Merge branch 'mlxsw-Preparations-for-VxLAN-support'

Ido Schimmel says:

====================
mlxsw: Preparations for VxLAN support

This patchset prepares mlxsw for VxLAN support. It contains small and
mostly non-functional changes.

The first eight patches perform small changes in the code to make it
more receptive towards the actual VxLAN changes in the next patchset.

Patches 9-17 add the registers used to configure the device for VxLAN
offload.

Last two patches add the required resources and trap IDs.

The next patchset is available here [1].

1. https://github.com/idosch/linux/tree/vxlan
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net/ethernet/mellanox/mlxsw/reg.h

@@ -295,6 +295,7 @@ enum mlxsw_reg_sfd_rec_type {
 	MLXSW_REG_SFD_REC_TYPE_UNICAST = 0x0,
 	MLXSW_REG_SFD_REC_TYPE_UNICAST_LAG = 0x1,
 	MLXSW_REG_SFD_REC_TYPE_MULTICAST = 0x2,
+	MLXSW_REG_SFD_REC_TYPE_UNICAST_TUNNEL = 0xC,
 };
 
 /* reg_sfd_rec_type
@@ -525,6 +526,61 @@ mlxsw_reg_sfd_mc_pack(char *payload, int rec_index,
 	mlxsw_reg_sfd_mc_mid_set(payload, rec_index, mid);
 }
 
+/* reg_sfd_uc_tunnel_uip_msb
+ * When protocol is IPv4, the most significant byte of the underlay IPv4
+ * destination IP.
+ * When protocol is IPv6, reserved.
+ * Access: RW
+ */
+MLXSW_ITEM32_INDEXED(reg, sfd, uc_tunnel_uip_msb, MLXSW_REG_SFD_BASE_LEN, 24,
+		     8, MLXSW_REG_SFD_REC_LEN, 0x08, false);
+
+/* reg_sfd_uc_tunnel_fid
+ * Filtering ID.
+ * Access: Index
+ */
+MLXSW_ITEM32_INDEXED(reg, sfd, uc_tunnel_fid, MLXSW_REG_SFD_BASE_LEN, 0, 16,
+		     MLXSW_REG_SFD_REC_LEN, 0x08, false);
+
+enum mlxsw_reg_sfd_uc_tunnel_protocol {
+	MLXSW_REG_SFD_UC_TUNNEL_PROTOCOL_IPV4,
+	MLXSW_REG_SFD_UC_TUNNEL_PROTOCOL_IPV6,
+};
+
+/* reg_sfd_uc_tunnel_protocol
+ * IP protocol.
+ * Access: RW
+ */
+MLXSW_ITEM32_INDEXED(reg, sfd, uc_tunnel_protocol, MLXSW_REG_SFD_BASE_LEN, 27,
+		     1, MLXSW_REG_SFD_REC_LEN, 0x0C, false);
+
+/* reg_sfd_uc_tunnel_uip_lsb
+ * When protocol is IPv4, the least significant bytes of the underlay
+ * IPv4 destination IP.
+ * When protocol is IPv6, pointer to the underlay IPv6 destination IP
+ * which is configured by RIPS.
+ * Access: RW
+ */
+MLXSW_ITEM32_INDEXED(reg, sfd, uc_tunnel_uip_lsb, MLXSW_REG_SFD_BASE_LEN, 0,
+		     24, MLXSW_REG_SFD_REC_LEN, 0x0C, false);
+
+static inline void
+mlxsw_reg_sfd_uc_tunnel_pack(char *payload, int rec_index,
+			     enum mlxsw_reg_sfd_rec_policy policy,
+			     const char *mac, u16 fid,
+			     enum mlxsw_reg_sfd_rec_action action, u32 uip,
+			     enum mlxsw_reg_sfd_uc_tunnel_protocol proto)
+{
+	mlxsw_reg_sfd_rec_pack(payload, rec_index,
+			       MLXSW_REG_SFD_REC_TYPE_UNICAST_TUNNEL, mac,
+			       action);
+	mlxsw_reg_sfd_rec_policy_set(payload, rec_index, policy);
+	mlxsw_reg_sfd_uc_tunnel_uip_msb_set(payload, rec_index, uip >> 24);
+	mlxsw_reg_sfd_uc_tunnel_uip_lsb_set(payload, rec_index, uip);
+	mlxsw_reg_sfd_uc_tunnel_fid_set(payload, rec_index, fid);
+	mlxsw_reg_sfd_uc_tunnel_protocol_set(payload, rec_index, proto);
+}
+
 /* SFN - Switch FDB Notification Register
  * -------------------------------------------
  * The switch provides notifications on newly learned FDB entries and
@@ -1069,6 +1125,8 @@ enum mlxsw_reg_sfdf_flush_type {
 	MLXSW_REG_SFDF_FLUSH_PER_PORT_AND_FID,
 	MLXSW_REG_SFDF_FLUSH_PER_LAG,
 	MLXSW_REG_SFDF_FLUSH_PER_LAG_AND_FID,
+	MLXSW_REG_SFDF_FLUSH_PER_NVE,
+	MLXSW_REG_SFDF_FLUSH_PER_NVE_AND_FID,
 };
 
