Re: [PATCH v4 2/6] drm/format-helper: Add drm_fb_xrgb8888_to_mono_reversed()

[Javier Martinez Canillas]

Hello Andy,

Thanks for your feedback.

On 2/11/22 12:10, Andy Shevchenko wrote:

[snip]

>> +static void drm_fb_gray8_to_mono_reversed_line(u8 *dst, const u8 *src, unsigned int pixels,
>> +					       unsigned int start_offset, unsigned int end_len)
>> +{
>> +	unsigned int xb, i;
>> +
>> +	for (xb = 0; xb < pixels; xb++) {
>> +		unsigned int start = 0, end = 8;
>> +		u8 byte = 0x00;
> 
>> +		if (xb == 0 && start_offset)
>> +			start = start_offset;
> 
> This is invariant to the loop, can be moved out.
> 
>> +		if (xb == pixels - 1 && end_len)
>> +			end = end_len;
> 
> Ditto. However it may require to factor out the following loop to a helper.
>

Not sure I'm following, it's not invariant since it depends on the
loop iterator value. It only applies to the first and last pixels.
 
[snip]

>> +	/*
>> +	 * The reversed mono destination buffer contains 1 bit per pixel
>> +	 * and destination scanlines have to be in multiple of 8 pixels.
>> +	 */
>> +	if (!dst_pitch)
>> +		dst_pitch = DIV_ROUND_UP(linepixels, 8);
> 
> round_up() ?
> 

But it's not a round up operation but a div and round up.

>> +	WARN_ONCE(dst_pitch % 8 != 0, "dst_pitch is not a multiple of 8\n");
> 
> 
> I would move this to the if conditional, i.e.
> 
> 	if (dst_pitch)
> 		WARN_ONCE(dst_pitch % 8 != 0, "dst_pitch is not a multiple of 8\n");
> 	else
> 		dst_pitch = round_up(linepixels, 8);
>

No, because we always need to div and round up. The warning is just printed to
let know that the dst pitch is not a multiple of 8 as it should be. So callers
could be fixed.

>> +	/*
>> +	 * The cma memory is write-combined so reads are uncached.
> 
> CMA
>

Yes, this bug me too. But other format helpers (e.g: drm_fb_xrgb8888_to_rgb565
and drm_fb_xrgb8888_to_gray8) had this comment with CMA in lower case. So did
the same for consistency.

>> +	 * Speed up by fetching one line at a time.
>> +	 *
>> +	 * Also, format conversion from XR24 to reversed monochrome
>> +	 * are done line-by-line but are converted to 8-bit grayscale
>> +	 * as an intermediate step.
>> +	 *
>> +	 * Allocate a buffer to be used for both copying from the cma
>> +	 * memory and to store the intermediate grayscale line pixels.
>> +	 */
>> +	src32 = kmalloc(len_src32 + linepixels, GFP_KERNEL);
> 
> size_add() ?
>

I wasn't familiar with this macro and git grep returned nothing. Then I noticed
that it is fairly new, introduced in commit a66866cff71c ("overflow: Implement
size_t saturating arithmetic helpers").

git tag --contains a66866cff71c | head -1
next-20220207

So I can't really use it since isn't yet in the latest drm-misc-next base tag :)

>> +	if (!src32)
>> +		return;
> 
> ...
> 
>> +	/*
>> +	 * For damage handling, it is possible that only parts of the source
>> +	 * buffer is copied and this could lead to start and end pixels that
>> +	 * are not aligned to multiple of 8.
>> +	 *
>> +	 * Calculate if the start and end pixels are not aligned and set the
>> +	 * offsets for the reversed mono line conversion function to adjust.
>> +	 */
>> +	start_offset = clip->x1 % 8;
>> +	end_len = clip->x2 % 8;
> 
> ALIGN() ?
> 

But we don't want to align here but to know what's the start and end if is
not aligned since that would mean converting to mono in the middle of a byte.

Best regards,
-- 
Javier Martinez Canillas
Linux Engineering
Red Hat