Merge branch 'mlx5-hw-managed-flow-steering-in-fs-core-level'

Tariq Toukan says:

====================
mlx5 HW-Managed Flow Steering in FS core level

This patchset by Moshe follows Yevgeny's patchsets [1][2] on subject
"HW-Managed Flow Steering in mlx5 driver". As introduced there in HW
managed Flow Steering mode (HWS) the driver is configuring steering
rules directly to the HW using WQs with a special new type of WQE (Work
Queue Element). This way we can reach higher rule insertion/deletion
rate with much lower CPU utilization compared to SW Managed Flow
Steering (SWS).

This patchset adds API to manage namespace, flow tables, flow groups and
prepare FTE (Flow Table Entry) rules. It also adds caching and pool
mechanisms for HWS actions to allow sharing of steering actions among
different rules. The implementation of this API in FS layer, allows FS
core to use HW Managed Flow Steering in addition to the existing FW or
SW Managed Flow Steering.

Patch 13 of this series adds support for configuring HW Managed Flow
Steering mode through devlink param, similar to configuring SW Managed
Flow Steering mode:

 # devlink dev param set pci/0000:08:00.0 name flow_steering_mode \
      cmode runtime value hmfs

In addition, the series contains 2 HWS patches from Yevgeny that
implement flow update support.

[1] https://lore.kernel.org/netdev/20240903031948.78006-1-saeed@kernel.org/
[2] https://lore.kernel.org/all/20250102181415.1477316-1-tariqt@nvidia.com/
====================

Link: https://patch.msgid.link/20250109160546.1733647-1-tariqt@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

19 files changed, 2245 insertions, 284 deletions

diff --git a/Documentation/networking/devlink/mlx5.rst b/Documentation/networking/devlink/mlx5.rst
index 456985407475..41618538fc70 100644
--- a/Documentation/networking/devlink/mlx5.rst
+++ b/Documentation/networking/devlink/mlx5.rst
@@ -53,6 +53,9 @@ parameters.
        * ``smfs`` Software managed flow steering. In SMFS mode, the HW
          steering entities are created and manage through the driver without
          firmware intervention.
+       * ``hmfs`` Hardware managed flow steering. In HMFS mode, the driver
+         is configuring steering rules directly to the HW using Work Queues with
+         a special new type of WQE (Work Queue Element).
 
        SMFS mode is faster and provides better rule insertion rate compared to
        default DMFS mode.
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/Makefile b/drivers/net/ethernet/mellanox/mlx5/core/Makefile
index 10a763e668ed..d9a8817bb33c 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/Makefile
+++ b/drivers/net/ethernet/mellanox/mlx5/core/Makefile
@@ -151,8 +151,9 @@ mlx5_core-$(CONFIG_MLX5_HW_STEERING) += steering/hws/cmd.o \
 					steering/hws/bwc.o \
 					steering/hws/debug.o \
 					steering/hws/vport.o \
-					steering/hws/bwc_complex.o
-
+					steering/hws/bwc_complex.o \
+					steering/hws/fs_hws_pools.o \
+					steering/hws/fs_hws.o
 
 #
 # SF device
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c b/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c
index 41b5e98a0495..f43fd96a680d 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c
@@ -3535,35 +3535,42 @@ static int mlx5_fs_mode_validate(struct devlink *devlink, u32 id,
 {
 	struct mlx5_core_dev *dev = devlink_priv(devlink);
 	char *value = val.vstr;
-	int err = 0;
+	u8 eswitch_mode;
 
-	if (!strcmp(value, "dmfs")) {
+	if (!strcmp(value, "dmfs"))
 		return 0;
-	} else if (!strcmp(value, "smfs")) {
-		u8 eswitch_mode;
-		bool smfs_cap;
 
-		eswitch_mode = mlx5_eswitch_mode(dev);
-		smfs_cap = mlx5_fs_dr_is_supported(dev);
+	if (!strcmp(value, "smfs")) {
+		bool smfs_cap = mlx5_fs_dr_is_supported(dev);
 
 		if (!smfs_cap) {
-			err = -EOPNOTSUPP;
 			NL_SET_ERR_MSG_MOD(extack,
 					   "Software managed steering is not supported by current device");
+			return -EOPNOTSUPP;
 		}
+	} else if (!strcmp(value, "hmfs")) {
+		bool hmfs_cap = mlx5_fs_hws_is_supported(dev);
 
-		else if (eswitch_mode == MLX5_ESWITCH_OFFLOADS) {
+		if (!hmfs_cap) {
 			NL_SET_ERR_MSG_MOD(extack,
-					   "Software managed steering is not supported when eswitch offloads enabled.");
-			err = -EOPNOTSUPP;
+					   "Hardware steering is not supported by current device");
+			return -EOPNOTSUPP;
 		}
 	} else {
 		NL_SET_ERR_MSG_MOD(extack,
-				   "Bad parameter: supported values are [\"dmfs\", \"smfs\"]");
-		err = -EINVAL;
+				   "Bad parameter: supported values are [\"dmfs\", \"smfs\", \"hmfs\"]");
+		return -EINVAL;
 	}
 
-	return err;
+	eswitch_mode = mlx5_eswitch_mode(dev);
+	if (eswitch_mode == MLX5_ESWITCH_OFFLOADS) {
+		NL_SET_ERR_MSG_FMT_MOD(extack,
+				       "Moving to %s is not supported when eswitch offloads enabled.",
+				       value);
+		return -EOPNOTSUPP;
+	}
+
+	return 0;
 }
 
 static int mlx5_fs_mode_set(struct devlink *devlink, u32 id,
@@ -3575,6 +3582,8 @@ static int mlx5_fs_mode_set(struct devlink *devlink, u32 id,
 
 	if (!strcmp(ctx->val.vstr, "smfs"))
 		mode = MLX5_FLOW_STEERING_MODE_SMFS;
+	else if (!strcmp(ctx->val.vstr, "hmfs"))
+		mode = MLX5_FLOW_STEERING_MODE_HMFS;
 	else
 		mode = MLX5_FLOW_STEERING_MODE_DMFS;
 	dev->priv.steering->mode = mode;
@@ -3587,10 +3596,17 @@ static int mlx5_fs_mode_get(struct devlink *devlink, u32 id,
 {
 	struct mlx5_core_dev *dev = devlink_priv(devlink);
 
-	if (dev->priv.steering->mode == MLX5_FLOW_STEERING_MODE_SMFS)
+	switch (dev->priv.steering->mode) {
+	case MLX5_FLOW_STEERING_MODE_SMFS:
 		strscpy(ctx->val.vstr, "smfs", sizeof(ctx->val.vstr));
-	else
+		break;
+	case MLX5_FLOW_STEERING_MODE_HMFS:
+		strscpy(ctx->val.vstr, "hmfs", sizeof(ctx->val.vstr));
+		break;
+	default:
 		strscpy(ctx->val.vstr, "dmfs", sizeof(ctx->val.vstr));
+	}
+
 	return 0;
 }
 
@@ -4009,6 +4025,8 @@ int mlx5_flow_namespace_set_mode(struct mlx5_flow_namespace *ns,
 
 	if (mode == MLX5_FLOW_STEERING_MODE_SMFS)
 		cmds = mlx5_fs_cmd_get_dr_cmds();
+	else if (mode == MLX5_FLOW_STEERING_MODE_HMFS)
+		cmds = mlx5_fs_cmd_get_hws_cmds();
 	else
 		cmds = mlx5_fs_cmd_get_fw_cmds();
 	if (!cmds)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fs_core.h b/drivers/net/ethernet/mellanox/mlx5/core/fs_core.h
index bad2df0715ec..20837e526679 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/fs_core.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fs_core.h
@@ -38,6 +38,7 @@
 #include <linux/rhashtable.h>
 #include <linux/llist.h>
 #include <steering/sws/fs_dr.h>
+#include <steering/hws/fs_hws.h>
 
 #define FDB_TC_MAX_CHAIN 3
 #define FDB_FT_CHAIN (FDB_TC_MAX_CHAIN + 1)
@@ -64,6 +65,7 @@ struct mlx5_modify_hdr {
 	enum mlx5_flow_resource_owner owner;
 	union {
 		struct mlx5_fs_dr_action fs_dr_action;
+		struct mlx5_fs_hws_action fs_hws_action;
 		u32 id;
 	};
 };
@@ -74,6 +76,7 @@ struct mlx5_pkt_reformat {
 	enum mlx5_flow_resource_owner owner;
 	union {
 		struct mlx5_fs_dr_action fs_dr_action;
+		struct mlx5_fs_hws_action fs_hws_action;
 		u32 id;
 	};
 };
@@ -126,7 +129,8 @@ enum fs_fte_status {
 
 enum mlx5_flow_steering_mode {
 	MLX5_FLOW_STEERING_MODE_DMFS,
-	MLX5_FLOW_STEERING_MODE_SMFS
+	MLX5_FLOW_STEERING_MODE_SMFS,
+	MLX5_FLOW_STEERING_MODE_HMFS,
 };
 
 enum mlx5_flow_steering_capabilty {
@@ -190,7 +194,10 @@ struct mlx5_flow_handle {
 /* Type of children is mlx5_flow_group */
 struct mlx5_flow_table {
 	struct fs_node			node;
-	struct mlx5_fs_dr_table		fs_dr_table;
+	union {
+		struct mlx5_fs_dr_table		fs_dr_table;
+		struct mlx5_fs_hws_table	fs_hws_table;
+	};
 	u32				id;
 	u16				vport;
 	unsigned int			max_fte;
@@ -247,7 +254,10 @@ struct fs_fte_dup {
 /* Type of children is mlx5_flow_rule */
 struct fs_fte {
 	struct fs_node			node;
-	struct mlx5_fs_dr_rule		fs_dr_rule;
+	union {
+		struct mlx5_fs_dr_rule		fs_dr_rule;
+		struct mlx5_fs_hws_rule		fs_hws_rule;
+	};
 	u32				val[MLX5_ST_SZ_DW_MATCH_PARAM];
 	struct fs_fte_action		act_dests;
 	struct fs_fte_dup		*dup;
@@ -280,7 +290,10 @@ struct mlx5_flow_group_mask {
 /* Type of children is fs_fte */
 struct mlx5_flow_group {
 	struct fs_node			node;
-	struct mlx5_fs_dr_matcher	fs_dr_matcher;
+	union {
+		struct mlx5_fs_dr_matcher	fs_dr_matcher;
+		struct mlx5_fs_hws_matcher	fs_hws_matcher;
+	};
 	struct mlx5_flow_group_mask	mask;
 	u32				start_index;
 	u32				max_ftes;
@@ -293,7 +306,10 @@ struct mlx5_flow_group {
 struct mlx5_flow_root_namespace {
 	struct mlx5_flow_namespace	ns;
 	enum   mlx5_flow_steering_mode	mode;
-	struct mlx5_fs_dr_domain	fs_dr_domain;
+	union {
+		struct mlx5_fs_dr_domain	fs_dr_domain;
+		struct mlx5_fs_hws_context	fs_hws_context;
+	};
 	enum   fs_flow_table_type	table_type;
 	struct mlx5_core_dev		*dev;
 	struct mlx5_flow_table		*root_ft;
@@ -303,6 +319,42 @@ struct mlx5_flow_root_namespace {
 	const struct mlx5_flow_cmds	*cmds;
 };
 
+enum mlx5_fc_type {
+	MLX5_FC_TYPE_ACQUIRED = 0,
+	MLX5_FC_TYPE_LOCAL,
+};
+
+struct mlx5_fc_cache {
+	u64 packets;
+	u64 bytes;
+	u64 lastuse;
+};
+
+struct mlx5_fc {
+	u32 id;
+	bool aging;
+	enum mlx5_fc_type type;
+	struct mlx5_fc_bulk *bulk;
+	struct mlx5_fc_cache cache;
+	/* last{packets,bytes} are used for calculating deltas since last reading. */
+	u64 lastpackets;
+	u64 lastbytes;
+};
+
+struct mlx5_fc_bulk_hws_data {
+	struct mlx5hws_action *hws_action;
+	struct mutex lock; /* protects hws_action */
+	refcount_t hws_action_refcount;
+};
+
+struct mlx5_fc_bulk {
+	struct mlx5_fs_bulk fs_bulk;
+	u32 base_id;
+	struct mlx5_fc_bulk_hws_data hws_data;
+	struct mlx5_fc fcs[];
+};
+
+u32 mlx5_fc_get_base_id(struct mlx5_fc *counter);
 int mlx5_init_fc_stats(struct mlx5_core_dev *dev);
 void mlx5_cleanup_fc_stats(struct mlx5_core_dev *dev);
 void mlx5_fc_queue_stats_work(struct mlx5_core_dev *dev,
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fs_counters.c b/drivers/net/ethernet/mellanox/mlx5/core/fs_counters.c
index d8e1c4ebd364..492775d3d193 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/fs_counters.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fs_counters.c
@@ -44,28 +44,6 @@
 #define MLX5_FC_POOL_MAX_THRESHOLD BIT(18)
 #define MLX5_FC_POOL_USED_BUFF_RATIO 10
 
-enum mlx5_fc_type {
-	MLX5_FC_TYPE_ACQUIRED = 0,
-	MLX5_FC_TYPE_LOCAL,
-};
-
-struct mlx5_fc_cache {
-	u64 packets;
-	u64 bytes;
-	u64 lastuse;
-};
-
-struct mlx5_fc {
-	u32 id;
-	bool aging;
-	enum mlx5_fc_type type;
-	struct mlx5_fc_bulk *bulk;
-	struct mlx5_fc_cache cache;
-	/* last{packets,bytes} are used for calculating deltas since last reading. */
-	u64 lastpackets;
-	u64 lastbytes;
-};
-
 struct mlx5_fc_stats {
 	struct xarray counters;
 
@@ -434,13 +412,7 @@ void mlx5_fc_update_sampling_interval(struct mlx5_core_dev *dev,
 					    fc_stats->sampling_interval);
 }
 
-/* Flow counter bluks */
-
-struct mlx5_fc_bulk {
-	struct mlx5_fs_bulk fs_bulk;
-	u32 base_id;
-	struct mlx5_fc fcs[];
-};
+/* Flow counter bulks */
 
 static void mlx5_fc_init(struct mlx5_fc *counter, struct mlx5_fc_bulk *bulk,
 			 u32 id)
@@ -449,7 +421,13 @@ static void mlx5_fc_init(struct mlx5_fc *counter, struct mlx5_fc_bulk *bulk,
 	counter->id = id;
 }
 
-static struct mlx5_fs_bulk *mlx5_fc_bulk_create(struct mlx5_core_dev *dev)
+u32 mlx5_fc_get_base_id(struct mlx5_fc *counter)
+{
+	return counter->bulk->base_id;
+}
+
+static struct mlx5_fs_bulk *mlx5_fc_bulk_create(struct mlx5_core_dev *dev,
+						void *pool_ctx)
 {
 	enum mlx5_fc_bulk_alloc_bitmask alloc_bitmask;
 	struct mlx5_fc_bulk *fc_bulk;
@@ -473,6 +451,8 @@ static struct mlx5_fs_bulk *mlx5_fc_bulk_create(struct mlx5_core_dev *dev)
 	for (i = 0; i < bulk_len; i++)
 		mlx5_fc_init(&fc_bulk->fcs[i], fc_bulk, base_id + i);
 
+	refcount_set(&fc_bulk->hws_data.hws_action_refcount, 0);
+	mutex_init(&fc_bulk->hws_data.lock);
 	return &fc_bulk->fs_bulk;
 
 fs_bulk_cleanup:
@@ -518,7 +498,7 @@ static const struct mlx5_fs_pool_ops mlx5_fc_pool_ops = {
 static void
 mlx5_fc_pool_init(struct mlx5_fs_pool *fc_pool, struct mlx5_core_dev *dev)
 {
-	mlx5_fs_pool_init(fc_pool, dev, &mlx5_fc_pool_ops);
+	mlx5_fs_pool_init(fc_pool, dev, &mlx5_fc_pool_ops, NULL);
 }
 
 static void mlx5_fc_pool_cleanup(struct mlx5_fs_pool *fc_pool)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fs_pool.c b/drivers/net/ethernet/mellanox/mlx5/core/fs_pool.c
index b891d7b9e3e0..f6c226664602 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/fs_pool.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fs_pool.c
@@ -56,11 +56,12 @@ static int mlx5_fs_bulk_release_index(struct mlx5_fs_bulk *fs_bulk, int index)
 }
 
 void mlx5_fs_pool_init(struct mlx5_fs_pool *pool, struct mlx5_core_dev *dev,
-		       const struct mlx5_fs_pool_ops *ops)
+		       const struct mlx5_fs_pool_ops *ops, void *pool_ctx)
 {
 	WARN_ON_ONCE(!ops || !ops->bulk_destroy || !ops->bulk_create ||
 		     !ops->update_threshold);
 	pool->dev = dev;
+	pool->pool_ctx = pool_ctx;
 	mutex_init(&pool->pool_lock);
 	INIT_LIST_HEAD(&pool->fully_used);
 	INIT_LIST_HEAD(&pool->partially_used);
@@ -91,7 +92,7 @@ mlx5_fs_pool_alloc_new_bulk(struct mlx5_fs_pool *fs_pool)
 	struct mlx5_core_dev *dev = fs_pool->dev;
 	struct mlx5_fs_bulk *new_bulk;
 
-	new_bulk = fs_pool->ops->bulk_create(dev);
+	new_bulk = fs_pool->ops->bulk_create(dev, fs_pool->pool_ctx);
 	if (new_bulk)
 		fs_pool->available_units += new_bulk->bulk_len;
 	fs_pool->ops->update_threshold(fs_pool);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fs_pool.h b/drivers/net/ethernet/mellanox/mlx5/core/fs_pool.h
index 3b149863260c..f04ec3107498 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/fs_pool.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fs_pool.h
@@ -21,7 +21,8 @@ struct mlx5_fs_pool;
 
 struct mlx5_fs_pool_ops {
 	int (*bulk_destroy)(struct mlx5_core_dev *dev, struct mlx5_fs_bulk *bulk);
-	struct mlx5_fs_bulk * (*bulk_create)(struct mlx5_core_dev *dev);
+	struct mlx5_fs_bulk * (*bulk_create)(struct mlx5_core_dev *dev,
+					     void *pool_ctx);
 	void (*update_threshold)(struct mlx5_fs_pool *pool);
 };
 