 /* reg_sfdf_flush_type
@@ -1079,6 +1137,10 @@ enum mlxsw_reg_sfdf_flush_type {
  * 3 - All FID dynamic entries pointing to port are flushed.
  * 4 - All dynamic entries pointing to LAG are flushed.
  * 5 - All FID dynamic entries pointing to LAG are flushed.
+ * 6 - All entries of type "Unicast Tunnel" or "Multicast Tunnel" are
+ *     flushed.
+ * 7 - All entries of type "Unicast Tunnel" or "Multicast Tunnel" are
+ *     flushed, per FID.
  * Access: RW
  */
 MLXSW_ITEM32(reg, sfdf, flush_type, 0x04, 28, 4);
@@ -1315,12 +1377,19 @@ MLXSW_ITEM32(reg, slcr, type, 0x00, 0, 4);
  */
 MLXSW_ITEM32(reg, slcr, lag_hash, 0x04, 0, 20);
 
-static inline void mlxsw_reg_slcr_pack(char *payload, u16 lag_hash)
+/* reg_slcr_seed
+ * LAG seed value. The seed is the same for all ports.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, slcr, seed, 0x08, 0, 32);
+
+static inline void mlxsw_reg_slcr_pack(char *payload, u16 lag_hash, u32 seed)
 {
 	MLXSW_REG_ZERO(slcr, payload);
 	mlxsw_reg_slcr_pp_set(payload, MLXSW_REG_SLCR_PP_GLOBAL);
 	mlxsw_reg_slcr_type_set(payload, MLXSW_REG_SLCR_TYPE_CRC);
 	mlxsw_reg_slcr_lag_hash_set(payload, lag_hash);
+	mlxsw_reg_slcr_seed_set(payload, seed);
 }
 
 /* SLCOR - Switch LAG Collector Register
@@ -8279,6 +8348,508 @@ static inline void mlxsw_reg_mgpc_pack(char *payload, u32 counter_index,
 	mlxsw_reg_mgpc_opcode_set(payload, opcode);
 }
 
