Re: [PATCH v3 2/4] media: i2c: Add ST VD55G0 global shutter sensor driver

[Benjamin Mugnier <benjamin.mugnier@foss.st.com>]

Hi Sergey,

Le 09/11/2026 à 14:27, Sergey Lebedev a écrit :
> Hi Sakari,
> 
> On Mon, Nov 09, 2026 at 02:59:44PM +0200, Sakari Ailus wrote:
>>>>> +static const s64 vd55g0_ev_bias_menu[] = {
>>>>> +	-3000, -2500, -2000, -1500, -1000, -500,
>>>>> +	    0,
>>>>> +	  500,  1000,  1500,  2000,  2500, 3000,
>>>>
>>>> Is this
>>>
>>> The sentence seems to have been cut off here. What did you mean? The
>>> table is vd55g1's, in the control's 0.001 EV units.
>>
>> Uh, yes. I meant to ask if this is a firmware limitation. I guess so?
> 
> No, it is the driver's. The firmware has both a target and a
> compensation register (UM2829, 14.7.3, Table 39 [1]):
> 
>   AE_COMPENSATION       VALUE  15:0  SFP 8.8  Compensation in EV
>   AE_TARGET_PERCENTAGE  VALUE  15:0  UI       Percentage relative to saturation
> 
> where "100 means saturated image". The driver writes the target from
> this table, 27 * 2^EV, as vd55g1 does:
> 
> 	static const unsigned int index2exposure_target[] = {
> 		3, 5, 7, 10, 14, 19, 27, 38, 54, 76, 108, 153, 216,
> 	};
> 
> So +2.5 and +3 EV ask for 153 and 216, above saturation; I have not
> tested what the AE does there.
> 
> v4 keeps vd55g1's table, unless you or Benjamin prefer the compensation
> register.

From a user perspective they are mostly interchangeable. vd55g1 uses
AE_TARGET_PERCENTAGE while vd56g3 uses AE_COMPENSATION for instance.

The downstream driver uses AE_TARGET_PERCENTAGE and that's why you have
it here. It would make more sense to use AE_COMPENSATION semantic wise,
but this is not a deal breaker.

> 
> [1] https://www.st.com/resource/en/user_manual/um2829-how-to-integrate-and-configure-the-vd55g0-device-from-a-hardware-and-software-perspective-stmicroelectronics.pdf
> 
> Sergey

-- 
Regards,
Benjamin