On Mon, Nov 20, 2023 at 05:55:12PM +0200, Ceclan Dumitru wrote:
> On 11/20/23 15:00, Andy Shevchenko wrote:
> > On Thu, Nov 16, 2023 at 03:46:55PM +0200, mitrutzceclan wrote:
...
> >> +struct ad7173_channel_config {
> >> + bool live;
> >> + u8 cfg_slot;
> >> + /* Following fields are used to compare equality. Bipolar must be first */
> >> + bool bipolar;
> >> + bool input_buf;
> >> + u8 odr;
> >> + u8 ref_sel;
> >
> > If you group better by types, it might save a few bytes on the architectures /
> > compilers where bool != byte.
> >
> Grouping by type will result in not being able to use memcmp() for
> comparing configs. But then there is the issue that I was under the
> assumption that bool=byte. If that is not the case, the config equality
> check might be comparing padding bytes.
It's most likely the case, BUT... it is not guaranteed by the C standard.
> In this case what do you suggest:
> - using the packed attribute
> - using only u8
> - drop memcmp, manually compare fields
Use struct_group() to show explicitly the group of members. If it's not an ABI
in any sense, then memcmp() is fine.
...
> >> + cmp_size = sizeof(*cfg) - offset;
> >
> > sizeof_field() from the above mentioned header?
>
> This computes the size of multiple fields, following cfg_slot. Better to
> group the fields that need to be compared into another struct then use
> sizeof_field()?
See above.
...
> >> + return vref / (MICRO/MILLI);
> >
> > What does the denominator mean and why you can't simply use MILL?
>
> Original vref values are in micro, I considered that it was adequate to
> represent the conversion from MICRO to MILLI as a fraction.
>
> >> + *val = st->info->sinc5_data_rates[reg] / (MICRO/MILLI);
> >> + *val2 = (st->info->sinc5_data_rates[reg] % MILLI) * (MICRO/MILLI);
> >
> > Same Q about denominator.
> >
> Here, a misunderstanding on my part of a suggestion from Jonathan in V2,
> will be removed.
You need to clarify with him that.
--
With Best Regards,
Andy Shevchenko
