2.6.29+ ath5k connect disconnect cycles?

2.6.29+ ath5k connect disconnect cycles?

[Soeren Sonnenburg]

Dear all,

does anyone else observer strange connect disconnect cycles with ath5k?
I have two different notebooks here one with an AR5414 and one with some
2424 (?) chipset both having the same problems to connect...

Soeren
-- 
For the one fact about the future of which we can be certain is that it
will be utterly fantastic. -- Arthur C. Clarke, 1962

Re: 2.6.29+ ath5k connect disconnect cycles?

[Tulio Magno Quites Machado Filho]

On Apr 06 09 10:55, Soeren Sonnenburg wrote:
> does anyone else observer strange connect disconnect cycles with ath5k?
> I have two different notebooks here one with an AR5414 and one with some
> 2424 (?) chipset both having the same problems to connect...

Hello Soeren,

Is there any information from dmesg?

-- 
Tulio Magno

Re: 2.6.29+ ath5k connect disconnect cycles?

[Soeren Sonnenburg]

On Mon, 2009-04-06 at 21:55 +0200, Tulio Magno Quites Machado Filho
wrote:
> On Apr 06 09 10:55, Soeren Sonnenburg wrote:
> > does anyone else observer strange connect disconnect cycles with ath5k?
> > I have two different notebooks here one with an AR5414 and one with some
> > 2424 (?) chipset both having the same problems to connect...
> 
> Hello Soeren,
> 
> Is there any information from dmesg?

I am seeing cycles of

airport: deauthenticated (Reason: 2)
airport: direct probe to AP XX:XX:XX:XX:XX:XX try 1
airport direct probe responded
airport: authenticate with AP XX:XX:XX:XX:XX:XX
airport: authenticated
airport: associate with AP XX:XX:XX:XX:XX:XX
airport: RX ReassocResp from XX:XX:XX:XX:XX:XX (capab=0x421 status=0 aid=79)
airport: associated
airport: deauthenticated (Reason: 2)
airport: direct probe to AP XX:XX:XX:XX:XX:XX try 1
airport direct probe responded
airport: authenticate with AP XX:XX:XX:XX:XX:XX
airport: authenticated
airport: associate with AP XX:XX:XX:XX:XX:XX
airport: RX ReassocResp from XX:XX:XX:XX:XX:XX (capab=0x421 status=0 aid=81)
airport: associated
airport: deauthenticated (Reason: 2)
airport: direct probe to AP XX:XX:XX:XX:XX:XX try 1
airport direct probe responded
airport: authenticate with AP XX:XX:XX:XX:XX:XX
airport: authenticated
airport: associate with AP XX:XX:XX:XX:XX:XX
airport: RX ReassocResp from XX:XX:XX:XX:XX:XX (capab=0x421 status=0 aid=82)
airport: associated
airport: deauthenticated (Reason: 2)
...

Not sure about the involvement of network-manager here... but I see that
the device gets a valid IP via dhcp.

Soeren
-- 
For the one fact about the future of which we can be certain is that it
will be utterly fantastic. -- Arthur C. Clarke, 1962

Re: 2.6.29+ ath5k connect disconnect cycles?

["Markus Koßmann"]

> On Mon, 2009-04-06 at 21:55 +0200, Tulio Magno Quites Machado Filho
> wrote:
> > On Apr 06 09 10:55, Soeren Sonnenburg wrote:
> > > does anyone else observer strange connect disconnect cycles with
> ath5k?
> > > I have two different notebooks here one with an AR5414 and one with
> some
> > > 2424 (?) chipset both having the same problems to connect...
> > 
> > Hello Soeren,
> > 
> > Is there any information from dmesg?
> 
> I am seeing cycles of
> 
> airport: deauthenticated (Reason: 2)
> airport: direct probe to AP XX:XX:XX:XX:XX:XX try 1
> airport direct probe responded
> airport: authenticate with AP XX:XX:XX:XX:XX:XX
> airport: authenticated
> airport: associate with AP XX:XX:XX:XX:XX:XX
> airport: RX ReassocResp from XX:XX:XX:XX:XX:XX (capab=0x421 status=0
> aid=79)
> airport: associated
> airport: deauthenticated (Reason: 2)
> airport: direct probe to AP XX:XX:XX:XX:XX:XX try 1
> airport direct probe responded
> airport: authenticate with AP XX:XX:XX:XX:XX:XX
> airport: authenticated
> airport: associate with AP XX:XX:XX:XX:XX:XX
> airport: RX ReassocResp from XX:XX:XX:XX:XX:XX (capab=0x421 status=0
> aid=81)
> airport: associated
> airport: deauthenticated (Reason: 2)
> airport: direct probe to AP XX:XX:XX:XX:XX:XX try 1
> airport direct probe responded
> airport: authenticate with AP XX:XX:XX:XX:XX:XX
> airport: authenticated
> airport: associate with AP XX:XX:XX:XX:XX:XX
> airport: RX ReassocResp from XX:XX:XX:XX:XX:XX (capab=0x421 status=0
> aid=82)
> airport: associated
> airport: deauthenticated (Reason: 2)
> ...
> 
> Not sure about the involvement of network-manager here... but I see that
> the device gets a valid IP via dhcp.
> 
I've seen such cycles also with intel ipw5300 WLAN and using ifconfig to start the network. To get WLAN working again, I have to ifdown wlan0 , unload and reload the driver module and  finally ifup wlan0. So it's probably neither a ath5k nor a network-manager problem.

