[pull request][net-next 00/14] Mellanox, mlx5 updates

[pull request][net-next 00/14] Mellanox, mlx5 updates

[Saeed Mahameed]

Hi Dave,

This series from Paul, Or and Mark provides the support for 
e-switch tc offloading of multiple priorities and chains, the series
is based on a merge commit with mlx5-next branch for patches that are
already submitted and reviewed through the netdev and rdma mailing
lists.

For more information please see tag log below.

Please pull and let me know if there is any problem.

Thanks,
saeed.

---

The following changes since commit 186daf0c20507072e72a3c74db4ac50a5b6dae07:

  Merge branch 'mlx5-next' of git://git.kernel.org/pub/scm/linux/kernel/git/mellanox/linux into net-next (2018-10-17 14:13:36 -0700)

are available in the Git repository at:

  git://git.kernel.org/pub/scm/linux/kernel/git/saeed/linux.git tags/mlx5-updates-2018-10-17

for you to fetch changes up to bf07aa730a04a375bc10d09df1e81357af1d4477:

  net/mlx5e: Support offloading tc priorities and chains for eswitch flows (2018-10-17 14:20:49 -0700)

----------------------------------------------------------------
mlx5-updates-2018-10-17

========================================================================

>From Or Gerlitz <ogerlitz@mellanox.com>:

This series from Paul adds support to mlx5 e-switch tc offloading of multiple priorities and chains.

This is made of four building blocks (along with few minor driver refactors):

[1] Split FDB fast path prio to multiple namespaces

Currently the FDB name-space contains two priorities, fast path (p0) and slow path (p1).
The slow path contains the per representor SQ send-to-vport TX rule and the match-all
RX miss rule. As a pre-step to support multi-chains and priorities, we split the FDB fast path
to multiple namespaces  (sub namespaces), each with multiple priorities.

[2] E-Switch chains and priorities

A chain is a group of priorities. We use the fdb parallel sub-namespaces to implement chains,
and a flow table for each priority in them.

Because these namespaces are parallel and in series to the slow path
fdb, the chains aren't connected to each other (but to the slow path),
and one must use a explicit goto action to reach a different chain.

Flow tables for the priorities are created on demand and destroyed
once not used.

[3] Add a no-append flow insertion mode, use it for TC offloads

Enhance the driver fs core, such that if a no-append flag is set by the caller,
we add a new FTE, instead of appending the actions of the inserted rule when
the same match already exists.

For encap rules, we defer the HW offloading till we have a valid neighbor. This can
result in the packet hitting a lower priority rule in the HW DP. Use the no-append API
to push these packets to the slow path FDB table, so they go to the TC kernel DP as done
before priorities where supported.

[4] Offloading tc priorities and chains for eswitch flows

Using [1], [2] and [3] above we add the support for offloading both chains
and priorities. To get to a new chain, use the tc goto action. We support
a fixed prio range 1-16, and chains 0-3.
=============================================================================

----------------------------------------------------------------
Mark Bloch (2):
      net/mlx5: E-Switch, Get counters for offloaded flows from callers
      net/mlx5: Use flow counter IDs and not the wrapping cache object

Or Gerlitz (2):
      net/mlx5: E-Switch, Have explicit API to delete fwd rules
      net/mlx5e: Avoid duplicated code for tc offloads add/del fdb rule

Paul Blakey (8):
      net/mlx5: Add cap bits for multi fdb encap
      net/mlx5: Split FDB fast path prio to multiple namespaces
      net/mlx5: E-Switch, Add chains and priorities
      net/mlx5: Add a no-append flow insertion mode
      net/mlx5e: For TC offloads, always add new flow instead of appending the actions
      net/mlx5: E-Switch, Enable setting goto slow path chain action
      net/mlx5e: Use a slow path rule instead if vxlan neighbour isn't available
      net/mlx5e: Support offloading tc priorities and chains for eswitch flows

Rabie Loulou (1):
      net/mlx5e: Change return type of tc add flow functions

Roi Dayan (1):
      net/mlx5e: Split TC add rule path for nic vs e-switch

 drivers/infiniband/hw/mlx5/main.c                  |  13 +-
 .../mellanox/mlx5/core/diag/fs_tracepoint.h        |   6 +-
 drivers/net/ethernet/mellanox/mlx5/core/en_main.c  |   3 -
 drivers/net/ethernet/mellanox/mlx5/core/en_rep.c   |   6 -
 drivers/net/ethernet/mellanox/mlx5/core/en_tc.c    | 490 +++++++++++++++------
 drivers/net/ethernet/mellanox/mlx5/core/eswitch.c  |   6 +-
 drivers/net/ethernet/mellanox/mlx5/core/eswitch.h  |  46 +-
 .../ethernet/mellanox/mlx5/core/eswitch_offloads.c | 410 +++++++++++++----
 .../net/ethernet/mellanox/mlx5/core/fpga/ipsec.c   |   2 +-
 drivers/net/ethernet/mellanox/mlx5/core/fs_cmd.c   |   2 +-
 drivers/net/ethernet/mellanox/mlx5/core/fs_core.c  | 122 +++--
 drivers/net/ethernet/mellanox/mlx5/core/fs_core.h  |  13 +
 .../net/ethernet/mellanox/mlx5/core/fs_counters.c  |   6 +
 include/linux/mlx5/fs.h                            |  20 +-
 include/linux/mlx5/mlx5_ifc.h                      |   4 +-
 15 files changed, 860 insertions(+), 289 deletions(-)

[net-next 01/14] net/mlx5: E-Switch, Get counters for offloaded flows from callers

[Saeed Mahameed]

From: Mark Bloch <markb@mellanox.com>

There's no real reason for the e-switch logic to manage the creation of
counters for offloaded flows. The API already has the directive for the
caller to denote they want to attach a counter to the created flow.
As such, we go and move the management of flow counters to the mlx5e
tc offload logic. This also lets us remove an inelegant interface where
the FS layer had to provide a way to retrieve a counter from a flow rule.

Signed-off-by: Mark Bloch <markb@mellanox.com>
Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
 .../net/ethernet/mellanox/mlx5/core/en_tc.c   | 32 +++++++++++++++++--
 .../net/ethernet/mellanox/mlx5/core/eswitch.h |  1 +
 .../mellanox/mlx5/core/eswitch_offloads.c     | 20 ++----------
 .../net/ethernet/mellanox/mlx5/core/fs_core.c | 15 ---------
 include/linux/mlx5/fs.h                       |  1 -
 5 files changed, 33 insertions(+), 36 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
index acf7a847f561..8a27c0813a18 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
@@ -61,6 +61,7 @@ struct mlx5_nic_flow_attr {
 	u32 hairpin_tirn;
 	u8 match_level;
 	struct mlx5_flow_table	*hairpin_ft;
+	struct mlx5_fc		*counter;
 };
 
 #define MLX5E_TC_FLOW_BASE (MLX5E_TC_LAST_EXPORTED_BIT + 1)
@@ -721,6 +722,7 @@ mlx5e_tc_add_nic_flow(struct mlx5e_priv *priv,
 		dest[dest_ix].type = MLX5_FLOW_DESTINATION_TYPE_COUNTER;
 		dest[dest_ix].counter = counter;
 		dest_ix++;
+		attr->counter = counter;
 	}
 
 	if (attr->action & MLX5_FLOW_CONTEXT_ACTION_MOD_HDR) {
@@ -797,7 +799,7 @@ static void mlx5e_tc_del_nic_flow(struct mlx5e_priv *priv,
 	struct mlx5_nic_flow_attr *attr = flow->nic_attr;
 	struct mlx5_fc *counter = NULL;
 
-	counter = mlx5_flow_rule_counter(flow->rule[0]);
+	counter = attr->counter;
 	mlx5_del_flow_rules(flow->rule[0]);
 	mlx5_fc_destroy(priv->mdev, counter);
 
@@ -833,6 +835,7 @@ mlx5e_tc_add_fdb_flow(struct mlx5e_priv *priv,
 	struct mlx5_esw_flow_attr *attr = flow->esw_attr;
 	struct net_device *out_dev, *encap_dev = NULL;
 	struct mlx5_flow_handle *rule = NULL;
+	struct mlx5_fc *counter = NULL;
 	struct mlx5e_rep_priv *rpriv;
 	struct mlx5e_priv *out_priv;
 	int err;
@@ -868,6 +871,16 @@ mlx5e_tc_add_fdb_flow(struct mlx5e_priv *priv,
 		}
 	}
 
+	if (attr->action & MLX5_FLOW_CONTEXT_ACTION_COUNT) {
+		counter = mlx5_fc_create(esw->dev, true);
+		if (IS_ERR(counter)) {
+			rule = ERR_CAST(counter);
+			goto err_create_counter;
+		}
+
+		attr->counter = counter;
+	}
+
 	/* we get here if (1) there's no error (rule being null) or when
 	 * (2) there's an encap action and we're on -EAGAIN (no valid neigh)
 	 */
@@ -888,6 +901,8 @@ mlx5e_tc_add_fdb_flow(struct mlx5e_priv *priv,
 	mlx5_eswitch_del_offloaded_rule(esw, rule, attr);
 	rule = flow->rule[1];
 err_add_rule:
+	mlx5_fc_destroy(esw->dev, counter);
+err_create_counter:
 	if (attr->action & MLX5_FLOW_CONTEXT_ACTION_MOD_HDR)
 		mlx5e_detach_mod_hdr(priv, flow);
 err_mod_hdr:
@@ -921,6 +936,9 @@ static void mlx5e_tc_del_fdb_flow(struct mlx5e_priv *priv,
 
 	if (attr->action & MLX5_FLOW_CONTEXT_ACTION_MOD_HDR)
 		mlx5e_detach_mod_hdr(priv, flow);
+
+	if (attr->action & MLX5_FLOW_CONTEXT_ACTION_COUNT)
+		mlx5_fc_destroy(esw->dev, attr->counter);
 }
 
 void mlx5e_tc_encap_flows_add(struct mlx5e_priv *priv,
@@ -992,6 +1010,14 @@ void mlx5e_tc_encap_flows_del(struct mlx5e_priv *priv,
 	}
 }
 
+static struct mlx5_fc *mlx5e_tc_get_counter(struct mlx5e_tc_flow *flow)
+{
+	if (flow->flags & MLX5E_TC_FLOW_ESWITCH)
+		return flow->esw_attr->counter;
+	else
+		return flow->nic_attr->counter;
+}
+
 void mlx5e_tc_update_neigh_used_value(struct mlx5e_neigh_hash_entry *nhe)
 {
 	struct mlx5e_neigh *m_neigh = &nhe->m_neigh;
@@ -1017,7 +1043,7 @@ void mlx5e_tc_update_neigh_used_value(struct mlx5e_neigh_hash_entry *nhe)
 			continue;
 		list_for_each_entry(flow, &e->flows, encap) {
 			if (flow->flags & MLX5E_TC_FLOW_OFFLOADED) {
-				counter = mlx5_flow_rule_counter(flow->rule[0]);
+				counter = mlx5e_tc_get_counter(flow);
 				mlx5_fc_query_cached(counter, &bytes, &packets, &lastuse);
 				if (time_after((unsigned long)lastuse, nhe->reported_lastuse)) {
 					neigh_used = true;
@@ -3019,7 +3045,7 @@ int mlx5e_stats_flower(struct mlx5e_priv *priv,
 	if (!(flow->flags & MLX5E_TC_FLOW_OFFLOADED))
 		return 0;
 
-	counter = mlx5_flow_rule_counter(flow->rule[0]);
+	counter = mlx5e_tc_get_counter(flow);
 	if (!counter)
 		return 0;
 
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h
index dfc642de4e6d..c1b627577003 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h
@@ -266,6 +266,7 @@ struct mlx5_esw_flow_attr {
 	u32	encap_id;
 	u32	mod_hdr_id;
 	u8	match_level;
+	struct mlx5_fc *counter;
 	struct mlx5e_tc_flow_parse_attr *parse_attr;
 };
 
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c b/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
index 0741683f7d70..a2f2d726c99b 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
@@ -51,7 +51,6 @@ mlx5_eswitch_add_offloaded_rule(struct mlx5_eswitch *esw,
 	struct mlx5_flow_destination dest[MLX5_MAX_FLOW_FWD_VPORTS + 1] = {};
 	struct mlx5_flow_act flow_act = {0};
 	struct mlx5_flow_table *ft = NULL;
-	struct mlx5_fc *counter = NULL;
 	struct mlx5_flow_handle *rule;
 	int j, i = 0;
 	void *misc;
@@ -91,13 +90,8 @@ mlx5_eswitch_add_offloaded_rule(struct mlx5_eswitch *esw,
 		}
 	}
 	if (flow_act.action & MLX5_FLOW_CONTEXT_ACTION_COUNT) {
-		counter = mlx5_fc_create(esw->dev, true);
-		if (IS_ERR(counter)) {
-			rule = ERR_CAST(counter);
-			goto err_counter_alloc;
-		}
 		dest[i].type = MLX5_FLOW_DESTINATION_TYPE_COUNTER;
-		dest[i].counter = counter;
+		dest[i].counter = attr->counter;
 		i++;
 	}
 
@@ -132,15 +126,11 @@ mlx5_eswitch_add_offloaded_rule(struct mlx5_eswitch *esw,
 
 	rule = mlx5_add_flow_rules(ft, spec, &flow_act, dest, i);
 	if (IS_ERR(rule))
-		goto err_add_rule;
+		goto out;
 	else
 		esw->offloads.num_flows++;
 
-	return rule;
-
-err_add_rule:
-	mlx5_fc_destroy(esw->dev, counter);
-err_counter_alloc:
+out:
 	return rule;
 }
 
@@ -200,11 +190,7 @@ mlx5_eswitch_del_offloaded_rule(struct mlx5_eswitch *esw,
 				struct mlx5_flow_handle *rule,
 				struct mlx5_esw_flow_attr *attr)
 {
-	struct mlx5_fc *counter = NULL;
-
-	counter = mlx5_flow_rule_counter(rule);
 	mlx5_del_flow_rules(rule);
-	mlx5_fc_destroy(esw->dev, counter);
 	esw->offloads.num_flows--;
 }
 
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c b/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c
index 8d340e5181f8..9e18e6c0a8b3 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c
@@ -1474,21 +1474,6 @@ static struct mlx5_flow_handle *add_rule_fg(struct mlx5_flow_group *fg,
 	return handle;
 }
 
-struct mlx5_fc *mlx5_flow_rule_counter(struct mlx5_flow_handle *handle)
-{
-	struct mlx5_flow_rule *dst;
-	struct fs_fte *fte;
-
-	fs_get_obj(fte, handle->rule[0]->node.parent);
-
-	fs_for_each_dst(dst, fte) {
-		if (dst->dest_attr.type == MLX5_FLOW_DESTINATION_TYPE_COUNTER)
-			return dst->dest_attr.counter;
-	}
-
-	return NULL;
-}
-
 static bool counter_is_valid(struct mlx5_fc *counter, u32 action)
 {
 	if (!(action & MLX5_FLOW_CONTEXT_ACTION_COUNT))
diff --git a/include/linux/mlx5/fs.h b/include/linux/mlx5/fs.h
index b1c026f1c8ba..74d0ea146c9a 100644
--- a/include/linux/mlx5/fs.h
+++ b/include/linux/mlx5/fs.h
@@ -186,7 +186,6 @@ int mlx5_modify_rule_destination(struct mlx5_flow_handle *handler,
 				 struct mlx5_flow_destination *new_dest,
 				 struct mlx5_flow_destination *old_dest);
 
-struct mlx5_fc *mlx5_flow_rule_counter(struct mlx5_flow_handle *handler);
 struct mlx5_fc *mlx5_fc_create(struct mlx5_core_dev *dev, bool aging);
 void mlx5_fc_destroy(struct mlx5_core_dev *dev, struct mlx5_fc *counter);
 void mlx5_fc_query_cached(struct mlx5_fc *counter,
-- 
2.17.2

[net-next 02/14] net/mlx5: Use flow counter IDs and not the wrapping cache object

[Saeed Mahameed]

From: Mark Bloch <markb@mellanox.com>

Currently, when a flow rule is created using the FS core layer, the caller
has to pass the entire flow counter object and not just the counter HW
handle (ID). This requires both the FS core and the caller to have
knowledge about the inner implementation of the FS layer flow counters
cache and limits the possible users.

Move to use the counter ID across the place when dealing with flows.

Doing this decoupling, now can we privatize the inner implementation
of the flow counters.

Signed-off-by: Mark Bloch <markb@mellanox.com>
Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
 drivers/infiniband/hw/mlx5/main.c                      |  7 +++++--
 .../ethernet/mellanox/mlx5/core/diag/fs_tracepoint.h   |  6 +++---
 drivers/net/ethernet/mellanox/mlx5/core/en_tc.c        |  2 +-
 drivers/net/ethernet/mellanox/mlx5/core/eswitch.c      |  4 ++--
 .../net/ethernet/mellanox/mlx5/core/eswitch_offloads.c |  2 +-
 drivers/net/ethernet/mellanox/mlx5/core/fs_cmd.c       |  2 +-
 drivers/net/ethernet/mellanox/mlx5/core/fs_core.c      | 10 ++--------
 drivers/net/ethernet/mellanox/mlx5/core/fs_counters.c  |  6 ++++++
 include/linux/mlx5/fs.h                                |  3 ++-
 9 files changed, 23 insertions(+), 19 deletions(-)

diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index 5d9b7f62a0ba..5ced0cc46ba1 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -3320,15 +3320,18 @@ static struct mlx5_ib_flow_handler *_create_flow_rule(struct mlx5_ib_dev *dev,
 	}
 
 	if (flow_act.action & MLX5_FLOW_CONTEXT_ACTION_COUNT) {
+		struct mlx5_ib_mcounters *mcounters;
+
 		err = flow_counters_set_data(flow_act.counters, ucmd);
 		if (err)
 			goto free;
 
+		mcounters = to_mcounters(flow_act.counters);
 		handler->ibcounters = flow_act.counters;
 		dest_arr[dest_num].type =
 			MLX5_FLOW_DESTINATION_TYPE_COUNTER;
-		dest_arr[dest_num].counter =
-			to_mcounters(flow_act.counters)->hw_cntrs_hndl;
+		dest_arr[dest_num].counter_id =
+			mlx5_fc_id(mcounters->hw_cntrs_hndl);
 		dest_num++;
 	}
 
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/diag/fs_tracepoint.h b/drivers/net/ethernet/mellanox/mlx5/core/diag/fs_tracepoint.h
index e83dda441a81..d027ce00c8ce 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/diag/fs_tracepoint.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/diag/fs_tracepoint.h
@@ -252,10 +252,10 @@ TRACE_EVENT(mlx5_fs_add_rule,
 			   memcpy(__entry->destination,
 				  &rule->dest_attr,
 				  sizeof(__entry->destination));
-			   if (rule->dest_attr.type & MLX5_FLOW_DESTINATION_TYPE_COUNTER &&
-			       rule->dest_attr.counter)
+			   if (rule->dest_attr.type &
+			       MLX5_FLOW_DESTINATION_TYPE_COUNTER)
 				__entry->counter_id =
-				rule->dest_attr.counter->id;
+					rule->dest_attr.counter_id;
 	    ),
 	    TP_printk("rule=%p fte=%p index=%u sw_action=<%s> [dst] %s\n",
 		      __entry->rule, __entry->fte, __entry->index,
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
index 8a27c0813a18..5ce87f54852d 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
@@ -720,7 +720,7 @@ mlx5e_tc_add_nic_flow(struct mlx5e_priv *priv,
 			goto err_fc_create;
 		}
 		dest[dest_ix].type = MLX5_FLOW_DESTINATION_TYPE_COUNTER;
-		dest[dest_ix].counter = counter;
+		dest[dest_ix].counter_id = mlx5_fc_id(counter);
 		dest_ix++;
 		attr->counter = counter;
 	}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.c b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.c
index e1d47fa5ab83..9c893d7d273e 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.c
@@ -1198,7 +1198,7 @@ static int esw_vport_ingress_config(struct mlx5_eswitch *esw,
 	if (counter) {
 		flow_act.action |= MLX5_FLOW_CONTEXT_ACTION_COUNT;
 		drop_ctr_dst.type = MLX5_FLOW_DESTINATION_TYPE_COUNTER;
-		drop_ctr_dst.counter = counter;
+		drop_ctr_dst.counter_id = mlx5_fc_id(counter);
 		dst = &drop_ctr_dst;
 		dest_num++;
 	}
@@ -1285,7 +1285,7 @@ static int esw_vport_egress_config(struct mlx5_eswitch *esw,
 	if (counter) {
 		flow_act.action |= MLX5_FLOW_CONTEXT_ACTION_COUNT;
 		drop_ctr_dst.type = MLX5_FLOW_DESTINATION_TYPE_COUNTER;
-		drop_ctr_dst.counter = counter;
+		drop_ctr_dst.counter_id = mlx5_fc_id(counter);
 		dst = &drop_ctr_dst;
 		dest_num++;
 	}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c b/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
index a2f2d726c99b..39932dce15cb 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
@@ -91,7 +91,7 @@ mlx5_eswitch_add_offloaded_rule(struct mlx5_eswitch *esw,
 	}
 	if (flow_act.action & MLX5_FLOW_CONTEXT_ACTION_COUNT) {
 		dest[i].type = MLX5_FLOW_DESTINATION_TYPE_COUNTER;
-		dest[i].counter = attr->counter;
+		dest[i].counter_id = mlx5_fc_id(attr->counter);
 		i++;
 	}
 
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fs_cmd.c b/drivers/net/ethernet/mellanox/mlx5/core/fs_cmd.c
index dc8d7f6b52c2..08a891f9aade 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/fs_cmd.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fs_cmd.c
@@ -419,7 +419,7 @@ static int mlx5_cmd_set_fte(struct mlx5_core_dev *dev,
 				continue;
 
 			MLX5_SET(flow_counter_list, in_dests, flow_counter_id,
-				 dst->dest_attr.counter->id);
+				 dst->dest_attr.counter_id);
 			in_dests += MLX5_ST_SZ_BYTES(dest_format_struct);
 			list_size++;
 		}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c b/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c
index 9e18e6c0a8b3..cdcbf9d0ae6c 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c
@@ -1474,14 +1474,8 @@ static struct mlx5_flow_handle *add_rule_fg(struct mlx5_flow_group *fg,
 	return handle;
 }
 
-static bool counter_is_valid(struct mlx5_fc *counter, u32 action)
+static bool counter_is_valid(u32 action)
 {
-	if (!(action & MLX5_FLOW_CONTEXT_ACTION_COUNT))
-		return !counter;
-
-	if (!counter)
-		return false;
-
 	return (action & (MLX5_FLOW_CONTEXT_ACTION_DROP |
 			  MLX5_FLOW_CONTEXT_ACTION_FWD_DEST));
 }
@@ -1491,7 +1485,7 @@ static bool dest_is_valid(struct mlx5_flow_destination *dest,
 			  struct mlx5_flow_table *ft)
 {
 	if (dest && (dest->type == MLX5_FLOW_DESTINATION_TYPE_COUNTER))
-		return counter_is_valid(dest->counter, action);
+		return counter_is_valid(action);
 
 	if (!(action & MLX5_FLOW_CONTEXT_ACTION_FWD_DEST))
 		return true;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fs_counters.c b/drivers/net/ethernet/mellanox/mlx5/core/fs_counters.c
index 09206c4acd9a..1329bc5b7969 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/fs_counters.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fs_counters.c
@@ -258,6 +258,12 @@ struct mlx5_fc *mlx5_fc_create(struct mlx5_core_dev *dev, bool aging)
 }
 EXPORT_SYMBOL(mlx5_fc_create);
 
+u32 mlx5_fc_id(struct mlx5_fc *counter)
+{
+	return counter->id;
+}
+EXPORT_SYMBOL(mlx5_fc_id);
+
 void mlx5_fc_destroy(struct mlx5_core_dev *dev, struct mlx5_fc *counter)
 {
 	struct mlx5_fc_stats *fc_stats = &dev->priv.fc_stats;
diff --git a/include/linux/mlx5/fs.h b/include/linux/mlx5/fs.h
index 74d0ea146c9a..a5fc62184195 100644
--- a/include/linux/mlx5/fs.h
+++ b/include/linux/mlx5/fs.h
@@ -92,7 +92,7 @@ struct mlx5_flow_destination {
 		u32			tir_num;
 		u32			ft_num;
 		struct mlx5_flow_table	*ft;
-		struct mlx5_fc		*counter;
+		u32			counter_id;
 		struct {
 			u16		num;
 			u16		vhca_id;
@@ -192,6 +192,7 @@ void mlx5_fc_query_cached(struct mlx5_fc *counter,
 			  u64 *bytes, u64 *packets, u64 *lastuse);
 int mlx5_fc_query(struct mlx5_core_dev *dev, struct mlx5_fc *counter,
 		  u64 *packets, u64 *bytes);
+u32 mlx5_fc_id(struct mlx5_fc *counter);
 
 int mlx5_fs_add_rx_underlay_qpn(struct mlx5_core_dev *dev, u32 underlay_qpn);
 int mlx5_fs_remove_rx_underlay_qpn(struct mlx5_core_dev *dev, u32 underlay_qpn);
-- 
2.17.2

[net-next 03/14] net/mlx5e: Change return type of tc add flow functions

[Saeed Mahameed]

From: Rabie Loulou <rabiel@mellanox.com>

Refactor the flow add utility functions to return err code instead of rule
pointers. This will allow for simpler logic when one tc rule is
duplicated to two HW rules in downstream patches.

Signed-off-by: Rabie Loulou <rabiel@mellanox.com>
Signed-off-by: Shahar Klein <shahark@mellanox.com>
Reviewed-by: Roi Dayan <roid@mellanox.com>
Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
 .../net/ethernet/mellanox/mlx5/core/en_tc.c   | 86 +++++++++----------
 1 file changed, 39 insertions(+), 47 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
index 5ce87f54852d..861986f82844 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
@@ -673,7 +673,7 @@ static void mlx5e_hairpin_flow_del(struct mlx5e_priv *priv,
 	}
 }
 
-static struct mlx5_flow_handle *
+static int
 mlx5e_tc_add_nic_flow(struct mlx5e_priv *priv,
 		      struct mlx5e_tc_flow_parse_attr *parse_attr,
 		      struct mlx5e_tc_flow *flow,
@@ -689,14 +689,12 @@ mlx5e_tc_add_nic_flow(struct mlx5e_priv *priv,
 		.reformat_id = 0,
 	};
 	struct mlx5_fc *counter = NULL;
-	struct mlx5_flow_handle *rule;
 	bool table_created = false;
 	int err, dest_ix = 0;
 
 	if (flow->flags & MLX5E_TC_FLOW_HAIRPIN) {
 		err = mlx5e_hairpin_flow_add(priv, flow, parse_attr, extack);
 		if (err) {
-			rule = ERR_PTR(err);
 			goto err_add_hairpin_flow;
 		}
 		if (flow->flags & MLX5E_TC_FLOW_HAIRPIN_RSS) {
@@ -716,7 +714,7 @@ mlx5e_tc_add_nic_flow(struct mlx5e_priv *priv,
 	if (attr->action & MLX5_FLOW_CONTEXT_ACTION_COUNT) {
 		counter = mlx5_fc_create(dev, true);
 		if (IS_ERR(counter)) {
-			rule = ERR_CAST(counter);
+			err = PTR_ERR(counter);
 			goto err_fc_create;
 		}
 		dest[dest_ix].type = MLX5_FLOW_DESTINATION_TYPE_COUNTER;
@@ -729,10 +727,8 @@ mlx5e_tc_add_nic_flow(struct mlx5e_priv *priv,
 		err = mlx5e_attach_mod_hdr(priv, flow, parse_attr);
 		flow_act.modify_id = attr->mod_hdr_id;
 		kfree(parse_attr->mod_hdr_actions);
-		if (err) {
-			rule = ERR_PTR(err);
+		if (err)
 			goto err_create_mod_hdr_id;
-		}
 	}
 
 	if (IS_ERR_OR_NULL(priv->fs.tc.t)) {
@@ -758,7 +754,7 @@ mlx5e_tc_add_nic_flow(struct mlx5e_priv *priv,
 					   "Failed to create tc offload table\n");
 			netdev_err(priv->netdev,
 				   "Failed to create tc offload table\n");
-			rule = ERR_CAST(priv->fs.tc.t);
+			err = PTR_ERR(priv->fs.tc.t);
 			goto err_create_ft;
 		}
 
@@ -768,13 +764,15 @@ mlx5e_tc_add_nic_flow(struct mlx5e_priv *priv,
 	if (attr->match_level != MLX5_MATCH_NONE)
 		parse_attr->spec.match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
 
-	rule = mlx5_add_flow_rules(priv->fs.tc.t, &parse_attr->spec,
-				   &flow_act, dest, dest_ix);
+	flow->rule[0] = mlx5_add_flow_rules(priv->fs.tc.t, &parse_attr->spec,
+					    &flow_act, dest, dest_ix);
 
-	if (IS_ERR(rule))
+	if (IS_ERR(flow->rule[0])) {
+		err = PTR_ERR(flow->rule[0]);
 		goto err_add_rule;
+	}
 
-	return rule;
+	return 0;
 
 err_add_rule:
 	if (table_created) {
@@ -790,7 +788,7 @@ mlx5e_tc_add_nic_flow(struct mlx5e_priv *priv,
 	if (flow->flags & MLX5E_TC_FLOW_HAIRPIN)
 		mlx5e_hairpin_flow_del(priv, flow);
 err_add_hairpin_flow:
-	return rule;
+	return err;
 }
 
 static void mlx5e_tc_del_nic_flow(struct mlx5e_priv *priv,
@@ -825,7 +823,7 @@ static int mlx5e_attach_encap(struct mlx5e_priv *priv,
 			      struct mlx5e_tc_flow *flow,
 			      struct netlink_ext_ack *extack);
 
-static struct mlx5_flow_handle *
+static int
 mlx5e_tc_add_fdb_flow(struct mlx5e_priv *priv,
 		      struct mlx5e_tc_flow_parse_attr *parse_attr,
 		      struct mlx5e_tc_flow *flow,
@@ -834,21 +832,20 @@ mlx5e_tc_add_fdb_flow(struct mlx5e_priv *priv,
 	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
 	struct mlx5_esw_flow_attr *attr = flow->esw_attr;
 	struct net_device *out_dev, *encap_dev = NULL;
-	struct mlx5_flow_handle *rule = NULL;
 	struct mlx5_fc *counter = NULL;
 	struct mlx5e_rep_priv *rpriv;
 	struct mlx5e_priv *out_priv;
-	int err;
+	int err = 0, encap_err = 0;
 
 	if (attr->action & MLX5_FLOW_CONTEXT_ACTION_PACKET_REFORMAT) {
 		out_dev = __dev_get_by_index(dev_net(priv->netdev),
 					     attr->parse_attr->mirred_ifindex);
-		err = mlx5e_attach_encap(priv, &parse_attr->tun_info,
-					 out_dev, &encap_dev, flow, extack);
-		if (err) {
-			rule = ERR_PTR(err);
-			if (err != -EAGAIN)
-				goto err_attach_encap;
+		encap_err = mlx5e_attach_encap(priv, &parse_attr->tun_info,
+					       out_dev, &encap_dev, flow,
+					       extack);
+		if (encap_err && encap_err != -EAGAIN) {
+			err = encap_err;
+			goto err_attach_encap;
 		}
 		out_priv = netdev_priv(encap_dev);
 		rpriv = out_priv->ppriv;
@@ -857,49 +854,49 @@ mlx5e_tc_add_fdb_flow(struct mlx5e_priv *priv,
 	}
 
 	err = mlx5_eswitch_add_vlan_action(esw, attr);
-	if (err) {
-		rule = ERR_PTR(err);
+	if (err)
 		goto err_add_vlan;
-	}
 
 	if (attr->action & MLX5_FLOW_CONTEXT_ACTION_MOD_HDR) {
 		err = mlx5e_attach_mod_hdr(priv, flow, parse_attr);
 		kfree(parse_attr->mod_hdr_actions);
-		if (err) {
-			rule = ERR_PTR(err);
+		if (err)
 			goto err_mod_hdr;
-		}
 	}
 
 	if (attr->action & MLX5_FLOW_CONTEXT_ACTION_COUNT) {
 		counter = mlx5_fc_create(esw->dev, true);
 		if (IS_ERR(counter)) {
-			rule = ERR_CAST(counter);
+			err = PTR_ERR(counter);
 			goto err_create_counter;
 		}
 
 		attr->counter = counter;
 	}
 
-	/* we get here if (1) there's no error (rule being null) or when
+	/* we get here if (1) there's no error or when
 	 * (2) there's an encap action and we're on -EAGAIN (no valid neigh)
 	 */
-	if (rule != ERR_PTR(-EAGAIN)) {
-		rule = mlx5_eswitch_add_offloaded_rule(esw, &parse_attr->spec, attr);
-		if (IS_ERR(rule))
+	if (encap_err != -EAGAIN) {
+		flow->rule[0] = mlx5_eswitch_add_offloaded_rule(esw, &parse_attr->spec, attr);
+		if (IS_ERR(flow->rule[0])) {
+			err = PTR_ERR(flow->rule[0]);
 			goto err_add_rule;
+		}
 
 		if (attr->mirror_count) {
 			flow->rule[1] = mlx5_eswitch_add_fwd_rule(esw, &parse_attr->spec, attr);
-			if (IS_ERR(flow->rule[1]))
+			if (IS_ERR(flow->rule[1])) {
+				err = PTR_ERR(flow->rule[1]);
 				goto err_fwd_rule;
+			}
 		}
 	}
-	return rule;
+
+	return encap_err;
 
 err_fwd_rule:
-	mlx5_eswitch_del_offloaded_rule(esw, rule, attr);
-	rule = flow->rule[1];
+	mlx5_eswitch_del_offloaded_rule(esw, flow->rule[0], attr);
 err_add_rule:
 	mlx5_fc_destroy(esw->dev, counter);
 err_create_counter:
@@ -911,7 +908,7 @@ mlx5e_tc_add_fdb_flow(struct mlx5e_priv *priv,
 	if (attr->action & MLX5_FLOW_CONTEXT_ACTION_PACKET_REFORMAT)
 		mlx5e_detach_encap(priv, flow);
 err_attach_encap:
-	return rule;
+	return err;
 }
 
 static void mlx5e_tc_del_fdb_flow(struct mlx5e_priv *priv,
@@ -2959,22 +2956,17 @@ int mlx5e_configure_flower(struct mlx5e_priv *priv,
 					   extack);
 		if (err < 0)
 			goto err_free;
-		flow->rule[0] = mlx5e_tc_add_fdb_flow(priv, parse_attr, flow,
-						      extack);
+		err = mlx5e_tc_add_fdb_flow(priv, parse_attr, flow, extack);
 	} else {
 		err = parse_tc_nic_actions(priv, f->exts, parse_attr, flow,
 					   extack);
 		if (err < 0)
 			goto err_free;
-		flow->rule[0] = mlx5e_tc_add_nic_flow(priv, parse_attr, flow,
-						      extack);
+		err = mlx5e_tc_add_nic_flow(priv, parse_attr, flow, extack);
 	}
 
