[PATCH net-next 0/9] nfp: flower: decap neighbour table rework

[PATCH net-next 0/9] nfp: flower: decap neighbour table rework

[Simon Horman]

Louis Peens says:

This patch series reworks the way in which flow rules that outputs to
OVS internal ports gets handled by the nfp driver.

Previously this made use of a small pre_tun_table, but this only used
destination MAC addresses, and made the implicit assumption that there is
only a single source MAC":"destination MAC" mapping per tunnel. In
hindsight this seems to be a pretty obvious oversight, but this was hidden
in plain sight for quite some time.

This series changes the implementation to make use of the same Neighbour
table for decap that is in use for the tunnel encap solution. It stores
any new Neighbour updates in this table. Previously this path was only
triggered for encapsulation candidates, and the entries were send and
forget, not saved on the host as it is after this series. It also keeps
track of any flow rule that outputs to OVS internal ports (and some
other criteria not worth mentioning here), very similar to how it was
done previously, except now these flows are kept track of in a list.

When a new Neighbour entry gets added this list gets iterated for
potential matches, in which case the table gets updated with a reference
to the flow, and the Neighbour entry on the card gets updated with the
relevant host_ctx. The same happens when a new qualifying flow gets
added - the Neighbour table gets iterated for applicable matches, and
once again the firmware gets updated with the host_ctx when any matches
are found.

Since this also requires a firmware change we add a new capability bit,
and keep the old behaviour in case of older firmware without this bit
set.

This series starts by doing some preparation, then adding the new list
and table entries. Next the functionality to link/unlink these entries
are added, and finally this new functionality is enabled by adding the
DECAP_V2 bit to the driver feature list.


Louis Peens (9):
  nfp: flower: add infrastructure for pre_tun rework
  nfp: flower: add/remove predt_list entries
  nfp: flower: enforce more strict pre_tun checks
  nfp: flower: fixup ipv6/ipv4 route lookup for neigh events
  nfp: flower: update nfp_tun_neigh structs
  nfp: flower: rework tunnel neighbour configuration
  nfp: flower: link pre_tun flow rules with neigh entries
  nfp: flower: remove unused neighbour cache
  nfp: flower: enable decap_v2 bit

 .../ethernet/netronome/nfp/flower/action.c    |   3 +-
 .../net/ethernet/netronome/nfp/flower/main.h  | 110 +++-
 .../ethernet/netronome/nfp/flower/metadata.c  |  19 +-
 .../ethernet/netronome/nfp/flower/offload.c   |  86 ++-
 .../netronome/nfp/flower/tunnel_conf.c        | 502 +++++++++---------
 5 files changed, 453 insertions(+), 267 deletions(-)

-- 
2.30.2

[PATCH net-next 1/9] nfp: flower: add infrastructure for pre_tun rework

[Simon Horman]

From: Louis Peens <louis.peens@corigine.com>

The previous implementation of using a pre_tun_table for decap has
some limitations, causing flows to end up unoffloaded when in fact
we are able to offload them. This is because the pre_tun_table does
not have enough matching resolution. The next step is to instead make
use of the neighbour table which already exists for the encap direction.
This patch prepares for this by:

- Moving nfp_tun_neigh/_v6 to main.h.
- Creating two new "wrapping" structures, one to keep track of neighbour
  entries (previously they were send-and-forget), and another to keep
  track of pre_tun flows.
- Create a new list in nfp_flower_priv to keep track of pre_tunnel flows
- Create a new table in nfp_flower_priv to keep track of next neighbour
  entries
- Initialising and destroying these new list/tables
- Extending nfp_fl_payload->pre_tun_rule to save more information for
  future use.

Signed-off-by: Louis Peens <louis.peens@corigine.com>
Signed-off-by: Yinjun Zhang <yinjun.zhang@corigine.com>
Signed-off-by: Simon Horman <simon.horman@corigine.com>
---
 .../net/ethernet/netronome/nfp/flower/main.h  | 87 +++++++++++++++++++
 .../ethernet/netronome/nfp/flower/metadata.c  | 19 +++-
 .../netronome/nfp/flower/tunnel_conf.c        | 32 -------
 3 files changed, 105 insertions(+), 33 deletions(-)

diff --git a/drivers/net/ethernet/netronome/nfp/flower/main.h b/drivers/net/ethernet/netronome/nfp/flower/main.h
index fa902ce2dd82..454fdb6ea4a5 100644
--- a/drivers/net/ethernet/netronome/nfp/flower/main.h
+++ b/drivers/net/ethernet/netronome/nfp/flower/main.h
@@ -51,6 +51,7 @@ struct nfp_app;
 #define NFP_FL_FEATS_VLAN_QINQ		BIT(8)
 #define NFP_FL_FEATS_QOS_PPS		BIT(9)
 #define NFP_FL_FEATS_QOS_METER		BIT(10)
+#define NFP_FL_FEATS_DECAP_V2		BIT(11)
 #define NFP_FL_FEATS_HOST_ACK		BIT(31)
 
 #define NFP_FL_ENABLE_FLOW_MERGE	BIT(0)
@@ -109,6 +110,80 @@ struct nfp_fl_tunnel_offloads {
 	struct notifier_block neigh_nb;
 };
 
+/**
+ * struct nfp_tun_neigh - neighbour/route entry on the NFP
+ * @dst_ipv4:	Destination IPv4 address
+ * @src_ipv4:	Source IPv4 address
+ * @dst_addr:	Destination MAC address
+ * @src_addr:	Source MAC address
+ * @port_id:	NFP port to output packet on - associated with source IPv4
+ * @vlan_tpid:	VLAN_TPID match field
+ * @vlan_tci:	VLAN_TCI match field
+ * @host_ctx:	Host context ID to be saved here
+ */
+struct nfp_tun_neigh {
+	__be32 dst_ipv4;
+	__be32 src_ipv4;
+	u8 dst_addr[ETH_ALEN];
+	u8 src_addr[ETH_ALEN];
+	__be32 port_id;
+	__be16 vlan_tpid;
+	__be16 vlan_tci;
+	__be32 host_ctx;
+};
+
+/**
+ * struct nfp_tun_neigh_v6 - neighbour/route entry on the NFP
+ * @dst_ipv6:	Destination IPv6 address
+ * @src_ipv6:	Source IPv6 address
+ * @dst_addr:	Destination MAC address
+ * @src_addr:	Source MAC address
+ * @port_id:	NFP port to output packet on - associated with source IPv6
+ * @vlan_tpid:	VLAN_TPID match field
+ * @vlan_tci:	VLAN_TCI match field
+ * @host_ctx:	Host context ID to be saved here
+ */
+struct nfp_tun_neigh_v6 {
+	struct in6_addr dst_ipv6;
+	struct in6_addr src_ipv6;
+	u8 dst_addr[ETH_ALEN];
+	u8 src_addr[ETH_ALEN];
+	__be32 port_id;
+	__be16 vlan_tpid;
+	__be16 vlan_tci;
+	__be32 host_ctx;
+};
+
+/**
+ * struct nfp_neigh_entry
+ * @neigh_cookie:	Cookie for hashtable lookup
+ * @ht_node:		rhash_head entry for hashtable
+ * @list_head:		Needed as member of linked_nn_entries list
+ * @payload:		The neighbour info payload
+ * @flow:		Linked flow rule
+ * @is_ipv6:		Flag to indicate if payload is ipv6 or ipv4
+ */
+struct nfp_neigh_entry {
+	unsigned long neigh_cookie;
+	struct rhash_head ht_node;
+	struct list_head list_head;
+	char *payload;
+	struct nfp_predt_entry *flow;
+	bool is_ipv6;
+};
+
+/**
+ * struct nfp_predt_entry
+ * @list_head:		List head to attach to predt_list
+ * @flow_pay:		Direct link to flow_payload
+ * @nn_list:		List of linked nfp_neigh_entries
+ */
+struct nfp_predt_entry {
+	struct list_head list_head;
+	struct nfp_fl_payload *flow_pay;
+	struct list_head nn_list;
+};
+
 /**
  * struct nfp_mtu_conf - manage MTU setting
  * @portnum:		NFP port number of repr with requested MTU change
@@ -202,6 +277,9 @@ struct nfp_fl_internal_ports {
  * @ct_zone_table:	Hash table used to store the different zones
  * @ct_zone_wc:		Special zone entry for wildcarded zone matches
  * @ct_map_table:	Hash table used to referennce ct flows
+ * @predt_list:		List to keep track of decap pretun flows
+ * @neigh_table:	Table to keep track of neighbor entries
+ * @predt_lock:		Lock to serialise predt/neigh table updates
  */
 struct nfp_flower_priv {
 	struct nfp_app *app;
@@ -241,6 +319,9 @@ struct nfp_flower_priv {
 	struct rhashtable ct_zone_table;
 	struct nfp_fl_ct_zone_entry *ct_zone_wc;
 	struct rhashtable ct_map_table;
+	struct list_head predt_list;
+	struct rhashtable neigh_table;
+	spinlock_t predt_lock; /* Lock to serialise predt/neigh table updates */
 };
 
 /**
@@ -344,9 +425,14 @@ struct nfp_fl_payload {
 	struct list_head linked_flows;
 	bool in_hw;
 	struct {
+		struct nfp_predt_entry *predt;
 		struct net_device *dev;
+		__be16 vlan_tpid;
 		__be16 vlan_tci;
 		__be16 port_idx;
+		u8 loc_mac[ETH_ALEN];
+		u8 rem_mac[ETH_ALEN];
+		bool is_ipv6;
 	} pre_tun_rule;
 };
 
@@ -369,6 +455,7 @@ struct nfp_fl_payload_link {
 
 extern const struct rhashtable_params nfp_flower_table_params;
 extern const struct rhashtable_params merge_table_params;
+extern const struct rhashtable_params neigh_table_params;
 
 struct nfp_merge_info {
 	u64 parent_ctx;
diff --git a/drivers/net/ethernet/netronome/nfp/flower/metadata.c b/drivers/net/ethernet/netronome/nfp/flower/metadata.c
index f448c5682594..74e1b279c13b 100644
--- a/drivers/net/ethernet/netronome/nfp/flower/metadata.c
+++ b/drivers/net/ethernet/netronome/nfp/flower/metadata.c
@@ -502,6 +502,12 @@ const struct rhashtable_params nfp_ct_map_params = {
 	.automatic_shrinking	= true,
 };
 