+/* MPRS - Monitoring Parsing State Register
+ * ----------------------------------------
+ * The MPRS register is used for setting up the parsing for hash,
+ * policy-engine and routing.
+ */
+#define MLXSW_REG_MPRS_ID 0x9083
+#define MLXSW_REG_MPRS_LEN 0x14
+
+MLXSW_REG_DEFINE(mprs, MLXSW_REG_MPRS_ID, MLXSW_REG_MPRS_LEN);
+
+/* reg_mprs_parsing_depth
+ * Minimum parsing depth.
+ * Need to enlarge parsing depth according to L3, MPLS, tunnels, ACL
+ * rules, traps, hash, etc. Default is 96 bytes. Reserved when SwitchX-2.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mprs, parsing_depth, 0x00, 0, 16);
+
+/* reg_mprs_parsing_en
+ * Parsing enable.
+ * Bit 0 - Enable parsing of NVE of types VxLAN, VxLAN-GPE, GENEVE and
+ * NVGRE. Default is enabled. Reserved when SwitchX-2.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mprs, parsing_en, 0x04, 0, 16);
+
+/* reg_mprs_vxlan_udp_dport
+ * VxLAN UDP destination port.
+ * Used for identifying VxLAN packets and for dport field in
+ * encapsulation. Default is 4789.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mprs, vxlan_udp_dport, 0x10, 0, 16);
+
+static inline void mlxsw_reg_mprs_pack(char *payload, u16 parsing_depth,
+				       u16 vxlan_udp_dport)
+{
+	MLXSW_REG_ZERO(mprs, payload);
+	mlxsw_reg_mprs_parsing_depth_set(payload, parsing_depth);
+	mlxsw_reg_mprs_parsing_en_set(payload, true);
+	mlxsw_reg_mprs_vxlan_udp_dport_set(payload, vxlan_udp_dport);
+}
+
+/* TNGCR - Tunneling NVE General Configuration Register
+ * ----------------------------------------------------
+ * The TNGCR register is used for setting up the NVE Tunneling configuration.
+ */
+#define MLXSW_REG_TNGCR_ID 0xA001
+#define MLXSW_REG_TNGCR_LEN 0x44
+
+MLXSW_REG_DEFINE(tngcr, MLXSW_REG_TNGCR_ID, MLXSW_REG_TNGCR_LEN);
+
+enum mlxsw_reg_tngcr_type {
+	MLXSW_REG_TNGCR_TYPE_VXLAN,
+	MLXSW_REG_TNGCR_TYPE_VXLAN_GPE,
+	MLXSW_REG_TNGCR_TYPE_GENEVE,
+	MLXSW_REG_TNGCR_TYPE_NVGRE,
+};
+
+/* reg_tngcr_type
+ * Tunnel type for encapsulation and decapsulation. The types are mutually
+ * exclusive.
+ * Note: For Spectrum the NVE parsing must be enabled in MPRS.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tngcr, type, 0x00, 0, 4);
+
+/* reg_tngcr_nve_valid
+ * The VTEP is valid. Allows adding FDB entries for tunnel encapsulation.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tngcr, nve_valid, 0x04, 31, 1);
+
+/* reg_tngcr_nve_ttl_uc
+ * The TTL for NVE tunnel encapsulation underlay unicast packets.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tngcr, nve_ttl_uc, 0x04, 0, 8);
+
+/* reg_tngcr_nve_ttl_mc
+ * The TTL for NVE tunnel encapsulation underlay multicast packets.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tngcr, nve_ttl_mc, 0x08, 0, 8);
+
+enum {
+	/* Do not copy flow label. Calculate flow label using nve_flh. */
+	MLXSW_REG_TNGCR_FL_NO_COPY,
+	/* Copy flow label from inner packet if packet is IPv6 and
+	 * encapsulation is by IPv6. Otherwise, calculate flow label using
+	 * nve_flh.
+	 */
+	MLXSW_REG_TNGCR_FL_COPY,
+};
+
+/* reg_tngcr_nve_flc
+ * For NVE tunnel encapsulation: Flow label copy from inner packet.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tngcr, nve_flc, 0x0C, 25, 1);
+
+enum {
+	/* Flow label is static. In Spectrum this means '0'. Spectrum-2
+	 * uses {nve_fl_prefix, nve_fl_suffix}.
+	 */
+	MLXSW_REG_TNGCR_FL_NO_HASH,
+	/* 8 LSBs of the flow label are calculated from ECMP hash of the
+	 * inner packet. 12 MSBs are configured by nve_fl_prefix.
+	 */
+	MLXSW_REG_TNGCR_FL_HASH,
+};
+
+/* reg_tngcr_nve_flh
+ * NVE flow label hash.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tngcr, nve_flh, 0x0C, 24, 1);
+
+/* reg_tngcr_nve_fl_prefix
+ * NVE flow label prefix. Constant 12 MSBs of the flow label.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tngcr, nve_fl_prefix, 0x0C, 8, 12);
+
+/* reg_tngcr_nve_fl_suffix
+ * NVE flow label suffix. Constant 8 LSBs of the flow label.
+ * Reserved when nve_flh=1 and for Spectrum.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tngcr, nve_fl_suffix, 0x0C, 0, 8);
+
+enum {
+	/* Source UDP port is fixed (default '0') */
+	MLXSW_REG_TNGCR_UDP_SPORT_NO_HASH,
+	/* Source UDP port is calculated based on hash */
+	MLXSW_REG_TNGCR_UDP_SPORT_HASH,
+};
+
+/* reg_tngcr_nve_udp_sport_type
+ * NVE UDP source port type.
+ * Spectrum uses LAG hash (SLCRv2). Spectrum-2 uses ECMP hash (RECRv2).
+ * When the source UDP port is calculated based on hash, then the 8 LSBs
+ * are calculated from hash the 8 MSBs are configured by
+ * nve_udp_sport_prefix.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tngcr, nve_udp_sport_type, 0x10, 24, 1);
+
+/* reg_tngcr_nve_udp_sport_prefix
+ * NVE UDP source port prefix. Constant 8 MSBs of the UDP source port.
+ * Reserved when NVE type is NVGRE.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tngcr, nve_udp_sport_prefix, 0x10, 8, 8);
+
+/* reg_tngcr_nve_group_size_mc
+ * The amount of sequential linked lists of MC entries. The first linked
+ * list is configured by SFD.underlay_mc_ptr.
+ * Valid values: 1, 2, 4, 8, 16, 32, 64
+ * The linked list are configured by TNUMT.
+ * The hash is set by LAG hash.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tngcr, nve_group_size_mc, 0x18, 0, 8);
+
+/* reg_tngcr_nve_group_size_flood
+ * The amount of sequential linked lists of flooding entries. The first
+ * linked list is configured by SFMR.nve_tunnel_flood_ptr
+ * Valid values: 1, 2, 4, 8, 16, 32, 64
+ * The linked list are configured by TNUMT.
+ * The hash is set by LAG hash.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tngcr, nve_group_size_flood, 0x1C, 0, 8);
+
+/* reg_tngcr_learn_enable
+ * During decapsulation, whether to learn from NVE port.
+ * Reserved when Spectrum-2. See TNPC.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tngcr, learn_enable, 0x20, 31, 1);
+
+/* reg_tngcr_underlay_virtual_router
+ * Underlay virtual router.
+ * Reserved when Spectrum-2.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tngcr, underlay_virtual_router, 0x20, 0, 16);
+
+/* reg_tngcr_underlay_rif
+ * Underlay ingress router interface. RIF type should be loopback generic.
+ * Reserved when Spectrum.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tngcr, underlay_rif, 0x24, 0, 16);
+
+/* reg_tngcr_usipv4
+ * Underlay source IPv4 address of the NVE.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tngcr, usipv4, 0x28, 0, 32);
+
+/* reg_tngcr_usipv6
+ * Underlay source IPv6 address of the NVE. For Spectrum, must not be
+ * modified under traffic of NVE tunneling encapsulation.
+ * Access: RW
+ */
+MLXSW_ITEM_BUF(reg, tngcr, usipv6, 0x30, 16);
+
+static inline void mlxsw_reg_tngcr_pack(char *payload,
+					enum mlxsw_reg_tngcr_type type,
+					bool valid, u8 ttl)
+{
+	MLXSW_REG_ZERO(tngcr, payload);
+	mlxsw_reg_tngcr_type_set(payload, type);
+	mlxsw_reg_tngcr_nve_valid_set(payload, valid);
+	mlxsw_reg_tngcr_nve_ttl_uc_set(payload, ttl);
+	mlxsw_reg_tngcr_nve_ttl_mc_set(payload, ttl);
+	mlxsw_reg_tngcr_nve_flc_set(payload, MLXSW_REG_TNGCR_FL_NO_COPY);
+	mlxsw_reg_tngcr_nve_flh_set(payload, 0);
+	mlxsw_reg_tngcr_nve_udp_sport_type_set(payload,
+					       MLXSW_REG_TNGCR_UDP_SPORT_HASH);
+	mlxsw_reg_tngcr_nve_udp_sport_prefix_set(payload, 0);
+	mlxsw_reg_tngcr_nve_group_size_mc_set(payload, 1);
+	mlxsw_reg_tngcr_nve_group_size_flood_set(payload, 1);
+}
+
+/* TNUMT - Tunneling NVE Underlay Multicast Table Register
+ * -------------------------------------------------------
+ * The TNUMT register is for building the underlay MC table. It is used
+ * for MC, flooding and BC traffic into the NVE tunnel.
+ */
+#define MLXSW_REG_TNUMT_ID 0xA003
+#define MLXSW_REG_TNUMT_LEN 0x20
+
+MLXSW_REG_DEFINE(tnumt, MLXSW_REG_TNUMT_ID, MLXSW_REG_TNUMT_LEN);
+
+enum mlxsw_reg_tnumt_record_type {
+	MLXSW_REG_TNUMT_RECORD_TYPE_IPV4,
+	MLXSW_REG_TNUMT_RECORD_TYPE_IPV6,
+	MLXSW_REG_TNUMT_RECORD_TYPE_LABEL,
+};
+
+/* reg_tnumt_record_type
+ * Record type.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tnumt, record_type, 0x00, 28, 4);
+
+enum mlxsw_reg_tnumt_tunnel_port {
+	MLXSW_REG_TNUMT_TUNNEL_PORT_NVE,
+	MLXSW_REG_TNUMT_TUNNEL_PORT_VPLS,
+	MLXSW_REG_TNUMT_TUNNEL_FLEX_TUNNEL0,
+	MLXSW_REG_TNUMT_TUNNEL_FLEX_TUNNEL1,
+};
+
+/* reg_tnumt_tunnel_port
+ * Tunnel port.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tnumt, tunnel_port, 0x00, 24, 4);
+
+/* reg_tnumt_underlay_mc_ptr
+ * Index to the underlay multicast table.
+ * For Spectrum the index is to the KVD linear.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, tnumt, underlay_mc_ptr, 0x00, 0, 24);
+
+/* reg_tnumt_vnext
+ * The next_underlay_mc_ptr is valid.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tnumt, vnext, 0x04, 31, 1);
+
+/* reg_tnumt_next_underlay_mc_ptr
+ * The next index to the underlay multicast table.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tnumt, next_underlay_mc_ptr, 0x04, 0, 24);
+
+/* reg_tnumt_record_size
+ * Number of IP addresses in the record.
+ * Range is 1..cap_max_nve_mc_entries_ipv{4,6}
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tnumt, record_size, 0x08, 0, 3);
+
+/* reg_tnumt_udip
+ * The underlay IPv4 addresses. udip[i] is reserved if i >= size
+ * Access: RW
+ */
+MLXSW_ITEM32_INDEXED(reg, tnumt, udip, 0x0C, 0, 32, 0x04, 0x00, false);
+
+/* reg_tnumt_udip_ptr
+ * The pointer to the underlay IPv6 addresses. udip_ptr[i] is reserved if
+ * i >= size. The IPv6 addresses are configured by RIPS.
+ * Access: RW
+ */
+MLXSW_ITEM32_INDEXED(reg, tnumt, udip_ptr, 0x0C, 0, 24, 0x04, 0x00, false);
+
+static inline void mlxsw_reg_tnumt_pack(char *payload,
+					enum mlxsw_reg_tnumt_record_type type,
+					enum mlxsw_reg_tnumt_tunnel_port tport,
+					u32 underlay_mc_ptr, bool vnext,
+					u32 next_underlay_mc_ptr,
+					u8 record_size)
+{
+	MLXSW_REG_ZERO(tnumt, payload);
+	mlxsw_reg_tnumt_record_type_set(payload, type);
+	mlxsw_reg_tnumt_tunnel_port_set(payload, tport);
+	mlxsw_reg_tnumt_underlay_mc_ptr_set(payload, underlay_mc_ptr);
+	mlxsw_reg_tnumt_vnext_set(payload, vnext);
+	mlxsw_reg_tnumt_next_underlay_mc_ptr_set(payload, next_underlay_mc_ptr);
+	mlxsw_reg_tnumt_record_size_set(payload, record_size);
+}
+
+/* TNQCR - Tunneling NVE QoS Configuration Register
+ * ------------------------------------------------
+ * The TNQCR register configures how QoS is set in encapsulation into the
+ * underlay network.
+ */
+#define MLXSW_REG_TNQCR_ID 0xA010
+#define MLXSW_REG_TNQCR_LEN 0x0C
+
+MLXSW_REG_DEFINE(tnqcr, MLXSW_REG_TNQCR_ID, MLXSW_REG_TNQCR_LEN);
+
+/* reg_tnqcr_enc_set_dscp
+ * For encapsulation: How to set DSCP field:
+ * 0 - Copy the DSCP from the overlay (inner) IP header to the underlay
+ * (outer) IP header. If there is no IP header, use TNQDR.dscp
+ * 1 - Set the DSCP field as TNQDR.dscp
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tnqcr, enc_set_dscp, 0x04, 28, 1);
+
+static inline void mlxsw_reg_tnqcr_pack(char *payload)
+{
+	MLXSW_REG_ZERO(tnqcr, payload);
+	mlxsw_reg_tnqcr_enc_set_dscp_set(payload, 0);
