Re: [PATCH v3] omap: beagle: add support for wl1271 on the board file

[Luciano Coelho <luciano.coelho@nokia.com>]

On Fri, 2010-09-24 at 21:11 +0200, ext Tony Lindgren wrote:
> * Luciano Coelho <luciano.coelho@nokia.com> [100924 11:42]:
> > On Fri, 2010-09-24 at 20:03 +0200, ext John W. Linville wrote:
> > > On Thu, Sep 23, 2010 at 01:03:35PM +0300, Luciano Coelho wrote:
> > > > Add board configuration for the wl1271 daughter board.  This patch is based
> > > > on Ohad Ben-Cohen's patches for Zoom boards.
> > > > 
> > > > Cc: Ohad Ben-Cohen <ohad@wizery.com>
> > > > Signed-off-by: Luciano Coelho <luciano.coelho@nokia.com>
> > > > ---
> > > > from v2 to v3: removed accidental change to wl1271.h
> > > 
> > > I presume that this should come through my tree as well (since it
> > > contains the move of wl12xx.h)?
> > 
> > Yes, this is based on your tree.  But please hold on, because there were
> > some comments (namely, the expansion board should be detected, not
> > configured for wl1271 by default), which I still have to fix.  I'm
> > waiting for the manufacturer of the expansion board to provide me with
> > some more info for proper detection.
> > 
> > Also, I know that there are some other work being done in the beagle
> > board file, so we may need to rethink how to sync this.
> > 
> > So, for now, let's skip this patch and I'll provide another one when
> > these issues are addressed.
> 
> You could also do it based on some kernel cmdline option. Of course
> detection during the runtime would be better.

Yes, this is more or less what Openembedded (?) guys seem to be doing.
You can check Robert Nelson's email in the "v2" thread.

The already have some changes in u-boot (which are apparently not in its
mainline yet) that does exactly that.  The bootloader probes the
expansion board and passes the result in a kernel cmdline option.  This
is "runtime", but I think the best solution would be in the end to let
the kernel do the probing itself.

As far as I could tell after some googling, having the bootloader probe
this kind of stuff is legacy because of problems with muxing in older
kernels, so the omap muxing had to be done in the bootloader.  Now that
muxing is working fine in newer kernel, this kind of thing should be
moved to the kernel, or what do you think?

-- 
Cheers,
Luca.