+const struct rhashtable_params neigh_table_params = {
+	.key_offset	= offsetof(struct nfp_neigh_entry, neigh_cookie),
+	.head_offset	= offsetof(struct nfp_neigh_entry, ht_node),
+	.key_len	= sizeof(unsigned long),
+};
+
 int nfp_flower_metadata_init(struct nfp_app *app, u64 host_ctx_count,
 			     unsigned int host_num_mems)
 {
@@ -530,6 +536,12 @@ int nfp_flower_metadata_init(struct nfp_app *app, u64 host_ctx_count,
 	if (err)
 		goto err_free_ct_zone_table;
 
+	err = rhashtable_init(&priv->neigh_table, &neigh_table_params);
+	if (err)
+		goto err_free_ct_map_table;
+
+	INIT_LIST_HEAD(&priv->predt_list);
+
 	get_random_bytes(&priv->mask_id_seed, sizeof(priv->mask_id_seed));
 
 	/* Init ring buffer and unallocated mask_ids. */
@@ -537,7 +549,7 @@ int nfp_flower_metadata_init(struct nfp_app *app, u64 host_ctx_count,
 		kmalloc_array(NFP_FLOWER_MASK_ENTRY_RS,
 			      NFP_FLOWER_MASK_ELEMENT_RS, GFP_KERNEL);
 	if (!priv->mask_ids.mask_id_free_list.buf)
-		goto err_free_ct_map_table;
+		goto err_free_neigh_table;
 
 	priv->mask_ids.init_unallocated = NFP_FLOWER_MASK_ENTRY_RS - 1;
 
@@ -565,6 +577,7 @@ int nfp_flower_metadata_init(struct nfp_app *app, u64 host_ctx_count,
 		goto err_free_ring_buf;
 
 	spin_lock_init(&priv->stats_lock);
+	spin_lock_init(&priv->predt_lock);
 
 	return 0;
 
@@ -574,6 +587,8 @@ int nfp_flower_metadata_init(struct nfp_app *app, u64 host_ctx_count,
 	kfree(priv->mask_ids.last_used);
 err_free_mask_id:
 	kfree(priv->mask_ids.mask_id_free_list.buf);
+err_free_neigh_table:
+	rhashtable_destroy(&priv->neigh_table);
 err_free_ct_map_table:
 	rhashtable_destroy(&priv->ct_map_table);
 err_free_ct_zone_table:
@@ -700,6 +715,8 @@ void nfp_flower_metadata_cleanup(struct nfp_app *app)
 
 	rhashtable_free_and_destroy(&priv->ct_map_table,
 				    nfp_free_map_table_entry, NULL);
+	rhashtable_free_and_destroy(&priv->neigh_table,
+				    nfp_check_rhashtable_empty, NULL);
 	kvfree(priv->stats);
 	kfree(priv->mask_ids.mask_id_free_list.buf);
 	kfree(priv->mask_ids.last_used);
diff --git a/drivers/net/ethernet/netronome/nfp/flower/tunnel_conf.c b/drivers/net/ethernet/netronome/nfp/flower/tunnel_conf.c
index c71bd555f482..f5e8ed14e517 100644
--- a/drivers/net/ethernet/netronome/nfp/flower/tunnel_conf.c
+++ b/drivers/net/ethernet/netronome/nfp/flower/tunnel_conf.c
@@ -76,38 +76,6 @@ struct nfp_tun_active_tuns_v6 {
 	} tun_info[];
 };
 
-/**
- * struct nfp_tun_neigh - neighbour/route entry on the NFP
- * @dst_ipv4:	destination IPv4 address
- * @src_ipv4:	source IPv4 address
- * @dst_addr:	destination MAC address
- * @src_addr:	source MAC address
- * @port_id:	NFP port to output packet on - associated with source IPv4
- */
-struct nfp_tun_neigh {
-	__be32 dst_ipv4;
-	__be32 src_ipv4;
-	u8 dst_addr[ETH_ALEN];
-	u8 src_addr[ETH_ALEN];
-	__be32 port_id;
-};
-
-/**
- * struct nfp_tun_neigh_v6 - neighbour/route entry on the NFP
- * @dst_ipv6:	destination IPv6 address
- * @src_ipv6:	source IPv6 address
- * @dst_addr:	destination MAC address
- * @src_addr:	source MAC address
- * @port_id:	NFP port to output packet on - associated with source IPv6
- */
-struct nfp_tun_neigh_v6 {
-	struct in6_addr dst_ipv6;
-	struct in6_addr src_ipv6;
-	u8 dst_addr[ETH_ALEN];
-	u8 src_addr[ETH_ALEN];
-	__be32 port_id;
-};
-
 /**
  * struct nfp_tun_req_route_ipv4 - NFP requests a route/neighbour lookup
  * @ingress_port:	ingress port of packet that signalled request
-- 
2.30.2

[PATCH net-next 2/9] nfp: flower: add/remove predt_list entries

[Simon Horman]

From: Louis Peens <louis.peens@corigine.com>

Add calls to add and remove flows to the predt_table. This very simply
just allocates and add a new pretun entry if detected as such, and
removes it when encountered on a delete flow.

Compatibility for older firmware is kept in place through the
DECAP_V2 feature bit.

Signed-off-by: Louis Peens <louis.peens@corigine.com>
Signed-off-by: Yinjun Zhang <yinjun.zhang@corigine.com>
Signed-off-by: Simon Horman <simon.horman@corigine.com>
---
 .../ethernet/netronome/nfp/flower/action.c    |  3 +-
 .../ethernet/netronome/nfp/flower/offload.c   | 43 ++++++++++++++++---
 2 files changed, 39 insertions(+), 7 deletions(-)

diff --git a/drivers/net/ethernet/netronome/nfp/flower/action.c b/drivers/net/ethernet/netronome/nfp/flower/action.c
index 1b9421e844a9..0147de405365 100644
--- a/drivers/net/ethernet/netronome/nfp/flower/action.c
+++ b/drivers/net/ethernet/netronome/nfp/flower/action.c
@@ -220,7 +220,8 @@ nfp_fl_output(struct nfp_app *app, struct nfp_fl_output *output,
 		}
 		output->port = cpu_to_be32(NFP_FL_LAG_OUT | gid);
 	} else if (nfp_flower_internal_port_can_offload(app, out_dev)) {
-		if (!(priv->flower_ext_feats & NFP_FL_FEATS_PRE_TUN_RULES)) {
+		if (!(priv->flower_ext_feats & NFP_FL_FEATS_PRE_TUN_RULES) &&
+		    !(priv->flower_ext_feats & NFP_FL_FEATS_DECAP_V2)) {
 			NL_SET_ERR_MSG_MOD(extack, "unsupported offload: pre-tunnel rules not supported in loaded firmware");
 			return -EOPNOTSUPP;
 		}
diff --git a/drivers/net/ethernet/netronome/nfp/flower/offload.c b/drivers/net/ethernet/netronome/nfp/flower/offload.c
index 92e8ade4854e..0fe018bef410 100644
--- a/drivers/net/ethernet/netronome/nfp/flower/offload.c
+++ b/drivers/net/ethernet/netronome/nfp/flower/offload.c
@@ -1362,11 +1362,29 @@ nfp_flower_add_offload(struct nfp_app *app, struct net_device *netdev,
 		goto err_release_metadata;
 	}
 
-	if (flow_pay->pre_tun_rule.dev)
-		err = nfp_flower_xmit_pre_tun_flow(app, flow_pay);
-	else
+	if (flow_pay->pre_tun_rule.dev) {
+		if (priv->flower_ext_feats & NFP_FL_FEATS_DECAP_V2) {
+			struct nfp_predt_entry *predt;
+
+			predt = kzalloc(sizeof(*predt), GFP_KERNEL);
+			if (!predt) {
+				err = -ENOMEM;
+				goto err_remove_rhash;
+			}
+			predt->flow_pay = flow_pay;
+			INIT_LIST_HEAD(&predt->nn_list);
+			spin_lock_bh(&priv->predt_lock);
+			list_add(&predt->list_head, &priv->predt_list);
+			spin_unlock_bh(&priv->predt_lock);
+			flow_pay->pre_tun_rule.predt = predt;
+		} else {
+			err = nfp_flower_xmit_pre_tun_flow(app, flow_pay);
+		}
+	} else {
 		err = nfp_flower_xmit_flow(app, flow_pay,
 					   NFP_FLOWER_CMSG_TYPE_FLOW_ADD);
+	}
+
 	if (err)
 		goto err_remove_rhash;
 
@@ -1538,11 +1556,24 @@ nfp_flower_del_offload(struct nfp_app *app, struct net_device *netdev,
 		goto err_free_merge_flow;
 	}
 
-	if (nfp_flow->pre_tun_rule.dev)
-		err = nfp_flower_xmit_pre_tun_del_flow(app, nfp_flow);
-	else
+	if (nfp_flow->pre_tun_rule.dev) {
+		if (priv->flower_ext_feats & NFP_FL_FEATS_DECAP_V2) {
+			struct nfp_predt_entry *predt;
+
+			predt = nfp_flow->pre_tun_rule.predt;
+			if (predt) {
+				spin_lock_bh(&priv->predt_lock);
+				list_del(&predt->list_head);
+				spin_unlock_bh(&priv->predt_lock);
+				kfree(predt);
+			}
+		} else {
+			err = nfp_flower_xmit_pre_tun_del_flow(app, nfp_flow);
+		}
+	} else {
 		err = nfp_flower_xmit_flow(app, nfp_flow,
 					   NFP_FLOWER_CMSG_TYPE_FLOW_DEL);
+	}
 	/* Fall through on error. */
 
 err_free_merge_flow:
-- 
2.30.2

[PATCH net-next 3/9] nfp: flower: enforce more strict pre_tun checks

[Simon Horman]

From: Louis Peens <louis.peens@corigine.com>

Make sure that the rule also matches on source MAC address. On top
of that also now save the src and dst MAC addresses similar to how
vlan_tci is saved - this will be used in later comparisons with
neighbour entries. Indicate if the flow matched on ipv4 or ipv6.
Populate the vlan_tpid field that got added to the pre_run_rule
struct as well.

