Re: [PATCH 2/2] block: enable zone-append for iov_iter of bvec type

[Kanchan Joshi]

On Fri, Jul 03, 2020 at 05:32:56AM +0000, Damien Le Moal wrote:
> On 2020/07/03 0:42, Kanchan Joshi wrote:
> > zone-append with bvec iov_iter gives WARN_ON, and returns -EINVAL.
> > Add new helper to process such iov_iter and add pages in bio honoring
> > zone-append specific constraints.
> >
> > Signed-off-by: Kanchan Joshi <joshi.k@xxxxxxxxxxx>
> > Signed-off-by: Selvakumar S <selvakuma.s1@xxxxxxxxxxx>
> > Signed-off-by: Nitesh Shetty <nj.shetty@xxxxxxxxxxx>
> > Signed-off-by: Javier Gonzalez <javier.gonz@xxxxxxxxxxx>
> > ---
> >  block/bio.c | 31 ++++++++++++++++++++++++++++---
> >  1 file changed, 28 insertions(+), 3 deletions(-)
> >
> > diff --git a/block/bio.c b/block/bio.c
> > index 0cecdbc..ade9da7 100644
> > --- a/block/bio.c
> > +++ b/block/bio.c
> > @@ -975,6 +975,30 @@ static int __bio_iov_bvec_add_pages(struct bio *bio, struct iov_iter *iter)
> >  	iov_iter_advance(iter, size);
> >  	return 0;
> >  }
> > +static int __bio_iov_bvec_append_add_pages(struct bio *bio, struct iov_iter *iter)
> > +{
> > +	const struct bio_vec *bv = iter->bvec;
> > +	unsigned int len;
> > +	size_t size;
> > +	struct request_queue *q = bio->bi_disk->queue;
> > +	unsigned int max_append_sectors = queue_max_zone_append_sectors(q);
> > +	bool same_page = false;
> > +
> > +	if (WARN_ON_ONCE(!max_append_sectors))
> > +		return -EINVAL;
> > +
> > +	if (WARN_ON_ONCE(iter->iov_offset > bv->bv_len))
> > +		return -EINVAL;
> > +
> > +	len = min_t(size_t, bv->bv_len - iter->iov_offset, iter->count);
> > +	size = bio_add_hw_page(q, bio, bv->bv_page, len,
> > +				bv->bv_offset + iter->iov_offset,
> > +				max_append_sectors, &same_page);
> > +	if (unlikely(size != len))
> > +		return -EINVAL;
> > +	iov_iter_advance(iter, size);
> > +	return 0;
> > +}
> >
> >  #define PAGE_PTRS_PER_BVEC     (sizeof(struct bio_vec) / sizeof(struct page *))
> >
> > @@ -1105,9 +1129,10 @@ int bio_iov_iter_get_pages(struct bio *bio, struct iov_iter *iter)
> >
> >  	do {
> >  		if (bio_op(bio) == REQ_OP_ZONE_APPEND) {
> > -			if (WARN_ON_ONCE(is_bvec))
> > -				return -EINVAL;
> > -			ret = __bio_iov_append_get_pages(bio, iter);
> > +			if (is_bvec)
> > +				ret = __bio_iov_bvec_append_add_pages(bio, iter);
> > +			else
> > +				ret = __bio_iov_append_get_pages(bio, iter);
> >  		} else {
> >  			if (is_bvec)
> >  				ret = __bio_iov_bvec_add_pages(bio, iter);
>
> The only user of this function that issues zone append requests is zonefs. The
> issued requests are not using bvec iter but a user direct IO buffer iter. So
> this change would have no user at all as far as I can see. Am I missing
> something ? What IO path makes this change necessary ?

Yes, zonefs does not use bvec iter. But while enabling io-uring path for
zone-append, I hit into this condition returning -EINVAL. 

Reference (from user zone-append series cover letter):
"Append using io_uring fixed-buffer --->
This is flagged as not-supported at the moment. Reason being, for fixed-buffer
io-uring sends iov_iter of bvec type. But current append-infra in block-layer
does not support such iov_iter."

And zone-append doesn't have a problem in using bvec iter as well, so
thought that this may make infra more complete/future-proof?

> --
> Damien Le Moal
> Western Digital Research