-	if (IS_ERR(flow->rule[0])) {
-		err = PTR_ERR(flow->rule[0]);
-		if (err != -EAGAIN)
-			goto err_free;
-	}
+	if (err && err != -EAGAIN)
+		goto err_free;
 
 	if (err != -EAGAIN)
 		flow->flags |= MLX5E_TC_FLOW_OFFLOADED;
-- 
2.17.2

[net-next 04/14] net/mlx5e: Split TC add rule path for nic vs e-switch

[Saeed Mahameed]

From: Roi Dayan <roid@mellanox.com>

Move to have clear separation on the code path to add nic vs e-switch
flows. While here we break the code that deals with adding offloaded
TC tool to few smaller stages, each on helper function.

Besides getting us simpler and readable code, these are pre-steps
for being able to have two HW flows serving one SW TC flow for some
e-switch use cases.

Signed-off-by: Roi Dayan <roid@mellanox.com>
Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
 .../net/ethernet/mellanox/mlx5/core/en_tc.c   | 185 +++++++++++++-----
 1 file changed, 138 insertions(+), 47 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
index 861986f82844..a4a432f02930 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
@@ -2907,34 +2907,15 @@ static struct rhashtable *get_tc_ht(struct mlx5e_priv *priv)
 		return &priv->fs.tc.ht;
 }
 
-int mlx5e_configure_flower(struct mlx5e_priv *priv,
-			   struct tc_cls_flower_offload *f, int flags)
+static int
+mlx5e_alloc_flow(struct mlx5e_priv *priv, int attr_size,
+		 struct tc_cls_flower_offload *f, u8 flow_flags,
+		 struct mlx5e_tc_flow_parse_attr **__parse_attr,
+		 struct mlx5e_tc_flow **__flow)
 {
-	struct netlink_ext_ack *extack = f->common.extack;
-	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
 	struct mlx5e_tc_flow_parse_attr *parse_attr;
-	struct rhashtable *tc_ht = get_tc_ht(priv);
 	struct mlx5e_tc_flow *flow;
-	int attr_size, err = 0;
-	u8 flow_flags = 0;
-
-	get_flags(flags, &flow_flags);
-
-	flow = rhashtable_lookup_fast(tc_ht, &f->cookie, tc_ht_params);
-	if (flow) {
-		NL_SET_ERR_MSG_MOD(extack,
-				   "flow cookie already exists, ignoring");
-		netdev_warn_once(priv->netdev, "flow cookie %lx already exists, ignoring\n", f->cookie);
-		return 0;
-	}
-
-	if (esw && esw->mode == SRIOV_OFFLOADS) {
-		flow_flags |= MLX5E_TC_FLOW_ESWITCH;
-		attr_size  = sizeof(struct mlx5_esw_flow_attr);
-	} else {
-		flow_flags |= MLX5E_TC_FLOW_NIC;
-		attr_size  = sizeof(struct mlx5_nic_flow_attr);
-	}
+	int err;
 
 	flow = kzalloc(sizeof(*flow) + attr_size, GFP_KERNEL);
 	parse_attr = kvzalloc(sizeof(*parse_attr), GFP_KERNEL);
@@ -2948,45 +2929,155 @@ int mlx5e_configure_flower(struct mlx5e_priv *priv,
 	flow->priv = priv;
 
 	err = parse_cls_flower(priv, flow, &parse_attr->spec, f);
-	if (err < 0)
+	if (err)
 		goto err_free;
 
-	if (flow->flags & MLX5E_TC_FLOW_ESWITCH) {
-		err = parse_tc_fdb_actions(priv, f->exts, parse_attr, flow,
-					   extack);
-		if (err < 0)
-			goto err_free;
-		err = mlx5e_tc_add_fdb_flow(priv, parse_attr, flow, extack);
-	} else {
-		err = parse_tc_nic_actions(priv, f->exts, parse_attr, flow,
-					   extack);
-		if (err < 0)
-			goto err_free;
-		err = mlx5e_tc_add_nic_flow(priv, parse_attr, flow, extack);
-	}
+	*__flow = flow;
+	*__parse_attr = parse_attr;
+
+	return 0;
 
+err_free:
+	kfree(flow);
+	kvfree(parse_attr);
+	return err;
+}
+
+static int
+mlx5e_add_fdb_flow(struct mlx5e_priv *priv,
+		   struct tc_cls_flower_offload *f,
+		   u8 flow_flags,
+		   struct mlx5e_tc_flow **__flow)
+{
+	struct netlink_ext_ack *extack = f->common.extack;
+	struct mlx5e_tc_flow_parse_attr *parse_attr;
+	struct mlx5e_tc_flow *flow;
+	int attr_size, err;
+
+	flow_flags |= MLX5E_TC_FLOW_ESWITCH;
+	attr_size  = sizeof(struct mlx5_esw_flow_attr);
+	err = mlx5e_alloc_flow(priv, attr_size, f, flow_flags,
+			       &parse_attr, &flow);
+	if (err)
+		goto out;
+
+	err = parse_tc_fdb_actions(priv, f->exts, parse_attr, flow, extack);
+	if (err)
+		goto err_free;
+
+	err = mlx5e_tc_add_fdb_flow(priv, parse_attr, flow, extack);
 	if (err && err != -EAGAIN)
 		goto err_free;
 
-	if (err != -EAGAIN)
+	if (!err)
 		flow->flags |= MLX5E_TC_FLOW_OFFLOADED;
 
-	if (!(flow->flags & MLX5E_TC_FLOW_ESWITCH) ||
-	    !(flow->esw_attr->action &
+	if (!(flow->esw_attr->action &
 	      MLX5_FLOW_CONTEXT_ACTION_PACKET_REFORMAT))
 		kvfree(parse_attr);
 
-	err = rhashtable_insert_fast(tc_ht, &flow->node, tc_ht_params);
-	if (err) {
-		mlx5e_tc_del_flow(priv, flow);
-		kfree(flow);
-	}
+	*__flow = flow;
 
+	return 0;
+
+err_free:
+	kfree(flow);
+	kvfree(parse_attr);
+out:
 	return err;
+}
+
+static int
+mlx5e_add_nic_flow(struct mlx5e_priv *priv,
+		   struct tc_cls_flower_offload *f,
+		   u8 flow_flags,
+		   struct mlx5e_tc_flow **__flow)
+{
+	struct netlink_ext_ack *extack = f->common.extack;
+	struct mlx5e_tc_flow_parse_attr *parse_attr;
+	struct mlx5e_tc_flow *flow;
+	int attr_size, err;
+
+	flow_flags |= MLX5E_TC_FLOW_NIC;
+	attr_size  = sizeof(struct mlx5_nic_flow_attr);
+	err = mlx5e_alloc_flow(priv, attr_size, f, flow_flags,
+			       &parse_attr, &flow);
+	if (err)
+		goto out;
+
+	err = parse_tc_nic_actions(priv, f->exts, parse_attr, flow, extack);
+	if (err)
+		goto err_free;
+
+	err = mlx5e_tc_add_nic_flow(priv, parse_attr, flow, extack);
+	if (err)
+		goto err_free;
+
+	flow->flags |= MLX5E_TC_FLOW_OFFLOADED;
+	kvfree(parse_attr);
+	*__flow = flow;
+
+	return 0;
 
 err_free:
+	kfree(flow);
 	kvfree(parse_attr);
+out:
+	return err;
+}
+
+static int
+mlx5e_tc_add_flow(struct mlx5e_priv *priv,
+		  struct tc_cls_flower_offload *f,
+		  int flags,
+		  struct mlx5e_tc_flow **flow)
+{
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	u8 flow_flags;
+	int err;
+
+	get_flags(flags, &flow_flags);
+
+	if (esw && esw->mode == SRIOV_OFFLOADS)
+		err = mlx5e_add_fdb_flow(priv, f, flow_flags, flow);
+	else
+		err = mlx5e_add_nic_flow(priv, f, flow_flags, flow);
+
+	return err;
+}
+
+int mlx5e_configure_flower(struct mlx5e_priv *priv,
+			   struct tc_cls_flower_offload *f, int flags)
+{
+	struct netlink_ext_ack *extack = f->common.extack;
+	struct rhashtable *tc_ht = get_tc_ht(priv);
+	struct mlx5e_tc_flow *flow;
+	int err = 0;
+
+	flow = rhashtable_lookup_fast(tc_ht, &f->cookie, tc_ht_params);
+	if (flow) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "flow cookie already exists, ignoring");
+		netdev_warn_once(priv->netdev,
+				 "flow cookie %lx already exists, ignoring\n",
+				 f->cookie);
+		goto out;
+	}
+
+	err = mlx5e_tc_add_flow(priv, f, flags, &flow);
+	if (err)
+		goto out;
+
+	err = rhashtable_insert_fast(tc_ht, &flow->node, tc_ht_params);
+	if (err)
+		goto err_free;
+
+	return 0;
+
+err_free:
+	mlx5e_tc_del_flow(priv, flow);
 	kfree(flow);
+out:
 	return err;
 }
 
-- 
2.17.2

[net-next 05/14] net/mlx5: Add cap bits for multi fdb encap

[Saeed Mahameed]

From: Paul Blakey <paulb@mellanox.com>

If set, the firmware supports creating of flow tables with encap
enabled while VFs are configured, if we already created one
(restriction still applies on the first creation).

Signed-off-by: Paul Blakey <paulb@mellanox.com>
Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
 include/linux/mlx5/mlx5_ifc.h | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/include/linux/mlx5/mlx5_ifc.h b/include/linux/mlx5/mlx5_ifc.h
index 15e36198f85f..963611820006 100644
--- a/include/linux/mlx5/mlx5_ifc.h
+++ b/include/linux/mlx5/mlx5_ifc.h
@@ -584,7 +584,9 @@ struct mlx5_ifc_flow_table_nic_cap_bits {
 struct mlx5_ifc_flow_table_eswitch_cap_bits {
 	u8      reserved_at_0[0x1c];
 	u8      fdb_multi_path_to_table[0x1];
-	u8      reserved_at_1d[0x1e3];
+	u8      reserved_at_1d[0x1];
+	u8      multi_fdb_encap[0x1];
+	u8      reserved_at_1e[0x1e1];
 
 	struct mlx5_ifc_flow_table_prop_layout_bits flow_table_properties_nic_esw_fdb;
 
-- 
2.17.2

[net-next 06/14] net/mlx5: Split FDB fast path prio to multiple namespaces

[Saeed Mahameed]

From: Paul Blakey <paulb@mellanox.com>

Towards supporting multi-chains and priorities, split the FDB fast path
to multiple namespaces (sub namespaces), each with multiple priorities.

This patch adds a new flow steering type, FS_TYPE_PRIO_CHAINS, which is
like current FS_TYPE_PRIO, but may contain only namespaces, and those
will be in parallel to one another in terms of managing of the flow
tables connections inside them. Meaning, while searching for the next
or previous flow table to connect for a new table inside such namespace
we skip the parallel namespaces in the same level under the
FS_TYPE_PRIO_CHAINS prio we originated from.

We use this new type for splitting the fast path prio into multiple
parallel namespaces, each containing normal prios.
The prios inside them (and their tables) will be connected to one
another, but not from one parallel namespace to another, instead the
last prio in each namespace will be connected to the next prio in
the containing FDB namespace, which is the slow path prio.

Signed-off-by: Paul Blakey <paulb@mellanox.com>
Acked-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
 .../net/ethernet/mellanox/mlx5/core/eswitch.c |  2 +-
 .../net/ethernet/mellanox/mlx5/core/eswitch.h |  7 ++
 .../net/ethernet/mellanox/mlx5/core/fs_core.c | 88 ++++++++++++++++---
 .../net/ethernet/mellanox/mlx5/core/fs_core.h | 13 +++
 include/linux/mlx5/fs.h                       |  2 +
 5 files changed, 101 insertions(+), 11 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.c b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.c
index 9c893d7d273e..d004957328f9 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.c
@@ -263,7 +263,7 @@ static int esw_create_legacy_fdb_table(struct mlx5_eswitch *esw)
 	esw_debug(dev, "Create FDB log_max_size(%d)\n",
 		  MLX5_CAP_ESW_FLOWTABLE_FDB(dev, log_max_ft_size));
 
-	root_ns = mlx5_get_flow_namespace(dev, MLX5_FLOW_NAMESPACE_FDB);
+	root_ns = mlx5_get_fdb_sub_ns(dev, 0);
 	if (!root_ns) {
 		esw_warn(dev, "Failed to get FDB flow namespace\n");
 		return -EOPNOTSUPP;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h
index c1b627577003..1698a322a7c4 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h
@@ -59,6 +59,9 @@
 #define mlx5_esw_has_fwd_fdb(dev) \
 	MLX5_CAP_ESW_FLOWTABLE(dev, fdb_multi_path_to_table)
 
+#define FDB_MAX_CHAIN 3
+#define FDB_MAX_PRIO 16
+
 struct vport_ingress {
 	struct mlx5_flow_table *acl;
 	struct mlx5_flow_group *allow_untagged_spoofchk_grp;
@@ -319,6 +322,10 @@ static inline void mlx5_eswitch_cleanup(struct mlx5_eswitch *esw) {}
 static inline void mlx5_eswitch_vport_event(struct mlx5_eswitch *esw, struct mlx5_eqe *eqe) {}
 static inline int  mlx5_eswitch_enable_sriov(struct mlx5_eswitch *esw, int nvfs, int mode) { return 0; }
 static inline void mlx5_eswitch_disable_sriov(struct mlx5_eswitch *esw) {}
+
+#define FDB_MAX_CHAIN 1
+#define FDB_MAX_PRIO 1
+
 #endif /* CONFIG_MLX5_ESWITCH */
 
 #endif /* __MLX5_ESWITCH_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c b/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c
index cdcbf9d0ae6c..7eb6d58733ac 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c
@@ -40,6 +40,7 @@
 #include "diag/fs_tracepoint.h"
 #include "accel/ipsec.h"
 #include "fpga/ipsec.h"
+#include "eswitch.h"
 
 #define INIT_TREE_NODE_ARRAY_SIZE(...)	(sizeof((struct init_tree_node[]){__VA_ARGS__}) /\
 					 sizeof(struct init_tree_node))
@@ -713,7 +714,7 @@ static struct mlx5_flow_table *find_closest_ft_recursive(struct fs_node  *root,
 	struct fs_node *iter = list_entry(start, struct fs_node, list);
 	struct mlx5_flow_table *ft = NULL;
 
-	if (!root)
+	if (!root || root->type == FS_TYPE_PRIO_CHAINS)
 		return NULL;
 
 	list_for_each_advance_continue(iter, &root->children, reverse) {
@@ -1973,6 +1974,18 @@ void mlx5_destroy_flow_group(struct mlx5_flow_group *fg)
 			       fg->id);
 }
 
+struct mlx5_flow_namespace *mlx5_get_fdb_sub_ns(struct mlx5_core_dev *dev,
+						int n)
+{
+	struct mlx5_flow_steering *steering = dev->priv.steering;
+
+	if (!steering || !steering->fdb_sub_ns)
+		return NULL;
+
+	return steering->fdb_sub_ns[n];
+}
+EXPORT_SYMBOL(mlx5_get_fdb_sub_ns);
+
 struct mlx5_flow_namespace *mlx5_get_flow_namespace(struct mlx5_core_dev *dev,
 						    enum mlx5_flow_namespace_type type)
 {
@@ -2051,8 +2064,10 @@ struct mlx5_flow_namespace *mlx5_get_flow_vport_acl_namespace(struct mlx5_core_d
 	}
 }
 
-static struct fs_prio *fs_create_prio(struct mlx5_flow_namespace *ns,
-				      unsigned int prio, int num_levels)
+static struct fs_prio *_fs_create_prio(struct mlx5_flow_namespace *ns,
+				       unsigned int prio,
+				       int num_levels,
+				       enum fs_node_type type)
 {
 	struct fs_prio *fs_prio;
 
@@ -2060,7 +2075,7 @@ static struct fs_prio *fs_create_prio(struct mlx5_flow_namespace *ns,
 	if (!fs_prio)
 		return ERR_PTR(-ENOMEM);
 
-	fs_prio->node.type = FS_TYPE_PRIO;
+	fs_prio->node.type = type;
 	tree_init_node(&fs_prio->node, NULL, del_sw_prio);
 	tree_add_node(&fs_prio->node, &ns->node);
 	fs_prio->num_levels = num_levels;
@@ -2070,6 +2085,19 @@ static struct fs_prio *fs_create_prio(struct mlx5_flow_namespace *ns,
 	return fs_prio;
 }
 
+static struct fs_prio *fs_create_prio_chained(struct mlx5_flow_namespace *ns,
+					      unsigned int prio,
+					      int num_levels)
+{
+	return _fs_create_prio(ns, prio, num_levels, FS_TYPE_PRIO_CHAINS);
+}
+
+static struct fs_prio *fs_create_prio(struct mlx5_flow_namespace *ns,
+				      unsigned int prio, int num_levels)
+{
+	return _fs_create_prio(ns, prio, num_levels, FS_TYPE_PRIO);
+}
+
 static struct mlx5_flow_namespace *fs_init_namespace(struct mlx5_flow_namespace
 						     *ns)
 {
@@ -2374,6 +2402,9 @@ void mlx5_cleanup_fs(struct mlx5_core_dev *dev)
 	cleanup_egress_acls_root_ns(dev);
 	cleanup_ingress_acls_root_ns(dev);
 	cleanup_root_ns(steering->fdb_root_ns);
+	steering->fdb_root_ns = NULL;
+	kfree(steering->fdb_sub_ns);
+	steering->fdb_sub_ns = NULL;
 	cleanup_root_ns(steering->sniffer_rx_root_ns);
 	cleanup_root_ns(steering->sniffer_tx_root_ns);
 	cleanup_root_ns(steering->egress_root_ns);
@@ -2419,27 +2450,64 @@ static int init_sniffer_rx_root_ns(struct mlx5_flow_steering *steering)
 
 static int init_fdb_root_ns(struct mlx5_flow_steering *steering)
 {
-	struct fs_prio *prio;
+	struct mlx5_flow_namespace *ns;
+	struct fs_prio *maj_prio;
+	struct fs_prio *min_prio;
+	int levels;
+	int chain;
+	int prio;
+	int err;
 
 	steering->fdb_root_ns = create_root_ns(steering, FS_FT_FDB);
 	if (!steering->fdb_root_ns)
 		return -ENOMEM;
 
-	prio = fs_create_prio(&steering->fdb_root_ns->ns, 0, 2);
-	if (IS_ERR(prio))
+	steering->fdb_sub_ns = kzalloc(sizeof(steering->fdb_sub_ns) *
+				       FDB_MAX_CHAIN + 1, GFP_KERNEL);
+	if (!steering->fdb_sub_ns)
+		return -ENOMEM;
+
+	levels = 2 * FDB_MAX_PRIO * (FDB_MAX_CHAIN + 1);
+	maj_prio = fs_create_prio_chained(&steering->fdb_root_ns->ns, 0,
+					  levels);
+	if (IS_ERR(maj_prio)) {
+		err = PTR_ERR(maj_prio);
 		goto out_err;
+	}
 
-	prio = fs_create_prio(&steering->fdb_root_ns->ns, 1, 1);
-	if (IS_ERR(prio))
+	for (chain = 0; chain <= FDB_MAX_CHAIN; chain++) {
+		ns = fs_create_namespace(maj_prio);
+		if (IS_ERR(ns)) {
+			err = PTR_ERR(ns);
+			goto out_err;
+		}
+
+		for (prio = 0; prio < FDB_MAX_PRIO * (chain + 1); prio++) {
+			min_prio = fs_create_prio(ns, prio, 2);
+			if (IS_ERR(min_prio)) {
+				err = PTR_ERR(min_prio);
+				goto out_err;
+			}
+		}
+
+		steering->fdb_sub_ns[chain] = ns;
+	}
+
+	maj_prio = fs_create_prio(&steering->fdb_root_ns->ns, 1, 1);
+	if (IS_ERR(maj_prio)) {
+		err = PTR_ERR(maj_prio);
 		goto out_err;
+	}
 
 	set_prio_attrs(steering->fdb_root_ns);
 	return 0;
 
 out_err:
 	cleanup_root_ns(steering->fdb_root_ns);
+	kfree(steering->fdb_sub_ns);
+	steering->fdb_sub_ns = NULL;
 	steering->fdb_root_ns = NULL;
-	return PTR_ERR(prio);
+	return err;
 }
 
 static int init_egress_acl_root_ns(struct mlx5_flow_steering *steering, int vport)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fs_core.h b/drivers/net/ethernet/mellanox/mlx5/core/fs_core.h
index a06f83c0c2b6..b51ad217da32 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/fs_core.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fs_core.h
@@ -38,9 +38,21 @@
 #include <linux/rhashtable.h>
 #include <linux/llist.h>
 
+/* FS_TYPE_PRIO_CHAINS is a PRIO that will have namespaces only,
+ * and those are in parallel to one another when going over them to connect
+ * a new flow table. Meaning the last flow table in a TYPE_PRIO prio in one
+ * parallel namespace will not automatically connect to the first flow table
+ * found in any prio in any next namespace, but skip the entire containing
+ * TYPE_PRIO_CHAINS prio.
+ *
+ * This is used to implement tc chains, each chain of prios is a different
+ * namespace inside a containing TYPE_PRIO_CHAINS prio.
+ */
+
 enum fs_node_type {
 	FS_TYPE_NAMESPACE,
 	FS_TYPE_PRIO,
+	FS_TYPE_PRIO_CHAINS,
 	FS_TYPE_FLOW_TABLE,
 	FS_TYPE_FLOW_GROUP,
 	FS_TYPE_FLOW_ENTRY,
@@ -73,6 +85,7 @@ struct mlx5_flow_steering {
 	struct kmem_cache               *ftes_cache;
 	struct mlx5_flow_root_namespace *root_ns;
 	struct mlx5_flow_root_namespace *fdb_root_ns;
+	struct mlx5_flow_namespace	**fdb_sub_ns;
 	struct mlx5_flow_root_namespace **esw_egress_root_ns;
 	struct mlx5_flow_root_namespace **esw_ingress_root_ns;
 	struct mlx5_flow_root_namespace	*sniffer_tx_root_ns;
diff --git a/include/linux/mlx5/fs.h b/include/linux/mlx5/fs.h
index a5fc62184195..f8d00872c7d3 100644
--- a/include/linux/mlx5/fs.h
+++ b/include/linux/mlx5/fs.h
@@ -101,6 +101,8 @@ struct mlx5_flow_destination {
 	};
 };
 
+struct mlx5_flow_namespace *
+mlx5_get_fdb_sub_ns(struct mlx5_core_dev *dev, int n);
 struct mlx5_flow_namespace *
 mlx5_get_flow_namespace(struct mlx5_core_dev *dev,
 			enum mlx5_flow_namespace_type type);
-- 
2.17.2

[net-next 07/14] net/mlx5: E-Switch, Have explicit API to delete fwd rules

[Saeed Mahameed]

From: Or Gerlitz <ogerlitz@mellanox.com>

Be symmetric with the e-switch API to add rules which has a
specific function to add fwd rules which are used as part of
vport mirroring.

This patch doesn't change any functionality.

Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Reviewed-by: Paul Blakey <paulb@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
 drivers/net/ethernet/mellanox/mlx5/core/en_tc.c           | 4 ++--
 drivers/net/ethernet/mellanox/mlx5/core/eswitch.h         | 4 ++++
 .../net/ethernet/mellanox/mlx5/core/eswitch_offloads.c    | 8 ++++++++
 3 files changed, 14 insertions(+), 2 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
index a4a432f02930..7487bdd55f23 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
@@ -920,7 +920,7 @@ static void mlx5e_tc_del_fdb_flow(struct mlx5e_priv *priv,
 	if (flow->flags & MLX5E_TC_FLOW_OFFLOADED) {
 		flow->flags &= ~MLX5E_TC_FLOW_OFFLOADED;
 		if (attr->mirror_count)
-			mlx5_eswitch_del_offloaded_rule(esw, flow->rule[1], attr);
+			mlx5_eswitch_del_fwd_rule(esw, flow->rule[1], attr);
 		mlx5_eswitch_del_offloaded_rule(esw, flow->rule[0], attr);
 	}
 
@@ -996,7 +996,7 @@ void mlx5e_tc_encap_flows_del(struct mlx5e_priv *priv,
 
 			flow->flags &= ~MLX5E_TC_FLOW_OFFLOADED;
 			if (attr->mirror_count)
-				mlx5_eswitch_del_offloaded_rule(esw, flow->rule[1], attr);
+				mlx5_eswitch_del_fwd_rule(esw, flow->rule[1], attr);
 			mlx5_eswitch_del_offloaded_rule(esw, flow->rule[0], attr);
 		}
 	}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h
index 1698a322a7c4..584e735bbad1 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h
@@ -231,6 +231,10 @@ void
 mlx5_eswitch_del_offloaded_rule(struct mlx5_eswitch *esw,
 				struct mlx5_flow_handle *rule,
 				struct mlx5_esw_flow_attr *attr);
+void
+mlx5_eswitch_del_fwd_rule(struct mlx5_eswitch *esw,
+			  struct mlx5_flow_handle *rule,
+			  struct mlx5_esw_flow_attr *attr);
 
 struct mlx5_flow_handle *
 mlx5_eswitch_create_vport_rx_rule(struct mlx5_eswitch *esw, int vport,
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c b/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
index 39932dce15cb..983bb8a80f75 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
@@ -194,6 +194,14 @@ mlx5_eswitch_del_offloaded_rule(struct mlx5_eswitch *esw,
 	esw->offloads.num_flows--;
 }
 
+void
+mlx5_eswitch_del_fwd_rule(struct mlx5_eswitch *esw,
+			  struct mlx5_flow_handle *rule,
+			  struct mlx5_esw_flow_attr *attr)
+{
+	mlx5_eswitch_del_offloaded_rule(esw, rule, attr);
+}
+
 static int esw_set_global_vlan_pop(struct mlx5_eswitch *esw, u8 val)
 {
 	struct mlx5_eswitch_rep *rep;
-- 
2.17.2

[net-next 08/14] net/mlx5: E-Switch, Add chains and priorities

[Saeed Mahameed]

From: Paul Blakey <paulb@mellanox.com>

A chain is a group of priorities, so use the fdb parallel
sub namespaces to implement chains, and a flow table for each
priority in them.

Because these namespaces are parallel and in series to the slow path
fdb, the chains aren't connected to one another (but to the slow path),
and one must use a explicit goto action to reach a different chain.

Flow tables for the priorities will be created on demand and destroyed
once not used.

The Firmware has four pools of tables for sizes S/XS/M/L (4k, 64k, 1m, 4m).
We maintain ghost copies of the pools occupancy.

When a new table is to be created, we scan the pools from large to small
and find the 1st table size which can be now created. When a table is
destroyed, we update the relevant pool.

Multi chain/prio isn't enabled yet by this patch, for now all flows
will use the default chain 0, and prio 1.

Signed-off-by: Paul Blakey <paulb@mellanox.com>
Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
 .../net/ethernet/mellanox/mlx5/core/en_tc.c   |   3 +
 .../net/ethernet/mellanox/mlx5/core/eswitch.h |  32 +-
 .../mellanox/mlx5/core/eswitch_offloads.c     | 390 ++++++++++++++----
 3 files changed, 339 insertions(+), 86 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
index 7487bdd55f23..6c04e11f9a05 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
@@ -837,6 +837,9 @@ mlx5e_tc_add_fdb_flow(struct mlx5e_priv *priv,
 	struct mlx5e_priv *out_priv;
 	int err = 0, encap_err = 0;
 
+	/* keep the old behaviour, use same prio for all offloaded rules */
+	attr->prio = 1;
+
 	if (attr->action & MLX5_FLOW_CONTEXT_ACTION_PACKET_REFORMAT) {
 		out_dev = __dev_get_by_index(dev_net(priv->netdev),
 					     attr->parse_attr->mirred_ifindex);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h
index 584e735bbad1..54215f4312fa 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h
@@ -123,6 +123,13 @@ struct mlx5_vport {
 	u16                     enabled_events;
 };
 
+enum offloads_fdb_flags {
+	ESW_FDB_CHAINS_AND_PRIOS_SUPPORTED = BIT(0),
+};
+
+extern const unsigned int ESW_POOLS[4];
+
+#define PRIO_LEVELS 2
 struct mlx5_eswitch_fdb {
 	union {
 		struct legacy_fdb {
@@ -133,16 +140,24 @@ struct mlx5_eswitch_fdb {
 		} legacy;
 
 		struct offloads_fdb {
-			struct mlx5_flow_table *fast_fdb;
-			struct mlx5_flow_table *fwd_fdb;
 			struct mlx5_flow_table *slow_fdb;
 			struct mlx5_flow_group *send_to_vport_grp;
 			struct mlx5_flow_group *miss_grp;
 			struct mlx5_flow_handle *miss_rule_uni;
 			struct mlx5_flow_handle *miss_rule_multi;
 			int vlan_push_pop_refcount;
+
+			struct {
+				struct mlx5_flow_table *fdb;
+				u32 num_rules;
+			} fdb_prio[FDB_MAX_CHAIN + 1][FDB_MAX_PRIO + 1][PRIO_LEVELS];
+			/* Protects fdb_prio table */
+			struct mutex fdb_prio_lock;
+
+			int fdb_left[ARRAY_SIZE(ESW_POOLS)];
 		} offloads;
 	};
+	u32 flags;
 };
 
 struct mlx5_esw_offload {
@@ -184,6 +199,7 @@ struct mlx5_eswitch {
 
 	struct mlx5_esw_offload offloads;
 	int                     mode;
+	int                     nvports;
 };
 
 void esw_offloads_cleanup(struct mlx5_eswitch *esw, int nvports);
@@ -236,6 +252,15 @@ mlx5_eswitch_del_fwd_rule(struct mlx5_eswitch *esw,
 			  struct mlx5_flow_handle *rule,
 			  struct mlx5_esw_flow_attr *attr);
 
+bool
+mlx5_eswitch_prios_supported(struct mlx5_eswitch *esw);
+
+u16
+mlx5_eswitch_get_prio_range(struct mlx5_eswitch *esw);
+
+u32
+mlx5_eswitch_get_chain_range(struct mlx5_eswitch *esw);
+
 struct mlx5_flow_handle *
 mlx5_eswitch_create_vport_rx_rule(struct mlx5_eswitch *esw, int vport,
 				  struct mlx5_flow_destination *dest);
@@ -274,6 +299,9 @@ struct mlx5_esw_flow_attr {
 	u32	mod_hdr_id;
 	u8	match_level;
 	struct mlx5_fc *counter;
+	u32	chain;
+	u16	prio;
+	u32	dest_chain;
 	struct mlx5e_tc_flow_parse_attr *parse_attr;
 };
 