Signed-off-by: Louis Peens <louis.peens@corigine.com>
Signed-off-by: Yinjun Zhang <yinjun.zhang@corigine.com>
Signed-off-by: Simon Horman <simon.horman@corigine.com>
---
 .../ethernet/netronome/nfp/flower/offload.c   | 41 +++++++++++++++----
 1 file changed, 34 insertions(+), 7 deletions(-)

diff --git a/drivers/net/ethernet/netronome/nfp/flower/offload.c b/drivers/net/ethernet/netronome/nfp/flower/offload.c
index 0fe018bef410..5fea3e3415fe 100644
--- a/drivers/net/ethernet/netronome/nfp/flower/offload.c
+++ b/drivers/net/ethernet/netronome/nfp/flower/offload.c
@@ -1170,6 +1170,11 @@ nfp_flower_validate_pre_tun_rule(struct nfp_app *app,
 		return -EOPNOTSUPP;
 	}
 
+	if (key_layer & NFP_FLOWER_LAYER_IPV6)
+		flow->pre_tun_rule.is_ipv6 = true;
+	else
+		flow->pre_tun_rule.is_ipv6 = false;
+
 	/* Skip fields known to exist. */
 	mask += sizeof(struct nfp_flower_meta_tci);
 	ext += sizeof(struct nfp_flower_meta_tci);
@@ -1180,13 +1185,6 @@ nfp_flower_validate_pre_tun_rule(struct nfp_app *app,
 	mask += sizeof(struct nfp_flower_in_port);
 	ext += sizeof(struct nfp_flower_in_port);
 
-	/* Ensure destination MAC address matches pre_tun_dev. */
-	mac = (struct nfp_flower_mac_mpls *)ext;
-	if (memcmp(&mac->mac_dst[0], flow->pre_tun_rule.dev->dev_addr, 6)) {
-		NL_SET_ERR_MSG_MOD(extack, "unsupported pre-tunnel rule: dest MAC must match output dev MAC");
-		return -EOPNOTSUPP;
-	}
-
 	/* Ensure destination MAC address is fully matched. */
 	mac = (struct nfp_flower_mac_mpls *)mask;
 	if (!is_broadcast_ether_addr(&mac->mac_dst[0])) {
@@ -1194,11 +1192,36 @@ nfp_flower_validate_pre_tun_rule(struct nfp_app *app,
 		return -EOPNOTSUPP;
 	}
 
+	/* Ensure source MAC address is fully matched. This is only needed
+	 * for firmware with the DECAP_V2 feature enabled. Don't do this
+	 * for firmware without this feature to keep old behaviour.
+	 */
+	if (priv->flower_ext_feats & NFP_FL_FEATS_DECAP_V2) {
+		mac = (struct nfp_flower_mac_mpls *)mask;
+		if (!is_broadcast_ether_addr(&mac->mac_src[0])) {
+			NL_SET_ERR_MSG_MOD(extack,
+					   "unsupported pre-tunnel rule: source MAC field must not be masked");
+			return -EOPNOTSUPP;
+		}
+	}
+
 	if (mac->mpls_lse) {
 		NL_SET_ERR_MSG_MOD(extack, "unsupported pre-tunnel rule: MPLS not supported");
 		return -EOPNOTSUPP;
 	}
 
+	/* Ensure destination MAC address matches pre_tun_dev. */
+	mac = (struct nfp_flower_mac_mpls *)ext;
+	if (memcmp(&mac->mac_dst[0], flow->pre_tun_rule.dev->dev_addr, 6)) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "unsupported pre-tunnel rule: dest MAC must match output dev MAC");
+		return -EOPNOTSUPP;
+	}
+
+	/* Save mac addresses in pre_tun_rule entry for later use */
+	memcpy(&flow->pre_tun_rule.loc_mac, &mac->mac_dst[0], ETH_ALEN);
+	memcpy(&flow->pre_tun_rule.rem_mac, &mac->mac_src[0], ETH_ALEN);
+
 	mask += sizeof(struct nfp_flower_mac_mpls);
 	ext += sizeof(struct nfp_flower_mac_mpls);
 	if (key_layer & NFP_FLOWER_LAYER_IPV4 ||
@@ -1227,17 +1250,21 @@ nfp_flower_validate_pre_tun_rule(struct nfp_app *app,
 	if ((priv->flower_ext_feats & NFP_FL_FEATS_VLAN_QINQ)) {
 		if (key_ls->key_layer_two & NFP_FLOWER_LAYER2_QINQ) {
 			struct nfp_flower_vlan *vlan_tags;
+			u16 vlan_tpid;
 			u16 vlan_tci;
 
 			vlan_tags = (struct nfp_flower_vlan *)ext;
 
 			vlan_tci = be16_to_cpu(vlan_tags->outer_tci);
+			vlan_tpid = be16_to_cpu(vlan_tags->outer_tpid);
 
 			vlan_tci &= ~NFP_FLOWER_MASK_VLAN_PRESENT;
 			flow->pre_tun_rule.vlan_tci = cpu_to_be16(vlan_tci);
+			flow->pre_tun_rule.vlan_tpid = cpu_to_be16(vlan_tpid);
 			vlan = true;
 		} else {
 			flow->pre_tun_rule.vlan_tci = cpu_to_be16(0xffff);
+			flow->pre_tun_rule.vlan_tpid = cpu_to_be16(0xffff);
 		}
 	}
 
-- 
2.30.2

[PATCH net-next 4/9] nfp: flower: fixup ipv6/ipv4 route lookup for neigh events

[Simon Horman]

From: Louis Peens <louis.peens@corigine.com>

When a callback is received to invalidate a neighbour entry
there is no need to try and populate any other flow information.
Only the flowX->daddr information is needed as lookup key to delete
an entry from the NFP neighbour table. Fix this by only doing the
lookup if the callback is for a new entry.

As part of this cleanup remove the setting of flow6.flowi6_proto, as
this is not needed either, it looks to be a possible leftover from a
previous implementation.

Signed-off-by: Louis Peens <louis.peens@corigine.com>
Signed-off-by: Yinjun Zhang <yinjun.zhang@corigine.com>
Signed-off-by: Simon Horman <simon.horman@corigine.com>
---
 .../netronome/nfp/flower/tunnel_conf.c        | 52 ++++++++++++-------
 1 file changed, 33 insertions(+), 19 deletions(-)

diff --git a/drivers/net/ethernet/netronome/nfp/flower/tunnel_conf.c b/drivers/net/ethernet/netronome/nfp/flower/tunnel_conf.c
index f5e8ed14e517..0cb016afbab3 100644
--- a/drivers/net/ethernet/netronome/nfp/flower/tunnel_conf.c
+++ b/drivers/net/ethernet/netronome/nfp/flower/tunnel_conf.c
@@ -494,7 +494,7 @@ nfp_tun_neigh_event_handler(struct notifier_block *nb, unsigned long event,
 	struct flowi6 flow6 = {};
 	struct neighbour *n;
 	struct nfp_app *app;
-	struct rtable *rt;
+	bool neigh_invalid;
 	bool ipv6 = false;
 	int err;
 
@@ -513,6 +513,8 @@ nfp_tun_neigh_event_handler(struct notifier_block *nb, unsigned long event,
 	if (n->tbl->family == AF_INET6)
 		ipv6 = true;
 
+	neigh_invalid = !(n->nud_state & NUD_VALID) || n->dead;
+
 	if (ipv6)
 		flow6.daddr = *(struct in6_addr *)n->primary_key;
 	else
@@ -533,29 +535,41 @@ nfp_tun_neigh_event_handler(struct notifier_block *nb, unsigned long event,
 #if IS_ENABLED(CONFIG_INET)
 	if (ipv6) {
 #if IS_ENABLED(CONFIG_IPV6)
-		struct dst_entry *dst;
-
-		dst = ipv6_stub->ipv6_dst_lookup_flow(dev_net(n->dev), NULL,
-						      &flow6, NULL);
-		if (IS_ERR(dst))
-			return NOTIFY_DONE;
-
-		dst_release(dst);
-		flow6.flowi6_proto = IPPROTO_UDP;
+		if (!neigh_invalid) {
+			struct dst_entry *dst;
+			/* Use ipv6_dst_lookup_flow to populate flow6->saddr
+			 * and other fields. This information is only needed
+			 * for new entries, lookup can be skipped when an entry
+			 * gets invalidated - as only the daddr is needed for
+			 * deleting.
+			 */
+			dst = ip6_dst_lookup_flow(dev_net(n->dev), NULL,
+						  &flow6, NULL);
+			if (IS_ERR(dst))
+				return NOTIFY_DONE;
+
+			dst_release(dst);
+		}
 		nfp_tun_write_neigh_v6(n->dev, app, &flow6, n, GFP_ATOMIC);
 #else
 		return NOTIFY_DONE;
 #endif /* CONFIG_IPV6 */
 	} else {
-		/* Do a route lookup to populate flow data. */
-		rt = ip_route_output_key(dev_net(n->dev), &flow4);
-		err = PTR_ERR_OR_ZERO(rt);
-		if (err)
-			return NOTIFY_DONE;
-
-		ip_rt_put(rt);
-
-		flow4.flowi4_proto = IPPROTO_UDP;
+		if (!neigh_invalid) {
+			struct rtable *rt;
+			/* Use ip_route_output_key to populate flow4->saddr and
+			 * other fields. This information is only needed for
+			 * new entries, lookup can be skipped when an entry
+			 * gets invalidated - as only the daddr is needed for
+			 * deleting.
+			 */
+			rt = ip_route_output_key(dev_net(n->dev), &flow4);
+			err = PTR_ERR_OR_ZERO(rt);
+			if (err)
+				return NOTIFY_DONE;
+
+			ip_rt_put(rt);
+		}
 		nfp_tun_write_neigh_v4(n->dev, app, &flow4, n, GFP_ATOMIC);
 	}
 #else
-- 
2.30.2

[PATCH net-next 5/9] nfp: flower: update nfp_tun_neigh structs

[Simon Horman]

From: Louis Peens <louis.peens@corigine.com>

Prepare for more rework in following patches by updating
the existing nfp_neigh_structs. The update allows for
the same headers to be used for both old and new firmware,
with a slight length adjustment when sending the control message
to the firmware.

