net/mlx5e: Ethtool steering, ip6 support
authorSaeed Mahameed <saeedm@mellanox.com>
Fri, 6 Jul 2018 00:11:09 +0000 (17:11 -0700)
committerSaeed Mahameed <saeedm@mellanox.com>
Mon, 13 Aug 2018 19:50:12 +0000 (12:50 -0700)
Add ip6 support for ethtool flow steering.

New supported flow types: ip6|tcp6|udp6|
Supported fields: src-ip|dst-ip|src-port|dst-port

Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
drivers/net/ethernet/mellanox/mlx5/core/en_fs_ethtool.c

index cf2491a..c81ab21 100644 (file)
@@ -66,11 +66,14 @@ static struct mlx5e_ethtool_table *get_flow_table(struct mlx5e_priv *priv,
        switch (fs->flow_type & ~(FLOW_EXT | FLOW_MAC_EXT)) {
        case TCP_V4_FLOW:
        case UDP_V4_FLOW:
+       case TCP_V6_FLOW:
+       case UDP_V6_FLOW:
                max_tuples = ETHTOOL_NUM_L3_L4_FTS;
                prio = MLX5E_ETHTOOL_L3_L4_PRIO + (max_tuples - num_tuples);
                eth_ft = &priv->fs.ethtool.l3_l4_ft[prio];
                break;
        case IP_USER_FLOW:
+       case IPV6_USER_FLOW:
                max_tuples = ETHTOOL_NUM_L3_L4_FTS;
                prio = MLX5E_ETHTOOL_L3_L4_PRIO + (max_tuples - num_tuples);
                eth_ft = &priv->fs.ethtool.l3_l4_ft[prio];
@@ -143,6 +146,29 @@ set_ip4(void *headers_c, void *headers_v, __be32 ip4src_m,
 }
 
 static void
+set_ip6(void *headers_c, void *headers_v, __be32 ip6src_m[4],
+       __be32 ip6src_v[4], __be32 ip6dst_m[4], __be32 ip6dst_v[4])
+{
+       u8 ip6_sz = MLX5_FLD_SZ_BYTES(ipv6_layout, ipv6);
+
+       if (!ipv6_addr_any((struct in6_addr *)ip6src_m)) {
+               memcpy(MLX5E_FTE_ADDR_OF(headers_v, src_ipv4_src_ipv6.ipv6_layout.ipv6),
+                      ip6src_v, ip6_sz);
+               memcpy(MLX5E_FTE_ADDR_OF(headers_c, src_ipv4_src_ipv6.ipv6_layout.ipv6),
+                      ip6src_m, ip6_sz);
+       }
+       if (!ipv6_addr_any((struct in6_addr *)ip6dst_m)) {
+               memcpy(MLX5E_FTE_ADDR_OF(headers_v, dst_ipv4_dst_ipv6.ipv6_layout.ipv6),
+                      ip6dst_v, ip6_sz);
+               memcpy(MLX5E_FTE_ADDR_OF(headers_c, dst_ipv4_dst_ipv6.ipv6_layout.ipv6),
+                      ip6dst_m, ip6_sz);
+       }
+
+       MLX5E_FTE_SET(headers_c, ethertype, 0xffff);
+       MLX5E_FTE_SET(headers_v, ethertype, ETH_P_IPV6);
+}
+
+static void
 set_tcp(void *headers_c, void *headers_v, __be16 psrc_m, __be16 psrc_v,
        __be16 pdst_m, __be16 pdst_v)
 {
@@ -214,6 +240,42 @@ parse_ip4(void *headers_c, void *headers_v, struct ethtool_rx_flow_spec *fs)
 }
 
