Re: [PATCH 14/20] drm: add convenience function to create an enum property

[Sascha Hauer <s.hauer@pengutronix.de>]

On Fri, Feb 03, 2012 at 10:08:12AM +0000, Dave Airlie wrote:
> On Wed, Feb 1, 2012 at 1:05 PM, Sascha Hauer <s.hauer@pengutronix.de> wrote:
> > On Wed, Feb 01, 2012 at 07:55:40AM -0500, David Airlie wrote:
> >>
> >>
> >> > > diff --git a/include/drm/drm_crtc.h b/include/drm/drm_crtc.h
> >> > > index 8d593ad..cdbbb40 100644
> >> > > --- a/include/drm/drm_crtc.h
> >> > > +++ b/include/drm/drm_crtc.h
> >> > > @@ -394,7 +394,7 @@ struct drm_crtc {
> >> > >   s64 framedur_ns, linedur_ns, pixeldur_ns;
> >> > >
> >> > >   /* if you are using the helper */
> >> > > - void *helper_private;
> >> > > + struct drm_crtc_helper_funcs *helper_private;
> >> > >  };
> >> > >
> >> > >
> >> > > @@ -481,7 +481,7 @@ struct drm_encoder {
> >> > >
> >> > >   struct drm_crtc *crtc;
> >> > >   const struct drm_encoder_funcs *funcs;
> >> > > - void *helper_private;
> >> > > + struct drm_encoder_helper_funcs *helper_private;
> >> > >  };
> >> > >
> >> > >  enum drm_connector_force {
> >> > > @@ -573,7 +573,7 @@ struct drm_connector {
> >> > >   /* requested DPMS state */
> >> > >   int dpms;
> >> > >
> >> > > - void *helper_private;
> >> > > + struct drm_connector_helper_funcs *helper_private;
> >> > >
> >> > >   /* forced on connector */
> >> > >   enum drm_connector_force force;
> >> >
> >> > This is a separate chunk.
> >>
> >> And totally wrong, using the helper should remain optional, and the
> >> helper includes should not be included into the main headers.
> >
> > The intention was to prevent people from thinking that they
> > should invent their own set of helper functions. As these
> > are only pointers we could also add a
> >
> > struct drm_crtc_helper_funcs;
> > struct drm_connector_helper_funcs;
> > struct drm_encoder_helper_funcs;
> >
> > to drm_crtc.h without having to include the helper includes.
> >
> 
> This might be better, though driver can invent their own helpers if
> they need it, its the whole point of using helpers and not forcing
> drivers to do stuff one single way.

I hope to get the drivers more uniform. When people want to invent
their own helpers I think the question we have to ask is what is wrong
with the helpers and what is missing and try to fix/implement this
before adding a new set of helpers. If someone really has a good reason
to implement new helpers we can easily revert this change, but we
shouldn't motivate him to do so.

Sascha


-- 
Pengutronix e.K.                           |                             |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |
Peiner Str. 6-8, 31137 Hildesheim, Germany | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |