Re: [V7, 2/2] media: i2c: dw9768: Add DW9768 VCM driver

[Dongchun Zhu <dongchun.zhu@mediatek.com>]

Hi Andy,

On Mon, 2020-06-08 at 16:27 +0300, Andy Shevchenko wrote:
> On Sat, Jun 06, 2020 at 02:19:18PM +0800, Dongchun Zhu wrote:
> > On Fri, 2020-06-05 at 15:46 +0300, Andy Shevchenko wrote:
> > > On Fri, Jun 05, 2020 at 06:54:12PM +0800, Dongchun Zhu wrote:
> 
> ...
> 
> > > > +	depends on I2C && VIDEO_V4L2
> > > 
> > > No compile test?
> > > 
> > 
> > Sorry?
> > Kconfig here is based on the current media tree master branch.
> > It is also what the other similar drivers from Dongwoon do. 
> 
> COMPILE_TEST.
> I dunno if it's established or not practice in media subsystem.
> 
> ...
> 
> > > > +/*
> > > > + * DW9768 requires waiting time (delay time) of t_OPR after power-up,
> > > > + * or in the case of PD reset taking place.
> > > > + */
> > > > +#define DW9768_T_OPR_US				1000
> > > > +#define DW9768_Tvib_MS_BASE10			(64 - 1)
> > > > +#define DW9768_AAC_MODE_DEFAULT			2
> > > 
> > > > +#define DW9768_AAC_TIME_DEFAULT			0x20
> > > 
> > > Hex? Why not decimal?
> > > 
> > 
> > There is one optional property 'dongwoon,aac-timing' defined in DT.
> > I don't know whether you have noticed that.
> > 
> > 'DW9768_AAC_TIME_DEFAULT' is the value set to AACT[5:0] register.
> > I thought the Hex unit should be proper as it is directly written to the
> > Hex register.
> 
> I see. I would rather put it like (BIT(6) / 2) to show explicitly that we
> choose half of the resolution.
> 

I knew your idea.
'(BIT(6) / 2)' may somewhat show the meaning of 'median of the total
range of AACT[5:0]'.

But this value is still very obscure relative to '0x20'.
As I thought that simple is the best, especially for kernel upstream
patch.

> ...
> 
> > > > +	val = ((unsigned char)ret & ~mask) | (val & mask);
> > > 
> > > This cast is weird.
> > > 
> > 
> > Thanks for the review, but this cast is using classical pattern from
> > your suggestion on OV02A10 v5:
> > https://patchwork.linuxtv.org/patch/59788/
> > 
> > So I wonder whether it is still required to be refined currently.
> > Or what would it be supposed to do for the cast?
> 
> Okay, does it produce a warning w/o cast?
> If yes, replace it at least to be the same type as mask and val.
> 

No.

> ...
> 
> > > > +	for ( ; val >= 0; val -= DW9768_MOVE_STEPS) {
> > > > +		ret = dw9768_set_dac(dw9768, val);
> > > > +		if (ret) {
> > > > +			dev_err(&client->dev, "I2C write fail: %d", ret);
> > > > +			return ret;
> > > > +		}
> > > > +		usleep_range(move_delay_us, move_delay_us + 1000);
> > > > +	}
> > > 
> > > 
> > > It will look more naturally in the multiplier kind of value.
> > > 
> > > 	unsigned int steps = DIV_ROUND_UP(...);
> > > 
> > > 	while (steps--) {
> > > 		...(..., steps * ..._MOVE_STEPS);
> > > 		...
> > > 	}
> > > 
> > > but double check arithmetics.
> > 
> > The current coding style is actually updated with reference to your
> > previous comments on DW9768 v3:
> > https://patchwork.linuxtv.org/patch/61856/
> 
> I understand, but can you consider to go further and see if the proposal works?
> 

In fact, your suggestion is something like one another method to set DAC
value to actuator.
It is like there may be several solutions to a question.

Yes. I just tried the new method and it works as expected.
u32 steps = DIV_ROUND_UP(dw9768->focus->val, DW9768_MOVE_STEPS);
while (steps--) {
	ret = dw9768_set_dac(dw9768, steps * DW9768_MOVE_STEPS);
	if (ret)
		return ret;
	usleep_range(move_delay_us, move_delay_us + 1000);
}

But from my perspective, I may prefer to the original method.
As here what we really care is the DAC value,
'dw9768_set_dac(dw9768, val)' seems more simple.