@@ -44,7 +45,7 @@ void mlx5_fs_bulk_cleanup(struct mlx5_fs_bulk *fs_bulk);
 int mlx5_fs_bulk_get_free_amount(struct mlx5_fs_bulk *bulk);
 
 void mlx5_fs_pool_init(struct mlx5_fs_pool *pool, struct mlx5_core_dev *dev,
-		       const struct mlx5_fs_pool_ops *ops);
+		       const struct mlx5_fs_pool_ops *ops, void *pool_ctx);
 void mlx5_fs_pool_cleanup(struct mlx5_fs_pool *pool);
 int mlx5_fs_pool_acquire_index(struct mlx5_fs_pool *fs_pool,
 			       struct mlx5_fs_pool_index *pool_index);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/debug.c b/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/debug.c
index 60ada3143d60..696275fd0ce2 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/debug.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/debug.c
@@ -148,8 +148,8 @@ static int hws_debug_dump_matcher(struct seq_file *f, struct mlx5hws_matcher *ma
 		   matcher->match_ste.rtc_1_id,
 		   (int)ste_1_id);
 
-	ste = &matcher->action_ste[0].ste;
-	ste_pool = matcher->action_ste[0].pool;
+	ste = &matcher->action_ste.ste;
+	ste_pool = matcher->action_ste.pool;
 	if (ste_pool) {
 		ste_0_id = mlx5hws_pool_chunk_get_base_id(ste_pool, ste);
 		if (tbl_type == MLX5HWS_TABLE_TYPE_FDB)
@@ -171,10 +171,8 @@ static int hws_debug_dump_matcher(struct seq_file *f, struct mlx5hws_matcher *ma
 		return ret;
 
