Any reason AID is not in the NL80211_STA_INFO enumeration?

Any reason AID is not in the NL80211_STA_INFO enumeration?

[Ben Greear]

To sniff /AX stations, it is helpful to know their AID.  So, any problems
with adding it to 'iw station dump' output?

Thanks,
Ben

-- 
Ben Greear <greearb@candelatech.com>
Candela Technologies Inc  http://www.candelatech.com

Re: Any reason AID is not in the NL80211_STA_INFO enumeration?

[Johannes Berg]

On Mon, 2020-02-17 at 17:11 -0800, Ben Greear wrote:
> To sniff /AX stations, it is helpful to know their AID.  So, any problems
> with adding it to 'iw station dump' output?

No, no problem at all. I guess it was just never added because it
originally comes from userspace (hostapd), and so didn't really need to
be sent back. But yeah, we have trivial infrastructure to send it back,
so sure, why not.

johannes

Re: Any reason AID is not in the NL80211_STA_INFO enumeration?

[Ben Greear]

On 02/28/2020 02:06 AM, Johannes Berg wrote:
> On Mon, 2020-02-17 at 17:11 -0800, Ben Greear wrote:
>> To sniff /AX stations, it is helpful to know their AID.  So, any problems
>> with adding it to 'iw station dump' output?
>
> No, no problem at all. I guess it was just never added because it
> originally comes from userspace (hostapd), and so didn't really need to
> be sent back. But yeah, we have trivial infrastructure to send it back,
> so sure, why not.

Ok, thanks for the response.

 From looking at debugfs for a station device, there is the netdev AID:

cat /debug/ieee80211/wiphy0/netdev\:wlan0/aid
1

And there is a peer AID:

cat /debug/ieee80211/wiphy0/netdev\:wlan0/stations/04\:f0\:21\:c2\:fd\:b0/aid
0

For sniffing AX, I'd need to use '1' here as far as I can tell.  Maybe for station
devices the peer is always zero?  And for AP devices, the peer IDs are what is useful
and the netdev AID is not?


Thanks,
Ben

>
> johannes
>

-- 
Ben Greear <greearb@candelatech.com>
Candela Technologies Inc  http://www.candelatech.com

Re: Any reason AID is not in the NL80211_STA_INFO enumeration?

[Johannes Berg]

On Fri, 2020-02-28 at 07:26 -0800, Ben Greear wrote:
> 
> On 02/28/2020 02:06 AM, Johannes Berg wrote:
> > On Mon, 2020-02-17 at 17:11 -0800, Ben Greear wrote:
> > > To sniff /AX stations, it is helpful to know their AID.  So, any problems
> > > with adding it to 'iw station dump' output?
> > 
> > No, no problem at all. I guess it was just never added because it
> > originally comes from userspace (hostapd), and so didn't really need to
> > be sent back. But yeah, we have trivial infrastructure to send it back,
> > so sure, why not.
> 
> Ok, thanks for the response.
> 
>  From looking at debugfs for a station device, there is the netdev AID:
> 
> cat /debug/ieee80211/wiphy0/netdev\:wlan0/aid
> 1
> 
> And there is a peer AID:
> 
> cat /debug/ieee80211/wiphy0/netdev\:wlan0/stations/04\:f0\:21\:c2\:fd\:b0/aid
> 0
> 
> For sniffing AX, I'd need to use '1' here as far as I can tell.  Maybe for station
> devices the peer is always zero?  And for AP devices, the peer IDs are what is useful
> and the netdev AID is not?

Yeah. The first is u.mgd.aid (i.e. the netdev data), and the latter is
the STA AID.

It makes sense the latter is not assigned in the client-side case.

Arguably, only the *latter* should be in nl80211 per STA, since in the
client-side case it's *our* AID, not the APs.

johannes

Re: Any reason AID is not in the NL80211_STA_INFO enumeration?

[Ben Greear]

On 03/03/2020 06:08 AM, Johannes Berg wrote:
> On Fri, 2020-02-28 at 07:26 -0800, Ben Greear wrote:
>>
>> On 02/28/2020 02:06 AM, Johannes Berg wrote:
>>> On Mon, 2020-02-17 at 17:11 -0800, Ben Greear wrote:
>>>> To sniff /AX stations, it is helpful to know their AID.  So, any problems
>>>> with adding it to 'iw station dump' output?
>>>
>>> No, no problem at all. I guess it was just never added because it
>>> originally comes from userspace (hostapd), and so didn't really need to
>>> be sent back. But yeah, we have trivial infrastructure to send it back,
>>> so sure, why not.
>>
>> Ok, thanks for the response.
>>
>>  From looking at debugfs for a station device, there is the netdev AID:
>>
>> cat /debug/ieee80211/wiphy0/netdev\:wlan0/aid
>> 1
>>
>> And there is a peer AID:
>>
>> cat /debug/ieee80211/wiphy0/netdev\:wlan0/stations/04\:f0\:21\:c2\:fd\:b0/aid
>> 0
>>
>> For sniffing AX, I'd need to use '1' here as far as I can tell.  Maybe for station
>> devices the peer is always zero?  And for AP devices, the peer IDs are what is useful
>> and the netdev AID is not?
>
> Yeah. The first is u.mgd.aid (i.e. the netdev data), and the latter is
> the STA AID.
>
> It makes sense the latter is not assigned in the client-side case.
>
> Arguably, only the *latter* should be in nl80211 per STA, since in the
> client-side case it's *our* AID, not the APs.

Although it is a bit weird, I was thinking to always return the AID as part of the station
dump, and for station netdevs, use the local netdev AID, and for APs, use the remote station AID.

That seems a bit cleaner to me than having two separate ways to get the AID for stations and APs,
but I can do two separate ways if you prefer that.

Thanks,
Ben

>
> johannes
>

-- 
Ben Greear <greearb@candelatech.com>
Candela Technologies Inc  http://www.candelatech.com

Re: Any reason AID is not in the NL80211_STA_INFO enumeration?

[Johannes Berg]

On Tue, 2020-03-03 at 06:57 -0800, Ben Greear wrote:
> 
> Although it is a bit weird, I was thinking to always return the AID as part of the station
> dump, and for station netdevs, use the local netdev AID, and for APs, use the remote station AID.
> 
> That seems a bit cleaner to me than having two separate ways to get the AID for stations and APs,
> but I can do two separate ways if you prefer that.

I don't really agree that it's cleaner - by definition, the data in a
station is data about that station, not data about "self".

We may not have a great way right now to show the latter ('data about
self')? But I still don't think we should mix it.

johannes

Re: Any reason AID is not in the NL80211_STA_INFO enumeration?

[Ben Greear]

On 03/03/2020 06:59 AM, Johannes Berg wrote:
> On Tue, 2020-03-03 at 06:57 -0800, Ben Greear wrote:
>>
>> Although it is a bit weird, I was thinking to always return the AID as part of the station
>> dump, and for station netdevs, use the local netdev AID, and for APs, use the remote station AID.
>>
>> That seems a bit cleaner to me than having two separate ways to get the AID for stations and APs,
>> but I can do two separate ways if you prefer that.
>
> I don't really agree that it's cleaner - by definition, the data in a
> station is data about that station, not data about "self".
>
> We may not have a great way right now to show the latter ('data about
> self')? But I still don't think we should mix it.

Ok, I'll go poke at the code and see what I can figure out.

Thanks,
Ben

-- 
Ben Greear <greearb@candelatech.com>
Candela Technologies Inc  http://www.candelatech.com