On 2020-06-18 11:36 AM, xiangxia.m....@gmail.com wrote:
> From: Tonghao Zhang <xiangxia.m....@gmail.com>
>
> The commit 0a7fcb78cc21 ("net/mlx5e: Support inner header rewrite with
> goto action"), will decapsulate the tunnel packets if there is a goto
> action in chain 0. But in some case, we don't want do that, for example:
>
> $ tc filter add dev $VXLAN protocol ip parent ffff: prio 1 chain 0 \
> flower enc_dst_ip 2.2.2.100 enc_dst_port 4789 \
> action goto chain 2
> $ tc filter add dev $VXLAN protocol ip parent ffff: prio 1 chain 2 \
> flower dst_mac 00:11:22:33:44:55 enc_src_ip 2.2.2.200 \
> enc_dst_ip 2.2.2.100 enc_dst_port 4789 enc_key_id 100 \
> action tunnel_key unset action mirred egress redirect dev enp130s0f0_0
> $ tc filter add dev $VXLAN protocol ip parent ffff: prio 1 chain 2 \
> flower dst_mac 00:11:22:33:44:66 enc_src_ip 2.2.2.200 \
> enc_dst_ip 2.2.2.100 enc_dst_port 4789 enc_key_id 200 \
> action tunnel_key unset action mirred egress redirect dev enp130s0f0_1
>
> If there are pedit and goto actions, do the decapsulate and id mapping action.
>
Hi Tonghao,
I think you might missed Paul's comments on V2?
Thanks,
Roi
> Signed-off-by: Tonghao Zhang <xiangxia.m....@gmail.com>
> ---
> .../ethernet/mellanox/mlx5/core/en/mapping.c | 24 ++++
> .../ethernet/mellanox/mlx5/core/en/mapping.h | 1 +
> .../net/ethernet/mellanox/mlx5/core/en_tc.c | 109 ++++++++++++------
> 3 files changed, 99 insertions(+), 35 deletions(-)
>
> diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/mapping.c
> b/drivers/net/ethernet/mellanox/mlx5/core/en/mapping.c
> index ea321e528749..90306dde6b60 100644
> --- a/drivers/net/ethernet/mellanox/mlx5/core/en/mapping.c
> +++ b/drivers/net/ethernet/mellanox/mlx5/core/en/mapping.c
> @@ -74,6 +74,30 @@ int mapping_add(struct mapping_ctx *ctx, void *data, u32
> *id)
> return err;
> }
>
> +int mapping_find_by_data(struct mapping_ctx *ctx, void *data, u32 *id)
> +{
> + struct mapping_item *mi;
> + u32 hash_key;
> +
> + mutex_lock(&ctx->lock);
> +
> + hash_key = jhash(data, ctx->data_size, 0);
> + hash_for_each_possible(ctx->ht, mi, node, hash_key) {
> + if (!memcmp(data, mi->data, ctx->data_size))
> + goto found;
> + }
> +
> + mutex_unlock(&ctx->lock);
> + return -ENOENT;
> +
> +found:
> + if (id)
> + *id = mi->id;
> +
> + mutex_unlock(&ctx->lock);
> + return 0;
> +}
> +
> static void mapping_remove_and_free(struct mapping_ctx *ctx,
> struct mapping_item *mi)
> {
> diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/mapping.h
> b/drivers/net/ethernet/mellanox/mlx5/core/en/mapping.h
> index 285525cc5470..af501c9796b7 100644
> --- a/drivers/net/ethernet/mellanox/mlx5/core/en/mapping.h
> +++ b/drivers/net/ethernet/mellanox/mlx5/core/en/mapping.h
> @@ -9,6 +9,7 @@ struct mapping_ctx;
> int mapping_add(struct mapping_ctx *ctx, void *data, u32 *id);
> int mapping_remove(struct mapping_ctx *ctx, u32 id);
> int mapping_find(struct mapping_ctx *ctx, u32 id, void *data);
> +int mapping_find_by_data(struct mapping_ctx *ctx, void *data, u32 *id);
>
> /* mapping uses an xarray to map data to ids in add(), and for find().