 	seq_printf(f, ",%d,%d,%d,%d,%d,0x%llx,0x%llx\n",
-		   matcher->action_ste[0].rtc_0_id,
-		   (int)ste_0_id,
-		   matcher->action_ste[0].rtc_1_id,
-		   (int)ste_1_id,
+		   matcher->action_ste.rtc_0_id, (int)ste_0_id,
+		   matcher->action_ste.rtc_1_id, (int)ste_1_id,
 		   0,
 		   mlx5hws_debug_icm_to_idx(icm_addr_0),
 		   mlx5hws_debug_icm_to_idx(icm_addr_1));
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/fs_hws.c b/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/fs_hws.c
new file mode 100644
index 000000000000..05329afeb9ea
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/fs_hws.c
@@ -0,0 +1,1377 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2025 NVIDIA Corporation & Affiliates */
+
+#include <linux/mlx5/vport.h>
+#include <mlx5_core.h>
+#include <fs_core.h>
+#include <fs_cmd.h>
+#include "fs_hws_pools.h"
+#include "mlx5hws.h"
+
+#define MLX5HWS_CTX_MAX_NUM_OF_QUEUES 16
+#define MLX5HWS_CTX_QUEUE_SIZE 256
+
+static struct mlx5hws_action *
+mlx5_fs_create_action_remove_header_vlan(struct mlx5hws_context *ctx);
+static void
+mlx5_fs_destroy_pr_pool(struct mlx5_fs_pool *pool, struct xarray *pr_pools,
+			unsigned long index);
+static void
+mlx5_fs_destroy_mh_pool(struct mlx5_fs_pool *pool, struct xarray *mh_pools,
+			unsigned long index);
+
+static int mlx5_fs_init_hws_actions_pool(struct mlx5_core_dev *dev,
+					 struct mlx5_fs_hws_context *fs_ctx)
+{
+	u32 flags = MLX5HWS_ACTION_FLAG_HWS_FDB | MLX5HWS_ACTION_FLAG_SHARED;
+	struct mlx5_fs_hws_actions_pool *hws_pool = &fs_ctx->hws_pool;
+	struct mlx5hws_action_reformat_header reformat_hdr = {};
+	struct mlx5hws_context *ctx = fs_ctx->hws_ctx;
+	enum mlx5hws_action_type action_type;
+	int err = -ENOSPC;
+
+	hws_pool->tag_action = mlx5hws_action_create_tag(ctx, flags);
+	if (!hws_pool->tag_action)
+		return err;
+	hws_pool->pop_vlan_action = mlx5hws_action_create_pop_vlan(ctx, flags);
+	if (!hws_pool->pop_vlan_action)
+		goto destroy_tag;
+	hws_pool->push_vlan_action = mlx5hws_action_create_push_vlan(ctx, flags);
+	if (!hws_pool->push_vlan_action)
+		goto destroy_pop_vlan;
+	hws_pool->drop_action = mlx5hws_action_create_dest_drop(ctx, flags);
+	if (!hws_pool->drop_action)
+		goto destroy_push_vlan;
+	action_type = MLX5HWS_ACTION_TYP_REFORMAT_TNL_L2_TO_L2;
+	hws_pool->decapl2_action =
+		mlx5hws_action_create_reformat(ctx, action_type, 1,
+					       &reformat_hdr, 0, flags);
+	if (!hws_pool->decapl2_action)
+		goto destroy_drop;
+	hws_pool->remove_hdr_vlan_action =
+		mlx5_fs_create_action_remove_header_vlan(ctx);
+	if (!hws_pool->remove_hdr_vlan_action)
+		goto destroy_decapl2;
+	err = mlx5_fs_hws_pr_pool_init(&hws_pool->insert_hdr_pool, dev, 0,
+				       MLX5HWS_ACTION_TYP_INSERT_HEADER);
+	if (err)
+		goto destroy_remove_hdr;
+	err = mlx5_fs_hws_pr_pool_init(&hws_pool->dl3tnltol2_pool, dev, 0,
+				       MLX5HWS_ACTION_TYP_REFORMAT_TNL_L3_TO_L2);
+	if (err)
+		goto cleanup_insert_hdr;
+	xa_init(&hws_pool->el2tol3tnl_pools);
+	xa_init(&hws_pool->el2tol2tnl_pools);
+	xa_init(&hws_pool->mh_pools);
+	xa_init(&hws_pool->table_dests);
+	xa_init(&hws_pool->vport_dests);
+	xa_init(&hws_pool->vport_vhca_dests);
+	return 0;
+
+cleanup_insert_hdr:
+	mlx5_fs_hws_pr_pool_cleanup(&hws_pool->insert_hdr_pool);
+destroy_remove_hdr:
+	mlx5hws_action_destroy(hws_pool->remove_hdr_vlan_action);
+destroy_decapl2:
+	mlx5hws_action_destroy(hws_pool->decapl2_action);
+destroy_drop:
+	mlx5hws_action_destroy(hws_pool->drop_action);
+destroy_push_vlan:
+	mlx5hws_action_destroy(hws_pool->push_vlan_action);
+destroy_pop_vlan:
+	mlx5hws_action_destroy(hws_pool->pop_vlan_action);
+destroy_tag:
+	mlx5hws_action_destroy(hws_pool->tag_action);
+	return err;
+}
+
+static void mlx5_fs_cleanup_hws_actions_pool(struct mlx5_fs_hws_context *fs_ctx)
+{
+	struct mlx5_fs_hws_actions_pool *hws_pool = &fs_ctx->hws_pool;
+	struct mlx5hws_action *action;
+	struct mlx5_fs_pool *pool;
+	unsigned long i;
+
+	xa_for_each(&hws_pool->vport_vhca_dests, i, action)
+		mlx5hws_action_destroy(action);
+	xa_destroy(&hws_pool->vport_vhca_dests);
+	xa_for_each(&hws_pool->vport_dests, i, action)
+		mlx5hws_action_destroy(action);
+	xa_destroy(&hws_pool->vport_dests);
+	xa_destroy(&hws_pool->table_dests);
+	xa_for_each(&hws_pool->mh_pools, i, pool)
+		mlx5_fs_destroy_mh_pool(pool, &hws_pool->mh_pools, i);
+	xa_destroy(&hws_pool->mh_pools);
+	xa_for_each(&hws_pool->el2tol2tnl_pools, i, pool)
+		mlx5_fs_destroy_pr_pool(pool, &hws_pool->el2tol2tnl_pools, i);
+	xa_destroy(&hws_pool->el2tol2tnl_pools);
+	xa_for_each(&hws_pool->el2tol3tnl_pools, i, pool)
+		mlx5_fs_destroy_pr_pool(pool, &hws_pool->el2tol3tnl_pools, i);
+	xa_destroy(&hws_pool->el2tol3tnl_pools);
+	mlx5_fs_hws_pr_pool_cleanup(&hws_pool->dl3tnltol2_pool);
+	mlx5_fs_hws_pr_pool_cleanup(&hws_pool->insert_hdr_pool);
+	mlx5hws_action_destroy(hws_pool->remove_hdr_vlan_action);
+	mlx5hws_action_destroy(hws_pool->decapl2_action);
+	mlx5hws_action_destroy(hws_pool->drop_action);
+	mlx5hws_action_destroy(hws_pool->push_vlan_action);
+	mlx5hws_action_destroy(hws_pool->pop_vlan_action);
+	mlx5hws_action_destroy(hws_pool->tag_action);
+}
+
+static int mlx5_cmd_hws_create_ns(struct mlx5_flow_root_namespace *ns)
+{
+	struct mlx5hws_context_attr hws_ctx_attr = {};
+	int err;
+
+	hws_ctx_attr.queues = min_t(int, num_online_cpus(),
+				    MLX5HWS_CTX_MAX_NUM_OF_QUEUES);
+	hws_ctx_attr.queue_size = MLX5HWS_CTX_QUEUE_SIZE;
+
+	ns->fs_hws_context.hws_ctx =
+		mlx5hws_context_open(ns->dev, &hws_ctx_attr);
+	if (!ns->fs_hws_context.hws_ctx) {
+		mlx5_core_err(ns->dev, "Failed to create hws flow namespace\n");
+		return -EINVAL;
+	}
+	err = mlx5_fs_init_hws_actions_pool(ns->dev, &ns->fs_hws_context);
+	if (err) {
+		mlx5_core_err(ns->dev, "Failed to init hws actions pool\n");
+		mlx5hws_context_close(ns->fs_hws_context.hws_ctx);
+		return err;
+	}
+	return 0;
+}
+
+static int mlx5_cmd_hws_destroy_ns(struct mlx5_flow_root_namespace *ns)
+{
+	mlx5_fs_cleanup_hws_actions_pool(&ns->fs_hws_context);
+	return mlx5hws_context_close(ns->fs_hws_context.hws_ctx);
+}
+
+static int mlx5_cmd_hws_set_peer(struct mlx5_flow_root_namespace *ns,
+				 struct mlx5_flow_root_namespace *peer_ns,
+				 u16 peer_vhca_id)
+{
+	struct mlx5hws_context *peer_ctx = NULL;
+
+	if (peer_ns)
+		peer_ctx = peer_ns->fs_hws_context.hws_ctx;
+	mlx5hws_context_set_peer(ns->fs_hws_context.hws_ctx, peer_ctx,
+				 peer_vhca_id);
+	return 0;
+}
+
+static int mlx5_fs_set_ft_default_miss(struct mlx5_flow_root_namespace *ns,
+				       struct mlx5_flow_table *ft,
+				       struct mlx5_flow_table *next_ft)
+{
+	struct mlx5hws_table *next_tbl;
+	int err;
+
+	if (!ns->fs_hws_context.hws_ctx)
+		return -EINVAL;
+
+	/* if no change required, return */
+	if (!next_ft && !ft->fs_hws_table.miss_ft_set)
+		return 0;
+
+	next_tbl = next_ft ? next_ft->fs_hws_table.hws_table : NULL;
+	err = mlx5hws_table_set_default_miss(ft->fs_hws_table.hws_table, next_tbl);
+	if (err) {
+		mlx5_core_err(ns->dev, "Failed setting FT default miss (%d)\n", err);
+		return err;
+	}
+	ft->fs_hws_table.miss_ft_set = !!next_tbl;
+	return 0;
+}
+
+static int mlx5_fs_add_flow_table_dest_action(struct mlx5_flow_root_namespace *ns,
+					      struct mlx5_flow_table *ft)
+{
+	u32 flags = MLX5HWS_ACTION_FLAG_HWS_FDB | MLX5HWS_ACTION_FLAG_SHARED;
+	struct mlx5_fs_hws_context *fs_ctx = &ns->fs_hws_context;
+	struct mlx5hws_action *dest_ft_action;
+	struct xarray *dests_xa;
+	int err;
+
+	dest_ft_action = mlx5hws_action_create_dest_table_num(fs_ctx->hws_ctx,
+							      ft->id, flags);
+	if (!dest_ft_action) {
+		mlx5_core_err(ns->dev, "Failed creating dest table action\n");
+		return -ENOMEM;
+	}
+
+	dests_xa = &fs_ctx->hws_pool.table_dests;
+	err = xa_insert(dests_xa, ft->id, dest_ft_action, GFP_KERNEL);
+	if (err)
+		mlx5hws_action_destroy(dest_ft_action);
+	return err;
+}
+
+static int mlx5_fs_del_flow_table_dest_action(struct mlx5_flow_root_namespace *ns,
+					      struct mlx5_flow_table *ft)
+{
+	struct mlx5_fs_hws_context *fs_ctx = &ns->fs_hws_context;
+	struct mlx5hws_action *dest_ft_action;
+	struct xarray *dests_xa;
+	int err;
+
+	dests_xa = &fs_ctx->hws_pool.table_dests;
+	dest_ft_action = xa_erase(dests_xa, ft->id);
+	if (!dest_ft_action) {
+		mlx5_core_err(ns->dev, "Failed to erase dest ft action\n");
+		return -ENOENT;
+	}
+
+	err = mlx5hws_action_destroy(dest_ft_action);
+	if (err)
+		mlx5_core_err(ns->dev, "Failed to destroy dest ft action\n");
+	return err;
+}
+
+static int mlx5_cmd_hws_create_flow_table(struct mlx5_flow_root_namespace *ns,
+					  struct mlx5_flow_table *ft,
+					  struct mlx5_flow_table_attr *ft_attr,
+					  struct mlx5_flow_table *next_ft)
+{
+	struct mlx5hws_context *ctx = ns->fs_hws_context.hws_ctx;
+	struct mlx5hws_table_attr tbl_attr = {};
+	struct mlx5hws_table *tbl;
+	int err;
+
+	if (mlx5_fs_cmd_is_fw_term_table(ft)) {
+		err = mlx5_fs_cmd_get_fw_cmds()->create_flow_table(ns, ft, ft_attr,
+								   next_ft);
+		if (err)
+			return err;
+		err = mlx5_fs_add_flow_table_dest_action(ns, ft);
+		if (err)
+			mlx5_fs_cmd_get_fw_cmds()->destroy_flow_table(ns, ft);
+		return err;
+	}
+
+	if (ns->table_type != FS_FT_FDB) {
+		mlx5_core_err(ns->dev, "Table type %d not supported for HWS\n",
+			      ns->table_type);
+		return -EOPNOTSUPP;
+	}
+
+	tbl_attr.type = MLX5HWS_TABLE_TYPE_FDB;
+	tbl_attr.level = ft_attr->level;
+	tbl = mlx5hws_table_create(ctx, &tbl_attr);
+	if (!tbl) {
+		mlx5_core_err(ns->dev, "Failed creating hws flow_table\n");
+		return -EINVAL;
+	}
+
+	ft->fs_hws_table.hws_table = tbl;
+	ft->id = mlx5hws_table_get_id(tbl);
+
+	if (next_ft) {
+		err = mlx5_fs_set_ft_default_miss(ns, ft, next_ft);
+		if (err)
+			goto destroy_table;
+	}
+
+	ft->max_fte = INT_MAX;
+
+	err = mlx5_fs_add_flow_table_dest_action(ns, ft);
+	if (err)
+		goto clear_ft_miss;
+	return 0;
+
+clear_ft_miss:
+	mlx5_fs_set_ft_default_miss(ns, ft, NULL);
+destroy_table:
+	mlx5hws_table_destroy(tbl);
+	ft->fs_hws_table.hws_table = NULL;
+	return err;
+}
+
+static int mlx5_cmd_hws_destroy_flow_table(struct mlx5_flow_root_namespace *ns,
+					   struct mlx5_flow_table *ft)
+{
+	int err;
+
+	err = mlx5_fs_del_flow_table_dest_action(ns, ft);
+	if (err)
+		mlx5_core_err(ns->dev, "Failed to remove dest action (%d)\n", err);
+
+	if (mlx5_fs_cmd_is_fw_term_table(ft))
+		return mlx5_fs_cmd_get_fw_cmds()->destroy_flow_table(ns, ft);
+
+	err = mlx5_fs_set_ft_default_miss(ns, ft, NULL);
+	if (err)
+		mlx5_core_err(ns->dev, "Failed to disconnect next table (%d)\n", err);
+
+	err = mlx5hws_table_destroy(ft->fs_hws_table.hws_table);
+	if (err)
+		mlx5_core_err(ns->dev, "Failed to destroy flow_table (%d)\n", err);
+
+	return err;
+}
+
+static int mlx5_cmd_hws_modify_flow_table(struct mlx5_flow_root_namespace *ns,
+					  struct mlx5_flow_table *ft,
+					  struct mlx5_flow_table *next_ft)
+{
+	if (mlx5_fs_cmd_is_fw_term_table(ft))
+		return mlx5_fs_cmd_get_fw_cmds()->modify_flow_table(ns, ft, next_ft);
+
+	return mlx5_fs_set_ft_default_miss(ns, ft, next_ft);
+}
+
+static int mlx5_cmd_hws_update_root_ft(struct mlx5_flow_root_namespace *ns,
+				       struct mlx5_flow_table *ft,
+				       u32 underlay_qpn,
+				       bool disconnect)
+{
+	return mlx5_fs_cmd_get_fw_cmds()->update_root_ft(ns, ft, underlay_qpn,
+							 disconnect);
+}
+
+static int mlx5_cmd_hws_create_flow_group(struct mlx5_flow_root_namespace *ns,
+					  struct mlx5_flow_table *ft, u32 *in,
+					  struct mlx5_flow_group *fg)
+{
+	struct mlx5hws_match_parameters mask;
+	struct mlx5hws_bwc_matcher *matcher;
+	u8 match_criteria_enable;
+	u32 priority;
+
+	if (mlx5_fs_cmd_is_fw_term_table(ft))
+		return mlx5_fs_cmd_get_fw_cmds()->create_flow_group(ns, ft, in, fg);
+
+	mask.match_buf = MLX5_ADDR_OF(create_flow_group_in, in, match_criteria);
+	mask.match_sz = sizeof(fg->mask.match_criteria);
+
+	match_criteria_enable = MLX5_GET(create_flow_group_in, in,
+					 match_criteria_enable);
+	priority = MLX5_GET(create_flow_group_in, in, start_flow_index);
+	matcher = mlx5hws_bwc_matcher_create(ft->fs_hws_table.hws_table,
+					     priority, match_criteria_enable,
+					     &mask);
+	if (!matcher) {
+		mlx5_core_err(ns->dev, "Failed creating matcher\n");
+		return -EINVAL;
+	}
+
+	fg->fs_hws_matcher.matcher = matcher;
+	return 0;
+}
+
+static int mlx5_cmd_hws_destroy_flow_group(struct mlx5_flow_root_namespace *ns,
+					   struct mlx5_flow_table *ft,
+					   struct mlx5_flow_group *fg)
+{
+	if (mlx5_fs_cmd_is_fw_term_table(ft))
+		return mlx5_fs_cmd_get_fw_cmds()->destroy_flow_group(ns, ft, fg);
+
+	return mlx5hws_bwc_matcher_destroy(fg->fs_hws_matcher.matcher);
+}
+
+static struct mlx5hws_action *
+mlx5_fs_get_dest_action_ft(struct mlx5_fs_hws_context *fs_ctx,
+			   struct mlx5_flow_rule *dst)
+{
+	return xa_load(&fs_ctx->hws_pool.table_dests, dst->dest_attr.ft->id);
+}
+
+static struct mlx5hws_action *
+mlx5_fs_get_dest_action_table_num(struct mlx5_fs_hws_context *fs_ctx,
+				  struct mlx5_flow_rule *dst)
+{
+	u32 table_num = dst->dest_attr.ft_num;
+
+	return xa_load(&fs_ctx->hws_pool.table_dests, table_num);
+}
+
+static struct mlx5hws_action *
+mlx5_fs_create_dest_action_table_num(struct mlx5_fs_hws_context *fs_ctx,
+				     struct mlx5_flow_rule *dst)
+{
+	u32 flags = MLX5HWS_ACTION_FLAG_HWS_FDB | MLX5HWS_ACTION_FLAG_SHARED;
+	struct mlx5hws_context *ctx = fs_ctx->hws_ctx;
+	u32 table_num = dst->dest_attr.ft_num;
+
+	return mlx5hws_action_create_dest_table_num(ctx, table_num, flags);
+}
+
+static struct mlx5hws_action *
+mlx5_fs_get_dest_action_vport(struct mlx5_fs_hws_context *fs_ctx,
+			      struct mlx5_flow_rule *dst,
+			      bool is_dest_type_uplink)
+{
+	u32 flags = MLX5HWS_ACTION_FLAG_HWS_FDB | MLX5HWS_ACTION_FLAG_SHARED;
+	struct mlx5_flow_destination *dest_attr = &dst->dest_attr;
+	struct mlx5hws_context *ctx = fs_ctx->hws_ctx;
+	struct mlx5hws_action *dest;
+	struct xarray *dests_xa;
+	bool vhca_id_valid;
+	unsigned long idx;
+	u16 vport_num;
+	int err;
+
+	vhca_id_valid = is_dest_type_uplink ||
+			(dest_attr->vport.flags & MLX5_FLOW_DEST_VPORT_VHCA_ID);
+	vport_num = is_dest_type_uplink ? MLX5_VPORT_UPLINK : dest_attr->vport.num;
+	if (vhca_id_valid) {
+		dests_xa = &fs_ctx->hws_pool.vport_vhca_dests;
+		idx = dest_attr->vport.vhca_id << 16 | vport_num;
+	} else {
+		dests_xa = &fs_ctx->hws_pool.vport_dests;
+		idx = vport_num;
+	}
+dest_load:
+	dest = xa_load(dests_xa, idx);
+	if (dest)
+		return dest;
+
+	dest = mlx5hws_action_create_dest_vport(ctx, vport_num,	vhca_id_valid,
+						dest_attr->vport.vhca_id, flags);
+
+	err = xa_insert(dests_xa, idx, dest, GFP_KERNEL);
+	if (err) {
+		mlx5hws_action_destroy(dest);
+		dest = NULL;
+
+		if (err == -EBUSY)
+			/* xarray entry was already stored by another thread */
+			goto dest_load;
+	}
+
+	return dest;
+}
+
+static struct mlx5hws_action *
+mlx5_fs_create_dest_action_range(struct mlx5hws_context *ctx,
+				 struct mlx5_flow_rule *dst)
+{
+	u32 flags = MLX5HWS_ACTION_FLAG_HWS_FDB | MLX5HWS_ACTION_FLAG_SHARED;
+	struct mlx5_flow_destination *dest_attr = &dst->dest_attr;
+
+	return mlx5hws_action_create_dest_match_range(ctx,
+						      dest_attr->range.field,
+						      dest_attr->range.hit_ft,
+						      dest_attr->range.miss_ft,
+						      dest_attr->range.min,
+						      dest_attr->range.max,
+						      flags);
+}
+
+static struct mlx5hws_action *
+mlx5_fs_create_action_dest_array(struct mlx5hws_context *ctx,
+				 struct mlx5hws_action_dest_attr *dests,
+				 u32 num_of_dests, bool ignore_flow_level,
+				 u32 flow_source)
+{
+	u32 flags = MLX5HWS_ACTION_FLAG_HWS_FDB | MLX5HWS_ACTION_FLAG_SHARED;
+
+	return mlx5hws_action_create_dest_array(ctx, num_of_dests, dests,
+						ignore_flow_level,
