Re: [PATCH v12 07/16] iio: afe: rescale: add INT_PLUS_{MICRO,NANO} support

From: Peter Rosin <peda@axentia.se>
To: Andy Shevchenko <andy.shevchenko@gmail.com>,
	Liam Beguin <liambeguin@gmail.com>
Cc: Jonathan Cameron <jic23@kernel.org>,
	Lars-Peter Clausen <lars@metafoo.de>,
	Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
	linux-iio <linux-iio@vger.kernel.org>,
	devicetree <devicetree@vger.kernel.org>,
	Rob Herring <robh+dt@kernel.org>
Subject: Re: [PATCH v12 07/16] iio: afe: rescale: add INT_PLUS_{MICRO,NANO} support
Date: Sun, 9 Jan 2022 21:20:09 +0100	[thread overview]
Message-ID: <8046b39f-28c1-7e46-e27c-6b9bc8824e21@axentia.se> (raw)
In-Reply-To: <CAHp75Vdi2bvCEtxpezt5L5JhO=8D+Za++CbQ8AximFaLnxnqyg@mail.gmail.com>

Hi!

On 2022-01-09 13:48, Andy Shevchenko wrote:
> On Sat, Jan 8, 2022 at 10:53 PM Liam Beguin <liambeguin@gmail.com> wrote:
>>
>> Some ADCs use IIO_VAL_INT_PLUS_{NANO,MICRO} scale types.
>> Add support for these to allow using the iio-rescaler with them.
> 
> ...
> 
>> +               mult = scale_type == IIO_VAL_INT_PLUS_NANO ? GIGA : MEGA;
>> +
>> +               /*
>> +                * For IIO_VAL_INT_PLUS_{MICRO,NANO} scale types if either *val
>> +                * OR *val2 is negative the schan scale is negative, i.e.
>> +                * *val = 1 and *val2 = -0.5 yields -1.5 not -0.5.
>> +                */
>> +               neg = *val < 0 || *val2 < 0;
>> +
>> +               tmp = (s64)abs(*val) * abs(rescale->numerator);
>> +               *val = div_s64_rem(tmp, abs(rescale->denominator), &rem);
>> +
>> +               tmp = (s64)rem * mult + (s64)abs(*val2) * abs(rescale->numerator);
>> +               tmp = div_s64(tmp, abs(rescale->denominator));
> 
> Isn't it too many repetitive abs() calls?
> 
> What about
> 
> // Create a macro and use for u16 (struct rn5t618_channel_ratios), s16
> (struct twl4030_prescale_divider_ratios), s32 (can be reused in struct
> rescale)
> struct u32_fract {
>   u32 numerator;
>   u32 denominator;
> };
> // (potential reuse in struct hclge_ptp_cycle) and so on...
> 
>   struct u32_fract fract = {
>     .numerator = abs(rescale->numerator),
>     .denominator = abs(rescale->denominator),
>   };
> 
> // obviously we can add a macro/inliner to abs() the fract struct and
> return original sign
> 
> and reuse fract.numerator, fract.denominator?

This feels a bit excessive when the "problem" is two extra abs calls.
I don't think the code will get any easier to read by changing
abs(rescale->denominator) into fract.denominator and with my maintainer
hat on, I vote for just letting the compiler exercise its CSE engine.

Cheers,
Peter