Re: [PATCH v9 4/6] mm, iomap: keep a partly covered page until an uncached write completes it

[Christoph Hellwig <hch@lst.de>]

On Sat, Oct 24, 2026 at 06:11:50PM -0400, Mike Snitzer wrote:
> An uncached (RWF_DONTCACHE) buffered write marks drop-behind every folio
> it creates, so writeback drops each one as soon as this write's data is
> on disk.

Please write proper commit message.  This sentence is absolutely
unreadable crap, and it gets worse from there.

If you can't state the problem you're fixing in your own words, you
really should not be touching this code.

> Measured locally on XFS (emulated 4Kn NVMe, 4 KiB pages), 32 threads

And avoid useless details.  What is emulated here, why does it matter?
Unlike SCSI/SATA nvme doesn't have the concept of native vs non-native
4k block sizses, and even for SCSI/SATA HDD that was purely a marketing
term which doesn't matter for code.

<snip pages and pages of irrelevant crap>

Do you really expect someone to read all this?  Please!

>  
>  	if (iter->flags & IOMAP_NOWAIT)
>  		fgp |= FGP_NOWAIT;
> -	if (iter->flags & IOMAP_DONTCACHE)
> +	if (iter->flags & IOMAP_DONTCACHE) {
> +		size_t unit = mapping_min_folio_nrbytes(mapping);
> +		size_t off = pos & (unit - 1);
> +
>  		fgp |= FGP_DONTCACHE;
> +		/*
> +		 * The smallest folio the mapping allows, a page or a block
> +		 * where the block is larger, that this write covers only
> +		 * partly is shared with a neighbouring write: get it as just
> +		 * that, which is kept until the write that completes it.

What the heck is this monster sentence supposed to mean?

> +	/*
> +	 * The page cache marks an edge it finds uptodate.  With blocks
> +	 * smaller than the folio, the edge this write completes is not
> +	 * uptodate yet: mark it if its other blocks are.
> +	 */

When did the page cache become an acting entity?