| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Date: Mon, 25 Jan 2021 10:41:10 +0900
From: Changheun Lee <nanich.lee@...sung.com>
To: ming.lei@...hat.com
Cc: Johannes.Thumshirn@....com, asml.silence@...il.com,
axboe@...nel.dk, damien.lemoal@....com, jisoo2146.oh@...sung.com,
junho89.kim@...sung.com, linux-block@...r.kernel.org,
linux-kernel@...r.kernel.org, mj0123.lee@...sung.com,
nanich.lee@...sung.com, osandov@...com, patchwork-bot@...nel.org,
seunghwan.hyun@...sung.com, sookwan7.kim@...sung.com,
tj@...nel.org, tom.leiming@...il.com, woosung2.lee@...sung.com,
yt0928.kim@...sung.com
Subject: Re: [PATCH v2] bio: limit bio max size
> On Thu, Jan 21, 2021 at 09:58:03AM +0900, Changheun Lee wrote:
> > bio size can grow up to 4GB when muli-page bvec is enabled.
> > but sometimes it would lead to inefficient behaviors.
> > in case of large chunk direct I/O, - 32MB chunk read in user space -
> > all pages for 32MB would be merged to a bio structure if memory address is
> > continued phsycally. it makes some delay to submit until merge complete.
> > bio max size should be limited as a proper size.
> >
> > When 32MB chunk read with direct I/O option is coming from userspace,
> > kernel behavior is below now. it's timeline.
>
> IMO, the issue should only exist on sync direct IO and writeback.
> Not sure if writeback cares this small delay because user data
> has been written to page cache already.
>
> Wrt. your big direct IO case, as I suggested, you may reduce the
> submission delay a lot by applying THP.
>
> Or you can just hardcode the limit in case of sync dio.
As you said, this small delay is not affect in case of writeback.
It's not common, but there are needs of direct I/O from userspace.
It will be operated optional, not default in v3 patch version.
And it'll be activated with queue flag, or sysfs node.
Please, review it again.
>
> >
> > | bio merge for 32MB. total 8,192 pages are merged.
> > | total elapsed time is over 2ms.
> > |------------------ ... ----------------------->|
> > | 8,192 pages merged a bio.
> > | at this time, first bio submit is done.
> > | 1 bio is split to 32 read request and issue.
> > |--------------->
> > |--------------->
> > |--------------->
> > ......
> > |--------------->
> > |--------------->|
> > total 19ms elapsed to complete 32MB read done from device. |
> >
> > If bio max size is limited with 1MB, behavior is changed below.
> >
> > | bio merge for 1MB. 256 pages are merged for each bio.
> > | total 32 bio will be made.
> > | total elapsed time is over 2ms. it's same.
> > | but, first bio submit timing is fast. about 100us.
> > |--->|--->|--->|---> ... -->|--->|--->|--->|--->|
> > | 256 pages merged a bio.
> > | at this time, first bio submit is done.
> > | and 1 read request is issued for 1 bio.
> > |--------------->
> > |--------------->
> > |--------------->
> > ......
> > |--------------->
> > |--------------->|
> > total 17ms elapsed to complete 32MB read done from device. |
> >
> > As a result, read request issue timing is faster if bio max size is limited.
> > Current kernel behavior with multipage bvec, super large bio can be created.
> > And it lead to delay first I/O request issue.
