Merge branch 'ct-offload' of git://git.kernel.org/pub/scm/linux/kernel/git/saeed/linux

drivers/infiniband/hw/mlx5/main.c

@@ -3570,7 +3570,8 @@ static void mlx5_ib_set_rule_source_port(struct mlx5_ib_dev *dev,
 		misc = MLX5_ADDR_OF(fte_match_param, spec->match_criteria,
 				    misc_parameters_2);
 
-		MLX5_SET_TO_ONES(fte_match_set_misc2, misc, metadata_reg_c_0);
+		MLX5_SET(fte_match_set_misc2, misc, metadata_reg_c_0,
+			 mlx5_eswitch_get_vport_metadata_mask());
 	} else {
 		misc = MLX5_ADDR_OF(fte_match_param, spec->match_value,
 				    misc_parameters);

drivers/net/ethernet/mellanox/mlx5/core/Makefile

@@ -34,7 +34,7 @@ mlx5_core-$(CONFIG_MLX5_EN_ARFS)     += en_arfs.o
 mlx5_core-$(CONFIG_MLX5_EN_RXNFC)    += en_fs_ethtool.o
 mlx5_core-$(CONFIG_MLX5_CORE_EN_DCB) += en_dcbnl.o en/port_buffer.o
 mlx5_core-$(CONFIG_MLX5_ESWITCH)     += en_rep.o en_tc.o en/tc_tun.o lib/port_tun.o lag_mp.o \
-					lib/geneve.o en/tc_tun_vxlan.o en/tc_tun_gre.o \
+					lib/geneve.o en/mapping.o en/tc_tun_vxlan.o en/tc_tun_gre.o \
 					en/tc_tun_geneve.o diag/en_tc_tracepoint.o
 mlx5_core-$(CONFIG_PCI_HYPERV_INTERFACE) += en/hv_vhca_stats.o
 

drivers/net/ethernet/mellanox/mlx5/core/en/mapping.c

+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2018 Mellanox Technologies */
+
+#include <linux/jhash.h>
+#include <linux/slab.h>
+#include <linux/xarray.h>
+#include <linux/hashtable.h>
+
+#include "mapping.h"
+
+#define MAPPING_GRACE_PERIOD 2000
+
+struct mapping_ctx {
+	struct xarray xarray;
+	DECLARE_HASHTABLE(ht, 8);
+	struct mutex lock; /* Guards hashtable and xarray */
+	unsigned long max_id;
+	size_t data_size;
+	bool delayed_removal;
+	struct delayed_work dwork;
+	struct list_head pending_list;
+	spinlock_t pending_list_lock; /* Guards pending list */
+};
+
+struct mapping_item {
+	struct rcu_head rcu;
+	struct list_head list;
+	unsigned long timeout;
+	struct hlist_node node;
+	int cnt;
+	u32 id;
+	char data[];
+};
+
+int mapping_add(struct mapping_ctx *ctx, void *data, u32 *id)
+{
+	struct mapping_item *mi;
+	int err = -ENOMEM;
+	u32 hash_key;
+
+	mutex_lock(&ctx->lock);
+
+	hash_key = jhash(data, ctx->data_size, 0);
+	hash_for_each_possible(ctx->ht, mi, node, hash_key) {
+		if (!memcmp(data, mi->data, ctx->data_size))
+			goto attach;
+	}
+
+	mi = kzalloc(sizeof(*mi) + ctx->data_size, GFP_KERNEL);
+	if (!mi)
+		goto err_alloc;
+
+	memcpy(mi->data, data, ctx->data_size);
+	hash_add(ctx->ht, &mi->node, hash_key);
+
+	err = xa_alloc(&ctx->xarray, &mi->id, mi, XA_LIMIT(1, ctx->max_id),
+		       GFP_KERNEL);
+	if (err)
+		goto err_assign;
+attach:
+	++mi->cnt;
+	*id = mi->id;
+
+	mutex_unlock(&ctx->lock);
+
+	return 0;
+
+err_assign:
+	hash_del(&mi->node);
+	kfree(mi);
+err_alloc:
+	mutex_unlock(&ctx->lock);
+
+	return err;
+}
+
+static void mapping_remove_and_free(struct mapping_ctx *ctx,
+				    struct mapping_item *mi)
+{
+	xa_erase(&ctx->xarray, mi->id);
+	kfree_rcu(mi, rcu);
+}
+
+static void mapping_free_item(struct mapping_ctx *ctx,
+			      struct mapping_item *mi)
+{
+	if (!ctx->delayed_removal) {
+		mapping_remove_and_free(ctx, mi);
+		return;
+	}
+
+	mi->timeout = jiffies + msecs_to_jiffies(MAPPING_GRACE_PERIOD);
+
+	spin_lock(&ctx->pending_list_lock);
+	list_add_tail(&mi->list, &ctx->pending_list);
+	spin_unlock(&ctx->pending_list_lock);
+
+	schedule_delayed_work(&ctx->dwork, MAPPING_GRACE_PERIOD);
+}
+
+int mapping_remove(struct mapping_ctx *ctx, u32 id)
+{
+	unsigned long index = id;
+	struct mapping_item *mi;
+	int err = -ENOENT;
+
+	mutex_lock(&ctx->lock);
+	mi = xa_load(&ctx->xarray, index);
+	if (!mi)
+		goto out;
+	err = 0;
+
+	if (--mi->cnt > 0)
+		goto out;
+
+	hash_del(&mi->node);
+	mapping_free_item(ctx, mi);
+out:
+	mutex_unlock(&ctx->lock);
+
+	return err;
+}
+
+int mapping_find(struct mapping_ctx *ctx, u32 id, void *data)
+{
+	unsigned long index = id;
+	struct mapping_item *mi;
+	int err = -ENOENT;
+
+	rcu_read_lock();
+	mi = xa_load(&ctx->xarray, index);
+	if (!mi)
+		goto err_find;
+
+	memcpy(data, mi->data, ctx->data_size);
+	err = 0;
+
+err_find:
+	rcu_read_unlock();
+	return err;
+}
+
+static void
+mapping_remove_and_free_list(struct mapping_ctx *ctx, struct list_head *list)
+{
+	struct mapping_item *mi;
+
+	list_for_each_entry(mi, list, list)
+		mapping_remove_and_free(ctx, mi);
+}
+
+static void mapping_work_handler(struct work_struct *work)
+{
+	unsigned long min_timeout = 0, now = jiffies;
+	struct mapping_item *mi, *next;
+	LIST_HEAD(pending_items);
+	struct mapping_ctx *ctx;
+
+	ctx = container_of(work, struct mapping_ctx, dwork.work);
+
+	spin_lock(&ctx->pending_list_lock);
+	list_for_each_entry_safe(mi, next, &ctx->pending_list, list) {
+		if (time_after(now, mi->timeout))
+			list_move(&mi->list, &pending_items);
+		else if (!min_timeout ||
+			 time_before(mi->timeout, min_timeout))
+			min_timeout = mi->timeout;
+	}
+	spin_unlock(&ctx->pending_list_lock);
+
+	mapping_remove_and_free_list(ctx, &pending_items);
+
+	if (min_timeout)
+		schedule_delayed_work(&ctx->dwork, abs(min_timeout - now));
+}
+
+static void mapping_flush_work(struct mapping_ctx *ctx)
+{
+	if (!ctx->delayed_removal)
+		return;
+
+	cancel_delayed_work_sync(&ctx->dwork);
+	mapping_remove_and_free_list(ctx, &ctx->pending_list);
+}
+
+struct mapping_ctx *
+mapping_create(size_t data_size, u32 max_id, bool delayed_removal)
+{
+	struct mapping_ctx *ctx;
+
+	ctx = kzalloc(sizeof(*ctx), GFP_KERNEL);
+	if (!ctx)
+		return ERR_PTR(-ENOMEM);
+
+	ctx->max_id = max_id ? max_id : UINT_MAX;
+	ctx->data_size = data_size;
+
+	if (delayed_removal) {
+		INIT_DELAYED_WORK(&ctx->dwork, mapping_work_handler);
+		INIT_LIST_HEAD(&ctx->pending_list);
+		spin_lock_init(&ctx->pending_list_lock);
+		ctx->delayed_removal = true;
+	}
+
+	mutex_init(&ctx->lock);
+	xa_init_flags(&ctx->xarray, XA_FLAGS_ALLOC1);
+
+	return ctx;
+}
+
+void mapping_destroy(struct mapping_ctx *ctx)
+{
+	mapping_flush_work(ctx);
+	xa_destroy(&ctx->xarray);
+	mutex_destroy(&ctx->lock);
+
+	kfree(ctx);
+}

drivers/net/ethernet/mellanox/mlx5/core/en/mapping.h

+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2019 Mellanox Technologies */
+
+#ifndef __MLX5_MAPPING_H__
+#define __MLX5_MAPPING_H__
+
+struct mapping_ctx;
+
+int mapping_add(struct mapping_ctx *ctx, void *data, u32 *id);
+int mapping_remove(struct mapping_ctx *ctx, u32 id);
+int mapping_find(struct mapping_ctx *ctx, u32 id, void *data);
+
+/* mapping uses an xarray to map data to ids in add(), and for find().
+ * For locking, it uses a internal xarray spin lock for add()/remove(),
+ * find() uses rcu_read_lock().
+ * Choosing delayed_removal postpones the removal of a previously mapped
+ * id by MAPPING_GRACE_PERIOD milliseconds.
+ * This is to avoid races against hardware, where we mark the packet in
+ * hardware with a previous id, and quick remove() and add() reusing the same
+ * previous id. Then find() will get the new mapping instead of the old
+ * which was used to mark the packet.
+ */
+struct mapping_ctx *mapping_create(size_t data_size, u32 max_id,
+				   bool delayed_removal);
+void mapping_destroy(struct mapping_ctx *ctx);
+
+#endif /* __MLX5_MAPPING_H__ */

drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun.c

@@ -469,10 +469,15 @@ int mlx5e_tc_tun_parse(struct net_device *filter_dev,
 		       struct mlx5e_priv *priv,
 		       struct mlx5_flow_spec *spec,
 		       struct flow_cls_offload *f,
-		       void *headers_c,
-		       void *headers_v, u8 *match_level)
+		       u8 *match_level)
 {
 	struct mlx5e_tc_tunnel *tunnel = mlx5e_get_tc_tun(filter_dev);
+	struct flow_rule *rule = flow_cls_offload_flow_rule(f);
+	void *headers_c = MLX5_ADDR_OF(fte_match_param, spec->match_criteria,
+				       outer_headers);
+	void *headers_v = MLX5_ADDR_OF(fte_match_param, spec->match_value,
+				       outer_headers);
+	struct netlink_ext_ack *extack = f->common.extack;
 	int err = 0;
 
 	if (!tunnel) {
@@ -499,6 +504,109 @@ int mlx5e_tc_tun_parse(struct net_device *filter_dev,
 			goto out;
 	}
 
+	if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_ENC_CONTROL)) {
+		struct flow_match_control match;
+		u16 addr_type;
+
+		flow_rule_match_enc_control(rule, &match);
+		addr_type = match.key->addr_type;
+
+		/* For tunnel addr_type used same key id`s as for non-tunnel */
+		if (addr_type == FLOW_DISSECTOR_KEY_IPV4_ADDRS) {
+			struct flow_match_ipv4_addrs match;
+
+			flow_rule_match_enc_ipv4_addrs(rule, &match);
+			MLX5_SET(fte_match_set_lyr_2_4, headers_c,
+				 src_ipv4_src_ipv6.ipv4_layout.ipv4,
+				 ntohl(match.mask->src));
+			MLX5_SET(fte_match_set_lyr_2_4, headers_v,
+				 src_ipv4_src_ipv6.ipv4_layout.ipv4,
+				 ntohl(match.key->src));
+
+			MLX5_SET(fte_match_set_lyr_2_4, headers_c,
+				 dst_ipv4_dst_ipv6.ipv4_layout.ipv4,
+				 ntohl(match.mask->dst));
+			MLX5_SET(fte_match_set_lyr_2_4, headers_v,
+				 dst_ipv4_dst_ipv6.ipv4_layout.ipv4,
+				 ntohl(match.key->dst));
+
+			MLX5_SET_TO_ONES(fte_match_set_lyr_2_4, headers_c,
+					 ethertype);
+			MLX5_SET(fte_match_set_lyr_2_4, headers_v, ethertype,
+				 ETH_P_IP);
+		} else if (addr_type == FLOW_DISSECTOR_KEY_IPV6_ADDRS) {
+			struct flow_match_ipv6_addrs match;
+
+			flow_rule_match_enc_ipv6_addrs(rule, &match);
+			memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_c,
+					    src_ipv4_src_ipv6.ipv6_layout.ipv6),
+			       &match.mask->src, MLX5_FLD_SZ_BYTES(ipv6_layout,
+								   ipv6));
+			memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_v,
+					    src_ipv4_src_ipv6.ipv6_layout.ipv6),
+			       &match.key->src, MLX5_FLD_SZ_BYTES(ipv6_layout,
+								  ipv6));
+
+			memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_c,
+					    dst_ipv4_dst_ipv6.ipv6_layout.ipv6),
+			       &match.mask->dst, MLX5_FLD_SZ_BYTES(ipv6_layout,
+								   ipv6));
+			memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_v,
+					    dst_ipv4_dst_ipv6.ipv6_layout.ipv6),
+			       &match.key->dst, MLX5_FLD_SZ_BYTES(ipv6_layout,
+								  ipv6));
+
+			MLX5_SET_TO_ONES(fte_match_set_lyr_2_4, headers_c,
+					 ethertype);
+			MLX5_SET(fte_match_set_lyr_2_4, headers_v, ethertype,
+				 ETH_P_IPV6);
+		}
+	}
+
+	if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_ENC_IP)) {
+		struct flow_match_ip match;
+
+		flow_rule_match_enc_ip(rule, &match);
+		MLX5_SET(fte_match_set_lyr_2_4, headers_c, ip_ecn,
+			 match.mask->tos & 0x3);
+		MLX5_SET(fte_match_set_lyr_2_4, headers_v, ip_ecn,
+			 match.key->tos & 0x3);
+
+		MLX5_SET(fte_match_set_lyr_2_4, headers_c, ip_dscp,
+			 match.mask->tos >> 2);
+		MLX5_SET(fte_match_set_lyr_2_4, headers_v, ip_dscp,
+			 match.key->tos  >> 2);
+
+		MLX5_SET(fte_match_set_lyr_2_4, headers_c, ttl_hoplimit,
+			 match.mask->ttl);
+		MLX5_SET(fte_match_set_lyr_2_4, headers_v, ttl_hoplimit,
+			 match.key->ttl);
+
+		if (match.mask->ttl &&
+		    !MLX5_CAP_ESW_FLOWTABLE_FDB
+			(priv->mdev,
+			 ft_field_support.outer_ipv4_ttl)) {
+			NL_SET_ERR_MSG_MOD(extack,
+					   "Matching on TTL is not supported");
+			err = -EOPNOTSUPP;
+			goto out;
+		}
+	}
+
+	/* Enforce DMAC when offloading incoming tunneled flows.
+	 * Flow counters require a match on the DMAC.
+	 */
+	MLX5_SET_TO_ONES(fte_match_set_lyr_2_4, headers_c, dmac_47_16);
+	MLX5_SET_TO_ONES(fte_match_set_lyr_2_4, headers_c, dmac_15_0);
+	ether_addr_copy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_v,
+				     dmac_47_16), priv->netdev->dev_addr);
+
+	/* let software handle IP fragments */
+	MLX5_SET(fte_match_set_lyr_2_4, headers_c, frag, 1);
+	MLX5_SET(fte_match_set_lyr_2_4, headers_v, frag, 0);
+
+	return 0;
+
 out:
 	return err;
 }

drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun.h

@@ -76,8 +76,7 @@ int mlx5e_tc_tun_parse(struct net_device *filter_dev,
 		       struct mlx5e_priv *priv,
 		       struct mlx5_flow_spec *spec,
 		       struct flow_cls_offload *f,
-		       void *headers_c,
-		       void *headers_v, u8 *match_level);
+		       u8 *match_level);
 
 int mlx5e_tc_tun_parse_udp_ports(struct mlx5e_priv *priv,
 				 struct mlx5_flow_spec *spec,

drivers/net/ethernet/mellanox/mlx5/core/en_rep.c

@@ -1952,7 +1952,7 @@ static const struct mlx5e_profile mlx5e_rep_profile = {
 	.update_rx		= mlx5e_update_rep_rx,
 	.update_stats           = mlx5e_update_ndo_stats,
 	.rx_handlers.handle_rx_cqe       = mlx5e_handle_rx_cqe_rep,
-	.rx_handlers.handle_rx_cqe_mpwqe = mlx5e_handle_rx_cqe_mpwrq,
+	.rx_handlers.handle_rx_cqe_mpwqe = mlx5e_handle_rx_cqe_mpwrq_rep,
 	.max_tc			= 1,
 	.rq_groups		= MLX5E_NUM_RQ_GROUPS(REGULAR),
 	.stats_grps		= mlx5e_rep_stats_grps,
@@ -1972,7 +1972,7 @@ static const struct mlx5e_profile mlx5e_uplink_rep_profile = {
 	.update_stats           = mlx5e_update_ndo_stats,
 	.update_carrier	        = mlx5e_update_carrier,
 	.rx_handlers.handle_rx_cqe       = mlx5e_handle_rx_cqe_rep,
-	.rx_handlers.handle_rx_cqe_mpwqe = mlx5e_handle_rx_cqe_mpwrq,
+	.rx_handlers.handle_rx_cqe_mpwqe = mlx5e_handle_rx_cqe_mpwrq_rep,
 	.max_tc			= MLX5E_MAX_NUM_TC,
 	.rq_groups		= MLX5E_NUM_RQ_GROUPS(REGULAR),
 	.stats_grps		= mlx5e_ul_rep_stats_grps,

drivers/net/ethernet/mellanox/mlx5/core/en_rep.h

@@ -81,6 +81,11 @@ struct mlx5_rep_uplink_priv {
 	struct mutex                unready_flows_lock;
 	struct list_head            unready_flows;
 	struct work_struct          reoffload_flows_work;
+
+	/* maps tun_info to a unique id*/
+	struct mapping_ctx *tunnel_mapping;
+	/* maps tun_enc_opts to a unique id*/
+	struct mapping_ctx *tunnel_enc_opts_mapping;
 };
 
 struct mlx5e_rep_priv {
@@ -192,6 +197,8 @@ int mlx5e_add_sqs_fwd_rules(struct mlx5e_priv *priv);
 void mlx5e_remove_sqs_fwd_rules(struct mlx5e_priv *priv);
 
 void mlx5e_handle_rx_cqe_rep(struct mlx5e_rq *rq, struct mlx5_cqe64 *cqe);
+void mlx5e_handle_rx_cqe_mpwrq_rep(struct mlx5e_rq *rq,
+				   struct mlx5_cqe64 *cqe);
 
 int mlx5e_rep_encap_entry_attach(struct mlx5e_priv *priv,
 				 struct mlx5e_encap_entry *e);

drivers/net/ethernet/mellanox/mlx5/core/en_rx.c

@@ -1195,6 +1195,7 @@ void mlx5e_handle_rx_cqe_rep(struct mlx5e_rq *rq, struct mlx5_cqe64 *cqe)
 	struct mlx5e_priv *priv = netdev_priv(netdev);
 	struct mlx5e_rep_priv *rpriv  = priv->ppriv;
 	struct mlx5_eswitch_rep *rep = rpriv->rep;
+	struct mlx5e_tc_update_priv tc_priv = {};
 	struct mlx5_wq_cyc *wq = &rq->wqe.wq;
 	struct mlx5e_wqe_frag_info *wi;
 	struct sk_buff *skb;
@@ -1227,13 +1228,78 @@ void mlx5e_handle_rx_cqe_rep(struct mlx5e_rq *rq, struct mlx5_cqe64 *cqe)
 	if (rep->vlan && skb_vlan_tag_present(skb))
 		skb_vlan_pop(skb);
 
+	if (!mlx5e_tc_rep_update_skb(cqe, skb, &tc_priv))
+		goto free_wqe;
+
 	napi_gro_receive(rq->cq.napi, skb);
 
+	mlx5_tc_rep_post_napi_receive(&tc_priv);
+
 free_wqe:
 	mlx5e_free_rx_wqe(rq, wi, true);
 wq_cyc_pop:
 	mlx5_wq_cyc_pop(wq);
 }
+
+void mlx5e_handle_rx_cqe_mpwrq_rep(struct mlx5e_rq *rq,
+				   struct mlx5_cqe64 *cqe)
+{
+	u16 cstrides       = mpwrq_get_cqe_consumed_strides(cqe);
+	u16 wqe_id         = be16_to_cpu(cqe->wqe_id);
+	struct mlx5e_mpw_info *wi = &rq->mpwqe.info[wqe_id];
+	u16 stride_ix      = mpwrq_get_cqe_stride_index(cqe);
+	u32 wqe_offset     = stride_ix << rq->mpwqe.log_stride_sz;
+	u32 head_offset    = wqe_offset & (PAGE_SIZE - 1);
+	u32 page_idx       = wqe_offset >> PAGE_SHIFT;
+	struct mlx5e_tc_update_priv tc_priv = {};
+	struct mlx5e_rx_wqe_ll *wqe;
+	struct mlx5_wq_ll *wq;
+	struct sk_buff *skb;
+	u16 cqe_bcnt;
+
+	wi->consumed_strides += cstrides;
+
+	if (unlikely(MLX5E_RX_ERR_CQE(cqe))) {
+		trigger_report(rq, cqe);
+		rq->stats->wqe_err++;
+		goto mpwrq_cqe_out;
+	}
+
+	if (unlikely(mpwrq_is_filler_cqe(cqe))) {
+		struct mlx5e_rq_stats *stats = rq->stats;
+
+		stats->mpwqe_filler_cqes++;
+		stats->mpwqe_filler_strides += cstrides;
+		goto mpwrq_cqe_out;
+	}
+
+	cqe_bcnt = mpwrq_get_cqe_byte_cnt(cqe);
+
+	skb = INDIRECT_CALL_2(rq->mpwqe.skb_from_cqe_mpwrq,
+			      mlx5e_skb_from_cqe_mpwrq_linear,
+			      mlx5e_skb_from_cqe_mpwrq_nonlinear,
+			      rq, wi, cqe_bcnt, head_offset, page_idx);
+	if (!skb)
+		goto mpwrq_cqe_out;
+
+	mlx5e_complete_rx_cqe(rq, cqe, cqe_bcnt, skb);
+
+	if (!mlx5e_tc_rep_update_skb(cqe, skb, &tc_priv))
+		goto mpwrq_cqe_out;
+
+	napi_gro_receive(rq->cq.napi, skb);
+
+	mlx5_tc_rep_post_napi_receive(&tc_priv);
+
+mpwrq_cqe_out:
+	if (likely(wi->consumed_strides < rq->mpwqe.num_strides))
+		return;
+
+	wq  = &rq->mpwqe.wq;
+	wqe = mlx5_wq_ll_get_wqe(wq, wqe_id);
+	mlx5e_free_rx_mpwqe(rq, wi, true);
+	mlx5_wq_ll_pop(wq, cqe->wqe_id, &wqe->next.next_wqe_index);
+}
 #endif
 
 struct sk_buff *

drivers/net/ethernet/mellanox/mlx5/core/en_tc.c

@@ -55,10 +55,13 @@
 #include "fs_core.h"
 #include "en/port.h"
 #include "en/tc_tun.h"
+#include "en/mapping.h"
 #include "lib/devcom.h"
 #include "lib/geneve.h"
 #include "diag/en_tc_tracepoint.h"
 
+#define MLX5_MH_ACT_SZ MLX5_UN_SZ_BYTES(set_action_in_add_action_in_auto)
+
 struct mlx5_nic_flow_attr {
 	u32 action;
 	u32 flow_tag;
@@ -134,6 +137,8 @@ struct mlx5e_tc_flow {
 	refcount_t		refcnt;
 	struct rcu_head		rcu_head;
 	struct completion	init_done;
+	int tunnel_id; /* the mapped tunnel id of this flow */
+
 	union {
 		struct mlx5_esw_flow_attr esw_attr[0];
 		struct mlx5_nic_flow_attr nic_attr[0];
@@ -144,15 +149,112 @@ struct mlx5e_tc_flow_parse_attr {
 	const struct ip_tunnel_info *tun_info[MLX5_MAX_FLOW_FWD_VPORTS];
 	struct net_device *filter_dev;
 	struct mlx5_flow_spec spec;
-	int num_mod_hdr_actions;
-	int max_mod_hdr_actions;
-	void *mod_hdr_actions;
+	struct mlx5e_tc_mod_hdr_acts mod_hdr_acts;
 	int mirred_ifindex[MLX5_MAX_FLOW_FWD_VPORTS];
 };
 
 #define MLX5E_TC_TABLE_NUM_GROUPS 4
 #define MLX5E_TC_TABLE_MAX_GROUP_SIZE BIT(16)
 
+struct tunnel_match_key {
+	struct flow_dissector_key_control enc_control;
+	struct flow_dissector_key_keyid enc_key_id;
+	struct flow_dissector_key_ports enc_tp;
+	struct flow_dissector_key_ip enc_ip;
+	union {
+		struct flow_dissector_key_ipv4_addrs enc_ipv4;
+		struct flow_dissector_key_ipv6_addrs enc_ipv6;
+	};
+
+	int filter_ifindex;
+};
+
+/* Tunnel_id mapping is TUNNEL_INFO_BITS + ENC_OPTS_BITS.
+ * Upper TUNNEL_INFO_BITS for general tunnel info.
+ * Lower ENC_OPTS_BITS bits for enc_opts.
+ */
+#define TUNNEL_INFO_BITS 6
+#define TUNNEL_INFO_BITS_MASK GENMASK(TUNNEL_INFO_BITS - 1, 0)
+#define ENC_OPTS_BITS 2
+#define ENC_OPTS_BITS_MASK GENMASK(ENC_OPTS_BITS - 1, 0)
+#define TUNNEL_ID_BITS (TUNNEL_INFO_BITS + ENC_OPTS_BITS)
+#define TUNNEL_ID_MASK GENMASK(TUNNEL_ID_BITS - 1, 0)
+
+struct mlx5e_tc_attr_to_reg_mapping mlx5e_tc_attr_to_reg_mappings[] = {
+	[CHAIN_TO_REG] = {
+		.mfield = MLX5_ACTION_IN_FIELD_METADATA_REG_C_0,
+		.moffset = 0,
+		.mlen = 2,
+	},
+	[TUNNEL_TO_REG] = {
+		.mfield = MLX5_ACTION_IN_FIELD_METADATA_REG_C_1,
+		.moffset = 3,
+		.mlen = 1,
+		.soffset = MLX5_BYTE_OFF(fte_match_param,
+					 misc_parameters_2.metadata_reg_c_1),
+	},
+};
+
+static void mlx5e_put_flow_tunnel_id(struct mlx5e_tc_flow *flow);
+
+void
+mlx5e_tc_match_to_reg_match(struct mlx5_flow_spec *spec,
+			    enum mlx5e_tc_attr_to_reg type,
+			    u32 data,
+			    u32 mask)
+{
+	int soffset = mlx5e_tc_attr_to_reg_mappings[type].soffset;
+	int match_len = mlx5e_tc_attr_to_reg_mappings[type].mlen;
+	void *headers_c = spec->match_criteria;
+	void *headers_v = spec->match_value;
+	void *fmask, *fval;
+
+	fmask = headers_c + soffset;
+	fval = headers_v + soffset;
+
+	mask = cpu_to_be32(mask) >> (32 - (match_len * 8));
+	data = cpu_to_be32(data) >> (32 - (match_len * 8));
+
+	memcpy(fmask, &mask, match_len);
+	memcpy(fval, &data, match_len);
+
+	spec->match_criteria_enable |= MLX5_MATCH_MISC_PARAMETERS_2;
+}
+
+int
+mlx5e_tc_match_to_reg_set(struct mlx5_core_dev *mdev,
+			  struct mlx5e_tc_mod_hdr_acts *mod_hdr_acts,
+			  enum mlx5e_tc_attr_to_reg type,
+			  u32 data)
+{
+	int moffset = mlx5e_tc_attr_to_reg_mappings[type].moffset;
+	int mfield = mlx5e_tc_attr_to_reg_mappings[type].mfield;
+	int mlen = mlx5e_tc_attr_to_reg_mappings[type].mlen;
+	char *modact;
+	int err;
+
+	err = alloc_mod_hdr_actions(mdev, MLX5_FLOW_NAMESPACE_FDB,
+				    mod_hdr_acts);
+	if (err)
+		return err;
+
+	modact = mod_hdr_acts->actions +
+		 (mod_hdr_acts->num_actions * MLX5_MH_ACT_SZ);
+
+	/* Firmware has 5bit length field and 0 means 32bits */
+	if (mlen == 4)
+		mlen = 0;
+
+	MLX5_SET(set_action_in, modact, action_type, MLX5_ACTION_TYPE_SET);
+	MLX5_SET(set_action_in, modact, field, mfield);
+	MLX5_SET(set_action_in, modact, offset, moffset * 8);
+	MLX5_SET(set_action_in, modact, length, mlen * 8);
+	MLX5_SET(set_action_in, modact, data, data);
+	mod_hdr_acts->num_actions++;
+
+	return 0;
+}
+
 struct mlx5e_hairpin {
 	struct mlx5_hairpin *pair;
 
@@ -210,8 +312,6 @@ struct mlx5e_mod_hdr_entry {
 	int compl_result;
 };
 
-#define MLX5_MH_ACT_SZ MLX5_UN_SZ_BYTES(set_action_in_add_action_in_auto)
-
 static void mlx5e_tc_del_flow(struct mlx5e_priv *priv,
 			      struct mlx5e_tc_flow *flow);
 