Signed-off-by: Louis Peens <louis.peens@corigine.com>
Signed-off-by: Yinjun Zhang <yinjun.zhang@corigine.com>
Signed-off-by: Simon Horman <simon.horman@corigine.com>
---
 .../net/ethernet/netronome/nfp/flower/main.h  | 50 +++++++++++--------
 .../netronome/nfp/flower/tunnel_conf.c        | 30 ++++++-----
 2 files changed, 47 insertions(+), 33 deletions(-)

diff --git a/drivers/net/ethernet/netronome/nfp/flower/main.h b/drivers/net/ethernet/netronome/nfp/flower/main.h
index 454fdb6ea4a5..2c011d60c212 100644
--- a/drivers/net/ethernet/netronome/nfp/flower/main.h
+++ b/drivers/net/ethernet/netronome/nfp/flower/main.h
@@ -111,47 +111,55 @@ struct nfp_fl_tunnel_offloads {
 };
 
 /**
- * struct nfp_tun_neigh - neighbour/route entry on the NFP
- * @dst_ipv4:	Destination IPv4 address
- * @src_ipv4:	Source IPv4 address
+ * struct nfp_tun_neigh - basic neighbour data
  * @dst_addr:	Destination MAC address
  * @src_addr:	Source MAC address
  * @port_id:	NFP port to output packet on - associated with source IPv4
- * @vlan_tpid:	VLAN_TPID match field
- * @vlan_tci:	VLAN_TCI match field
- * @host_ctx:	Host context ID to be saved here
  */
 struct nfp_tun_neigh {
-	__be32 dst_ipv4;
-	__be32 src_ipv4;
 	u8 dst_addr[ETH_ALEN];
 	u8 src_addr[ETH_ALEN];
 	__be32 port_id;
+};
+
+/**
+ * struct nfp_tun_neigh_ext - extended neighbour data
+ * @vlan_tpid:	VLAN_TPID match field
+ * @vlan_tci:	VLAN_TCI match field
+ * @host_ctx:	Host context ID to be saved here
+ */
+struct nfp_tun_neigh_ext {
 	__be16 vlan_tpid;
 	__be16 vlan_tci;
 	__be32 host_ctx;
 };
 
 /**
- * struct nfp_tun_neigh_v6 - neighbour/route entry on the NFP
+ * struct nfp_tun_neigh_v4 - neighbour/route entry on the NFP for IPv4
+ * @dst_ipv4:	Destination IPv4 address
+ * @src_ipv4:	Source IPv4 address
+ * @common:	Neighbour/route common info
+ * @ext:	Neighbour/route extended info
+ */
+struct nfp_tun_neigh_v4 {
+	__be32 dst_ipv4;
+	__be32 src_ipv4;
+	struct nfp_tun_neigh common;
+	struct nfp_tun_neigh_ext ext;
+};
+
+/**
+ * struct nfp_tun_neigh_v6 - neighbour/route entry on the NFP for IPv6
  * @dst_ipv6:	Destination IPv6 address
  * @src_ipv6:	Source IPv6 address
- * @dst_addr:	Destination MAC address
- * @src_addr:	Source MAC address
- * @port_id:	NFP port to output packet on - associated with source IPv6
- * @vlan_tpid:	VLAN_TPID match field
- * @vlan_tci:	VLAN_TCI match field
- * @host_ctx:	Host context ID to be saved here
+ * @common:	Neighbour/route common info
+ * @ext:	Neighbour/route extended info
  */
 struct nfp_tun_neigh_v6 {
 	struct in6_addr dst_ipv6;
 	struct in6_addr src_ipv6;
-	u8 dst_addr[ETH_ALEN];
-	u8 src_addr[ETH_ALEN];
-	__be32 port_id;
-	__be16 vlan_tpid;
-	__be16 vlan_tci;
-	__be32 host_ctx;
+	struct nfp_tun_neigh common;
+	struct nfp_tun_neigh_ext ext;
 };
 
 /**
diff --git a/drivers/net/ethernet/netronome/nfp/flower/tunnel_conf.c b/drivers/net/ethernet/netronome/nfp/flower/tunnel_conf.c
index 0cb016afbab3..174888272a30 100644
--- a/drivers/net/ethernet/netronome/nfp/flower/tunnel_conf.c
+++ b/drivers/net/ethernet/netronome/nfp/flower/tunnel_conf.c
@@ -281,9 +281,15 @@ static int
 nfp_flower_xmit_tun_conf(struct nfp_app *app, u8 mtype, u16 plen, void *pdata,
 			 gfp_t flag)
 {
+	struct nfp_flower_priv *priv = app->priv;
 	struct sk_buff *skb;
 	unsigned char *msg;
 
+	if (!(priv->flower_ext_feats & NFP_FL_FEATS_DECAP_V2) &&
+	    (mtype == NFP_FLOWER_CMSG_TYPE_TUN_NEIGH ||
+	     mtype == NFP_FLOWER_CMSG_TYPE_TUN_NEIGH_V6))
+		plen -= sizeof(struct nfp_tun_neigh_ext);
+
 	skb = nfp_flower_cmsg_alloc(app, plen, mtype, flag);
 	if (!skb)
 		return -ENOMEM;
@@ -416,14 +422,14 @@ static void
 nfp_tun_write_neigh_v4(struct net_device *netdev, struct nfp_app *app,
 		       struct flowi4 *flow, struct neighbour *neigh, gfp_t flag)
 {
-	struct nfp_tun_neigh payload;
+	struct nfp_tun_neigh_v4 payload;
 	u32 port_id;
 
 	port_id = nfp_flower_get_port_id_from_netdev(app, netdev);
 	if (!port_id)
 		return;
 
-	memset(&payload, 0, sizeof(struct nfp_tun_neigh));
+	memset(&payload, 0, sizeof(struct nfp_tun_neigh_v4));
 	payload.dst_ipv4 = flow->daddr;
 
 	/* If entry has expired send dst IP with all other fields 0. */
@@ -436,15 +442,15 @@ nfp_tun_write_neigh_v4(struct net_device *netdev, struct nfp_app *app,
 
 	/* Have a valid neighbour so populate rest of entry. */
 	payload.src_ipv4 = flow->saddr;
-	ether_addr_copy(payload.src_addr, netdev->dev_addr);
-	neigh_ha_snapshot(payload.dst_addr, neigh, netdev);
-	payload.port_id = cpu_to_be32(port_id);
+	ether_addr_copy(payload.common.src_addr, netdev->dev_addr);
+	neigh_ha_snapshot(payload.common.dst_addr, neigh, netdev);
+	payload.common.port_id = cpu_to_be32(port_id);
 	/* Add destination of new route to NFP cache. */
 	nfp_tun_add_route_to_cache_v4(app, &payload.dst_ipv4);
 
 send_msg:
 	nfp_flower_xmit_tun_conf(app, NFP_FLOWER_CMSG_TYPE_TUN_NEIGH,
-				 sizeof(struct nfp_tun_neigh),
+				 sizeof(struct nfp_tun_neigh_v4),
 				 (unsigned char *)&payload, flag);
 }
 
@@ -472,9 +478,9 @@ nfp_tun_write_neigh_v6(struct net_device *netdev, struct nfp_app *app,
 
 	/* Have a valid neighbour so populate rest of entry. */
 	payload.src_ipv6 = flow->saddr;
-	ether_addr_copy(payload.src_addr, netdev->dev_addr);
-	neigh_ha_snapshot(payload.dst_addr, neigh, netdev);
-	payload.port_id = cpu_to_be32(port_id);
+	ether_addr_copy(payload.common.src_addr, netdev->dev_addr);
+	neigh_ha_snapshot(payload.common.dst_addr, neigh, netdev);
+	payload.common.port_id = cpu_to_be32(port_id);
 	/* Add destination of new route to NFP cache. */
 	nfp_tun_add_route_to_cache_v6(app, &payload.dst_ipv6);
 
@@ -1372,7 +1378,7 @@ void nfp_tunnel_config_stop(struct nfp_app *app)
 	struct nfp_flower_priv *priv = app->priv;
 	struct nfp_ipv4_addr_entry *ip_entry;
 	struct nfp_tun_neigh_v6 ipv6_route;
-	struct nfp_tun_neigh ipv4_route;
+	struct nfp_tun_neigh_v4 ipv4_route;
 	struct list_head *ptr, *storage;
 
 	unregister_netevent_notifier(&priv->tun.neigh_nb);
@@ -1398,7 +1404,7 @@ void nfp_tunnel_config_stop(struct nfp_app *app)
 		kfree(route_entry);
 
 		nfp_flower_xmit_tun_conf(app, NFP_FLOWER_CMSG_TYPE_TUN_NEIGH,
-					 sizeof(struct nfp_tun_neigh),
+					 sizeof(struct nfp_tun_neigh_v4),
 					 (unsigned char *)&ipv4_route,
 					 GFP_KERNEL);
 	}
@@ -1412,7 +1418,7 @@ void nfp_tunnel_config_stop(struct nfp_app *app)
 		kfree(route_entry);
 
 		nfp_flower_xmit_tun_conf(app, NFP_FLOWER_CMSG_TYPE_TUN_NEIGH_V6,
-					 sizeof(struct nfp_tun_neigh),
+					 sizeof(struct nfp_tun_neigh_v6),
 					 (unsigned char *)&ipv6_route,
 					 GFP_KERNEL);
 	}
-- 
2.30.2

[PATCH net-next 6/9] nfp: flower: rework tunnel neighbour configuration

[Simon Horman]

From: Louis Peens <louis.peens@corigine.com>

This patch updates the way in which the tunnel neighbour entries
are handled. Previously they were mostly send-and-forget, with
just the destination IP's cached in a list. This update changes
to a scheme where the neighbour entry information is stored in
a hash table.

The reason for this is that the neighbour table will now also
be used on the decapsulation path, whereas previously it was
only used for encapsulation. We need to save more of the neighbour
information in order to link them with flower flows in follow
up patches.

Updating of the neighbour table is now also handled by the same
function, instead of separate  *_write_neigh_vX functions.

