Re: mpc5200b custom board upstreamable?

["Grant Likely" <grant.likely@secretlab.ca>]

On Thu, Apr 24, 2008 at 9:53 AM, Sascha Hauer <s.hauer@pengutronix.de> wrote:
> On Thu, Apr 24, 2008 at 09:13:45AM -0600, Grant Likely wrote:
>  > On Thu, Apr 24, 2008 at 9:12 AM, Sascha Hauer <s.hauer@pengutronix.de> wrote:
>  > >
>  > >  Hi all,
>  > >
>  > >  I had the intention to push the code for a custom mpc5200b board (freely
>  > >  available, no internal project) upstream. After cleaning up the code I
>  > >  realized that actually no board specific code is left and our board is
>  > >  well handled by the mpc5200_simple_platform machine.
>  > >
>  > >  The only issue is that the machine only matches things like
>  > >  "schindler,cm5200", there's no generic entry. Would it be possible to
>  > >  add a "generic-mpc52xx" entry to this list?
>  >
>  > I'm being cautious about this for the time being.  I'd like to have a
>  > generic match mechanism, but I don't want to do something that isn't
>  > easy to recover from if it turns out to be brain dead.  For now, just
>  > add your board name to the explicit match list.
>
>  The board is called "generic". No, just kidding ;)

/me slaps Sascha

Seriously though; I do intend to fix this, but I don't think adding a
generic entry to the compatible list is the right way to do it.  For
example, what would "mpc5200-generic" really mean anyway?  Convention
for usage of 'compatible' would indicate that it means the *entire
board* is compatible (obviously not true).  The use-case you're
talking about is simply "the board uses a 5200 and firmware is sane".
On the other hand, I may just be overthinking things and compatible is
the most appropriate place to specify that the board is a mpc5200
based board.  (please feel free to argue with my; my opinion can
probably be swayed... attaching promises of beer to your argument is
probably an effective strategy)

This is an issue that probably affects the other embedded platforms
too, so it would be nice to agree on a common method of handling it.

Regardless, whatever method is chosen, it is also important that it is
always possible for board specific fixups to override the generic
behavior.

Cheers,
g.

-- 
Grant Likely, B.Sc., P.Eng.
Secret Lab Technologies Ltd.