+}
+
+/* TNQDR - Tunneling NVE QoS Default Register
+ * ------------------------------------------
+ * The TNQDR register configures the default QoS settings for NVE
+ * encapsulation.
+ */
+#define MLXSW_REG_TNQDR_ID 0xA011
+#define MLXSW_REG_TNQDR_LEN 0x08
+
+MLXSW_REG_DEFINE(tnqdr, MLXSW_REG_TNQDR_ID, MLXSW_REG_TNQDR_LEN);
+
+/* reg_tnqdr_local_port
+ * Local port number (receive port). CPU port is supported.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, tnqdr, local_port, 0x00, 16, 8);
+
+/* reg_tnqdr_dscp
+ * For encapsulation, the default DSCP.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tnqdr, dscp, 0x04, 0, 6);
+
+static inline void mlxsw_reg_tnqdr_pack(char *payload, u8 local_port)
+{
+	MLXSW_REG_ZERO(tnqdr, payload);
+	mlxsw_reg_tnqdr_local_port_set(payload, local_port);
+	mlxsw_reg_tnqdr_dscp_set(payload, 0);
+}
+
+/* TNEEM - Tunneling NVE Encapsulation ECN Mapping Register
+ * --------------------------------------------------------
+ * The TNEEM register maps ECN of the IP header at the ingress to the
+ * encapsulation to the ECN of the underlay network.
+ */
+#define MLXSW_REG_TNEEM_ID 0xA012
+#define MLXSW_REG_TNEEM_LEN 0x0C
+
+MLXSW_REG_DEFINE(tneem, MLXSW_REG_TNEEM_ID, MLXSW_REG_TNEEM_LEN);
+
+/* reg_tneem_overlay_ecn
+ * ECN of the IP header in the overlay network.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, tneem, overlay_ecn, 0x04, 24, 2);
+
+/* reg_tneem_underlay_ecn
+ * ECN of the IP header in the underlay network.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tneem, underlay_ecn, 0x04, 16, 2);
+
+static inline void mlxsw_reg_tneem_pack(char *payload, u8 overlay_ecn,
+					u8 underlay_ecn)
+{
+	MLXSW_REG_ZERO(tneem, payload);
+	mlxsw_reg_tneem_overlay_ecn_set(payload, overlay_ecn);
+	mlxsw_reg_tneem_underlay_ecn_set(payload, underlay_ecn);
+}
+
+/* TNDEM - Tunneling NVE Decapsulation ECN Mapping Register
+ * --------------------------------------------------------
+ * The TNDEM register configures the actions that are done in the
+ * decapsulation.
+ */
+#define MLXSW_REG_TNDEM_ID 0xA013
+#define MLXSW_REG_TNDEM_LEN 0x0C
+
+MLXSW_REG_DEFINE(tndem, MLXSW_REG_TNDEM_ID, MLXSW_REG_TNDEM_LEN);
+
+/* reg_tndem_underlay_ecn
+ * ECN field of the IP header in the underlay network.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, tndem, underlay_ecn, 0x04, 24, 2);
+
+/* reg_tndem_overlay_ecn
+ * ECN field of the IP header in the overlay network.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, tndem, overlay_ecn, 0x04, 16, 2);
+
+/* reg_tndem_eip_ecn
+ * Egress IP ECN. ECN field of the IP header of the packet which goes out
+ * from the decapsulation.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tndem, eip_ecn, 0x04, 8, 2);
+
+/* reg_tndem_trap_en
+ * Trap enable:
+ * 0 - No trap due to decap ECN
+ * 1 - Trap enable with trap_id
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tndem, trap_en, 0x08, 28, 4);
+
+/* reg_tndem_trap_id
+ * Trap ID. Either DECAP_ECN0 or DECAP_ECN1.
+ * Reserved when trap_en is '0'.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tndem, trap_id, 0x08, 0, 9);
+
+static inline void mlxsw_reg_tndem_pack(char *payload, u8 underlay_ecn,
+					u8 overlay_ecn, u8 ecn, bool trap_en,
+					u16 trap_id)
+{
+	MLXSW_REG_ZERO(tndem, payload);
+	mlxsw_reg_tndem_underlay_ecn_set(payload, underlay_ecn);
+	mlxsw_reg_tndem_overlay_ecn_set(payload, overlay_ecn);
+	mlxsw_reg_tndem_eip_ecn_set(payload, ecn);
+	mlxsw_reg_tndem_trap_en_set(payload, trap_en);
+	mlxsw_reg_tndem_trap_id_set(payload, trap_id);
+}
+
+/* TNPC - Tunnel Port Configuration Register
+ * -----------------------------------------
+ * The TNPC register is used for tunnel port configuration.
+ * Reserved when Spectrum.
+ */
+#define MLXSW_REG_TNPC_ID 0xA020
+#define MLXSW_REG_TNPC_LEN 0x18
+
+MLXSW_REG_DEFINE(tnpc, MLXSW_REG_TNPC_ID, MLXSW_REG_TNPC_LEN);
+
+enum mlxsw_reg_tnpc_tunnel_port {
+	MLXSW_REG_TNPC_TUNNEL_PORT_NVE,
+	MLXSW_REG_TNPC_TUNNEL_PORT_VPLS,
+	MLXSW_REG_TNPC_TUNNEL_FLEX_TUNNEL0,
+	MLXSW_REG_TNPC_TUNNEL_FLEX_TUNNEL1,
+};
+
+/* reg_tnpc_tunnel_port
+ * Tunnel port.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, tnpc, tunnel_port, 0x00, 0, 4);
+
+/* reg_tnpc_learn_enable_v6
+ * During IPv6 underlay decapsulation, whether to learn from tunnel port.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tnpc, learn_enable_v6, 0x04, 1, 1);
+
+/* reg_tnpc_learn_enable_v4
+ * During IPv4 underlay decapsulation, whether to learn from tunnel port.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, tnpc, learn_enable_v4, 0x04, 0, 1);
+
+static inline void mlxsw_reg_tnpc_pack(char *payload,
+				       enum mlxsw_reg_tnpc_tunnel_port tport,
+				       bool learn_enable)
+{
+	MLXSW_REG_ZERO(tnpc, payload);
+	mlxsw_reg_tnpc_tunnel_port_set(payload, tport);
+	mlxsw_reg_tnpc_learn_enable_v4_set(payload, learn_enable);
+	mlxsw_reg_tnpc_learn_enable_v6_set(payload, learn_enable);
+}
+
 /* TIGCR - Tunneling IPinIP General Configuration Register
  * -------------------------------------------------------
  * The TIGCR register is used for setting up the IPinIP Tunnel configuration.
@@ -8828,6 +9399,14 @@ static const struct mlxsw_reg_info *mlxsw_reg_infos[] = {
 	MLXSW_REG(mcc),
 	MLXSW_REG(mcda),
 	MLXSW_REG(mgpc),
+	MLXSW_REG(mprs),
+	MLXSW_REG(tngcr),
+	MLXSW_REG(tnumt),
+	MLXSW_REG(tnqcr),
+	MLXSW_REG(tnqdr),
+	MLXSW_REG(tneem),
+	MLXSW_REG(tndem),
+	MLXSW_REG(tnpc),
 	MLXSW_REG(tigcr),
 	MLXSW_REG(sbpr),
 	MLXSW_REG(sbcm),

