Skip to contentSkip to conversation

current status of 802.11s mesh in ath5k ?

DISCUSSION Hans Maes 5 participants 9 messages Jun 19, 2009 – Jul 7, 2009

lore
Share & tools
lore

Hans Maes

author
#0 lore raw

Hello,

I've been reading conflicting articles on several linux wireless websites about this, so I'm asking the question here:

What is the status of 802.11s mesh support in ath5k in the 2.6.30 kernel, in the current wireless-compat, and in the current wireless-testing ? Does it work, or is it broken ?
Can/should it be used in a production environment ?

Also, what is the best place to track the status/progress on 802.11s mesh in ath5k ?

Thanks!

Regards,

Hans

#1 lore raw
Quoted 3 lines · Hans Maes

What is the status of 802.11s mesh support in ath5k in the 2.6.30 kernel, in the current wireless-compat, and in the current wireless-testing ? Does it work, or is it broken ?

You'll want wireless-testing (or the latest compat), after commit 41ba08509. However ath5k will not beacon until you issue a scan due to a bug. You can bring up an ath5k node like this:

# iw phy phy0 interface add mesh type mp mesh_id mymesh
# ifconfig mesh up
# iwlist mesh scan
Quoted 1 line

Can/should it be used in a production environment ?

Things seem to work well aside from the scan issue, however I haven't fully verified multi-hop yet. 802.11s is currently a draft and the mac80211 implementation does not implement the full draft at this time.

Quoted 2 lines

Also, what is the best place to track the status/progress on 802.11s mesh in ath5k ?

We're keeping the mesh-specific drivers list up to date here:
http://o11s.org/trac
...however you may also be interested in the ath5k mailing list or the open80211s mailing list. Debugging effort and patches to resolve the scan issue or any other ath5k issues are very much welcome! Thanks,

-Andrey

Quoted 5 lines

--
To unsubscribe from this list: send the line "unsubscribe linux-wireless" in the body of a message to [email protected]
More majordomo info at  http://vger.kernel.org/majordomo-info.html

#2 lore raw
Quoted 3 lines · Andrey Yurovsky

You'll want wireless-testing (or the latest compat), after commit 41ba08509.  However ath5k will not beacon until you issue a scan due to a bug.

By the way, I imagine this is a side effect of calling ath5k_beacon_config inside ath5k_reset(). If so, we probably want to call it at the appropriate time when the interface is brought up.

I have been working on a patch to rework the beaconing, namely using bss_conf->enable_beacon and calling beacon_config whenever the beacon interval or beacon changes, that might fix it but I'm not too familiar with mesh details to say.

Signature · 1 line
Bob Copeland %% www.bobcopeland.com
#3 lore raw
Quoted 8 lines · Bob Copeland
Quoted 3 lines · Andrey Yurovsky

You'll want wireless-testing (or the latest compat), after commit 41ba08509.  However ath5k will not beacon until you issue a scan due to a bug.

By the way, I imagine this is a side effect of calling ath5k_beacon_config inside ath5k_reset().  If so, we probably want to call it at the appropriate time when the interface is brought up.

That sounds like it's on the right track, also there's a similar issue with rt2x00.

Quoted 4 lines

I have been working on a patch to rework the beaconing, namely using bss_conf->enable_beacon and calling beacon_config whenever the beacon interval or beacon changes, that might fix it but I'm not too familiar with mesh details to say.

'looking forward to that patch. From the driver side, there's not much difference between mesh beaconing and AP or IBSS beaconing. The mesh-specific requirements are listed on the driver API page:
http://wireless.kernel.org/en/developers/Documentation/mac80211/API -- if there's something we missed there, I'll be happy to edit it.

-Andrey

author
#6 lore raw
In reply to Andrey Yurovsky
Quoted 9 lines · Andrey Yurovsky

However ath5k will not beacon until you issue a scan due
to a bug. You can bring up an ath5k node like this:

# iw phy phy0 interface add mesh type mp mesh_id mymesh
# ifconfig mesh up
# iwlist mesh scan

So, once the first scan triggers the beaconing, everything should keep working ok ?
Or should I throw together a small script that triggers a scan once every 10minutes until this bug is fixed ? (In my case the mesh nodes should stay online 24/7 without user interaction)

Quoted 3 lines

Things seem to work well aside from the scan issue, however I haven't fully verified multi-hop yet.

I have a semi production setup with about 10 mesh nodes, in a building where multihop is required due to the extensive use of metal plating and therefore 'poor' wireless performance. In a typical scenario I need about 4 to 5 hops to reach the nodes at the other end of the building.