@@ -361,10 +461,10 @@ static int mlx5e_attach_mod_hdr(struct mlx5e_priv *priv,
 	struct mod_hdr_key key;
 	u32 hash_key;
 
-	num_actions  = parse_attr->num_mod_hdr_actions;
+	num_actions  = parse_attr->mod_hdr_acts.num_actions;
 	actions_size = MLX5_MH_ACT_SZ * num_actions;
 
-	key.actions = parse_attr->mod_hdr_actions;
+	key.actions = parse_attr->mod_hdr_acts.actions;
 	key.num_actions = num_actions;
 
 	hash_key = hash_mod_hdr_info(&key);
@@ -954,7 +1054,7 @@ mlx5e_tc_add_nic_flow(struct mlx5e_priv *priv,
 	if (attr->action & MLX5_FLOW_CONTEXT_ACTION_MOD_HDR) {
 		err = mlx5e_attach_mod_hdr(priv, flow, parse_attr);
 		flow_act.modify_hdr = attr->modify_hdr;
-		kfree(parse_attr->mod_hdr_actions);
+		dealloc_mod_hdr_actions(&parse_attr->mod_hdr_acts);
 		if (err)
 			return err;
 	}
@@ -1224,7 +1324,7 @@ mlx5e_tc_add_fdb_flow(struct mlx5e_priv *priv,
 
 	if (attr->action & MLX5_FLOW_CONTEXT_ACTION_MOD_HDR) {
 		err = mlx5e_attach_mod_hdr(priv, flow, parse_attr);
-		kfree(parse_attr->mod_hdr_actions);
+		dealloc_mod_hdr_actions(&parse_attr->mod_hdr_acts);
 		if (err)
 			return err;
 	}
@@ -1274,6 +1374,8 @@ static void mlx5e_tc_del_fdb_flow(struct mlx5e_priv *priv,
 	struct mlx5_esw_flow_attr *attr = flow->esw_attr;
 	int out_index;
 
+	mlx5e_put_flow_tunnel_id(flow);
+
 	if (flow_flag_test(flow, NOT_READY)) {
 		remove_unready_flow(flow);
 		kvfree(attr->parse_attr);
@@ -1662,139 +1764,250 @@ static void mlx5e_tc_del_flow(struct mlx5e_priv *priv,
 	}
 }
 
+static int flow_has_tc_fwd_action(struct flow_cls_offload *f)
+{
+	struct flow_rule *rule = flow_cls_offload_flow_rule(f);
+	struct flow_action *flow_action = &rule->action;
+	const struct flow_action_entry *act;
+	int i;
 
-static int parse_tunnel_attr(struct mlx5e_priv *priv,
-			     struct mlx5_flow_spec *spec,
+	flow_action_for_each(i, act, flow_action) {
+		switch (act->id) {
+		case FLOW_ACTION_GOTO:
+			return true;
+		default:
+			continue;
+		}
+	}
+
+	return false;
+}
+
+static int
+enc_opts_is_dont_care_or_full_match(struct mlx5e_priv *priv,
+				    struct flow_dissector_key_enc_opts *opts,
+				    struct netlink_ext_ack *extack,
+				    bool *dont_care)
+{
+	struct geneve_opt *opt;
+	int off = 0;
+
+	*dont_care = true;
+
+	while (opts->len > off) {
+		opt = (struct geneve_opt *)&opts->data[off];
+
+		if (!(*dont_care) || opt->opt_class || opt->type ||
+		    memchr_inv(opt->opt_data, 0, opt->length * 4)) {
+			*dont_care = false;
+
+			if (opt->opt_class != U16_MAX ||
+			    opt->type != U8_MAX ||
+			    memchr_inv(opt->opt_data, 0xFF,
+				       opt->length * 4)) {
+				NL_SET_ERR_MSG(extack,
+					       "Partial match of tunnel options in chain > 0 isn't supported");
+				netdev_warn(priv->netdev,
+					    "Partial match of tunnel options in chain > 0 isn't supported");
+				return -EOPNOTSUPP;
+			}
+		}
+
+		off += sizeof(struct geneve_opt) + opt->length * 4;
+	}
+
+	return 0;
+}
+
+#define COPY_DISSECTOR(rule, diss_key, dst)\
+({ \
+	struct flow_rule *__rule = (rule);\
+	typeof(dst) __dst = dst;\
+\
+	memcpy(__dst,\
+	       skb_flow_dissector_target(__rule->match.dissector,\
+					 diss_key,\
+					 __rule->match.key),\
+	       sizeof(*__dst));\
+})
+
+static int mlx5e_get_flow_tunnel_id(struct mlx5e_priv *priv,
+				    struct mlx5e_tc_flow *flow,
 				    struct flow_cls_offload *f,
-			     struct net_device *filter_dev, u8 *match_level)
+				    struct net_device *filter_dev)
 {
-	struct netlink_ext_ack *extack = f->common.extack;
-	void *headers_c = MLX5_ADDR_OF(fte_match_param, spec->match_criteria,
-				       outer_headers);
-	void *headers_v = MLX5_ADDR_OF(fte_match_param, spec->match_value,
-				       outer_headers);
 	struct flow_rule *rule = flow_cls_offload_flow_rule(f);
+	struct netlink_ext_ack *extack = f->common.extack;
+	struct mlx5_esw_flow_attr *attr = flow->esw_attr;
+	struct mlx5e_tc_mod_hdr_acts *mod_hdr_acts;
+	struct flow_match_enc_opts enc_opts_match;
+	struct mlx5_rep_uplink_priv *uplink_priv;
+	struct mlx5e_rep_priv *uplink_rpriv;
+	struct tunnel_match_key tunnel_key;
+	bool enc_opts_is_dont_care = true;
+	u32 tun_id, enc_opts_id = 0;
+	struct mlx5_eswitch *esw;
+	u32 value, mask;
 	int err;
 
-	err = mlx5e_tc_tun_parse(filter_dev, priv, spec, f,
-				 headers_c, headers_v, match_level);
-	if (err) {
-		NL_SET_ERR_MSG_MOD(extack,
-				   "failed to parse tunnel attributes");
+	esw = priv->mdev->priv.eswitch;
+	uplink_rpriv = mlx5_eswitch_get_uplink_priv(esw, REP_ETH);
+	uplink_priv = &uplink_rpriv->uplink_priv;
+
+	memset(&tunnel_key, 0, sizeof(tunnel_key));
+	COPY_DISSECTOR(rule, FLOW_DISSECTOR_KEY_ENC_CONTROL,
+		       &tunnel_key.enc_control);
+	if (tunnel_key.enc_control.addr_type == FLOW_DISSECTOR_KEY_IPV4_ADDRS)
+		COPY_DISSECTOR(rule, FLOW_DISSECTOR_KEY_ENC_IPV4_ADDRS,
+			       &tunnel_key.enc_ipv4);
+	else
+		COPY_DISSECTOR(rule, FLOW_DISSECTOR_KEY_ENC_IPV6_ADDRS,
+			       &tunnel_key.enc_ipv6);
+	COPY_DISSECTOR(rule, FLOW_DISSECTOR_KEY_ENC_IP, &tunnel_key.enc_ip);
+	COPY_DISSECTOR(rule, FLOW_DISSECTOR_KEY_ENC_PORTS,
+		       &tunnel_key.enc_tp);
+	COPY_DISSECTOR(rule, FLOW_DISSECTOR_KEY_ENC_KEYID,
+		       &tunnel_key.enc_key_id);
+	tunnel_key.filter_ifindex = filter_dev->ifindex;
+
+	err = mapping_add(uplink_priv->tunnel_mapping, &tunnel_key, &tun_id);
+	if (err)
 		return err;
-	}
 
-	if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_ENC_CONTROL)) {
-		struct flow_match_control match;
-		u16 addr_type;
+	flow_rule_match_enc_opts(rule, &enc_opts_match);
+	err = enc_opts_is_dont_care_or_full_match(priv,
+						  enc_opts_match.mask,
+						  extack,
+						  &enc_opts_is_dont_care);
+	if (err)
+		goto err_enc_opts;
 
-		flow_rule_match_enc_control(rule, &match);
-		addr_type = match.key->addr_type;
+	if (!enc_opts_is_dont_care) {
+		err = mapping_add(uplink_priv->tunnel_enc_opts_mapping,
+				  enc_opts_match.key, &enc_opts_id);
+		if (err)
+			goto err_enc_opts;
+	}
 
-		/* For tunnel addr_type used same key id`s as for non-tunnel */
-		if (addr_type == FLOW_DISSECTOR_KEY_IPV4_ADDRS) {
-			struct flow_match_ipv4_addrs match;
+	value = tun_id << ENC_OPTS_BITS | enc_opts_id;
+	mask = enc_opts_id ? TUNNEL_ID_MASK :
+			     (TUNNEL_ID_MASK & ~ENC_OPTS_BITS_MASK);
 
-			flow_rule_match_enc_ipv4_addrs(rule, &match);
-			MLX5_SET(fte_match_set_lyr_2_4, headers_c,
-				 src_ipv4_src_ipv6.ipv4_layout.ipv4,
-				 ntohl(match.mask->src));
-			MLX5_SET(fte_match_set_lyr_2_4, headers_v,
-				 src_ipv4_src_ipv6.ipv4_layout.ipv4,
-				 ntohl(match.key->src));
+	if (attr->chain) {
+		mlx5e_tc_match_to_reg_match(&attr->parse_attr->spec,
+					    TUNNEL_TO_REG, value, mask);
+	} else {
+		mod_hdr_acts = &attr->parse_attr->mod_hdr_acts;
+		err = mlx5e_tc_match_to_reg_set(priv->mdev,
+						mod_hdr_acts,
+						TUNNEL_TO_REG, value);
+		if (err)
+			goto err_set;
 
-			MLX5_SET(fte_match_set_lyr_2_4, headers_c,
-				 dst_ipv4_dst_ipv6.ipv4_layout.ipv4,
-				 ntohl(match.mask->dst));
-			MLX5_SET(fte_match_set_lyr_2_4, headers_v,
-				 dst_ipv4_dst_ipv6.ipv4_layout.ipv4,
-				 ntohl(match.key->dst));
+		attr->action |= MLX5_FLOW_CONTEXT_ACTION_MOD_HDR;
+	}
 
-			MLX5_SET_TO_ONES(fte_match_set_lyr_2_4, headers_c,
-					 ethertype);
-			MLX5_SET(fte_match_set_lyr_2_4, headers_v, ethertype,
-				 ETH_P_IP);
-		} else if (addr_type == FLOW_DISSECTOR_KEY_IPV6_ADDRS) {
-			struct flow_match_ipv6_addrs match;
+	flow->tunnel_id = value;
+	return 0;
 
-			flow_rule_match_enc_ipv6_addrs(rule, &match);
-			memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_c,
-					    src_ipv4_src_ipv6.ipv6_layout.ipv6),
-			       &match.mask->src, MLX5_FLD_SZ_BYTES(ipv6_layout,
-								   ipv6));
-			memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_v,
-					    src_ipv4_src_ipv6.ipv6_layout.ipv6),
-			       &match.key->src, MLX5_FLD_SZ_BYTES(ipv6_layout,
-								  ipv6));
+err_set:
+	if (enc_opts_id)
+		mapping_remove(uplink_priv->tunnel_enc_opts_mapping,
+			       enc_opts_id);
+err_enc_opts:
+	mapping_remove(uplink_priv->tunnel_mapping, tun_id);
+	return err;
+}
 
-			memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_c,
-					    dst_ipv4_dst_ipv6.ipv6_layout.ipv6),
-			       &match.mask->dst, MLX5_FLD_SZ_BYTES(ipv6_layout,
-								   ipv6));
-			memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_v,
-					    dst_ipv4_dst_ipv6.ipv6_layout.ipv6),
-			       &match.key->dst, MLX5_FLD_SZ_BYTES(ipv6_layout,
-								  ipv6));
+static void mlx5e_put_flow_tunnel_id(struct mlx5e_tc_flow *flow)
+{
+	u32 enc_opts_id = flow->tunnel_id & ENC_OPTS_BITS_MASK;
+	u32 tun_id = flow->tunnel_id >> ENC_OPTS_BITS;
+	struct mlx5_rep_uplink_priv *uplink_priv;
+	struct mlx5e_rep_priv *uplink_rpriv;
+	struct mlx5_eswitch *esw;
 
-			MLX5_SET_TO_ONES(fte_match_set_lyr_2_4, headers_c,
-					 ethertype);
-			MLX5_SET(fte_match_set_lyr_2_4, headers_v, ethertype,
-				 ETH_P_IPV6);
-		}
-	}
+	esw = flow->priv->mdev->priv.eswitch;
+	uplink_rpriv = mlx5_eswitch_get_uplink_priv(esw, REP_ETH);
+	uplink_priv = &uplink_rpriv->uplink_priv;
 
-	if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_ENC_IP)) {
-		struct flow_match_ip match;
+	if (tun_id)
+		mapping_remove(uplink_priv->tunnel_mapping, tun_id);
+	if (enc_opts_id)
+		mapping_remove(uplink_priv->tunnel_enc_opts_mapping,
+			       enc_opts_id);
+}
 
-		flow_rule_match_enc_ip(rule, &match);
-		MLX5_SET(fte_match_set_lyr_2_4, headers_c, ip_ecn,
-			 match.mask->tos & 0x3);
-		MLX5_SET(fte_match_set_lyr_2_4, headers_v, ip_ecn,
-			 match.key->tos & 0x3);
+static int parse_tunnel_attr(struct mlx5e_priv *priv,
+			     struct mlx5e_tc_flow *flow,
+			     struct mlx5_flow_spec *spec,
+			     struct flow_cls_offload *f,
+			     struct net_device *filter_dev,
+			     u8 *match_level,
+			     bool *match_inner)
+{
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	struct netlink_ext_ack *extack = f->common.extack;
+	bool needs_mapping, sets_mapping;
+	int err;
 
-		MLX5_SET(fte_match_set_lyr_2_4, headers_c, ip_dscp,
-			 match.mask->tos >> 2);
-		MLX5_SET(fte_match_set_lyr_2_4, headers_v, ip_dscp,
-			 match.key->tos  >> 2);
+	if (!mlx5e_is_eswitch_flow(flow))
+		return -EOPNOTSUPP;
 
-		MLX5_SET(fte_match_set_lyr_2_4, headers_c, ttl_hoplimit,
-			 match.mask->ttl);
-		MLX5_SET(fte_match_set_lyr_2_4, headers_v, ttl_hoplimit,
-			 match.key->ttl);
+	needs_mapping = !!flow->esw_attr->chain;
+	sets_mapping = !flow->esw_attr->chain && flow_has_tc_fwd_action(f);
+	*match_inner = !needs_mapping;
 