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c b/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
index 983bb8a80f75..8501b6c31c02 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
@@ -37,32 +37,59 @@
 #include <linux/mlx5/fs.h>
 #include "mlx5_core.h"
 #include "eswitch.h"
+#include "en.h"
+#include "fs_core.h"
 
 enum {
 	FDB_FAST_PATH = 0,
 	FDB_SLOW_PATH
 };
 
+#define fdb_prio_table(esw, chain, prio, level) \
+	(esw)->fdb_table.offloads.fdb_prio[(chain)][(prio)][(level)]
+
+static struct mlx5_flow_table *
+esw_get_prio_table(struct mlx5_eswitch *esw, u32 chain, u16 prio, int level);
+static void
+esw_put_prio_table(struct mlx5_eswitch *esw, u32 chain, u16 prio, int level);
+
+bool mlx5_eswitch_prios_supported(struct mlx5_eswitch *esw)
+{
+	return (!!(esw->fdb_table.flags & ESW_FDB_CHAINS_AND_PRIOS_SUPPORTED));
+}
+
+u32 mlx5_eswitch_get_chain_range(struct mlx5_eswitch *esw)
+{
+	if (esw->fdb_table.flags & ESW_FDB_CHAINS_AND_PRIOS_SUPPORTED)
+		return FDB_MAX_CHAIN;
+
+	return 0;
+}
+
+u16 mlx5_eswitch_get_prio_range(struct mlx5_eswitch *esw)
+{
+	if (esw->fdb_table.flags & ESW_FDB_CHAINS_AND_PRIOS_SUPPORTED)
+		return FDB_MAX_PRIO;
+
+	return U16_MAX;
+}
+
 struct mlx5_flow_handle *
 mlx5_eswitch_add_offloaded_rule(struct mlx5_eswitch *esw,
 				struct mlx5_flow_spec *spec,
 				struct mlx5_esw_flow_attr *attr)
 {
 	struct mlx5_flow_destination dest[MLX5_MAX_FLOW_FWD_VPORTS + 1] = {};
+	bool mirror = !!(attr->mirror_count);
 	struct mlx5_flow_act flow_act = {0};
-	struct mlx5_flow_table *ft = NULL;
 	struct mlx5_flow_handle *rule;
+	struct mlx5_flow_table *fdb;
 	int j, i = 0;
 	void *misc;
 
 	if (esw->mode != SRIOV_OFFLOADS)
 		return ERR_PTR(-EOPNOTSUPP);
 
-	if (attr->mirror_count)
-		ft = esw->fdb_table.offloads.fwd_fdb;
-	else
-		ft = esw->fdb_table.offloads.fast_fdb;
-
 	flow_act.action = attr->action;
 	/* if per flow vlan pop/push is emulated, don't set that into the firmware */
 	if (!mlx5_eswitch_vlan_actions_supported(esw->dev, 1))
@@ -80,13 +107,28 @@ mlx5_eswitch_add_offloaded_rule(struct mlx5_eswitch *esw,
 	}
 
 	if (flow_act.action & MLX5_FLOW_CONTEXT_ACTION_FWD_DEST) {
-		for (j = attr->mirror_count; j < attr->out_count; j++) {
-			dest[i].type = MLX5_FLOW_DESTINATION_TYPE_VPORT;
-			dest[i].vport.num = attr->out_rep[j]->vport;
-			dest[i].vport.vhca_id =
-				MLX5_CAP_GEN(attr->out_mdev[j], vhca_id);
-			dest[i].vport.vhca_id_valid = !!MLX5_CAP_ESW(esw->dev, merged_eswitch);
+		if (attr->dest_chain) {
+			struct mlx5_flow_table *ft;
+
+			ft = esw_get_prio_table(esw, attr->dest_chain, 1, 0);
+			if (IS_ERR(ft)) {
+				rule = ERR_CAST(ft);
+				goto err_create_goto_table;
+			}
+
+			dest[i].type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
+			dest[i].ft = ft;
 			i++;
+		} else {
+			for (j = attr->mirror_count; j < attr->out_count; j++) {
+				dest[i].type = MLX5_FLOW_DESTINATION_TYPE_VPORT;
+				dest[i].vport.num = attr->out_rep[j]->vport;
+				dest[i].vport.vhca_id =
+					MLX5_CAP_GEN(attr->out_mdev[j], vhca_id);
+				dest[i].vport.vhca_id_valid =
+					!!MLX5_CAP_ESW(esw->dev, merged_eswitch);
+				i++;
+			}
 		}
 	}
 	if (flow_act.action & MLX5_FLOW_CONTEXT_ACTION_COUNT) {
@@ -124,13 +166,26 @@ mlx5_eswitch_add_offloaded_rule(struct mlx5_eswitch *esw,
 	if (flow_act.action & MLX5_FLOW_CONTEXT_ACTION_PACKET_REFORMAT)
 		flow_act.reformat_id = attr->encap_id;
 
-	rule = mlx5_add_flow_rules(ft, spec, &flow_act, dest, i);
+	fdb = esw_get_prio_table(esw, attr->chain, attr->prio, !!mirror);
+	if (IS_ERR(fdb)) {
+		rule = ERR_CAST(fdb);
+		goto err_esw_get;
+	}
+
+	rule = mlx5_add_flow_rules(fdb, spec, &flow_act, dest, i);
 	if (IS_ERR(rule))
-		goto out;
+		goto err_add_rule;
 	else
 		esw->offloads.num_flows++;
 
-out:
+	return rule;
+
+err_add_rule:
+	esw_put_prio_table(esw, attr->chain, attr->prio, !!mirror);
+err_esw_get:
+	if (attr->dest_chain)
+		esw_put_prio_table(esw, attr->dest_chain, 1, 0);
+err_create_goto_table:
 	return rule;
 }
 
@@ -141,10 +196,24 @@ mlx5_eswitch_add_fwd_rule(struct mlx5_eswitch *esw,
 {
 	struct mlx5_flow_destination dest[MLX5_MAX_FLOW_FWD_VPORTS + 1] = {};
 	struct mlx5_flow_act flow_act = {0};
+	struct mlx5_flow_table *fast_fdb;
+	struct mlx5_flow_table *fwd_fdb;
 	struct mlx5_flow_handle *rule;
 	void *misc;
 	int i;
 
+	fast_fdb = esw_get_prio_table(esw, attr->chain, attr->prio, 0);
+	if (IS_ERR(fast_fdb)) {
+		rule = ERR_CAST(fast_fdb);
+		goto err_get_fast;
+	}
+
+	fwd_fdb = esw_get_prio_table(esw, attr->chain, attr->prio, 1);
+	if (IS_ERR(fwd_fdb)) {
+		rule = ERR_CAST(fwd_fdb);
+		goto err_get_fwd;
+	}
+
 	flow_act.action = MLX5_FLOW_CONTEXT_ACTION_FWD_DEST;
 	for (i = 0; i < attr->mirror_count; i++) {
 		dest[i].type = MLX5_FLOW_DESTINATION_TYPE_VPORT;
@@ -154,7 +223,7 @@ mlx5_eswitch_add_fwd_rule(struct mlx5_eswitch *esw,
 		dest[i].vport.vhca_id_valid = !!MLX5_CAP_ESW(esw->dev, merged_eswitch);
 	}
 	dest[i].type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
-	dest[i].ft = esw->fdb_table.offloads.fwd_fdb,
+	dest[i].ft = fwd_fdb,
 	i++;
 
 	misc = MLX5_ADDR_OF(fte_match_param, spec->match_value, misc_parameters);
@@ -177,21 +246,49 @@ mlx5_eswitch_add_fwd_rule(struct mlx5_eswitch *esw,
 		spec->match_criteria_enable = MLX5_MATCH_OUTER_HEADERS |
 					      MLX5_MATCH_MISC_PARAMETERS;
 
-	rule = mlx5_add_flow_rules(esw->fdb_table.offloads.fast_fdb, spec, &flow_act, dest, i);
+	rule = mlx5_add_flow_rules(fast_fdb, spec, &flow_act, dest, i);
 
-	if (!IS_ERR(rule))
-		esw->offloads.num_flows++;
+	if (IS_ERR(rule))
+		goto add_err;
 
+	esw->offloads.num_flows++;
+
+	return rule;
+add_err:
+	esw_put_prio_table(esw, attr->chain, attr->prio, 1);
+err_get_fwd:
+	esw_put_prio_table(esw, attr->chain, attr->prio, 0);
+err_get_fast:
 	return rule;
 }
 
+static void
+__mlx5_eswitch_del_rule(struct mlx5_eswitch *esw,
+			struct mlx5_flow_handle *rule,
+			struct mlx5_esw_flow_attr *attr,
+			bool fwd_rule)
+{
+	bool mirror = (attr->mirror_count > 0);
+
+	mlx5_del_flow_rules(rule);
+	esw->offloads.num_flows--;
+
+	if (fwd_rule)  {
+		esw_put_prio_table(esw, attr->chain, attr->prio, 1);
+		esw_put_prio_table(esw, attr->chain, attr->prio, 0);
+	} else {
+		esw_put_prio_table(esw, attr->chain, attr->prio, !!mirror);
+		if (attr->dest_chain)
+			esw_put_prio_table(esw, attr->dest_chain, 1, 0);
+	}
+}
+
 void
 mlx5_eswitch_del_offloaded_rule(struct mlx5_eswitch *esw,
 				struct mlx5_flow_handle *rule,
 				struct mlx5_esw_flow_attr *attr)
 {
-	mlx5_del_flow_rules(rule);
-	esw->offloads.num_flows--;
+	__mlx5_eswitch_del_rule(esw, rule, attr, false);
 }
 
 void
@@ -199,7 +296,7 @@ mlx5_eswitch_del_fwd_rule(struct mlx5_eswitch *esw,
 			  struct mlx5_flow_handle *rule,
 			  struct mlx5_esw_flow_attr *attr)
 {
-	mlx5_eswitch_del_offloaded_rule(esw, rule, attr);
+	__mlx5_eswitch_del_rule(esw, rule, attr, true);
 }
 
 static int esw_set_global_vlan_pop(struct mlx5_eswitch *esw, u8 val)
@@ -288,7 +385,8 @@ int mlx5_eswitch_add_vlan_action(struct mlx5_eswitch *esw,
 
 	push = !!(attr->action & MLX5_FLOW_CONTEXT_ACTION_VLAN_PUSH);
 	pop  = !!(attr->action & MLX5_FLOW_CONTEXT_ACTION_VLAN_POP);
-	fwd  = !!(attr->action & MLX5_FLOW_CONTEXT_ACTION_FWD_DEST);
+	fwd  = !!((attr->action & MLX5_FLOW_CONTEXT_ACTION_FWD_DEST) &&
+		   !attr->dest_chain);
 
 	err = esw_add_vlan_action_check(attr, push, pop, fwd);
 	if (err)
@@ -495,75 +593,164 @@ static int esw_add_fdb_miss_rule(struct mlx5_eswitch *esw)
 
 #define ESW_OFFLOADS_NUM_GROUPS  4
 
-static int esw_create_offloads_fast_fdb_table(struct mlx5_eswitch *esw)
+/* Firmware currently has 4 pool of 4 sizes that it supports (ESW_POOLS),
+ * and a virtual memory region of 16M (ESW_SIZE), this region is duplicated
+ * for each flow table pool. We can allocate up to 16M of each pool,
+ * and we keep track of how much we used via put/get_sz_to_pool.
+ * Firmware doesn't report any of this for now.
+ * ESW_POOL is expected to be sorted from large to small
+ */
+#define ESW_SIZE (16 * 1024 * 1024)
+const unsigned int ESW_POOLS[4] = { 4 * 1024 * 1024, 1 * 1024 * 1024,
+				    64 * 1024, 4 * 1024 };
+
+static int
+get_sz_from_pool(struct mlx5_eswitch *esw)
+{
+	int sz = 0, i;
+
+	for (i = 0; i < ARRAY_SIZE(ESW_POOLS); i++) {
+		if (esw->fdb_table.offloads.fdb_left[i]) {
+			--esw->fdb_table.offloads.fdb_left[i];
+			sz = ESW_POOLS[i];
+			break;
+		}
+	}
+
+	return sz;
+}
+
+static void
+put_sz_to_pool(struct mlx5_eswitch *esw, int sz)
+{
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(ESW_POOLS); i++) {
+		if (sz >= ESW_POOLS[i]) {
+			++esw->fdb_table.offloads.fdb_left[i];
+			break;
+		}
+	}
+}
+
+static struct mlx5_flow_table *
+create_next_size_table(struct mlx5_eswitch *esw,
+		       struct mlx5_flow_namespace *ns,
+		       u16 table_prio,
+		       int level,
+		       u32 flags)
+{
+	struct mlx5_flow_table *fdb;
+	int sz;
+
+	sz = get_sz_from_pool(esw);
+	if (!sz)
+		return ERR_PTR(-ENOSPC);
+
+	fdb = mlx5_create_auto_grouped_flow_table(ns,
+						  table_prio,
+						  sz,
+						  ESW_OFFLOADS_NUM_GROUPS,
+						  level,
+						  flags);
+	if (IS_ERR(fdb)) {
+		esw_warn(esw->dev, "Failed to create FDB Table err %d (table prio: %d, level: %d, size: %d)\n",
+			 (int)PTR_ERR(fdb), table_prio, level, sz);
+		put_sz_to_pool(esw, sz);
+	}
+
+	return fdb;
+}
+
+static struct mlx5_flow_table *
+esw_get_prio_table(struct mlx5_eswitch *esw, u32 chain, u16 prio, int level)
 {
 	struct mlx5_core_dev *dev = esw->dev;
-	struct mlx5_flow_namespace *root_ns;
 	struct mlx5_flow_table *fdb = NULL;
-	int esw_size, err = 0;
+	struct mlx5_flow_namespace *ns;
+	int table_prio, l = 0;
 	u32 flags = 0;
-	u32 max_flow_counter = (MLX5_CAP_GEN(dev, max_flow_counter_31_16) << 16) |
-				MLX5_CAP_GEN(dev, max_flow_counter_15_0);
 
-	root_ns = mlx5_get_flow_namespace(dev, MLX5_FLOW_NAMESPACE_FDB);
-	if (!root_ns) {
-		esw_warn(dev, "Failed to get FDB flow namespace\n");
-		err = -EOPNOTSUPP;
-		goto out_namespace;
-	}
-
-	esw_debug(dev, "Create offloads FDB table, min (max esw size(2^%d), max counters(%d)*groups(%d))\n",
-		  MLX5_CAP_ESW_FLOWTABLE_FDB(dev, log_max_ft_size),
-		  max_flow_counter, ESW_OFFLOADS_NUM_GROUPS);
+	mutex_lock(&esw->fdb_table.offloads.fdb_prio_lock);
 
-	esw_size = min_t(int, max_flow_counter * ESW_OFFLOADS_NUM_GROUPS,
-			 1 << MLX5_CAP_ESW_FLOWTABLE_FDB(dev, log_max_ft_size));
+	fdb = fdb_prio_table(esw, chain, prio, level).fdb;
+	if (fdb) {
+		/* take ref on earlier levels as well */
+		while (level >= 0)
+			fdb_prio_table(esw, chain, prio, level--).num_rules++;
+		mutex_unlock(&esw->fdb_table.offloads.fdb_prio_lock);
+		return fdb;
+	}
 
-	if (mlx5_esw_has_fwd_fdb(dev))
-		esw_size >>= 1;
+	ns = mlx5_get_fdb_sub_ns(dev, chain);
+	if (!ns) {
+		esw_warn(dev, "Failed to get FDB sub namespace\n");
+		mutex_unlock(&esw->fdb_table.offloads.fdb_prio_lock);
+		return ERR_PTR(-EOPNOTSUPP);
+	}
 
 	if (esw->offloads.encap != DEVLINK_ESWITCH_ENCAP_MODE_NONE)
 		flags |= (MLX5_FLOW_TABLE_TUNNEL_EN_REFORMAT |
 			  MLX5_FLOW_TABLE_TUNNEL_EN_DECAP);
 
-	fdb = mlx5_create_auto_grouped_flow_table(root_ns, FDB_FAST_PATH,
-						  esw_size,
-						  ESW_OFFLOADS_NUM_GROUPS, 0,
-						  flags);
-	if (IS_ERR(fdb)) {
-		err = PTR_ERR(fdb);
-		esw_warn(dev, "Failed to create Fast path FDB Table err %d\n", err);
-		goto out_namespace;
-	}
-	esw->fdb_table.offloads.fast_fdb = fdb;
+	table_prio = (chain * FDB_MAX_PRIO) + prio - 1;
 
-	if (!mlx5_esw_has_fwd_fdb(dev))
-		goto out_namespace;
+	/* create earlier levels for correct fs_core lookup when
+	 * connecting tables
+	 */
+	for (l = 0; l <= level; l++) {
+		if (fdb_prio_table(esw, chain, prio, l).fdb) {
+			fdb_prio_table(esw, chain, prio, l).num_rules++;
+			continue;
+		}
 