drivers/net/ethernet/mellanox/mlxsw/resources.h

@@ -46,6 +46,8 @@ enum mlxsw_res_id {
 	MLXSW_RES_ID_MAX_RIFS,
 	MLXSW_RES_ID_MC_ERIF_LIST_ENTRIES,
 	MLXSW_RES_ID_MAX_LPM_TREES,
+	MLXSW_RES_ID_MAX_NVE_MC_ENTRIES_IPV4,
+	MLXSW_RES_ID_MAX_NVE_MC_ENTRIES_IPV6,
 
 	/* Internal resources.
 	 * Determined by the SW, not queried from the HW.
@@ -96,6 +98,8 @@ static u16 mlxsw_res_ids[] = {
 	[MLXSW_RES_ID_MAX_RIFS] = 0x2C02,
 	[MLXSW_RES_ID_MC_ERIF_LIST_ENTRIES] = 0x2C10,
 	[MLXSW_RES_ID_MAX_LPM_TREES] = 0x2C30,
+	[MLXSW_RES_ID_MAX_NVE_MC_ENTRIES_IPV4] = 0x2E02,
+	[MLXSW_RES_ID_MAX_NVE_MC_ENTRIES_IPV6] = 0x2E03,
 };
 
 struct mlxsw_res {

drivers/net/ethernet/mellanox/mlxsw/spectrum.c

@@ -21,6 +21,7 @@
 #include <linux/dcbnl.h>
 #include <linux/inetdevice.h>
 #include <linux/netlink.h>
+#include <linux/random.h>
 #include <net/switchdev.h>
 #include <net/pkt_cls.h>
 #include <net/tc_act/tc_mirred.h>
@@ -3469,6 +3470,7 @@ static const struct mlxsw_listener mlxsw_sp_listener[] = {
 	MLXSW_SP_RXL_MARK(ROUTER_ALERT_IPV4, TRAP_TO_CPU, ROUTER_EXP, false),
 	MLXSW_SP_RXL_MARK(ROUTER_ALERT_IPV6, TRAP_TO_CPU, ROUTER_EXP, false),
 	MLXSW_SP_RXL_MARK(IPIP_DECAP_ERROR, TRAP_TO_CPU, ROUTER_EXP, false),
+	MLXSW_SP_RXL_MARK(DECAP_ECN0, TRAP_TO_CPU, ROUTER_EXP, false),
 	MLXSW_SP_RXL_MARK(IPV4_VRRP, TRAP_TO_CPU, ROUTER_EXP, false),
 	MLXSW_SP_RXL_MARK(IPV6_VRRP, TRAP_TO_CPU, ROUTER_EXP, false),
 	/* PKT Sample trap */
@@ -3482,6 +3484,8 @@ static const struct mlxsw_listener mlxsw_sp_listener[] = {
 	MLXSW_SP_RXL_MARK(RPF, TRAP_TO_CPU, RPF, false),
 	MLXSW_SP_RXL_MARK(ACL1, TRAP_TO_CPU, MULTICAST, false),
 	MLXSW_SP_RXL_MR_MARK(ACL2, TRAP_TO_CPU, MULTICAST, false),
+	/* NVE traps */
+	MLXSW_SP_RXL_MARK(NVE_ENCAP_ARP, TRAP_TO_CPU, ARP, false),
 };
 
 static int mlxsw_sp_cpu_policers_set(struct mlxsw_core *mlxsw_core)
@@ -3666,8 +3670,10 @@ static void mlxsw_sp_traps_fini(struct mlxsw_sp *mlxsw_sp)
 static int mlxsw_sp_lag_init(struct mlxsw_sp *mlxsw_sp)
 {
 	char slcr_pl[MLXSW_REG_SLCR_LEN];
+	u32 seed;
 	int err;
 