Signed-off-by: Louis Peens <louis.peens@corigine.com>
Signed-off-by: Yinjun Zhang <yinjun.zhang@corigine.com>
Signed-off-by: Simon Horman <simon.horman@corigine.com>
---
 .../netronome/nfp/flower/tunnel_conf.c        | 174 ++++++++++++------
 1 file changed, 115 insertions(+), 59 deletions(-)

diff --git a/drivers/net/ethernet/netronome/nfp/flower/tunnel_conf.c b/drivers/net/ethernet/netronome/nfp/flower/tunnel_conf.c
index 174888272a30..4ca149501409 100644
--- a/drivers/net/ethernet/netronome/nfp/flower/tunnel_conf.c
+++ b/drivers/net/ethernet/netronome/nfp/flower/tunnel_conf.c
@@ -419,75 +419,131 @@ nfp_tun_del_route_from_cache_v6(struct nfp_app *app, struct in6_addr *ipv6_addr)
 }
 
 static void
-nfp_tun_write_neigh_v4(struct net_device *netdev, struct nfp_app *app,
-		       struct flowi4 *flow, struct neighbour *neigh, gfp_t flag)
+nfp_tun_write_neigh(struct net_device *netdev, struct nfp_app *app,
+		    void *flow, struct neighbour *neigh, bool is_ipv6)
 {
-	struct nfp_tun_neigh_v4 payload;
+	bool neigh_invalid = !(neigh->nud_state & NUD_VALID) || neigh->dead;
+	size_t neigh_size = is_ipv6 ? sizeof(struct nfp_tun_neigh_v6) :
+			    sizeof(struct nfp_tun_neigh_v4);
+	unsigned long cookie = (unsigned long)neigh;
+	struct nfp_flower_priv *priv = app->priv;
+	struct nfp_neigh_entry *nn_entry;
 	u32 port_id;
+	u8 mtype;
 
 	port_id = nfp_flower_get_port_id_from_netdev(app, netdev);
 	if (!port_id)
 		return;
 
-	memset(&payload, 0, sizeof(struct nfp_tun_neigh_v4));
-	payload.dst_ipv4 = flow->daddr;
+	spin_lock_bh(&priv->predt_lock);
+	nn_entry = rhashtable_lookup_fast(&priv->neigh_table, &cookie,
+					  neigh_table_params);
+	if (!nn_entry && !neigh_invalid) {
+		struct nfp_tun_neigh_ext *ext;
+		struct nfp_tun_neigh *common;
+
+		nn_entry = kzalloc(sizeof(*nn_entry) + neigh_size,
+				   GFP_ATOMIC);
+		if (!nn_entry)
+			goto err;
+
+		nn_entry->payload = (char *)&nn_entry[1];
+		nn_entry->neigh_cookie = cookie;
+		nn_entry->is_ipv6 = is_ipv6;
+		nn_entry->flow = NULL;
+		if (is_ipv6) {
+			struct flowi6 *flowi6 = (struct flowi6 *)flow;
+			struct nfp_tun_neigh_v6 *payload;
+
+			payload = (struct nfp_tun_neigh_v6 *)nn_entry->payload;
+			payload->src_ipv6 = flowi6->saddr;
+			payload->dst_ipv6 = flowi6->daddr;
+			common = &payload->common;
+			ext = &payload->ext;
+			mtype = NFP_FLOWER_CMSG_TYPE_TUN_NEIGH_V6;
+		} else {
+			struct flowi4 *flowi4 = (struct flowi4 *)flow;
+			struct nfp_tun_neigh_v4 *payload;
+
+			payload = (struct nfp_tun_neigh_v4 *)nn_entry->payload;
+			payload->src_ipv4 = flowi4->saddr;
+			payload->dst_ipv4 = flowi4->daddr;
+			common = &payload->common;
+			ext = &payload->ext;
+			mtype = NFP_FLOWER_CMSG_TYPE_TUN_NEIGH;
+		}
+		ext->host_ctx = cpu_to_be32(U32_MAX);
+		ext->vlan_tpid = cpu_to_be16(U16_MAX);
+		ext->vlan_tci = cpu_to_be16(U16_MAX);
+		ether_addr_copy(common->src_addr, netdev->dev_addr);
+		neigh_ha_snapshot(common->dst_addr, neigh, netdev);
+		common->port_id = cpu_to_be32(port_id);
+
+		if (rhashtable_insert_fast(&priv->neigh_table,
+					   &nn_entry->ht_node,
+					   neigh_table_params))
+			goto err;
+
+		/* Add entries to the relevant route cache */
+		if (is_ipv6) {
+			struct nfp_tun_neigh_v6 *payload;
+
+			payload = (struct nfp_tun_neigh_v6 *)nn_entry->payload;
+			nfp_tun_add_route_to_cache_v6(app, &payload->dst_ipv6);
+		} else {
+			struct nfp_tun_neigh_v4 *payload;
+
+			payload = (struct nfp_tun_neigh_v4 *)nn_entry->payload;
+			nfp_tun_add_route_to_cache_v4(app, &payload->dst_ipv4);
+		}
 
-	/* If entry has expired send dst IP with all other fields 0. */
-	if (!(neigh->nud_state & NUD_VALID) || neigh->dead) {
-		nfp_tun_del_route_from_cache_v4(app, &payload.dst_ipv4);
+		nfp_flower_xmit_tun_conf(app, mtype, neigh_size,
+					 nn_entry->payload,
+					 GFP_ATOMIC);
+	} else if (nn_entry && neigh_invalid) {
+		if (is_ipv6) {
+			struct flowi6 *flowi6 = (struct flowi6 *)flow;
+			struct nfp_tun_neigh_v6 *payload;
+
+			payload = (struct nfp_tun_neigh_v6 *)nn_entry->payload;
+			memset(payload, 0, sizeof(struct nfp_tun_neigh_v6));
+			payload->dst_ipv6 = flowi6->daddr;
+			mtype = NFP_FLOWER_CMSG_TYPE_TUN_NEIGH_V6;
+			nfp_tun_del_route_from_cache_v6(app,
+							&payload->dst_ipv6);
+		} else {
+			struct flowi4 *flowi4 = (struct flowi4 *)flow;
+			struct nfp_tun_neigh_v4 *payload;
+
+			payload = (struct nfp_tun_neigh_v4 *)nn_entry->payload;
+			memset(payload, 0, sizeof(struct nfp_tun_neigh_v4));
+			payload->dst_ipv4 = flowi4->daddr;
+			mtype = NFP_FLOWER_CMSG_TYPE_TUN_NEIGH;
+			nfp_tun_del_route_from_cache_v4(app,
+							&payload->dst_ipv4);
+		}
 		/* Trigger ARP to verify invalid neighbour state. */
 		neigh_event_send(neigh, NULL);
-		goto send_msg;
-	}
-
-	/* Have a valid neighbour so populate rest of entry. */
-	payload.src_ipv4 = flow->saddr;
-	ether_addr_copy(payload.common.src_addr, netdev->dev_addr);
-	neigh_ha_snapshot(payload.common.dst_addr, neigh, netdev);
-	payload.common.port_id = cpu_to_be32(port_id);
-	/* Add destination of new route to NFP cache. */
-	nfp_tun_add_route_to_cache_v4(app, &payload.dst_ipv4);
-
-send_msg:
-	nfp_flower_xmit_tun_conf(app, NFP_FLOWER_CMSG_TYPE_TUN_NEIGH,
-				 sizeof(struct nfp_tun_neigh_v4),
-				 (unsigned char *)&payload, flag);
-}
+		rhashtable_remove_fast(&priv->neigh_table,
+				       &nn_entry->ht_node,
+				       neigh_table_params);
 
-static void
-nfp_tun_write_neigh_v6(struct net_device *netdev, struct nfp_app *app,
-		       struct flowi6 *flow, struct neighbour *neigh, gfp_t flag)
-{
-	struct nfp_tun_neigh_v6 payload;
-	u32 port_id;
+		nfp_flower_xmit_tun_conf(app, mtype, neigh_size,
+					 nn_entry->payload,
+					 GFP_ATOMIC);
 
-	port_id = nfp_flower_get_port_id_from_netdev(app, netdev);
-	if (!port_id)
-		return;
-
-	memset(&payload, 0, sizeof(struct nfp_tun_neigh_v6));
-	payload.dst_ipv6 = flow->daddr;
-
-	/* If entry has expired send dst IP with all other fields 0. */
-	if (!(neigh->nud_state & NUD_VALID) || neigh->dead) {
-		nfp_tun_del_route_from_cache_v6(app, &payload.dst_ipv6);
-		/* Trigger probe to verify invalid neighbour state. */
-		neigh_event_send(neigh, NULL);
-		goto send_msg;
+		if (nn_entry->flow)
+			list_del(&nn_entry->list_head);
+		kfree(nn_entry);
 	}
 
-	/* Have a valid neighbour so populate rest of entry. */
-	payload.src_ipv6 = flow->saddr;
-	ether_addr_copy(payload.common.src_addr, netdev->dev_addr);
-	neigh_ha_snapshot(payload.common.dst_addr, neigh, netdev);
-	payload.common.port_id = cpu_to_be32(port_id);
-	/* Add destination of new route to NFP cache. */
-	nfp_tun_add_route_to_cache_v6(app, &payload.dst_ipv6);
-
-send_msg:
-	nfp_flower_xmit_tun_conf(app, NFP_FLOWER_CMSG_TYPE_TUN_NEIGH_V6,
-				 sizeof(struct nfp_tun_neigh_v6),
-				 (unsigned char *)&payload, flag);
+	spin_unlock_bh(&priv->predt_lock);
+	return;
+
+err:
+	kfree(nn_entry);
+	spin_unlock_bh(&priv->predt_lock);
+	nfp_flower_cmsg_warn(app, "Neighbour configuration failed.\n");
 }
 
 static int
@@ -556,7 +612,7 @@ nfp_tun_neigh_event_handler(struct notifier_block *nb, unsigned long event,
 
 			dst_release(dst);
 		}
-		nfp_tun_write_neigh_v6(n->dev, app, &flow6, n, GFP_ATOMIC);
+		nfp_tun_write_neigh(n->dev, app, &flow6, n, true);
 #else
 		return NOTIFY_DONE;
 #endif /* CONFIG_IPV6 */
@@ -576,7 +632,7 @@ nfp_tun_neigh_event_handler(struct notifier_block *nb, unsigned long event,
 
 			ip_rt_put(rt);
 		}
-		nfp_tun_write_neigh_v4(n->dev, app, &flow4, n, GFP_ATOMIC);
+		nfp_tun_write_neigh(n->dev, app, &flow4, n, false);
 	}
 #else
 	return NOTIFY_DONE;
@@ -619,7 +675,7 @@ void nfp_tunnel_request_route_v4(struct nfp_app *app, struct sk_buff *skb)
 	ip_rt_put(rt);
 	if (!n)
 		goto fail_rcu_unlock;
-	nfp_tun_write_neigh_v4(n->dev, app, &flow, n, GFP_ATOMIC);
+	nfp_tun_write_neigh(n->dev, app, &flow, n, false);
 	neigh_release(n);
 	rcu_read_unlock();
 	return;
@@ -661,7 +717,7 @@ void nfp_tunnel_request_route_v6(struct nfp_app *app, struct sk_buff *skb)
 	if (!n)
 		goto fail_rcu_unlock;
 