-	fdb = mlx5_create_auto_grouped_flow_table(root_ns, FDB_FAST_PATH,
-						  esw_size,
-						  ESW_OFFLOADS_NUM_GROUPS, 1,
-						  flags);
-	if (IS_ERR(fdb)) {
-		err = PTR_ERR(fdb);
-		esw_warn(dev, "Failed to create fwd table err %d\n", err);
-		goto out_ft;
+		fdb = create_next_size_table(esw, ns, table_prio, l, flags);
+		if (IS_ERR(fdb)) {
+			l--;
+			goto err_create_fdb;
+		}
+
+		fdb_prio_table(esw, chain, prio, l).fdb = fdb;
+		fdb_prio_table(esw, chain, prio, l).num_rules = 1;
 	}
-	esw->fdb_table.offloads.fwd_fdb = fdb;
 
-	return err;
+	mutex_unlock(&esw->fdb_table.offloads.fdb_prio_lock);
+	return fdb;
 
-out_ft:
-	mlx5_destroy_flow_table(esw->fdb_table.offloads.fast_fdb);
-out_namespace:
-	return err;
+err_create_fdb:
+	mutex_unlock(&esw->fdb_table.offloads.fdb_prio_lock);
+	if (l >= 0)
+		esw_put_prio_table(esw, chain, prio, l);
+
+	return fdb;
 }
 
-static void esw_destroy_offloads_fast_fdb_table(struct mlx5_eswitch *esw)
+static void
+esw_put_prio_table(struct mlx5_eswitch *esw, u32 chain, u16 prio, int level)
 {
-	if (mlx5_esw_has_fwd_fdb(esw->dev))
-		mlx5_destroy_flow_table(esw->fdb_table.offloads.fwd_fdb);
-	mlx5_destroy_flow_table(esw->fdb_table.offloads.fast_fdb);
+	int l;
+
+	mutex_lock(&esw->fdb_table.offloads.fdb_prio_lock);
+
+	for (l = level; l >= 0; l--) {
+		if (--(fdb_prio_table(esw, chain, prio, l).num_rules) > 0)
+			continue;
+
+		put_sz_to_pool(esw, fdb_prio_table(esw, chain, prio, l).fdb->max_fte);
+		mlx5_destroy_flow_table(fdb_prio_table(esw, chain, prio, l).fdb);
+		fdb_prio_table(esw, chain, prio, l).fdb = NULL;
+	}
+
+	mutex_unlock(&esw->fdb_table.offloads.fdb_prio_lock);
+}
+
+static void esw_destroy_offloads_fast_fdb_tables(struct mlx5_eswitch *esw)
+{
+	/* If lazy creation isn't supported, deref the fast path tables */
+	if (!(esw->fdb_table.flags & ESW_FDB_CHAINS_AND_PRIOS_SUPPORTED)) {
+		esw_put_prio_table(esw, 0, 1, 1);
+		esw_put_prio_table(esw, 0, 1, 0);
+	}
 }
 
 #define MAX_PF_SQ 256
@@ -574,12 +761,13 @@ static int esw_create_offloads_fdb_tables(struct mlx5_eswitch *esw, int nvports)
 	int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
 	struct mlx5_flow_table_attr ft_attr = {};
 	struct mlx5_core_dev *dev = esw->dev;
+	u32 *flow_group_in, max_flow_counter;
 	struct mlx5_flow_namespace *root_ns;
 	struct mlx5_flow_table *fdb = NULL;
-	int table_size, ix, err = 0;
+	int table_size, ix, err = 0, i;
 	struct mlx5_flow_group *g;
+	u32 flags = 0, fdb_max;
 	void *match_criteria;
-	u32 *flow_group_in;
 	u8 *dmac;
 
 	esw_debug(esw->dev, "Create offloads FDB Tables\n");
@@ -594,12 +782,29 @@ static int esw_create_offloads_fdb_tables(struct mlx5_eswitch *esw, int nvports)
 		goto ns_err;
 	}
 
-	err = esw_create_offloads_fast_fdb_table(esw);
-	if (err)
-		goto fast_fdb_err;
+	max_flow_counter = (MLX5_CAP_GEN(dev, max_flow_counter_31_16) << 16) |
+			    MLX5_CAP_GEN(dev, max_flow_counter_15_0);
+	fdb_max = 1 << MLX5_CAP_ESW_FLOWTABLE_FDB(dev, log_max_ft_size);
+
+	esw_debug(dev, "Create offloads FDB table, min (max esw size(2^%d), max counters(%d), groups(%d), max flow table size(2^%d))\n",
+		  MLX5_CAP_ESW_FLOWTABLE_FDB(dev, log_max_ft_size),
+		  max_flow_counter, ESW_OFFLOADS_NUM_GROUPS,
+		  fdb_max);
+
+	for (i = 0; i < ARRAY_SIZE(ESW_POOLS); i++)
+		esw->fdb_table.offloads.fdb_left[i] =
+			ESW_POOLS[i] <= fdb_max ? ESW_SIZE / ESW_POOLS[i] : 0;
 
 	table_size = nvports * MAX_SQ_NVPORTS + MAX_PF_SQ + 2;
 
+	/* create the slow path fdb with encap set, so further table instances
+	 * can be created at run time while VFs are probed if the FW allows that.
+	 */
+	if (esw->offloads.encap != DEVLINK_ESWITCH_ENCAP_MODE_NONE)
+		flags |= (MLX5_FLOW_TABLE_TUNNEL_EN_REFORMAT |
+			  MLX5_FLOW_TABLE_TUNNEL_EN_DECAP);
+
+	ft_attr.flags = flags;
 	ft_attr.max_fte = table_size;
 	ft_attr.prio = FDB_SLOW_PATH;
 
@@ -611,6 +816,18 @@ static int esw_create_offloads_fdb_tables(struct mlx5_eswitch *esw, int nvports)
 	}
 	esw->fdb_table.offloads.slow_fdb = fdb;
 
+	/* If lazy creation isn't supported, open the fast path tables now */
+	if (!MLX5_CAP_ESW_FLOWTABLE(esw->dev, multi_fdb_encap) &&
+	    esw->offloads.encap != DEVLINK_ESWITCH_ENCAP_MODE_NONE) {
+		esw->fdb_table.flags &= ~ESW_FDB_CHAINS_AND_PRIOS_SUPPORTED;
+		esw_warn(dev, "Lazy creation of flow tables isn't supported, ignoring priorities\n");
+		esw_get_prio_table(esw, 0, 1, 0);
+		esw_get_prio_table(esw, 0, 1, 1);
+	} else {
+		esw_debug(dev, "Lazy creation of flow tables supported, deferring table opening\n");
+		esw->fdb_table.flags |= ESW_FDB_CHAINS_AND_PRIOS_SUPPORTED;
+	}
+
 	/* create send-to-vport group */
 	memset(flow_group_in, 0, inlen);
 	MLX5_SET(create_flow_group_in, flow_group_in, match_criteria_enable,
@@ -658,6 +875,7 @@ static int esw_create_offloads_fdb_tables(struct mlx5_eswitch *esw, int nvports)
 	if (err)
 		goto miss_rule_err;
 
+	esw->nvports = nvports;
 	kvfree(flow_group_in);
 	return 0;
 
@@ -666,10 +884,9 @@ static int esw_create_offloads_fdb_tables(struct mlx5_eswitch *esw, int nvports)
 miss_err:
 	mlx5_destroy_flow_group(esw->fdb_table.offloads.send_to_vport_grp);
 send_vport_err:
+	esw_destroy_offloads_fast_fdb_tables(esw);
 	mlx5_destroy_flow_table(esw->fdb_table.offloads.slow_fdb);
 slow_fdb_err:
-	esw_destroy_offloads_fast_fdb_table(esw);
-fast_fdb_err:
 ns_err:
 	kvfree(flow_group_in);
 	return err;
@@ -677,7 +894,7 @@ static int esw_create_offloads_fdb_tables(struct mlx5_eswitch *esw, int nvports)
 
 static void esw_destroy_offloads_fdb_tables(struct mlx5_eswitch *esw)
 {
-	if (!esw->fdb_table.offloads.fast_fdb)
+	if (!esw->fdb_table.offloads.slow_fdb)
 		return;
 
 	esw_debug(esw->dev, "Destroy offloads FDB Tables\n");
@@ -687,7 +904,7 @@ static void esw_destroy_offloads_fdb_tables(struct mlx5_eswitch *esw)
 	mlx5_destroy_flow_group(esw->fdb_table.offloads.miss_grp);
 
 	mlx5_destroy_flow_table(esw->fdb_table.offloads.slow_fdb);
-	esw_destroy_offloads_fast_fdb_table(esw);
+	esw_destroy_offloads_fast_fdb_tables(esw);
 }
 
 static int esw_create_offloads_table(struct mlx5_eswitch *esw)
@@ -944,6 +1161,8 @@ int esw_offloads_init(struct mlx5_eswitch *esw, int nvports)
 {
 	int err;
 
+	mutex_init(&esw->fdb_table.offloads.fdb_prio_lock);
+
 	err = esw_create_offloads_fdb_tables(esw, nvports);
 	if (err)
 		return err;
@@ -1272,16 +1491,19 @@ int mlx5_devlink_eswitch_encap_mode_set(struct devlink *devlink, u8 encap,
 		return -EOPNOTSUPP;
 	}
 
-	esw_destroy_offloads_fast_fdb_table(esw);
+	esw_destroy_offloads_fdb_tables(esw);
 
 	esw->offloads.encap = encap;
-	err = esw_create_offloads_fast_fdb_table(esw);
+
+	err = esw_create_offloads_fdb_tables(esw, esw->nvports);
+
 	if (err) {
 		NL_SET_ERR_MSG_MOD(extack,
 				   "Failed re-creating fast FDB table");
 		esw->offloads.encap = !encap;
-		(void)esw_create_offloads_fast_fdb_table(esw);
+		(void)esw_create_offloads_fdb_tables(esw, esw->nvports);
 	}
+
 	return err;
 }
 
-- 
2.17.2

[net-next 09/14] net/mlx5: Add a no-append flow insertion mode

[Saeed Mahameed]

From: Paul Blakey <paulb@mellanox.com>

If no-append flag is set, we will add a new FTE, instead of appending
the actions of the inserted rule when the same match already exists.

While here, move the has_flow_tag boolean indicator to be a flag too.

This patch doesn't change any functionality.

Signed-off-by: Paul Blakey <paulb@mellanox.com>
Reviewed-by: Or Gerlitz <ogerlitz@mellanmox.com>
Reviewed-by: Mark Bloch <markb@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
 drivers/infiniband/hw/mlx5/main.c                  |  6 +++---
 drivers/net/ethernet/mellanox/mlx5/core/en_tc.c    |  2 +-
 .../net/ethernet/mellanox/mlx5/core/fpga/ipsec.c   |  2 +-
 drivers/net/ethernet/mellanox/mlx5/core/fs_core.c  |  9 ++++++++-
 include/linux/mlx5/fs.h                            | 14 +++++++++++---
 5 files changed, 24 insertions(+), 9 deletions(-)

diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index 5ced0cc46ba1..af32899bb72a 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -2793,7 +2793,7 @@ static int parse_flow_attr(struct mlx5_core_dev *mdev, u32 *match_c,
 			return -EINVAL;
 
 		action->flow_tag = ib_spec->flow_tag.tag_id;
-		action->has_flow_tag = true;
+		action->flags |= FLOW_ACT_HAS_TAG;
 		break;
 	case IB_FLOW_SPEC_ACTION_DROP:
 		if (FIELDS_NOT_SUPPORTED(ib_spec->drop,
@@ -2886,7 +2886,7 @@ is_valid_esp_aes_gcm(struct mlx5_core_dev *mdev,
 		return egress ? VALID_SPEC_INVALID : VALID_SPEC_NA;
 
 	return is_crypto && is_ipsec &&
-		(!egress || (!is_drop && !flow_act->has_flow_tag)) ?
+		(!egress || (!is_drop && !(flow_act->flags & FLOW_ACT_HAS_TAG))) ?
 		VALID_SPEC_VALID : VALID_SPEC_INVALID;
 }
 
@@ -3349,7 +3349,7 @@ static struct mlx5_ib_flow_handler *_create_flow_rule(struct mlx5_ib_dev *dev,
 					MLX5_FLOW_CONTEXT_ACTION_FWD_NEXT_PRIO;
 	}
 
-	if (flow_act.has_flow_tag &&
+	if ((flow_act.flags & FLOW_ACT_HAS_TAG)  &&
 	    (flow_attr->type == IB_FLOW_ATTR_ALL_DEFAULT ||
 	     flow_attr->type == IB_FLOW_ATTR_MC_DEFAULT)) {
 		mlx5_ib_warn(dev, "Flow tag %u and attribute type %x isn't allowed in leftovers\n",
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
index 6c04e11f9a05..a9c68b7859b4 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
@@ -684,9 +684,9 @@ mlx5e_tc_add_nic_flow(struct mlx5e_priv *priv,
 	struct mlx5_flow_destination dest[2] = {};
 	struct mlx5_flow_act flow_act = {
 		.action = attr->action,
-		.has_flow_tag = true,
 		.flow_tag = attr->flow_tag,
 		.reformat_id = 0,
+		.flags    = FLOW_ACT_HAS_TAG,
 	};
 	struct mlx5_fc *counter = NULL;
 	bool table_created = false;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fpga/ipsec.c b/drivers/net/ethernet/mellanox/mlx5/core/fpga/ipsec.c
index 5645a4facad2..28aa8c968a80 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/fpga/ipsec.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fpga/ipsec.c
@@ -650,7 +650,7 @@ static bool mlx5_is_fpga_egress_ipsec_rule(struct mlx5_core_dev *dev,
 	    (match_criteria_enable &
 	     ~(MLX5_MATCH_OUTER_HEADERS | MLX5_MATCH_MISC_PARAMETERS)) ||
 	    (flow_act->action & ~(MLX5_FLOW_CONTEXT_ACTION_ENCRYPT | MLX5_FLOW_CONTEXT_ACTION_ALLOW)) ||
-	     flow_act->has_flow_tag)
+	     (flow_act->flags & FLOW_ACT_HAS_TAG))
 		return false;
 
 	return true;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c b/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c
index 7eb6d58733ac..67ba4c975d81 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c
@@ -1428,7 +1428,7 @@ static int check_conflicting_ftes(struct fs_fte *fte, const struct mlx5_flow_act
 		return -EEXIST;
 	}
 
-	if (flow_act->has_flow_tag &&
+	if ((flow_act->flags & FLOW_ACT_HAS_TAG) &&
 	    fte->action.flow_tag != flow_act->flow_tag) {
 		mlx5_core_warn(get_dev(&fte->node),
 			       "FTE flow tag %u already exists with different flow tag %u\n",
@@ -1628,6 +1628,8 @@ try_add_to_existing_fg(struct mlx5_flow_table *ft,
 
 search_again_locked:
 	version = matched_fgs_get_version(match_head);
+	if (flow_act->flags & FLOW_ACT_NO_APPEND)
+		goto skip_search;
 	/* Try to find a fg that already contains a matching fte */
 	list_for_each_entry(iter, match_head, list) {
 		struct fs_fte *fte_tmp;
@@ -1644,6 +1646,11 @@ try_add_to_existing_fg(struct mlx5_flow_table *ft,
 		return rule;
 	}
 
+skip_search:
+	/* No group with matching fte found, or we skipped the search.
+	 * Try to add a new fte to any matching fg.
+	 */
+
 	/* Check the ft version, for case that new flow group
 	 * was added while the fgs weren't locked
 	 */
diff --git a/include/linux/mlx5/fs.h b/include/linux/mlx5/fs.h
index f8d00872c7d3..5660f07d3be0 100644
--- a/include/linux/mlx5/fs.h
+++ b/include/linux/mlx5/fs.h
@@ -158,20 +158,28 @@ struct mlx5_fs_vlan {
 
 #define MLX5_FS_VLAN_DEPTH	2
 
+enum {
+	FLOW_ACT_HAS_TAG   = BIT(0),
+	FLOW_ACT_NO_APPEND = BIT(1),
+};
+
 struct mlx5_flow_act {
 	u32 action;
-	bool has_flow_tag;
 	u32 flow_tag;
 	u32 reformat_id;
 	u32 modify_id;
 	uintptr_t esp_id;
+	u32 flags;
 	struct mlx5_fs_vlan vlan[MLX5_FS_VLAN_DEPTH];
 	struct ib_counters *counters;
 };
 
 #define MLX5_DECLARE_FLOW_ACT(name) \
-	struct mlx5_flow_act name = {MLX5_FLOW_CONTEXT_ACTION_FWD_DEST,\
-				     MLX5_FS_DEFAULT_FLOW_TAG, 0, 0}
+	struct mlx5_flow_act name = { .action = MLX5_FLOW_CONTEXT_ACTION_FWD_DEST,\
+				      .flow_tag = MLX5_FS_DEFAULT_FLOW_TAG, \
+				      .reformat_id = 0, \
+				      .modify_id = 0, \
+				      .flags =  0, }
 
 /* Single destination per rule.
  * Group ID is implied by the match criteria.
-- 
2.17.2

[net-next 10/14] net/mlx5e: For TC offloads, always add new flow instead of appending the actions

[Saeed Mahameed]

From: Paul Blakey <paulb@mellanox.com>

When replacing a tc flower rule, flower first requests to add the
new rule (new action), then deletes the old one.
But currently when asked to add a new tc flower flow, we append the
actions (and counters to it).

