Re: [PATCH 2/4] drm/format-helper: Add drm_fb_gray8_to_mono_reversed()

[Javier Martinez Canillas <javierm@redhat.com>]

Hello Geert,

On 3/14/22 14:40, Geert Uytterhoeven wrote:
> Hi Javier,
> 
> On Mon, Jan 31, 2022 at 9:12 PM Javier Martinez Canillas
> <javierm@redhat.com> wrote:
>> Add support to convert 8-bit grayscale to reversed monochrome for drivers
>> that control monochromatic displays, that only have 1 bit per pixel depth.
>>
>> This helper function was based on repaper_gray8_to_mono_reversed() from
>> the drivers/gpu/drm/tiny/repaper.c driver.
>>
>> Signed-off-by: Javier Martinez Canillas <javierm@redhat.com>
> 
>> --- a/drivers/gpu/drm/drm_format_helper.c
>> +++ b/drivers/gpu/drm/drm_format_helper.c
>> @@ -584,3 +584,38 @@ int drm_fb_blit_toio(void __iomem *dst, unsigned int dst_pitch, uint32_t dst_for
>>         return -EINVAL;
>>  }
>>  EXPORT_SYMBOL(drm_fb_blit_toio);
>> +
>> +/**
>> + * drm_fb_gray8_to_mono_reversed - Convert grayscale to reversed monochrome
>> + * @dst: reversed monochrome destination buffer
> 
> What's the meaning of "reversed"?

Originally I took this helper from the repaper driver and it was called
repaper_gray8_to_mono_reversed(), so the naming just got carried over.

> During the last few days, I've been balancing between (a) "reverse
> video" and (b) "reverse bit order", but none of them seems to be true.
>
> (a) The code maps 0-127 to 0 and 8-255 to 1, which just reduces from
>     256 to 2 grayscale levels, without inversion. The result is also
>     white-on-black on my ssd130x OLED.
> (b) On little-endian, the CFB drawops use little-endian bit order,
>     which is what ends up in "byte" in the code below.
>

As far as I understand is (a) from the options since the repaper display
uses black-on-white. But as you pointed out the ssd130x OLEDs instead do
white-on-black.

I should had probably just name the helper drm_fb_gray8_to_monochrome()
or maybe drm_fb_gray8_to_gray1() ?

-- 
Best regards,

Javier Martinez Canillas
Linux Engineering
Red Hat