Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
IronPort-SDR: SPOc5ZhZ/+rilHOFMgB7LQv7i5/UcYbwWQ7TN8072KkuAAdsHWfK8LaqAvz3MEjhnvwFZijXLS
 QzYI+g7A/O7qmFAKyLWlCNhQWZ0GTTmNU9uetTb/VJeqJbQmIWATAcFlCUr9ryWkj1+MPvEAkL
 OvGaG+iTrvfVBUawqzsU6eecrHHleh87K7+bl/jz/hPvGPS1+oG60XPrfIalGY17vFUz/Y4i58
 xyGoAn+8ZqQYGkxG7fsoHID/rm/GK3c980jm1hXP/wrh68SQis2p4WMw7uIbDTx/9BAlMa7k9L
 4yE=
From:   Damien Le Moal <Damien.LeMoal@wdc.com>
To:     Changheun Lee <nanich.lee@samsung.com>,
        "hch@infradead.org" <hch@infradead.org>,
        Johannes Thumshirn <Johannes.Thumshirn@wdc.com>,
        "alex_y_xu@yahoo.ca" <alex_y_xu@yahoo.ca>,
        "asml.silence@gmail.com" <asml.silence@gmail.com>,
        "axboe@kernel.dk" <axboe@kernel.dk>,
        "bgoncalv@redhat.com" <bgoncalv@redhat.com>,
        "bvanassche@acm.org" <bvanassche@acm.org>,
        "gregkh@linuxfoundation.org" <gregkh@linuxfoundation.org>,
        "jaegeuk@kernel.org" <jaegeuk@kernel.org>,
        "linux-block@vger.kernel.org" <linux-block@vger.kernel.org>,
        "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
        "ming.lei@redhat.com" <ming.lei@redhat.com>,
        "osandov@fb.com" <osandov@fb.com>,
        "patchwork-bot@kernel.org" <patchwork-bot@kernel.org>,
        "tj@kernel.org" <tj@kernel.org>,
        "tom.leiming@gmail.com" <tom.leiming@gmail.com>,
        "yi.zhang@redhat.com" <yi.zhang@redhat.com>
CC:     "jisoo2146.oh@samsung.com" <jisoo2146.oh@samsung.com>,
        "junho89.kim@samsung.com" <junho89.kim@samsung.com>,
        "mj0123.lee@samsung.com" <mj0123.lee@samsung.com>,
        "seunghwan.hyun@samsung.com" <seunghwan.hyun@samsung.com>,
        "sookwan7.kim@samsung.com" <sookwan7.kim@samsung.com>,
        "woosung2.lee@samsung.com" <woosung2.lee@samsung.com>,
        "yt0928.kim@samsung.com" <yt0928.kim@samsung.com>
Subject: Re: [PATCH v11 1/3] bio: control bio max size
Thread-Topic: [PATCH v11 1/3] bio: control bio max size
Thread-Index: AQHXV6rsD3TU7ripHkSbAptiY/rq5A==
Date:   Wed, 2 Jun 2021 12:51:17 +0000
Message-ID: <DM6PR04MB70819D1D9C6301E399FC1A67E73D9@DM6PR04MB7081.namprd04.prod.outlook.com>
References: <20210602121037.11083-1-nanich.lee@samsung.com>
 <CGME20210602122910epcas1p17ab868175c38cf3f143b64f4ec59f75c@epcas1p1.samsung.com>
 <20210602121037.11083-2-nanich.lee@samsung.com>
Accept-Language: en-US
Content-Language: en-US
wdcipoutbound: EOP-TRUE
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: quoted-printable
MIME-Version: 1.0
X-MS-Exchange-CrossTenant-Network-Message-Id: 4325f35d-5898-459c-cf00-08d925c51d42
X-MS-Exchange-CrossTenant-originalarrivaltime: 02 Jun 2021 12:51:17.8223
 (UTC)
X-MS-Exchange-CrossTenant-fromentityheader: Hosted
X-MS-Exchange-CrossTenant-id: b61c8803-16f3-4c35-9b17-6f65f441df86
X-MS-Exchange-CrossTenant-mailboxtype: HOSTED
X-MS-Exchange-CrossTenant-userprincipalname: o7wCIZgSVCh9qhw4rwZceJjKofS+OkJ+8Y7wYiWcCzHH6gMpsCmZ/w8l4hgtHfdnMji6a6CYo1TbQj1GCv+3og==
X-MS-Exchange-Transport-CrossTenantHeadersStamped: DM6PR04MB5882
Precedence: bulk
List-ID: <linux-kernel.vger.kernel.org>
X-Mailing-List: linux-kernel@vger.kernel.org

On 2021/06/02 21:29, Changheun Lee wrote:=0A=
> bio size can grow up to 4GB after muli-page bvec has been enabled.=0A=
> But sometimes large size of bio would lead to inefficient behaviors.=0A=