This can result in a fte with two flow counters or conflicting
actions (drop and encap action) which firmware complains/errs
about and isn't achieving what the user aimed for.

Instead, insert the flow using the new no-append flag which will add a
new HW rule, the old flow and rule will be deleted later by flower

Signed-off-by: Paul Blakey <paulb@mellanox.com>
Reviewed-by: Or Gerlitz <ogerlitz@mellanmox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
 drivers/net/ethernet/mellanox/mlx5/core/en_tc.c            | 2 +-
 drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c | 4 ++--
 2 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
index a9c68b7859b4..8b25850cbf6a 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
@@ -686,7 +686,7 @@ mlx5e_tc_add_nic_flow(struct mlx5e_priv *priv,
 		.action = attr->action,
 		.flow_tag = attr->flow_tag,
 		.reformat_id = 0,
-		.flags    = FLOW_ACT_HAS_TAG,
+		.flags    = FLOW_ACT_HAS_TAG | FLOW_ACT_NO_APPEND,
 	};
 	struct mlx5_fc *counter = NULL;
 	bool table_created = false;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c b/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
index 8501b6c31c02..289f1992f624 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
@@ -80,8 +80,8 @@ mlx5_eswitch_add_offloaded_rule(struct mlx5_eswitch *esw,
 				struct mlx5_esw_flow_attr *attr)
 {
 	struct mlx5_flow_destination dest[MLX5_MAX_FLOW_FWD_VPORTS + 1] = {};
+	struct mlx5_flow_act flow_act = { .flags = FLOW_ACT_NO_APPEND, };
 	bool mirror = !!(attr->mirror_count);
-	struct mlx5_flow_act flow_act = {0};
 	struct mlx5_flow_handle *rule;
 	struct mlx5_flow_table *fdb;
 	int j, i = 0;
@@ -195,7 +195,7 @@ mlx5_eswitch_add_fwd_rule(struct mlx5_eswitch *esw,
 			  struct mlx5_esw_flow_attr *attr)
 {
 	struct mlx5_flow_destination dest[MLX5_MAX_FLOW_FWD_VPORTS + 1] = {};
-	struct mlx5_flow_act flow_act = {0};
+	struct mlx5_flow_act flow_act = { .flags = FLOW_ACT_NO_APPEND, };
 	struct mlx5_flow_table *fast_fdb;
 	struct mlx5_flow_table *fwd_fdb;
 	struct mlx5_flow_handle *rule;
-- 
2.17.2

[net-next 11/14] net/mlx5e: Avoid duplicated code for tc offloads add/del fdb rule

[Saeed Mahameed]

From: Or Gerlitz <ogerlitz@mellanox.com>

The code for adding/deleting fdb flow is repeated when
user-space does flow add/del and when we add/del from
the neigh update path - unify them to avoid the duplication.

Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Paul Blakey <paulb@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
 .../net/ethernet/mellanox/mlx5/core/en_tc.c   | 91 ++++++++++---------
 1 file changed, 50 insertions(+), 41 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
index 8b25850cbf6a..1786e25644ac 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
@@ -823,6 +823,43 @@ static int mlx5e_attach_encap(struct mlx5e_priv *priv,
 			      struct mlx5e_tc_flow *flow,
 			      struct netlink_ext_ack *extack);
 
+static struct mlx5_flow_handle *
+mlx5e_tc_offload_fdb_rules(struct mlx5_eswitch *esw,
+			   struct mlx5e_tc_flow *flow,
+			   struct mlx5_flow_spec *spec,
+			   struct mlx5_esw_flow_attr *attr)
+{
+	struct mlx5_flow_handle *rule;
+
+	rule = mlx5_eswitch_add_offloaded_rule(esw, spec, attr);
+	if (IS_ERR(rule))
+		return rule;
+
+	if (attr->mirror_count) {
+		flow->rule[1] = mlx5_eswitch_add_fwd_rule(esw, spec, attr);
+		if (IS_ERR(flow->rule[1])) {
+			mlx5_eswitch_del_offloaded_rule(esw, rule, attr);
+			return flow->rule[1];
+		}
+	}
+
+	flow->flags |= MLX5E_TC_FLOW_OFFLOADED;
+	return rule;
+}
+
+static void
+mlx5e_tc_unoffload_fdb_rules(struct mlx5_eswitch *esw,
+			     struct mlx5e_tc_flow *flow,
+			   struct mlx5_esw_flow_attr *attr)
+{
+	flow->flags &= ~MLX5E_TC_FLOW_OFFLOADED;
+
+	if (attr->mirror_count)
+		mlx5_eswitch_del_fwd_rule(esw, flow->rule[1], attr);
+
+	mlx5_eswitch_del_offloaded_rule(esw, flow->rule[0], attr);
+}
+
 static int
 mlx5e_tc_add_fdb_flow(struct mlx5e_priv *priv,
 		      struct mlx5e_tc_flow_parse_attr *parse_attr,
@@ -881,25 +918,15 @@ mlx5e_tc_add_fdb_flow(struct mlx5e_priv *priv,
 	 * (2) there's an encap action and we're on -EAGAIN (no valid neigh)
 	 */
 	if (encap_err != -EAGAIN) {
-		flow->rule[0] = mlx5_eswitch_add_offloaded_rule(esw, &parse_attr->spec, attr);
+		flow->rule[0] = mlx5e_tc_offload_fdb_rules(esw, flow, &parse_attr->spec, attr);
 		if (IS_ERR(flow->rule[0])) {
 			err = PTR_ERR(flow->rule[0]);
 			goto err_add_rule;
 		}
-
-		if (attr->mirror_count) {
-			flow->rule[1] = mlx5_eswitch_add_fwd_rule(esw, &parse_attr->spec, attr);
-			if (IS_ERR(flow->rule[1])) {
-				err = PTR_ERR(flow->rule[1]);
-				goto err_fwd_rule;
-			}
-		}
 	}
 
 	return encap_err;
 
-err_fwd_rule:
-	mlx5_eswitch_del_offloaded_rule(esw, flow->rule[0], attr);
 err_add_rule:
 	mlx5_fc_destroy(esw->dev, counter);
 err_create_counter:
@@ -920,12 +947,8 @@ static void mlx5e_tc_del_fdb_flow(struct mlx5e_priv *priv,
 	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
 	struct mlx5_esw_flow_attr *attr = flow->esw_attr;
 
-	if (flow->flags & MLX5E_TC_FLOW_OFFLOADED) {
-		flow->flags &= ~MLX5E_TC_FLOW_OFFLOADED;
-		if (attr->mirror_count)
-			mlx5_eswitch_del_fwd_rule(esw, flow->rule[1], attr);
-		mlx5_eswitch_del_offloaded_rule(esw, flow->rule[0], attr);
-	}
+	if (flow->flags & MLX5E_TC_FLOW_OFFLOADED)
+		mlx5e_tc_unoffload_fdb_rules(esw, flow, flow->esw_attr);
 
 	mlx5_eswitch_del_vlan_action(esw, attr);
 
@@ -946,6 +969,8 @@ void mlx5e_tc_encap_flows_add(struct mlx5e_priv *priv,
 {
 	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
 	struct mlx5_esw_flow_attr *esw_attr;
+	struct mlx5_flow_handle *rule;
+	struct mlx5_flow_spec *spec;
 	struct mlx5e_tc_flow *flow;
 	int err;
 
@@ -964,26 +989,16 @@ void mlx5e_tc_encap_flows_add(struct mlx5e_priv *priv,
 	list_for_each_entry(flow, &e->flows, encap) {
 		esw_attr = flow->esw_attr;
 		esw_attr->encap_id = e->encap_id;
-		flow->rule[0] = mlx5_eswitch_add_offloaded_rule(esw, &esw_attr->parse_attr->spec, esw_attr);
-		if (IS_ERR(flow->rule[0])) {
-			err = PTR_ERR(flow->rule[0]);
+		spec = &esw_attr->parse_attr->spec;
+
+		rule = mlx5e_tc_offload_fdb_rules(esw, flow, spec, esw_attr);
+		if (IS_ERR(rule)) {
+			err = PTR_ERR(rule);
 			mlx5_core_warn(priv->mdev, "Failed to update cached encapsulation flow, %d\n",
 				       err);
 			continue;
 		}
-
-		if (esw_attr->mirror_count) {
-			flow->rule[1] = mlx5_eswitch_add_fwd_rule(esw, &esw_attr->parse_attr->spec, esw_attr);
-			if (IS_ERR(flow->rule[1])) {
-				mlx5_eswitch_del_offloaded_rule(esw, flow->rule[0], esw_attr);
-				err = PTR_ERR(flow->rule[1]);
-				mlx5_core_warn(priv->mdev, "Failed to update cached mirror flow, %d\n",
-					       err);
-				continue;
-			}
-		}
-
-		flow->flags |= MLX5E_TC_FLOW_OFFLOADED;
+		flow->rule[0] = rule;
 	}
 }
 
@@ -994,14 +1009,8 @@ void mlx5e_tc_encap_flows_del(struct mlx5e_priv *priv,
 	struct mlx5e_tc_flow *flow;
 
 	list_for_each_entry(flow, &e->flows, encap) {
-		if (flow->flags & MLX5E_TC_FLOW_OFFLOADED) {
-			struct mlx5_esw_flow_attr *attr = flow->esw_attr;
-
-			flow->flags &= ~MLX5E_TC_FLOW_OFFLOADED;
-			if (attr->mirror_count)
-				mlx5_eswitch_del_fwd_rule(esw, flow->rule[1], attr);
-			mlx5_eswitch_del_offloaded_rule(esw, flow->rule[0], attr);
-		}
+		if (flow->flags & MLX5E_TC_FLOW_OFFLOADED)
+			mlx5e_tc_unoffload_fdb_rules(esw, flow, flow->esw_attr);
 	}
 
 	if (e->flags & MLX5_ENCAP_ENTRY_VALID) {
-- 
2.17.2

[net-next 12/14] net/mlx5: E-Switch, Enable setting goto slow path chain action

[Saeed Mahameed]

From: Paul Blakey <paulb@mellanox.com>

A pre-step for the tc offloads code to use this when a neigh is
not available for encap rules.

Signed-off-by: Paul Blakey <paulb@mellanox.com>
Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
 drivers/net/ethernet/mellanox/mlx5/core/eswitch.h          | 2 ++
 drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c | 6 ++++++
 2 files changed, 8 insertions(+)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h
index 54215f4312fa..aaafc9f17115 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h
@@ -60,6 +60,7 @@
 	MLX5_CAP_ESW_FLOWTABLE(dev, fdb_multi_path_to_table)
 
 #define FDB_MAX_CHAIN 3
+#define FDB_SLOW_PATH_CHAIN (FDB_MAX_CHAIN + 1)
 #define FDB_MAX_PRIO 16
 
 struct vport_ingress {
@@ -356,6 +357,7 @@ static inline int  mlx5_eswitch_enable_sriov(struct mlx5_eswitch *esw, int nvfs,
 static inline void mlx5_eswitch_disable_sriov(struct mlx5_eswitch *esw) {}
 
 #define FDB_MAX_CHAIN 1
+#define FDB_SLOW_PATH_CHAIN (FDB_MAX_CHAIN + 1)
 #define FDB_MAX_PRIO 1
 
 #endif /* CONFIG_MLX5_ESWITCH */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c b/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
index 289f1992f624..42a130455ef8 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
@@ -671,6 +671,9 @@ esw_get_prio_table(struct mlx5_eswitch *esw, u32 chain, u16 prio, int level)
 	int table_prio, l = 0;
 	u32 flags = 0;
 
+	if (chain == FDB_SLOW_PATH_CHAIN)
+		return esw->fdb_table.offloads.slow_fdb;
+
 	mutex_lock(&esw->fdb_table.offloads.fdb_prio_lock);
 
 	fdb = fdb_prio_table(esw, chain, prio, level).fdb;
@@ -730,6 +733,9 @@ esw_put_prio_table(struct mlx5_eswitch *esw, u32 chain, u16 prio, int level)
 {
 	int l;
 
+	if (chain == FDB_SLOW_PATH_CHAIN)
+		return;
+
 	mutex_lock(&esw->fdb_table.offloads.fdb_prio_lock);
 
 	for (l = level; l >= 0; l--) {
-- 
2.17.2

[net-next 13/14] net/mlx5e: Use a slow path rule instead if vxlan neighbour isn't available

[Saeed Mahameed]

From: Paul Blakey <paulb@mellanox.com>

When adding a vxlan tc rule, and a neighbour isn't available, we
don't insert any rule to hardware. Once we enable offloading flows
with multiple priorities, a packet that should have matched this rule
will continue in hardware pipeline and might match a wrong one.

This is unlike in tc software path where it will be matched and
forwarded to the vxlan device (which will cause a ARP lookup
eventually) and stop processing further tc filters.

To address that, when when a neighbour isn't available (EAGAIN from
attach_encap), or gets deleted, change the original action to be a
forward to slow path instead. Neighbour update will restore the original
action once the neighbour becomes available. This will be done atomically
so at any given time we will have a the correct match.

Signed-off-by: Paul Blakey <paulb@mellanox.com>
Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
 .../net/ethernet/mellanox/mlx5/core/en_tc.c   | 104 ++++++++++++++----
 1 file changed, 82 insertions(+), 22 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
index 1786e25644ac..cb66964aa1ff 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
@@ -74,6 +74,7 @@ enum {
 	MLX5E_TC_FLOW_OFFLOADED	= BIT(MLX5E_TC_FLOW_BASE + 2),
 	MLX5E_TC_FLOW_HAIRPIN	= BIT(MLX5E_TC_FLOW_BASE + 3),
 	MLX5E_TC_FLOW_HAIRPIN_RSS = BIT(MLX5E_TC_FLOW_BASE + 4),
+	MLX5E_TC_FLOW_SLOW	  = BIT(MLX5E_TC_FLOW_BASE + 5),
 };
 
 #define MLX5E_TC_MAX_SPLITS 1
@@ -82,7 +83,7 @@ struct mlx5e_tc_flow {
 	struct rhash_head	node;
 	struct mlx5e_priv	*priv;
 	u64			cookie;
-	u8			flags;
+	u16			flags;
 	struct mlx5_flow_handle *rule[MLX5E_TC_MAX_SPLITS + 1];
 	struct list_head	encap;   /* flows sharing the same encap ID */
 	struct list_head	mod_hdr; /* flows sharing the same mod hdr ID */
@@ -860,6 +861,36 @@ mlx5e_tc_unoffload_fdb_rules(struct mlx5_eswitch *esw,
 	mlx5_eswitch_del_offloaded_rule(esw, flow->rule[0], attr);
 }
 