-		if (match.mask->ttl &&
-		    !MLX5_CAP_ESW_FLOWTABLE_FDB
-			(priv->mdev,
-			 ft_field_support.outer_ipv4_ttl)) {
-			NL_SET_ERR_MSG_MOD(extack,
-					   "Matching on TTL is not supported");
+	if ((needs_mapping || sets_mapping) &&
+	    !mlx5_eswitch_vport_match_metadata_enabled(esw)) {
+		NL_SET_ERR_MSG(extack,
+			       "Chains on tunnel devices isn't supported without register metadata support");
+		netdev_warn(priv->netdev,
+			    "Chains on tunnel devices isn't supported without register metadata support");
 		return -EOPNOTSUPP;
 	}
 
+	if (!flow->esw_attr->chain) {
+		err = mlx5e_tc_tun_parse(filter_dev, priv, spec, f,
+					 match_level);
+		if (err) {
+			NL_SET_ERR_MSG_MOD(extack,
+					   "Failed to parse tunnel attributes");
+			netdev_warn(priv->netdev,
+				    "Failed to parse tunnel attributes");
+			return err;
 		}
 
-	/* Enforce DMAC when offloading incoming tunneled flows.
-	 * Flow counters require a match on the DMAC.
-	 */
-	MLX5_SET_TO_ONES(fte_match_set_lyr_2_4, headers_c, dmac_47_16);
-	MLX5_SET_TO_ONES(fte_match_set_lyr_2_4, headers_c, dmac_15_0);
-	ether_addr_copy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_v,
-				     dmac_47_16), priv->netdev->dev_addr);
-
-	/* let software handle IP fragments */
-	MLX5_SET(fte_match_set_lyr_2_4, headers_c, frag, 1);
-	MLX5_SET(fte_match_set_lyr_2_4, headers_v, frag, 0);
+		flow->esw_attr->action |= MLX5_FLOW_CONTEXT_ACTION_DECAP;
+	}
 
+	if (!needs_mapping && !sets_mapping)
 		return 0;
+
+	return mlx5e_get_flow_tunnel_id(priv, flow, f, filter_dev);
 }
 
-static void *get_match_headers_criteria(u32 flags,
-					struct mlx5_flow_spec *spec)
+static void *get_match_inner_headers_criteria(struct mlx5_flow_spec *spec)
 {
-	return (flags & MLX5_FLOW_CONTEXT_ACTION_DECAP) ?
-		MLX5_ADDR_OF(fte_match_param, spec->match_criteria,
-			     inner_headers) :
-		MLX5_ADDR_OF(fte_match_param, spec->match_criteria,
+	return MLX5_ADDR_OF(fte_match_param, spec->match_criteria,
+			    inner_headers);
+}
+
+static void *get_match_inner_headers_value(struct mlx5_flow_spec *spec)
+{
+	return MLX5_ADDR_OF(fte_match_param, spec->match_value,
+			    inner_headers);
+}
+
+static void *get_match_outer_headers_criteria(struct mlx5_flow_spec *spec)
+{
+	return MLX5_ADDR_OF(fte_match_param, spec->match_criteria,
+			    outer_headers);
+}
+
+static void *get_match_outer_headers_value(struct mlx5_flow_spec *spec)
+{
+	return MLX5_ADDR_OF(fte_match_param, spec->match_value,
 			    outer_headers);
 }
 
@@ -1802,10 +2015,16 @@ static void *get_match_headers_value(u32 flags,
 				     struct mlx5_flow_spec *spec)
 {
 	return (flags & MLX5_FLOW_CONTEXT_ACTION_DECAP) ?
-		MLX5_ADDR_OF(fte_match_param, spec->match_value,
-			     inner_headers) :
-		MLX5_ADDR_OF(fte_match_param, spec->match_value,
-			     outer_headers);
+		get_match_inner_headers_value(spec) :
+		get_match_outer_headers_value(spec);
+}
+
+static void *get_match_headers_criteria(u32 flags,
+					struct mlx5_flow_spec *spec)
+{
+	return (flags & MLX5_FLOW_CONTEXT_ACTION_DECAP) ?
+		get_match_inner_headers_criteria(spec) :
+		get_match_outer_headers_criteria(spec);
 }
 
 static int mlx5e_flower_parse_meta(struct net_device *filter_dev,
@@ -1843,6 +2062,7 @@ static int mlx5e_flower_parse_meta(struct net_device *filter_dev,
 }
 
 static int __parse_cls_flower(struct mlx5e_priv *priv,
+			      struct mlx5e_tc_flow *flow,
 			      struct mlx5_flow_spec *spec,
 			      struct flow_cls_offload *f,
 			      struct net_device *filter_dev,
@@ -1892,18 +2112,22 @@ static int __parse_cls_flower(struct mlx5e_priv *priv,
 	}
 
 	if (mlx5e_get_tc_tun(filter_dev)) {
-		if (parse_tunnel_attr(priv, spec, f, filter_dev,
-				      outer_match_level))
-			return -EOPNOTSUPP;
+		bool match_inner = false;
 
-		/* At this point, header pointers should point to the inner
-		 * headers, outer header were already set by parse_tunnel_attr
+		err = parse_tunnel_attr(priv, flow, spec, f, filter_dev,
+					outer_match_level, &match_inner);
+		if (err)
+			return err;
+
+		if (match_inner) {
+			/* header pointers should point to the inner headers
+			 * if the packet was decapsulated already.
+			 * outer headers are set by parse_tunnel_attr.
 			 */
 			match_level = inner_match_level;
-		headers_c = get_match_headers_criteria(MLX5_FLOW_CONTEXT_ACTION_DECAP,
-						       spec);
-		headers_v = get_match_headers_value(MLX5_FLOW_CONTEXT_ACTION_DECAP,
-						    spec);
+			headers_c = get_match_inner_headers_criteria(spec);
+			headers_v = get_match_inner_headers_value(spec);
+		}
 	}
 
 	err = mlx5e_flower_parse_meta(filter_dev, f);
@@ -2220,8 +2444,8 @@ static int parse_cls_flower(struct mlx5e_priv *priv,
 	inner_match_level = MLX5_MATCH_NONE;
 	outer_match_level = MLX5_MATCH_NONE;
 
-	err = __parse_cls_flower(priv, spec, f, filter_dev, &inner_match_level,
-				 &outer_match_level);
+	err = __parse_cls_flower(priv, flow, spec, f, filter_dev,
+				 &inner_match_level, &outer_match_level);
 	non_tunnel_match_level = (inner_match_level == MLX5_MATCH_NONE) ?
 				 outer_match_level : inner_match_level;
 
@@ -2381,25 +2605,26 @@ static struct mlx5_fields fields[] = {
 	OFFLOAD(UDP_DPORT, 16, U16_MAX, udp.dest,   0, udp_dport),
 };
 
-/* On input attr->max_mod_hdr_actions tells how many HW actions can be parsed at
- * max from the SW pedit action. On success, attr->num_mod_hdr_actions
- * says how many HW actions were actually parsed.
- */
-static int offload_pedit_fields(struct pedit_headers_action *hdrs,
+static int offload_pedit_fields(struct mlx5e_priv *priv,
+				int namespace,
+				struct pedit_headers_action *hdrs,
 				struct mlx5e_tc_flow_parse_attr *parse_attr,
 				u32 *action_flags,
 				struct netlink_ext_ack *extack)
 {
 	struct pedit_headers *set_masks, *add_masks, *set_vals, *add_vals;
-	int i, action_size, nactions, max_actions, first, last, next_z;
+	int i, action_size, first, last, next_z;
 	void *headers_c, *headers_v, *action, *vals_p;
 	u32 *s_masks_p, *a_masks_p, s_mask, a_mask;
+	struct mlx5e_tc_mod_hdr_acts *mod_acts;
 	struct mlx5_fields *f;
 	unsigned long mask;
 	__be32 mask_be32;
 	__be16 mask_be16;
+	int err;
 	u8 cmd;
 
+	mod_acts = &parse_attr->mod_hdr_acts;
 	headers_c = get_match_headers_criteria(*action_flags, &parse_attr->spec);
 	headers_v = get_match_headers_value(*action_flags, &parse_attr->spec);
 
@@ -2409,11 +2634,6 @@ static int offload_pedit_fields(struct pedit_headers_action *hdrs,
 	add_vals = &hdrs[1].vals;
 
 	action_size = MLX5_UN_SZ_BYTES(set_action_in_add_action_in_auto);
-	action = parse_attr->mod_hdr_actions +
-		 parse_attr->num_mod_hdr_actions * action_size;
-
-	max_actions = parse_attr->max_mod_hdr_actions;
-	nactions = parse_attr->num_mod_hdr_actions;
 
 	for (i = 0; i < ARRAY_SIZE(fields); i++) {
 		bool skip;
@@ -2439,13 +2659,6 @@ static int offload_pedit_fields(struct pedit_headers_action *hdrs,
 			return -EOPNOTSUPP;
 		}
 
-		if (nactions == max_actions) {
-			NL_SET_ERR_MSG_MOD(extack,
-					   "too many pedit actions, can't offload");
-			printk(KERN_WARNING "mlx5: parsed %d pedit actions, can't do more\n", nactions);
-			return -EOPNOTSUPP;
-		}
-
 		skip = false;
 		if (s_mask) {
 			void *match_mask = headers_c + f->match_offset;
@@ -2492,6 +2705,18 @@ static int offload_pedit_fields(struct pedit_headers_action *hdrs,
 			return -EOPNOTSUPP;
 		}
 
+		err = alloc_mod_hdr_actions(priv->mdev, namespace, mod_acts);
+		if (err) {
+			NL_SET_ERR_MSG_MOD(extack,
+					   "too many pedit actions, can't offload");
+			mlx5_core_warn(priv->mdev,
+				       "mlx5: parsed %d pedit actions, can't do more\n",
+				       mod_acts->num_actions);
+			return err;
+		}
+
+		action = mod_acts->actions +
+			 (mod_acts->num_actions * action_size);
 		MLX5_SET(set_action_in, action, action_type, cmd);
 		MLX5_SET(set_action_in, action, field, f->field);
 
@@ -2514,11 +2739,9 @@ static int offload_pedit_fields(struct pedit_headers_action *hdrs,
 		else if (f->field_bsize == 8)
 			MLX5_SET(set_action_in, action, data, *(u8 *)vals_p >> first);
 
-		action += action_size;
-		nactions++;
+		++mod_acts->num_actions;
 	}
 
-	parse_attr->num_mod_hdr_actions = nactions;
 	return 0;
 }
 
@@ -2531,29 +2754,48 @@ static int mlx5e_flow_namespace_max_modify_action(struct mlx5_core_dev *mdev,
 		return MLX5_CAP_FLOWTABLE_NIC_RX(mdev, max_modify_header_actions);
 }
 
-static int alloc_mod_hdr_actions(struct mlx5e_priv *priv,
-				 struct pedit_headers_action *hdrs,
+int alloc_mod_hdr_actions(struct mlx5_core_dev *mdev,
 			  int namespace,
-				 struct mlx5e_tc_flow_parse_attr *parse_attr)
+			  struct mlx5e_tc_mod_hdr_acts *mod_hdr_acts)
 {
-	int nkeys, action_size, max_actions;
+	int action_size, new_num_actions, max_hw_actions;
+	size_t new_sz, old_sz;
+	void *ret;
 
-	nkeys = hdrs[TCA_PEDIT_KEY_EX_CMD_SET].pedits +
-		hdrs[TCA_PEDIT_KEY_EX_CMD_ADD].pedits;
-	action_size = MLX5_UN_SZ_BYTES(set_action_in_add_action_in_auto);
+	if (mod_hdr_acts->num_actions < mod_hdr_acts->max_actions)
+		return 0;
 
-	max_actions = mlx5e_flow_namespace_max_modify_action(priv->mdev, namespace);
-	/* can get up to crazingly 16 HW actions in 32 bits pedit SW key */
-	max_actions = min(max_actions, nkeys * 16);
+	action_size = MLX5_UN_SZ_BYTES(set_action_in_add_action_in_auto);
 
-	parse_attr->mod_hdr_actions = kcalloc(max_actions, action_size, GFP_KERNEL);
-	if (!parse_attr->mod_hdr_actions)
+	max_hw_actions = mlx5e_flow_namespace_max_modify_action(mdev,
+								namespace);
+	new_num_actions = min(max_hw_actions,
+			      mod_hdr_acts->actions ?
+			      mod_hdr_acts->max_actions * 2 : 1);
+	if (mod_hdr_acts->max_actions == new_num_actions)
+		return -ENOSPC;
+
+	new_sz = action_size * new_num_actions;
+	old_sz = mod_hdr_acts->max_actions * action_size;
+	ret = krealloc(mod_hdr_acts->actions, new_sz, GFP_KERNEL);
+	if (!ret)
 		return -ENOMEM;
 
-	parse_attr->max_mod_hdr_actions = max_actions;
+	memset(ret + old_sz, 0, new_sz - old_sz);
+	mod_hdr_acts->actions = ret;
+	mod_hdr_acts->max_actions = new_num_actions;
+
 	return 0;
 }
 
+void dealloc_mod_hdr_actions(struct mlx5e_tc_mod_hdr_acts *mod_hdr_acts)
+{
+	kfree(mod_hdr_acts->actions);
+	mod_hdr_acts->actions = NULL;
+	mod_hdr_acts->num_actions = 0;
+	mod_hdr_acts->max_actions = 0;
+}
+
 static const struct pedit_headers zero_masks = {};
 
 static int parse_tc_pedit_action(struct mlx5e_priv *priv,
@@ -2605,13 +2847,8 @@ static int alloc_tc_pedit_action(struct mlx5e_priv *priv, int namespace,
 	int err;
 	u8 cmd;
 
-	if (!parse_attr->mod_hdr_actions) {
-		err = alloc_mod_hdr_actions(priv, hdrs, namespace, parse_attr);
-		if (err)
-			goto out_err;
-	}
-
-	err = offload_pedit_fields(hdrs, parse_attr, action_flags, extack);
+	err = offload_pedit_fields(priv, namespace, hdrs, parse_attr,
+				   action_flags, extack);
 	if (err < 0)
 		goto out_dealloc_parsed_actions;
 
@@ -2631,8 +2868,7 @@ static int alloc_tc_pedit_action(struct mlx5e_priv *priv, int namespace,
 	return 0;
 
 out_dealloc_parsed_actions:
-	kfree(parse_attr->mod_hdr_actions);
-out_err:
+	dealloc_mod_hdr_actions(&parse_attr->mod_hdr_acts);
 	return err;
 }
 
@@ -2761,6 +2997,8 @@ static bool actions_match_supported(struct mlx5e_priv *priv,
 				    struct mlx5e_tc_flow *flow,
 				    struct netlink_ext_ack *extack)
 {
+	struct net_device *filter_dev = parse_attr->filter_dev;
+	bool drop_action, pop_action;
 	u32 actions;
 
 	if (mlx5e_is_eswitch_flow(flow))
@@ -2768,11 +3006,17 @@ static bool actions_match_supported(struct mlx5e_priv *priv,
 	else
 		actions = flow->nic_attr->action;
 
