[PATCH net v4 0/2] net/mlx5: Bridge, fix remaining switchdev ownership gaps on merged eswitch

Bernardo Soares <[email protected]>
Newsgroups org.kernel.vger.netdev
Message-ID <[email protected]>
v4 of "net/mlx5: Bridge, don't fail switchdev events of sibling
eswitch ports" (5f324c5d1b12), addressing reviewer feedback on v3:

 - The v3 commit message for patch 1 wrongly attributed the recursive
   lower-device walk fix to LAG bond enslavement order. As pointed
   out in review, mlx5_esw_bridge_lag_rep_get() already filters on
   mlx5_esw_bridge_dev_same_esw() per candidate and cannot select a
   sibling's rep. The actual bug is in the generic recursive walk
   used when attribute changes are emitted against the bridge master
   netdevice directly (a bridge with representors of more than one
   eswitch instance enslaved, no LAG involved): the walk returns as
   soon as any lower device yields a rep, and the underlying base
   case only checks same-HW, not ownership. Commit message rewritten
   to describe this correctly; no functional change from v3.
 - The v3 commit message for patch 2 claimed a "replayed/duplicate"
   NETDEV_CHANGEUPPER unlink as one of the reachable cases. As
   pointed out in review, netdevice notifiers are not replayed, so
   there is no such duplicate delivery. The actual (and only)
   reachable case is a sibling instance whose bridge offload notifier
   registers after a peer port was already enslaved, so it misses
   the link event and never tracks the port, then genuinely receives
   the later unlink event. Commit message rewritten accordingly; no
   functional change from v3.

Tested Patch 1 on a ConnectX-7 NIC (MT2910) on my single NIC system.
Patch 2 requires a multiple eswitch instance setup, so I wasn't able
to exercise its code paths.

Note: v1/v2 were sent From/Signed-off-by [email protected]; v3
and v4 are sent from my personal address ([email protected])
instead, for unrelated mail delivery reasons. Same author, same
person.

Bernardo Soares (2):
  net/mlx5: Bridge, don't fail switchdev events of sibling eswitch ports
  net/mlx5: Bridge, don't fail unlink of untracked/unsupported peer
    ports

 .../mellanox/mlx5/core/en/rep/bridge.c        | 45 +++++++++++++++----
 .../ethernet/mellanox/mlx5/core/esw/bridge.c  | 15 +++++--
 .../ethernet/mellanox/mlx5/core/esw/bridge.h  |  2 +
 3 files changed, 49 insertions(+), 13 deletions(-)

-- 
2.43.0
lmpx.com only provides a reader for public news (NNTP) servers. It is not affiliated with the servers or forums shown here and is not responsible for the content of articles, which is written by their respective authors.