Re: CVS (STABLE_0_3) works for Madwifi/Atheros, but starts too late

[Dan Williams <dcbw redhat com>]

On Thu, 2005-05-12 at 14:14 -0400, Derek Atkins wrote:
> Quoting Dan Williams <dcbw redhat com>:
> 
> > On Thu, 2005-05-12 at 12:34 -0400, Derek Atkins wrote:
> > > Any chance you could tag a 0.3.5 release and push an update out to FC3? 
> > (Not
> > > that it matters to me; I built my own from the CVS code, but others might
> > want
> > > it).
> > 
> > Yay :)  I'll check out stability on my FC3 laptop and investigate.
> 
> I did have a few issues with it today..  I was connected to a wireless network
> and then suspended the machine (which tearsdown the wireless drivers).  When I
> resumed (which reloads the wireless drivers) NM didn't notice and didn't try to
> connect to my (different) wireless network.
> 
> Now, this COULD have been due to the fact that I didn't realize I forgot to
> restart NMInfo, so it's POSSIBLE that that is part of the problem.  I'll be
> able to test tonight when I get back to my hotel room and resume.  Looking at
> the logs, however, it just resumed the DHCP and didn't notice that the AP had
> gone away:

NM is likely to be very unhappy if you just suspend it, and wireless
drivers in general have issues with suspend.  There really is no method
for NM to know when you've put the computer to sleep since nothing sends
global events out (PowerManager+HAL will eventually do this the Right
Way using dbus).

NM does have a "sleep" dbus method that will make NM stop and down all
devices, and a "wake" method to wake it up.  Those should be able to be
done in the ACPI scripts fairly simply:

/usr/bin/dbus-send --system --dest=org.freedesktop.NetworkManager
--type=method-call /org/freedesktop/NetworkManager
org.freedesktop.NetworkManager.sleep
sleep 5;

and:

/usr/bin/dbus-send --system --dest=org.freedesktop.NetworkManager
--type=method-call /org/freedesktop/NetworkManager
org.freedesktop.NetworkManager.wake

Dan