-	if (flow_flag_test(flow, EGRESS) &&
-	    !((actions & MLX5_FLOW_CONTEXT_ACTION_DECAP) ||
-	      (actions & MLX5_FLOW_CONTEXT_ACTION_VLAN_POP) ||
-	      (actions & MLX5_FLOW_CONTEXT_ACTION_DROP)))
+	drop_action = actions & MLX5_FLOW_CONTEXT_ACTION_DROP;
+	pop_action = actions & MLX5_FLOW_CONTEXT_ACTION_VLAN_POP;
+
+	if (flow_flag_test(flow, EGRESS) && !drop_action) {
+		/* We only support filters on tunnel device, or on vlan
+		 * devices if they have pop/drop action
+		 */
+		if (!mlx5e_get_tc_tun(filter_dev) ||
+		    (is_vlan_dev(filter_dev) && !pop_action))
 			return false;
+	}
 
 	if (actions & MLX5_FLOW_CONTEXT_ACTION_MOD_HDR)
 		return modify_header_match_supported(&parse_attr->spec,
@@ -2968,9 +3212,9 @@ static int parse_tc_nic_actions(struct mlx5e_priv *priv,
 		/* in case all pedit actions are skipped, remove the MOD_HDR
 		 * flag.
 		 */
-		if (parse_attr->num_mod_hdr_actions == 0) {
+		if (parse_attr->mod_hdr_acts.num_actions == 0) {
 			action &= ~MLX5_FLOW_CONTEXT_ACTION_MOD_HDR;
-			kfree(parse_attr->mod_hdr_actions);
+			dealloc_mod_hdr_actions(&parse_attr->mod_hdr_acts);
 		}
 	}
 
@@ -3366,9 +3610,9 @@ static int parse_tc_fdb_actions(struct mlx5e_priv *priv,
 	int ifindexes[MLX5_MAX_FLOW_FWD_VPORTS];
 	bool ft_flow = mlx5e_is_ft_flow(flow);
 	const struct flow_action_entry *act;
+	bool encap = false, decap = false;
+	u32 action = attr->action;
 	int err, i, if_count = 0;
-	bool encap = false;
-	u32 action = 0;
 
 	if (!flow_action_has_entries(flow_action))
 		return -EINVAL;
@@ -3571,7 +3815,7 @@ static int parse_tc_fdb_actions(struct mlx5e_priv *priv,
 			attr->split_count = attr->out_count;
 			break;
 		case FLOW_ACTION_TUNNEL_DECAP:
-			action |= MLX5_FLOW_CONTEXT_ACTION_DECAP;
+			decap = true;
 			break;
 		case FLOW_ACTION_GOTO:
 			err = mlx5_validate_goto_chain(esw, flow, act, action,
@@ -3610,9 +3854,9 @@ static int parse_tc_fdb_actions(struct mlx5e_priv *priv,
 		 * flag. we might have set split_count either by pedit or
 		 * pop/push. if there is no pop/push either, reset it too.
 		 */
-		if (parse_attr->num_mod_hdr_actions == 0) {
+		if (parse_attr->mod_hdr_acts.num_actions == 0) {
 			action &= ~MLX5_FLOW_CONTEXT_ACTION_MOD_HDR;
-			kfree(parse_attr->mod_hdr_actions);
+			dealloc_mod_hdr_actions(&parse_attr->mod_hdr_acts);
 			if (!((action & MLX5_FLOW_CONTEXT_ACTION_VLAN_POP) ||
 			      (action & MLX5_FLOW_CONTEXT_ACTION_VLAN_PUSH)))
 				attr->split_count = 0;
@@ -3624,6 +3868,22 @@ static int parse_tc_fdb_actions(struct mlx5e_priv *priv,
 		return -EOPNOTSUPP;
 
 	if (attr->dest_chain) {
+		if (decap) {
+			/* It can be supported if we'll create a mapping for
+			 * the tunnel device only (without tunnel), and set
+			 * this tunnel id with this decap flow.
+			 *
+			 * On restore (miss), we'll just set this saved tunnel
+			 * device.
+			 */
+
+			NL_SET_ERR_MSG(extack,
+				       "Decap with goto isn't supported");
+			netdev_warn(priv->netdev,
+				    "Decap with goto isn't supported");
+			return -EOPNOTSUPP;
+		}
+
 		if (attr->action & MLX5_FLOW_CONTEXT_ACTION_FWD_DEST) {
 			NL_SET_ERR_MSG_MOD(extack,
 					   "Mirroring goto chain rules isn't supported");
@@ -4353,12 +4613,55 @@ void mlx5e_tc_nic_cleanup(struct mlx5e_priv *priv)
 
 int mlx5e_tc_esw_init(struct rhashtable *tc_ht)
 {
-	return rhashtable_init(tc_ht, &tc_ht_params);
+	const size_t sz_enc_opts = sizeof(struct flow_dissector_key_enc_opts);
+	struct mlx5_rep_uplink_priv *uplink_priv;
+	struct mlx5e_rep_priv *priv;
+	struct mapping_ctx *mapping;
+	int err;
+
+	uplink_priv = container_of(tc_ht, struct mlx5_rep_uplink_priv, tc_ht);
+	priv = container_of(uplink_priv, struct mlx5e_rep_priv, uplink_priv);
+
+	mapping = mapping_create(sizeof(struct tunnel_match_key),
+				 TUNNEL_INFO_BITS_MASK, true);
+	if (IS_ERR(mapping)) {
+		err = PTR_ERR(mapping);
+		goto err_tun_mapping;
+	}
+	uplink_priv->tunnel_mapping = mapping;
+
+	mapping = mapping_create(sz_enc_opts, ENC_OPTS_BITS_MASK, true);
+	if (IS_ERR(mapping)) {
+		err = PTR_ERR(mapping);
+		goto err_enc_opts_mapping;
+	}
+	uplink_priv->tunnel_enc_opts_mapping = mapping;
+
+	err = rhashtable_init(tc_ht, &tc_ht_params);
+	if (err)
+		goto err_ht_init;
+
+	return err;
+
+err_ht_init:
+	mapping_destroy(uplink_priv->tunnel_enc_opts_mapping);
+err_enc_opts_mapping:
+	mapping_destroy(uplink_priv->tunnel_mapping);
+err_tun_mapping:
+	netdev_warn(priv->netdev,
+		    "Failed to initialize tc (eswitch), err: %d", err);
+	return err;
 }
 
 void mlx5e_tc_esw_cleanup(struct rhashtable *tc_ht)
 {
+	struct mlx5_rep_uplink_priv *uplink_priv;
+
 	rhashtable_free_and_destroy(tc_ht, _mlx5e_tc_del_flow, NULL);
+
+	uplink_priv = container_of(tc_ht, struct mlx5_rep_uplink_priv, tc_ht);
+	mapping_destroy(uplink_priv->tunnel_enc_opts_mapping);
+	mapping_destroy(uplink_priv->tunnel_mapping);
 }
 
 int mlx5e_tc_num_filters(struct mlx5e_priv *priv, unsigned long flags)
@@ -4390,3 +4693,138 @@ void mlx5e_tc_reoffload_flows_work(struct work_struct *work)
 	}
 	mutex_unlock(&rpriv->unready_flows_lock);
 }
+
+#if IS_ENABLED(CONFIG_NET_TC_SKB_EXT)
+static bool mlx5e_restore_tunnel(struct mlx5e_priv *priv, struct sk_buff *skb,
+				 struct mlx5e_tc_update_priv *tc_priv,
+				 u32 tunnel_id)
+{
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	struct flow_dissector_key_enc_opts enc_opts = {};
+	struct mlx5_rep_uplink_priv *uplink_priv;
+	struct mlx5e_rep_priv *uplink_rpriv;
+	struct metadata_dst *tun_dst;
+	struct tunnel_match_key key;
+	u32 tun_id, enc_opts_id;
+	struct net_device *dev;
+	int err;
+
+	enc_opts_id = tunnel_id & ENC_OPTS_BITS_MASK;
+	tun_id = tunnel_id >> ENC_OPTS_BITS;
+
+	if (!tun_id)
+		return true;
+
+	uplink_rpriv = mlx5_eswitch_get_uplink_priv(esw, REP_ETH);
+	uplink_priv = &uplink_rpriv->uplink_priv;
+
+	err = mapping_find(uplink_priv->tunnel_mapping, tun_id, &key);
+	if (err) {
+		WARN_ON_ONCE(true);
+		netdev_dbg(priv->netdev,
+			   "Couldn't find tunnel for tun_id: %d, err: %d\n",
+			   tun_id, err);
+		return false;
+	}
+
+	if (enc_opts_id) {
+		err = mapping_find(uplink_priv->tunnel_enc_opts_mapping,
+				   enc_opts_id, &enc_opts);
+		if (err) {
+			netdev_dbg(priv->netdev,
+				   "Couldn't find tunnel (opts) for tun_id: %d, err: %d\n",
+				   enc_opts_id, err);
+			return false;
+		}
+	}
+
+	tun_dst = tun_rx_dst(enc_opts.len);
+	if (!tun_dst) {
+		WARN_ON_ONCE(true);
+		return false;
+	}
+
+	ip_tunnel_key_init(&tun_dst->u.tun_info.key,
+			   key.enc_ipv4.src, key.enc_ipv4.dst,
+			   key.enc_ip.tos, key.enc_ip.ttl,
+			   0, /* label */
+			   key.enc_tp.src, key.enc_tp.dst,
+			   key32_to_tunnel_id(key.enc_key_id.keyid),
+			   TUNNEL_KEY);
+
+	if (enc_opts.len)
+		ip_tunnel_info_opts_set(&tun_dst->u.tun_info, enc_opts.data,
+					enc_opts.len, enc_opts.dst_opt_type);
+
+	skb_dst_set(skb, (struct dst_entry *)tun_dst);
+	dev = dev_get_by_index(&init_net, key.filter_ifindex);
+	if (!dev) {
+		netdev_dbg(priv->netdev,
+			   "Couldn't find tunnel device with ifindex: %d\n",
+			   key.filter_ifindex);
+		return false;
+	}
+
+	/* Set tun_dev so we do dev_put() after datapath */
+	tc_priv->tun_dev = dev;
+
+	skb->dev = dev;
+
+	return true;
+}
+#endif /* CONFIG_NET_TC_SKB_EXT */
+
+bool mlx5e_tc_rep_update_skb(struct mlx5_cqe64 *cqe,
+			     struct sk_buff *skb,
+			     struct mlx5e_tc_update_priv *tc_priv)
+{
+#if IS_ENABLED(CONFIG_NET_TC_SKB_EXT)
+	u32 chain = 0, reg_c0, reg_c1, tunnel_id;
+	struct tc_skb_ext *tc_skb_ext;
+	struct mlx5_eswitch *esw;
+	struct mlx5e_priv *priv;
+	int tunnel_moffset;
+	int err;
+
+	reg_c0 = (be32_to_cpu(cqe->sop_drop_qpn) & MLX5E_TC_FLOW_ID_MASK);
+	if (reg_c0 == MLX5_FS_DEFAULT_FLOW_TAG)
+		reg_c0 = 0;
+	reg_c1 = be32_to_cpu(cqe->imm_inval_pkey);
+
+	if (!reg_c0)
+		return true;
+
+	priv = netdev_priv(skb->dev);
+	esw = priv->mdev->priv.eswitch;
+
+	err = mlx5_eswitch_get_chain_for_tag(esw, reg_c0, &chain);
+	if (err) {
+		netdev_dbg(priv->netdev,
+			   "Couldn't find chain for chain tag: %d, err: %d\n",
+			   reg_c0, err);
+		return false;
+	}
+
+	if (chain) {
+		tc_skb_ext = skb_ext_add(skb, TC_SKB_EXT);
+		if (!tc_skb_ext) {
+			WARN_ON(1);
+			return false;
+		}
+
+		tc_skb_ext->chain = chain;
+	}
+
+	tunnel_moffset = mlx5e_tc_attr_to_reg_mappings[TUNNEL_TO_REG].moffset;
+	tunnel_id = reg_c1 >> (8 * tunnel_moffset);
+	return mlx5e_restore_tunnel(priv, skb, tc_priv, tunnel_id);
+#endif /* CONFIG_NET_TC_SKB_EXT */
+
+	return true;
+}
+
+void mlx5_tc_rep_post_napi_receive(struct mlx5e_tc_update_priv *tc_priv)
+{
+	if (tc_priv->tun_dev)
+		dev_put(tc_priv->tun_dev);
+}

drivers/net/ethernet/mellanox/mlx5/core/en_tc.h

@@ -91,9 +91,54 @@ int mlx5e_tc_num_filters(struct mlx5e_priv *priv, unsigned long flags);
 
 void mlx5e_tc_reoffload_flows_work(struct work_struct *work);
 
+enum mlx5e_tc_attr_to_reg {
+	CHAIN_TO_REG,
+	TUNNEL_TO_REG,
+};
+
+struct mlx5e_tc_attr_to_reg_mapping {
+	int mfield; /* rewrite field */
+	int moffset; /* offset of mfield */
+	int mlen; /* bytes to rewrite/match */
+
+	int soffset; /* offset of spec for match */
+};
+
+extern struct mlx5e_tc_attr_to_reg_mapping mlx5e_tc_attr_to_reg_mappings[];
+
 bool mlx5e_is_valid_eswitch_fwd_dev(struct mlx5e_priv *priv,
 				    struct net_device *out_dev);
 
+struct mlx5e_tc_update_priv {
+	struct net_device *tun_dev;
+};
+
+bool mlx5e_tc_rep_update_skb(struct mlx5_cqe64 *cqe, struct sk_buff *skb,
+			     struct mlx5e_tc_update_priv *tc_priv);
+
+void mlx5_tc_rep_post_napi_receive(struct mlx5e_tc_update_priv *tc_priv);
+
+struct mlx5e_tc_mod_hdr_acts {
+	int num_actions;
+	int max_actions;
+	void *actions;
+};
+
+int mlx5e_tc_match_to_reg_set(struct mlx5_core_dev *mdev,
+			      struct mlx5e_tc_mod_hdr_acts *mod_hdr_acts,
+			      enum mlx5e_tc_attr_to_reg type,
+			      u32 data);
+
+void mlx5e_tc_match_to_reg_match(struct mlx5_flow_spec *spec,
+				 enum mlx5e_tc_attr_to_reg type,
+				 u32 data,
+				 u32 mask);
+
+int alloc_mod_hdr_actions(struct mlx5_core_dev *mdev,
+			  int namespace,
+			  struct mlx5e_tc_mod_hdr_acts *mod_hdr_acts);
+void dealloc_mod_hdr_actions(struct mlx5e_tc_mod_hdr_acts *mod_hdr_acts);
+
 #else /* CONFIG_MLX5_ESWITCH */
 static inline int  mlx5e_tc_nic_init(struct mlx5e_priv *priv) { return 0; }
 static inline void mlx5e_tc_nic_cleanup(struct mlx5e_priv *priv) {}

drivers/net/ethernet/mellanox/mlx5/core/eswitch.h

@@ -197,6 +197,10 @@ struct mlx5_eswitch_fdb {
 };
 
 struct mlx5_esw_offload {
+	struct mlx5_flow_table *ft_offloads_restore;
+	struct mlx5_flow_group *restore_group;
+	struct mlx5_modify_hdr *restore_copy_hdr_id;
+
 	struct mlx5_flow_table *ft_offloads;
 	struct mlx5_flow_group *vport_rx_group;
 	struct mlx5_eswitch_rep *vport_reps;
@@ -636,6 +640,11 @@ esw_vport_destroy_offloads_acl_tables(struct mlx5_eswitch *esw,
 int mlx5_esw_vport_tbl_get(struct mlx5_eswitch *esw);
 void mlx5_esw_vport_tbl_put(struct mlx5_eswitch *esw);
 
