mbox series

[v2,0/6] realtek: fix management of mdb entries

Message ID 20230506230535.1150881-1-jan@3e8.eu
Headers show
Series realtek: fix management of mdb entries | expand

Message

Jan Hoffmann May 6, 2023, 11:05 p.m. UTC 
This series fixes multiple issues related to the L2 table and multicast 
table. That includes an issue that causes corruption of the port mask 
for unknown multicast forwarding, which can occur even when multicast 
snooping is disabled.

As discussed previously, these changes don't make multicast snooping
fully work, as that would still need additional changes in the kernel
for all DSA switches (see the follow-up messages to the original patch
series at [1] and [2] for details).

v2:
 - properly deal with redundant is_lagmember check by removing it

[1] http://lists.openwrt.org/pipermail/openwrt-devel/2023-March/040621.html
[2] https://lore.kernel.org/netdev/20230306134636.p2ufzoqk6kf3hu3y@skbuf/

Jan Hoffmann (6):
  realtek: fix writing/deletion of CAM entries
  realtek: don't treat first multicast portmask entry as reserved
  realtek: actually remove port from multicast portmask
  realtek: don't add CPU port to multicast portmasks
  realtek: remove store_mcgroups/load_mcgroups
  realtek: remove redundant is_lagmember checks

 .../files-5.10/drivers/net/dsa/rtl83xx/dsa.c  | 62 +++----------------
 .../drivers/net/dsa/rtl83xx/rtl838x.h         |  1 -
 .../files-5.15/drivers/net/dsa/rtl83xx/dsa.c  | 58 +++--------------
 .../drivers/net/dsa/rtl83xx/rtl838x.h         |  1 -
 4 files changed, 18 insertions(+), 104 deletions(-)

Comments

Bjørn Mork May 7, 2023, 7:20 a.m. UTC | #1 
Jan Hoffmann <jan@3e8.eu> writes:

> This series fixes multiple issues related to the L2 table and multicast 
> table. That includes an issue that causes corruption of the port mask 
> for unknown multicast forwarding, which can occur even when multicast 
> snooping is disabled.

Thanks for cleaning up this mess.

I was curious and started looking at how all these bugs were
introduced.  And, uhm, I should know better by now...

THANKS A LOT!


Bjørn
Sander Vanheule May 7, 2023, 5:22 p.m. UTC | #2 
On Sun, 2023-05-07 at 01:05 +0200, Jan Hoffmann wrote:
> This series fixes multiple issues related to the L2 table and multicast 
> table. That includes an issue that causes corruption of the port mask 
> for unknown multicast forwarding, which can occur even when multicast 
> snooping is disabled.
> 
> As discussed previously, these changes don't make multicast snooping
> fully work, as that would still need additional changes in the kernel
> for all DSA switches (see the follow-up messages to the original patch
> series at [1] and [2] for details).
> 
> v2:
>  - properly deal with redundant is_lagmember check by removing it
> 
> [1] http://lists.openwrt.org/pipermail/openwrt-devel/2023-March/040621.html
> [2] https://lore.kernel.org/netdev/20230306134636.p2ufzoqk6kf3hu3y@skbuf/
> 
> Jan Hoffmann (6):
>   realtek: fix writing/deletion of CAM entries
>   realtek: don't treat first multicast portmask entry as reserved
>   realtek: actually remove port from multicast portmask
>   realtek: don't add CPU port to multicast portmasks
>   realtek: remove store_mcgroups/load_mcgroups

git-am didn't like this patch as submitted, but I did apply these patches after
the other series ("realtek: fix multiple issues with L2 forwarding"). Should be
resolved though, looks like it was just due to some code moving around.

>   realtek: remove redundant is_lagmember checks

Thanks for fixing these multicast issues! Applied to master.

Best,
Sander