-	nfp_tun_write_neigh_v6(n->dev, app, &flow, n, GFP_ATOMIC);
+	nfp_tun_write_neigh(n->dev, app, &flow, n, true);
 	neigh_release(n);
 	rcu_read_unlock();
 	return;
-- 
2.30.2

[PATCH net-next 7/9] nfp: flower: link pre_tun flow rules with neigh entries

[Simon Horman]

From: Louis Peens <louis.peens@corigine.com>

Add helper functions that can create links between flow rules
and cached neighbour entries. Also add the relevant calls to
these functions.

* When a new neighbour entry gets added cycle through the saved
  pre_tun flow list and link any relevant matches. Update the
  neighbour table on the nfp with this new information.
* When a new pre_tun flow rule gets added iterate through the
  save neighbour entries and link any relevant matches. Once
  again update the nfp neighbour table with any new links.
* Do the inverse when deleting - remove any created links and
  also inform the nfp of this.

Signed-off-by: Louis Peens <louis.peens@corigine.com>
Signed-off-by: Yinjun Zhang <yinjun.zhang@corigine.com>
Signed-off-by: Simon Horman <simon.horman@corigine.com>
---
 .../net/ethernet/netronome/nfp/flower/main.h  |   4 +
 .../ethernet/netronome/nfp/flower/offload.c   |   4 +-
 .../netronome/nfp/flower/tunnel_conf.c        | 147 ++++++++++++++++++
 3 files changed, 154 insertions(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/netronome/nfp/flower/main.h b/drivers/net/ethernet/netronome/nfp/flower/main.h
index 2c011d60c212..6bc7a9cbf131 100644
--- a/drivers/net/ethernet/netronome/nfp/flower/main.h
+++ b/drivers/net/ethernet/netronome/nfp/flower/main.h
@@ -675,6 +675,10 @@ void
 nfp_flower_non_repr_priv_put(struct nfp_app *app, struct net_device *netdev);
 u32 nfp_flower_get_port_id_from_netdev(struct nfp_app *app,
 				       struct net_device *netdev);
+void nfp_tun_link_and_update_nn_entries(struct nfp_app *app,
+					struct nfp_predt_entry *predt);
+void nfp_tun_unlink_and_update_nn_entries(struct nfp_app *app,
+					  struct nfp_predt_entry *predt);
 int nfp_flower_xmit_pre_tun_flow(struct nfp_app *app,
 				 struct nfp_fl_payload *flow);
 int nfp_flower_xmit_pre_tun_del_flow(struct nfp_app *app,
diff --git a/drivers/net/ethernet/netronome/nfp/flower/offload.c b/drivers/net/ethernet/netronome/nfp/flower/offload.c
index 5fea3e3415fe..9d65459bdba5 100644
--- a/drivers/net/ethernet/netronome/nfp/flower/offload.c
+++ b/drivers/net/ethernet/netronome/nfp/flower/offload.c
@@ -1402,8 +1402,9 @@ nfp_flower_add_offload(struct nfp_app *app, struct net_device *netdev,
 			INIT_LIST_HEAD(&predt->nn_list);
 			spin_lock_bh(&priv->predt_lock);
 			list_add(&predt->list_head, &priv->predt_list);
-			spin_unlock_bh(&priv->predt_lock);
 			flow_pay->pre_tun_rule.predt = predt;
+			nfp_tun_link_and_update_nn_entries(app, predt);
+			spin_unlock_bh(&priv->predt_lock);
 		} else {
 			err = nfp_flower_xmit_pre_tun_flow(app, flow_pay);
 		}
@@ -1590,6 +1591,7 @@ nfp_flower_del_offload(struct nfp_app *app, struct net_device *netdev,
 			predt = nfp_flow->pre_tun_rule.predt;
 			if (predt) {
 				spin_lock_bh(&priv->predt_lock);
+				nfp_tun_unlink_and_update_nn_entries(app, predt);
 				list_del(&predt->list_head);
 				spin_unlock_bh(&priv->predt_lock);
 				kfree(predt);
diff --git a/drivers/net/ethernet/netronome/nfp/flower/tunnel_conf.c b/drivers/net/ethernet/netronome/nfp/flower/tunnel_conf.c
index 4ca149501409..fa9df24bec27 100644
--- a/drivers/net/ethernet/netronome/nfp/flower/tunnel_conf.c
+++ b/drivers/net/ethernet/netronome/nfp/flower/tunnel_conf.c
@@ -301,6 +301,150 @@ nfp_flower_xmit_tun_conf(struct nfp_app *app, u8 mtype, u16 plen, void *pdata,
 	return 0;
 }
 
+static void
+nfp_tun_mutual_link(struct nfp_predt_entry *predt,
+		    struct nfp_neigh_entry *neigh)
+{
+	struct nfp_fl_payload *flow_pay = predt->flow_pay;
+	struct nfp_tun_neigh_ext *ext;
+	struct nfp_tun_neigh *common;
+
+	if (flow_pay->pre_tun_rule.is_ipv6 != neigh->is_ipv6)
+		return;
+
+	/* In the case of bonding it is possible that there might already
+	 * be a flow linked (as the MAC address gets shared). If a flow
+	 * is already linked just return.
+	 */
+	if (neigh->flow)
+		return;
+
+	common = neigh->is_ipv6 ?
+		 &((struct nfp_tun_neigh_v6 *)neigh->payload)->common :
+		 &((struct nfp_tun_neigh_v4 *)neigh->payload)->common;
+	ext = neigh->is_ipv6 ?
+		 &((struct nfp_tun_neigh_v6 *)neigh->payload)->ext :
+		 &((struct nfp_tun_neigh_v4 *)neigh->payload)->ext;
+
+	if (memcmp(flow_pay->pre_tun_rule.loc_mac,
+		   common->src_addr, ETH_ALEN) ||
+	    memcmp(flow_pay->pre_tun_rule.rem_mac,
+		   common->dst_addr, ETH_ALEN))
+		return;
+
+	list_add(&neigh->list_head, &predt->nn_list);
+	neigh->flow = predt;
+	ext->host_ctx = flow_pay->meta.host_ctx_id;
+	ext->vlan_tci = flow_pay->pre_tun_rule.vlan_tci;
+	ext->vlan_tpid = flow_pay->pre_tun_rule.vlan_tpid;
+}
+
+static void
+nfp_tun_link_predt_entries(struct nfp_app *app,
+			   struct nfp_neigh_entry *nn_entry)
+{
+	struct nfp_flower_priv *priv = app->priv;
+	struct nfp_predt_entry *predt, *tmp;
+
+	list_for_each_entry_safe(predt, tmp, &priv->predt_list, list_head) {
+		nfp_tun_mutual_link(predt, nn_entry);
+	}
+}
+
+void nfp_tun_link_and_update_nn_entries(struct nfp_app *app,
+					struct nfp_predt_entry *predt)
+{
+	struct nfp_flower_priv *priv = app->priv;
+	struct nfp_neigh_entry *nn_entry;
+	struct rhashtable_iter iter;
+	size_t neigh_size;
+	u8 type;
+
+	rhashtable_walk_enter(&priv->neigh_table, &iter);
+	rhashtable_walk_start(&iter);
+	while ((nn_entry = rhashtable_walk_next(&iter)) != NULL) {
+		if (IS_ERR(nn_entry))
+			continue;
+		nfp_tun_mutual_link(predt, nn_entry);
+		neigh_size = nn_entry->is_ipv6 ?
+			     sizeof(struct nfp_tun_neigh_v6) :
+			     sizeof(struct nfp_tun_neigh_v4);
+		type = nn_entry->is_ipv6 ? NFP_FLOWER_CMSG_TYPE_TUN_NEIGH_V6 :
+					   NFP_FLOWER_CMSG_TYPE_TUN_NEIGH;
+		nfp_flower_xmit_tun_conf(app, type, neigh_size,
+					 nn_entry->payload,
+					 GFP_ATOMIC);
+	}
+	rhashtable_walk_stop(&iter);
+	rhashtable_walk_exit(&iter);
+}
+
+static void nfp_tun_cleanup_nn_entries(struct nfp_app *app)
+{
+	struct nfp_flower_priv *priv = app->priv;
+	struct nfp_neigh_entry *neigh;
+	struct nfp_tun_neigh_ext *ext;
+	struct rhashtable_iter iter;
+	size_t neigh_size;
+	u8 type;
+
+	rhashtable_walk_enter(&priv->neigh_table, &iter);
+	rhashtable_walk_start(&iter);
+	while ((neigh = rhashtable_walk_next(&iter)) != NULL) {
+		if (IS_ERR(neigh))
+			continue;
+		ext = neigh->is_ipv6 ?
+			 &((struct nfp_tun_neigh_v6 *)neigh->payload)->ext :
+			 &((struct nfp_tun_neigh_v4 *)neigh->payload)->ext;
+		ext->host_ctx = cpu_to_be32(U32_MAX);
+		ext->vlan_tpid = cpu_to_be16(U16_MAX);
+		ext->vlan_tci = cpu_to_be16(U16_MAX);
+
+		neigh_size = neigh->is_ipv6 ?
+			     sizeof(struct nfp_tun_neigh_v6) :
+			     sizeof(struct nfp_tun_neigh_v4);
+		type = neigh->is_ipv6 ? NFP_FLOWER_CMSG_TYPE_TUN_NEIGH_V6 :
+					   NFP_FLOWER_CMSG_TYPE_TUN_NEIGH;
+		nfp_flower_xmit_tun_conf(app, type, neigh_size, neigh->payload,
+					 GFP_ATOMIC);
+
+		rhashtable_remove_fast(&priv->neigh_table, &neigh->ht_node,
+				       neigh_table_params);
+		if (neigh->flow)
+			list_del(&neigh->list_head);
+		kfree(neigh);
+	}
+	rhashtable_walk_stop(&iter);
+	rhashtable_walk_exit(&iter);
+}
+
+void nfp_tun_unlink_and_update_nn_entries(struct nfp_app *app,
+					  struct nfp_predt_entry *predt)
+{
+	struct nfp_neigh_entry *neigh, *tmp;
+	struct nfp_tun_neigh_ext *ext;
+	size_t neigh_size;
+	u8 type;
+
+	list_for_each_entry_safe(neigh, tmp, &predt->nn_list, list_head) {
+		ext = neigh->is_ipv6 ?
+			 &((struct nfp_tun_neigh_v6 *)neigh->payload)->ext :
+			 &((struct nfp_tun_neigh_v4 *)neigh->payload)->ext;
+		neigh->flow = NULL;
+		ext->host_ctx = cpu_to_be32(U32_MAX);
+		ext->vlan_tpid = cpu_to_be16(U16_MAX);
+		ext->vlan_tci = cpu_to_be16(U16_MAX);
+		list_del(&neigh->list_head);
+		neigh_size = neigh->is_ipv6 ?
+			     sizeof(struct nfp_tun_neigh_v6) :
+			     sizeof(struct nfp_tun_neigh_v4);
+		type = neigh->is_ipv6 ? NFP_FLOWER_CMSG_TYPE_TUN_NEIGH_V6 :
+					   NFP_FLOWER_CMSG_TYPE_TUN_NEIGH;
+		nfp_flower_xmit_tun_conf(app, type, neigh_size, neigh->payload,
+					 GFP_ATOMIC);
+	}
+}
+
 static bool
 __nfp_tun_has_route(struct list_head *route_list, spinlock_t *list_lock,
 		    void *add, int add_len)
@@ -497,6 +641,7 @@ nfp_tun_write_neigh(struct net_device *netdev, struct nfp_app *app,
 			nfp_tun_add_route_to_cache_v4(app, &payload->dst_ipv4);
 		}
 
+		nfp_tun_link_predt_entries(app, nn_entry);
 		nfp_flower_xmit_tun_conf(app, mtype, neigh_size,
 					 nn_entry->payload,
 					 GFP_ATOMIC);
@@ -1482,4 +1627,6 @@ void nfp_tunnel_config_stop(struct nfp_app *app)
 	/* Destroy rhash. Entries should be cleaned on netdev notifier unreg. */
 	rhashtable_free_and_destroy(&priv->tun.offloaded_macs,
 				    nfp_check_rhashtable_empty, NULL);
+
+	nfp_tun_cleanup_nn_entries(app);
 }
-- 
2.30.2

[PATCH net-next 8/9] nfp: flower: remove unused neighbour cache

[Simon Horman]

From: Louis Peens <louis.peens@corigine.com>

With the neighbour entries now stored in a dedicated table there
is no use to make use of the tunnel route cache anymore, so remove
this.

Signed-off-by: Louis Peens <louis.peens@corigine.com>
Signed-off-by: Yinjun Zhang <yinjun.zhang@corigine.com>
Signed-off-by: Simon Horman <simon.horman@corigine.com>
---
 .../net/ethernet/netronome/nfp/flower/main.h  |   8 -
 .../netronome/nfp/flower/tunnel_conf.c        | 175 ------------------
 2 files changed, 183 deletions(-)

diff --git a/drivers/net/ethernet/netronome/nfp/flower/main.h b/drivers/net/ethernet/netronome/nfp/flower/main.h
index 6bc7a9cbf131..66f847414693 100644
--- a/drivers/net/ethernet/netronome/nfp/flower/main.h
+++ b/drivers/net/ethernet/netronome/nfp/flower/main.h
@@ -87,12 +87,8 @@ struct nfp_fl_stats_id {
  * @offloaded_macs:	Hashtable of the offloaded MAC addresses
  * @ipv4_off_list:	List of IPv4 addresses to offload
  * @ipv6_off_list:	List of IPv6 addresses to offload
- * @neigh_off_list_v4:	List of IPv4 neighbour offloads
- * @neigh_off_list_v6:	List of IPv6 neighbour offloads
  * @ipv4_off_lock:	Lock for the IPv4 address list
  * @ipv6_off_lock:	Lock for the IPv6 address list
- * @neigh_off_lock_v4:	Lock for the IPv4 neighbour address list
- * @neigh_off_lock_v6:	Lock for the IPv6 neighbour address list
  * @mac_off_ids:	IDA to manage id assignment for offloaded MACs
  * @neigh_nb:		Notifier to monitor neighbour state
  */
@@ -100,12 +96,8 @@ struct nfp_fl_tunnel_offloads {
 	struct rhashtable offloaded_macs;
 	struct list_head ipv4_off_list;
 	struct list_head ipv6_off_list;
-	struct list_head neigh_off_list_v4;
-	struct list_head neigh_off_list_v6;
 	struct mutex ipv4_off_lock;
 	struct mutex ipv6_off_lock;
-	spinlock_t neigh_off_lock_v4;
-	spinlock_t neigh_off_lock_v6;
 	struct ida mac_off_ids;
 	struct notifier_block neigh_nb;
 };
diff --git a/drivers/net/ethernet/netronome/nfp/flower/tunnel_conf.c b/drivers/net/ethernet/netronome/nfp/flower/tunnel_conf.c
index fa9df24bec27..9c37ed6943d0 100644
--- a/drivers/net/ethernet/netronome/nfp/flower/tunnel_conf.c
+++ b/drivers/net/ethernet/netronome/nfp/flower/tunnel_conf.c
@@ -445,123 +445,6 @@ void nfp_tun_unlink_and_update_nn_entries(struct nfp_app *app,
 	}
 }
 
-static bool
-__nfp_tun_has_route(struct list_head *route_list, spinlock_t *list_lock,
-		    void *add, int add_len)
-{
-	struct nfp_offloaded_route *entry;
-
-	spin_lock_bh(list_lock);
-	list_for_each_entry(entry, route_list, list)
-		if (!memcmp(entry->ip_add, add, add_len)) {
-			spin_unlock_bh(list_lock);
-			return true;
-		}
-	spin_unlock_bh(list_lock);
-	return false;
-}
-
-static int
-__nfp_tun_add_route_to_cache(struct list_head *route_list,
-			     spinlock_t *list_lock, void *add, int add_len)
-{
-	struct nfp_offloaded_route *entry;
-
-	spin_lock_bh(list_lock);
-	list_for_each_entry(entry, route_list, list)
-		if (!memcmp(entry->ip_add, add, add_len)) {
-			spin_unlock_bh(list_lock);
-			return 0;
-		}
-
-	entry = kmalloc(struct_size(entry, ip_add, add_len), GFP_ATOMIC);
-	if (!entry) {
-		spin_unlock_bh(list_lock);
-		return -ENOMEM;
-	}
-
-	memcpy(entry->ip_add, add, add_len);
-	list_add_tail(&entry->list, route_list);
-	spin_unlock_bh(list_lock);
-
-	return 0;
-}
-
-static void
-__nfp_tun_del_route_from_cache(struct list_head *route_list,
-			       spinlock_t *list_lock, void *add, int add_len)
-{
-	struct nfp_offloaded_route *entry;
-
-	spin_lock_bh(list_lock);
-	list_for_each_entry(entry, route_list, list)
-		if (!memcmp(entry->ip_add, add, add_len)) {
-			list_del(&entry->list);
-			kfree(entry);
-			break;
-		}
-	spin_unlock_bh(list_lock);
-}
-
-static bool nfp_tun_has_route_v4(struct nfp_app *app, __be32 *ipv4_addr)
-{
-	struct nfp_flower_priv *priv = app->priv;
-
-	return __nfp_tun_has_route(&priv->tun.neigh_off_list_v4,
-				   &priv->tun.neigh_off_lock_v4, ipv4_addr,
-				   sizeof(*ipv4_addr));
-}
-
-static bool
-nfp_tun_has_route_v6(struct nfp_app *app, struct in6_addr *ipv6_addr)
-{
-	struct nfp_flower_priv *priv = app->priv;
-
-	return __nfp_tun_has_route(&priv->tun.neigh_off_list_v6,
-				   &priv->tun.neigh_off_lock_v6, ipv6_addr,
-				   sizeof(*ipv6_addr));
-}
-
-static void
-nfp_tun_add_route_to_cache_v4(struct nfp_app *app, __be32 *ipv4_addr)
-{
-	struct nfp_flower_priv *priv = app->priv;
-
-	__nfp_tun_add_route_to_cache(&priv->tun.neigh_off_list_v4,
-				     &priv->tun.neigh_off_lock_v4, ipv4_addr,
-				     sizeof(*ipv4_addr));
-}
-
-static void
-nfp_tun_add_route_to_cache_v6(struct nfp_app *app, struct in6_addr *ipv6_addr)
-{
-	struct nfp_flower_priv *priv = app->priv;
-
-	__nfp_tun_add_route_to_cache(&priv->tun.neigh_off_list_v6,
-				     &priv->tun.neigh_off_lock_v6, ipv6_addr,
-				     sizeof(*ipv6_addr));
-}
-
-static void
-nfp_tun_del_route_from_cache_v4(struct nfp_app *app, __be32 *ipv4_addr)
-{
-	struct nfp_flower_priv *priv = app->priv;
-
-	__nfp_tun_del_route_from_cache(&priv->tun.neigh_off_list_v4,
-				       &priv->tun.neigh_off_lock_v4, ipv4_addr,
-				       sizeof(*ipv4_addr));
-}
-
-static void
-nfp_tun_del_route_from_cache_v6(struct nfp_app *app, struct in6_addr *ipv6_addr)
-{
-	struct nfp_flower_priv *priv = app->priv;
-
-	__nfp_tun_del_route_from_cache(&priv->tun.neigh_off_list_v6,
-				       &priv->tun.neigh_off_lock_v6, ipv6_addr,
-				       sizeof(*ipv6_addr));
-}
-
 static void
 nfp_tun_write_neigh(struct net_device *netdev, struct nfp_app *app,
 		    void *flow, struct neighbour *neigh, bool is_ipv6)
@@ -628,19 +511,6 @@ nfp_tun_write_neigh(struct net_device *netdev, struct nfp_app *app,
 					   neigh_table_params))
 			goto err;
 
