[PATCH net-next 0/3] net/mlx5: Offload setting/matching on tunnel tos/ttl

[PATCH net-next 0/3] net/mlx5: Offload setting/matching on tunnel tos/ttl

[Or Gerlitz]

This series enables mlx5 offloading of tc eswitch rules that set tos/ttl (encap) or 
match on them (decap) for tunnels.

Or Gerlitz (3):
  net/mlx5e: Use ttl from route lookup on tc encap offload only if needed
  net/mlx5e: Support setup of tos and ttl for tunnel key TC action offload
  net/mlx5e: Offload TC matching on tos/ttl for ip tunnels

 drivers/net/ethernet/mellanox/mlx5/core/en_tc.c |   59 +++++++++++++++++-----
 1 files changed, 45 insertions(+), 14 deletions(-)

[PATCH net-next 1/3] net/mlx5e: Use ttl from route lookup on tc encap offload only if needed

[Or Gerlitz]

Currnetly, the ttl for the encapsulation headers is taken from the
route lookup result. As a pre-step to allow for an offload case when
the user specifies the ttl, take it from the route lookup only if
not zero. While here, also move to use u8 instead int for the ttl.

Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Reviewed-by: Roi Dayan <roid@mellanox.com>
---
 drivers/net/ethernet/mellanox/mlx5/core/en_tc.c |   27 ++++++++++++++--------
 1 files changed, 17 insertions(+), 10 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
index b5f854f..d00f8f8 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
@@ -2121,7 +2121,7 @@ static int mlx5e_route_lookup_ipv4(struct mlx5e_priv *priv,
 				   struct net_device **out_dev,
 				   struct flowi4 *fl4,
 				   struct neighbour **out_n,
-				   int *out_ttl)
+				   u8 *out_ttl)
 {
 	struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
 	struct mlx5e_rep_priv *uplink_rpriv;
@@ -2145,7 +2145,8 @@ static int mlx5e_route_lookup_ipv4(struct mlx5e_priv *priv,
 	else
 		*out_dev = rt->dst.dev;
 
-	*out_ttl = ip4_dst_hoplimit(&rt->dst);
+	if (!(*out_ttl))
+		*out_ttl = ip4_dst_hoplimit(&rt->dst);
 	n = dst_neigh_lookup(&rt->dst, &fl4->daddr);
 	ip_rt_put(rt);
 	if (!n)
@@ -2174,7 +2175,7 @@ static int mlx5e_route_lookup_ipv6(struct mlx5e_priv *priv,
 				   struct net_device **out_dev,
 				   struct flowi6 *fl6,
 				   struct neighbour **out_n,
-				   int *out_ttl)
+				   u8 *out_ttl)
 {
 	struct neighbour *n = NULL;
 	struct dst_entry *dst;
@@ -2189,7 +2190,8 @@ static int mlx5e_route_lookup_ipv6(struct mlx5e_priv *priv,
 	if (ret < 0)
 		return ret;
 
-	*out_ttl = ip6_dst_hoplimit(dst);
+	if (!(*out_ttl))
+		*out_ttl = ip6_dst_hoplimit(dst);
 
 	uplink_rpriv = mlx5_eswitch_get_uplink_priv(esw, REP_ETH);
 	/* if the egress device isn't on the same HW e-switch, we use the uplink */
@@ -2213,7 +2215,7 @@ static int mlx5e_route_lookup_ipv6(struct mlx5e_priv *priv,
 static void gen_vxlan_header_ipv4(struct net_device *out_dev,
 				  char buf[], int encap_size,
 				  unsigned char h_dest[ETH_ALEN],
-				  int ttl,
+				  u8 ttl,
 				  __be32 daddr,
 				  __be32 saddr,
 				  __be16 udp_dst_port,
@@ -2246,7 +2248,7 @@ static void gen_vxlan_header_ipv4(struct net_device *out_dev,
 static void gen_vxlan_header_ipv6(struct net_device *out_dev,
 				  char buf[], int encap_size,
 				  unsigned char h_dest[ETH_ALEN],
-				  int ttl,
+				  u8 ttl,
 				  struct in6_addr *daddr,
 				  struct in6_addr *saddr,
 				  __be16 udp_dst_port,
@@ -2286,8 +2288,8 @@ static int mlx5e_create_encap_header_ipv4(struct mlx5e_priv *priv,
 	struct neighbour *n = NULL;
 	struct flowi4 fl4 = {};
 	char *encap_header;
-	int ttl, err;
-	u8 nud_state;
+	u8 nud_state, ttl;
+	int err;
 