> * For locking, it uses a internal xarray spin lock for add()/remove(),
> diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
> b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
> index 7fc84f58e28a..05f8df8b53af 100644
> --- a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
> +++ b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
> @@ -1836,7 +1836,8 @@ static void mlx5e_tc_del_flow(struct mlx5e_priv *priv,
> }
> }
>
> -static int flow_has_tc_fwd_action(struct flow_cls_offload *f)
> +static int flow_has_tc_action(struct flow_cls_offload *f,
> + enum flow_action_id action)
> {
> struct flow_rule *rule = flow_cls_offload_flow_rule(f);
> struct flow_action *flow_action = &rule->action;
> @@ -1844,12 +1845,8 @@ static int flow_has_tc_fwd_action(struct
> flow_cls_offload *f)
> int i;
>
> flow_action_for_each(i, act, flow_action) {
> - switch (act->id) {
> - case FLOW_ACTION_GOTO:
> + if (act->id == action)
> return true;
> - default:
> - continue;
> - }
> }
>
> return false;
> @@ -1901,10 +1898,37 @@ enc_opts_is_dont_care_or_full_match(struct mlx5e_priv
> *priv,
> sizeof(*__dst));\
> })
>
> +static void mlx5e_make_tunnel_match_key(struct flow_cls_offload *f,
> + struct net_device *filter_dev,
> + struct tunnel_match_key *tunnel_key)
> +{
> + struct flow_rule *rule = flow_cls_offload_flow_rule(f);
> +
> + memset(tunnel_key, 0, sizeof(*tunnel_key));
> + COPY_DISSECTOR(rule, FLOW_DISSECTOR_KEY_ENC_CONTROL,
> + &tunnel_key->enc_control);
> + if (tunnel_key->enc_control.addr_type == FLOW_DISSECTOR_KEY_IPV4_ADDRS)
> + COPY_DISSECTOR(rule, FLOW_DISSECTOR_KEY_ENC_IPV4_ADDRS,
> + &tunnel_key->enc_ipv4);
> + else
> + COPY_DISSECTOR(rule, FLOW_DISSECTOR_KEY_ENC_IPV6_ADDRS,
> + &tunnel_key->enc_ipv6);
> +
> + COPY_DISSECTOR(rule, FLOW_DISSECTOR_KEY_ENC_IP, &tunnel_key->enc_ip);
> + COPY_DISSECTOR(rule, FLOW_DISSECTOR_KEY_ENC_PORTS,
> + &tunnel_key->enc_tp);
> + COPY_DISSECTOR(rule, FLOW_DISSECTOR_KEY_ENC_KEYID,
> + &tunnel_key->enc_key_id);
> +
> + tunnel_key->filter_ifindex = filter_dev->ifindex;
> +}
> +
> static int mlx5e_get_flow_tunnel_id(struct mlx5e_priv *priv,
> struct mlx5e_tc_flow *flow,
> struct flow_cls_offload *f,
> - struct net_device *filter_dev)
> + struct net_device *filter_dev,
> + bool sets_mapping,
> + bool needs_mapping)
> {
> struct flow_rule *rule = flow_cls_offload_flow_rule(f);
> struct netlink_ext_ack *extack = f->common.extack;
> @@ -1925,22 +1949,7 @@ static int mlx5e_get_flow_tunnel_id(struct mlx5e_priv
> *priv,
> uplink_rpriv = mlx5_eswitch_get_uplink_priv(esw, REP_ETH);
> uplink_priv = &uplink_rpriv->uplink_priv;
>
> - memset(&tunnel_key, 0, sizeof(tunnel_key));
> - COPY_DISSECTOR(rule, FLOW_DISSECTOR_KEY_ENC_CONTROL,
> - &tunnel_key.enc_control);
> - if (tunnel_key.enc_control.addr_type == FLOW_DISSECTOR_KEY_IPV4_ADDRS)
> - COPY_DISSECTOR(rule, FLOW_DISSECTOR_KEY_ENC_IPV4_ADDRS,
> - &tunnel_key.enc_ipv4);
> - else
> - COPY_DISSECTOR(rule, FLOW_DISSECTOR_KEY_ENC_IPV6_ADDRS,
> - &tunnel_key.enc_ipv6);
> - COPY_DISSECTOR(rule, FLOW_DISSECTOR_KEY_ENC_IP, &tunnel_key.enc_ip);
> - COPY_DISSECTOR(rule, FLOW_DISSECTOR_KEY_ENC_PORTS,
> - &tunnel_key.enc_tp);
> - COPY_DISSECTOR(rule, FLOW_DISSECTOR_KEY_ENC_KEYID,
> - &tunnel_key.enc_key_id);
> - tunnel_key.filter_ifindex = filter_dev->ifindex;
> -
> + mlx5e_make_tunnel_match_key(f, filter_dev, &tunnel_key);
> err = mapping_add(uplink_priv->tunnel_mapping, &tunnel_key, &tun_id);
> if (err)
> return err;
> @@ -1970,10 +1979,10 @@ static int mlx5e_get_flow_tunnel_id(struct mlx5e_priv
> *priv,
