Re: [PATCH net] hsr: broadcast netlink notifications in the device's net namespace

[Fernando Fernandez Mancera <fmancera@suse.de>]

On 5/31/26 12:23 PM, Maoyi Xie wrote:
> On Thu, 28 May 2026 07:18, Jakub Kicinski <kuba@kernel.org> wrote:
>> Not sure TBH, we'd need to take a ref on the netns and allocate
>> a tracker (on DEBUG kernels). One could go either way.
> 
> On the RCU side, you're right that moving the net out of the lock
> means taking a ref, and this isn't a hot path where that really pays
> off. So I'd lean towards keeping it as posted, with the multicast
> still inside the rcu_read_lock. Fernando, thanks for the suggestion
> either way.
> 

In such case:

Reviewed-by: Fernando Fernandez Mancera <fmancera@suse.de>

Keep it if re-posted to net-next please.

Thanks!

>> I'm replying because I wanted to question whether this is Fixes+stable@
>> worthy. Sending the notifications to the namespace where the device is
>> makes sense. But it's as much a behavior changes as it is a fix.
>> The commit in question was merged to 5.6, real users clearly don't care.
> 
> On the Fixes and stable tags, my thinking was that the init_net side
> is an information leak. A privileged listener there ends up seeing
> ring error and node down events from devices in other netns. The
> payload carries the peer MAC and the slave ifindex. That was my reason
> for tagging it.
> 
> But I see your point that it is as much a behavior change as a fix,
> and if nobody has hit it since 5.6, the risk is clearly low. I don't
> feel strongly here. If you'd rather take it as a plain net-next
> improvement without the two tags, that is completely fine by me and I
> will respin it that way.
> 
> Thanks,
> Maoyi