+						flow_source, flags);
+}
+
+static struct mlx5hws_action *
+mlx5_fs_get_action_push_vlan(struct mlx5_fs_hws_context *fs_ctx)
+{
+	return fs_ctx->hws_pool.push_vlan_action;
+}
+
+static u32 mlx5_fs_calc_vlan_hdr(struct mlx5_fs_vlan *vlan)
+{
+	u16 n_ethtype = vlan->ethtype;
+	u8 prio = vlan->prio;
+	u16 vid = vlan->vid;
+
+	return (u32)n_ethtype << 16 | (u32)(prio) << 12 | (u32)vid;
+}
+
+static struct mlx5hws_action *
+mlx5_fs_get_action_pop_vlan(struct mlx5_fs_hws_context *fs_ctx)
+{
+	return fs_ctx->hws_pool.pop_vlan_action;
+}
+
+static struct mlx5hws_action *
+mlx5_fs_get_action_decap_tnl_l2_to_l2(struct mlx5_fs_hws_context *fs_ctx)
+{
+	return fs_ctx->hws_pool.decapl2_action;
+}
+
+static struct mlx5hws_action *
+mlx5_fs_get_dest_action_drop(struct mlx5_fs_hws_context *fs_ctx)
+{
+	return fs_ctx->hws_pool.drop_action;
+}
+
+static struct mlx5hws_action *
+mlx5_fs_get_action_tag(struct mlx5_fs_hws_context *fs_ctx)
+{
+	return fs_ctx->hws_pool.tag_action;
+}
+
+static struct mlx5hws_action *
+mlx5_fs_create_action_last(struct mlx5hws_context *ctx)
+{
+	u32 flags = MLX5HWS_ACTION_FLAG_HWS_FDB | MLX5HWS_ACTION_FLAG_SHARED;
+
+	return mlx5hws_action_create_last(ctx, flags);
+}
+
+static void mlx5_fs_destroy_fs_action(struct mlx5_fs_hws_rule_action *fs_action)
+{
+	switch (mlx5hws_action_get_type(fs_action->action)) {
+	case MLX5HWS_ACTION_TYP_CTR:
+		mlx5_fc_put_hws_action(fs_action->counter);
+		break;
+	default:
+		mlx5hws_action_destroy(fs_action->action);
+	}
+}
+
+static void
+mlx5_fs_destroy_fs_actions(struct mlx5_fs_hws_rule_action **fs_actions,
+			   int *num_fs_actions)
+{
+	int i;
+
+	/* Free in reverse order to handle action dependencies */
+	for (i = *num_fs_actions - 1; i >= 0; i--)
+		mlx5_fs_destroy_fs_action(*fs_actions + i);
+	*num_fs_actions = 0;
+	kfree(*fs_actions);
+	*fs_actions = NULL;
+}
+
+/* Splits FTE's actions into cached, rule and destination actions.
+ * The cached and destination actions are saved on the fte hws rule.
+ * The rule actions are returned as a parameter, together with their count.
+ * We want to support a rule with 32 destinations, which means we need to
+ * account for 32 destinations plus usually a counter plus one more action
+ * for a multi-destination flow table.
+ * 32 is SW limitation for array size, keep. HWS limitation is 16M STEs per matcher
+ */
+#define MLX5_FLOW_CONTEXT_ACTION_MAX 34
+static int mlx5_fs_fte_get_hws_actions(struct mlx5_flow_root_namespace *ns,
+				       struct mlx5_flow_table *ft,
+				       struct mlx5_flow_group *group,
+				       struct fs_fte *fte,
+				       struct mlx5hws_rule_action **ractions)
+{
+	struct mlx5_flow_act *fte_action = &fte->act_dests.action;
+	struct mlx5_fs_hws_context *fs_ctx = &ns->fs_hws_context;
+	struct mlx5hws_action_dest_attr *dest_actions;
+	struct mlx5hws_context *ctx = fs_ctx->hws_ctx;
+	struct mlx5_fs_hws_rule_action *fs_actions;
+	struct mlx5_core_dev *dev = ns->dev;
+	struct mlx5hws_action *dest_action;
+	struct mlx5hws_action *tmp_action;
+	struct mlx5_fs_hws_pr *pr_data;
+	struct mlx5_fs_hws_mh *mh_data;
+	bool delay_encap_set = false;
+	struct mlx5_flow_rule *dst;
+	int num_dest_actions = 0;
+	int num_fs_actions = 0;
+	int num_actions = 0;
+	int err;
+
+	*ractions = kcalloc(MLX5_FLOW_CONTEXT_ACTION_MAX, sizeof(**ractions),
+			    GFP_KERNEL);
+	if (!*ractions) {
+		err = -ENOMEM;
+		goto out_err;
+	}
+
+	fs_actions = kcalloc(MLX5_FLOW_CONTEXT_ACTION_MAX,
+			     sizeof(*fs_actions), GFP_KERNEL);
+	if (!fs_actions) {
+		err = -ENOMEM;
+		goto free_actions_alloc;
+	}
+
+	dest_actions = kcalloc(MLX5_FLOW_CONTEXT_ACTION_MAX,
+			       sizeof(*dest_actions), GFP_KERNEL);
+	if (!dest_actions) {
+		err = -ENOMEM;
+		goto free_fs_actions_alloc;
+	}
+
+	/* The order of the actions are must to be kept, only the following
+	 * order is supported by HW steering:
+	 * HWS: decap -> remove_hdr -> pop_vlan -> modify header -> push_vlan
+	 *      -> reformat (insert_hdr/encap) -> ctr -> tag -> aso
+	 *      -> drop -> FWD:tbl/vport/sampler/tbl_num/range -> dest_array -> last
+	 */
+	if (fte_action->action & MLX5_FLOW_CONTEXT_ACTION_DECAP) {
+		tmp_action = mlx5_fs_get_action_decap_tnl_l2_to_l2(fs_ctx);
+		if (!tmp_action) {
+			err = -ENOMEM;
+			goto free_dest_actions_alloc;
+		}
+		(*ractions)[num_actions++].action = tmp_action;
+	}
+
+	if (fte_action->action & MLX5_FLOW_CONTEXT_ACTION_PACKET_REFORMAT) {
+		int reformat_type = fte_action->pkt_reformat->reformat_type;
+
+		if (fte_action->pkt_reformat->owner == MLX5_FLOW_RESOURCE_OWNER_FW) {
+			mlx5_core_err(dev, "FW-owned reformat can't be used in HWS rule\n");
+			err = -EINVAL;
+			goto free_actions;
+		}
+
+		if (reformat_type == MLX5_REFORMAT_TYPE_L3_TUNNEL_TO_L2) {
+			pr_data = fte_action->pkt_reformat->fs_hws_action.pr_data;
+			(*ractions)[num_actions].reformat.offset = pr_data->offset;
+			(*ractions)[num_actions].reformat.hdr_idx = pr_data->hdr_idx;
+			(*ractions)[num_actions].reformat.data = pr_data->data;
+			(*ractions)[num_actions++].action =
+				fte_action->pkt_reformat->fs_hws_action.hws_action;
+		} else if (reformat_type == MLX5_REFORMAT_TYPE_REMOVE_HDR) {
+			(*ractions)[num_actions++].action =
+				fte_action->pkt_reformat->fs_hws_action.hws_action;
+		} else {
+			delay_encap_set = true;
+		}
+	}
+
+	if (fte_action->action & MLX5_FLOW_CONTEXT_ACTION_VLAN_POP) {
+		tmp_action = mlx5_fs_get_action_pop_vlan(fs_ctx);
+		if (!tmp_action) {
+			err = -ENOMEM;
+			goto free_actions;
+		}
+		(*ractions)[num_actions++].action = tmp_action;
+	}
+
+	if (fte_action->action & MLX5_FLOW_CONTEXT_ACTION_VLAN_POP_2) {
+		tmp_action = mlx5_fs_get_action_pop_vlan(fs_ctx);
+		if (!tmp_action) {
+			err = -ENOMEM;
+			goto free_actions;
+		}
+		(*ractions)[num_actions++].action = tmp_action;
+	}
+
+	if (fte_action->action & MLX5_FLOW_CONTEXT_ACTION_MOD_HDR) {
+		mh_data = fte_action->modify_hdr->fs_hws_action.mh_data;
+		(*ractions)[num_actions].modify_header.offset = mh_data->offset;
+		(*ractions)[num_actions].modify_header.data = mh_data->data;
+		(*ractions)[num_actions++].action =
+			fte_action->modify_hdr->fs_hws_action.hws_action;
+	}
+
+	if (fte_action->action & MLX5_FLOW_CONTEXT_ACTION_VLAN_PUSH) {
+		tmp_action = mlx5_fs_get_action_push_vlan(fs_ctx);
+		if (!tmp_action) {
+			err = -ENOMEM;
+			goto free_actions;
+		}
+		(*ractions)[num_actions].push_vlan.vlan_hdr =
+			htonl(mlx5_fs_calc_vlan_hdr(&fte_action->vlan[0]));
+		(*ractions)[num_actions++].action = tmp_action;
+	}
+
+	if (fte_action->action & MLX5_FLOW_CONTEXT_ACTION_VLAN_PUSH_2) {
+		tmp_action = mlx5_fs_get_action_push_vlan(fs_ctx);
+		if (!tmp_action) {
+			err = -ENOMEM;
+			goto free_actions;
+		}
+		(*ractions)[num_actions].push_vlan.vlan_hdr =
+			htonl(mlx5_fs_calc_vlan_hdr(&fte_action->vlan[1]));
+		(*ractions)[num_actions++].action = tmp_action;
+	}
+
+	if (delay_encap_set) {
+		pr_data = fte_action->pkt_reformat->fs_hws_action.pr_data;
+		(*ractions)[num_actions].reformat.offset = pr_data->offset;
+		(*ractions)[num_actions].reformat.data = pr_data->data;
+		(*ractions)[num_actions++].action =
+			fte_action->pkt_reformat->fs_hws_action.hws_action;
+	}
+
+	if (fte_action->action & MLX5_FLOW_CONTEXT_ACTION_COUNT) {
+		list_for_each_entry(dst, &fte->node.children, node.list) {
+			struct mlx5_fc *counter;
+
+			if (dst->dest_attr.type !=
+			    MLX5_FLOW_DESTINATION_TYPE_COUNTER)
+				continue;
+
+			if (num_actions == MLX5_FLOW_CONTEXT_ACTION_MAX) {
+				err = -EOPNOTSUPP;
+				goto free_actions;
+			}
+
+			counter = dst->dest_attr.counter;
+			tmp_action = mlx5_fc_get_hws_action(ctx, counter);
+			if (!tmp_action) {
+				err = -EINVAL;
+				goto free_actions;
+			}
+
+			(*ractions)[num_actions].counter.offset =
+				mlx5_fc_id(counter) - mlx5_fc_get_base_id(counter);
+			(*ractions)[num_actions++].action = tmp_action;
+			fs_actions[num_fs_actions].action = tmp_action;
+			fs_actions[num_fs_actions++].counter = counter;
+		}
+	}
+
+	if (fte->act_dests.flow_context.flow_tag) {
+		if (num_actions == MLX5_FLOW_CONTEXT_ACTION_MAX) {
+			err = -EOPNOTSUPP;
+			goto free_actions;
+		}
+		tmp_action = mlx5_fs_get_action_tag(fs_ctx);
+		if (!tmp_action) {
+			err = -ENOMEM;
+			goto free_actions;
+		}
+		(*ractions)[num_actions].tag.value = fte->act_dests.flow_context.flow_tag;
+		(*ractions)[num_actions++].action = tmp_action;
+	}
+
+	if (fte_action->action & MLX5_FLOW_CONTEXT_ACTION_EXECUTE_ASO) {
+		err = -EOPNOTSUPP;
+		goto free_actions;
+	}
+
+	if (fte_action->action & MLX5_FLOW_CONTEXT_ACTION_DROP) {
+		dest_action = mlx5_fs_get_dest_action_drop(fs_ctx);
+		if (!dest_action) {
+			err = -ENOMEM;
+			goto free_actions;
+		}
+		dest_actions[num_dest_actions++].dest = dest_action;
+	}
+
+	if (fte_action->action & MLX5_FLOW_CONTEXT_ACTION_FWD_DEST) {
+		list_for_each_entry(dst, &fte->node.children, node.list) {
+			struct mlx5_flow_destination *attr = &dst->dest_attr;
+			bool type_uplink =
+				attr->type == MLX5_FLOW_DESTINATION_TYPE_UPLINK;
+
+			if (num_fs_actions == MLX5_FLOW_CONTEXT_ACTION_MAX ||
+			    num_dest_actions == MLX5_FLOW_CONTEXT_ACTION_MAX) {
+				err = -EOPNOTSUPP;
+				goto free_actions;
+			}
+			if (attr->type == MLX5_FLOW_DESTINATION_TYPE_COUNTER)
+				continue;
+
+			switch (attr->type) {
+			case MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE:
+				dest_action = mlx5_fs_get_dest_action_ft(fs_ctx, dst);
+				break;
+			case MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE_NUM:
+				dest_action = mlx5_fs_get_dest_action_table_num(fs_ctx,
+										dst);
+				if (dest_action)
+					break;
+				dest_action = mlx5_fs_create_dest_action_table_num(fs_ctx,
+										   dst);
+				fs_actions[num_fs_actions++].action = dest_action;
+				break;
+			case MLX5_FLOW_DESTINATION_TYPE_RANGE:
+				dest_action = mlx5_fs_create_dest_action_range(ctx, dst);
+				fs_actions[num_fs_actions++].action = dest_action;
+				break;
+			case MLX5_FLOW_DESTINATION_TYPE_UPLINK:
+			case MLX5_FLOW_DESTINATION_TYPE_VPORT:
+				dest_action = mlx5_fs_get_dest_action_vport(fs_ctx, dst,
+									    type_uplink);
+				break;
+			default:
+				err = -EOPNOTSUPP;
+				goto free_actions;
+			}
+			if (!dest_action) {
+				err = -ENOMEM;
+				goto free_actions;
+			}
+			dest_actions[num_dest_actions++].dest = dest_action;
+		}
+	}
+
+	if (num_dest_actions == 1) {
+		if (num_actions == MLX5_FLOW_CONTEXT_ACTION_MAX) {
+			err = -EOPNOTSUPP;
+			goto free_actions;
+		}
+		(*ractions)[num_actions++].action = dest_actions->dest;
+	} else if (num_dest_actions > 1) {
+		u32 flow_source = fte->act_dests.flow_context.flow_source;
+		bool ignore_flow_level;
+
+		if (num_actions == MLX5_FLOW_CONTEXT_ACTION_MAX ||
+		    num_fs_actions == MLX5_FLOW_CONTEXT_ACTION_MAX) {
+			err = -EOPNOTSUPP;
+			goto free_actions;
+		}
+		ignore_flow_level =
+			!!(fte_action->flags & FLOW_ACT_IGNORE_FLOW_LEVEL);
+		tmp_action = mlx5_fs_create_action_dest_array(ctx, dest_actions,
+							      num_dest_actions,
+							      ignore_flow_level,
+							      flow_source);
+		if (!tmp_action) {
+			err = -EOPNOTSUPP;
+			goto free_actions;
+		}
+		fs_actions[num_fs_actions++].action = tmp_action;
+		(*ractions)[num_actions++].action = tmp_action;
+	}
+
+	if (num_actions == MLX5_FLOW_CONTEXT_ACTION_MAX ||
+	    num_fs_actions == MLX5_FLOW_CONTEXT_ACTION_MAX) {
+		err = -EOPNOTSUPP;
+		goto free_actions;
+	}
+
+	tmp_action = mlx5_fs_create_action_last(ctx);
+	if (!tmp_action) {
+		err = -ENOMEM;
+		goto free_actions;
+	}
+	fs_actions[num_fs_actions++].action = tmp_action;
+	(*ractions)[num_actions++].action = tmp_action;
+
+	kfree(dest_actions);
+
+	/* Actions created specifically for this rule will be destroyed
+	 * once rule is deleted.
+	 */
+	fte->fs_hws_rule.num_fs_actions = num_fs_actions;
+	fte->fs_hws_rule.hws_fs_actions = fs_actions;
+
+	return 0;
+
+free_actions:
+	mlx5_fs_destroy_fs_actions(&fs_actions, &num_fs_actions);
+free_dest_actions_alloc:
+	kfree(dest_actions);
+free_fs_actions_alloc:
+	kfree(fs_actions);
+free_actions_alloc:
+	kfree(*ractions);
+	*ractions = NULL;
+out_err:
+	return err;
+}
+
+static int mlx5_cmd_hws_create_fte(struct mlx5_flow_root_namespace *ns,
+				   struct mlx5_flow_table *ft,
+				   struct mlx5_flow_group *group,
+				   struct fs_fte *fte)
+{
+	struct mlx5hws_match_parameters params;
+	struct mlx5hws_rule_action *ractions;
+	struct mlx5hws_bwc_rule *rule;
+	int err = 0;
+
+	if (mlx5_fs_cmd_is_fw_term_table(ft)) {
+		/* Packet reformat on terminamtion table not supported yet */
+		if (fte->act_dests.action.action &
+		    MLX5_FLOW_CONTEXT_ACTION_PACKET_REFORMAT)
+			return -EOPNOTSUPP;
+		return mlx5_fs_cmd_get_fw_cmds()->create_fte(ns, ft, group, fte);
+	}
+
+	err = mlx5_fs_fte_get_hws_actions(ns, ft, group, fte, &ractions);
+	if (err)
+		goto out_err;
+
+	params.match_sz = sizeof(fte->val);
+	params.match_buf = fte->val;
+
+	rule = mlx5hws_bwc_rule_create(group->fs_hws_matcher.matcher, &params,
+				       fte->act_dests.flow_context.flow_source,
+				       ractions);
+	kfree(ractions);
+	if (!rule) {
+		err = -EINVAL;
+		goto free_actions;
+	}
+
+	fte->fs_hws_rule.bwc_rule = rule;
+	return 0;
+
+free_actions:
+	mlx5_fs_destroy_fs_actions(&fte->fs_hws_rule.hws_fs_actions,
+				   &fte->fs_hws_rule.num_fs_actions);
+out_err:
+	mlx5_core_err(ns->dev, "Failed to create hws rule err(%d)\n", err);
+	return err;
+}
+
+static int mlx5_cmd_hws_delete_fte(struct mlx5_flow_root_namespace *ns,
+				   struct mlx5_flow_table *ft,
+				   struct fs_fte *fte)
+{
+	struct mlx5_fs_hws_rule *rule = &fte->fs_hws_rule;
+	int err;
+
+	if (mlx5_fs_cmd_is_fw_term_table(ft))
+		return mlx5_fs_cmd_get_fw_cmds()->delete_fte(ns, ft, fte);
+
+	err = mlx5hws_bwc_rule_destroy(rule->bwc_rule);
+	rule->bwc_rule = NULL;
+
+	mlx5_fs_destroy_fs_actions(&rule->hws_fs_actions, &rule->num_fs_actions);
+
+	return err;
+}
+
+static int mlx5_cmd_hws_update_fte(struct mlx5_flow_root_namespace *ns,
+				   struct mlx5_flow_table *ft,
+				   struct mlx5_flow_group *group,
+				   int modify_mask,
+				   struct fs_fte *fte)
+{
+	int allowed_mask = BIT(MLX5_SET_FTE_MODIFY_ENABLE_MASK_ACTION) |
+		BIT(MLX5_SET_FTE_MODIFY_ENABLE_MASK_DESTINATION_LIST) |
+		BIT(MLX5_SET_FTE_MODIFY_ENABLE_MASK_FLOW_COUNTERS);
+	struct mlx5_fs_hws_rule_action *saved_hws_fs_actions;
+	struct mlx5hws_rule_action *ractions;
+	int saved_num_fs_actions;
+	int ret;
+
+	if (mlx5_fs_cmd_is_fw_term_table(ft))
+		return mlx5_fs_cmd_get_fw_cmds()->update_fte(ns, ft, group,
+							     modify_mask, fte);
+
+	if ((modify_mask & ~allowed_mask) != 0)
+		return -EINVAL;
+
+	saved_hws_fs_actions = fte->fs_hws_rule.hws_fs_actions;
+	saved_num_fs_actions = fte->fs_hws_rule.num_fs_actions;
+
+	ret = mlx5_fs_fte_get_hws_actions(ns, ft, group, fte, &ractions);
+	if (ret)
+		return ret;
+
+	ret = mlx5hws_bwc_rule_action_update(fte->fs_hws_rule.bwc_rule, ractions);
+	kfree(ractions);
+	if (ret)
+		goto restore_actions;
+
+	mlx5_fs_destroy_fs_actions(&saved_hws_fs_actions, &saved_num_fs_actions);
+	return ret;
+
+restore_actions:
+	mlx5_fs_destroy_fs_actions(&fte->fs_hws_rule.hws_fs_actions,
+				   &fte->fs_hws_rule.num_fs_actions);
+	fte->fs_hws_rule.hws_fs_actions = saved_hws_fs_actions;
+	fte->fs_hws_rule.num_fs_actions = saved_num_fs_actions;
+	return ret;
+}
+
+static struct mlx5hws_action *
+mlx5_fs_create_action_remove_header_vlan(struct mlx5hws_context *ctx)
+{
+	u32 flags = MLX5HWS_ACTION_FLAG_HWS_FDB | MLX5HWS_ACTION_FLAG_SHARED;
+	struct mlx5hws_action_remove_header_attr remove_hdr_vlan = {};
+
+	/* MAC anchor not supported in HWS reformat, use VLAN anchor */
+	remove_hdr_vlan.anchor = MLX5_REFORMAT_CONTEXT_ANCHOR_VLAN_START;
+	remove_hdr_vlan.offset = 0;