> Control of bio max size will be helpful to improve inefficiency.=0A=
> =0A=
> Below is a example for inefficient behaviours.=0A=
> In case of large chunk direct I/O, - 32MB chunk read in user space -=0A=
> all pages for 32MB would be merged to a bio structure if the pages=0A=
> physical addresses are contiguous. It makes some delay to submit=0A=
> until merge complete. bio max size should be limited to a proper size.=0A=
> =0A=
> When 32MB chunk read with direct I/O option is coming from userspace,=0A=
> kernel behavior is below now in do_direct_IO() loop. It's timeline.=0A=
> =0A=
>  | bio merge for 32MB. total 8,192 pages are merged.=0A=
>  | total elapsed time is over 2ms.=0A=
>  |------------------ ... ----------------------->|=0A=
>                                                  | 8,192 pages merged a b=
io.=0A=
>                                                  | at this time, first bi=
o submit is done.=0A=
>                                                  | 1 bio is split to 32 r=
ead request and issue.=0A=
>                                                  |--------------->=0A=
>                                                   |--------------->=0A=
>                                                    |--------------->=0A=
>                                                               ......=0A=
>                                                                    |-----=
---------->=0A=
>                                                                     |----=
----------->|=0A=
>                           total 19ms elapsed to complete 32MB read done f=
rom device. |=0A=
> =0A=
> If bio max size is limited with 1MB, behavior is changed below.=0A=
> =0A=
>  | bio merge for 1MB. 256 pages are merged for each bio.=0A=
>  | total 32 bio will be made.=0A=
>  | total elapsed time is over 2ms. it's same.=0A=
>  | but, first bio submit timing is fast. about 100us.=0A=
>  |--->|--->|--->|---> ... -->|--->|--->|--->|--->|=0A=
>       | 256 pages merged a bio.=0A=
>       | at this time, first bio submit is done.=0A=
>       | and 1 read request is issued for 1 bio.=0A=
>       |--------------->=0A=
>            |--------------->=0A=
>                 |--------------->=0A=
>                                       ......=0A=
>                                                  |--------------->=0A=
>                                                   |--------------->|=0A=
>         total 17ms elapsed to complete 32MB read done from device. |=0A=
> =0A=
> As a result, read request issue timing is faster if bio max size is limit=
ed.=0A=
> Current kernel behavior with multipage bvec, super large bio can be creat=
ed.=0A=
> And it lead to delay first I/O request issue.=0A=
> =0A=
> Signed-off-by: Changheun Lee <nanich.lee@samsung.com>=0A=
> ---=0A=
>  block/bio.c            | 17 ++++++++++++++---=0A=
>  block/blk-settings.c   | 19 +++++++++++++++++++=0A=
>  include/linux/bio.h    |  4 +++-=0A=
>  include/linux/blkdev.h |  3 +++=0A=
>  4 files changed, 39 insertions(+), 4 deletions(-)=0A=
> =0A=
> diff --git a/block/bio.c b/block/bio.c=0A=
> index 44205dfb6b60..c52639bb80cd 100644=0A=
> --- a/block/bio.c=0A=
> +++ b/block/bio.c=0A=
> @@ -255,6 +255,13 @@ void bio_init(struct bio *bio, struct bio_vec *table=
