From: Ming Lei <ming.lei@redhat.com>
To: Christoph Hellwig <hch@lst.de>
Cc: Jens Axboe <axboe@kernel.dk>,
linux-block@vger.kernel.org, linux-kernel@vger.kernel.org,
linux-mm@kvack.org, Theodore Ts'o <tytso@mit.edu>,
Omar Sandoval <osandov@fb.com>, Sagi Grimberg <sagi@grimberg.me>,
Dave Chinner <dchinner@redhat.com>,
Kent Overstreet <kent.overstreet@gmail.com>,
Mike Snitzer <snitzer@redhat.com>,
dm-devel@redhat.com, Alexander Viro <viro@zeniv.linux.org.uk>,
linux-fsdevel@vger.kernel.org, Shaohua Li <shli@kernel.org>,
linux-raid@vger.kernel.org, David Sterba <dsterba@suse.com>,
linux-btrfs@vger.kernel.org,
"Darrick J . Wong" <darrick.wong@oracle.com>,
linux-xfs@vger.kernel.org, Gao Xiang <gaoxiang25@huawei.com>,
linux-ext4@vger.kernel.org, Coly Li <colyli@suse.de>,
linux-bcache@vger.kernel.org, Boaz Harrosh <ooo@electrozaur.com>,
Bob Peterson <rpeterso@redhat.com>,
cluster-devel@redhat.com
Subject: Re: [PATCH V11 14/19] block: handle non-cluster bio out of blk_bio_segment_split
Date: Wed, 21 Nov 2018 23:37:27 +0800 [thread overview]
Message-ID: <20181121153726.GC19111@ming.t460p> (raw)
In-Reply-To: <20181121143355.GB2594@lst.de>
On Wed, Nov 21, 2018 at 03:33:55PM +0100, Christoph Hellwig wrote:
> > + non-cluster.o
>
> Do we really need a new source file for these few functions?
>
> > default:
> > + if (!blk_queue_cluster(q)) {
> > + blk_queue_non_cluster_bio(q, bio);
> > + return;
>
> I'd name this blk_bio_segment_split_singlepage or similar.
OK.
>
> > +static __init int init_non_cluster_bioset(void)
> > +{
> > + WARN_ON(bioset_init(&non_cluster_bio_set, BIO_POOL_SIZE, 0,
> > + BIOSET_NEED_BVECS));
> > + WARN_ON(bioset_integrity_create(&non_cluster_bio_set, BIO_POOL_SIZE));
> > + WARN_ON(bioset_init(&non_cluster_bio_split, BIO_POOL_SIZE, 0, 0));
>
> Please only allocate the resources once a queue without the cluster
> flag is registered, there are only very few modern drivers that do that.
OK.
>
> > +static void non_cluster_end_io(struct bio *bio)
> > +{
> > + struct bio *bio_orig = bio->bi_private;
> > +
> > + bio_orig->bi_status = bio->bi_status;
> > + bio_endio(bio_orig);
> > + bio_put(bio);
> > +}
>
> Why can't we use bio_chain for the split bios?
The parent bio is multi-page bvec, we can't submit it for non-cluster.
>
> > + bio_for_each_segment(from, *bio_orig, iter) {
> > + if (i++ < max_segs)
> > + sectors += from.bv_len >> 9;
> > + else
> > + break;
> > + }
>
> The easy to read way would be:
>
> bio_for_each_segment(from, *bio_orig, iter) {
> if (i++ == max_segs)
> break;
> sectors += from.bv_len >> 9;
> }
OK.
>
> > + if (sectors < bio_sectors(*bio_orig)) {
> > + bio = bio_split(*bio_orig, sectors, GFP_NOIO,
> > + &non_cluster_bio_split);
> > + bio_chain(bio, *bio_orig);
> > + generic_make_request(*bio_orig);
> > + *bio_orig = bio;
>
> I don't think this is very efficient, as this means we now
> clone the bio twice, first to split it at the sector boundary,
> and then again when converting it to single-page bio_vec.
