On 06/25/2013 09:28 AM, Linus Walleij wrote:

On Fri, Jun 21, 2013 at 11:17 PM, Stephen Warren [off-list ref] wrote:

quoted

And finally, I don't really like using pin groups for the purpose of
defining these mappings, since I intended them to purely represent the
mapping from register fields to the set of affected pins. However, I can
see an argument for doing this, since the pin groups are in fact still
representing /some/ aspect of the pinctrl internal HW.

The groups concept was part of the first pinctrl commit
2744e8afb3b76343e7eb8197e8b3e085036010a5
and there the intention was clearly just to define a discrete
set of pins.

And some drivers use it like this still, with no connection to
physical registers, i.MX come to mind.

But I do seem to recall some endless discussions about this,
I think we need to agree to disagree.

But the whole point of a subsystem is to provide clear common semantics
across all the different drivers that comprise it. IMHO, it's a great
failing of pinctrl that it doesn't clearly define its data model at all,
and just leaves individual driver authors to use groups in whatever
random fashion they want. We really should have different entries in the
pinctrl data model for these different concepts (real HW groups, and
logical/virtual/SW groups) since they're entirely different things with
different semantics.

Perhaps it's simplest if I just step out of pinctrl and let it exist as
it is.