Re: [PATCH 1/3] hwsim: remove 'optimization' sending to only known MACs

[James Prestwood <prestwoj@gmail.com>]

Hi Denis,

On 6/28/23 9:57 AM, Denis Kenzior wrote:
> Hi James,
> 
>>> Then this may be a conversation you have to start on linux-wireless.  
>>> I would have thought that there should only be a single wmediumd 
>>> instance on a system regardless of namespaces, but maybe I'm wrong here.
>>
>> Yeah, I'll do that as well. This would simplify things if the 
>> namespaces didn't come into play.
> 
> Check
> 100cb9ff40e0 ("mac80211_hwsim: Allow managing radios from non-initial 
> namespaces")
> f21e4d8ed16b ("mac80211_hwsim: Allow wmediumd to attach to radios 
> created in its netns")

Yeah, the second one does mention frames specifically, which I am 
getting from radios outside the namespace. But I don't see a difference 
in how ADD/DEL_MAC_ADDR works. But yeah, this is something I'll look into.

> 
>  From those two commits I get the impression that the intent is to allow 
> a single wmediumd across namespaces.  If for some reason the 
> ADD_MAC_ADDR messages are not being sent, then I would look into fixing 
> that first.
> 
>>
>> Nevertheless, was the distributed hwsim concept something you'd be 
>> interested in accepting upstream? Like I said, I'm already doing it 
>> and I'm happy to re-use hwsim and extend to support this external 
>> socket concept.
> 
> I'm not against it, but as you point out, it sounds super complicated.  
> So I'd have to look into it more once you are ready to share something.

Ok fair enough.

> 
> Regards,
> -Denis
>