> >
> > Signed-off-by: Changheun Lee <nanich.lee@...sung.com>
> >
> > ---
> > block/bio.c | 17 ++++++++++++++++-
> > include/linux/bio.h | 13 +++----------
> > include/linux/blk_types.h | 1 +
> > 3 files changed, 20 insertions(+), 11 deletions(-)
> >
> > diff --git a/block/bio.c b/block/bio.c
> > index 1f2cc1fbe283..027503c2e2e7 100644
> > --- a/block/bio.c
> > +++ b/block/bio.c
> > @@ -284,9 +284,24 @@ void bio_init(struct bio *bio, struct bio_vec *table,
> >
> > bio->bi_io_vec = table;
> > bio->bi_max_vecs = max_vecs;
> > + bio->bi_max_size = UINT_MAX;
> > }
> > EXPORT_SYMBOL(bio_init);
> >
> > +void bio_set_dev(struct bio *bio, struct block_device *bdev)
> > +{
> > + if (bio->bi_disk != bdev->bd_disk)
> > + bio_clear_flag(bio, BIO_THROTTLED);
> > +
> > + bio->bi_disk = bdev->bd_disk;
> > + bio->bi_partno = bdev->bd_partno;
> > + bio->bi_max_size = blk_queue_get_max_sectors(bio->bi_disk->queue,
> > + bio_op(bio)) << SECTOR_SHIFT;
> > +
> > + bio_associate_blkg(bio);
> > +}
> > +EXPORT_SYMBOL(bio_set_dev);
> > +
> > /**
> > * bio_reset - reinitialize a bio
> > * @bio: bio to reset
> > @@ -877,7 +892,7 @@ bool __bio_try_merge_page(struct bio *bio, struct page *page,
> > struct bio_vec *bv = &bio->bi_io_vec[bio->bi_vcnt - 1];
> >
> > if (page_is_mergeable(bv, page, len, off, same_page)) {
> > - if (bio->bi_iter.bi_size > UINT_MAX - len) {
> > + if (bio->bi_iter.bi_size > bio->bi_max_size - len)
> > *same_page = false;
> > return false;
> > }
> > diff --git a/include/linux/bio.h b/include/linux/bio.h
> > index 1edda614f7ce..b9803e80c259 100644
> > --- a/include/linux/bio.h
> > +++ b/include/linux/bio.h
> > @@ -113,7 +113,7 @@ static inline bool bio_full(struct bio *bio, unsigned len)
> > if (bio->bi_vcnt >= bio->bi_max_vecs)
> > return true;
> >
> > - if (bio->bi_iter.bi_size > UINT_MAX - len)
> > + if (bio->bi_iter.bi_size > bio->bi_max_size - len)
> > return true;
> >
> > return false;
> > @@ -482,20 +482,13 @@ extern struct bio_vec *bvec_alloc(gfp_t, int, unsigned long *, mempool_t *);
> > extern void bvec_free(mempool_t *, struct bio_vec *, unsigned int);
> > extern unsigned int bvec_nr_vecs(unsigned short idx);
> > extern const char *bio_devname(struct bio *bio, char *buffer);
> > -
> > -#define bio_set_dev(bio, bdev) \
> > -do { \
> > - if ((bio)->bi_disk != (bdev)->bd_disk) \
> > - bio_clear_flag(bio, BIO_THROTTLED);\
> > - (bio)->bi_disk = (bdev)->bd_disk; \
> > - (bio)->bi_partno = (bdev)->bd_partno; \
> > - bio_associate_blkg(bio); \
> > -} while (0)
> > +extern void bio_set_dev(struct bio *bio, struct block_device *bdev);
> >
> > #define bio_copy_dev(dst, src) \
> > do { \
> > (dst)->bi_disk = (src)->bi_disk; \
> > (dst)->bi_partno = (src)->bi_partno; \
> > + (dst)->bi_max_size = (src)->bi_max_size;\
> > bio_clone_blkg_association(dst, src); \
> > } while (0)
> >
> > diff --git a/include/linux/blk_types.h b/include/linux/blk_types.h
> > index 866f74261b3b..e5dd5b7d8fc1 100644
> > --- a/include/linux/blk_types.h
> > +++ b/include/linux/blk_types.h
> > @@ -270,6 +270,7 @@ struct bio {
> > */
> >
> > unsigned short bi_max_vecs; /* max bvl_vecs we can hold */
> > + unsigned int bi_max_size; /* max data size we can hold */
>
> People don't like to extend bio which can be fit in two cachelines
> exactly, and adding one 'int' will make it cross 3 cache lines.
>
Yes, you right. good comment.
I'll not add a new variable in bio. :)
>
> Thanks,
> Ming
>
>
---
Changheun Lee
Samsung Electronics
Powered by blists - more mailing lists