[v3] pinctrl: renesas: checker: Rework drive and bias pin iteration

The checker code to iterate over all drive strength and bias register
description items is cumbersome, due to the repeated calculation of
indices, and the use of hardcoded array sizes.  The latter was done
under the assumption they would never need to be changed, which turned
out to be false.

Increase readability by introducing helper macros to access drive
strength and bias register description items.
Increase maintainability by replacing hardcoded numbers by array sizes
calculated at compile-time.

Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
---
v3:
  - Add and use drive_ofs() and bias_ofs() helpers, as suggested by
    Wolfram,

v2:
  - Replace "pinctrl: renesas: checker: Fix for drive reg field
    increase" by independent proper solution.

To be inserted into renesas-pinctrl-for-v5.19 before commit
d5c9688095d29a6c ("pinctrl: renesas: Allow up to 10 fields for
drive_regs") in renesas-pinctrl-for-v5.19.
---
 drivers/pinctrl/renesas/core.c | 61 ++++++++++++++++++----------------
 1 file changed, 33 insertions(+), 28 deletions(-)

On Tue, Apr 19, 2022 at 10:08:32AM +0200, Geert Uytterhoeven wrote:
> The checker code to iterate over all drive strength and bias register
> description items is cumbersome, due to the repeated calculation of
> indices, and the use of hardcoded array sizes.  The latter was done
> under the assumption they would never need to be changed, which turned
> out to be false.
> 
> Increase readability by introducing helper macros to access drive
> strength and bias register description items.
> Increase maintainability by replacing hardcoded numbers by array sizes
> calculated at compile-time.
> 
> Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>

Yay, no hardcoded maximum anymore and much better to read. Win-win!

Reviewed-by: Wolfram Sang <wsa+renesas@sang-engineering.com>

On Tue, Apr 19, 2022 at 10:08 AM Geert Uytterhoeven
<geert+renesas@glider.be> wrote:
> The checker code to iterate over all drive strength and bias register
> description items is cumbersome, due to the repeated calculation of
> indices, and the use of hardcoded array sizes.  The latter was done
> under the assumption they would never need to be changed, which turned
> out to be false.
>
> Increase readability by introducing helper macros to access drive
> strength and bias register description items.
> Increase maintainability by replacing hardcoded numbers by array sizes
> calculated at compile-time.
>
> Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
> ---
> v3:
>   - Add and use drive_ofs() and bias_ofs() helpers, as suggested by
>     Wolfram,

> --- a/drivers/pinctrl/renesas/core.c
> +++ b/drivers/pinctrl/renesas/core.c
> @@ -1007,7 +1007,18 @@ static void __init sh_pfc_compare_groups(const char *drvname,
>  static void __init sh_pfc_check_info(const struct sh_pfc_soc_info *info)
>  {
>         const struct pinmux_drive_reg *drive_regs = info->drive_regs;
> +#define drive_nfields  ARRAY_SIZE(drive_regs->fields)
> +#define drive_ofs(i)   drive_regs[(i) / drive_nfields]
> +#define drive_reg(i)   drive_ofs(i).reg
> +#define drive_bit(i)   ((i) % drive_nfields)
> +#define drive_field(i) drive_ofs(i).fields[drive_bit(i)]
>         const struct pinmux_bias_reg *bias_regs = info->bias_regs;
> +#define bias_npins     ARRAY_SIZE(bias_regs->pins)
> +#define bias_ofs(i)    bias_regs[(i) / bias_npins]
> +#define bias_puen(i)   bias_ofs(i).puen
> +#define bias_pud(i)    bias_ofs(i).pud
> +#define bias_bit(i)    ((i) % bias_npins)
> +#define bias_pin(i)    bias_ofs(i).pins[bias_bit(i)]
>         const char *drvname = info->name;
>         unsigned int *refcnts;
>         unsigned int i, j, k;

> @@ -1164,20 +1175,17 @@ static void __init sh_pfc_check_info(const struct sh_pfc_soc_info *info)
>         for (i = 0; drive_regs && drive_regs[i].reg; i++)
>                 sh_pfc_check_drive_reg(info, &drive_regs[i]);
>
> -       for (i = 0; drive_regs && drive_regs[i / 8].reg; i++) {
> -               if (!drive_regs[i / 8].fields[i % 8].pin &&
> -                   !drive_regs[i / 8].fields[i % 8].offset &&
> -                   !drive_regs[i / 8].fields[i % 8].size)
> +       for (i = 0; drive_regs && drive_reg(i); i++) {
> +               if (!drive_field(i).pin && !drive_field(i).offset &&
> +                   !drive_field(i).size)
>                         continue;
>
>                 for (j = 0; j < i; j++) {
> -                       if (drive_regs[i / 8].fields[i % 8].pin ==
> -                           drive_regs[j / 8].fields[j % 8].pin &&
> -                           drive_regs[j / 8].fields[j % 8].offset &&
> -                           drive_regs[j / 8].fields[j % 8].size) {
> +                       if (drive_field(i).pin == drive_field(j).pin &&
> +                           drive_field(j).offset && drive_field(j).size) {
>                                 sh_pfc_err("drive_reg 0x%x:%u/0x%x:%u: pin conflict\n",
                                                             ^^      ^^
> -                                          drive_regs[i / 8].reg, i % 8,
> -                                          drive_regs[j / 8].reg, j % 8);
> +                                          drive_reg(i), drive_bit(i),
> +                                          drive_reg(j), drive_bit(j));

Whoops, as reported by kernel test robot for 64-bit builds, drive_bit()
is no longer unsigned int, but size_t[*], hence "%zu" should be used
for printing.  The same is true for bias_bit().

Will fix up tomorrow...

[*] A bit counter-intuitive from the mathematical point of view,
    but as "size_t" is either "unsigned int" or "unsigned long",
    "unsigned int % size_t" is ... "size_t"!

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds