100 Jiri Pirko wrote:

This patchset includes couple of patches in reaction to the discussions to the original HW stats patchset. The first patch is a fix, the other two patches are basically cosmetics.

Quoted 8 lines

This problem already exists, but writing a patch for nfp I noticed that there is no way for this:

	if (!flow_action_hw_stats_types_check(flow_action, extack,
					      FLOW_ACTION_HW_STATS_TYPE_DELAYED_BIT))
		return -EOPNOTSUPP;

to fit on a line for either bit, which kind of sucks.

Yeah, I was thinking about having flow_action_hw_stats_types_check as a macro and then just simply have:

	if (!flow_action_hw_stats_types_check(flow_action, extack, DELAYED))
		return -EOPNOTSUPP;

WDYT?

Quoted 2 lines

I may send a rename...

maintainer
#7 lore raw
Quoted 17 lines · On Wed, 11 Mar 2020 08:19:55 +0100 Jiri Pirko wrote:
Quoted 7 lines · Tue, Mar 10, 2020 at 08:05:19PM CET, [email protected] wrote:
Quoted 3 lines · On Tue, 10 Mar 2020 16:49:06 +0100 Jiri Pirko wrote:

This patchset includes couple of patches in reaction to the discussions to the original HW stats patchset. The first patch is a fix, the other two patches are basically cosmetics.

Quoted 8 lines

This problem already exists, but writing a patch for nfp I noticed that there is no way for this:

	if (!flow_action_hw_stats_types_check(flow_action, extack,
					      FLOW_ACTION_HW_STATS_TYPE_DELAYED_BIT))
		return -EOPNOTSUPP;

to fit on a line for either bit, which kind of sucks.

Quoted 7 lines

Yeah, I was thinking about having flow_action_hw_stats_types_check as a macro and then just simply have:

	if (!flow_action_hw_stats_types_check(flow_action, extack, DELAYED))
		return -EOPNOTSUPP;

WDYT?

I'd rather have the 80+ lines than not be able to grep for it :(

What's wrong with flow_action_stats_ok()? Also perhaps, flow_act as a prefix?

author
#8 lore raw
Quoted 24 lines · Wed, Mar 11, 2020 at 09:30:28PM CET, [email protected] wrote:
Quoted 23 lines · On Wed, 11 Mar 2020 08:19:55 +0100 Jiri Pirko wrote:
Quoted 7 lines · Tue, Mar 10, 2020 at 08:05:19PM CET, [email protected] wrote:
Quoted 3 lines · On Tue, 10 Mar 2020 16:49:06 +0100 Jiri Pirko wrote:

This patchset includes couple of patches in reaction to the discussions to the original HW stats patchset. The first patch is a fix, the other two patches are basically cosmetics.

Quoted 8 lines

This problem already exists, but writing a patch for nfp I noticed that there is no way for this:

	if (!flow_action_hw_stats_types_check(flow_action, extack,
					      FLOW_ACTION_HW_STATS_TYPE_DELAYED_BIT))
		return -EOPNOTSUPP;

to fit on a line for either bit, which kind of sucks.

Yeah, I was thinking about having flow_action_hw_stats_types_check as a macro and then just simply have:

	if (!flow_action_hw_stats_types_check(flow_action, extack, DELAYED))
		return -EOPNOTSUPP;
Quoted 6 lines
Quoted 1 line

WDYT?

I'd rather have the 80+ lines than not be able to grep for it :(

What's wrong with flow_action_stats_ok()? Also perhaps, flow_act as a prefix?

Well nothing, just that we'd loose consistency. Everything is "flow_action_*" and also, the name you suggest might indicate that you are checking sw stats. :/

maintainer
#9 lore raw
Quoted 27 lines · On Thu, 12 Mar 2020 08:03:59 +0100 Jiri Pirko wrote:
Quoted 26 lines · Wed, Mar 11, 2020 at 09:30:28PM CET, [email protected] wrote:
Quoted 24 lines · On Wed, 11 Mar 2020 08:19:55 +0100 Jiri Pirko wrote:
Quoted 7 lines · Tue, Mar 10, 2020 at 08:05:19PM CET, [email protected] wrote:
Quoted 3 lines · On Tue, 10 Mar 2020 16:49:06 +0100 Jiri Pirko wrote:

This patchset includes couple of patches in reaction to the discussions to the original HW stats patchset. The first patch is a fix, the other two patches are basically cosmetics.

Quoted 8 lines

This problem already exists, but writing a patch for nfp I noticed that there is no way for this:

	if (!flow_action_hw_stats_types_check(flow_action, extack,
					      FLOW_ACTION_HW_STATS_TYPE_DELAYED_BIT))
		return -EOPNOTSUPP;

to fit on a line for either bit, which kind of sucks.

Yeah, I was thinking about having flow_action_hw_stats_types_check as a macro and then just simply have:

	if (!flow_action_hw_stats_types_check(flow_action, extack, DELAYED))
		return -EOPNOTSUPP;

WDYT?

Quoted 8 lines
Quoted 4 lines

I'd rather have the 80+ lines than not be able to grep for it :(

What's wrong with flow_action_stats_ok()? Also perhaps, flow_act as a prefix?

Well nothing, just that we'd loose consistency. Everything is "flow_action_*" and also, the name you suggest might indicate that you are checking sw stats. :/

SW stats in flow action? flow stuff is an abstraction for HW/drivers.

maintainer
#10 lore raw
In reply to Jiri Pirko

From: Jiri Pirko <[email protected]>
Date: Tue, 10 Mar 2020 16:49:06 +0100

Quoted 3 lines

This patchset includes couple of patches in reaction to the discussions to the original HW stats patchset. The first patch is a fix, the other two patches are basically cosmetics.

Series applied, thanks Jiri.