 	if (max_encap_size < ipv4_encap_size) {
 		mlx5_core_warn(priv->mdev, "encap size %d too big, max supported is %d\n",
@@ -2308,6 +2310,9 @@ static int mlx5e_create_encap_header_ipv4(struct mlx5e_priv *priv,
 		err = -EOPNOTSUPP;
 		goto free_encap;
 	}
+
+	ttl = 0;
+
 	fl4.flowi4_tos = tun_key->tos;
 	fl4.daddr = tun_key->u.ipv4.dst;
 	fl4.saddr = tun_key->u.ipv4.src;
@@ -2391,8 +2396,8 @@ static int mlx5e_create_encap_header_ipv6(struct mlx5e_priv *priv,
 	struct neighbour *n = NULL;
 	struct flowi6 fl6 = {};
 	char *encap_header;
-	int err, ttl = 0;
-	u8 nud_state;
+	u8 ttl, nud_state;
+	int err;
 
 	if (max_encap_size < ipv6_encap_size) {
 		mlx5_core_warn(priv->mdev, "encap size %d too big, max supported is %d\n",
@@ -2414,6 +2419,8 @@ static int mlx5e_create_encap_header_ipv6(struct mlx5e_priv *priv,
 		goto free_encap;
 	}
 
+	ttl = 0;
+
 	fl6.flowlabel = ip6_make_flowinfo(RT_TOS(tun_key->tos), tun_key->label);
 	fl6.daddr = tun_key->u.ipv6.dst;
 	fl6.saddr = tun_key->u.ipv6.src;
-- 
1.7.1

[PATCH net-next 2/3] net/mlx5e: Support setup of tos and ttl for tunnel key TC action offload

[Or Gerlitz]

Use the values provided by user-space for the encapsulation headers.

Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Reviewed-by: Roi Dayan <roid@mellanox.com>
---
 drivers/net/ethernet/mellanox/mlx5/core/en_tc.c |   21 ++++++++++++---------
 1 files changed, 12 insertions(+), 9 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
index d00f8f8..de23758 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
@@ -2215,7 +2215,7 @@ static int mlx5e_route_lookup_ipv6(struct mlx5e_priv *priv,
 static void gen_vxlan_header_ipv4(struct net_device *out_dev,
 				  char buf[], int encap_size,
 				  unsigned char h_dest[ETH_ALEN],
-				  u8 ttl,
+				  u8 tos, u8 ttl,
 				  __be32 daddr,
 				  __be32 saddr,
 				  __be16 udp_dst_port,
@@ -2235,6 +2235,7 @@ static void gen_vxlan_header_ipv4(struct net_device *out_dev,
 	ip->daddr = daddr;
 	ip->saddr = saddr;
 
+	ip->tos = tos;
 	ip->ttl = ttl;
 	ip->protocol = IPPROTO_UDP;
 	ip->version = 0x4;
@@ -2248,7 +2249,7 @@ static void gen_vxlan_header_ipv4(struct net_device *out_dev,
 static void gen_vxlan_header_ipv6(struct net_device *out_dev,
 				  char buf[], int encap_size,
 				  unsigned char h_dest[ETH_ALEN],
-				  u8 ttl,
+				  u8 tos, u8 ttl,
 				  struct in6_addr *daddr,
 				  struct in6_addr *saddr,
 				  __be16 udp_dst_port,
@@ -2265,7 +2266,7 @@ static void gen_vxlan_header_ipv6(struct net_device *out_dev,
 	ether_addr_copy(eth->h_source, out_dev->dev_addr);
 	eth->h_proto = htons(ETH_P_IPV6);
 
-	ip6_flow_hdr(ip6h, 0, 0);
+	ip6_flow_hdr(ip6h, tos, 0);
 	/* the HW fills up ipv6 payload len */
 	ip6h->nexthdr     = IPPROTO_UDP;
 	ip6h->hop_limit   = ttl;
@@ -2287,8 +2288,8 @@ static int mlx5e_create_encap_header_ipv4(struct mlx5e_priv *priv,
 	struct net_device *out_dev;
 	struct neighbour *n = NULL;
 	struct flowi4 fl4 = {};
+	u8 nud_state, tos, ttl;
 	char *encap_header;
-	u8 nud_state, ttl;
 	int err;
 
 	if (max_encap_size < ipv4_encap_size) {
@@ -2311,7 +2312,8 @@ static int mlx5e_create_encap_header_ipv4(struct mlx5e_priv *priv,
 		goto free_encap;
 	}
 
-	ttl = 0;
+	tos = tun_key->tos;
+	ttl = tun_key->ttl;
 
 	fl4.flowi4_tos = tun_key->tos;
 	fl4.daddr = tun_key->u.ipv4.dst;
@@ -2347,7 +2349,7 @@ static int mlx5e_create_encap_header_ipv4(struct mlx5e_priv *priv,
 	switch (e->tunnel_type) {
 	case MLX5_HEADER_TYPE_VXLAN:
 		gen_vxlan_header_ipv4(out_dev, encap_header,
-				      ipv4_encap_size, e->h_dest, ttl,
+				      ipv4_encap_size, e->h_dest, tos, ttl,
 				      fl4.daddr,
 				      fl4.saddr, tun_key->tp_dst,
 				      tunnel_id_to_key32(tun_key->tun_id));
@@ -2395,8 +2397,8 @@ static int mlx5e_create_encap_header_ipv6(struct mlx5e_priv *priv,
 	struct net_device *out_dev;
 	struct neighbour *n = NULL;
 	struct flowi6 fl6 = {};
+	u8 nud_state, tos, ttl;
 	char *encap_header;
-	u8 ttl, nud_state;
 	int err;
 
 	if (max_encap_size < ipv6_encap_size) {
@@ -2419,7 +2421,8 @@ static int mlx5e_create_encap_header_ipv6(struct mlx5e_priv *priv,
 		goto free_encap;
 	}
 
-	ttl = 0;
+	tos = tun_key->tos;
+	ttl = tun_key->ttl;
 
 	fl6.flowlabel = ip6_make_flowinfo(RT_TOS(tun_key->tos), tun_key->label);
 	fl6.daddr = tun_key->u.ipv6.dst;
@@ -2455,7 +2458,7 @@ static int mlx5e_create_encap_header_ipv6(struct mlx5e_priv *priv,
 	switch (e->tunnel_type) {
 	case MLX5_HEADER_TYPE_VXLAN:
 		gen_vxlan_header_ipv6(out_dev, encap_header,
-				      ipv6_encap_size, e->h_dest, ttl,
+				      ipv6_encap_size, e->h_dest, tos, ttl,
 				      &fl6.daddr,
 				      &fl6.saddr, tun_key->tp_dst,
 				      tunnel_id_to_key32(tun_key->tun_id));
-- 
1.7.1

[PATCH net-next 3/3] net/mlx5e: Offload TC matching on tos/ttl for ip tunnels

[Or Gerlitz]

Enable offloading of TC matching on tos/ttl for ipv4/6 tunnels.

Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Reviewed-by: Roi Dayan <roid@mellanox.com>
---
 drivers/net/ethernet/mellanox/mlx5/core/en_tc.c |   23 ++++++++++++++++++++++-
 1 files changed, 22 insertions(+), 1 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
index de23758..bdec95d 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
@@ -1207,6 +1207,26 @@ static int parse_tunnel_attr(struct mlx5e_priv *priv,
 		MLX5_SET(fte_match_set_lyr_2_4, headers_v, ethertype, ETH_P_IPV6);
 	}
 
+	if (dissector_uses_key(f->dissector, FLOW_DISSECTOR_KEY_ENC_IP)) {
+		struct flow_dissector_key_ip *key =
+			skb_flow_dissector_target(f->dissector,
+						  FLOW_DISSECTOR_KEY_ENC_IP,
+						  f->key);
+		struct flow_dissector_key_ip *mask =
+			skb_flow_dissector_target(f->dissector,
+						  FLOW_DISSECTOR_KEY_ENC_IP,
+						  f->mask);
+
+		MLX5_SET(fte_match_set_lyr_2_4, headers_c, ip_ecn, mask->tos & 0x3);
+		MLX5_SET(fte_match_set_lyr_2_4, headers_v, ip_ecn, key->tos & 0x3);
+
+		MLX5_SET(fte_match_set_lyr_2_4, headers_c, ip_dscp, mask->tos >> 2);
+		MLX5_SET(fte_match_set_lyr_2_4, headers_v, ip_dscp, key->tos  >> 2);
+
+		MLX5_SET(fte_match_set_lyr_2_4, headers_c, ttl_hoplimit, mask->ttl);
+		MLX5_SET(fte_match_set_lyr_2_4, headers_v, ttl_hoplimit, key->ttl);
+	}
+
 	/* Enforce DMAC when offloading incoming tunneled flows.
 	 * Flow counters require a match on the DMAC.
 	 */
@@ -1255,7 +1275,8 @@ static int __parse_cls_flower(struct mlx5e_priv *priv,
 	      BIT(FLOW_DISSECTOR_KEY_ENC_PORTS)	|
 	      BIT(FLOW_DISSECTOR_KEY_ENC_CONTROL) |
 	      BIT(FLOW_DISSECTOR_KEY_TCP) |
-	      BIT(FLOW_DISSECTOR_KEY_IP))) {
+	      BIT(FLOW_DISSECTOR_KEY_IP)  |
+	      BIT(FLOW_DISSECTOR_KEY_ENC_IP))) {
 		netdev_warn(priv->netdev, "Unsupported key used: 0x%x\n",
 			    f->dissector->used_keys);
 		return -EOPNOTSUPP;
-- 
1.7.1

Re: [PATCH net-next 0/3] net/mlx5: Offload setting/matching on tunnel tos/ttl

[David Miller]

From: Or Gerlitz <ogerlitz@mellanox.com>
Date: Tue, 24 Jul 2018 13:59:32 +0300

> This series enables mlx5 offloading of tc eswitch rules that set
> tos/ttl (encap) or match on them (decap) for tunnels.

Series applied, thanks.