Re: [PATCH V11 15/19] block: enable multipage bvecs

[Ming Lei]

On Wed, Nov 21, 2018 at 03:55:02PM +0100, Christoph Hellwig wrote:
> On Wed, Nov 21, 2018 at 11:23:23AM +0800, Ming Lei wrote:
> >  	if (bio->bi_vcnt > 0) {
> > -		struct bio_vec *bv = &bio->bi_io_vec[bio->bi_vcnt - 1];
> > +		struct bio_vec bv;
> > +		struct bio_vec *seg = &bio->bi_io_vec[bio->bi_vcnt - 1];
> >  
> > -		if (page == bv->bv_page && off == bv->bv_offset + bv->bv_len) {
> > -			bv->bv_len += len;
> > +		bvec_last_segment(seg, &bv);
> > +
> > +		if (page == bv.bv_page && off == bv.bv_offset + bv.bv_len) {
> 
> I think this we can simplify the try to merge into bio case a bit,
> and also document it better with something like this:
> 
> diff --git a/block/bio.c b/block/bio.c
> index 854676edc438..cc913281a723 100644
> --- a/block/bio.c
> +++ b/block/bio.c
> @@ -822,54 +822,40 @@ EXPORT_SYMBOL(bio_add_pc_page);
>   * @page: page to add
>   * @len: length of the data to add
>   * @off: offset of the data in @page
> + * @same_page: if %true only merge if the new data is in the same physical
> + *		page as the last segment of the bio.
>   *
> - * Try to add the data at @page + @off to the last page of @bio.  This is a
> + * Try to add the data at @page + @off to the last bvec of @bio.  This is a
>   * a useful optimisation for file systems with a block size smaller than the
>   * page size.
>   *
>   * Return %true on success or %false on failure.
>   */
>  bool __bio_try_merge_page(struct bio *bio, struct page *page,
> -		unsigned int len, unsigned int off)
> +		unsigned int len, unsigned int off, bool same_page)
>  {
>  	if (WARN_ON_ONCE(bio_flagged(bio, BIO_CLONED)))
>  		return false;
>  
>  	if (bio->bi_vcnt > 0) {
> -		struct bio_vec bv;
> -		struct bio_vec *seg = &bio->bi_io_vec[bio->bi_vcnt - 1];
> -
> -		bvec_last_segment(seg, &bv);
> -
> -		if (page == bv.bv_page && off == bv.bv_offset + bv.bv_len) {
> -			seg->bv_len += len;
> -			bio->bi_iter.bi_size += len;
> -			return true;
> -		}
> +		struct bio_vec *bv = &bio->bi_io_vec[bio->bi_vcnt - 1];
> +		phys_addr_t vec_addr = page_to_phys(bv->bv_page);
> +		phys_addr_t page_addr = page_to_phys(page);
> +
> +		if (vec_addr + bv->bv_offset + bv->bv_len != page_addr + off)
> +			return false;
> +		if (same_page &&
> +		    (vec_addr & PAGE_SIZE) != (page_addr & PAGE_SIZE))
> +			return false;

I guess the correct check should be:

		end_addr = vec_addr + bv->bv_offset + bv->bv_len;
		if (same_page &&
		    (end_addr & PAGE_MASK) != (page_addr & PAGE_MASK))
			return false;

And this approach is good, will take it in V12.

Thanks,
Ming