turning off IBSS (with wext)

turning off IBSS (with wext)

[Johannes Berg]

[-- Attachment #1: Type: text/plain, Size: 455 bytes --]

Hi,

so I was testing IBSS, and found this strange behaviour:

 * wpa_supplicant (or NM?) will try to do SIWMLME(disassoc) in IBSS
   mode, which seems kinda wrong. Why do we accept it? We send out a
   frame and then do nothing else.
 * Then I thought "iwconfig wlan1 essid off" - but that fails with
   -EINVAL due to ieee80211_sta_find_ibss

It seems to me that either one of those, or both, should actually turn
off beaconing.

johannes

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: turning off IBSS (with wext)

[Dan Williams]

On Fri, 2009-02-06 at 14:57 +0100, Johannes Berg wrote:
> Hi,
> 
> so I was testing IBSS, and found this strange behaviour:
> 
>  * wpa_supplicant (or NM?) will try to do SIWMLME(disassoc) in IBSS
>    mode, which seems kinda wrong. Why do we accept it? We send out a
>    frame and then do nothing else.
>  * Then I thought "iwconfig wlan1 essid off" - but that fails with
>    -EINVAL due to ieee80211_sta_find_ibss
> 
> It seems to me that either one of those, or both, should actually turn
> off beaconing.

Either one; the supplicant uses SIWMLME(disassoc) at various times (like
when leaving a BSS), and NM will sometimes set a NULL SSID when
deactivating the device.

Dan

Re: turning off IBSS (with wext)

[Johannes Berg]

[-- Attachment #1: Type: text/plain, Size: 829 bytes --]

On Fri, 2009-02-06 at 11:41 -0500, Dan Williams wrote:
> On Fri, 2009-02-06 at 14:57 +0100, Johannes Berg wrote:
> > Hi,
> > 
> > so I was testing IBSS, and found this strange behaviour:
> > 
> >  * wpa_supplicant (or NM?) will try to do SIWMLME(disassoc) in IBSS
> >    mode, which seems kinda wrong. Why do we accept it? We send out a
> >    frame and then do nothing else.
> >  * Then I thought "iwconfig wlan1 essid off" - but that fails with
> >    -EINVAL due to ieee80211_sta_find_ibss
> > 
> > It seems to me that either one of those, or both, should actually turn
> > off beaconing.
> 
> Either one; the supplicant uses SIWMLME(disassoc) at various times (like
> when leaving a BSS), and NM will sometimes set a NULL SSID when
> deactivating the device.

Fun. Neither one actually works.

johannes

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: turning off IBSS (with wext)

[Dan Williams]

On Fri, 2009-02-06 at 21:40 +0100, Johannes Berg wrote:
> On Fri, 2009-02-06 at 11:41 -0500, Dan Williams wrote:
> > On Fri, 2009-02-06 at 14:57 +0100, Johannes Berg wrote:
> > > Hi,
> > > 
> > > so I was testing IBSS, and found this strange behaviour:
> > > 
> > >  * wpa_supplicant (or NM?) will try to do SIWMLME(disassoc) in IBSS
> > >    mode, which seems kinda wrong. Why do we accept it? We send out a
> > >    frame and then do nothing else.
> > >  * Then I thought "iwconfig wlan1 essid off" - but that fails with
> > >    -EINVAL due to ieee80211_sta_find_ibss
> > > 
> > > It seems to me that either one of those, or both, should actually turn
> > > off beaconing.
> > 
> > Either one; the supplicant uses SIWMLME(disassoc) at various times (like
> > when leaving a BSS), and NM will sometimes set a NULL SSID when
> > deactivating the device.
> 
> Fun. Neither one actually works.

Yeah; I had to recently patch wpa_supplicant to really disassociate at
Jouni's suggestion; just telling the driver to disassoc isn't adequate
in all cases, because some drivers will just try to reassoc.  Instead,
unlocking the BSSID and setting a bogus SSID will usually actually
accomplish the disassociation in addition to the MLME disassoc...

Dan

Re: turning off IBSS (with wext)

[Johannes Berg]

[-- Attachment #1: Type: text/plain, Size: 1416 bytes --]

On Fri, 2009-02-06 at 15:49 -0500, Dan Williams wrote:

> > > >  * wpa_supplicant (or NM?) will try to do SIWMLME(disassoc) in IBSS
> > > >    mode, which seems kinda wrong. Why do we accept it? We send out a
> > > >    frame and then do nothing else.
> > > >  * Then I thought "iwconfig wlan1 essid off" - but that fails with
> > > >    -EINVAL due to ieee80211_sta_find_ibss
> > > > 
> > > > It seems to me that either one of those, or both, should actually turn
> > > > off beaconing.
> > > 
> > > Either one; the supplicant uses SIWMLME(disassoc) at various times (like
> > > when leaving a BSS), and NM will sometimes set a NULL SSID when
> > > deactivating the device.
> > 
> > Fun. Neither one actually works.
> 
> Yeah; I had to recently patch wpa_supplicant to really disassociate at
> Jouni's suggestion; just telling the driver to disassoc isn't adequate
> in all cases, because some drivers will just try to reassoc.  Instead,
> unlocking the BSSID and setting a bogus SSID will usually actually
> accomplish the disassociation in addition to the MLME disassoc...

To be honest, I'm not sure I like disassoc, it doesn't actually make any
sense since we never associated. OTOH, wext claims we did... It's all a
little fishy. What do you mean by "bogus SSID" here? Hopefully the empty
"off" SSID I had in point 2? If so, I guess I'll make that work, or
something...

johannes

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 836 bytes --]