The IP's next protocol will be specified by user when set IP-in-IP
tunnel header via set raw decap/encap commands.
Currently this field is wrongly set to zero if there is no upper layer.
This leads to the encapsulated IP-in-IP tunnel header is not correct.
This next protocol field should be leave it as-is if there is no upper
layer or value is already set.
Fixes: 30626def03d6 ("app/testpmd: support raw encap/decap actions")
Cc: sta...@dpdk.org
Signed-off-by: Xiaoyu Min <jack...@mellanox.com>
---
app/test-pmd/cmdline_flow.c | 8 ++++++--
1 file changed, 6 insertions(+), 2 deletions(-)
diff --git a/app/test-pmd/cmdline_flow.c b/app/test-pmd/cmdline_flow.c
index 0d0bc0a5b2..d5ad64bc8b 100644
--- a/app/test-pmd/cmdline_flow.c
+++ b/app/test-pmd/cmdline_flow.c
@@ -5996,11 +5996,13 @@ update_fields(uint8_t *buf, struct rte_flow_item *item,
uint16_t next_proto)
case RTE_FLOW_ITEM_TYPE_IPV4:
ipv4 = (struct rte_flow_item_ipv4 *)buf;
ipv4->hdr.version_ihl = 0x45;
- ipv4->hdr.next_proto_id = (uint8_t)next_proto;
+ if (next_proto && ipv4->hdr.next_proto_id == 0)
+ ipv4->hdr.next_proto_id = (uint8_t)next_proto;
break;
case RTE_FLOW_ITEM_TYPE_IPV6:
ipv6 = (struct rte_flow_item_ipv6 *)buf;
- ipv6->hdr.proto = (uint8_t)next_proto;
+ if (next_proto && ipv6->hdr.proto == 0)
+ ipv6->hdr.proto = (uint8_t)next_proto;
ipv6_vtc_flow = rte_be_to_cpu_32(ipv6->hdr.vtc_flow);
ipv6_vtc_flow &= 0x0FFFFFFF; /*< reset version bits. */
ipv6_vtc_flow |= 0x60000000; /*< set ipv6 version. */
@@ -6182,9 +6184,11 @@ cmd_set_raw_parsed(const struct buffer *in)
break;
case RTE_FLOW_ITEM_TYPE_GRE_KEY:
size = sizeof(rte_be32_t);
+ proto = 0x0;
break;
case RTE_FLOW_ITEM_TYPE_MPLS:
size = sizeof(struct rte_flow_item_mpls);
+ proto = 0x0;
break;
case RTE_FLOW_ITEM_TYPE_NVGRE:
size = sizeof(struct rte_flow_item_nvgre);
--
2.24.0.rc0.3.g12a4aeaad8