-		/* Add entries to the relevant route cache */
-		if (is_ipv6) {
-			struct nfp_tun_neigh_v6 *payload;
-
-			payload = (struct nfp_tun_neigh_v6 *)nn_entry->payload;
-			nfp_tun_add_route_to_cache_v6(app, &payload->dst_ipv6);
-		} else {
-			struct nfp_tun_neigh_v4 *payload;
-
-			payload = (struct nfp_tun_neigh_v4 *)nn_entry->payload;
-			nfp_tun_add_route_to_cache_v4(app, &payload->dst_ipv4);
-		}
-
 		nfp_tun_link_predt_entries(app, nn_entry);
 		nfp_flower_xmit_tun_conf(app, mtype, neigh_size,
 					 nn_entry->payload,
@@ -654,8 +524,6 @@ nfp_tun_write_neigh(struct net_device *netdev, struct nfp_app *app,
 			memset(payload, 0, sizeof(struct nfp_tun_neigh_v6));
 			payload->dst_ipv6 = flowi6->daddr;
 			mtype = NFP_FLOWER_CMSG_TYPE_TUN_NEIGH_V6;
-			nfp_tun_del_route_from_cache_v6(app,
-							&payload->dst_ipv6);
 		} else {
 			struct flowi4 *flowi4 = (struct flowi4 *)flow;
 			struct nfp_tun_neigh_v4 *payload;
@@ -664,8 +532,6 @@ nfp_tun_write_neigh(struct net_device *netdev, struct nfp_app *app,
 			memset(payload, 0, sizeof(struct nfp_tun_neigh_v4));
 			payload->dst_ipv4 = flowi4->daddr;
 			mtype = NFP_FLOWER_CMSG_TYPE_TUN_NEIGH;
-			nfp_tun_del_route_from_cache_v4(app,
-							&payload->dst_ipv4);
 		}
 		/* Trigger ARP to verify invalid neighbour state. */
 		neigh_event_send(neigh, NULL);
@@ -734,11 +600,6 @@ nfp_tun_neigh_event_handler(struct notifier_block *nb, unsigned long event,
 	    !nfp_flower_internal_port_can_offload(app, n->dev))
 		return NOTIFY_DONE;
 
-	/* Only concerned with changes to routes already added to NFP. */
-	if ((ipv6 && !nfp_tun_has_route_v6(app, &flow6.daddr)) ||
-	    (!ipv6 && !nfp_tun_has_route_v4(app, &flow4.daddr)))
-		return NOTIFY_DONE;
-
 #if IS_ENABLED(CONFIG_INET)
 	if (ipv6) {
 #if IS_ENABLED(CONFIG_IPV6)
@@ -1557,10 +1418,6 @@ int nfp_tunnel_config_start(struct nfp_app *app)
 	INIT_LIST_HEAD(&priv->tun.ipv6_off_list);
 
 	/* Initialise priv data for neighbour offloading. */
-	spin_lock_init(&priv->tun.neigh_off_lock_v4);
-	INIT_LIST_HEAD(&priv->tun.neigh_off_list_v4);
-	spin_lock_init(&priv->tun.neigh_off_lock_v6);
-	INIT_LIST_HEAD(&priv->tun.neigh_off_list_v6);
 	priv->tun.neigh_nb.notifier_call = nfp_tun_neigh_event_handler;
 
 	err = register_netevent_notifier(&priv->tun.neigh_nb);
@@ -1575,11 +1432,8 @@ int nfp_tunnel_config_start(struct nfp_app *app)
 
 void nfp_tunnel_config_stop(struct nfp_app *app)
 {
-	struct nfp_offloaded_route *route_entry, *temp;
 	struct nfp_flower_priv *priv = app->priv;
 	struct nfp_ipv4_addr_entry *ip_entry;
-	struct nfp_tun_neigh_v6 ipv6_route;
-	struct nfp_tun_neigh_v4 ipv4_route;
 	struct list_head *ptr, *storage;
 
 	unregister_netevent_notifier(&priv->tun.neigh_nb);
@@ -1595,35 +1449,6 @@ void nfp_tunnel_config_stop(struct nfp_app *app)
 
 	mutex_destroy(&priv->tun.ipv6_off_lock);
 
-	/* Free memory in the route list and remove entries from fw cache. */
-	list_for_each_entry_safe(route_entry, temp,
-				 &priv->tun.neigh_off_list_v4, list) {
-		memset(&ipv4_route, 0, sizeof(ipv4_route));
-		memcpy(&ipv4_route.dst_ipv4, &route_entry->ip_add,
-		       sizeof(ipv4_route.dst_ipv4));
-		list_del(&route_entry->list);
-		kfree(route_entry);
-
-		nfp_flower_xmit_tun_conf(app, NFP_FLOWER_CMSG_TYPE_TUN_NEIGH,
-					 sizeof(struct nfp_tun_neigh_v4),
-					 (unsigned char *)&ipv4_route,
-					 GFP_KERNEL);
-	}
-
-	list_for_each_entry_safe(route_entry, temp,
-				 &priv->tun.neigh_off_list_v6, list) {
-		memset(&ipv6_route, 0, sizeof(ipv6_route));
-		memcpy(&ipv6_route.dst_ipv6, &route_entry->ip_add,
-		       sizeof(ipv6_route.dst_ipv6));
-		list_del(&route_entry->list);
-		kfree(route_entry);
-
-		nfp_flower_xmit_tun_conf(app, NFP_FLOWER_CMSG_TYPE_TUN_NEIGH_V6,
-					 sizeof(struct nfp_tun_neigh_v6),
-					 (unsigned char *)&ipv6_route,
-					 GFP_KERNEL);
-	}
-
 	/* Destroy rhash. Entries should be cleaned on netdev notifier unreg. */
 	rhashtable_free_and_destroy(&priv->tun.offloaded_macs,
 				    nfp_check_rhashtable_empty, NULL);
-- 
2.30.2

[PATCH net-next 9/9] nfp: flower: enable decap_v2 bit

[Simon Horman]

From: Louis Peens <louis.peens@corigine.com>

Finally enable the decap_v2 feature bit now that all the
other bits are in place to configure it correctly.

Signed-off-by: Louis Peens <louis.peens@corigine.com>
Signed-off-by: Yinjun Zhang <yinjun.zhang@corigine.com>
Signed-off-by: Simon Horman <simon.horman@corigine.com>
---
 drivers/net/ethernet/netronome/nfp/flower/main.h | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/netronome/nfp/flower/main.h b/drivers/net/ethernet/netronome/nfp/flower/main.h
index 66f847414693..cb799d18682d 100644
--- a/drivers/net/ethernet/netronome/nfp/flower/main.h
+++ b/drivers/net/ethernet/netronome/nfp/flower/main.h
@@ -68,7 +68,8 @@ struct nfp_app;
 	NFP_FL_FEATS_IPV6_TUN | \
 	NFP_FL_FEATS_VLAN_QINQ | \
 	NFP_FL_FEATS_QOS_PPS | \
-	NFP_FL_FEATS_QOS_METER)
+	NFP_FL_FEATS_QOS_METER | \
+	NFP_FL_FEATS_DECAP_V2)
 
 struct nfp_fl_mask_id {
 	struct circ_buf mask_id_free_list;
-- 
2.30.2

Re: [PATCH net-next 0/9] nfp: flower: decap neighbour table rework

[patchwork-bot+netdevbpf]

Hello:

This series was applied to netdev/net-next.git (master)
by David S. Miller <davem@davemloft.net>:

On Thu,  5 May 2022 14:43:39 +0900 you wrote:
> Louis Peens says:
> 
> This patch series reworks the way in which flow rules that outputs to
> OVS internal ports gets handled by the nfp driver.
> 
> Previously this made use of a small pre_tun_table, but this only used
> destination MAC addresses, and made the implicit assumption that there is
> only a single source MAC":"destination MAC" mapping per tunnel. In
> hindsight this seems to be a pretty obvious oversight, but this was hidden
> in plain sight for quite some time.
> 
> [...]

Here is the summary with links:
  - [net-next,1/9] nfp: flower: add infrastructure for pre_tun rework
    https://git.kernel.org/netdev/net-next/c/29c691347e38
  - [net-next,2/9] nfp: flower: add/remove predt_list entries
    https://git.kernel.org/netdev/net-next/c/e30b2b68c14f
  - [net-next,3/9] nfp: flower: enforce more strict pre_tun checks
    https://git.kernel.org/netdev/net-next/c/38fc158e172b
  - [net-next,4/9] nfp: flower: fixup ipv6/ipv4 route lookup for neigh events
    https://git.kernel.org/netdev/net-next/c/9d5447ed44b5
  - [net-next,5/9] nfp: flower: update nfp_tun_neigh structs
    https://git.kernel.org/netdev/net-next/c/9ee7c42183d1
  - [net-next,6/9] nfp: flower: rework tunnel neighbour configuration
    https://git.kernel.org/netdev/net-next/c/f1df7956c11f
  - [net-next,7/9] nfp: flower: link pre_tun flow rules with neigh entries
    https://git.kernel.org/netdev/net-next/c/591c90a1d0b0
  - [net-next,8/9] nfp: flower: remove unused neighbour cache
    https://git.kernel.org/netdev/net-next/c/c83a0fbe9766
  - [net-next,9/9] nfp: flower: enable decap_v2 bit
    https://git.kernel.org/netdev/net-next/c/a7da2a864a4f

You are awesome, thank you!
-- 
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html