2011-03-31 15:43:04

by Stanislaw Gruszka

[permalink] [raw]
Subject: ec15e68ba6a regresion?

Seems commit:

commit ec15e68ba6a505631016f230899bafbb7b8cd0d6
Author: Jouni Malinen <[email protected]>
Date: Wed Mar 23 15:29:52 2011 +0200

cfg80211: Add nl80211 event for deletion of a station entry

is causing problems on my system. Sometimes when I disconnect
from wireless network, I can not connect again. Without commit
I can not reproduce the problem.

Stanislaw


2011-03-31 15:46:46

by Jouni Malinen

[permalink] [raw]
Subject: Re: ec15e68ba6a regresion?

On Thu, Mar 31, 2011 at 05:42:41PM +0200, Stanislaw Gruszka wrote:
> Seems commit:
>
> commit ec15e68ba6a505631016f230899bafbb7b8cd0d6
> Author: Jouni Malinen <[email protected]>
> Date: Wed Mar 23 15:29:52 2011 +0200
>
> cfg80211: Add nl80211 event for deletion of a station entry
>
> is causing problems on my system. Sometimes when I disconnect
> from wireless network, I can not connect again. Without commit
> I can not reproduce the problem.

That sounds strange.. That commit is supposed to just add a new nl80211
event that existing user space tools would ignore (and new
wpa_supplicant snapshot would only use in the case of IBSS). Would it be
possible to get a wpa_supplicant debug log showing this problem?

--
Jouni Malinen PGP id EFC895FA

2011-04-01 09:54:38

by Jouni Malinen

[permalink] [raw]
Subject: Re: ec15e68ba6a regresion?

On Fri, Apr 01, 2011 at 11:25:08AM +0200, Stanislaw Gruszka wrote:

> Seems when disconnect event is handled, we remove some key. I'm
> attaching non-debug and debug logs. Note this is kinda old
> version of wpa_supplicant.

> wpa_driver_wext_disassociate

Huh.. This is with WEXT and the commit
ec15e68ba6a505631016f230899bafbb7b8cd0d6 is adding an nl80211 event. It
should have no change whatsoever to WEXT. I'm not sure whether I
understood the debug log correctly, so I'm not sure what was the area
that behaved incorrectly among multiple successful connections. Anyway,
I do not see how this particular commit could have anything to do with
this. Are you sure that this issues shows only when this commit is in
place and does not show if you revert just this commit?

--
Jouni Malinen PGP id EFC895FA

2011-04-01 09:25:30

by Stanislaw Gruszka

[permalink] [raw]
Subject: Re: ec15e68ba6a regresion?

On Thu, Mar 31, 2011 at 06:46:38PM +0300, Jouni Malinen wrote:
> On Thu, Mar 31, 2011 at 05:42:41PM +0200, Stanislaw Gruszka wrote:
> > Seems commit:
> >
> > commit ec15e68ba6a505631016f230899bafbb7b8cd0d6
> > Author: Jouni Malinen <[email protected]>
> > Date: Wed Mar 23 15:29:52 2011 +0200
> >
> > cfg80211: Add nl80211 event for deletion of a station entry
> >
> > is causing problems on my system. Sometimes when I disconnect
> > from wireless network, I can not connect again. Without commit
> > I can not reproduce the problem.
>
> That sounds strange.. That commit is supposed to just add a new nl80211
> event that existing user space tools would ignore (and new
> wpa_supplicant snapshot would only use in the case of IBSS). Would it be
> possible to get a wpa_supplicant debug log showing this problem?

Seems when disconnect event is handled, we remove some key. I'm
attaching non-debug and debug logs. Note this is kinda old
version of wpa_supplicant.

NetworkManager-0.8.1-7.el6.x86_64
wpa_supplicant-0.6.8-10.el6.x86_64
kernel-latest wireless-testing

Stanislaw


Attachments:
(No filename) (1.07 kB)
wpa_supplicant.log (7.22 kB)
wpa_supplicant.log.debug (180.32 kB)
Download all attachments

2011-04-01 11:11:13

by Stanislaw Gruszka

[permalink] [raw]
Subject: Re: ec15e68ba6a regresion?

On Fri, Apr 01, 2011 at 12:54:28PM +0300, Jouni Malinen wrote:
> On Fri, Apr 01, 2011 at 11:25:08AM +0200, Stanislaw Gruszka wrote:
>
> > Seems when disconnect event is handled, we remove some key. I'm
> > attaching non-debug and debug logs. Note this is kinda old
> > version of wpa_supplicant.
>
> > wpa_driver_wext_disassociate
>
> Huh.. This is with WEXT and the commit
> ec15e68ba6a505631016f230899bafbb7b8cd0d6 is adding an nl80211 event. It
> should have no change whatsoever to WEXT. I'm not sure whether I
> understood the debug log correctly, so I'm not sure what was the area
> that behaved incorrectly among multiple successful connections. Anyway,
> I do not see how this particular commit could have anything to do with
> this. Are you sure that this issues shows only when this commit is in
> place and does not show if you revert just this commit?

I just reproduced the problem with commit reverted. Sorry for noise,
I'm going to investigate issue a bit more.

Stanislaw