Unset CONFIG_P80211

[Karl Relton]

On Mon, 2009-04-20 at 17:03 +0200, Stefan Bader wrote:
> Tim Gardner wrote:
> > Stefan Bader wrote:
> >> Tim Gardner wrote:
> >>> Karl Relton wrote:
> >>>> In the Jaunty kernel configs, CONFIG_P80211 can now be un-set.
> >>>>
> >>>> This config enables the p80211 module in the Ubuntu sauce, which used to
> >>>> be a prerequisite of the prism2_usb module in the Ubuntu sauce.
> >>>>
> >>>> Both of these are now superseeded by the new prism2_usb module in
> >>>> mainline (that has its own p80211 code compiled in). The Jaunty configs
> >>>> unset the old prism2_usb, but have forgot to unset the old p80211, so it
> >>>> gets compiled and installed unnecessarily.
> >>>>
> >>>> Going forward (Jaunty+1), I guess the Ubuntu sauce for these can be
> >>>> dispensed with altogether.
> >>>>
> >>>> Regards
> >>>> Karl
> >>>>
> >>>>
> >>> Well, that is interesting. Even though CONFIG_STAGING_EXCLUDE_BUILD=y,
> >>> drivers/staging/wlan-ng still gets built. I suspect it is because it
> >>> uses the same config option CONFIG_PRISM2_USB. This is _not_ what I
> >>> intended, but it is what it is now (and will be for the released
> >>> kernel). I guess if its working, then we ought to prefer the staging
> >>> driver over the one in ubuntu/misc/wireless/prism2_usb.
> >>>
> >>> Stefan - I think you should consider completely removing the
> >>> ubuntu/misc/wireless/prism2_usb and ubuntu/misc/wireless/p80211
> >>> directories in order to avoid future confusion. They are already gone in
> >>> Karmic.
> >>>
> >> The question is how well old and new are working. It seems currently the
> >> one in ubuntu/ is used. Removing everything will change it to use the
> >> kernel driver and if that acts differently we have one more regression.
> >> So if things are working for now, even if not perfectly I would tend to
> >> leave it the way it is.
> >>
> >> Stefan
> > 
> > That sounds reasonable to me. Perhaps we should change the config option
> >  in drivers/staging/wlan-ng so that it doesn't get built, thereby
> > reducing opportunities for confusion.
> > 
> > rtg
> 
> Yes, that makes sense, so nobody stumbles over duplicate drivers.
> Noted on my todo.
> 
> 
> Stefan
> 

Right - leaving it using the ubuntu/ version makes sense for this kernel
version and will continue to work well for end users (like myself).
Going forward (i.e. karmic) we can switch to the staging/ version.
Functionally it is the same code, though there are some debug warning
message cleanups that need sorting out in it.

I am also trying to work with upstream to improve staging/ version
functionality, namely moving firmware loading to in-kernel (its
currently done from a userspace utility launched from udev). I'll let
you know if/when upstream have accepted this.

Karl