That is exactly what bounce code does. The problem for both bounce
and non-cluster is same actually because the bvec table itself has
to be changed.
>
> I think this could be something like this (totally untested):
>
> diff --git a/block/non-cluster.c b/block/non-cluster.c
> index 9c2910be9404..60389f275c43 100644
> --- a/block/non-cluster.c
> +++ b/block/non-cluster.c
> @@ -13,58 +13,59 @@
>
> #include "blk.h"
>
> -static struct bio_set non_cluster_bio_set, non_cluster_bio_split;
> +static struct bio_set non_cluster_bio_set;
>
> static __init int init_non_cluster_bioset(void)
> {
> WARN_ON(bioset_init(&non_cluster_bio_set, BIO_POOL_SIZE, 0,
> BIOSET_NEED_BVECS));
> WARN_ON(bioset_integrity_create(&non_cluster_bio_set, BIO_POOL_SIZE));
> - WARN_ON(bioset_init(&non_cluster_bio_split, BIO_POOL_SIZE, 0, 0));
>
> return 0;
> }
> __initcall(init_non_cluster_bioset);
>
> -static void non_cluster_end_io(struct bio *bio)
> -{
> - struct bio *bio_orig = bio->bi_private;
> -
> - bio_orig->bi_status = bio->bi_status;
> - bio_endio(bio_orig);
> - bio_put(bio);
> -}
> -
> void blk_queue_non_cluster_bio(struct request_queue *q, struct bio **bio_orig)
> {
> - struct bio *bio;
> struct bvec_iter iter;
> - struct bio_vec from;
> - unsigned i = 0;
> - unsigned sectors = 0;
> - unsigned short max_segs = min_t(unsigned short, BIO_MAX_PAGES,
> - queue_max_segments(q));
> + struct bio *bio;
> + struct bio_vec bv;
> + unsigned short max_segs, segs = 0;
> +
> + bio = bio_alloc_bioset(GFP_NOIO, bio_segments(*bio_orig),
> + &non_cluster_bio_set);
bio_segments(*bio_orig) may be > 256, so bio_alloc_bioset() may fail.
Thanks,
Ming
next prev parent reply other threads:[~2018-11-21 15:37 UTC|newest]
Thread overview: 58+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-11-21 3:23 [PATCH V11 00/19] block: support multi-page bvec Ming Lei
2018-11-21 3:23 ` [PATCH V11 01/19] block: don't use bio->bi_vcnt to figure out segment number Ming Lei
2018-11-21 3:23 ` [PATCH V11 02/19] block: introduce multi-page bvec helpers Ming Lei
2018-11-21 13:19 ` Christoph Hellwig
2018-11-21 15:06 ` Ming Lei
2018-11-21 16:08 ` Christoph Hellwig
2018-11-22 1:09 ` Ming Lei
2018-11-21 3:23 ` [PATCH V11 03/19] block: introduce bio_for_each_bvec() Ming Lei
2018-11-21 13:32 ` Christoph Hellwig
2018-11-21 15:31 ` Ming Lei
2018-11-21 16:10 ` Christoph Hellwig
2018-11-21 17:12 ` Christoph Hellwig
2018-11-22 10:15 ` Ming Lei
2018-11-22 10:23 ` Christoph Hellwig
2018-11-22 10:30 ` Christoph Hellwig
2018-11-22 14:57 ` Ming Lei
2018-11-22 1:17 ` Ming Lei
2018-11-21 3:23 ` [PATCH V11 04/19] block: use bio_for_each_bvec() to compute multi-page bvec count Ming Lei
2018-11-21 3:23 ` [PATCH V11 05/19] block: use bio_for_each_bvec() to map sg Ming Lei
2018-11-21 3:23 ` [PATCH V11 06/19] block: introduce bvec_last_segment() Ming Lei
2018-11-21 3:23 ` [PATCH V11 07/19] fs/buffer.c: use bvec iterator to truncate the bio Ming Lei