+struct mlx5_flow_handle *
+esw_add_restore_rule(struct mlx5_eswitch *esw, u32 tag);
+u32
+esw_get_max_restore_tag(struct mlx5_eswitch *esw);
+
 #else  /* CONFIG_MLX5_ESWITCH */
 /* eswitch API stubs */
 static inline int  mlx5_eswitch_init(struct mlx5_core_dev *dev) { return 0; }
@@ -651,6 +660,12 @@ static inline const u32 *mlx5_esw_query_functions(struct mlx5_core_dev *dev)
 
 static inline void mlx5_eswitch_update_num_of_vfs(struct mlx5_eswitch *esw, const int num_vfs) {}
 
+static struct mlx5_flow_handle *
+esw_add_restore_rule(struct mlx5_eswitch *esw, u32 tag)
+{
+	return ERR_PTR(-EOPNOTSUPP);
+}
+
 #endif /* CONFIG_MLX5_ESWITCH */
 
 #endif /* __MLX5_ESWITCH_H__ */

drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c

@@ -260,7 +260,8 @@ mlx5_eswitch_set_rule_source_port(struct mlx5_eswitch *esw,
 								   attr->in_rep->vport));
 
 		misc2 = MLX5_ADDR_OF(fte_match_param, spec->match_criteria, misc_parameters_2);
-		MLX5_SET_TO_ONES(fte_match_set_misc2, misc2, metadata_reg_c_0);
+		MLX5_SET(fte_match_set_misc2, misc2, metadata_reg_c_0,
+			 mlx5_eswitch_get_vport_metadata_mask());
 
 		spec->match_criteria_enable |= MLX5_MATCH_MISC_PARAMETERS_2;
 		misc = MLX5_ADDR_OF(fte_match_param, spec->match_criteria, misc_parameters);
@@ -781,9 +782,11 @@ static int esw_set_passing_vport_metadata(struct mlx5_eswitch *esw, bool enable)
 					 esw_vport_context.fdb_to_vport_reg_c_id);
 
 	if (enable)
-		fdb_to_vport_reg_c_id |= MLX5_FDB_TO_VPORT_REG_C_0;
+		fdb_to_vport_reg_c_id |= MLX5_FDB_TO_VPORT_REG_C_0 |
+					 MLX5_FDB_TO_VPORT_REG_C_1;
 	else
-		fdb_to_vport_reg_c_id &= ~MLX5_FDB_TO_VPORT_REG_C_0;
+		fdb_to_vport_reg_c_id &= ~(MLX5_FDB_TO_VPORT_REG_C_0 |
+					   MLX5_FDB_TO_VPORT_REG_C_1);
 
 	MLX5_SET(modify_esw_vport_context_in, in,
 		 esw_vport_context.fdb_to_vport_reg_c_id, fdb_to_vport_reg_c_id);
@@ -805,7 +808,8 @@ static void peer_miss_rules_setup(struct mlx5_eswitch *esw,
 	if (mlx5_eswitch_vport_match_metadata_enabled(esw)) {
 		misc = MLX5_ADDR_OF(fte_match_param, spec->match_criteria,
 				    misc_parameters_2);
-		MLX5_SET_TO_ONES(fte_match_set_misc2, misc, metadata_reg_c_0);
+		MLX5_SET(fte_match_set_misc2, misc, metadata_reg_c_0,
+			 mlx5_eswitch_get_vport_metadata_mask());
 
 		spec->match_criteria_enable = MLX5_MATCH_MISC_PARAMETERS_2;
 	} else {
@@ -1020,6 +1024,56 @@ static int esw_add_fdb_miss_rule(struct mlx5_eswitch *esw)
 	return err;
 }
 
+struct mlx5_flow_handle *
+esw_add_restore_rule(struct mlx5_eswitch *esw, u32 tag)
+{
+	struct mlx5_flow_act flow_act = { .flags = FLOW_ACT_NO_APPEND, };
+	struct mlx5_flow_table *ft = esw->offloads.ft_offloads_restore;
+	struct mlx5_flow_context *flow_context;
+	struct mlx5_flow_handle *flow_rule;
+	struct mlx5_flow_destination dest;
+	struct mlx5_flow_spec *spec;
+	void *misc;
+
+	spec = kzalloc(sizeof(*spec), GFP_KERNEL);
+	if (!spec)
+		return ERR_PTR(-ENOMEM);
+
+	misc = MLX5_ADDR_OF(fte_match_param, spec->match_criteria,
+			    misc_parameters_2);
+	MLX5_SET(fte_match_set_misc2, misc, metadata_reg_c_0,
+		 ESW_CHAIN_TAG_METADATA_MASK);
+	misc = MLX5_ADDR_OF(fte_match_param, spec->match_value,
+			    misc_parameters_2);
+	MLX5_SET(fte_match_set_misc2, misc, metadata_reg_c_0, tag);
+	spec->match_criteria_enable = MLX5_MATCH_MISC_PARAMETERS_2;
+	flow_act.action = MLX5_FLOW_CONTEXT_ACTION_FWD_DEST |
+			  MLX5_FLOW_CONTEXT_ACTION_MOD_HDR;
+	flow_act.modify_hdr = esw->offloads.restore_copy_hdr_id;
+
+	flow_context = &spec->flow_context;
+	flow_context->flags |= FLOW_CONTEXT_HAS_TAG;
+	flow_context->flow_tag = tag;
+	dest.type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
+	dest.ft = esw->offloads.ft_offloads;
+
+	flow_rule = mlx5_add_flow_rules(ft, spec, &flow_act, &dest, 1);
+	kfree(spec);
+
+	if (IS_ERR(flow_rule))
+		esw_warn(esw->dev,
+			 "Failed to create restore rule for tag: %d, err(%d)\n",
+			 tag, (int)PTR_ERR(flow_rule));
+
+	return flow_rule;
+}
+
+u32
+esw_get_max_restore_tag(struct mlx5_eswitch *esw)
+{
+	return ESW_CHAIN_TAG_METADATA_MASK;
+}
+
 #define MAX_PF_SQ 256
 #define MAX_SQ_NVPORTS 32
 
@@ -1035,8 +1089,9 @@ static void esw_set_flow_group_source_port(struct mlx5_eswitch *esw,
 			 match_criteria_enable,
 			 MLX5_MATCH_MISC_PARAMETERS_2);
 
-		MLX5_SET_TO_ONES(fte_match_param, match_criteria,
-				 misc_parameters_2.metadata_reg_c_0);
+		MLX5_SET(fte_match_param, match_criteria,
+			 misc_parameters_2.metadata_reg_c_0,
+			 mlx5_eswitch_get_vport_metadata_mask());
 	} else {
 		MLX5_SET(create_flow_group_in, flow_group_in,
 			 match_criteria_enable,
@@ -1241,6 +1296,7 @@ static int esw_create_offloads_table(struct mlx5_eswitch *esw, int nvports)
 	}
 
 	ft_attr.max_fte = nvports + MLX5_ESW_MISS_FLOWS;
+	ft_attr.prio = 1;
 
 	ft_offloads = mlx5_create_flow_table(ns, &ft_attr);
 	if (IS_ERR(ft_offloads)) {
@@ -1318,7 +1374,8 @@ mlx5_eswitch_create_vport_rx_rule(struct mlx5_eswitch *esw, u16 vport,
 			 mlx5_eswitch_get_vport_metadata_for_match(esw, vport));
 
 		misc = MLX5_ADDR_OF(fte_match_param, spec->match_criteria, misc_parameters_2);
-		MLX5_SET_TO_ONES(fte_match_set_misc2, misc, metadata_reg_c_0);
+		MLX5_SET(fte_match_set_misc2, misc, metadata_reg_c_0,
+			 mlx5_eswitch_get_vport_metadata_mask());
 
 		spec->match_criteria_enable = MLX5_MATCH_MISC_PARAMETERS_2;
 	} else {
@@ -1344,6 +1401,7 @@ mlx5_eswitch_create_vport_rx_rule(struct mlx5_eswitch *esw, u16 vport,
 	return flow_rule;
 }
 
+
 static int mlx5_eswitch_inline_mode_get(const struct mlx5_eswitch *esw, u8 *mode)
 {
 	u8 prev_mlx5_mode, mlx5_mode = MLX5_INLINE_MODE_L2;
@@ -1381,6 +1439,102 @@ static int mlx5_eswitch_inline_mode_get(const struct mlx5_eswitch *esw, u8 *mode
 	return 0;
 }       
 
+static void esw_destroy_restore_table(struct mlx5_eswitch *esw)
+{
+	struct mlx5_esw_offload *offloads = &esw->offloads;
+
+	mlx5_modify_header_dealloc(esw->dev, offloads->restore_copy_hdr_id);
+	mlx5_destroy_flow_group(offloads->restore_group);
+	mlx5_destroy_flow_table(offloads->ft_offloads_restore);
+}
+
+static int esw_create_restore_table(struct mlx5_eswitch *esw)
+{
+	u8 modact[MLX5_UN_SZ_BYTES(set_action_in_add_action_in_auto)] = {};
+	int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
+	struct mlx5_flow_table_attr ft_attr = {};
+	struct mlx5_core_dev *dev = esw->dev;
+	struct mlx5_flow_namespace *ns;
+	struct mlx5_modify_hdr *mod_hdr;
+	void *match_criteria, *misc;
+	struct mlx5_flow_table *ft;
+	struct mlx5_flow_group *g;
+	u32 *flow_group_in;
+	int err = 0;
+
+	ns = mlx5_get_flow_namespace(dev, MLX5_FLOW_NAMESPACE_OFFLOADS);
+	if (!ns) {
+		esw_warn(esw->dev, "Failed to get offloads flow namespace\n");
+		return -EOPNOTSUPP;
+	}
+
+	flow_group_in = kvzalloc(inlen, GFP_KERNEL);
+	if (!flow_group_in) {
+		err = -ENOMEM;
+		goto out_free;
+	}
+
+	ft_attr.max_fte = 1 << ESW_CHAIN_TAG_METADATA_BITS;
+	ft = mlx5_create_flow_table(ns, &ft_attr);
+	if (IS_ERR(ft)) {
+		err = PTR_ERR(ft);
+		esw_warn(esw->dev, "Failed to create restore table, err %d\n",
+			 err);
+		goto out_free;
+	}
+
+	memset(flow_group_in, 0, inlen);
+	match_criteria = MLX5_ADDR_OF(create_flow_group_in, flow_group_in,
+				      match_criteria);
+	misc = MLX5_ADDR_OF(fte_match_param, match_criteria,
+			    misc_parameters_2);
+
+	MLX5_SET(fte_match_set_misc2, misc, metadata_reg_c_0,
+		 ESW_CHAIN_TAG_METADATA_MASK);
+	MLX5_SET(create_flow_group_in, flow_group_in, start_flow_index, 0);
+	MLX5_SET(create_flow_group_in, flow_group_in, end_flow_index,
+		 ft_attr.max_fte - 1);
+	MLX5_SET(create_flow_group_in, flow_group_in, match_criteria_enable,
+		 MLX5_MATCH_MISC_PARAMETERS_2);
+	g = mlx5_create_flow_group(ft, flow_group_in);
+	if (IS_ERR(g)) {
+		err = PTR_ERR(g);
+		esw_warn(dev, "Failed to create restore flow group, err: %d\n",
+			 err);
+		goto err_group;
+	}
+
+	MLX5_SET(copy_action_in, modact, action_type, MLX5_ACTION_TYPE_COPY);
+	MLX5_SET(copy_action_in, modact, src_field,
+		 MLX5_ACTION_IN_FIELD_METADATA_REG_C_1);
+	MLX5_SET(copy_action_in, modact, dst_field,
+		 MLX5_ACTION_IN_FIELD_METADATA_REG_B);
+	mod_hdr = mlx5_modify_header_alloc(esw->dev,
+					   MLX5_FLOW_NAMESPACE_KERNEL, 1,
+					   modact);
+	if (IS_ERR(mod_hdr)) {
+		esw_warn(dev, "Failed to create restore mod header, err: %d\n",
+			 err);
+		err = PTR_ERR(mod_hdr);
+		goto err_mod_hdr;
+	}
+
+	esw->offloads.ft_offloads_restore = ft;
+	esw->offloads.restore_group = g;
+	esw->offloads.restore_copy_hdr_id = mod_hdr;
+
+	return 0;
+
+err_mod_hdr:
+	mlx5_destroy_flow_group(g);
+err_group:
+	mlx5_destroy_flow_table(ft);
+out_free:
+	kvfree(flow_group_in);
+
+	return err;
+}
+
 static int esw_offloads_start(struct mlx5_eswitch *esw,
 			      struct netlink_ext_ack *extack)
 {
@@ -1825,11 +1979,19 @@ static int esw_vport_add_ingress_acl_modify_metadata(struct mlx5_eswitch *esw,
 	static const struct mlx5_flow_spec spec = {};
 	struct mlx5_flow_act flow_act = {};
 	int err = 0;
+	u32 key;
+
+	key = mlx5_eswitch_get_vport_metadata_for_match(esw, vport->vport);
+	key >>= ESW_SOURCE_PORT_METADATA_OFFSET;
 
 	MLX5_SET(set_action_in, action, action_type, MLX5_ACTION_TYPE_SET);
-	MLX5_SET(set_action_in, action, field, MLX5_ACTION_IN_FIELD_METADATA_REG_C_0);
-	MLX5_SET(set_action_in, action, data,
-		 mlx5_eswitch_get_vport_metadata_for_match(esw, vport->vport));
+	MLX5_SET(set_action_in, action, field,
+		 MLX5_ACTION_IN_FIELD_METADATA_REG_C_0);
+	MLX5_SET(set_action_in, action, data, key);
+	MLX5_SET(set_action_in, action, offset,
+		 ESW_SOURCE_PORT_METADATA_OFFSET);
+	MLX5_SET(set_action_in, action, length,
+		 ESW_SOURCE_PORT_METADATA_BITS);
 
 	vport->ingress.offloads.modify_metadata =
 		mlx5_modify_header_alloc(esw->dev, MLX5_FLOW_NAMESPACE_ESW_INGRESS,
@@ -2144,13 +2306,17 @@ static int esw_offloads_steering_init(struct mlx5_eswitch *esw)
 	if (err)
 		return err;
 
-	err = esw_create_offloads_fdb_tables(esw, total_vports);
+	err = esw_create_offloads_table(esw, total_vports);
 	if (err)
-		goto create_fdb_err;
+		goto create_offloads_err;
 
