Re: [PATCH] pinctrl: pinctrl-single: Fix pcs_request_gpio() when bits_per_mux != 0

From: Tony Lindgren
Date: Tue Jan 23 2018 - 12:45:28 EST


* Andy Shevchenko <andy.shevchenko@xxxxxxxxx> [180123 09:25]:
> On Mon, Jan 22, 2018 at 7:06 PM, David Lechner <david@xxxxxxxxxxxxxx> wrote:
> > On 01/22/2018 08:49 AM, Andy Shevchenko wrote:
> >> On Mon, Jan 22, 2018 at 1:03 AM, David Lechner <david@xxxxxxxxxxxxxx>
> >> wrote:
>
> >>> Fixes: 4e7e8017a80e ("pinctrl: pinctrl-single: enhance to configure
> >>> multiple pins of different modules")
>
> >> One line?
>
> > One line is more important that wrapping to 75 chars?
>
> It's *special* line, i.e. tag. Tag per line is a rule I know, did I
> miss any new change to that?

I think the rule for merge commits is to try to limit them to 75
characters for git log.

> >>> + byte_num = (pcs->bits_per_pin * pin) /
> >>> BITS_PER_BYTE;
> >>> + offset = (byte_num / mux_bytes) * mux_bytes;
> >>> + pin_shift = pin % (pcs->width /
> >>> pcs->bits_per_pin) *
> >>> + pcs->bits_per_pin;
>
> >> Sounds like playing around pretty well defined macro and functions,
> >> e.g. DIV_ROUND_UP(), round_up().
>
> > I admit, I just copied existing code (which may be a reason to leave this
> > the
> > way it is).
>
> Ah, fair enough.
>
> > But, I only see once place to do this:
> >
> > offset = round_down(byte_num, mux_bytes);
> >
> > Did I miss another?
>
> I meant that you may use macros to make code cleaner. Though, taking
> above into consideration, it would be done as a separate patch later.

Yeah makes sense to me to do them separately. It would be good to
get Haojian's ack or tested-by on this one, I don't think I have
hardware where I can test the GPIO features with this driver.

Regards,

Tony