,=0A=
>  }=0A=
>  EXPORT_SYMBOL(bio_init);=0A=
>  =0A=
> +unsigned int bio_max_size(struct bio *bio)=0A=
=0A=
It would be nice to call this function the same as the limit name: bio_max_=
bytes().=0A=
=0A=
> +{=0A=
> +	struct block_device *bdev =3D bio->bi_bdev;=0A=
> +=0A=
> +	return bdev ? bdev->bd_disk->queue->limits.max_bio_bytes : UINT_MAX;=0A=
=0A=
My personal preference goes to a plain if() instead of this.=0A=
=0A=
> +}=0A=
> +=0A=
>  /**=0A=
>   * bio_reset - reinitialize a bio=0A=
>   * @bio:	bio to reset=0A=
> @@ -866,7 +873,7 @@ bool __bio_try_merge_page(struct bio *bio, struct pag=
e *page,=0A=
>  		struct bio_vec *bv =3D &bio->bi_io_vec[bio->bi_vcnt - 1];=0A=
>  =0A=
>  		if (page_is_mergeable(bv, page, len, off, same_page)) {=0A=
> -			if (bio->bi_iter.bi_size > UINT_MAX - len) {=0A=
> +			if (bio->bi_iter.bi_size > bio_max_size(bio) - len) {=0A=
>  				*same_page =3D false;=0A=
>  				return false;=0A=
>  			}=0A=
> @@ -995,6 +1002,7 @@ static int __bio_iov_iter_get_pages(struct bio *bio,=
 struct iov_iter *iter)=0A=
>  {=0A=
>  	unsigned short nr_pages =3D bio->bi_max_vecs - bio->bi_vcnt;=0A=
>  	unsigned short entries_left =3D bio->bi_max_vecs - bio->bi_vcnt;=0A=
> +	unsigned int bytes_left =3D bio_max_size(bio) - bio->bi_iter.bi_size;=
=0A=
>  	struct bio_vec *bv =3D bio->bi_io_vec + bio->bi_vcnt;=0A=
>  	struct page **pages =3D (struct page **)bv;=0A=
>  	bool same_page =3D false;=0A=
> @@ -1010,7 +1018,8 @@ static int __bio_iov_iter_get_pages(struct bio *bio=
, struct iov_iter *iter)=0A=
>  	BUILD_BUG_ON(PAGE_PTRS_PER_BVEC < 2);=0A=
>  	pages +=3D entries_left * (PAGE_PTRS_PER_BVEC - 1);=0A=
>  =0A=
> -	size =3D iov_iter_get_pages(iter, pages, LONG_MAX, nr_pages, &offset);=
=0A=
> +	size =3D iov_iter_get_pages(iter, pages, bytes_left, nr_pages,=0A=
> +				  &offset);=0A=
>  	if (unlikely(size <=3D 0))=0A=
>  		return size ? size : -EFAULT;=0A=
>  =0A=
> @@ -1038,6 +1047,7 @@ static int __bio_iov_append_get_pages(struct bio *b=
io, struct iov_iter *iter)=0A=
>  {=0A=
>  	unsigned short nr_pages =3D bio->bi_max_vecs - bio->bi_vcnt;=0A=
>  	unsigned short entries_left =3D bio->bi_max_vecs - bio->bi_vcnt;=0A=
> +	unsigned int bytes_left =3D bio_max_size(bio) - bio->bi_iter.bi_size;=
=0A=
>  	struct request_queue *q =3D bio->bi_bdev->bd_disk->queue;=0A=
>  	unsigned int max_append_sectors =3D queue_max_zone_append_sectors(q);=
=0A=
>  	struct bio_vec *bv =3D bio->bi_io_vec + bio->bi_vcnt;=0A=
> @@ -1058,7 +1068,8 @@ static int __bio_iov_append_get_pages(struct bio *b=
io, struct iov_iter *iter)=0A=
>  	BUILD_BUG_ON(PAGE_PTRS_PER_BVEC < 2);=0A=
>  	pages +=3D entries_left * (PAGE_PTRS_PER_BVEC - 1);=0A=
>  =0A=
> -	size =3D iov_iter_get_pages(iter, pages, LONG_MAX, nr_pages, &offset);=
=0A=
> +	size =3D iov_iter_get_pages(iter, pages, bytes_left, nr_pages,=0A=
> +				  &offset);=0A=
>  	if (unlikely(size <=3D 0))=0A=
>  		return size ? size : -EFAULT;=0A=
>  =0A=
> diff --git a/block/blk-settings.c b/block/blk-settings.c=0A=
> index 902c40d67120..e270e31519a1 100644=0A=
> --- a/block/blk-settings.c=0A=
> +++ b/block/blk-settings.c=0A=
> @@ -32,6 +32,7 @@ EXPORT_SYMBOL_GPL(blk_queue_rq_timeout);=0A=
>   */=0A=
>  void blk_set_default_limits(struct queue_limits *lim)=0A=
>  {=0A=
> +	lim->max_bio_bytes =3D UINT_MAX;=0A=
