Assuming a case with the following pattern:
eth / ipv4 / udp / vxlan / eth / ipv4 proto is (IPv4 | IPv6)
MLX5 PMD incorrectly recognized this as a packet encapsulated as
VXLAN and IP in IP at the same packet header level.
This caused a flow validation error.
This patch fixes the inner IP in IP detection logic.
Fixes: 2db234e769e1 ("net/mlx5: fix IP-in-IP tunnels recognition")
Cc: sta...@dpdk.org
Signed-off-by: Gregory Etelson <getel...@nvidia.com>
Acked-by: Dariusz Sosnowski <dsosnow...@nvidia.com>
---
drivers/net/mlx5/mlx5_flow_dv.c | 10 ++++++++--
1 file changed, 8 insertions(+), 2 deletions(-)
diff --git a/drivers/net/mlx5/mlx5_flow_dv.c b/drivers/net/mlx5/mlx5_flow_dv.c
index 61d3101ce8..2d03a3f480 100644
--- a/drivers/net/mlx5/mlx5_flow_dv.c
+++ b/drivers/net/mlx5/mlx5_flow_dv.c
@@ -7925,7 +7925,10 @@ flow_dv_validate(struct rte_eth_dev *dev, const struct
rte_flow_attr *attr,
mlx5_flow_tunnel_ip_check(items, next_protocol,
item_flags,
&l3_tunnel_flag);
- if (l3_tunnel_detection == l3_tunnel_inner) {
+ /*
+ * explicitly allow inner IPIP match
+ */
+ if (l3_tunnel_detection == l3_tunnel_outer) {
item_flags |= l3_tunnel_flag;
tunnel = 1;
}
@@ -7949,7 +7952,10 @@ flow_dv_validate(struct rte_eth_dev *dev, const struct
rte_flow_attr *attr,
mlx5_flow_tunnel_ip_check(items, next_protocol,
item_flags,
&l3_tunnel_flag);
- if (l3_tunnel_detection == l3_tunnel_inner) {
+ /*
+ * explicitly allow inner IPIP match
+ */
+ if (l3_tunnel_detection == l3_tunnel_outer) {
item_flags |= l3_tunnel_flag;
tunnel = 1;
}
--
2.45.2
