Re: NM 0.6.5 unable to reconnect to WPA2 network (first connection fine) - take 2

[Dan Williams <dcbw redhat com>]

On Thu, 2007-06-07 at 12:37 -0400, Jon Nettleton wrote:
> On Thu, 2007-06-07 at 12:08 -0400, Dan Williams wrote:
> > On Thu, 2007-06-07 at 16:42 +0100, Jon Escombe wrote:
> > > ----- "Dan Williams" <dcbw redhat com> wrote:
> > > 
> > > > > I believe Tomas is using 6.4, which doesn't work (for him) because
> > > > it always tries to get a password from the keyring. 6.5 includes a
> > > > patch to never get a password from the keyring (which I'm assuming
> > > > would work for Tomas - but breaks the private key password scenario).
> > > > This *should* now be the best of both worlds  ;)
> > > > 
> > > > Ah, ok.  Can you test out latest SVN for 0.6.x and confirm that I didn't
> > > > break anything when modifying your patch?
> > > > 
> > > > Thanks,
> > > > Dan
> > > 
> > > Yep, the patch from SVN works fine here, thanks.
> > > 
> > > Any thoughts on applying the device-link-race patch to 0.6.x?
> > 
> > Working on that now; I'd rather figure out why nm_device_is_up doesn't
> > return the correct value though...  In any case, it's also better to
> > move the sleep(1) to real_act_stage2_config() in
> > nm-device-802-11-wireless.c after the ap_need_key() call since only
> > wireless devices appear to need this.  Can you try the attached patch
> > and see if that works for you _instead_ of device-link-race.patch?
> > 
> 
> Dan,
> 
> your patch works for me.  I have also found that this problem exists on
> a couple of mac80211 stack drivers, as well as my prism54 fullmac card.

I just tried running a quick test program on 2.6.20 with prism54 with
inconclusive results...  It just runs system("/sbin/ip link set dev eth1
[up/down]") and then uses the nm_device_is_up() function to print out
status.  Seems to work OK there.  Grr.  dev-link-wait.patch committed in
revision 2579 just because.

Dan