-	err = esw_create_offloads_table(esw, total_vports);
+	err = esw_create_restore_table(esw);
+	if (err)
+		goto create_restore_err;
+
+	err = esw_create_offloads_fdb_tables(esw, total_vports);
 	if (err)
-		goto create_ft_err;
+		goto create_fdb_err;
 
 	err = esw_create_vport_rx_group(esw, total_vports);
 	if (err)
@@ -2162,12 +2328,12 @@ static int esw_offloads_steering_init(struct mlx5_eswitch *esw)
 	return 0;
 
 create_fg_err:
-	esw_destroy_offloads_table(esw);
-
-create_ft_err:
 	esw_destroy_offloads_fdb_tables(esw);
-
 create_fdb_err:
+	esw_destroy_restore_table(esw);
+create_restore_err:
+	esw_destroy_offloads_table(esw);
+create_offloads_err:
 	esw_destroy_uplink_offloads_acl_tables(esw);
 
 	return err;
@@ -2177,8 +2343,9 @@ static void esw_offloads_steering_cleanup(struct mlx5_eswitch *esw)
 {
 	mutex_destroy(&esw->fdb_table.offloads.vports.lock);
 	esw_destroy_vport_rx_group(esw);
-	esw_destroy_offloads_table(esw);
 	esw_destroy_offloads_fdb_tables(esw);
+	esw_destroy_restore_table(esw);
+	esw_destroy_offloads_table(esw);
 	esw_destroy_uplink_offloads_acl_tables(esw);
 }
 
@@ -2670,9 +2837,41 @@ bool mlx5_eswitch_vport_match_metadata_enabled(const struct mlx5_eswitch *esw)
 }
 EXPORT_SYMBOL(mlx5_eswitch_vport_match_metadata_enabled);
 
-u32 mlx5_eswitch_get_vport_metadata_for_match(const struct mlx5_eswitch *esw,
+u32 mlx5_eswitch_get_vport_metadata_for_match(struct mlx5_eswitch *esw,
 					      u16 vport_num)
 {
-	return ((MLX5_CAP_GEN(esw->dev, vhca_id) & 0xffff) << 16) | vport_num;
+	u32 vport_num_mask = GENMASK(ESW_VPORT_BITS - 1, 0);
+	u32 vhca_id_mask = GENMASK(ESW_VHCA_ID_BITS - 1, 0);
+	u32 vhca_id = MLX5_CAP_GEN(esw->dev, vhca_id);
+	u32 val;
+
+	/* Make sure the vhca_id fits the ESW_VHCA_ID_BITS */
+	WARN_ON_ONCE(vhca_id >= BIT(ESW_VHCA_ID_BITS));
+
+	/* Trim vhca_id to ESW_VHCA_ID_BITS */
+	vhca_id &= vhca_id_mask;
+
+	/* Make sure pf and ecpf map to end of ESW_VPORT_BITS range so they
+	 * don't overlap with VF numbers, and themselves, after trimming.
+	 */
+	WARN_ON_ONCE((MLX5_VPORT_UPLINK & vport_num_mask) <
+		     vport_num_mask - 1);
+	WARN_ON_ONCE((MLX5_VPORT_ECPF & vport_num_mask) <
+		     vport_num_mask - 1);
+	WARN_ON_ONCE((MLX5_VPORT_UPLINK & vport_num_mask) ==
+		     (MLX5_VPORT_ECPF & vport_num_mask));
+
+	/* Make sure that the VF vport_num fits ESW_VPORT_BITS and don't
+	 * overlap with pf and ecpf.
+	 */
+	if (vport_num != MLX5_VPORT_UPLINK &&
+	    vport_num != MLX5_VPORT_ECPF)
+		WARN_ON_ONCE(vport_num >= vport_num_mask - 1);
+
+	/* We can now trim vport_num to ESW_VPORT_BITS */
+	vport_num &= vport_num_mask;
+
+	val = (vhca_id << ESW_VPORT_BITS) | vport_num;
+	return val << (32 - ESW_SOURCE_PORT_METADATA_BITS);
 }
 EXPORT_SYMBOL(mlx5_eswitch_get_vport_metadata_for_match);

drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads_chains.c

@@ -6,14 +6,17 @@
 #include <linux/mlx5/fs.h>
 
 #include "eswitch_offloads_chains.h"
+#include "en/mapping.h"
 #include "mlx5_core.h"
 #include "fs_core.h"
 #include "eswitch.h"
 #include "en.h"
+#include "en_tc.h"
 
 #define esw_chains_priv(esw) ((esw)->fdb_table.offloads.esw_chains_priv)
 #define esw_chains_lock(esw) (esw_chains_priv(esw)->lock)
 #define esw_chains_ht(esw) (esw_chains_priv(esw)->chains_ht)
+#define esw_chains_mapping(esw) (esw_chains_priv(esw)->chains_mapping)
 #define esw_prios_ht(esw) (esw_chains_priv(esw)->prios_ht)
 #define fdb_pool_left(esw) (esw_chains_priv(esw)->fdb_left)
 #define tc_slow_fdb(esw) ((esw)->fdb_table.offloads.slow_fdb)
@@ -43,6 +46,7 @@ struct mlx5_esw_chains_priv {
 	struct mutex lock;
 
 	struct mlx5_flow_table *tc_end_fdb;
+	struct mapping_ctx *chains_mapping;
 
 	int fdb_left[ARRAY_SIZE(ESW_POOLS)];
 };
@@ -53,9 +57,12 @@ struct fdb_chain {
 	u32 chain;
 
 	int ref;
+	int id;
 
 	struct mlx5_eswitch *esw;
 	struct list_head prios_list;
+	struct mlx5_flow_handle *restore_rule;
+	struct mlx5_modify_hdr *miss_modify_hdr;
 };
 
 struct fdb_prio_key {
@@ -261,6 +268,70 @@ mlx5_esw_chains_destroy_fdb_table(struct mlx5_eswitch *esw,
 	mlx5_destroy_flow_table(fdb);
 }
 
+static int
+create_fdb_chain_restore(struct fdb_chain *fdb_chain)
+{
+	char modact[MLX5_UN_SZ_BYTES(set_action_in_add_action_in_auto)];
+	struct mlx5_eswitch *esw = fdb_chain->esw;
+	struct mlx5_modify_hdr *mod_hdr;
+	u32 index;
+	int err;
+
+	if (fdb_chain->chain == mlx5_esw_chains_get_ft_chain(esw))
+		return 0;
+
+	err = mapping_add(esw_chains_mapping(esw), &fdb_chain->chain, &index);
+	if (err)
+		return err;
+	if (index == MLX5_FS_DEFAULT_FLOW_TAG) {
+		/* we got the special default flow tag id, so we won't know
+		 * if we actually marked the packet with the restore rule
+		 * we create.
+		 *
+		 * This case isn't possible with MLX5_FS_DEFAULT_FLOW_TAG = 0.
+		 */
+		err = mapping_add(esw_chains_mapping(esw),
+				  &fdb_chain->chain, &index);
+		mapping_remove(esw_chains_mapping(esw),
+			       MLX5_FS_DEFAULT_FLOW_TAG);
+		if (err)
+			return err;
+	}
+
+	fdb_chain->id = index;
+
+	MLX5_SET(set_action_in, modact, action_type, MLX5_ACTION_TYPE_SET);
+	MLX5_SET(set_action_in, modact, field,
+		 mlx5e_tc_attr_to_reg_mappings[CHAIN_TO_REG].mfield);
+	MLX5_SET(set_action_in, modact, offset,
+		 mlx5e_tc_attr_to_reg_mappings[CHAIN_TO_REG].moffset * 8);
+	MLX5_SET(set_action_in, modact, length,
+		 mlx5e_tc_attr_to_reg_mappings[CHAIN_TO_REG].mlen * 8);
+	MLX5_SET(set_action_in, modact, data, fdb_chain->id);
+	mod_hdr = mlx5_modify_header_alloc(esw->dev, MLX5_FLOW_NAMESPACE_FDB,
+					   1, modact);
+	if (IS_ERR(mod_hdr)) {
+		err = PTR_ERR(mod_hdr);
+		goto err_mod_hdr;
+	}
+	fdb_chain->miss_modify_hdr = mod_hdr;
+
+	fdb_chain->restore_rule = esw_add_restore_rule(esw, fdb_chain->id);
+	if (IS_ERR(fdb_chain->restore_rule)) {
+		err = PTR_ERR(fdb_chain->restore_rule);
+		goto err_rule;
+	}
+
+	return 0;
+
+err_rule:
+	mlx5_modify_header_dealloc(esw->dev, fdb_chain->miss_modify_hdr);
+err_mod_hdr:
+	/* Datapath can't find this mapping, so we can safely remove it */
+	mapping_remove(esw_chains_mapping(esw), fdb_chain->id);
+	return err;
+}
+
 static struct fdb_chain *
 mlx5_esw_chains_create_fdb_chain(struct mlx5_eswitch *esw, u32 chain)
 {
@@ -275,6 +346,10 @@ mlx5_esw_chains_create_fdb_chain(struct mlx5_eswitch *esw, u32 chain)
 	fdb_chain->chain = chain;
 	INIT_LIST_HEAD(&fdb_chain->prios_list);
 
+	err = create_fdb_chain_restore(fdb_chain);
+	if (err)
+		goto err_restore;
+
 	err = rhashtable_insert_fast(&esw_chains_ht(esw), &fdb_chain->node,
 				     chain_params);
 	if (err)
@@ -283,6 +358,12 @@ mlx5_esw_chains_create_fdb_chain(struct mlx5_eswitch *esw, u32 chain)
 	return fdb_chain;
 
 err_insert:
+	if (fdb_chain->chain != mlx5_esw_chains_get_ft_chain(esw)) {
+		mlx5_del_flow_rules(fdb_chain->restore_rule);
+		mlx5_modify_header_dealloc(esw->dev,
+					   fdb_chain->miss_modify_hdr);
+	}
+err_restore:
 	kvfree(fdb_chain);
 	return ERR_PTR(err);
 }
@@ -294,6 +375,15 @@ mlx5_esw_chains_destroy_fdb_chain(struct fdb_chain *fdb_chain)
 
 	rhashtable_remove_fast(&esw_chains_ht(esw), &fdb_chain->node,
 			       chain_params);
+
+	if (fdb_chain->chain != mlx5_esw_chains_get_ft_chain(esw)) {
+		mlx5_del_flow_rules(fdb_chain->restore_rule);
+		mlx5_modify_header_dealloc(esw->dev,
+					   fdb_chain->miss_modify_hdr);
+
+		mapping_remove(esw_chains_mapping(esw), fdb_chain->id);
+	}
+
 	kvfree(fdb_chain);
 }
 
@@ -316,10 +406,12 @@ mlx5_esw_chains_get_fdb_chain(struct mlx5_eswitch *esw, u32 chain)
 }
 
 static struct mlx5_flow_handle *
-mlx5_esw_chains_add_miss_rule(struct mlx5_flow_table *fdb,
+mlx5_esw_chains_add_miss_rule(struct fdb_chain *fdb_chain,
+			      struct mlx5_flow_table *fdb,
 			      struct mlx5_flow_table *next_fdb)
 {
 	static const struct mlx5_flow_spec spec = {};
+	struct mlx5_eswitch *esw = fdb_chain->esw;
 	struct mlx5_flow_destination dest = {};
 	struct mlx5_flow_act act = {};
 
@@ -328,6 +420,11 @@ mlx5_esw_chains_add_miss_rule(struct mlx5_flow_table *fdb,
 	dest.type  = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
 	dest.ft = next_fdb;
 
+	if (fdb_chain->chain != mlx5_esw_chains_get_ft_chain(esw)) {
+		act.modify_hdr = fdb_chain->miss_modify_hdr;
+		act.action |= MLX5_FLOW_CONTEXT_ACTION_MOD_HDR;
+	}
+
 	return mlx5_add_flow_rules(fdb, &spec, &act, &dest, 1);
 }
 
@@ -351,7 +448,8 @@ mlx5_esw_chains_update_prio_prevs(struct fdb_prio *fdb_prio,
 	list_for_each_entry_continue_reverse(pos,
 					     &fdb_chain->prios_list,
 					     list) {
-		miss_rules[n] = mlx5_esw_chains_add_miss_rule(pos->fdb,
+		miss_rules[n] = mlx5_esw_chains_add_miss_rule(fdb_chain,
+							      pos->fdb,
 							      next_fdb);
 		if (IS_ERR(miss_rules[n])) {
 			err = PTR_ERR(miss_rules[n]);
@@ -465,7 +563,7 @@ mlx5_esw_chains_create_fdb_prio(struct mlx5_eswitch *esw,
 	}
 
 	/* Add miss rule to next_fdb */
-	miss_rule = mlx5_esw_chains_add_miss_rule(fdb, next_fdb);
+	miss_rule = mlx5_esw_chains_add_miss_rule(fdb_chain, fdb, next_fdb);
 	if (IS_ERR(miss_rule)) {
 		err = PTR_ERR(miss_rule);
 		goto err_miss_rule;
@@ -630,6 +728,7 @@ mlx5_esw_chains_init(struct mlx5_eswitch *esw)
 	struct mlx5_esw_chains_priv *chains_priv;
 	struct mlx5_core_dev *dev = esw->dev;
 	u32 max_flow_counter, fdb_max;
+	struct mapping_ctx *mapping;
 	int err;
 
 	chains_priv = kzalloc(sizeof(*chains_priv), GFP_KERNEL);
@@ -666,10 +765,20 @@ mlx5_esw_chains_init(struct mlx5_eswitch *esw)
 	if (err)
 		goto init_prios_ht_err;
 
+	mapping = mapping_create(sizeof(u32), esw_get_max_restore_tag(esw),
+				 true);
+	if (IS_ERR(mapping)) {
+		err = PTR_ERR(mapping);
+		goto mapping_err;
+	}
+	esw_chains_mapping(esw) = mapping;
+
 	mutex_init(&esw_chains_lock(esw));
 
 	return 0;
 
+mapping_err:
+	rhashtable_destroy(&esw_prios_ht(esw));
 init_prios_ht_err:
 	rhashtable_destroy(&esw_chains_ht(esw));
 init_chains_ht_err:
@@ -681,6 +790,7 @@ static void
 mlx5_esw_chains_cleanup(struct mlx5_eswitch *esw)
 {
 	mutex_destroy(&esw_chains_lock(esw));
+	mapping_destroy(esw_chains_mapping(esw));
 	rhashtable_destroy(&esw_prios_ht(esw));
 	rhashtable_destroy(&esw_chains_ht(esw));
 
@@ -759,3 +869,17 @@ mlx5_esw_chains_destroy(struct mlx5_eswitch *esw)
 	mlx5_esw_chains_close(esw);
 	mlx5_esw_chains_cleanup(esw);
 }
+
+int mlx5_eswitch_get_chain_for_tag(struct mlx5_eswitch *esw, u32 tag,
+				   u32 *chain)
+{
+	int err;
+
+	err = mapping_find(esw_chains_mapping(esw), tag, chain);
+	if (err) {
+		esw_warn(esw->dev, "Can't find chain for tag: %d\n", tag);
+		return -ENOENT;
+	}
+
+	return 0;
+}

drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads_chains.h

@@ -28,5 +28,7 @@ mlx5_esw_chains_get_tc_end_ft(struct mlx5_eswitch *esw);
 int mlx5_esw_chains_create(struct mlx5_eswitch *esw);
 void mlx5_esw_chains_destroy(struct mlx5_eswitch *esw);
 
