Hi Mark,
On Thu, Jun 11, 2020 at 10:50 AM Mark Brown <broonie@xxxxxxxxxx> wrote:
> On Wed, Jun 10, 2020 at 09:18:19PM +0200, Geert Uytterhoeven wrote:
> > On Wed, Jun 10, 2020 at 6:49 PM Mark Brown <broonie@xxxxxxxxxx> wrote:
>
> > > I'm much more comfortable explicitly listing the new compatible so that
> > > even if someone makes a DT that doesn't bother listing the fallbacks
> > > things will work.
>
> > Adding all of them would cause even more churn when adding support for
> > a new SoC... There are already more than 700 "renesas," compatible
> > values documented that are not directly matched by drivers.
>
> I'm not sure it's a particular concern, especially since you'll be
> sending this stuff in the same series as a bindings update and an extra
> patch in a series makes very little difference.
Until the DT bindings are split off into their own project...
Listing unneeded compatible values in drivers also increases binary size.
For RSPI and MSIOF that would be +2.5 KiB each. Times tens of drivers.
Considering the RSPI driver itself is only 9 KiB, and some RZ/A1 systems
are really memory-constrained, I think it's better to avoid that.
> > Nowadays we have "make dtbs_check", so if a DTS doesn't conform to the
> > binding, it will be flagged.
>
> For things that are upstream.
The DT bindings apply to out-of-tree DTS files, too ;-)
If they're not compliant, all odds are off.
Gr{oetje,eeting}s,
Geert
--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@xxxxxxxxxxxxxx
In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds
