[RFC PATCH linux dev-4.13 2/3] dts: aspeed-g5: Expose VGA scratch registers

Andrew Jeffery andrew at aj.id.au
Thu Apr 19 14:17:37 AEST 2018

> >
> >>
> >>
> >> > +               bit-mask = <0xffffffff>;
> >> > +               bit-shift = <0>;
> >>
> >> Should we assume this is the default? Or something else? Having a
> >> default for the common case would be nice.
> >
> > Whilst the nodes I've added in these examples expose an entire register, I think this would actually be a bit of an unusual case. More likely you'll want to expose a single bit from some register, in which case a default for mask (though maybe 0x1?) or shift isn't going to make any sense.
> Ok. If you can think of something that makes the bindings less verbose
> I'm all for that.

Part of the issue (IMO) is we're stretching the intent of the driver here to expose the scratch registers. Hence the bullet point on sysfs vs chardev mentioned in the cover letter. Using the driver for the scratch registers was driven by some conversations with BenH; we could not do this and go the chardev route instead. I'd like Ben to chime in, as I know JK is more for the chardev approach.


More information about the openbmc mailing list