+	get_random_bytes(&seed, sizeof(seed));
 	mlxsw_reg_slcr_pack(slcr_pl, MLXSW_REG_SLCR_LAG_HASH_SMAC |
 				     MLXSW_REG_SLCR_LAG_HASH_DMAC |
 				     MLXSW_REG_SLCR_LAG_HASH_ETHERTYPE |
@@ -3676,7 +3682,7 @@ static int mlxsw_sp_lag_init(struct mlxsw_sp *mlxsw_sp)
 				     MLXSW_REG_SLCR_LAG_HASH_DIP |
 				     MLXSW_REG_SLCR_LAG_HASH_SPORT |
 				     MLXSW_REG_SLCR_LAG_HASH_DPORT |
-				     MLXSW_REG_SLCR_LAG_HASH_IPPROTO);
+				     MLXSW_REG_SLCR_LAG_HASH_IPPROTO, seed);
 	err = mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(slcr), slcr_pl);
 	if (err)
 		return err;

drivers/net/ethernet/mellanox/mlxsw/spectrum.h

@@ -383,6 +383,17 @@ static inline void mlxsw_sp_port_dcb_fini(struct mlxsw_sp_port *mlxsw_sp_port)
 #endif
 
 /* spectrum_router.c */
+enum mlxsw_sp_l3proto {
+	MLXSW_SP_L3_PROTO_IPV4,
+	MLXSW_SP_L3_PROTO_IPV6,
+#define MLXSW_SP_L3_PROTO_MAX	(MLXSW_SP_L3_PROTO_IPV6 + 1)
+};
+
+union mlxsw_sp_l3addr {
+	__be32 addr4;
+	struct in6_addr addr6;
+};
+
 int mlxsw_sp_router_init(struct mlxsw_sp *mlxsw_sp);
 void mlxsw_sp_router_fini(struct mlxsw_sp *mlxsw_sp);
 int mlxsw_sp_netdevice_router_port_event(struct net_device *dev);
@@ -416,6 +427,10 @@ mlxsw_sp_port_vlan_router_leave(struct mlxsw_sp_port_vlan *mlxsw_sp_port_vlan);
 void mlxsw_sp_rif_destroy(struct mlxsw_sp_rif *rif);
 void mlxsw_sp_rif_destroy_by_dev(struct mlxsw_sp *mlxsw_sp,
 				 struct net_device *dev);
+struct mlxsw_sp_rif *mlxsw_sp_rif_find_by_dev(const struct mlxsw_sp *mlxsw_sp,
+					      const struct net_device *dev);
+u8 mlxsw_sp_router_port(const struct mlxsw_sp *mlxsw_sp);
+struct mlxsw_sp_fid *mlxsw_sp_rif_fid(const struct mlxsw_sp_rif *rif);
 
 /* spectrum_kvdl.c */
 enum mlxsw_sp_kvdl_entry_type {
@@ -423,6 +438,7 @@ enum mlxsw_sp_kvdl_entry_type {
 	MLXSW_SP_KVDL_ENTRY_TYPE_ACTSET,
 	MLXSW_SP_KVDL_ENTRY_TYPE_PBS,
 	MLXSW_SP_KVDL_ENTRY_TYPE_MCRIGR,
+	MLXSW_SP_KVDL_ENTRY_TYPE_TNUMT,
 };
 
 static inline unsigned int
@@ -433,6 +449,7 @@ mlxsw_sp_kvdl_entry_size(enum mlxsw_sp_kvdl_entry_type type)
 	case MLXSW_SP_KVDL_ENTRY_TYPE_ACTSET: /* fall through */
 	case MLXSW_SP_KVDL_ENTRY_TYPE_PBS: /* fall through */
 	case MLXSW_SP_KVDL_ENTRY_TYPE_MCRIGR: /* fall through */
+	case MLXSW_SP_KVDL_ENTRY_TYPE_TNUMT: /* fall through */
 	default:
 		return 1;
 	}

drivers/net/ethernet/mellanox/mlxsw/spectrum2_kvdl.c

@@ -35,6 +35,7 @@ static const struct mlxsw_sp2_kvdl_part_info mlxsw_sp2_kvdl_parts_info[] = {
 				 MAX_KVD_ACTION_SETS),
 	MLXSW_SP2_KVDL_PART_INFO(PBS, 0x24, KVD_SIZE, KVD_SIZE),
 	MLXSW_SP2_KVDL_PART_INFO(MCRIGR, 0x26, KVD_SIZE, KVD_SIZE),
+	MLXSW_SP2_KVDL_PART_INFO(TNUMT, 0x29, KVD_SIZE, KVD_SIZE),
 };
 
 #define MLXSW_SP2_KVDL_PARTS_INFO_LEN ARRAY_SIZE(mlxsw_sp2_kvdl_parts_info)

drivers/net/ethernet/mellanox/mlxsw/spectrum_router.h

@@ -7,17 +7,6 @@
 #include "spectrum.h"
 #include "reg.h"
 
-enum mlxsw_sp_l3proto {
-	MLXSW_SP_L3_PROTO_IPV4,
-	MLXSW_SP_L3_PROTO_IPV6,
-#define MLXSW_SP_L3_PROTO_MAX	(MLXSW_SP_L3_PROTO_IPV6 + 1)
-};
-
-union mlxsw_sp_l3addr {
-	__be32 addr4;
-	struct in6_addr addr6;
-};
-
 struct mlxsw_sp_rif_ipip_lb;
 struct mlxsw_sp_rif_ipip_lb_config {
 	enum mlxsw_reg_ritr_loopback_ipip_type lb_ipipt;
@@ -35,8 +24,6 @@ struct mlxsw_sp_neigh_entry;
 struct mlxsw_sp_nexthop;
 struct mlxsw_sp_ipip_entry;
 
