Re: [PATCH v11 2/3] NFSD: Implement NFSD_IO_DIRECT for NFS WRITE

[Mike Snitzer <snitzer@kernel.org>]

On Mon, Nov 10, 2025 at 01:17:19AM -0800, Christoph Hellwig wrote:
> On Fri, Nov 07, 2025 at 03:28:06PM -0500, Chuck Lever wrote:
> > +1 for an explanatory comment, but I'm not getting what is "counter-
> > intuitive" about leaving the content of the end segments in the page
> > cache. The end segments are small and simply cannot be handled by direct
> > I/O.
> 
> The downside if of course extra page cache usage / pollution.  If these
> segments are aligned to the file system minimum I/O size, but not to
> the memory requirements there is not going to be any RMW cycle that
> leaving them in the page cache for would be useful.

Straw man: the entire IO would be issued using DONTCACHE (due to it
being "case 2"). You too keep conflating "case 1" and "case 2":
https://lore.kernel.org/linux-nfs/20251107153422.4373-1-cel@kernel.org/T/#mca19127d102dd7d24b2e44df4d9417b2cc61b340

Your broader point of "extra page cache usage / pollution".  DONTCACHE
is already using page cache, it just will hopefully drop-behind its
pages.

But for "case 1", yes if we use cached buffered IO for the subpage
ends _BUT_ the workload isn't ever going to actually need to RMW
(e.g. because it isn't part of a continuous stream of unaligned
WRITEs) then: Yes, those individual pages associated with the
unaligned ends will polute. BUT they are individual pages that MM is
quite good about handling.

> Similarly if the single segment is not aligned in the file logic space,
> chances of having another RMW come in are the same as for a large
> one with unaligned head and/or tail.

Not following you, please reword and be clearer.