2018-11-22 10:58 ` Christoph Hellwig
2018-11-23 1:48 ` Ming Lei
2018-11-21 3:23 ` [PATCH V11 08/19] btrfs: use bvec_last_segment to get bio's last page Ming Lei
2018-11-21 3:23 ` [PATCH V11 09/19] btrfs: move bio_pages_all() to btrfs Ming Lei
2018-11-21 3:23 ` [PATCH V11 10/19] block: loop: pass multi-page bvec to iov_iter Ming Lei
2018-11-21 14:00 ` Christoph Hellwig
2018-11-21 3:23 ` [PATCH V11 11/19] bcache: avoid to use bio_for_each_segment_all() in bch_bio_alloc_pages() Ming Lei
2018-11-21 14:01 ` Christoph Hellwig
2018-11-21 3:23 ` [PATCH V11 12/19] block: allow bio_for_each_segment_all() to iterate over multi-page bvec Ming Lei
2018-11-21 14:02 ` Christoph Hellwig
2018-11-22 11:03 ` Christoph Hellwig
2018-11-23 2:19 ` Ming Lei
2018-11-21 3:23 ` [PATCH V11 13/19] block: move bounce_clone_bio into bio.c Ming Lei
2018-11-21 3:23 ` [PATCH V11 14/19] block: handle non-cluster bio out of blk_bio_segment_split Ming Lei
2018-11-21 14:33 ` Christoph Hellwig
2018-11-21 15:37 ` Ming Lei [this message]
2018-11-21 16:11 ` Christoph Hellwig
2018-11-21 17:46 ` Christoph Hellwig
2018-11-22 9:33 ` Ming Lei
2018-11-22 10:04 ` Christoph Hellwig
2018-11-22 10:26 ` Ming Lei
2018-11-22 10:40 ` Christoph Hellwig
2018-11-22 10:32 ` Ming Lei
2018-11-22 10:41 ` Christoph Hellwig
2018-11-22 10:46 ` Ming Lei
2018-11-22 10:47 ` Christoph Hellwig
2018-11-21 3:23 ` [PATCH V11 15/19] block: enable multipage bvecs Ming Lei
2018-11-21 14:55 ` Christoph Hellwig
2018-11-21 15:48 ` Ming Lei
2018-11-21 16:12 ` Christoph Hellwig
2018-11-23 10:50 ` Ming Lei
2018-11-21 3:23 ` [PATCH V11 16/19] block: always define BIO_MAX_PAGES as 256 Ming Lei
2018-11-21 3:23 ` [PATCH V11 17/19] block: document usage of bio iterator helpers Ming Lei
2018-11-21 7:45 ` Nikolay Borisov
2018-11-21 14:34 ` Christoph Hellwig
2018-11-21 3:23 ` [PATCH V11 18/19] block: kill QUEUE_FLAG_NO_SG_MERGE Ming Lei
2018-11-21 3:23 ` [PATCH V11 19/19] block: kill BLK_MQ_F_SG_MERGE Ming Lei
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20181121153726.GC19111@ming.t460p \
--to=ming.lei@redhat.com \
--cc=axboe@kernel.dk \
--cc=cluster-devel@redhat.com \
--cc=colyli@suse.de \
--cc=darrick.wong@oracle.com \
--cc=dchinner@redhat.com \
--cc=dm-devel@redhat.com \
--cc=dsterba@suse.com \
--cc=gaoxiang25@huawei.com \
--cc=hch@lst.de \
--cc=kent.overstreet@gmail.com \
--cc=linux-bcache@vger.kernel.org \
--cc=linux-block@vger.kernel.org \
--cc=linux-btrfs@vger.kernel.org \
--cc=linux-ext4@vger.kernel.org \
--cc=linux-fsdevel@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=linux-raid@vger.kernel.org \
--cc=linux-xfs@vger.kernel.org \
--cc=ooo@electrozaur.com \
--cc=osandov@fb.com \
--cc=rpeterso@redhat.com \
--cc=sagi@grimberg.me \
--cc=shli@kernel.org \
--cc=snitzer@redhat.com \
--cc=tytso@mit.edu \
--cc=viro@zeniv.linux.org.uk \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).