netdev - Re: Question on IGMP snooping on switch managment port

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <793efa88-2a97-4cc3-9f84-101eef51962d@lunn.ch>
Date: Mon, 17 Jul 2023 14:53:10 +0200
From: Andrew Lunn <andrew@...n.ch>
To: Riccardo Laiolo <laiolo@...ff.it>
Cc: netdev@...r.kernel.org
Subject: Re: Question on IGMP snooping on switch managment port

On Mon, Jul 17, 2023 at 09:32:35AM +0200, Riccardo Laiolo wrote:
> Hi,
> 
> I'm working on a NXP-based embedded board (imx8mp) with a Marvell mv88e6390 switch.
> I'm running Linux 5.15.71 from the NXP downstream git repository (which is a year behind
> the upstream 5.15.y LTS release, I think). I've applied all the commits related to the Marvell
> driver from the 5.15LTS upstream that I was missing into my codebase.
> 
> I can't get the IGMP snooping to works properly. On front facing ports, it appears to work fine:
> MDB rules get correctly updated and multicast packets get blocked or routed accordingly. But when
> the subscribed is my embedded device (so the subscribed device is the switch
> management port) it doesn't work. The first IGMP packet get correctly routed and
> propagated through te network and all the interested node update their MDB entry list.
> 
> From now on all the outgoing IGMP packets get dropped.
> 
> 
> Adding and removing MDB rules by hand I found the offending rule appears to be
> 	dev br0 port br0 grp 224.0.1.185 temp
> 
> this rule gets correctly appended when I open a multicast rx socket,
> but my device fails to answer to any IMGP membership query until I remove said rule.
> 
> What am I missing? Is it possible for a linux network switch to be a multicast recipient device?

Hi Riccardo

It is a good idea to test the latest kernel before reporting problems
to mainline. You can then determine if its a known and fixed issue,
and the back port is missing, or it is something new. Any fix will be
applied to the latest kernel, and will then need back porting.

I would be interested in knowing if:

commit 7bcad0f0e6fbc1d613e49e0ee35c8e5f2e685bb0
Author: Steffen Bätz <steffen@...osonix.de>
Date:   Wed Mar 29 12:01:40 2023 -0300

    net: dsa: mv88e6xxx: Enable IGMP snooping on user ports only
    
    Do not set the MV88E6XXX_PORT_CTL0_IGMP_MLD_SNOOP bit on CPU or DSA ports.
    
    This allows the host CPU port to be a regular IGMP listener by sending out
    IGMP Membership Reports, which would otherwise not be forwarded by the
    mv88exxx chip, but directly looped back to the CPU port itself.


helps.

	Andrew