Hello Gaëtan, On Fri, Mar 19, 2021 at 5:59 PM Gaetan Rivet <[email protected]> wrote: > Our rte_flow implementation uses ICM mappings to program our hardware, > which requires super privileged access. We are looking into ways to avoid it.
Ok, thanks for looking into it. > > In the meantime, we failed to properly communicate this need in the rte_flow > API. > We will improve the documentation and the error path in DPDK. Without this capa, mlx5 rte_flow full hw offloading errors with logs like: 2021-03-22T14:12:40.274Z|00001|netdev_offload_dpdk(dp_netdev_flow_9)|WARN|dpdk0-rep0: rte_flow creation failed: 1 ((null)). 2021-03-22T14:12:40.274Z|00002|netdev_offload_dpdk(dp_netdev_flow_9)|WARN|dpdk0-rep0: Failed flow: flow create 3 ingress priority 0 group 0 transfer pattern eth src is 6a:20:8f:82:52:49 dst is 0c:42:a1:00:a8:7c type is 0x0800 / ipv4 / end actions count / port_id original 0 id 2 / end First log is useless. This is more bugfixing than enhancement. Though logs do not need to tell the full story, they can point at the mlx5 pmd documentation where the full explanation is. > > I can also update OVS documentation if anyone thinks it could help, but it is > vendor-specific. > I would expect it to be more relevant at the DPDK level. Yes, OVS should be device agnostic. DPDK is enough for me. -- David Marchand _______________________________________________ dev mailing list [email protected] https://mail.openvswitch.org/mailman/listinfo/ovs-dev
