Re: [PATCH v2 2/5] media: ov02c10: Adjust x-win/y-win when changing flipping to preserve bayer-pattern

[Hans de Goede <johannes.goede@oss.qualcomm.com>]

Hi Sakari,

On 7-Jan-26 12:48, Sakari Ailus wrote:
> Hi Hans,
> 
> On Wed, Dec 10, 2025 at 12:24:33PM +0100, Hans de Goede wrote:
>> The ov02c10 is capable of having its (crop) window shifted around with 1
>> pixel precision while streaming.
>>
>> This allows changing the x/y window coordinates when changing flipping to
>> preserve the bayer-pattern.
> 
> Ideally we'd use the crop selection to configure this, but given these
> sensors (and drivers) are generally what they are, I guess this is probably
> a reasonable way to go.

Even in sensor drivers where we allow setting the crop selection
to get arbritrary modes, we always round the coordinates to a multiple
of 2, to avoid changing the bayer-pattern seen by userspace when
userspace changes the crop. See e.g. ov2680_set_selection().

And then when doing flipping we might add 1 to either the x and/or y
coordinate to the userspace provided crop x, y before sending it to
the sensor to make flipping not change the bayer order, see e.g.
ov01a10_set_hflip() after the v2 series you've here:

https://git.retiisi.eu/?p=~sailus/linux.git;a=shortlog;h=refs/heads/ov01a10

which does (simplified):

        offset = crop->left;
        if (hflip)
                offset++;

        cci_write(ov01a10->regmap, OV01A10_REG_X_WIN, offset, &ret);

IOW we are trying to not make userspace be able to affect the bayer-pattern
through setting the crop-selection and/or flip.

So I'm not sure what you mean with "Ideally we'd use the crop selection"
because we are actively trying to avoid to have the crop-selection change
the bayer order ?

Generally speaking I think we should avoid any settings change the bayer-order
whenever possible.

Regards,

Hans


here?