In a typical setup like:
guest A <-virtio-> OVS A <-vxlan-> OVS B <-virtio-> guest B
TSO packets from guest A are segmented against the OVS A physical port
mtu adjusted by the vxlan tunnel header size, regardless of guest A
interface mtu.
As an example, let's say guest A and guest B mtu are set to 1500 bytes.
OVS A and OVS B physical ports mtu are set to 1600 bytes.
Guest A will request TCP segmentation for 1448 bytes segments.
On the other hand, OVS A will request 1498 bytes segments to the HW.
This results in OVS B dropping packets because decapsulated packets
are larger than the vhost-user port (serving guest B) mtu.
2024-04-17T14:13:01.239Z|00002|netdev_dpdk(pmd-c03/id:7)|WARN|vhost0:
Too big size 1564 max_packet_len 1518
vhost-user ports expose a guest mtu by filling mbuf->tso_segsz.
Use it as a hint.
This may result in segments (on the wire) slightly shorter than the
optimal size.
Reported-at: https://github.com/openvswitch/ovs-issues/issues/321
Signed-off-by: David Marchand <[email protected]>
---
Changes since v3:
- removed check on mbuf->tso_segsz == 0,
---
lib/netdev-dpdk.c | 11 ++++++++---
1 file changed, 8 insertions(+), 3 deletions(-)
diff --git a/lib/netdev-dpdk.c b/lib/netdev-dpdk.c
index 0dfd685467..bda3fa94b6 100644
--- a/lib/netdev-dpdk.c
+++ b/lib/netdev-dpdk.c
@@ -2670,14 +2670,19 @@ netdev_dpdk_prep_hwol_packet(struct netdev_dpdk *dev,
struct rte_mbuf *mbuf)
if (mbuf->ol_flags & RTE_MBUF_F_TX_TCP_SEG) {
struct tcp_header *th = dp_packet_l4(pkt);
+ uint16_t link_tso_segsz;
int hdr_len;
if (tunnel_type) {
- mbuf->tso_segsz = dev->mtu - mbuf->l2_len - mbuf->l3_len -
- mbuf->l4_len - mbuf->outer_l3_len;
+ link_tso_segsz = dev->mtu - mbuf->l2_len - mbuf->l3_len -
+ mbuf->l4_len - mbuf->outer_l3_len;
} else {
mbuf->l4_len = TCP_OFFSET(th->tcp_ctl) * 4;
- mbuf->tso_segsz = dev->mtu - mbuf->l3_len - mbuf->l4_len;
+ link_tso_segsz = dev->mtu - mbuf->l3_len - mbuf->l4_len;
+ }
+
+ if (mbuf->tso_segsz > link_tso_segsz) {
+ mbuf->tso_segsz = link_tso_segsz;
}
hdr_len = mbuf->l2_len + mbuf->l3_len + mbuf->l4_len;
--
2.44.0
_______________________________________________
dev mailing list
[email protected]
https://mail.openvswitch.org/mailman/listinfo/ovs-dev
