Hi, thank you!
1. I got to ESTAB the link this way:
a) start meshd-nl80211 on each station (2 stations)
b) got error (the same as in the first e-mail)
c) killed meshd-nl80211 on 1st station (second still running)
d) start meshd-nl80211 on 2nd station again
e) link ESTABlished and PING works
2. There’s still an error:
estab with 02:00:00:00:00:02
set auth flag (seq num=1445545109)
mesh plink with 02:00:00:00:00:02 established
nlerror, cmd 18, seq 1445545113: Invalid argument
Mesh plink timer for 02:00:00:00:00:02 fired on state ESTAB
3. I’m using open ath9k_htc firmware from:
https://github.com/qca/open-ath9k-htc-firmware
I’ve heard you can actually build some bigger network (some person said about
20+ nodes) using these modules&firmware? With 80211s I tried only with 4
stations, but next 6 of them are waiting in the office so I will certainly
check it out.
The open80211s implementation in kernel I use (4.09) doesn’t provide any
configurable min. RSSI mesh parameter (not present if iw $m_if get mesh_param).
It’s just not present in the kernel. Did you mean manually altering the min.
RSSI in the kernel via patch?
Could you please share with me some links which would help me to go in-depth
how to apply open80211s patches etc. so I could use more recent version?
According to wpa_supplicant and setting manually next_hop - I tried to do it
but packets were being lost whole the time. For example if the path was set
manually, PING took several seconds to get any packet back. I was waiting
literally several minutes for iperf to begin transfer between nodes. I think it
was connected with ESTABed links between these particular nodes. If you’ve got
a direct link with other station, manually changing the next_hop address only
makes it worst - at least with my hw/sw configuration.
Do you think these module could work in IBSS-RSN mode (using wpa_supplicant)?
There are some information about IBSS-RSN in the net but I didn’t find anything
actually working (rather some info like „it could use WPA2 encryption”, but no
evidences).
Best regards,
Michal
> Wiadomość napisana przez Bob Copeland <[email protected]> w dniu
> 22.10.2015, o godz. 20:49:
>
> On Wed, Oct 21, 2015 at 05:58:58PM +0200, Michał Kołodziej via Devel wrote:
>> Hello,
>
> Hi!
>
>> we’re trying to configure secured mesh network based on several RPis, each
>> with wn722-n wireless module.
>> We did compile firmware for these modules, we’ve got working iw tool, mesh
>> enabled wpa_supplicant and 4.09 kernel.
>> At the beginning we build ad-hoc and babel protocol based mesh, but we
>> wasn’t able to setup WPA2 encryption (IBSS-RSN).
>
> Ok, so there is a lot to cover here :)
>
> How many is several? The ath9k_htc can support only 4 stations so a lot
> of people have reported issues here on the mailing list when that limit
> is reached.
>
> Anyway, first a couple of things to try:
>
> * roll back a few recent commits of authsae, and see if you still have
> the same issues
> * try debug=0xffff in your authsae.cfg to see which of the two paths
> are calling fin().
>
>> joining mesh mesh on freq 2412, mode 0
>> NL80211_CMD_FRAME (1445439047.424875)
>
> CMD_FRAME output shows four auth frames, that looks normal for SAE so far.
>
>> fin: 1, key len:0 peer:00:00:00:00:00:02 me:00:00:00:00:00:01
>
> WLAN_STATUS_UNSPECIFIED_FAILURE -- not terribly helpful. Finding the
> caller would be useful information.
>
> (aside, you might want to set the 'locally administered' bit on your
> mac addresses, i.e. 02:...)
>
>> removing peer candidate 1e25a66:28:00:00:14:00
>
> (This output is a bug, I have a patch for it)
>
> So there are a couple of things we can do in wpa_supplicant that we
> aren't doing today:
>
> * we can pass rssi threshold into the kernel so you don't get
> candidate events for far away stations
>
> * we could support explicit plink block command (when using authsae or
> wpa_s, the userspace is in charge of the peering manager, so messing with
> iw here will do nothing).
>
> I would think forcing next hop on the mpaths should work, though.
>
> --
> Bob Copeland %% http://bobcopeland.com/
_______________________________________________
Devel mailing list
[email protected]
http://lists.open80211s.org/cgi-bin/mailman/listinfo/devel
