Hi Laurent,
You are right, and the two extra entries should not be in the binding.
Both drivers take exactly three GPIOs:
drivers/gpu/drm/bridge/ti-tpd12s015.c:150,157,164
drivers/video/fbdev/omap2/omapfb/displays/encoder-tpd12s015.c:237,246,255
They ask for indices 0, 1 and 2 -- CT_CP_HPD, LS_OE and HPD. Nothing reads
index 3 or 4, and neither file mentions scl, sda or ddc.
The text binding I replaced said the same thing:
- gpios: CT CP HPD, LS OE and HPD gpios
omap5-igep0050.dts passes two more, and there is no pinctrl group for them
and nothing in-tree consumes them. v3 raised maxItems to five only so that
dtbs_check would keep passing for that board. That was the wrong direction:
it wrote a defect of the DT source into the binding instead of holding the
binding to what the hardware and the driver use.
So it splits the way you pointed at earlier:
- the conversion keeps gpios at three entries, which is what the driver
requires, and
- a separate patch drops the two extra entries from omap5-igep0050.dts.
My earlier question still stands: would you rather have that as a v4 from
me, or should I leave it to Eduard's series?
Thanks,
Chaosheng Qu