 static void
+parse_ip6(void *headers_c, void *headers_v, struct ethtool_rx_flow_spec *fs)
+{
+       struct ethtool_usrip6_spec *l3_mask = &fs->m_u.usr_ip6_spec;
+       struct ethtool_usrip6_spec *l3_val  = &fs->h_u.usr_ip6_spec;
+
+       set_ip6(headers_c, headers_v, l3_mask->ip6src,
+               l3_val->ip6src, l3_mask->ip6dst, l3_val->ip6dst);
+}
+
+static void
+parse_tcp6(void *headers_c, void *headers_v, struct ethtool_rx_flow_spec *fs)
+{
+       struct ethtool_tcpip6_spec *l4_mask = &fs->m_u.tcp_ip6_spec;
+       struct ethtool_tcpip6_spec *l4_val  = &fs->h_u.tcp_ip6_spec;
+
+       set_ip6(headers_c, headers_v, l4_mask->ip6src,
+               l4_val->ip6src, l4_mask->ip6dst, l4_val->ip6dst);
+
+       set_tcp(headers_c, headers_v, l4_mask->psrc, l4_val->psrc,
+               l4_mask->pdst, l4_val->pdst);
+}
+
+static void
+parse_udp6(void *headers_c, void *headers_v, struct ethtool_rx_flow_spec *fs)
+{
+       struct ethtool_tcpip6_spec *l4_mask = &fs->m_u.udp_ip6_spec;
+       struct ethtool_tcpip6_spec *l4_val  = &fs->h_u.udp_ip6_spec;
+
+       set_ip6(headers_c, headers_v, l4_mask->ip6src,
+               l4_val->ip6src, l4_mask->ip6dst, l4_val->ip6dst);
+
+       set_udp(headers_c, headers_v, l4_mask->psrc, l4_val->psrc,
+               l4_mask->pdst, l4_val->pdst);
+}
+
+static void
 parse_ether(void *headers_c, void *headers_v, struct ethtool_rx_flow_spec *fs)
 {
        struct ethhdr *eth_mask = &fs->m_u.ether_spec;
@@ -264,6 +326,15 @@ static int set_flow_attrs(u32 *match_c, u32 *match_v,
        case IP_USER_FLOW:
                parse_ip4(outer_headers_c, outer_headers_v, fs);
                break;
+       case TCP_V6_FLOW:
+               parse_tcp6(outer_headers_c, outer_headers_v, fs);
+               break;
+       case UDP_V6_FLOW:
+               parse_udp6(outer_headers_c, outer_headers_v, fs);
+               break;
+       case IPV6_USER_FLOW:
+               parse_ip6(outer_headers_c, outer_headers_v, fs);
+               break;
        case ETHER_FLOW:
                parse_ether(outer_headers_c, outer_headers_v, fs);
                break;
@@ -473,6 +544,51 @@ static int validate_ip4(struct ethtool_rx_flow_spec *fs)
        return ++ntuples;
 }
 
+static int validate_ip6(struct ethtool_rx_flow_spec *fs)
+{
+       struct ethtool_usrip6_spec *l3_mask = &fs->m_u.usr_ip6_spec;
+       int ntuples = 0;
+
+       if (l3_mask->l4_4_bytes || l3_mask->tclass || l3_mask->l4_proto)
+               return -EINVAL;
+       if (!ipv6_addr_any((struct in6_addr *)l3_mask->ip6src))
+               ntuples++;
+
+       if (!ipv6_addr_any((struct in6_addr *)l3_mask->ip6dst))
+               ntuples++;
+
+       /* Flow is IPv6 */
+       return ++ntuples;
+}
+
+static int validate_tcpudp6(struct ethtool_rx_flow_spec *fs)
+{
+       struct ethtool_tcpip6_spec *l4_mask = &fs->m_u.tcp_ip6_spec;
+       int ntuples = 0;
+
+       if (l4_mask->tclass)
+               return -EINVAL;
+
+       if (!ipv6_addr_any((struct in6_addr *)l4_mask->ip6src))
+               ntuples++;
+
+       if (!ipv6_addr_any((struct in6_addr *)l4_mask->ip6dst))
+               ntuples++;
+
+       if (l4_mask->psrc) {
+               if (!all_ones(l4_mask->psrc))
+                       return -EINVAL;
+               ntuples++;
+       }
+       if (l4_mask->pdst) {
+               if (!all_ones(l4_mask->pdst))
+                       return -EINVAL;
+               ntuples++;
+       }
+       /* Flow is TCP/UDP */
+       return ++ntuples;
+}
+
 static int validate_vlan(struct ethtool_rx_flow_spec *fs)
 {
        if (fs->m_ext.vlan_etype ||
@@ -516,6 +632,19 @@ static int validate_flow(struct mlx5e_priv *priv,
                        return ret;
                num_tuples += ret;
                break;
+       case TCP_V6_FLOW:
+       case UDP_V6_FLOW:
+               ret = validate_tcpudp6(fs);
+               if (ret < 0)
+                       return ret;
+               num_tuples += ret;
+               break;
+       case IPV6_USER_FLOW:
+               ret = validate_ip6(fs);
+               if (ret < 0)
+                       return ret;
+               num_tuples += ret;
+               break;
        default:
                return -ENOTSUPP;
        }