Re: 2.6.29+ ath5k connect disconnect cycles?

[Maxim Levitsky]

On Mon, 2009-04-06 at 10:55 +0200, Soeren Sonnenburg wrote:
> Dear all,
> 
> does anyone else observer strange connect disconnect cycles with ath5k?
> I have two different notebooks here one with an AR5414 and one with some
> 2424 (?) chipset both having the same problems to connect...
> 
> Soeren

I use latest -git of wireless-testing, and AR2425.
Thanks to hard work of ath5k developers, it works better and better.
Yet phy seems to die when loaded with traffic, but its reset brings is
back alive. ath5k does a reset every 10 seconds.

If you use WPA, then you might have bigger problems, I mean, during the
period when phy is dead, AP might ask for group rekey, but ath5k won't
hear this, and continue using old keys. This causes much longer delays,
and sometimes reconnect cycles you see.

Best regards,
	Maxim Levitsky

Re: 2.6.29+ ath5k connect disconnect cycles?

[Soeren Sonnenburg]

[-- Attachment #1: Type: text/plain, Size: 1175 bytes --]

On Tue, 2009-04-07 at 12:29 +0200, "Markus Koßmann" wrote:
> > On Mon, 2009-04-06 at 21:55 +0200, Tulio Magno Quites Machado Filho
> > wrote:
> > > On Apr 06 09 10:55, Soeren Sonnenburg wrote:
> > > > does anyone else observer strange connect disconnect cycles with
> > ath5k?
> > > > I have two different notebooks here one with an AR5414 and one with
> > some
> > > > 2424 (?) chipset both having the same problems to connect...
> > > 
> > > Hello Soeren,
> > > 
> > > Is there any information from dmesg?
> > 
> > I am seeing cycles of
[connecting and reconnecting]
> > Not sure about the involvement of network-manager here... but I see that
> > the device gets a valid IP via dhcp.
> > 
> I've seen such cycles also with intel ipw5300 WLAN and using ifconfig to start the network. To get WLAN working again, I have to ifdown wlan0 , unload and reload the driver module and  finally ifup wlan0. So it's probably neither a ath5k nor a network-manager problem.

A rmmod ath5k / reload didn't help in my case.

Soeren
-- 
For the one fact about the future of which we can be certain is that it
will be utterly fantastic. -- Arthur C. Clarke, 1962

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 197 bytes --]

Re: 2.6.29+ ath5k connect disconnect cycles?

[Soeren Sonnenburg]

[-- Attachment #1: Type: text/plain, Size: 1223 bytes --]

On Tue, 2009-04-07 at 16:00 +0300, Maxim Levitsky wrote:
> On Mon, 2009-04-06 at 10:55 +0200, Soeren Sonnenburg wrote:
> > Dear all,
> > 
> > does anyone else observer strange connect disconnect cycles with ath5k?
> > I have two different notebooks here one with an AR5414 and one with some
> > 2424 (?) chipset both having the same problems to connect...
> > 
> > Soeren
> 
> I use latest -git of wireless-testing, and AR2425.
> Thanks to hard work of ath5k developers, it works better and better.
> Yet phy seems to die when loaded with traffic, but its reset brings is
> back alive. ath5k does a reset every 10 seconds.
> 
> If you use WPA, then you might have bigger problems, I mean, during the
> period when phy is dead, AP might ask for group rekey, but ath5k won't
> hear this, and continue using old keys. This causes much longer delays,
> and sometimes reconnect cycles you see.

Unfortunately, this happened with a open network. But YMMV as it seems
to work very reliably in other locations (even with WPA)... with both
the 2425 and 5414 chipset.

Soeren
-- 
For the one fact about the future of which we can be certain is that it
will be utterly fantastic. -- Arthur C. Clarke, 1962

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 197 bytes --]

Re: 2.6.29+ ath5k connect disconnect cycles?

[Soeren Sonnenburg]

[-- Attachment #1: Type: text/plain, Size: 1225 bytes --]

On Tue, 2009-04-07 at 16:00 +0300, Maxim Levitsky wrote: 
> On Mon, 2009-04-06 at 10:55 +0200, Soeren Sonnenburg wrote:
> > Dear all,
> > 
> > does anyone else observer strange connect disconnect cycles with ath5k?
> > I have two different notebooks here one with an AR5414 and one with some
> > 2424 (?) chipset both having the same problems to connect...
> > 
> > Soeren
> 
> I use latest -git of wireless-testing, and AR2425.
> Thanks to hard work of ath5k developers, it works better and better.
> Yet phy seems to die when loaded with traffic, but its reset brings is
> back alive. ath5k does a reset every 10 seconds.
> 
> If you use WPA, then you might have bigger problems, I mean, during the
> period when phy is dead, AP might ask for group rekey, but ath5k won't
> hear this, and continue using old keys. This causes much longer delays,
> and sometimes reconnect cycles you see.

Unfortunately, this happened with an open network. But YMMV as it seems
to work very reliably in other locations (even with WPA)... with both
the 2425 and 5414 chipset.

Soeren
-- 
For the one fact about the future of which we can be certain is that it
will be utterly fantastic. -- Arthur C. Clarke, 1962

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 197 bytes --]