Re: [PATCH 12/13] regulator: add pm8008 pmic regulator driver

[Andy Shevchenko <andy.shevchenko@xxxxxxxxx>]

On Tue, May 14, 2024 at 6:05 PM Satya Priya Kakitapalli (Temp)
<quic_skakitap@xxxxxxxxxxx> wrote:
> On 5/14/2024 7:48 PM, Andy Shevchenko wrote:
> > On Tue, May 14, 2024 at 5:05 PM Satya Priya Kakitapalli
> > <quic_skakitap@xxxxxxxxxxx> wrote:
> >>> On Thu, May 09, 2024 at 03:07:02PM +0300, Andy Shevchenko wrote:
> >>>> Wed, May 08, 2024 at 10:37:50PM +0000, Stephen Boyd kirjoitti:
> >>>>> Quoting Johan Hovold (2024-05-06 08:08:29)

...

> >>>>>> +               BUILD_BUG_ON((ARRAY_SIZE(pldo_ranges) != 1) ||
> >>>>> This should be an && not || right?
> >>>>>> +                               (ARRAY_SIZE(nldo_ranges) != 1));
> >>>> In any case BUILD_BUG_ON() is not encouraged for such cases, it would be much
> >>>> better to have a static_assert() near to one of those arrays.
> >>> I think the reason it is placed here is that the above line reads:
> >>>
> >>>        rdesc->n_linear_ranges = 1;
> >>>
> >>> and that would need to change if anyone expands the arrays.
> >> Correct. static_assert() cannot be used in the middle of code here, it can only be used at the declarations part which doesn't serve the purpose.
> > I didn't get this. The ARRAY_SIZE():s are defined at compile time
> > globally. How does this prevent from using static_assert()?

> The reason we added it here is to make sure the nlod_ranges and
> pldo_ranges doesn't become larger, and we forget updating the
> n_linear_ranges.

> Adding static_assert here is not feasible so adding a
> BUILD_BUG_ON at this point makes sure the n_linear_ranges is proper.

No, static_assert() will do _exactly_ the same with better error
reporting and location, but what you are trying to say is that the
location is chosen to be near to the n_liner_ranges assignment which
happens at runtime, that's why it can't be used as an argument to
BUILD_BUG_ON(). Based on this discussion I think the comment is
missing before BUILD_BUG_ON() to explain the semantics of 1 and all
that was just said.

> >> So, BUILD_BUG_ON is the only way to go here.
> > I don't think so.

As i said.

-- 
With Best Regards,
Andy Shevchenko