Re: ipw3945 with wpa-psk and NM



On Mon, 2006-10-09 at 21:49 -0400, Louis Garcia wrote:
> On Mon, 2006-10-09 at 13:41 -0400, Louis Garcia wrote:
> > On Mon, 2006-10-09 at 09:57 -0400, Dan Williams wrote:
> > > On Sun, 2006-10-08 at 22:42 -0400, Louis Garcia wrote:
> > > > On Sat, 2006-10-07 at 23:26 -0400, Louis Garcia wrote:
> > > > > Just got a new HP dv2000 laptop with ipw3945. Setup winXP with wpa-psk
> > > > > and a netgear router and it worked. Couldn't wait to get FC6 on it.
> > > > > After a bit of a struggle with the ipw3935 driver and the reg-daemon I
> > > > > finally managed to get the wireless working but only the old way. I
> > > > > created a network-script ifcfg-eth1 with an open network and it worked.
> > > > > Now I would like to get NM working with wpa-psk.
> > > > > 
> > > > > Should I remove the network script and start new? Do I have to do
> > > > > anything with wpa_supplicant? Is there a fedora specific howto
> > > > > somewhere?
> > > > > 
> > > > > Thanks, --Louis 
> > > > > 
> > > > > _______________________________________________
> > > > > NetworkManager-list mailing list
> > > > > NetworkManager-list gnome org
> > > > > http://mail.gnome.org/mailman/listinfo/networkmanager-list
> > > > 
> > > > This is the log of NM: I believe for some reason ipw3945 is off. I have
> > > > ask the ipw3945 developers what is a frequency kill switch and how do I
> > > > turn it off.  
> > > 
> > > Most new laptops with builtin wireless have an RF kill switch these
> > > days.  It allows you to hit a hardware button on the laptop and
> > > immediately cease transmission of anything from the card.  For
> > > airplanes, for example.  In any case, there isn't yet a good interface
> > > for wireless cards & drivers under Linux.  It would be great if there
> > > was, because then we could detect whether or not the card's RF was off,
> > > and then do the right thing.  But we're not going to hack up solutions
> > > to every different driver.  Furthermore, not all drivers that support RF
> > > kill switches actually expose them to userspace at all.  Worse yet, kill
> > > switches are often OEM-dependent and the interface with which they
> > > connect is different between say HP and Dell even when the same exact
> > > wireless hardware is used.
> > > 
> > > It's a mess.
> > > 
> > > Dan
> > 
> > There is a switch in front of the laptop which I have on all the time. I
> > have found the rf_kill file in /sys and it was set to 0. What I don't
> > get is if I bring the wireless interface up the old way, with a
> > network-script, it works. It doesn't work for NM though. I hope there is
> > a model-specific code for HP somewhere.
> > 
> > 
> > -Louis
> 
> To confirm this, I created a network-script
> called /etc/sysconfig/network-scripts/ifcfg-eth1 then i started it,
> #ifup eth1, the led lit up, got the ip and the card was connected. I set
> the router to open for this test. Now I stopped eth1, #ifdown eth1. I
> started NetworkManager and tried to connect to the same open router and
> couldn't. The led of the wireless was off.
> 
> Is Network-manager bringing up the wireless card the same way as the
> network-script? 

NM brings up the card by setting the interface IFF_UP, just like a
'/sbin/ifconfig eth1 up' would.  Maybe the ifup scripts load/unload the
module or something?  That's pretty crack if true, but it would be very
interesting to know what happens here.  AFAIK with ipw adapters, there's
a "soft" kill that's driver-based, and a "hard" kill that's hardware
switch based.  But it OEM-dependent whether or not the hardware switch
just sends an event to the driver, which does softkill, or whether the
switch is actually wired up to the card in some way.  If
unloading/loading the module turns the card back on, we have our answer.

The question is, does the driver send any events out when the radio
dies?  If not, it should, but that interface is still being standardized
in hte kernel.  In the end, DavidZ is going to integrate this with HAL I
think, so that it will be easy for things to figure out for
NetworkManager.

Dan




[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]