On Mon, Mar 06, 2023 at 02:16:24PM +0200, Andy Shevchenko wrote:
> On Mon, Mar 06, 2023 at 02:11:48PM +0200, Andy Shevchenko wrote:
> > On Sat, Mar 04, 2023 at 05:57:51PM +0000, Jonathan Cameron wrote:
> > > On Tue, 28 Feb 2023 07:31:51 +0100
> > > Mike Looijmans <mike.looijmans@xxxxxxxx> wrote:
...
> > > > + for (i = 0; i < 4; i++) {
> > > > + if (BIT(i) == gain) {
> > > > + ads1100_set_config_bits(data, ADS1100_PGA_MASK, i);
> > > > + return 0;
> > > > + }
> > > > + }
> > > Andy's suggestion of something like..
> > > if (!gain)
> > > return -EINVAL;
> > > i = ffs(gain);
> > > if (i >= 4 || BIT(i) != gain)
> > > return -EINVAL;
> > >
> > > ads...
> > >
> > > Is perhaps nicer than the loop.
> >
> > Even better:
> >
> > if (!gain || !is_power_of_2(gain))
> > return -EINVAL;
>
> Or if you want to combine all checks:
> if (clamp_val(gain, BIT(0), BIT(3)) != gain || !is_power_of_2(gain))
> return -EINVAL;
Just a side note. I have been wanting to have is_in_range() for a long time.
Perhaps we can add a such to the kernel (math.h)
/* Check if in the range [start .. end] */
#define is_in_range(value, start, end) (value >= start && value <= end)
With it, the above will probably better look
if (!is_in_range(gain, BIT(0), BIT(3) || !is_power_of_2(gain))
> ads1100_set_config_bits(data, ADS1100_PGA_MASK, ffs(gain));
> return 0;
>
> (You can play with bloat-o-meter for the code generation and see which one is
> better from that aspect)
--
With Best Regards,
Andy Shevchenko