> mask = enc_opts_id ? TUNNEL_ID_MASK :
> (TUNNEL_ID_MASK & ~ENC_OPTS_BITS_MASK);
>
> - if (attr->chain) {
> + if (needs_mapping) {
> mlx5e_tc_match_to_reg_match(&attr->parse_attr->spec,
> TUNNEL_TO_REG, value, mask);
> - } else {
> + } else if (sets_mapping) {
> mod_hdr_acts = &attr->parse_attr->mod_hdr_acts;
> err = mlx5e_tc_match_to_reg_set(priv->mdev,
> mod_hdr_acts,
> @@ -1996,6 +2005,25 @@ static int mlx5e_get_flow_tunnel_id(struct mlx5e_priv
> *priv,
> return err;
> }
>
> +static int mlx5e_lookup_flow_tunnel_id(struct mlx5e_priv *priv,
> + struct mlx5e_tc_flow *flow,
> + struct flow_cls_offload *f,
> + struct net_device *filter_dev,
> + u32 *tun_id)
> +{
> + struct mlx5_rep_uplink_priv *uplink_priv;
> + struct mlx5e_rep_priv *uplink_rpriv;
> + struct tunnel_match_key tunnel_key;
> + struct mlx5_eswitch *esw;
> +
> + esw = priv->mdev->priv.eswitch;
> + uplink_rpriv = mlx5_eswitch_get_uplink_priv(esw, REP_ETH);
> + uplink_priv = &uplink_rpriv->uplink_priv;
> +
> + mlx5e_make_tunnel_match_key(f, filter_dev, &tunnel_key);
> + return mapping_find_by_data(uplink_priv->tunnel_mapping, &tunnel_key,
> tun_id);
> +}
> +
> static void mlx5e_put_flow_tunnel_id(struct mlx5e_tc_flow *flow)
> {
> u32 enc_opts_id = flow->tunnel_id & ENC_OPTS_BITS_MASK;
> @@ -2057,13 +2085,19 @@ static int parse_tunnel_attr(struct mlx5e_priv *priv,
> struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
> struct netlink_ext_ack *extack = f->common.extack;
> bool needs_mapping, sets_mapping;
> + bool pedit_action;
> int err;
>
> if (!mlx5e_is_eswitch_flow(flow))
> return -EOPNOTSUPP;
>
> - needs_mapping = !!flow->esw_attr->chain;
> - sets_mapping = !flow->esw_attr->chain && flow_has_tc_fwd_action(f);
> + pedit_action = flow_has_tc_action(f, FLOW_ACTION_MANGLE) ||
> + flow_has_tc_action(f, FLOW_ACTION_ADD);
> + sets_mapping = pedit_action &&
> + flow_has_tc_action(f, FLOW_ACTION_GOTO);
> + needs_mapping = !!flow->esw_attr->chain &&
> + !mlx5e_lookup_flow_tunnel_id(priv, flow, f,
> + filter_dev, NULL);
> *match_inner = !needs_mapping;
>
> if ((needs_mapping || sets_mapping) &&
> @@ -2075,7 +2109,7 @@ static int parse_tunnel_attr(struct mlx5e_priv *priv,
> return -EOPNOTSUPP;
> }
>
> - if (!flow->esw_attr->chain) {
> + if (*match_inner) {
> err = mlx5e_tc_tun_parse(filter_dev, priv, spec, f,
> match_level);
> if (err) {
> @@ -2085,18 +2119,20 @@ static int parse_tunnel_attr(struct mlx5e_priv *priv,
> "Failed to parse tunnel attributes");
> return err;
> }
> -
> - /* With mpls over udp we decapsulate using packet reformat
> - * object
> - */
> - if (!netif_is_bareudp(filter_dev))
> - flow->esw_attr->action |=
> MLX5_FLOW_CONTEXT_ACTION_DECAP;
> }
>
> + /* With mpls over udp we decapsulate using packet reformat
> + * object
> + */
> + if (!netif_is_bareudp(filter_dev) &&
> + sets_mapping && !needs_mapping)
> + flow->esw_attr->action |= MLX5_FLOW_CONTEXT_ACTION_DECAP;
> +
> if (!needs_mapping && !sets_mapping)
> return 0;
>
> - return mlx5e_get_flow_tunnel_id(priv, flow, f, filter_dev);
> + return mlx5e_get_flow_tunnel_id(priv, flow, f, filter_dev,
> + sets_mapping, needs_mapping);
> }
>
> static void *get_match_inner_headers_criteria(struct mlx5_flow_spec *spec)
> @@ -4309,6 +4345,9 @@ static int parse_tc_fdb_actions(struct mlx5e_priv *priv,
> attr->action |= MLX5_FLOW_CONTEXT_ACTION_FWD_DEST;
> }
>
> + if (decap)
> + attr->action |= MLX5_FLOW_CONTEXT_ACTION_DECAP;
> +
> if (!(attr->action &
> (MLX5_FLOW_CONTEXT_ACTION_FWD_DEST |
> MLX5_FLOW_CONTEXT_ACTION_DROP))) {
> NL_SET_ERR_MSG_MOD(extack,
>