-struct mlxsw_sp_rif *mlxsw_sp_rif_find_by_dev(const struct mlxsw_sp *mlxsw_sp,
-					      const struct net_device *dev);
 struct mlxsw_sp_rif *mlxsw_sp_rif_by_index(const struct mlxsw_sp *mlxsw_sp,
 					   u16 rif_index);
 u16 mlxsw_sp_rif_index(const struct mlxsw_sp_rif *rif);
@@ -44,9 +31,7 @@ u16 mlxsw_sp_ipip_lb_rif_index(const struct mlxsw_sp_rif_ipip_lb *rif);
 u16 mlxsw_sp_ipip_lb_ul_vr_id(const struct mlxsw_sp_rif_ipip_lb *rif);
 u32 mlxsw_sp_ipip_dev_ul_tb_id(const struct net_device *ol_dev);
 int mlxsw_sp_rif_dev_ifindex(const struct mlxsw_sp_rif *rif);
-u8 mlxsw_sp_router_port(const struct mlxsw_sp *mlxsw_sp);
 const struct net_device *mlxsw_sp_rif_dev(const struct mlxsw_sp_rif *rif);
-struct mlxsw_sp_fid *mlxsw_sp_rif_fid(const struct mlxsw_sp_rif *rif);
 int mlxsw_sp_rif_counter_value_get(struct mlxsw_sp *mlxsw_sp,
 				   struct mlxsw_sp_rif *rif,
 				   enum mlxsw_sp_rif_counter_dir dir,

drivers/net/ethernet/mellanox/mlxsw/spectrum_switchdev.c

@@ -17,7 +17,6 @@
 #include <net/switchdev.h>
 
 #include "spectrum_span.h"
-#include "spectrum_router.h"
 #include "spectrum_switchdev.h"
 #include "spectrum.h"
 #include "core.h"
@@ -2289,7 +2288,7 @@ struct mlxsw_sp_switchdev_event_work {
 	unsigned long event;
 };
 
-static void mlxsw_sp_switchdev_event_work(struct work_struct *work)
+static void mlxsw_sp_switchdev_bridge_fdb_event_work(struct work_struct *work)
 {
 	struct mlxsw_sp_switchdev_event_work *switchdev_work =
 		container_of(work, struct mlxsw_sp_switchdev_event_work, work);
@@ -2344,16 +2343,23 @@ static int mlxsw_sp_switchdev_event(struct notifier_block *unused,
 {
 	struct net_device *dev = switchdev_notifier_info_to_dev(ptr);
 	struct mlxsw_sp_switchdev_event_work *switchdev_work;
-	struct switchdev_notifier_fdb_info *fdb_info = ptr;
+	struct switchdev_notifier_fdb_info *fdb_info;
+	struct switchdev_notifier_info *info = ptr;
+	struct net_device *br_dev;
 
-	if (!mlxsw_sp_port_dev_lower_find_rcu(dev))
+	/* Tunnel devices are not our uppers, so check their master instead */
+	br_dev = netdev_master_upper_dev_get_rcu(dev);
+	if (!br_dev)
+		return NOTIFY_DONE;
+	if (!netif_is_bridge_master(br_dev))
+		return NOTIFY_DONE;
+	if (!mlxsw_sp_port_dev_lower_find_rcu(br_dev))
 		return NOTIFY_DONE;
 
 	switchdev_work = kzalloc(sizeof(*switchdev_work), GFP_ATOMIC);
 	if (!switchdev_work)
 		return NOTIFY_BAD;
 
-	INIT_WORK(&switchdev_work->work, mlxsw_sp_switchdev_event_work);
 	switchdev_work->dev = dev;
 	switchdev_work->event = event;
 
@@ -2362,6 +2368,11 @@ static int mlxsw_sp_switchdev_event(struct notifier_block *unused,
 	case SWITCHDEV_FDB_DEL_TO_DEVICE: /* fall through */
 	case SWITCHDEV_FDB_ADD_TO_BRIDGE: /* fall through */
 	case SWITCHDEV_FDB_DEL_TO_BRIDGE:
+		fdb_info = container_of(info,
+					struct switchdev_notifier_fdb_info,
+					info);
+		INIT_WORK(&switchdev_work->work,
+			  mlxsw_sp_switchdev_bridge_fdb_event_work);
 		memcpy(&switchdev_work->fdb_info, ptr,
 		       sizeof(switchdev_work->fdb_info));
 		switchdev_work->fdb_info.addr = kzalloc(ETH_ALEN, GFP_ATOMIC);

drivers/net/ethernet/mellanox/mlxsw/trap.h

@@ -24,6 +24,7 @@ enum {
 	MLXSW_TRAP_ID_IGMP_V3_REPORT = 0x34,
 	MLXSW_TRAP_ID_PKT_SAMPLE = 0x38,
 	MLXSW_TRAP_ID_FID_MISS = 0x3D,
+	MLXSW_TRAP_ID_DECAP_ECN0 = 0x40,
 	MLXSW_TRAP_ID_ARPBC = 0x50,
 	MLXSW_TRAP_ID_ARPUC = 0x51,
 	MLXSW_TRAP_ID_MTUERROR = 0x52,
@@ -59,6 +60,7 @@ enum {
 	MLXSW_TRAP_ID_IPV6_MC_LINK_LOCAL_DEST = 0x91,
 	MLXSW_TRAP_ID_HOST_MISS_IPV6 = 0x92,
 	MLXSW_TRAP_ID_IPIP_DECAP_ERROR = 0xB1,
+	MLXSW_TRAP_ID_NVE_ENCAP_ARP = 0xBD,
 	MLXSW_TRAP_ID_ROUTER_ALERT_IPV4 = 0xD6,
 	MLXSW_TRAP_ID_ROUTER_ALERT_IPV6 = 0xD7,
 	MLXSW_TRAP_ID_ACL0 = 0x1C0,