-#endif /* __ML5_ESW_CHAINS_H__ */
+int
+mlx5_eswitch_get_chain_for_tag(struct mlx5_eswitch *esw, u32 tag, u32 *chain);
 
+#endif /* __ML5_ESW_CHAINS_H__ */

drivers/net/ethernet/mellanox/mlx5/core/fs_core.c

@@ -111,8 +111,8 @@
 #define ANCHOR_MIN_LEVEL (BY_PASS_MIN_LEVEL + 1)
 
 #define OFFLOADS_MAX_FT 2
-#define OFFLOADS_NUM_PRIOS 1
-#define OFFLOADS_MIN_LEVEL (ANCHOR_MIN_LEVEL + 1)
+#define OFFLOADS_NUM_PRIOS 2
+#define OFFLOADS_MIN_LEVEL (ANCHOR_MIN_LEVEL + OFFLOADS_NUM_PRIOS)
 
 #define LAG_PRIO_NUM_LEVELS 1
 #define LAG_NUM_PRIOS 1

include/linux/mlx5/eswitch.h

@@ -71,7 +71,28 @@ enum devlink_eswitch_encap_mode
 mlx5_eswitch_get_encap_mode(const struct mlx5_core_dev *dev);
 
 bool mlx5_eswitch_vport_match_metadata_enabled(const struct mlx5_eswitch *esw);
-u32 mlx5_eswitch_get_vport_metadata_for_match(const struct mlx5_eswitch *esw,
+
+/* Reg C0 usage:
+ * Reg C0 = < ESW_VHCA_ID_BITS(8) | ESW_VPORT BITS(8) | ESW_CHAIN_TAG(16) >
+ *
+ * Highest 8 bits of the reg c0 is the vhca_id, next 8 bits is vport_num,
+ * the rest (lowest 16 bits) is left for tc chain tag restoration.
+ * VHCA_ID + VPORT comprise the SOURCE_PORT matching.
+ */
+#define ESW_VHCA_ID_BITS 8
+#define ESW_VPORT_BITS 8
+#define ESW_SOURCE_PORT_METADATA_BITS (ESW_VHCA_ID_BITS + ESW_VPORT_BITS)
+#define ESW_SOURCE_PORT_METADATA_OFFSET (32 - ESW_SOURCE_PORT_METADATA_BITS)
+#define ESW_CHAIN_TAG_METADATA_BITS (32 - ESW_SOURCE_PORT_METADATA_BITS)
+#define ESW_CHAIN_TAG_METADATA_MASK GENMASK(ESW_CHAIN_TAG_METADATA_BITS - 1,\
+					    0)
+
+static inline u32 mlx5_eswitch_get_vport_metadata_mask(void)
+{
+	return GENMASK(31, 32 - ESW_SOURCE_PORT_METADATA_BITS);
+}
+
+u32 mlx5_eswitch_get_vport_metadata_for_match(struct mlx5_eswitch *esw,
 					      u16 vport_num);
 u8 mlx5_eswitch_mode(struct mlx5_eswitch *esw);
 #else  /* CONFIG_MLX5_ESWITCH */
@@ -94,11 +115,17 @@ mlx5_eswitch_vport_match_metadata_enabled(const struct mlx5_eswitch *esw)
 };
 
 static inline u32
-mlx5_eswitch_get_vport_metadata_for_match(const struct mlx5_eswitch *esw,
+mlx5_eswitch_get_vport_metadata_for_match(struct mlx5_eswitch *esw,
 					  int vport_num)
 {
 	return 0;
 };
+
+static inline u32
+mlx5_eswitch_get_vport_metadata_mask(void)
+{
+	return 0;
+}
 #endif /* CONFIG_MLX5_ESWITCH */
 
 #endif

include/net/pkt_cls.h

@@ -72,6 +72,10 @@ static inline struct Qdisc *tcf_block_q(struct tcf_block *block)
 
 int tcf_classify(struct sk_buff *skb, const struct tcf_proto *tp,
 		 struct tcf_result *res, bool compat_mode);
+int tcf_classify_ingress(struct sk_buff *skb,
+			 const struct tcf_block *ingress_block,
+			 const struct tcf_proto *tp, struct tcf_result *res,
+			 bool compat_mode);
 
 #else
 static inline bool tcf_block_shared(struct tcf_block *block)
@@ -133,6 +137,15 @@ static inline int tcf_classify(struct sk_buff *skb, const struct tcf_proto *tp,
 {
 	return TC_ACT_UNSPEC;
 }
+
+static inline int tcf_classify_ingress(struct sk_buff *skb,
+				       const struct tcf_block *ingress_block,
+				       const struct tcf_proto *tp,
+				       struct tcf_result *res, bool compat_mode)
+{
+	return TC_ACT_UNSPEC;
+}
+
 #endif
 
 static inline unsigned long

include/net/sch_generic.h

@@ -1269,6 +1269,7 @@ static inline void psched_ratecfg_getrate(struct tc_ratespec *res,
  */
 struct mini_Qdisc {
 	struct tcf_proto *filter_list;
+	struct tcf_block *block;
 	struct gnet_stats_basic_cpu __percpu *cpu_bstats;
 	struct gnet_stats_queue	__percpu *cpu_qstats;
 	struct rcu_head rcu;
@@ -1295,6 +1296,8 @@ void mini_qdisc_pair_swap(struct mini_Qdisc_pair *miniqp,
 			  struct tcf_proto *tp_head);
 void mini_qdisc_pair_init(struct mini_Qdisc_pair *miniqp, struct Qdisc *qdisc,
 			  struct mini_Qdisc __rcu **p_miniq);
+void mini_qdisc_pair_block_init(struct mini_Qdisc_pair *miniqp,
+				struct tcf_block *block);
 
 static inline int skb_tc_reinsert(struct sk_buff *skb, struct tcf_result *res)
 {

net/core/dev.c

@@ -4848,7 +4848,8 @@ sch_handle_ingress(struct sk_buff *skb, struct packet_type **pt_prev, int *ret,
 	skb->tc_at_ingress = 1;
 	mini_qdisc_bstats_cpu_update(miniq, skb);
 
-	switch (tcf_classify(skb, miniq->filter_list, &cl_res, false)) {
+	switch (tcf_classify_ingress(skb, miniq->block, miniq->filter_list,
+				     &cl_res, false)) {
 	case TC_ACT_OK:
 	case TC_ACT_RECLASSIFY:
 		skb->tc_index = TC_H_MIN(cl_res.classid);

net/sched/cls_api.c

@@ -22,6 +22,7 @@
 #include <linux/idr.h>
 #include <linux/rhashtable.h>
 #include <linux/jhash.h>
+#include <linux/rculist.h>
 #include <net/net_namespace.h>
 #include <net/sock.h>
 #include <net/netlink.h>
@@ -354,7 +355,7 @@ static struct tcf_chain *tcf_chain_create(struct tcf_block *block,
 	chain = kzalloc(sizeof(*chain), GFP_KERNEL);
 	if (!chain)
 		return NULL;
-	list_add_tail(&chain->list, &block->chain_list);
+	list_add_tail_rcu(&chain->list, &block->chain_list);
 	mutex_init(&chain->filter_chain_lock);
 	chain->block = block;
 	chain->index = chain_index;
@@ -394,7 +395,7 @@ static bool tcf_chain_detach(struct tcf_chain *chain)
 
 	ASSERT_BLOCK_LOCKED(block);
 
-	list_del(&chain->list);
+	list_del_rcu(&chain->list);
 	if (!chain->index)
 		block->chain0.chain = NULL;
 
@@ -453,6 +454,20 @@ static struct tcf_chain *tcf_chain_lookup(struct tcf_block *block,
 	return NULL;
 }
 
+#if IS_ENABLED(CONFIG_NET_TC_SKB_EXT)
+static struct tcf_chain *tcf_chain_lookup_rcu(const struct tcf_block *block,
+					      u32 chain_index)
+{
+	struct tcf_chain *chain;
+
+	list_for_each_entry_rcu(chain, &block->chain_list, list) {
+		if (chain->index == chain_index)
+			return chain;
+	}
+	return NULL;
+}
+#endif
+
 static int tc_chain_notify(struct tcf_chain *chain, struct sk_buff *oskb,
 			   u32 seq, u16 flags, int event, bool unicast);
 
@@ -1559,12 +1574,15 @@ static int tcf_block_setup(struct tcf_block *block,
  * to this qdisc, (optionally) tests for protocol and asks
  * specific classifiers.
  */
-int tcf_classify(struct sk_buff *skb, const struct tcf_proto *tp,
-		 struct tcf_result *res, bool compat_mode)
+static inline int __tcf_classify(struct sk_buff *skb,
+				 const struct tcf_proto *tp,
+				 const struct tcf_proto *orig_tp,
+				 struct tcf_result *res,
+				 bool compat_mode,
+				 u32 *last_executed_chain)
 {
 #ifdef CONFIG_NET_CLS_ACT
 	const int max_reclassify_loop = 4;
-	const struct tcf_proto *orig_tp = tp;
 	const struct tcf_proto *first_tp;
 	int limit = 0;
 
@@ -1582,21 +1600,11 @@ int tcf_classify(struct sk_buff *skb, const struct tcf_proto *tp,
 #ifdef CONFIG_NET_CLS_ACT
 		if (unlikely(err == TC_ACT_RECLASSIFY && !compat_mode)) {
 			first_tp = orig_tp;
+			*last_executed_chain = first_tp->chain->index;
 			goto reset;
 		} else if (unlikely(TC_ACT_EXT_CMP(err, TC_ACT_GOTO_CHAIN))) {
 			first_tp = res->goto_tp;
-
-#if IS_ENABLED(CONFIG_NET_TC_SKB_EXT)
-			{
-				struct tc_skb_ext *ext;
-
-				ext = skb_ext_add(skb, TC_SKB_EXT);
-				if (WARN_ON_ONCE(!ext))
-					return TC_ACT_SHOT;
-
-				ext->chain = err & TC_ACT_EXT_VAL_MASK;
-			}
-#endif
+			*last_executed_chain = err & TC_ACT_EXT_VAL_MASK;
 			goto reset;
 		}
 #endif
@@ -1619,8 +1627,64 @@ int tcf_classify(struct sk_buff *skb, const struct tcf_proto *tp,
 	goto reclassify;
 #endif
 }
+
+int tcf_classify(struct sk_buff *skb, const struct tcf_proto *tp,
+		 struct tcf_result *res, bool compat_mode)
+{
+	u32 last_executed_chain = 0;
+
+	return __tcf_classify(skb, tp, tp, res, compat_mode,
+			      &last_executed_chain);
+}
 EXPORT_SYMBOL(tcf_classify);
 
+int tcf_classify_ingress(struct sk_buff *skb,
+			 const struct tcf_block *ingress_block,
+			 const struct tcf_proto *tp,
+			 struct tcf_result *res, bool compat_mode)
+{
+#if !IS_ENABLED(CONFIG_NET_TC_SKB_EXT)
+	u32 last_executed_chain = 0;
+
+	return __tcf_classify(skb, tp, tp, res, compat_mode,
+			      &last_executed_chain);
+#else
+	u32 last_executed_chain = tp ? tp->chain->index : 0;
+	const struct tcf_proto *orig_tp = tp;
+	struct tc_skb_ext *ext;
+	int ret;
+
+	ext = skb_ext_find(skb, TC_SKB_EXT);
+
+	if (ext && ext->chain) {
+		struct tcf_chain *fchain;
+
+		fchain = tcf_chain_lookup_rcu(ingress_block, ext->chain);
+		if (!fchain)
+			return TC_ACT_SHOT;
+
+		/* Consume, so cloned/redirect skbs won't inherit ext */
+		skb_ext_del(skb, TC_SKB_EXT);
+
+		tp = rcu_dereference_bh(fchain->filter_chain);
+	}
+
+	ret = __tcf_classify(skb, tp, orig_tp, res, compat_mode,
+			     &last_executed_chain);
+
+	/* If we missed on some chain */
+	if (ret == TC_ACT_UNSPEC && last_executed_chain) {
+		ext = skb_ext_add(skb, TC_SKB_EXT);
+		if (WARN_ON_ONCE(!ext))
+			return TC_ACT_SHOT;
+		ext->chain = last_executed_chain;
+	}
+
+	return ret;
+#endif
+}
+EXPORT_SYMBOL(tcf_classify_ingress);
+
 struct tcf_chain_info {
 	struct tcf_proto __rcu **pprev;
 	struct tcf_proto __rcu *next;

net/sched/sch_generic.c

@@ -1391,6 +1391,14 @@ void mini_qdisc_pair_swap(struct mini_Qdisc_pair *miniqp,
 }
 EXPORT_SYMBOL(mini_qdisc_pair_swap);
 
+void mini_qdisc_pair_block_init(struct mini_Qdisc_pair *miniqp,
+				struct tcf_block *block)
+{
+	miniqp->miniq1.block = block;
+	miniqp->miniq2.block = block;
+}
+EXPORT_SYMBOL(mini_qdisc_pair_block_init);
+
 void mini_qdisc_pair_init(struct mini_Qdisc_pair *miniqp, struct Qdisc *qdisc,
 			  struct mini_Qdisc __rcu **p_miniq)
 {

net/sched/sch_ingress.c

@@ -78,6 +78,7 @@ static int ingress_init(struct Qdisc *sch, struct nlattr *opt,
 {
 	struct ingress_sched_data *q = qdisc_priv(sch);
 	struct net_device *dev = qdisc_dev(sch);
+	int err;
 
 	net_inc_ingress_queue();
 
@@ -87,7 +88,13 @@ static int ingress_init(struct Qdisc *sch, struct nlattr *opt,
 	q->block_info.chain_head_change = clsact_chain_head_change;
 	q->block_info.chain_head_change_priv = &q->miniqp;
 
-	return tcf_block_get_ext(&q->block, sch, &q->block_info, extack);
+	err = tcf_block_get_ext(&q->block, sch, &q->block_info, extack);
+	if (err)
+		return err;
+
+	mini_qdisc_pair_block_init(&q->miniqp, q->block);
+
+	return 0;
 }
 
 static void ingress_destroy(struct Qdisc *sch)
@@ -226,6 +233,8 @@ static int clsact_init(struct Qdisc *sch, struct nlattr *opt,
 	if (err)
 		return err;
 
+	mini_qdisc_pair_block_init(&q->miniqp_ingress, q->ingress_block);
+
 	mini_qdisc_pair_init(&q->miniqp_egress, sch, &dev->miniq_egress);
 
 	q->egress_block_info.binder_type = FLOW_BLOCK_BINDER_TYPE_CLSACT_EGRESS;