+	remove_hdr_vlan.size = sizeof(struct vlan_hdr);
+	return mlx5hws_action_create_remove_header(ctx, &remove_hdr_vlan, flags);
+}
+
+static struct mlx5hws_action *
+mlx5_fs_get_action_remove_header_vlan(struct mlx5_fs_hws_context *fs_ctx,
+				      struct mlx5_pkt_reformat_params *params)
+{
+	if (!params ||
+	    params->param_0 != MLX5_REFORMAT_CONTEXT_ANCHOR_MAC_START ||
+	    params->param_1 != offsetof(struct vlan_ethhdr, h_vlan_proto) ||
+	    params->size != sizeof(struct vlan_hdr))
+		return NULL;
+
+	return fs_ctx->hws_pool.remove_hdr_vlan_action;
+}
+
+static int
+mlx5_fs_verify_insert_header_params(struct mlx5_core_dev *mdev,
+				    struct mlx5_pkt_reformat_params *params)
+{
+	if ((!params->data && params->size) || (params->data && !params->size) ||
+	    MLX5_CAP_GEN_2(mdev, max_reformat_insert_size) < params->size ||
+	    MLX5_CAP_GEN_2(mdev, max_reformat_insert_offset) < params->param_1) {
+		mlx5_core_err(mdev, "Invalid reformat params for INSERT_HDR\n");
+		return -EINVAL;
+	}
+	if (params->param_0 != MLX5_FS_INSERT_HDR_VLAN_ANCHOR ||
+	    params->param_1 != MLX5_FS_INSERT_HDR_VLAN_OFFSET ||
+	    params->size != MLX5_FS_INSERT_HDR_VLAN_SIZE) {
+		mlx5_core_err(mdev, "Only vlan insert header supported\n");
+		return -EOPNOTSUPP;
+	}
+	return 0;
+}
+
+static int
+mlx5_fs_verify_encap_decap_params(struct mlx5_core_dev *dev,
+				  struct mlx5_pkt_reformat_params *params)
+{
+	if (params->param_0 || params->param_1) {
+		mlx5_core_err(dev, "Invalid reformat params\n");
+		return -EINVAL;
+	}
+	return 0;
+}
+
+static struct mlx5_fs_pool *
+mlx5_fs_get_pr_encap_pool(struct mlx5_core_dev *dev, struct xarray *pr_pools,
+			  enum mlx5hws_action_type reformat_type, size_t size)
+{
+	struct mlx5_fs_pool *pr_pool;
+	unsigned long index = size;
+	int err;
+
+	pr_pool = xa_load(pr_pools, index);
+	if (pr_pool)
+		return pr_pool;
+
+	pr_pool = kzalloc(sizeof(*pr_pool), GFP_KERNEL);
+	if (!pr_pool)
+		return ERR_PTR(-ENOMEM);
+	err = mlx5_fs_hws_pr_pool_init(pr_pool, dev, size, reformat_type);
+	if (err)
+		goto free_pr_pool;
+	err = xa_insert(pr_pools, index, pr_pool, GFP_KERNEL);
+	if (err)
+		goto cleanup_pr_pool;
+	return pr_pool;
+
+cleanup_pr_pool:
+	mlx5_fs_hws_pr_pool_cleanup(pr_pool);
+free_pr_pool:
+	kfree(pr_pool);
+	return ERR_PTR(err);
+}
+
+static void
+mlx5_fs_destroy_pr_pool(struct mlx5_fs_pool *pool, struct xarray *pr_pools,
+			unsigned long index)
+{
+	xa_erase(pr_pools, index);
+	mlx5_fs_hws_pr_pool_cleanup(pool);
+	kfree(pool);
+}
+
+static int
+mlx5_cmd_hws_packet_reformat_alloc(struct mlx5_flow_root_namespace *ns,
+				   struct mlx5_pkt_reformat_params *params,
+				   enum mlx5_flow_namespace_type namespace,
+				   struct mlx5_pkt_reformat *pkt_reformat)
+{
+	struct mlx5_fs_hws_context *fs_ctx = &ns->fs_hws_context;
+	struct mlx5_fs_hws_actions_pool *hws_pool;
+	struct mlx5hws_action *hws_action = NULL;
+	struct mlx5_fs_hws_pr *pr_data = NULL;
+	struct mlx5_fs_pool *pr_pool = NULL;
+	struct mlx5_core_dev *dev = ns->dev;
+	u8 hdr_idx = 0;
+	int err;
+
+	if (!params)
+		return -EINVAL;
+
+	hws_pool = &fs_ctx->hws_pool;
+
+	switch (params->type) {
+	case MLX5_REFORMAT_TYPE_L2_TO_VXLAN:
+	case MLX5_REFORMAT_TYPE_L2_TO_NVGRE:
+	case MLX5_REFORMAT_TYPE_L2_TO_L2_TUNNEL:
+		if (mlx5_fs_verify_encap_decap_params(dev, params))
+			return -EINVAL;
+		pr_pool = mlx5_fs_get_pr_encap_pool(dev, &hws_pool->el2tol2tnl_pools,
+						    MLX5HWS_ACTION_TYP_REFORMAT_L2_TO_TNL_L2,
+						    params->size);
+		if (IS_ERR(pr_pool))
+			return PTR_ERR(pr_pool);
+		break;
+	case MLX5_REFORMAT_TYPE_L2_TO_L3_TUNNEL:
+		if (mlx5_fs_verify_encap_decap_params(dev, params))
+			return -EINVAL;
+		pr_pool = mlx5_fs_get_pr_encap_pool(dev, &hws_pool->el2tol3tnl_pools,
+						    MLX5HWS_ACTION_TYP_REFORMAT_L2_TO_TNL_L3,
+						    params->size);
+		if (IS_ERR(pr_pool))
+			return PTR_ERR(pr_pool);
+		break;
+	case MLX5_REFORMAT_TYPE_L3_TUNNEL_TO_L2:
+		if (mlx5_fs_verify_encap_decap_params(dev, params))
+			return -EINVAL;
+		pr_pool = &hws_pool->dl3tnltol2_pool;
+		hdr_idx = params->size == ETH_HLEN ?
+			  MLX5_FS_DL3TNLTOL2_MAC_HDR_IDX :
+			  MLX5_FS_DL3TNLTOL2_MAC_VLAN_HDR_IDX;
+		break;
+	case MLX5_REFORMAT_TYPE_INSERT_HDR:
+		err = mlx5_fs_verify_insert_header_params(dev, params);
+		if (err)
+			return err;
+		pr_pool = &hws_pool->insert_hdr_pool;
+		break;
+	case MLX5_REFORMAT_TYPE_REMOVE_HDR:
+		hws_action = mlx5_fs_get_action_remove_header_vlan(fs_ctx, params);
+		if (!hws_action)
+			mlx5_core_err(dev, "Only vlan remove header supported\n");
+		break;
+	default:
+		mlx5_core_err(ns->dev, "Packet-reformat not supported(%d)\n",
+			      params->type);
+		return -EOPNOTSUPP;
+	}
+
+	if (pr_pool) {
+		pr_data = mlx5_fs_hws_pr_pool_acquire_pr(pr_pool);
+		if (IS_ERR_OR_NULL(pr_data))
+			return !pr_data ? -EINVAL : PTR_ERR(pr_data);
+		hws_action = pr_data->bulk->hws_action;
+		if (!hws_action) {
+			mlx5_core_err(dev,
+				      "Failed allocating packet-reformat action\n");
+			err = -EINVAL;
+			goto release_pr;
+		}
+		pr_data->data = kmemdup(params->data, params->size, GFP_KERNEL);
+		if (!pr_data->data) {
+			err = -ENOMEM;
+			goto release_pr;
+		}
+		pr_data->hdr_idx = hdr_idx;
+		pr_data->data_size = params->size;
+		pkt_reformat->fs_hws_action.pr_data = pr_data;
+	}
+
+	pkt_reformat->owner = MLX5_FLOW_RESOURCE_OWNER_SW;
+	pkt_reformat->fs_hws_action.hws_action = hws_action;
+	return 0;
+
+release_pr:
+	if (pr_pool && pr_data)
+		mlx5_fs_hws_pr_pool_release_pr(pr_pool, pr_data);
+	return err;
+}
+
+static void mlx5_cmd_hws_packet_reformat_dealloc(struct mlx5_flow_root_namespace *ns,
+						 struct mlx5_pkt_reformat *pkt_reformat)
+{
+	struct mlx5_fs_hws_actions_pool *hws_pool = &ns->fs_hws_context.hws_pool;
+	struct mlx5_core_dev *dev = ns->dev;
+	struct mlx5_fs_hws_pr *pr_data;
+	struct mlx5_fs_pool *pr_pool;
+
+	if (pkt_reformat->reformat_type == MLX5_REFORMAT_TYPE_REMOVE_HDR)
+		return;
+
+	if (!pkt_reformat->fs_hws_action.pr_data) {
+		mlx5_core_err(ns->dev, "Failed release packet-reformat\n");
+		return;
+	}
+	pr_data = pkt_reformat->fs_hws_action.pr_data;
+
+	switch (pkt_reformat->reformat_type) {
+	case MLX5_REFORMAT_TYPE_L2_TO_VXLAN:
+	case MLX5_REFORMAT_TYPE_L2_TO_NVGRE:
+	case MLX5_REFORMAT_TYPE_L2_TO_L2_TUNNEL:
+		pr_pool = mlx5_fs_get_pr_encap_pool(dev, &hws_pool->el2tol2tnl_pools,
+						    MLX5HWS_ACTION_TYP_REFORMAT_L2_TO_TNL_L2,
+						    pr_data->data_size);
+		break;
+	case MLX5_REFORMAT_TYPE_L2_TO_L3_TUNNEL:
+		pr_pool = mlx5_fs_get_pr_encap_pool(dev, &hws_pool->el2tol2tnl_pools,
+						    MLX5HWS_ACTION_TYP_REFORMAT_L2_TO_TNL_L2,
+						    pr_data->data_size);
+		break;
+	case MLX5_REFORMAT_TYPE_L3_TUNNEL_TO_L2:
+		pr_pool = &hws_pool->dl3tnltol2_pool;
+		break;
+	case MLX5_REFORMAT_TYPE_INSERT_HDR:
+		pr_pool = &hws_pool->insert_hdr_pool;
+		break;
+	default:
+		mlx5_core_err(ns->dev, "Unknown packet-reformat type\n");
+		return;
+	}
+	if (!pkt_reformat->fs_hws_action.pr_data || IS_ERR(pr_pool)) {
+		mlx5_core_err(ns->dev, "Failed release packet-reformat\n");
+		return;
+	}
+	kfree(pr_data->data);
+	mlx5_fs_hws_pr_pool_release_pr(pr_pool, pr_data);
+	pkt_reformat->fs_hws_action.pr_data = NULL;
+}
+
+static struct mlx5_fs_pool *
+mlx5_fs_create_mh_pool(struct mlx5_core_dev *dev,
+		       struct mlx5hws_action_mh_pattern *pattern,
+		       struct xarray *mh_pools, unsigned long index)
+{
+	struct mlx5_fs_pool *pool;
+	int err;
+
+	pool = kzalloc(sizeof(*pool), GFP_KERNEL);
+	if (!pool)
+		return ERR_PTR(-ENOMEM);
+	err = mlx5_fs_hws_mh_pool_init(pool, dev, pattern);
+	if (err)
+		goto free_pool;
+	err = xa_insert(mh_pools, index, pool, GFP_KERNEL);
+	if (err)
+		goto cleanup_pool;
+	return pool;
+
+cleanup_pool:
+	mlx5_fs_hws_mh_pool_cleanup(pool);
+free_pool:
+	kfree(pool);
+	return ERR_PTR(err);
+}
+
+static void
+mlx5_fs_destroy_mh_pool(struct mlx5_fs_pool *pool, struct xarray *mh_pools,
+			unsigned long index)
+{
+	xa_erase(mh_pools, index);
+	mlx5_fs_hws_mh_pool_cleanup(pool);
+	kfree(pool);
+}
+
+static int mlx5_cmd_hws_modify_header_alloc(struct mlx5_flow_root_namespace *ns,
+					    u8 namespace, u8 num_actions,
+					    void *modify_actions,
+					    struct mlx5_modify_hdr *modify_hdr)
+{
+	struct mlx5_fs_hws_actions_pool *hws_pool = &ns->fs_hws_context.hws_pool;
+	struct mlx5hws_action_mh_pattern pattern = {};
+	struct mlx5_fs_hws_mh *mh_data = NULL;
+	struct mlx5hws_action *hws_action;
+	struct mlx5_fs_pool *pool;
+	unsigned long i, cnt = 0;
+	bool known_pattern;
+	int err;
+
+	pattern.sz = MLX5_UN_SZ_BYTES(set_add_copy_action_in_auto) * num_actions;
+	pattern.data = modify_actions;
+
+	known_pattern = false;
+	xa_for_each(&hws_pool->mh_pools, i, pool) {
+		if (mlx5_fs_hws_mh_pool_match(pool, &pattern)) {
+			known_pattern = true;
+			break;
+		}
+		cnt++;
+	}
+
+	if (!known_pattern) {
+		pool = mlx5_fs_create_mh_pool(ns->dev, &pattern,
+					      &hws_pool->mh_pools, cnt);
+		if (IS_ERR(pool))
+			return PTR_ERR(pool);
+	}
+	mh_data = mlx5_fs_hws_mh_pool_acquire_mh(pool);
+	if (IS_ERR(mh_data)) {
+		err = PTR_ERR(mh_data);
+		goto destroy_pool;
+	}
+	hws_action = mh_data->bulk->hws_action;
+	mh_data->data = kmemdup(pattern.data, pattern.sz, GFP_KERNEL);
+	if (!mh_data->data) {
+		err = -ENOMEM;
+		goto release_mh;
+	}
+	modify_hdr->fs_hws_action.mh_data = mh_data;
+	modify_hdr->fs_hws_action.fs_pool = pool;
+	modify_hdr->owner = MLX5_FLOW_RESOURCE_OWNER_SW;
+	modify_hdr->fs_hws_action.hws_action = hws_action;
+
+	return 0;
+
+release_mh:
+	mlx5_fs_hws_mh_pool_release_mh(pool, mh_data);
+destroy_pool:
+	if (!known_pattern)
+		mlx5_fs_destroy_mh_pool(pool, &hws_pool->mh_pools, cnt);
+	return err;
+}
+
+static void mlx5_cmd_hws_modify_header_dealloc(struct mlx5_flow_root_namespace *ns,
+					       struct mlx5_modify_hdr *modify_hdr)
+{
+	struct mlx5_fs_hws_mh *mh_data;
+	struct mlx5_fs_pool *pool;
+
+	if (!modify_hdr->fs_hws_action.fs_pool || !modify_hdr->fs_hws_action.mh_data) {
+		mlx5_core_err(ns->dev, "Failed release modify-header\n");
+		return;
+	}
+
+	mh_data = modify_hdr->fs_hws_action.mh_data;
+	kfree(mh_data->data);
+	pool = modify_hdr->fs_hws_action.fs_pool;
+	mlx5_fs_hws_mh_pool_release_mh(pool, mh_data);
+	modify_hdr->fs_hws_action.mh_data = NULL;
+}
+
+static int mlx5_cmd_hws_create_match_definer(struct mlx5_flow_root_namespace *ns,
+					     u16 format_id, u32 *match_mask)
+{
+	return -EOPNOTSUPP;
+}
+
+static int mlx5_cmd_hws_destroy_match_definer(struct mlx5_flow_root_namespace *ns,
+					      int definer_id)
+{
+	return -EOPNOTSUPP;
+}
+
+static u32 mlx5_cmd_hws_get_capabilities(struct mlx5_flow_root_namespace *ns,
+					 enum fs_flow_table_type ft_type)
+{
+	if (ft_type != FS_FT_FDB)
+		return 0;
+
+	return MLX5_FLOW_STEERING_CAP_VLAN_PUSH_ON_RX |
+	       MLX5_FLOW_STEERING_CAP_VLAN_POP_ON_TX |
+	       MLX5_FLOW_STEERING_CAP_MATCH_RANGES;
+}
+
+bool mlx5_fs_hws_is_supported(struct mlx5_core_dev *dev)
+{
+	return mlx5hws_is_supported(dev);
+}
+
+static const struct mlx5_flow_cmds mlx5_flow_cmds_hws = {
+	.create_flow_table = mlx5_cmd_hws_create_flow_table,
+	.destroy_flow_table = mlx5_cmd_hws_destroy_flow_table,
+	.modify_flow_table = mlx5_cmd_hws_modify_flow_table,
+	.update_root_ft = mlx5_cmd_hws_update_root_ft,
+	.create_flow_group = mlx5_cmd_hws_create_flow_group,
+	.destroy_flow_group = mlx5_cmd_hws_destroy_flow_group,
+	.create_fte = mlx5_cmd_hws_create_fte,
+	.delete_fte = mlx5_cmd_hws_delete_fte,
+	.update_fte = mlx5_cmd_hws_update_fte,
+	.packet_reformat_alloc = mlx5_cmd_hws_packet_reformat_alloc,
+	.packet_reformat_dealloc = mlx5_cmd_hws_packet_reformat_dealloc,
+	.modify_header_alloc = mlx5_cmd_hws_modify_header_alloc,
+	.modify_header_dealloc = mlx5_cmd_hws_modify_header_dealloc,
+	.create_match_definer = mlx5_cmd_hws_create_match_definer,
+	.destroy_match_definer = mlx5_cmd_hws_destroy_match_definer,
+	.create_ns = mlx5_cmd_hws_create_ns,
+	.destroy_ns = mlx5_cmd_hws_destroy_ns,
+	.set_peer = mlx5_cmd_hws_set_peer,
+	.get_capabilities = mlx5_cmd_hws_get_capabilities,
+};
+
+const struct mlx5_flow_cmds *mlx5_fs_cmd_get_hws_cmds(void)
+{
+	return &mlx5_flow_cmds_hws;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/fs_hws.h b/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/fs_hws.h
new file mode 100644
index 000000000000..cbddb72d4362
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/fs_hws.h
@@ -0,0 +1,80 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2025 NVIDIA Corporation & Affiliates */
+
+#ifndef _MLX5_FS_HWS_
+#define _MLX5_FS_HWS_
+
+#include "mlx5hws.h"
+#include "fs_hws_pools.h"
+
+struct mlx5_fs_hws_actions_pool {
+	struct mlx5hws_action *tag_action;
+	struct mlx5hws_action *pop_vlan_action;
+	struct mlx5hws_action *push_vlan_action;
+	struct mlx5hws_action *drop_action;
+	struct mlx5hws_action *decapl2_action;
+	struct mlx5hws_action *remove_hdr_vlan_action;
+	struct mlx5_fs_pool insert_hdr_pool;
+	struct mlx5_fs_pool dl3tnltol2_pool;
+	struct xarray el2tol3tnl_pools;
+	struct xarray el2tol2tnl_pools;
+	struct xarray mh_pools;
+	struct xarray table_dests;
+	struct xarray vport_vhca_dests;
+	struct xarray vport_dests;
+};
+
+struct mlx5_fs_hws_context {
+	struct mlx5hws_context	*hws_ctx;
+	struct mlx5_fs_hws_actions_pool hws_pool;
+};
+
+struct mlx5_fs_hws_table {
+	struct mlx5hws_table *hws_table;
+	bool miss_ft_set;
+};
+
+struct mlx5_fs_hws_action {
+	struct mlx5hws_action *hws_action;
+	struct mlx5_fs_pool *fs_pool;
+	struct mlx5_fs_hws_pr *pr_data;
+	struct mlx5_fs_hws_mh *mh_data;
+};
+
+struct mlx5_fs_hws_matcher {
+	struct mlx5hws_bwc_matcher *matcher;
+};
+
+struct mlx5_fs_hws_rule_action {
+	struct mlx5hws_action *action;
+	union {
+		struct mlx5_fc *counter;
+	};
+};
+
+struct mlx5_fs_hws_rule {
+	struct mlx5hws_bwc_rule *bwc_rule;
+	struct mlx5_fs_hws_rule_action *hws_fs_actions;
+	int num_fs_actions;
+};
+
+#ifdef CONFIG_MLX5_HW_STEERING
+
+bool mlx5_fs_hws_is_supported(struct mlx5_core_dev *dev);
+
+const struct mlx5_flow_cmds *mlx5_fs_cmd_get_hws_cmds(void);
+
+#else
+
+static inline bool mlx5_fs_hws_is_supported(struct mlx5_core_dev *dev)
+{
+	return false;
+}
+
+static inline const struct mlx5_flow_cmds *mlx5_fs_cmd_get_hws_cmds(void)
+{
+	return NULL;
+}
+
+#endif /* CONFIG_MLX5_HWS_STEERING */
+#endif
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/fs_hws_pools.c b/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/fs_hws_pools.c
new file mode 100644
index 000000000000..2ae4ac62b0e2
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/fs_hws_pools.c
@@ -0,0 +1,450 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2025 NVIDIA Corporation & Affiliates */
+
+#include <mlx5_core.h>
+#include "fs_hws_pools.h"
+
+#define MLX5_FS_HWS_DEFAULT_BULK_LEN 65536
+#define MLX5_FS_HWS_POOL_MAX_THRESHOLD BIT(18)
+#define MLX5_FS_HWS_POOL_USED_BUFF_RATIO 10
+
+static struct mlx5hws_action *
+mlx5_fs_dl3tnltol2_bulk_action_create(struct mlx5hws_context *ctx)
+{
+	struct mlx5hws_action_reformat_header reformat_hdr[2] = {};
+	u32 flags = MLX5HWS_ACTION_FLAG_HWS_FDB;
+	enum mlx5hws_action_type reformat_type;
+	u32 log_bulk_size;
+
+	reformat_type = MLX5HWS_ACTION_TYP_REFORMAT_TNL_L3_TO_L2;
+	reformat_hdr[MLX5_FS_DL3TNLTOL2_MAC_HDR_IDX].sz = ETH_HLEN;
+	reformat_hdr[MLX5_FS_DL3TNLTOL2_MAC_VLAN_HDR_IDX].sz = ETH_HLEN + VLAN_HLEN;
+
+	log_bulk_size = ilog2(MLX5_FS_HWS_DEFAULT_BULK_LEN);
+	return mlx5hws_action_create_reformat(ctx, reformat_type, 2,
+					      reformat_hdr, log_bulk_size, flags);
+}
+
+static struct mlx5hws_action *
+mlx5_fs_el2tol3tnl_bulk_action_create(struct mlx5hws_context *ctx, size_t data_size)
+{
+	struct mlx5hws_action_reformat_header reformat_hdr = {};
+	u32 flags = MLX5HWS_ACTION_FLAG_HWS_FDB;
