Re: NM fails to find access points

[Rene Rask <rene grain dk>]

Dan Williams wrote:
> On Thu, 2008-03-27 at 20:28 +0100, René Rask wrote:
> > > > I can replicate this reliably by restarting the NM daemon. That always
> >  >> leaves my AP off the list. I can also reliably find my AP again by using
> >  >> "Create Net Wireless Network". When the new network fails to connect it
> >  >> will always show my AP (and the rest of them) again. Sometimes it takes
> >  >> a short while for it to appear, but it always does. Seems like a
> >  >> scanning delay.
> >  >>
> >  >> Stopping NM, removing the wireless kernel module, insmod it and starting
> >  >> NM again also has the effect of making it find my AP again.
> >  >
> >  > So this is pointing more towards a driver problem.  If you ever have to
> >  > rmmod/insmod and stuff starts magically working again, it is almost
> >  > always a driver issue.
> >  >
> >  > One more question: does your AP almost always show with 'iwlist', but
> >  > usually never when doing the dbus-send to the supplicant?
> >  >
> >  > You might also try to do an 'iwlist' and then right afterwards, poke the
> >  > supplicant with the dbus-send command and see if your AP is in the list.
> >  > If the iwconfig and the supplicant-reported APs are quite different,
> >  > then the supplicant is the problem.
> >  >
> >  > If you can poke around a bit with these things, let me know what happens
> >  > so we can narrow down the issue further.
> >  >
> >
> > I can't get this to work anymore. Last time wpa_supplicant wasn't 
> > started a bootup so I started it manually.
> >
> > [root rr-laptop ~]# dbus-send --system 
> > --dest=fi.epitest.hostap.WPASupplicant --print-reply 
> > /fi/epitest/hostap/WPASupplicant/Interfaces/1 
> > fi.epitest.hostap.WPASupplicant.Interface.scanResults
> > Error org.freedesktop.DBus.Error.UnknownMethod: Method "scanResults" 
> > with signature "" on interface 
> > "fi.epitest.hostap.WPASupplicant.Interface" doesn't exist
> >
> > I can't restart wpa_supplicant unless I stop NM first. And I still can 
> > get the dbus command to work.
> >
> > (a while later..)
> > Hmm... now it works again. I was rmmodding/insmodding the iwl4965 modul 
> > with debug and other paramaters and stopping/starting NM while this 
> > happened.
> >
> > iwlist scan and the supplicant list match pretty good. Having the same 
> > number of APs after a iwlist scan.
> >
> > iwlist doesn't always show my AP an neither does supplicant. But they 
> > seem to agree on showing the same APs. The list of AP also goes from 3 
> > to 14 found and anything in between.
> > The scans have no effect on what is show in the applet.
> >
> > When the bug is in effect the number of found APs is 4-8 and my AP isn't 
> > found by NM or supplicant.
>
> But it _is_ seen by iwlist?  If it's not seen by iwlist either, then
> it's a driver bug.  It's important to note that not every AP will be in
> the scan results every time, because wireless is just hard and not
> reliable.  However, I would certainly expect a non-hidden AP to show up
> after 2 or 3 back-to-back scans at most.

No it isn't found at all when the bug is in effect. Doing "Create new.." 
dance makes it show up again.

I tried many times. But it doesn't show up.

Rene