Thread (3 messages) 3 messages, 2 authors, 18h ago
HOTtoday

[RFC PATCH net-next 0/1] qca8k host FDB handling

From: Brandon Mahdavi <hidden>
Date: 2026-07-25 03:22:17
Also in: lkml

This is an RFC because the configurations which exercise both parts of
the change currently rely on QCA8K patches carried downstream by OpenWrt.

OpenWrt enables assisted learning on the QCA8K CPU port and carries
multi-conduit support which floods unknown traffic to all CPU ports. The
host-FDB skip was tested there on multi-conduit QCA8K devices. The stale
dynamic-entry purge was tested on IPQ4019, whose downstream DSA driver
uses qca8k-common.c but is not yet present upstream.

The patch applies independently to net-next, but mainline QCA8K does not
currently enable assisted_learning_on_cpu_port, so the roaming path does
not yet have an in-tree user. I would appreciate guidance on whether this
change should be folded into a renewed QCA8K multi-conduit and assisted
learning series, or structured differently.

The CPU-port host-FDB handling was originally developed by Olaf Marzocchi.
The dynamic-entry purge and IPQ4019 testing were done by Brandon Mahdavi.

Brandon Mahdavi (1):
  net: dsa: qca8k: fix host FDB handling

 drivers/net/dsa/qca/qca8k-common.c | 56 ++++++++++++++++++++++++++++++
 1 file changed, 56 insertions(+)


base-commit: 04026c998c24ac47eb76886b9790c5710b603eb4
-- 
2.43.0
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help