net/mlx5: lay groundwork for switch offloads
authorAdrien Mazarguil <adrien.mazarguil@6wind.com>
Fri, 13 Jul 2018 09:40:37 +0000 (11:40 +0200)
committerThomas Monjalon <thomas@monjalon.net>
Thu, 26 Jul 2018 12:05:52 +0000 (14:05 +0200)
commit20b71e92ef8e0bac42143ae023f12c895829e44e
tree460cda6570876f58c55e7e643925b10cbdbdae07
parent3edc242d459b57aed3c63311f17a9138bec4b4cb
net/mlx5: lay groundwork for switch offloads

With mlx5, unlike normal flow rules implemented through Verbs for traffic
emitted and received by the application, those targeting different logical
ports of the device (VF representors for instance) are offloaded at the
switch level and must be configured through Netlink (TC interface).

This patch adds preliminary support to manage such flow rules through the
flow API (rte_flow).

Instead of rewriting tons of Netlink helpers and as previously suggested by
Stephen [1], this patch introduces a new dependency to libmnl [2]
(LGPL-2.1) when compiling mlx5.

[1] https://mails.dpdk.org/archives/dev/2018-March/092676.html
[2] https://netfilter.org/projects/libmnl/

Signed-off-by: Adrien Mazarguil <adrien.mazarguil@6wind.com>
Acked-by: Nelio Laranjeiro <nelio.laranjeiro@6wind.com>
Acked-by: Yongseok Koh <yskoh@mellanox.com>
doc/guides/nics/mlx5.rst
drivers/net/mlx5/Makefile
drivers/net/mlx5/mlx5.c
drivers/net/mlx5/mlx5.h
drivers/net/mlx5/mlx5_nl_flow.c [new file with mode: 0644]
mk/rte.app.mk