The nodes only have a single antenna connected, so I need to change some code to disable antenna switching, but I'm not a C expert/module developer at all so that might take a while to figure out...
I thought I managed to do it about 2 months ago but broke something since the nodes seemed to work perfectly after updating the wireless modules but they all went offline 3 to 5 days later and only came back online after reverting to the stable 2.6.29 wireless modules.

Anyway, when I get that issue worked out without breaking anything else I'll see how the multihop performs and report back in a few days.

Thanks!

Hans

#7 lore raw

2009/6/20 Hans Maes <[email protected]>:

Quoted 20 lines
Quoted 9 lines · Andrey Yurovsky

 However ath5k will not beacon until you issue a scan due to a bug.  You can bring up an ath5k node like this:

# iw phy phy0 interface add mesh type mp mesh_id mymesh
# ifconfig mesh up
# iwlist mesh scan

So, once the first scan triggers the beaconing, everything should keep working ok ?
Or should I throw together a small script that triggers a scan once every 10minutes until this bug is fixed ?
(In my case the mesh nodes should stay online 24/7 without user interaction)

Quoted 3 lines

Things seem to work well aside from the scan issue, however I haven't fully verified multi-hop yet.

Quoted 8 lines

I have a semi production setup with about 10 mesh nodes, in a building where multihop is required due to the extensive use of metal plating and therefore 'poor' wireless performance. In a typical scenario I need about 4 to 5 hops to reach the nodes at the other end of the building.

The nodes only have a single antenna connected, so I need to change some code to disable antenna switching, but I'm not a C expert/module developer at all so that might take a while to figure out...

Antenna controls are there, we just don't have a user interface yet (configfs/debugfs)
Check out ath5k_hw_set_antenna_mode inside phy.c, available modes are on enum ath5k_ant_mode inside ath5k.h and we call it on ath5k_config inside base.c ;-)

Quoted 9 lines

I thought I managed to do it about 2 months ago but broke something since the nodes seemed to work perfectly after updating the wireless modules but they all went offline 3 to 5 days later and only came back online after reverting to the stable 2.6.29 wireless modules.

Anyway, when I get that issue worked out without breaking anything else I'll see how the multihop performs and report back in a few days.

Thanks!

Signature · 3 lines
GPG ID: 0xD21DB2DB
As you read this post global entropy rises. Have Fun ;-)
Nick
#8 lore raw

Hi all,

I am using and testing the Open 80211s project with the ath5k driver. I have an atheros AR2413 chipset and an atheros AR5212/AR5213 chipset.

I installed my mesh network using this command lines on both devices:

# iw phy phy0 interface add mesh0 type mp mesh_id mesh
# ifconfig mesh0 192.168.1.1 (192.168.1.2 on the other device)
# iw dev mesh0 scan

The mesh network is unstable. I always get this warning messages:
ath5k phy0: unsupported jumbo

When I check on wireshark, it's look like some messages are lost on the radio media.

What can I do to solve this problems ? And get my mesh network stable ?

Thanks,
Stephane.

#4 lore raw
In reply to Andrey Yurovsky
Quoted 7 lines · Andrey Yurovsky
Quoted 4 lines

I have been working on a patch to rework the beaconing, namely using bss_conf->enable_beacon and calling beacon_config whenever the beacon interval or beacon changes, that might fix it but I'm not too familiar with mesh details to say.

'looking forward to that patch. From the driver side, there's not much difference between mesh beaconing and AP or IBSS beaconing.

Can you check whether my most recent patch (I think it was 2/5 in the ath5k series) helps? I looked at mac80211 and it seems we set enable_beacon properly in mesh mode so that should configure beaconing to start.

Signature · 1 line
Bob Copeland %% www.bobcopeland.com
#5 lore raw
Quoted 6 lines · Bob Copeland
Quoted 5 lines · Andrey Yurovsky
Quoted 4 lines

I have been working on a patch to rework the beaconing, namely using bss_conf->enable_beacon and calling beacon_config whenever the beacon interval or beacon changes, that might fix it but I'm not too familiar with mesh details to say.

Quoted 7 lines
Quoted 2 lines

'looking forward to that patch.  From the driver side, there's not much difference between mesh beaconing and AP or IBSS beaconing.

Can you check whether my most recent patch (I think it was 2/5 in the ath5k series) helps?  I looked at mac80211 and it seems we set enable_beacon properly in mesh mode so that should configure beaconing to start.

Thanks Bob, I confirm that your patch fixes mesh beaconing.

-Andrey