On Tue, Aug 23, 2016 at 11:39:06AM +0200, Hans de Goede wrote:
> Hi,
>
> On 23-08-16 11:26, Maxime Ripard wrote:
> >On Mon, Aug 22, 2016 at 09:03:57PM +0200, Hans de Goede wrote:
> >>Hi,
> >>
> >>On 22-08-16 20:30, Maxime Ripard wrote:
> >>>On Mon, Aug 08, 2016 at 09:43:14PM +0200, Hans de Goede wrote:
> >>>>The gt90h tablet has a gsl3675 touchscreen, add a dt node describing it.
> >>>>
> >>>>Signed-off-by: Hans de Goede <hdegoede@xxxxxxxxxx>
> >>>>---
> >>>>arch/arm/boot/dts/sun8i-a23-gt90h-v4.dts | 8 ++++++++
> >>>>1 file changed, 8 insertions(+)
> >>>>
> >>>>diff --git a/arch/arm/boot/dts/sun8i-a23-gt90h-v4.dts b/arch/arm/boot/dts/sun8i-a23-gt90h-v4.dts
> >>>>index f27ebbb..da55b5a 100644
> >>>>--- a/arch/arm/boot/dts/sun8i-a23-gt90h-v4.dts
> >>>>+++ b/arch/arm/boot/dts/sun8i-a23-gt90h-v4.dts
> >>>>@@ -53,6 +53,14 @@
> >>>> status = "okay";
> >>>>};
> >>>>
> >>>>+&gsl1680 {
> >>>>+ compatible = "silead,gsl3675";
> >>>>+ touchscreen-fw-name = "silead/gsl3675-gt90h.fw";
> >>>
> >>>That's not documented anywhere, and looks really suspicious.
> >>
> >>Ugh, that should have been in:
> >>
> >>Documentation/devicetree/bindings/input/touchscreen/silead_gsl1680.txt
> >>
> >>But somehow it is not (I believe it was there in earlier revisions of
> >>the patch), I'll send a patch to fix this.
> >>
> >>About it being suspicious, this is not really firmware it is a bunch
> >>of configuration data / lookup tables for the controller which tell
> >>it in which order the touchscreen horizontal / vertical sensor
> >>lines are connected to its sense pins, and what values to send
> >>for finger x% between line z and line z+1, which differs per
> >>tablet model, since not all tablets use the same digitizer.
> >
> >It's not really the firmware itself that I find suspicious, but more
> >the encoding of a path to a file in the DT,
>
> It is not a path it is a filename. We could drop the "silead/" prefix
> and put that in the driver instead to really make it a filename.
>
> >especially when you can
> >apparently derive it from other informations already found in the DT
> >(<vendor>/<compatible>-<board>.fw)
>
> That will not work, sometimes different boards use the same digitizer
> and thus the same firmware. Also in case of the q8 tablets, we need
> different firmware for different variants (this is to be dealt with
> by the q8-hardware-manager I'm working on), since although they
> all use the same digitizer they do not wire it up to the controllers
> pins the same in all models, so we need different firmware files
> corresponding to different wirings.
>
> TL;DR: There is no 1:1 mapping between board and the firmware filename.
The point still holds. It's exactly the same case than the broadcom
nvram filename discussion, and it raised the same issues.
> Also note that this is not the first driver to introduce a firmware
> filename in a dts binding, a quick grep shows at least:
>
> Documentation/devicetree/bindings/remoteproc/wkup_m3_rproc.txt
>
> Also does this, as well as several dt-bindings where board specific
> firmware gets embedded in the device tree itself.
Which has never been acked by any of the DT maintainers.
Maxime
--
Maxime Ripard, Free Electrons
Embedded Linux and Kernel engineering
http://free-electrons.com
Attachment:
signature.asc
Description: PGP signature