+	enum mlx5hws_action_type reformat_type;
+	u32 log_bulk_size;
+
+	reformat_type = MLX5HWS_ACTION_TYP_REFORMAT_L2_TO_TNL_L3;
+	reformat_hdr.sz = data_size;
+
+	log_bulk_size = ilog2(MLX5_FS_HWS_DEFAULT_BULK_LEN);
+	return mlx5hws_action_create_reformat(ctx, reformat_type, 1,
+					      &reformat_hdr, log_bulk_size, flags);
+}
+
+static struct mlx5hws_action *
+mlx5_fs_el2tol2tnl_bulk_action_create(struct mlx5hws_context *ctx, size_t data_size)
+{
+	struct mlx5hws_action_reformat_header reformat_hdr = {};
+	u32 flags = MLX5HWS_ACTION_FLAG_HWS_FDB;
+	enum mlx5hws_action_type reformat_type;
+	u32 log_bulk_size;
+
+	reformat_type = MLX5HWS_ACTION_TYP_REFORMAT_L2_TO_TNL_L2;
+	reformat_hdr.sz = data_size;
+
+	log_bulk_size = ilog2(MLX5_FS_HWS_DEFAULT_BULK_LEN);
+	return mlx5hws_action_create_reformat(ctx, reformat_type, 1,
+					      &reformat_hdr, log_bulk_size, flags);
+}
+
+static struct mlx5hws_action *
+mlx5_fs_insert_hdr_bulk_action_create(struct mlx5hws_context *ctx)
+{
+	struct mlx5hws_action_insert_header insert_hdr = {};
+	u32 flags = MLX5HWS_ACTION_FLAG_HWS_FDB;
+	u32 log_bulk_size;
+
+	log_bulk_size = ilog2(MLX5_FS_HWS_DEFAULT_BULK_LEN);
+	insert_hdr.hdr.sz = MLX5_FS_INSERT_HDR_VLAN_SIZE;
+	insert_hdr.anchor = MLX5_FS_INSERT_HDR_VLAN_ANCHOR;
+	insert_hdr.offset = MLX5_FS_INSERT_HDR_VLAN_OFFSET;
+
+	return mlx5hws_action_create_insert_header(ctx, 1, &insert_hdr,
+						   log_bulk_size, flags);
+}
+
+static struct mlx5hws_action *
+mlx5_fs_pr_bulk_action_create(struct mlx5_core_dev *dev,
+			      struct mlx5_fs_hws_pr_pool_ctx *pr_pool_ctx)
+{
+	struct mlx5_flow_root_namespace *root_ns;
+	struct mlx5hws_context *ctx;
+	size_t encap_data_size;
+
+	root_ns = mlx5_get_root_namespace(dev, MLX5_FLOW_NAMESPACE_FDB);
+	if (!root_ns || root_ns->mode != MLX5_FLOW_STEERING_MODE_HMFS)
+		return NULL;
+
+	ctx = root_ns->fs_hws_context.hws_ctx;
+	if (!ctx)
+		return NULL;
+
+	encap_data_size = pr_pool_ctx->encap_data_size;
+	switch (pr_pool_ctx->reformat_type) {
+	case MLX5HWS_ACTION_TYP_REFORMAT_TNL_L3_TO_L2:
+		return mlx5_fs_dl3tnltol2_bulk_action_create(ctx);
+	case MLX5HWS_ACTION_TYP_REFORMAT_L2_TO_TNL_L3:
+		return mlx5_fs_el2tol3tnl_bulk_action_create(ctx, encap_data_size);
+	case MLX5HWS_ACTION_TYP_REFORMAT_L2_TO_TNL_L2:
+		return mlx5_fs_el2tol2tnl_bulk_action_create(ctx, encap_data_size);
+	case MLX5HWS_ACTION_TYP_INSERT_HEADER:
+		return mlx5_fs_insert_hdr_bulk_action_create(ctx);
+	default:
+		return NULL;
+	}
+	return NULL;
+}
+
+static struct mlx5_fs_bulk *
+mlx5_fs_hws_pr_bulk_create(struct mlx5_core_dev *dev, void *pool_ctx)
+{
+	struct mlx5_fs_hws_pr_pool_ctx *pr_pool_ctx;
+	struct mlx5_fs_hws_pr_bulk *pr_bulk;
+	int bulk_len;
+	int i;
+
+	if (!pool_ctx)
+		return NULL;
+	pr_pool_ctx = pool_ctx;
+	bulk_len = MLX5_FS_HWS_DEFAULT_BULK_LEN;
+	pr_bulk = kvzalloc(struct_size(pr_bulk, prs_data, bulk_len), GFP_KERNEL);
+	if (!pr_bulk)
+		return NULL;
+
+	if (mlx5_fs_bulk_init(dev, &pr_bulk->fs_bulk, bulk_len))
+		goto free_pr_bulk;
+
+	for (i = 0; i < bulk_len; i++) {
+		pr_bulk->prs_data[i].bulk = pr_bulk;
+		pr_bulk->prs_data[i].offset = i;
+	}
+
+	pr_bulk->hws_action = mlx5_fs_pr_bulk_action_create(dev, pr_pool_ctx);
+	if (!pr_bulk->hws_action)
+		goto cleanup_fs_bulk;
+
+	return &pr_bulk->fs_bulk;
+
+cleanup_fs_bulk:
+	mlx5_fs_bulk_cleanup(&pr_bulk->fs_bulk);
+free_pr_bulk:
+	kvfree(pr_bulk);
+	return NULL;
+}
+
+static int
+mlx5_fs_hws_pr_bulk_destroy(struct mlx5_core_dev *dev, struct mlx5_fs_bulk *fs_bulk)
+{
+	struct mlx5_fs_hws_pr_bulk *pr_bulk;
+
+	pr_bulk = container_of(fs_bulk, struct mlx5_fs_hws_pr_bulk, fs_bulk);
+	if (mlx5_fs_bulk_get_free_amount(fs_bulk) < fs_bulk->bulk_len) {
+		mlx5_core_err(dev, "Freeing bulk before all reformats were released\n");
+		return -EBUSY;
+	}
+
+	mlx5hws_action_destroy(pr_bulk->hws_action);
+	mlx5_fs_bulk_cleanup(fs_bulk);
+	kvfree(pr_bulk);
+
+	return 0;
+}
+
+static void mlx5_hws_pool_update_threshold(struct mlx5_fs_pool *hws_pool)
+{
+	hws_pool->threshold = min_t(int, MLX5_FS_HWS_POOL_MAX_THRESHOLD,
+				    hws_pool->used_units / MLX5_FS_HWS_POOL_USED_BUFF_RATIO);
+}
+
+static const struct mlx5_fs_pool_ops mlx5_fs_hws_pr_pool_ops = {
+	.bulk_create = mlx5_fs_hws_pr_bulk_create,
+	.bulk_destroy = mlx5_fs_hws_pr_bulk_destroy,
+	.update_threshold = mlx5_hws_pool_update_threshold,
+};
+
+int mlx5_fs_hws_pr_pool_init(struct mlx5_fs_pool *pr_pool,
+			     struct mlx5_core_dev *dev, size_t encap_data_size,
+			     enum mlx5hws_action_type reformat_type)
+{
+	struct mlx5_fs_hws_pr_pool_ctx *pr_pool_ctx;
+
+	if (reformat_type != MLX5HWS_ACTION_TYP_INSERT_HEADER &&
+	    reformat_type != MLX5HWS_ACTION_TYP_REFORMAT_TNL_L3_TO_L2 &&
+	    reformat_type != MLX5HWS_ACTION_TYP_REFORMAT_L2_TO_TNL_L3 &&
+	    reformat_type != MLX5HWS_ACTION_TYP_REFORMAT_L2_TO_TNL_L2)
+		return -EOPNOTSUPP;
+
+	pr_pool_ctx = kzalloc(sizeof(*pr_pool_ctx), GFP_KERNEL);
+	if (!pr_pool_ctx)
+		return -ENOMEM;
+	pr_pool_ctx->reformat_type = reformat_type;
+	pr_pool_ctx->encap_data_size = encap_data_size;
+	mlx5_fs_pool_init(pr_pool, dev, &mlx5_fs_hws_pr_pool_ops, pr_pool_ctx);
+	return 0;
+}
+
+void mlx5_fs_hws_pr_pool_cleanup(struct mlx5_fs_pool *pr_pool)
+{
+	struct mlx5_fs_hws_pr_pool_ctx *pr_pool_ctx;
+
+	mlx5_fs_pool_cleanup(pr_pool);
+	pr_pool_ctx = pr_pool->pool_ctx;
+	if (!pr_pool_ctx)
+		return;
+	kfree(pr_pool_ctx);
+}
+
+struct mlx5_fs_hws_pr *
+mlx5_fs_hws_pr_pool_acquire_pr(struct mlx5_fs_pool *pr_pool)
+{
+	struct mlx5_fs_pool_index pool_index = {};
+	struct mlx5_fs_hws_pr_bulk *pr_bulk;
+	int err;
+
+	err = mlx5_fs_pool_acquire_index(pr_pool, &pool_index);
+	if (err)
+		return ERR_PTR(err);
+	pr_bulk = container_of(pool_index.fs_bulk, struct mlx5_fs_hws_pr_bulk,
+			       fs_bulk);
+	return &pr_bulk->prs_data[pool_index.index];
+}
+
+void mlx5_fs_hws_pr_pool_release_pr(struct mlx5_fs_pool *pr_pool,
+				    struct mlx5_fs_hws_pr *pr_data)
+{
+	struct mlx5_fs_bulk *fs_bulk = &pr_data->bulk->fs_bulk;
+	struct mlx5_fs_pool_index pool_index = {};
+	struct mlx5_core_dev *dev = pr_pool->dev;
+
+	pool_index.fs_bulk = fs_bulk;
+	pool_index.index = pr_data->offset;
+	if (mlx5_fs_pool_release_index(pr_pool, &pool_index))
+		mlx5_core_warn(dev, "Attempted to release packet reformat which is not acquired\n");
+}
+
+struct mlx5hws_action *mlx5_fs_hws_pr_get_action(struct mlx5_fs_hws_pr *pr_data)
+{
+	return pr_data->bulk->hws_action;
+}
+
+static struct mlx5hws_action *
+mlx5_fs_mh_bulk_action_create(struct mlx5hws_context *ctx,
+			      struct mlx5hws_action_mh_pattern *pattern)
+{
+	u32 flags = MLX5HWS_ACTION_FLAG_HWS_FDB;
+	u32 log_bulk_size;
+
+	log_bulk_size = ilog2(MLX5_FS_HWS_DEFAULT_BULK_LEN);
+	return mlx5hws_action_create_modify_header(ctx, 1, pattern,
+						   log_bulk_size, flags);
+}
+
+static struct mlx5_fs_bulk *
+mlx5_fs_hws_mh_bulk_create(struct mlx5_core_dev *dev, void *pool_ctx)
+{
+	struct mlx5hws_action_mh_pattern *pattern;
+	struct mlx5_flow_root_namespace *root_ns;
+	struct mlx5_fs_hws_mh_bulk *mh_bulk;
+	struct mlx5hws_context *ctx;
+	int bulk_len;
+
+	if (!pool_ctx)
+		return NULL;
+
+	root_ns = mlx5_get_root_namespace(dev, MLX5_FLOW_NAMESPACE_FDB);
+	if (!root_ns || root_ns->mode != MLX5_FLOW_STEERING_MODE_HMFS)
+		return NULL;
+
+	ctx = root_ns->fs_hws_context.hws_ctx;
+	if (!ctx)
+		return NULL;
+
+	pattern = pool_ctx;
+	bulk_len = MLX5_FS_HWS_DEFAULT_BULK_LEN;
+	mh_bulk = kvzalloc(struct_size(mh_bulk, mhs_data, bulk_len), GFP_KERNEL);
+	if (!mh_bulk)
+		return NULL;
+
+	if (mlx5_fs_bulk_init(dev, &mh_bulk->fs_bulk, bulk_len))
+		goto free_mh_bulk;
+
+	for (int i = 0; i < bulk_len; i++) {
+		mh_bulk->mhs_data[i].bulk = mh_bulk;
+		mh_bulk->mhs_data[i].offset = i;
+	}
+
+	mh_bulk->hws_action = mlx5_fs_mh_bulk_action_create(ctx, pattern);
+	if (!mh_bulk->hws_action)
+		goto cleanup_fs_bulk;
+
+	return &mh_bulk->fs_bulk;
+
+cleanup_fs_bulk:
+	mlx5_fs_bulk_cleanup(&mh_bulk->fs_bulk);
+free_mh_bulk:
+	kvfree(mh_bulk);
+	return NULL;
+}
+
+static int
+mlx5_fs_hws_mh_bulk_destroy(struct mlx5_core_dev *dev,
+			    struct mlx5_fs_bulk *fs_bulk)
+{
+	struct mlx5_fs_hws_mh_bulk *mh_bulk;
+
+	mh_bulk = container_of(fs_bulk, struct mlx5_fs_hws_mh_bulk, fs_bulk);
+	if (mlx5_fs_bulk_get_free_amount(fs_bulk) < fs_bulk->bulk_len) {
+		mlx5_core_err(dev, "Freeing bulk before all modify header were released\n");
+		return -EBUSY;
+	}
+
+	mlx5hws_action_destroy(mh_bulk->hws_action);
+	mlx5_fs_bulk_cleanup(fs_bulk);
+	kvfree(mh_bulk);
+
+	return 0;
+}
+
+static const struct mlx5_fs_pool_ops mlx5_fs_hws_mh_pool_ops = {
+	.bulk_create = mlx5_fs_hws_mh_bulk_create,
+	.bulk_destroy = mlx5_fs_hws_mh_bulk_destroy,
+	.update_threshold = mlx5_hws_pool_update_threshold,
+};
+
+int mlx5_fs_hws_mh_pool_init(struct mlx5_fs_pool *fs_hws_mh_pool,
+			     struct mlx5_core_dev *dev,
+			     struct mlx5hws_action_mh_pattern *pattern)
+{
+	struct mlx5hws_action_mh_pattern *pool_pattern;
+
+	pool_pattern = kzalloc(sizeof(*pool_pattern), GFP_KERNEL);
+	if (!pool_pattern)
+		return -ENOMEM;
+	pool_pattern->data = kmemdup(pattern->data, pattern->sz, GFP_KERNEL);
+	if (!pool_pattern->data) {
+		kfree(pool_pattern);
+		return -ENOMEM;
+	}
+	pool_pattern->sz = pattern->sz;
+	mlx5_fs_pool_init(fs_hws_mh_pool, dev, &mlx5_fs_hws_mh_pool_ops,
+			  pool_pattern);
+	return 0;
+}
+
+void mlx5_fs_hws_mh_pool_cleanup(struct mlx5_fs_pool *fs_hws_mh_pool)
+{
+	struct mlx5hws_action_mh_pattern *pool_pattern;
+
+	mlx5_fs_pool_cleanup(fs_hws_mh_pool);
+	pool_pattern = fs_hws_mh_pool->pool_ctx;
+	if (!pool_pattern)
+		return;
+	kfree(pool_pattern->data);
+	kfree(pool_pattern);
+}
+
+struct mlx5_fs_hws_mh *
+mlx5_fs_hws_mh_pool_acquire_mh(struct mlx5_fs_pool *mh_pool)
+{
+	struct mlx5_fs_pool_index pool_index = {};
+	struct mlx5_fs_hws_mh_bulk *mh_bulk;
+	int err;
+
+	err = mlx5_fs_pool_acquire_index(mh_pool, &pool_index);
+	if (err)
+		return ERR_PTR(err);
+	mh_bulk = container_of(pool_index.fs_bulk, struct mlx5_fs_hws_mh_bulk,
+			       fs_bulk);
+	return &mh_bulk->mhs_data[pool_index.index];
+}
+
+void mlx5_fs_hws_mh_pool_release_mh(struct mlx5_fs_pool *mh_pool,
+				    struct mlx5_fs_hws_mh *mh_data)
+{
+	struct mlx5_fs_bulk *fs_bulk = &mh_data->bulk->fs_bulk;
+	struct mlx5_fs_pool_index pool_index = {};
+	struct mlx5_core_dev *dev = mh_pool->dev;
+
+	pool_index.fs_bulk = fs_bulk;
+	pool_index.index = mh_data->offset;
+	if (mlx5_fs_pool_release_index(mh_pool, &pool_index))
+		mlx5_core_warn(dev, "Attempted to release modify header which is not acquired\n");
+}
+
+bool mlx5_fs_hws_mh_pool_match(struct mlx5_fs_pool *mh_pool,
+			       struct mlx5hws_action_mh_pattern *pattern)
+{
+	struct mlx5hws_action_mh_pattern *pool_pattern;
+	int num_actions, i;
+
+	pool_pattern = mh_pool->pool_ctx;
+	if (WARN_ON_ONCE(!pool_pattern))
+		return false;
+
+	if (pattern->sz != pool_pattern->sz)
+		return false;
+	num_actions = pattern->sz / MLX5_UN_SZ_BYTES(set_add_copy_action_in_auto);
+	for (i = 0; i < num_actions; i++) {
+		if ((__force __be32)pattern->data[i] !=
+		    (__force __be32)pool_pattern->data[i])
+			return false;
+	}
+	return true;
+}
+
+struct mlx5hws_action *mlx5_fc_get_hws_action(struct mlx5hws_context *ctx,
+					      struct mlx5_fc *counter)
+{
+	u32 flags = MLX5HWS_ACTION_FLAG_HWS_FDB | MLX5HWS_ACTION_FLAG_SHARED;
+	struct mlx5_fc_bulk *fc_bulk = counter->bulk;
+	struct mlx5_fc_bulk_hws_data *fc_bulk_hws;
+
+	fc_bulk_hws = &fc_bulk->hws_data;
+	/* try avoid locking if not necessary */
+	if (refcount_inc_not_zero(&fc_bulk_hws->hws_action_refcount))
+		return fc_bulk_hws->hws_action;
+
+	mutex_lock(&fc_bulk_hws->lock);
+	if (refcount_inc_not_zero(&fc_bulk_hws->hws_action_refcount)) {
+		mutex_unlock(&fc_bulk_hws->lock);
+		return fc_bulk_hws->hws_action;
+	}
+	fc_bulk_hws->hws_action =
+		mlx5hws_action_create_counter(ctx, fc_bulk->base_id, flags);
+	if (!fc_bulk_hws->hws_action) {
+		mutex_unlock(&fc_bulk_hws->lock);
+		return NULL;
+	}
+	refcount_set(&fc_bulk_hws->hws_action_refcount, 1);
+	mutex_unlock(&fc_bulk_hws->lock);
+
+	return fc_bulk_hws->hws_action;
+}
+
+void mlx5_fc_put_hws_action(struct mlx5_fc *counter)
+{
+	struct mlx5_fc_bulk_hws_data *fc_bulk_hws = &counter->bulk->hws_data;
+
+	/* try avoid locking if not necessary */
+	if (refcount_dec_not_one(&fc_bulk_hws->hws_action_refcount))
+		return;
+
+	mutex_lock(&fc_bulk_hws->lock);
+	if (!refcount_dec_and_test(&fc_bulk_hws->hws_action_refcount)) {
+		mutex_unlock(&fc_bulk_hws->lock);
+		return;
+	}
+	mlx5hws_action_destroy(fc_bulk_hws->hws_action);
+	fc_bulk_hws->hws_action = NULL;
+	mutex_unlock(&fc_bulk_hws->lock);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/fs_hws_pools.h b/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/fs_hws_pools.h
new file mode 100644
index 000000000000..34072551dd21
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/fs_hws_pools.h
@@ -0,0 +1,73 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2025 NVIDIA Corporation & Affiliates */
+
+#ifndef __MLX5_FS_HWS_POOLS_H__
+#define __MLX5_FS_HWS_POOLS_H__
+
+#include <linux/if_vlan.h>
+#include "fs_pool.h"
+#include "fs_core.h"
+
+#define MLX5_FS_INSERT_HDR_VLAN_ANCHOR MLX5_REFORMAT_CONTEXT_ANCHOR_MAC_START
+#define MLX5_FS_INSERT_HDR_VLAN_OFFSET offsetof(struct vlan_ethhdr, h_vlan_proto)
+#define MLX5_FS_INSERT_HDR_VLAN_SIZE sizeof(struct vlan_hdr)
+
+enum {
+	MLX5_FS_DL3TNLTOL2_MAC_HDR_IDX = 0,
+	MLX5_FS_DL3TNLTOL2_MAC_VLAN_HDR_IDX,
+};
+
+struct mlx5_fs_hws_pr {
+	struct mlx5_fs_hws_pr_bulk *bulk;
+	u32 offset;
+	u8 hdr_idx;
+	u8 *data;
+	size_t data_size;
+};
+
+struct mlx5_fs_hws_pr_bulk {
+	struct mlx5_fs_bulk fs_bulk;
+	struct mlx5hws_action *hws_action;
+	struct mlx5_fs_hws_pr prs_data[];
+};
+
+struct mlx5_fs_hws_pr_pool_ctx {
+	enum mlx5hws_action_type reformat_type;
+	size_t encap_data_size;
+};
+
+struct mlx5_fs_hws_mh {
+	struct mlx5_fs_hws_mh_bulk *bulk;
+	u32 offset;
+	u8 *data;
+};
+
+struct mlx5_fs_hws_mh_bulk {
+	struct mlx5_fs_bulk fs_bulk;
+	struct mlx5_fs_pool *mh_pool;
+	struct mlx5hws_action *hws_action;
+	struct mlx5_fs_hws_mh mhs_data[];
+};
+
+int mlx5_fs_hws_pr_pool_init(struct mlx5_fs_pool *pr_pool,
+			     struct mlx5_core_dev *dev, size_t encap_data_size,
+			     enum mlx5hws_action_type reformat_type);
+void mlx5_fs_hws_pr_pool_cleanup(struct mlx5_fs_pool *pr_pool);
+
+struct mlx5_fs_hws_pr *mlx5_fs_hws_pr_pool_acquire_pr(struct mlx5_fs_pool *pr_pool);
+void mlx5_fs_hws_pr_pool_release_pr(struct mlx5_fs_pool *pr_pool,
+				    struct mlx5_fs_hws_pr *pr_data);
+struct mlx5hws_action *mlx5_fs_hws_pr_get_action(struct mlx5_fs_hws_pr *pr_data);
+int mlx5_fs_hws_mh_pool_init(struct mlx5_fs_pool *fs_hws_mh_pool,
+			     struct mlx5_core_dev *dev,
+			     struct mlx5hws_action_mh_pattern *pattern);
+void mlx5_fs_hws_mh_pool_cleanup(struct mlx5_fs_pool *fs_hws_mh_pool);
+struct mlx5_fs_hws_mh *mlx5_fs_hws_mh_pool_acquire_mh(struct mlx5_fs_pool *mh_pool);
+void mlx5_fs_hws_mh_pool_release_mh(struct mlx5_fs_pool *mh_pool,
+				    struct mlx5_fs_hws_mh *mh_data);
+bool mlx5_fs_hws_mh_pool_match(struct mlx5_fs_pool *mh_pool,
+			       struct mlx5hws_action_mh_pattern *pattern);
+struct mlx5hws_action *mlx5_fc_get_hws_action(struct mlx5hws_context *ctx,
+					      struct mlx5_fc *counter);
+void mlx5_fc_put_hws_action(struct mlx5_fc *counter);
+#endif /* __MLX5_FS_HWS_POOLS_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/internal.h b/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/internal.h
index 3c8635f286ce..30ccd635b505 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/internal.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/internal.h
@@ -39,7 +39,6 @@
 #define mlx5hws_dbg(ctx, arg...) mlx5_core_dbg((ctx)->mdev, ##arg)
 
 #define MLX5HWS_TABLE_TYPE_BASE 2
-#define MLX5HWS_ACTION_STE_IDX_ANY 0
 
 static inline bool is_mem_zero(const u8 *mem, size_t size)
 {
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/matcher.c b/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/matcher.c
index 4419c72ad314..80157a29a076 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/matcher.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/matcher.c
@@ -200,7 +200,7 @@ static void hws_matcher_set_rtc_attr_sz(struct mlx5hws_matcher *matcher,
 					enum mlx5hws_matcher_rtc_type rtc_type,
 					bool is_mirror)
 {
-	struct mlx5hws_pool_chunk *ste = &matcher->action_ste[MLX5HWS_ACTION_STE_IDX_ANY].ste;
+	struct mlx5hws_pool_chunk *ste = &matcher->action_ste.ste;
 	enum mlx5hws_matcher_flow_src flow_src = matcher->attr.optimize_flow_src;
 	bool is_match_rtc = rtc_type == HWS_MATCHER_RTC_TYPE_MATCH;
 
@@ -217,8 +217,7 @@ static void hws_matcher_set_rtc_attr_sz(struct mlx5hws_matcher *matcher,
 }
 
 static int hws_matcher_create_rtc(struct mlx5hws_matcher *matcher,
-				  enum mlx5hws_matcher_rtc_type rtc_type,
-				  u8 action_ste_selector)
+				  enum mlx5hws_matcher_rtc_type rtc_type)
 {
 	struct mlx5hws_matcher_attr *attr = &matcher->attr;
 	struct mlx5hws_cmd_rtc_create_attr rtc_attr = {0};
@@ -278,14 +277,19 @@ static int hws_matcher_create_rtc(struct mlx5hws_matcher *matcher,
 		break;
 
 	case HWS_MATCHER_RTC_TYPE_STE_ARRAY:
-		action_ste = &matcher->action_ste[action_ste_selector];
+		action_ste = &matcher->action_ste;
 
 		rtc_0_id = &action_ste->rtc_0_id;
 		rtc_1_id = &action_ste->rtc_1_id;
 		ste_pool = action_ste->pool;
 		ste = &action_ste->ste;
+		/* Action RTC size calculation:
+		 * log((max number of rules in matcher) *
+		 *     (max number of action STEs per rule) *
+		 *     (2 to support writing new STEs for update rule))
+		 */
 		ste->order = ilog2(roundup_pow_of_two(action_ste->max_stes)) +
-			     attr->table.sz_row_log;
+			     attr->table.sz_row_log + 1;
 		rtc_attr.log_size = ste->order;
 		rtc_attr.log_depth = 0;
 		rtc_attr.update_index_mode = MLX5_IFC_RTC_STE_UPDATE_MODE_BY_OFFSET;
@@ -350,8 +354,7 @@ free_ste:
 }
 
 static void hws_matcher_destroy_rtc(struct mlx5hws_matcher *matcher,
-				    enum mlx5hws_matcher_rtc_type rtc_type,
-				    u8 action_ste_selector)
+				    enum mlx5hws_matcher_rtc_type rtc_type)
 {
 	struct mlx5hws_matcher_action_ste *action_ste;
 	struct mlx5hws_table *tbl = matcher->tbl;
@@ -367,7 +370,7 @@ static void hws_matcher_destroy_rtc(struct mlx5hws_matcher *matcher,
 		ste = &matcher->match_ste.ste;
 		break;
 	case HWS_MATCHER_RTC_TYPE_STE_ARRAY:
-		action_ste = &matcher->action_ste[action_ste_selector];
+		action_ste = &matcher->action_ste;
 		rtc_0_id = action_ste->rtc_0_id;
 		rtc_1_id = action_ste->rtc_1_id;
 		ste_pool = action_ste->pool;
@@ -458,20 +461,13 @@ static int hws_matcher_resize_init(struct mlx5hws_matcher *src_matcher)
 	if (!resize_data)
 		return -ENOMEM;
 
-	resize_data->max_stes = src_matcher->action_ste[MLX5HWS_ACTION_STE_IDX_ANY].max_stes;
-
-	resize_data->action_ste[0].stc = src_matcher->action_ste[0].stc;
-	resize_data->action_ste[0].rtc_0_id = src_matcher->action_ste[0].rtc_0_id;
-	resize_data->action_ste[0].rtc_1_id = src_matcher->action_ste[0].rtc_1_id;
-	resize_data->action_ste[0].pool = src_matcher->action_ste[0].max_stes ?
-					  src_matcher->action_ste[0].pool :
-					  NULL;
-	resize_data->action_ste[1].stc = src_matcher->action_ste[1].stc;
-	resize_data->action_ste[1].rtc_0_id = src_matcher->action_ste[1].rtc_0_id;
-	resize_data->action_ste[1].rtc_1_id = src_matcher->action_ste[1].rtc_1_id;
-	resize_data->action_ste[1].pool = src_matcher->action_ste[1].max_stes ?
-					  src_matcher->action_ste[1].pool :
-					   NULL;
+	resize_data->max_stes = src_matcher->action_ste.max_stes;
+
+	resize_data->stc = src_matcher->action_ste.stc;
+	resize_data->rtc_0_id = src_matcher->action_ste.rtc_0_id;
+	resize_data->rtc_1_id = src_matcher->action_ste.rtc_1_id;
+	resize_data->pool = src_matcher->action_ste.max_stes ?
+			    src_matcher->action_ste.pool : NULL;
 
 	/* Place the new resized matcher on the dst matcher's list */
 	list_add(&resize_data->list_node, &src_matcher->resize_dst->resize_data);
@@ -504,49 +500,68 @@ static void hws_matcher_resize_uninit(struct mlx5hws_matcher *matcher)
 		if (resize_data->max_stes) {
 			mlx5hws_action_free_single_stc(matcher->tbl->ctx,
 						       matcher->tbl->type,
-						       &resize_data->action_ste[1].stc);
-			mlx5hws_action_free_single_stc(matcher->tbl->ctx,
-						       matcher->tbl->type,
-						       &resize_data->action_ste[0].stc);
+						       &resize_data->stc);
 
-			if (matcher->tbl->type == MLX5HWS_TABLE_TYPE_FDB) {
+			if (matcher->tbl->type == MLX5HWS_TABLE_TYPE_FDB)
 				mlx5hws_cmd_rtc_destroy(matcher->tbl->ctx->mdev,
-							resize_data->action_ste[1].rtc_1_id);
-				mlx5hws_cmd_rtc_destroy(matcher->tbl->ctx->mdev,
-							resize_data->action_ste[0].rtc_1_id);
-			}
-			mlx5hws_cmd_rtc_destroy(matcher->tbl->ctx->mdev,
-						resize_data->action_ste[1].rtc_0_id);
+							resize_data->rtc_1_id);
+
 			mlx5hws_cmd_rtc_destroy(matcher->tbl->ctx->mdev,
-						resize_data->action_ste[0].rtc_0_id);
-			if (resize_data->action_ste[MLX5HWS_ACTION_STE_IDX_ANY].pool) {
-				mlx5hws_pool_destroy(resize_data->action_ste[1].pool);
-				mlx5hws_pool_destroy(resize_data->action_ste[0].pool);
-			}
+						resize_data->rtc_0_id);
+
+			if (resize_data->pool)
+				mlx5hws_pool_destroy(resize_data->pool);
 		}
 
 		kfree(resize_data);
 	}
 }
 
-static int
-hws_matcher_bind_at_idx(struct mlx5hws_matcher *matcher, u8 action_ste_selector)
+static int hws_matcher_bind_at(struct mlx5hws_matcher *matcher)
 {
+	bool is_jumbo = mlx5hws_matcher_mt_is_jumbo(matcher->mt);
 	struct mlx5hws_cmd_stc_modify_attr stc_attr = {0};
 	struct mlx5hws_matcher_action_ste *action_ste;
 	struct mlx5hws_table *tbl = matcher->tbl;
 	struct mlx5hws_pool_attr pool_attr = {0};
 	struct mlx5hws_context *ctx = tbl->ctx;
-	int ret;
+	u32 required_stes;
+	u8 max_stes = 0;
+	int i, ret;
+
+	if (matcher->flags & MLX5HWS_MATCHER_FLAGS_COLLISION)
+		return 0;
+
+	for (i = 0; i < matcher->num_of_at; i++) {
+		struct mlx5hws_action_template *at = &matcher->at[i];
 
-	action_ste = &matcher->action_ste[action_ste_selector];
+		ret = hws_matcher_check_and_process_at(matcher, at);
+		if (ret) {
+			mlx5hws_err(ctx, "Invalid at %d", i);
+			return ret;
+		}
+
+		required_stes = at->num_of_action_stes - (!is_jumbo || at->only_term);
+		max_stes = max(max_stes, required_stes);
+
+		/* Future: Optimize reparse */
+	}
+
+	/* There are no additional STEs required for matcher */
+	if (!max_stes)
+		return 0;
+
+	matcher->action_ste.max_stes = max_stes;
+
+	action_ste = &matcher->action_ste;
 
 	/* Allocate action STE mempool */
 	pool_attr.table_type = tbl->type;
 	pool_attr.pool_type = MLX5HWS_POOL_TYPE_STE;
 	pool_attr.flags = MLX5HWS_POOL_FLAGS_FOR_STE_ACTION_POOL;
+	/* Pool size is similar to action RTC size */
 	pool_attr.alloc_log_sz = ilog2(roundup_pow_of_two(action_ste->max_stes)) +
-				 matcher->attr.table.sz_row_log;
+				 matcher->attr.table.sz_row_log + 1;
 	hws_matcher_set_pool_attr(&pool_attr, matcher);
 	action_ste->pool = mlx5hws_pool_create(ctx, &pool_attr);
 	if (!action_ste->pool) {
@@ -555,7 +570,7 @@ hws_matcher_bind_at_idx(struct mlx5hws_matcher *matcher, u8 action_ste_selector)
 	}
 
 	/* Allocate action RTC */
-	ret = hws_matcher_create_rtc(matcher, HWS_MATCHER_RTC_TYPE_STE_ARRAY, action_ste_selector);
+	ret = hws_matcher_create_rtc(matcher, HWS_MATCHER_RTC_TYPE_STE_ARRAY);
 	if (ret) {
 		mlx5hws_err(ctx, "Failed to create action RTC\n");
 		goto free_ste_pool;
@@ -579,18 +594,18 @@ hws_matcher_bind_at_idx(struct mlx5hws_matcher *matcher, u8 action_ste_selector)
 	return 0;
 
 free_rtc:
-	hws_matcher_destroy_rtc(matcher, HWS_MATCHER_RTC_TYPE_STE_ARRAY, action_ste_selector);
+	hws_matcher_destroy_rtc(matcher, HWS_MATCHER_RTC_TYPE_STE_ARRAY);
 free_ste_pool:
 	mlx5hws_pool_destroy(action_ste->pool);
 	return ret;
 }
 
-static void hws_matcher_unbind_at_idx(struct mlx5hws_matcher *matcher, u8 action_ste_selector)
+static void hws_matcher_unbind_at(struct mlx5hws_matcher *matcher)
 {
 	struct mlx5hws_matcher_action_ste *action_ste;
 	struct mlx5hws_table *tbl = matcher->tbl;
 
-	action_ste = &matcher->action_ste[action_ste_selector];
+	action_ste = &matcher->action_ste;
 
 	if (!action_ste->max_stes ||
 	    matcher->flags & MLX5HWS_MATCHER_FLAGS_COLLISION ||
@@ -598,65 +613,10 @@ static void hws_matcher_unbind_at_idx(struct mlx5hws_matcher *matcher, u8 action
 		return;
 
 	mlx5hws_action_free_single_stc(tbl->ctx, tbl->type, &action_ste->stc);
-	hws_matcher_destroy_rtc(matcher, HWS_MATCHER_RTC_TYPE_STE_ARRAY, action_ste_selector);
+	hws_matcher_destroy_rtc(matcher, HWS_MATCHER_RTC_TYPE_STE_ARRAY);
 	mlx5hws_pool_destroy(action_ste->pool);
 }
 
-static int hws_matcher_bind_at(struct mlx5hws_matcher *matcher)
-{
-	bool is_jumbo = mlx5hws_matcher_mt_is_jumbo(matcher->mt);
-	struct mlx5hws_table *tbl = matcher->tbl;
-	struct mlx5hws_context *ctx = tbl->ctx;
-	u32 required_stes;
-	u8 max_stes = 0;
-	int i, ret;
-
-	if (matcher->flags & MLX5HWS_MATCHER_FLAGS_COLLISION)
-		return 0;
-
-	for (i = 0; i < matcher->num_of_at; i++) {
-		struct mlx5hws_action_template *at = &matcher->at[i];
-
-		ret = hws_matcher_check_and_process_at(matcher, at);
-		if (ret) {
-			mlx5hws_err(ctx, "Invalid at %d", i);
-			return ret;
-		}
-
-		required_stes = at->num_of_action_stes - (!is_jumbo || at->only_term);
-		max_stes = max(max_stes, required_stes);
-
-		/* Future: Optimize reparse */
-	}
-
-	/* There are no additional STEs required for matcher */
-	if (!max_stes)
-		return 0;
-
-	matcher->action_ste[0].max_stes = max_stes;
-	matcher->action_ste[1].max_stes = max_stes;
-
-	ret = hws_matcher_bind_at_idx(matcher, 0);
-	if (ret)
-		return ret;
-
-	ret = hws_matcher_bind_at_idx(matcher, 1);
-	if (ret)
-		goto free_at_0;
-
-	return 0;
-
-free_at_0:
-	hws_matcher_unbind_at_idx(matcher, 0);
-	return ret;
-}
-
-static void hws_matcher_unbind_at(struct mlx5hws_matcher *matcher)
-{
-	hws_matcher_unbind_at_idx(matcher, 1);
-	hws_matcher_unbind_at_idx(matcher, 0);
-}
-
 static int hws_matcher_bind_mt(struct mlx5hws_matcher *matcher)
 {
 	struct mlx5hws_context *ctx = matcher->tbl->ctx;
@@ -802,7 +762,7 @@ static int hws_matcher_create_and_connect(struct mlx5hws_matcher *matcher)
 		goto unbind_at;
 
 	/* Allocate the RTC for the new matcher */
-	ret = hws_matcher_create_rtc(matcher, HWS_MATCHER_RTC_TYPE_MATCH, 0);
+	ret = hws_matcher_create_rtc(matcher, HWS_MATCHER_RTC_TYPE_MATCH);
 	if (ret)
 		goto destroy_end_ft;
 
@@ -814,7 +774,7 @@ static int hws_matcher_create_and_connect(struct mlx5hws_matcher *matcher)
 	return 0;
 
 destroy_rtc:
-	hws_matcher_destroy_rtc(matcher, HWS_MATCHER_RTC_TYPE_MATCH, 0);
+	hws_matcher_destroy_rtc(matcher, HWS_MATCHER_RTC_TYPE_MATCH);
 destroy_end_ft:
 	hws_matcher_destroy_end_ft(matcher);
 unbind_at:
@@ -828,7 +788,7 @@ static void hws_matcher_destroy_and_disconnect(struct mlx5hws_matcher *matcher)
 {
 	hws_matcher_resize_uninit(matcher);
 	hws_matcher_disconnect(matcher);
-	hws_matcher_destroy_rtc(matcher, HWS_MATCHER_RTC_TYPE_MATCH, 0);
+	hws_matcher_destroy_rtc(matcher, HWS_MATCHER_RTC_TYPE_MATCH);
 	hws_matcher_destroy_end_ft(matcher);
 	hws_matcher_unbind_at(matcher);
 	hws_matcher_unbind_mt(matcher);
@@ -962,10 +922,9 @@ int mlx5hws_matcher_attach_at(struct mlx5hws_matcher *matcher,
 		return ret;
 
 	required_stes = at->num_of_action_stes - (!is_jumbo || at->only_term);
-	if (matcher->action_ste[MLX5HWS_ACTION_STE_IDX_ANY].max_stes < required_stes) {
+	if (matcher->action_ste.max_stes < required_stes) {
 		mlx5hws_dbg(ctx, "Required STEs [%d] exceeds initial action template STE [%d]\n",
-			    required_stes,
-			    matcher->action_ste[MLX5HWS_ACTION_STE_IDX_ANY].max_stes);
+			    required_stes, matcher->action_ste.max_stes);
 		return -ENOMEM;
 	}
 
@@ -1149,8 +1108,7 @@ static int hws_matcher_resize_precheck(struct mlx5hws_matcher *src_matcher,
 		return -EINVAL;
 	}
 
-	if (src_matcher->action_ste[MLX5HWS_ACTION_STE_IDX_ANY].max_stes >
-	    dst_matcher->action_ste[0].max_stes) {
+	if (src_matcher->action_ste.max_stes > dst_matcher->action_ste.max_stes) {
 		mlx5hws_err(ctx, "Src/dst matcher max STEs mismatch\n");
 		return -EINVAL;
 	}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/matcher.h b/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/matcher.h
index 81ff487f57be..cff4ae854a79 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/matcher.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/matcher.h
@@ -52,15 +52,11 @@ struct mlx5hws_matcher_action_ste {
 	u8 max_stes;
 };
 
-struct mlx5hws_matcher_resize_data_node {
+struct mlx5hws_matcher_resize_data {
 	struct mlx5hws_pool_chunk stc;
 	u32 rtc_0_id;
 	u32 rtc_1_id;
 	struct mlx5hws_pool *pool;
-};
-
-struct mlx5hws_matcher_resize_data {
-	struct mlx5hws_matcher_resize_data_node action_ste[2];
 	u8 max_stes;
 	struct list_head list_node;
 };
@@ -78,7 +74,7 @@ struct mlx5hws_matcher {
 	struct mlx5hws_matcher *col_matcher;
 	struct mlx5hws_matcher *resize_dst;
 	struct mlx5hws_matcher_match_ste match_ste;
-	struct mlx5hws_matcher_action_ste action_ste[2];
+	struct mlx5hws_matcher_action_ste action_ste;
 	struct list_head list_node;
 	struct list_head resize_data;
 };
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/rule.c b/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/rule.c
index 14f6307a1772..a27a2d5ffc7b 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/rule.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/rule.c
@@ -129,27 +129,18 @@ static void hws_rule_gen_comp(struct mlx5hws_send_engine *queue,
 
 static void
 hws_rule_save_resize_info(struct mlx5hws_rule *rule,
-			  struct mlx5hws_send_ste_attr *ste_attr,
-			  bool is_update)
+			  struct mlx5hws_send_ste_attr *ste_attr)
 {
 	if (!mlx5hws_matcher_is_resizable(rule->matcher))
 		return;
 
-	if (likely(!is_update)) {
+	/* resize_info might already exist (if we're in update flow) */
+	if (likely(!rule->resize_info)) {
 		rule->resize_info = kzalloc(sizeof(*rule->resize_info), GFP_KERNEL);
 		if (unlikely(!rule->resize_info)) {
 			pr_warn("HWS: resize info isn't allocated for rule\n");
 			return;
 		}
-
-		rule->resize_info->max_stes =
-			rule->matcher->action_ste[MLX5HWS_ACTION_STE_IDX_ANY].max_stes;
-		rule->resize_info->action_ste_pool[0] = rule->matcher->action_ste[0].max_stes ?
-							rule->matcher->action_ste[0].pool :
-							NULL;
-		rule->resize_info->action_ste_pool[1] = rule->matcher->action_ste[1].max_stes ?
-							rule->matcher->action_ste[1].pool :
-							NULL;
 	}
 
 	memcpy(rule->resize_info->ctrl_seg, ste_attr->wqe_ctrl,
@@ -204,15 +195,14 @@ hws_rule_load_delete_info(struct mlx5hws_rule *rule,
 	}
 }
 
-static int hws_rule_alloc_action_ste_idx(struct mlx5hws_rule *rule,
-					 u8 action_ste_selector)
+static int hws_rule_alloc_action_ste(struct mlx5hws_rule *rule)
 {
 	struct mlx5hws_matcher *matcher = rule->matcher;
 	struct mlx5hws_matcher_action_ste *action_ste;
 	struct mlx5hws_pool_chunk ste = {0};
 	int ret;
 
-	action_ste = &matcher->action_ste[action_ste_selector];
+	action_ste = &matcher->action_ste;
 	ste.order = ilog2(roundup_pow_of_two(action_ste->max_stes));
 	ret = mlx5hws_pool_chunk_alloc(action_ste->pool, &ste);
 	if (unlikely(ret)) {
@@ -220,68 +210,29 @@ static int hws_rule_alloc_action_ste_idx(struct mlx5hws_rule *rule,
 			    "Failed to allocate STE for rule actions");
 		return ret;
 	}
-	rule->action_ste_idx = ste.offset;
+
+	rule->action_ste.pool = matcher->action_ste.pool;
+	rule->action_ste.num_stes = matcher->action_ste.max_stes;
+	rule->action_ste.index = ste.offset;
 
 	return 0;
 }
 
-static void hws_rule_free_action_ste_idx(struct mlx5hws_rule *rule,
-					 u8 action_ste_selector)
+void mlx5hws_rule_free_action_ste(struct mlx5hws_rule_action_ste_info *action_ste)
 {
-	struct mlx5hws_matcher *matcher = rule->matcher;
 	struct mlx5hws_pool_chunk ste = {0};
-	struct mlx5hws_pool *pool;
-	u8 max_stes;
-
-	if (mlx5hws_matcher_is_resizable(matcher)) {
-		/* Free the original action pool if rule was resized */
-		max_stes = rule->resize_info->max_stes;
-		pool = rule->resize_info->action_ste_pool[action_ste_selector];
-	} else {
-		max_stes = matcher->action_ste[action_ste_selector].max_stes;
-		pool = matcher->action_ste[action_ste_selector].pool;
-	}
-
-	/* This release is safe only when the rule match part was deleted */
-	ste.order = ilog2(roundup_pow_of_two(max_stes));
-	ste.offset = rule->action_ste_idx;
-
-	mlx5hws_pool_chunk_free(pool, &ste);
-}
 
-static int hws_rule_alloc_action_ste(struct mlx5hws_rule *rule,
-				     struct mlx5hws_rule_attr *attr)
-{
-	int action_ste_idx;
-	int ret;
-
-	ret = hws_rule_alloc_action_ste_idx(rule, 0);
-	if (unlikely(ret))
-		return ret;
-
-	action_ste_idx = rule->action_ste_idx;
-
-	ret = hws_rule_alloc_action_ste_idx(rule, 1);
-	if (unlikely(ret)) {
-		hws_rule_free_action_ste_idx(rule, 0);
-		return ret;
-	}
-
-	/* Both pools have to return the same index */
-	if (unlikely(rule->action_ste_idx != action_ste_idx)) {
-		pr_warn("HWS: allocation of action STE failed - pool indexes mismatch\n");
-		return -EINVAL;
-	}
+	if (!action_ste->num_stes)
+		return;
 
-	return 0;
-}
+	ste.order = ilog2(roundup_pow_of_two(action_ste->num_stes));
+	ste.offset = action_ste->index;
 
-void mlx5hws_rule_free_action_ste(struct mlx5hws_rule *rule)
-{
-	if (rule->action_ste_idx > -1) {
-		hws_rule_free_action_ste_idx(rule, 1);
-		hws_rule_free_action_ste_idx(rule, 0);
-	}
+	/* This release is safe only when the rule match STE was deleted
+	 * (when the rule is being deleted) or replaced with the new STE that
+	 * isn't pointing to old action STEs (when the rule is being updated).
+	 */
+	mlx5hws_pool_chunk_free(action_ste->pool, &ste);
 }
 
 static void hws_rule_create_init(struct mlx5hws_rule *rule,
@@ -298,14 +249,24 @@ static void hws_rule_create_init(struct mlx5hws_rule *rule,
 		/* In update we use these rtc's */
 		rule->rtc_0 = 0;
 		rule->rtc_1 = 0;
-		rule->action_ste_selector = 0;
+
+		rule->action_ste.pool = NULL;
+		rule->action_ste.num_stes = 0;
+		rule->action_ste.index = -1;
+
+		rule->status = MLX5HWS_RULE_STATUS_CREATING;
 	} else {
-		rule->action_ste_selector = !rule->action_ste_selector;
+		rule->status = MLX5HWS_RULE_STATUS_UPDATING;
 	}
 
+	/* Initialize the old action STE info - shallow-copy action_ste.
+	 * In create flow this will set old_action_ste fields to initial values.
+	 * In update flow this will save the existing action STE info,
+	 * so that we will later use it to free old STEs.
+	 */
+	rule->old_action_ste = rule->action_ste;
+
 	rule->pending_wqes = 0;
-	rule->action_ste_idx = -1;
-	rule->status = MLX5HWS_RULE_STATUS_CREATING;
 
 	/* Init default send STE attributes */
 	ste_attr->gta_opcode = MLX5HWS_WQE_GTA_OP_ACTIVATE;
@@ -316,7 +277,7 @@ static void hws_rule_create_init(struct mlx5hws_rule *rule,
 	/* Init default action apply */
 	apply->tbl_type = tbl->type;
 	apply->common_res = &ctx->common_res;
-	apply->jump_to_action_stc = matcher->action_ste[0].stc.offset;
+	apply->jump_to_action_stc = matcher->action_ste.stc.offset;
 	apply->require_dep = 0;
 }
 
@@ -332,8 +293,6 @@ static void hws_rule_move_init(struct mlx5hws_rule *rule,
 	rule->rtc_1 = 0;
 
 	rule->pending_wqes = 0;
-	rule->action_ste_idx = -1;
-	rule->action_ste_selector = 0;
 	rule->status = MLX5HWS_RULE_STATUS_CREATING;
 	rule->resize_info->state = MLX5HWS_RULE_RESIZE_STATE_WRITING;
 }
@@ -394,21 +353,17 @@ static int hws_rule_create_hws(struct mlx5hws_rule *rule,
 
 	if (action_stes) {
 		/* Allocate action STEs for rules that need more than match STE */
-		if (!is_update) {
-			ret = hws_rule_alloc_action_ste(rule, attr);
-			if (ret) {
-				mlx5hws_err(ctx, "Failed to allocate action memory %d", ret);
-				mlx5hws_send_abort_new_dep_wqe(queue);
-				return ret;
-			}
+		ret = hws_rule_alloc_action_ste(rule);
+		if (ret) {
+			mlx5hws_err(ctx, "Failed to allocate action memory %d", ret);
+			mlx5hws_send_abort_new_dep_wqe(queue);
+			return ret;
 		}
 		/* Skip RX/TX based on the dep_wqe init */
-		ste_attr.rtc_0 = dep_wqe->rtc_0 ?
-				 matcher->action_ste[rule->action_ste_selector].rtc_0_id : 0;
-		ste_attr.rtc_1 = dep_wqe->rtc_1 ?
-				 matcher->action_ste[rule->action_ste_selector].rtc_1_id : 0;
+		ste_attr.rtc_0 = dep_wqe->rtc_0 ? matcher->action_ste.rtc_0_id : 0;
+		ste_attr.rtc_1 = dep_wqe->rtc_1 ? matcher->action_ste.rtc_1_id : 0;
 		/* Action STEs are written to a specific index last to first */
-		ste_attr.direct_index = rule->action_ste_idx + action_stes;
+		ste_attr.direct_index = rule->action_ste.index + action_stes;
 		apply.next_direct_idx = ste_attr.direct_index;
 	} else {
 		apply.next_direct_idx = 0;
@@ -459,7 +414,7 @@ static int hws_rule_create_hws(struct mlx5hws_rule *rule,
 	if (!is_update)
 		hws_rule_save_delete_info(rule, &ste_attr);
 
-	hws_rule_save_resize_info(rule, &ste_attr, is_update);
+	hws_rule_save_resize_info(rule, &ste_attr);
 	mlx5hws_send_engine_inc_rule(queue);
 
 	if (!attr->burst)
@@ -480,7 +435,10 @@ static void hws_rule_destroy_failed_hws(struct mlx5hws_rule *rule,
 			  attr->user_data, MLX5HWS_RULE_STATUS_DELETED);
 
 	/* Rule failed now we can safely release action STEs */
-	mlx5hws_rule_free_action_ste(rule);
+	mlx5hws_rule_free_action_ste(&rule->action_ste);
+
+	/* Perhaps the rule failed updating - release old action STEs as well */
+	mlx5hws_rule_free_action_ste(&rule->old_action_ste);
 
 	/* Clear complex tag */
 	hws_rule_clear_delete_info(rule);
@@ -517,7 +475,8 @@ static int hws_rule_destroy_hws(struct mlx5hws_rule *rule,
 	}
 
 	/* Rule is not completed yet */
-	if (rule->status == MLX5HWS_RULE_STATUS_CREATING)
+	if (rule->status == MLX5HWS_RULE_STATUS_CREATING ||
+	    rule->status == MLX5HWS_RULE_STATUS_UPDATING)
 		return -EBUSY;
 
 	/* Rule failed and doesn't require cleanup */
@@ -534,7 +493,7 @@ static int hws_rule_destroy_hws(struct mlx5hws_rule *rule,
 		hws_rule_gen_comp(queue, rule, false,
 				  attr->user_data, MLX5HWS_RULE_STATUS_DELETED);
 
-		mlx5hws_rule_free_action_ste(rule);
+		mlx5hws_rule_free_action_ste(&rule->action_ste);
 		mlx5hws_rule_clear_resize_info(rule);
 		return 0;
 	}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/rule.h b/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/rule.h
index 495cdd17e9f3..b5ee94ac449b 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/rule.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/rule.h
@@ -15,6 +15,8 @@ enum mlx5hws_rule_status {
 	MLX5HWS_RULE_STATUS_UNKNOWN,
 	MLX5HWS_RULE_STATUS_CREATING,
 	MLX5HWS_RULE_STATUS_CREATED,
+	MLX5HWS_RULE_STATUS_UPDATING,
+	MLX5HWS_RULE_STATUS_UPDATED,
 	MLX5HWS_RULE_STATUS_DELETING,
 	MLX5HWS_RULE_STATUS_DELETED,
 	MLX5HWS_RULE_STATUS_FAILING,
@@ -41,13 +43,17 @@ struct mlx5hws_rule_match_tag {
 	};
 };
 
+struct mlx5hws_rule_action_ste_info {
+	struct mlx5hws_pool *pool;
+	int index; /* STE array index */
+	u8 num_stes;
+};
+
 struct mlx5hws_rule_resize_info {
-	struct mlx5hws_pool *action_ste_pool[2];
 	u32 rtc_0;
 	u32 rtc_1;
 	u32 rule_idx;
 	u8 state;
-	u8 max_stes;
 	u8 ctrl_seg[MLX5HWS_WQE_SZ_GTA_CTRL]; /* Ctrl segment of STE: 48 bytes */
 	u8 data_seg[MLX5HWS_WQE_SZ_GTA_DATA]; /* Data segment of STE: 64 bytes */
 };
@@ -58,18 +64,18 @@ struct mlx5hws_rule {
 		struct mlx5hws_rule_match_tag tag;
 		struct mlx5hws_rule_resize_info *resize_info;
 	};
+	struct mlx5hws_rule_action_ste_info action_ste;
+	struct mlx5hws_rule_action_ste_info old_action_ste;
 	u32 rtc_0; /* The RTC into which the STE was inserted */
 	u32 rtc_1; /* The RTC into which the STE was inserted */
-	int action_ste_idx; /* STE array index */
 	u8 status; /* enum mlx5hws_rule_status */
-	u8 action_ste_selector; /* For rule update - which action STE is in use */
 	u8 pending_wqes;
 	bool skip_delete; /* For complex rules - another rule with same tag
 			   * still exists, so don't actually delete this rule.
 			   */
 };
 
-void mlx5hws_rule_free_action_ste(struct mlx5hws_rule *rule);
+void mlx5hws_rule_free_action_ste(struct mlx5hws_rule_action_ste_info *action_ste);
 
 int mlx5hws_rule_move_hws_remove(struct mlx5hws_rule *rule,
 				 void *queue, void *user_data);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/send.c b/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/send.c
index c680b7f984e1..cb6abc4ab7df 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/send.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/send.c
@@ -377,17 +377,25 @@ static void hws_send_engine_update_rule(struct mlx5hws_send_engine *queue,
 
 			*status = MLX5HWS_FLOW_OP_ERROR;
 		} else {
-			/* Increase the status, this only works on good flow as the enum
-			 * is arrange it away creating -> created -> deleting -> deleted
+			/* Increase the status, this only works on good flow as
+			 * the enum is arranged this way:
+			 *  - creating -> created
+			 *  - updating -> updated
+			 *  - deleting -> deleted
 			 */
 			priv->rule->status++;
 			*status = MLX5HWS_FLOW_OP_SUCCESS;
-			/* Rule was deleted now we can safely release action STEs
-			 * and clear resize info
-			 */
 			if (priv->rule->status == MLX5HWS_RULE_STATUS_DELETED) {
-				mlx5hws_rule_free_action_ste(priv->rule);
+				/* Rule was deleted, now we can safely release
+				 * action STEs and clear resize info
+				 */
+				mlx5hws_rule_free_action_ste(&priv->rule->action_ste);
 				mlx5hws_rule_clear_resize_info(priv->rule);
+			} else if (priv->rule->status == MLX5HWS_RULE_STATUS_UPDATED) {
+				/* Rule was updated, free the old action STEs */
+				mlx5hws_rule_free_action_ste(&priv->rule->old_action_ste);
+				/* Update completed - move the rule back to "created" */
+				priv->rule->status = MLX5HWS_RULE_STATUS_CREATED;
 			}
 		}
 	}
diff --git a/include/linux/mlx5/mlx5_ifc.h b/include/linux/mlx5/mlx5_ifc.h
index 370f533da107..bb99a35fc6a2 100644
--- a/include/linux/mlx5/mlx5_ifc.h
+++ b/include/linux/mlx5/mlx5_ifc.h
@@ -7025,6 +7025,7 @@ struct mlx5_ifc_alloc_packet_reformat_context_out_bits {
 
 enum {
 	MLX5_REFORMAT_CONTEXT_ANCHOR_MAC_START = 0x1,
+	MLX5_REFORMAT_CONTEXT_ANCHOR_VLAN_START = 0x2,
 	MLX5_REFORMAT_CONTEXT_ANCHOR_IP_START = 0x7,
 	MLX5_REFORMAT_CONTEXT_ANCHOR_TCP_UDP_START = 0x9,
 };