+static struct mlx5_flow_handle *
+mlx5e_tc_offload_to_slow_path(struct mlx5_eswitch *esw,
+			      struct mlx5e_tc_flow *flow,
+			      struct mlx5_flow_spec *spec,
+			      struct mlx5_esw_flow_attr *slow_attr)
+{
+	struct mlx5_flow_handle *rule;
+
+	memcpy(slow_attr, flow->esw_attr, sizeof(*slow_attr));
+	slow_attr->action = MLX5_FLOW_CONTEXT_ACTION_FWD_DEST,
+	slow_attr->mirror_count = 0,
+	slow_attr->dest_chain = FDB_SLOW_PATH_CHAIN,
+
+	rule = mlx5e_tc_offload_fdb_rules(esw, flow, spec, slow_attr);
+	if (!IS_ERR(rule))
+		flow->flags |= MLX5E_TC_FLOW_SLOW;
+
+	return rule;
+}
+
+static void
+mlx5e_tc_unoffload_from_slow_path(struct mlx5_eswitch *esw,
+				  struct mlx5e_tc_flow *flow,
+				  struct mlx5_esw_flow_attr *slow_attr)
+{
+	memcpy(slow_attr, flow->esw_attr, sizeof(*slow_attr));
+	mlx5e_tc_unoffload_fdb_rules(esw, flow, slow_attr);
+	flow->flags &= ~MLX5E_TC_FLOW_SLOW;
+}
+
 static int
 mlx5e_tc_add_fdb_flow(struct mlx5e_priv *priv,
 		      struct mlx5e_tc_flow_parse_attr *parse_attr,
@@ -917,15 +948,21 @@ mlx5e_tc_add_fdb_flow(struct mlx5e_priv *priv,
 	/* we get here if (1) there's no error or when
 	 * (2) there's an encap action and we're on -EAGAIN (no valid neigh)
 	 */
-	if (encap_err != -EAGAIN) {
+	if (encap_err == -EAGAIN) {
+		/* continue with goto slow path rule instead */
+		struct mlx5_esw_flow_attr slow_attr;
+
+		flow->rule[0] = mlx5e_tc_offload_to_slow_path(esw, flow, &parse_attr->spec, &slow_attr);
+	} else {
 		flow->rule[0] = mlx5e_tc_offload_fdb_rules(esw, flow, &parse_attr->spec, attr);
-		if (IS_ERR(flow->rule[0])) {
-			err = PTR_ERR(flow->rule[0]);
-			goto err_add_rule;
-		}
 	}
 
-	return encap_err;
+	if (IS_ERR(flow->rule[0])) {
+		err = PTR_ERR(flow->rule[0]);
+		goto err_add_rule;
+	}
+
+	return 0;
 
 err_add_rule:
 	mlx5_fc_destroy(esw->dev, counter);
@@ -946,9 +983,14 @@ static void mlx5e_tc_del_fdb_flow(struct mlx5e_priv *priv,
 {
 	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
 	struct mlx5_esw_flow_attr *attr = flow->esw_attr;
+	struct mlx5_esw_flow_attr slow_attr;
 
-	if (flow->flags & MLX5E_TC_FLOW_OFFLOADED)
-		mlx5e_tc_unoffload_fdb_rules(esw, flow, flow->esw_attr);
+	if (flow->flags & MLX5E_TC_FLOW_OFFLOADED) {
+		if (flow->flags & MLX5E_TC_FLOW_SLOW)
+			mlx5e_tc_unoffload_from_slow_path(esw, flow, &slow_attr);
+		else
+			mlx5e_tc_unoffload_fdb_rules(esw, flow, attr);
+	}
 
 	mlx5_eswitch_del_vlan_action(esw, attr);
 
@@ -968,7 +1010,7 @@ void mlx5e_tc_encap_flows_add(struct mlx5e_priv *priv,
 			      struct mlx5e_encap_entry *e)
 {
 	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
-	struct mlx5_esw_flow_attr *esw_attr;
+	struct mlx5_esw_flow_attr slow_attr, *esw_attr;
 	struct mlx5_flow_handle *rule;
 	struct mlx5_flow_spec *spec;
 	struct mlx5e_tc_flow *flow;
@@ -991,6 +1033,7 @@ void mlx5e_tc_encap_flows_add(struct mlx5e_priv *priv,
 		esw_attr->encap_id = e->encap_id;
 		spec = &esw_attr->parse_attr->spec;
 
+		/* update from slow path rule to encap rule */
 		rule = mlx5e_tc_offload_fdb_rules(esw, flow, spec, esw_attr);
 		if (IS_ERR(rule)) {
 			err = PTR_ERR(rule);
@@ -998,6 +1041,9 @@ void mlx5e_tc_encap_flows_add(struct mlx5e_priv *priv,
 				       err);
 			continue;
 		}
+
+		mlx5e_tc_unoffload_from_slow_path(esw, flow, &slow_attr);
+		flow->flags |= MLX5E_TC_FLOW_OFFLOADED; /* was unset when slow path rule removed */
 		flow->rule[0] = rule;
 	}
 }
@@ -1006,11 +1052,28 @@ void mlx5e_tc_encap_flows_del(struct mlx5e_priv *priv,
 			      struct mlx5e_encap_entry *e)
 {
 	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	struct mlx5_esw_flow_attr slow_attr;
+	struct mlx5_flow_handle *rule;
+	struct mlx5_flow_spec *spec;
 	struct mlx5e_tc_flow *flow;
+	int err;
 
 	list_for_each_entry(flow, &e->flows, encap) {
-		if (flow->flags & MLX5E_TC_FLOW_OFFLOADED)
-			mlx5e_tc_unoffload_fdb_rules(esw, flow, flow->esw_attr);
+		spec = &flow->esw_attr->parse_attr->spec;
+
+		/* update from encap rule to slow path rule */
+		rule = mlx5e_tc_offload_to_slow_path(esw, flow, spec, &slow_attr);
+
+		if (IS_ERR(rule)) {
+			err = PTR_ERR(rule);
+			mlx5_core_warn(priv->mdev, "Failed to update slow path (encap) flow, %d\n",
+				       err);
+			continue;
+		}
+
+		mlx5e_tc_unoffload_fdb_rules(esw, flow, flow->esw_attr);
+		flow->flags |= MLX5E_TC_FLOW_OFFLOADED; /* was unset when fast path rule removed */
+		flow->rule[0] = rule;
 	}
 
 	if (e->flags & MLX5_ENCAP_ENTRY_VALID) {
@@ -2888,9 +2951,9 @@ static int parse_tc_fdb_actions(struct mlx5e_priv *priv, struct tcf_exts *exts,
 	return 0;
 }
 
-static void get_flags(int flags, u8 *flow_flags)
+static void get_flags(int flags, u16 *flow_flags)
 {
-	u8 __flow_flags = 0;
+	u16 __flow_flags = 0;
 
 	if (flags & MLX5E_TC_INGRESS)
 		__flow_flags |= MLX5E_TC_FLOW_INGRESS;
@@ -2921,7 +2984,7 @@ static struct rhashtable *get_tc_ht(struct mlx5e_priv *priv)
 
 static int
 mlx5e_alloc_flow(struct mlx5e_priv *priv, int attr_size,
-		 struct tc_cls_flower_offload *f, u8 flow_flags,
+		 struct tc_cls_flower_offload *f, u16 flow_flags,
 		 struct mlx5e_tc_flow_parse_attr **__parse_attr,
 		 struct mlx5e_tc_flow **__flow)
 {
@@ -2958,7 +3021,7 @@ mlx5e_alloc_flow(struct mlx5e_priv *priv, int attr_size,
 static int
 mlx5e_add_fdb_flow(struct mlx5e_priv *priv,
 		   struct tc_cls_flower_offload *f,
-		   u8 flow_flags,
+		   u16 flow_flags,
 		   struct mlx5e_tc_flow **__flow)
 {
 	struct netlink_ext_ack *extack = f->common.extack;
@@ -2978,12 +3041,9 @@ mlx5e_add_fdb_flow(struct mlx5e_priv *priv,
 		goto err_free;
 
 	err = mlx5e_tc_add_fdb_flow(priv, parse_attr, flow, extack);
-	if (err && err != -EAGAIN)
+	if (err)
 		goto err_free;
 
-	if (!err)
-		flow->flags |= MLX5E_TC_FLOW_OFFLOADED;
-
 	if (!(flow->esw_attr->action &
 	      MLX5_FLOW_CONTEXT_ACTION_PACKET_REFORMAT))
 		kvfree(parse_attr);
@@ -3002,7 +3062,7 @@ mlx5e_add_fdb_flow(struct mlx5e_priv *priv,
 static int
 mlx5e_add_nic_flow(struct mlx5e_priv *priv,
 		   struct tc_cls_flower_offload *f,
-		   u8 flow_flags,
+		   u16 flow_flags,
 		   struct mlx5e_tc_flow **__flow)
 {
 	struct netlink_ext_ack *extack = f->common.extack;
@@ -3045,7 +3105,7 @@ mlx5e_tc_add_flow(struct mlx5e_priv *priv,
 		  struct mlx5e_tc_flow **flow)
 {
 	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
-	u8 flow_flags;
+	u16 flow_flags;
 	int err;
 
 	get_flags(flags, &flow_flags);
-- 
2.17.2

[net-next 14/14] net/mlx5e: Support offloading tc priorities and chains for eswitch flows

[Saeed Mahameed]

From: Paul Blakey <paulb@mellanox.com>

Currently we fail when user specify a non-zero chain, this patch adds the
support for it and tc priorities. To get to a new chain, use the tc
goto action.

Currently we support a fixed prio range 1-16, and chain range 0-3.

Signed-off-by: Paul Blakey <paulb@mellanox.com>
Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
 .../net/ethernet/mellanox/mlx5/core/en_main.c |  3 --
 .../net/ethernet/mellanox/mlx5/core/en_rep.c  |  6 ---
 .../net/ethernet/mellanox/mlx5/core/en_tc.c   | 51 ++++++++++++++++++-
 .../mellanox/mlx5/core/eswitch_offloads.c     |  2 +-
 4 files changed, 50 insertions(+), 12 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
index c9848e333450..1243edbedc9e 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
@@ -3392,9 +3392,6 @@ static int mlx5e_setup_tc_block_cb(enum tc_setup_type type, void *type_data,
 {
 	struct mlx5e_priv *priv = cb_priv;
 
-	if (!tc_cls_can_offload_and_chain0(priv->netdev, type_data))
-		return -EOPNOTSUPP;
-
 	switch (type) {
 	case TC_SETUP_CLSFLOWER:
 		return mlx5e_setup_tc_cls_flower(priv, type_data, MLX5E_TC_INGRESS);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c b/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c
index 64c2b9ea8b1e..c3c657548824 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c
@@ -853,9 +853,6 @@ static int mlx5e_rep_setup_tc_cb_egdev(enum tc_setup_type type, void *type_data,
 {
 	struct mlx5e_priv *priv = cb_priv;
 
-	if (!tc_cls_can_offload_and_chain0(priv->netdev, type_data))
-		return -EOPNOTSUPP;
-
 	switch (type) {
 	case TC_SETUP_CLSFLOWER:
 		return mlx5e_rep_setup_tc_cls_flower(priv, type_data, MLX5E_TC_EGRESS);
@@ -869,9 +866,6 @@ static int mlx5e_rep_setup_tc_cb(enum tc_setup_type type, void *type_data,
 {
 	struct mlx5e_priv *priv = cb_priv;
 
-	if (!tc_cls_can_offload_and_chain0(priv->netdev, type_data))
-		return -EOPNOTSUPP;
-
 	switch (type) {
 	case TC_SETUP_CLSFLOWER:
 		return mlx5e_rep_setup_tc_cls_flower(priv, type_data, MLX5E_TC_INGRESS);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
index cb66964aa1ff..608025ca5c04 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
@@ -898,15 +898,32 @@ mlx5e_tc_add_fdb_flow(struct mlx5e_priv *priv,
 		      struct netlink_ext_ack *extack)
 {
 	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+	u32 max_chain = mlx5_eswitch_get_chain_range(esw);
 	struct mlx5_esw_flow_attr *attr = flow->esw_attr;
+	u16 max_prio = mlx5_eswitch_get_prio_range(esw);
 	struct net_device *out_dev, *encap_dev = NULL;
 	struct mlx5_fc *counter = NULL;
 	struct mlx5e_rep_priv *rpriv;
 	struct mlx5e_priv *out_priv;
 	int err = 0, encap_err = 0;
 
-	/* keep the old behaviour, use same prio for all offloaded rules */
-	attr->prio = 1;
+	/* if prios are not supported, keep the old behaviour of using same prio
+	 * for all offloaded rules.
+	 */
+	if (!mlx5_eswitch_prios_supported(esw))
+		attr->prio = 1;
+
+	if (attr->chain > max_chain) {
+		NL_SET_ERR_MSG(extack, "Requested chain is out of supported range");
+		err = -EOPNOTSUPP;
+		goto err_max_prio_chain;
+	}
+
+	if (attr->prio > max_prio) {
+		NL_SET_ERR_MSG(extack, "Requested priority is out of supported range");
+		err = -EOPNOTSUPP;
+		goto err_max_prio_chain;
+	}
 
 	if (attr->action & MLX5_FLOW_CONTEXT_ACTION_PACKET_REFORMAT) {
 		out_dev = __dev_get_by_index(dev_net(priv->netdev),
@@ -975,6 +992,7 @@ mlx5e_tc_add_fdb_flow(struct mlx5e_priv *priv,
 	if (attr->action & MLX5_FLOW_CONTEXT_ACTION_PACKET_REFORMAT)
 		mlx5e_detach_encap(priv, flow);
 err_attach_encap:
+err_max_prio_chain:
 	return err;
 }
 
@@ -2826,6 +2844,7 @@ static int parse_tc_fdb_actions(struct mlx5e_priv *priv, struct tcf_exts *exts,
 				struct mlx5e_tc_flow *flow,
 				struct netlink_ext_ack *extack)
 {
+	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
 	struct mlx5_esw_flow_attr *attr = flow->esw_attr;
 	struct mlx5e_rep_priv *rpriv = priv->ppriv;
 	struct ip_tunnel_info *info = NULL;
@@ -2934,6 +2953,25 @@ static int parse_tc_fdb_actions(struct mlx5e_priv *priv, struct tcf_exts *exts,
 			continue;
 		}
 
+		if (is_tcf_gact_goto_chain(a)) {
+			u32 dest_chain = tcf_gact_goto_chain_index(a);
+			u32 max_chain = mlx5_eswitch_get_chain_range(esw);
+
+			if (dest_chain <= attr->chain) {
+				NL_SET_ERR_MSG(extack, "Goto earlier chain isn't supported");
+				return -EOPNOTSUPP;
+			}
+			if (dest_chain > max_chain) {
+				NL_SET_ERR_MSG(extack, "Requested destination chain is out of supported range");
+				return -EOPNOTSUPP;
+			}
+			action |= MLX5_FLOW_CONTEXT_ACTION_FWD_DEST |
+				  MLX5_FLOW_CONTEXT_ACTION_COUNT;
+			attr->dest_chain = dest_chain;
+
+			continue;
+		}
+
 		return -EINVAL;
 	}
 
@@ -3036,6 +3074,8 @@ mlx5e_add_fdb_flow(struct mlx5e_priv *priv,
 	if (err)
 		goto out;
 
+	flow->esw_attr->chain = f->common.chain_index;
+	flow->esw_attr->prio = TC_H_MAJ(f->common.prio) >> 16;
 	err = parse_tc_fdb_actions(priv, f->exts, parse_attr, flow, extack);
 	if (err)
 		goto err_free;
@@ -3070,6 +3110,10 @@ mlx5e_add_nic_flow(struct mlx5e_priv *priv,
 	struct mlx5e_tc_flow *flow;
 	int attr_size, err;
 
+	/* multi-chain not supported for NIC rules */
+	if (!tc_cls_can_offload_and_chain0(priv->netdev, &f->common))
+		return -EOPNOTSUPP;
+
 	flow_flags |= MLX5E_TC_FLOW_NIC;
 	attr_size  = sizeof(struct mlx5_nic_flow_attr);
 	err = mlx5e_alloc_flow(priv, attr_size, f, flow_flags,
@@ -3110,6 +3154,9 @@ mlx5e_tc_add_flow(struct mlx5e_priv *priv,
 
 	get_flags(flags, &flow_flags);
 
+	if (!tc_can_offload_extack(priv->netdev, f->common.extack))
+		return -EOPNOTSUPP;
+
 	if (esw && esw->mode == SRIOV_OFFLOADS)
 		err = mlx5e_add_fdb_flow(priv, f, flow_flags, flow);
 	else
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c b/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
index 42a130455ef8..9eac137790f5 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
@@ -71,7 +71,7 @@ u16 mlx5_eswitch_get_prio_range(struct mlx5_eswitch *esw)
 	if (esw->fdb_table.flags & ESW_FDB_CHAINS_AND_PRIOS_SUPPORTED)
 		return FDB_MAX_PRIO;
 
-	return U16_MAX;
+	return 1;
 }
 
 struct mlx5_flow_handle *
-- 
2.17.2

Re: [pull request][net-next 00/14] Mellanox, mlx5 updates

[David Miller]

From: Saeed Mahameed <saeedm@mellanox.com>
Date: Wed, 17 Oct 2018 17:08:45 -0700

> This series from Paul, Or and Mark provides the support for 
> e-switch tc offloading of multiple priorities and chains, the series
> is based on a merge commit with mlx5-next branch for patches that are
> already submitted and reviewed through the netdev and rdma mailing
> lists.
> 
> For more information please see tag log below.
> 
> Please pull and let me know if there is any problem.

Pulled, thanks Saeed.