Re: [PATCH v6 6/7] iio: adc: Add support for AD4000

[Jonathan Cameron <jic23@xxxxxxxxxx>]

On Mon, 1 Jul 2024 15:10:54 -0300
Marcelo Schmitt <marcelo.schmitt1@xxxxxxxxx> wrote:

> On 06/30, Jonathan Cameron wrote:
> > On Sat, 29 Jun 2024 16:06:59 -0300
> > Marcelo Schmitt <marcelo.schmitt@xxxxxxxxxx> wrote:
> >   
> > > Add support for AD4000 series of low noise, low power, high speed,
> > > successive approximation register (SAR) ADCs.
> > > 
> > > Signed-off-by: Marcelo Schmitt <marcelo.schmitt@xxxxxxxxxx>  
> > 
> > Hi Marcelo
> > 
> > A few comments inline. However, the spi_w8r8 etc can easily be a follow up
> > optimization patch (if you agree it's a good improvement) and the
> > other changes are so trivial I could tweak whilst applying.
> >   
> ...
> > > +	/*
> > > +	 * The gain is stored as a fraction of 1000 and, as we need to
> > > +	 * divide vref_mv by the gain, we invert the gain/1000 fraction.
> > > +	 * Also multiply by an extra MILLI to preserve precision.
> > > +	 * Thus, we have MILLI * MILLI equals MICRO as fraction numerator.
> > > +	 */
> > > +	val = mult_frac(st->vref_mv, MICRO, st->gain_milli);  
> > 
> > If you are rolling a v7 for other reasons, stick some line breaks in here!
> > It's a bit of a mass of text that is hard for my eyes to parse!
> >   
> Ack
> 
> ...
> 
> >   
> > > +static int ad4000_read_reg(struct ad4000_state *st, unsigned int *val)
> > > +{
> > > +	struct spi_transfer t = {
> > > +		.tx_buf = st->tx_buf,
> > > +		.rx_buf = st->rx_buf,
> > > +		.len = 2,
> > > +	};
> > > +	int ret;
> > > +
> > > +	st->tx_buf[0] = AD4000_READ_COMMAND;
> > > +	ret = spi_sync_transfer(st->spi, &t, 1);
> > > +	if (ret < 0)
> > > +		return ret;
> > > +
> > > +	*val = st->rx_buf[1];
> > > +	return ret;  
> > 
> > I'd be tempted to do
> > 
> > 	ssize_t ret;
> > 
> > 	ret = spi_w8r8(AD4000_READ_COMMAND);
> > 	if (ret < 0)
> > 		return ret;
> > 	*val = ret;
> > 	
> > 	return 0;
> >   
> I tried this when working on v6. Only difference was I had declared ret as int.
> Then reg values were not read correctly with spi_w8r8().
> I'm either missing something or reg access must be 16-bit transfer.
> Datasheet sais:
> "The AD4000/AD4004/AD4008 configuration register is read from and written to
> with a 16-bit SPI instruction."
> Yet, besides possible delay between first and last 8 SCLK pulses, I don't see
> any transfer level differences between current and spi_w8r8() versions.
Ah. If you go around again, throw in a comment so we don't 'fix' it in
the future!

> 
> > 
> >   
> ...
> > > +			ret = ad4000_write_reg(st, reg_val);
> > > +			if (ret < 0)
> > > +				return ret;
> > > +
> > > +			st->span_comp = span_comp_en;
> > > +			return ret;  
> > 
> > If you are spinning for another reason, make it clear this is always good.
> > The spi_write() never returns positive so current code is correct but I had
> > to go check which this would have avoided.
> > 
> > 			return 0;  
> 
> Ack
> > 
> > If nothing else comes up, I'll probably tweak whilst applying.
> > 
> > J
> >   
> > > +		}
> > > +		unreachable();
> > > +	default:
> > > +		return -EINVAL;
> > > +	}
> > > +}  
> >