>  	lim->max_segments =3D BLK_MAX_SEGMENTS;=0A=
>  	lim->max_discard_segments =3D 1;=0A=
>  	lim->max_integrity_segments =3D 0;=0A=
=0A=
What about the limit setup for stacked devices ? Leaving it to UINT_MAX is =
OK ?=0A=
If for your use case you add dm-linear on top of the device and rerun your =
test,=0A=
don't you get again slow performance ?=0A=
=0A=
> @@ -100,6 +101,24 @@ void blk_queue_bounce_limit(struct request_queue *q,=
 enum blk_bounce bounce)=0A=
>  }=0A=
>  EXPORT_SYMBOL(blk_queue_bounce_limit);=0A=
>  =0A=
> +/**=0A=
> + * blk_queue_max_bio_bytes - set bio max size for queue=0A=
> + * @q: the request queue for the device=0A=
> + * @bytes : bio max bytes to be set=0A=
> + *=0A=
> + * Description:=0A=
> + *    Set proper bio max size to optimize queue operating.=0A=
> + **/=0A=
> +void blk_queue_max_bio_bytes(struct request_queue *q, unsigned int bytes=
)=0A=
> +{=0A=
> +	struct queue_limits *limits =3D &q->limits;=0A=
> +	unsigned int max_bio_bytes =3D round_up(bytes, PAGE_SIZE);=0A=
> +=0A=
> +	limits->max_bio_bytes =3D max_t(unsigned int, max_bio_bytes,=0A=
> +				      BIO_MAX_VECS * PAGE_SIZE);=0A=
> +}=0A=
> +EXPORT_SYMBOL(blk_queue_max_bio_bytes);=0A=
=0A=
Why does this need to be exported ?=0A=
=0A=
> +=0A=
>  /**=0A=
>   * blk_queue_max_hw_sectors - set max sectors for a request for this que=
ue=0A=
>   * @q:  the request queue for the device=0A=
> diff --git a/include/linux/bio.h b/include/linux/bio.h=0A=
> index a0b4cfdf62a4..f1a99f0a240c 100644=0A=
> --- a/include/linux/bio.h=0A=
> +++ b/include/linux/bio.h=0A=
> @@ -106,6 +106,8 @@ static inline void *bio_data(struct bio *bio)=0A=
>  	return NULL;=0A=
>  }=0A=
>  =0A=
> +extern unsigned int bio_max_size(struct bio *bio);=0A=
=0A=
No need for extern.=0A=
=0A=
> +=0A=
>  /**=0A=
>   * bio_full - check if the bio is full=0A=
>   * @bio:	bio to check=0A=
> @@ -119,7 +121,7 @@ static inline bool bio_full(struct bio *bio, unsigned=
 len)=0A=
>  	if (bio->bi_vcnt >=3D bio->bi_max_vecs)=0A=
>  		return true;=0A=
>  =0A=
> -	if (bio->bi_iter.bi_size > UINT_MAX - len)=0A=
> +	if (bio->bi_iter.bi_size > bio_max_size(bio) - len)=0A=
>  		return true;=0A=
>  =0A=
>  	return false;=0A=
> diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h=0A=
> index 1255823b2bc0..861888501fc0 100644=0A=
> --- a/include/linux/blkdev.h=0A=
> +++ b/include/linux/blkdev.h=0A=
> @@ -326,6 +326,8 @@ enum blk_bounce {=0A=
>  };=0A=
>  =0A=
>  struct queue_limits {=0A=
> +	unsigned int		max_bio_bytes;=0A=
> +=0A=
>  	enum blk_bounce		bounce;=0A=
>  	unsigned long		seg_boundary_mask;=0A=
>  	unsigned long		virt_boundary_mask;=0A=
> @@ -1132,6 +1134,7 @@ extern void blk_abort_request(struct request *);=0A=
>   * Access functions for manipulating queue properties=0A=
>   */=0A=
>  extern void blk_cleanup_queue(struct request_queue *);=0A=
> +extern void blk_queue_max_bio_bytes(struct request_queue *, unsigned int=
);=0A=
=0A=
No need for extern.=0A=
=0A=
>  void blk_queue_bounce_limit(struct request_queue *q, enum blk_bounce lim=
it);=0A=
>  extern void blk_queue_max_hw_sectors(struct request_queue *, unsigned in=
t);=0A=
>  extern void blk_queue_chunk_sectors(struct request_queue *, unsigned int=
);=0A=
> =0A=
=0A=
=0A=
-- =0A=
Damien Le Moal=0A=
Western Digital Research=0A=