From: Jianbo Liu In the commit 5e466345291a ("net/mlx5e: IPsec: Add IPsec steering in local NIC RX"), the decrypted packets are handled in RX error flow table. There is only one rule in the table, which forwards packets to the default ESP TIR. This patch updates the design to allow RSS after decryption. For ESP traffic, SPI and IP addresses are the fields selected for RSS hash, and it's common that only one SPI is configured in RX direction, so RSS can't work properly as all the packets are hashed to one key in this case. To take advantage of RSS and improve performance, the decrypted packets need to be forwarded back to TTC table, where RSS can work based on the decrypted packet types. Signed-off-by: Jianbo Liu Reviewed-by: Dragos Tatulea Signed-off-by: Tariq Toukan --- .../mellanox/mlx5/core/en_accel/ipsec_fs.c | 24 +++++++++++++++---- .../mellanox/mlx5/core/lib/ipsec_fs_roce.c | 4 ++++ 2 files changed, 23 insertions(+), 5 deletions(-) diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_fs.c b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_fs.c index 98b6a3a623f9..a06929852296 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_fs.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_fs.c @@ -585,6 +585,20 @@ static int ipsec_miss_create(struct mlx5_core_dev *mdev, return err; } +static struct mlx5_flow_destination +ipsec_rx_decrypted_pkt_def_dest(struct mlx5_ttc_table *ttc, u32 family) +{ + struct mlx5_flow_destination dest; + + if (!mlx5_ttc_has_esp_flow_group(ttc)) + return mlx5_ttc_get_default_dest(ttc, family2tt(family)); + + dest.ft = mlx5_get_ttc_flow_table(ttc); + dest.type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE; + + return dest; +} + static void ipsec_rx_update_default_dest(struct mlx5e_ipsec_rx *rx, struct mlx5_flow_destination *old_dest, struct mlx5_flow_destination *new_dest) @@ -598,10 +612,10 @@ static void handle_ipsec_rx_bringup(struct mlx5e_ipsec *ipsec, u32 family) { struct mlx5e_ipsec_rx *rx = ipsec_rx(ipsec, family, XFRM_DEV_OFFLOAD_PACKET); struct mlx5_flow_namespace *ns = mlx5e_fs_get_ns(ipsec->fs, false); + struct mlx5_ttc_table *ttc = mlx5e_fs_get_ttc(ipsec->fs, false); struct mlx5_flow_destination old_dest, new_dest; - old_dest = mlx5_ttc_get_default_dest(mlx5e_fs_get_ttc(ipsec->fs, false), - family2tt(family)); + old_dest = ipsec_rx_decrypted_pkt_def_dest(ttc, family); mlx5_ipsec_fs_roce_rx_create(ipsec->mdev, ipsec->roce, ns, &old_dest, family, MLX5E_ACCEL_FS_ESP_FT_ROCE_LEVEL, MLX5E_NIC_PRIO); @@ -614,12 +628,12 @@ static void handle_ipsec_rx_bringup(struct mlx5e_ipsec *ipsec, u32 family) static void handle_ipsec_rx_cleanup(struct mlx5e_ipsec *ipsec, u32 family) { struct mlx5e_ipsec_rx *rx = ipsec_rx(ipsec, family, XFRM_DEV_OFFLOAD_PACKET); + struct mlx5_ttc_table *ttc = mlx5e_fs_get_ttc(ipsec->fs, false); struct mlx5_flow_destination old_dest, new_dest; old_dest.ft = mlx5_ipsec_fs_roce_ft_get(ipsec->roce, family); old_dest.type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE; - new_dest = mlx5_ttc_get_default_dest(mlx5e_fs_get_ttc(ipsec->fs, false), - family2tt(family)); + new_dest = ipsec_rx_decrypted_pkt_def_dest(ttc, family); ipsec_rx_update_default_dest(rx, &old_dest, &new_dest); mlx5_ipsec_fs_roce_rx_destroy(ipsec->roce, family, ipsec->mdev); @@ -763,7 +777,7 @@ static int ipsec_rx_status_pass_dest_get(struct mlx5e_ipsec *ipsec, if (rx == ipsec->rx_esw) return mlx5_esw_ipsec_rx_status_pass_dest_get(ipsec, dest); - *dest = mlx5_ttc_get_default_dest(attr->ttc, family2tt(attr->family)); + *dest = ipsec_rx_decrypted_pkt_def_dest(attr->ttc, attr->family); err = mlx5_ipsec_fs_roce_rx_create(ipsec->mdev, ipsec->roce, attr->ns, dest, attr->family, MLX5E_ACCEL_FS_ESP_FT_ROCE_LEVEL, attr->prio); diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/ipsec_fs_roce.c b/drivers/net/ethernet/mellanox/mlx5/core/lib/ipsec_fs_roce.c index b7d4b1a2baf2..d524f0220513 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/lib/ipsec_fs_roce.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/ipsec_fs_roce.c @@ -164,6 +164,8 @@ ipsec_fs_roce_rx_rule_setup(struct mlx5_core_dev *mdev, roce->rule = rule; memset(spec, 0, sizeof(*spec)); + if (default_dst->type == MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE) + flow_act.flags |= FLOW_ACT_IGNORE_FLOW_LEVEL; rule = mlx5_add_flow_rules(roce->ft, spec, &flow_act, default_dst, 1); if (IS_ERR(rule)) { err = PTR_ERR(rule); @@ -178,6 +180,8 @@ ipsec_fs_roce_rx_rule_setup(struct mlx5_core_dev *mdev, goto out; flow_act.action = MLX5_FLOW_CONTEXT_ACTION_FWD_DEST; + if (default_dst->type == MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE) + flow_act.flags &= ~FLOW_ACT_IGNORE_FLOW_LEVEL; dst.type = MLX5_FLOW_DESTINATION_TYPE_TABLE_TYPE; dst.ft = roce->ft_rdma; rule = mlx5_add_flow_rules(roce->nic_master_ft, NULL, &flow_act, &dst, -- 2.31.1