[PATCH v2] of: Add videomode helper
Sascha Hauer
s.hauer at pengutronix.de
Thu Sep 13 04:19:54 PDT 2012
On Thu, Sep 13, 2012 at 01:54:07PM +0300, Tomi Valkeinen wrote:
> On Wed, 2012-07-04 at 09:56 +0200, Sascha Hauer wrote:
> > This patch adds a helper function for parsing videomodes from the devicetree.
> > The videomode can be either converted to a struct drm_display_mode or a
> > struct fb_videomode.
>
> I have more or less the same generic comment for this as for the power
> sequences series discussed: this would add panel specific information
> into DT data, instead of the driver handling it. But, as also discussed
> in the thread, there are differing opinions on which way is better.
With the panel timings I think the approach of moving them into DT is
the best we can do. There are so many displays out there, patching the
kernel each time a customer comes with some new display is no fun.
>
> > +int of_get_video_mode(struct device_node *np, struct drm_display_mode *dmode,
> > + struct fb_videomode *fbmode);
>
> From caller's point of view I think it'd make more sense to have
> separate functions to get drm mode or fb mode. I don't think there's a
> case where the caller would want both.
Ok, that makes sense.
>
> Then again, even better would be to have a common video mode struct used
> by both fb and drm... But I think that's been on the todo list of
> Laurent for a long time already =).
Yes, indeed. We should go into that direction. We already realized that
we want to have ranges instead of fixed values for the timings.
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 |
More information about the dri-devel
mailing list