Re: [v4 PATCH 1/2] crypto: scatterwalk - Use nth_page instead of doing it by hand

[Eric Biggers <ebiggers@kernel.org>]

On Fri, Mar 14, 2025 at 11:27:20AM +0800, Herbert Xu wrote:
> Curiously, the Crypto API scatterwalk incremented pages by hand
> rather than using nth_page.  Possibly because scatterwalk predates
> nth_page (the following commit is from the history tree):
> 
> 	commit 3957f2b34960d85b63e814262a8be7d5ad91444d
> 	Author: James Morris <jmorris@intercode.com.au>
> 	Date:   Sun Feb 2 07:35:32 2003 -0800
> 
> 	    [CRYPTO]: in/out scatterlist support for ciphers.
> 
> Fix this by using nth_page.

As I said on v2, this needs an explanation of what it's actually fixing.

> @@ -189,14 +195,18 @@ static inline void scatterwalk_done_dst(struct scatter_walk *walk,
>  	 * reliably optimized out or not.
>  	 */
>  	if (ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE) {
> -		struct page *base_page, *start_page, *end_page, *page;
> +		struct page *base_page;
> +		unsigned int offset;
> +		int start, end, i;
>  
>  		base_page = sg_page(walk->sg);
> -		start_page = base_page + (walk->offset >> PAGE_SHIFT);
> -		end_page = base_page + ((walk->offset + nbytes +
> -					 PAGE_SIZE - 1) >> PAGE_SHIFT);
> -		for (page = start_page; page < end_page; page++)
> -			flush_dcache_page(page);
> +		offset = walk->offset;
> +		start = offset >> PAGE_SHIFT;
> +		end = start + (nbytes >> PAGE_SHIFT);
> +		end += (offset_in_page(offset) + offset_in_page(nbytes) +
> +			PAGE_SIZE - 1) >> PAGE_SHIFT;

The change to how the end page index is calculated is unrelated to the nth_page
fix, and it makes the code slower and harder to understand.  My original code
just rounded the new offset up to a page boundary to get the end page index.

- Eric