Re: [PATCH 5/7] block: Rework bio_for_each_folio_all()

[Kent Overstreet]

On Fri, May 26, 2023 at 10:36:03AM +1000, Dave Chinner wrote:
> On Thu, May 25, 2023 at 05:48:20PM -0400, Kent Overstreet wrote:
> > This reimplements bio_for_each_folio_all() on top of the newly-reworked
> > bvec_iter_all, and since it's now trivial we also provide
> > bio_for_each_folio.
> > 
> > Signed-off-by: Kent Overstreet <kent.overstreet@xxxxxxxxx>
> > Cc: Matthew Wilcox <willy@xxxxxxxxxxxxx>
> > Cc: linux-block@xxxxxxxxxxxxxxx
> > ---
> >  fs/crypto/bio.c        |  9 +++--
> >  fs/iomap/buffered-io.c | 14 ++++---
> >  fs/verity/verify.c     |  9 +++--
> >  include/linux/bio.h    | 91 +++++++++++++++++++++---------------------
> >  include/linux/bvec.h   | 15 +++++--
> >  5 files changed, 75 insertions(+), 63 deletions(-)
> ....
> > diff --git a/include/linux/bio.h b/include/linux/bio.h
> > index f86c7190c3..7ced281734 100644
> > --- a/include/linux/bio.h
> > +++ b/include/linux/bio.h
> > @@ -169,6 +169,42 @@ static inline void bio_advance(struct bio *bio, unsigned int nbytes)
> >  #define bio_for_each_segment(bvl, bio, iter)				\
> >  	__bio_for_each_segment(bvl, bio, iter, (bio)->bi_iter)
> >  
> > +struct folio_vec {
> > +	struct folio	*fv_folio;
> > +	size_t		fv_offset;
> > +	size_t		fv_len;
> > +};
> 
> Can we drop the "fv_" variable prefix here? It's just unnecessary
> verbosity when we know we have a folio_vec structure. i.e fv->folio
> is easier to read and type than fv->fv_folio...

That's actually one of the things I like about bio/biovec, it's been
handy in the past for grepping and block layer refactorings...

(I would _kill_ for a tool that let me do that kind of type-aware grep.
ctags can in theory produce that kind of an index but I never figured
out how to get vim to use it properly. I believe the lsp-server stuff
that uses the compiler as a backend can do it; I've started using that
stuff for Rust coding and it works amazingly, don't think I've tried it
for struct members - I wonder if that stuff works at all on a codebase
the size of the kernel or just dies...)

> Hmmm, this is probably not a good name considering "struct pagevec" is
> something completely different - the equivalent is "struct
> folio_batch" but I can see this being confusing for people who
> largely expect some symmetry between page<->folio naming
> conventions...

Yeah, good point. folio_seg, perhaps?

(I think Matthew may have already made that suggestion...)

> Also, why is this in bio.h and not in a mm/folio related header
> file?

Is it worth moving it there considering it's only used in bio.h/bvec.h?
Perhaps we could keep it where it's used for now and move it if it gains
more users?