Re: [PATCH v2] block : add larger order folio size instead of pages

[Christoph Hellwig <hch@lst.de>]

On Thu, May 02, 2024 at 08:45:33AM +0200, Hannes Reinecke wrote:
>> -		nr_pages = (fi.offset + fi.length - 1) / PAGE_SIZE -
>> -			   fi.offset / PAGE_SIZE + 1;
>> -		do {
>> -			bio_release_page(bio, page++);
>> -		} while (--nr_pages != 0);
>> +		bio_release_page(bio, page);
>
> Errm. I guess you need to call 'folio_put()' here, otherwise the page 
> reference counting will be messed up.

It shouldn't.  See the rfc patch and explanation that Keith sent in reply
to the previous version.  But as I wrote earlier it should be a separate
prep patch including a commit log clearly explaining the reason for it
and how it works.

> That is not a valid conversion.
> bvec_try_to_merge_pages() will try to merge a page with an existing
> bvec. If the logic is switch to folios you would need to iterate over
> all pages in a folio, and call bvec_try_to_merge_pages() for each page
> in the folio.
> Or convert / add a function 'bvec_try_to_merge_folio()'.
> But with the above patch it will only ever try to merge the first page
> in the folio.

You mean bvec_try_merge_page?  The only place where it uses more
informtation than just the physical address is for the same_page
logic.  That beeing said converting it to a folio would still be
a good idea and a good prep patch.