Re: [PATCH -next 4/8] md/raid1: switch to use md_account_bio() for io accounting

[Yu Kuai]

Hi,

在 2023/06/20 17:07, Xiao Ni 写道:
> On Mon, Jun 19, 2023 at 8:49 PM Yu Kuai <yukuai1@xxxxxxxxxxxxxxx> wrote:
> > From: Yu Kuai <yukuai3@xxxxxxxxxx>
> >
> > Two problems can be fixed this way:
> >
> > 1) 'active_io' will represent inflight io instead of io that is
> > dispatching.
> >
> > 2) If io accounting is enabled or disabled while io is still inflight,
> > bio_start_io_acct() and bio_end_io_acct() is not balanced and io
> > inflight counter will be leaked.
> >
> > Signed-off-by: Yu Kuai <yukuai3@xxxxxxxxxx>
> > ---
> >   drivers/md/raid1.c | 14 ++++++--------
> >   drivers/md/raid1.h |  1 -
> >   2 files changed, 6 insertions(+), 9 deletions(-)
> >
> > diff --git a/drivers/md/raid1.c b/drivers/md/raid1.c
> > index dd25832eb045..06fa1580501f 100644
> > --- a/drivers/md/raid1.c
> > +++ b/drivers/md/raid1.c
> > @@ -304,8 +304,6 @@ static void call_bio_endio(struct r1bio *r1_bio)
> >          if (!test_bit(R1BIO_Uptodate, &r1_bio->state))
> >                  bio->bi_status = BLK_STS_IOERR;
> >
> > -       if (blk_queue_io_stat(bio->bi_bdev->bd_disk->queue))
> > -               bio_end_io_acct(bio, r1_bio->start_time);
> >          bio_endio(bio);
> >   }
> >
> > @@ -1303,10 +1301,10 @@ static void raid1_read_request(struct mddev *mddev, struct bio *bio,
> >          }
> >
> >          r1_bio->read_disk = rdisk;
> > -
> > -       if (!r1bio_existed && blk_queue_io_stat(bio->bi_bdev->bd_disk->queue))
> > -               r1_bio->start_time = bio_start_io_acct(bio);
> > -
> > +       if (!r1bio_existed) {
> > +               md_account_bio(mddev, &bio);
> > +               r1_bio->master_bio = bio;
> > +       }
> >          read_bio = bio_alloc_clone(mirror->rdev->bdev, bio, gfp,
> >                                     &mddev->bio_set);
> >
> > @@ -1500,8 +1498,8 @@ static void raid1_write_request(struct mddev *mddev, struct bio *bio,
> >                  r1_bio->sectors = max_sectors;
> >          }
> >
> > -       if (blk_queue_io_stat(bio->bi_bdev->bd_disk->queue))
> > -               r1_bio->start_time = bio_start_io_acct(bio);
> > +       md_account_bio(mddev, &bio);
> > +       r1_bio->master_bio = bio;
> >          atomic_set(&r1_bio->remaining, 1);
> >          atomic_set(&r1_bio->behind_remaining, 0);
> >
> > diff --git a/drivers/md/raid1.h b/drivers/md/raid1.h
> > index 468f189da7a0..14d4211a123a 100644
> > --- a/drivers/md/raid1.h
> > +++ b/drivers/md/raid1.h
> > @@ -157,7 +157,6 @@ struct r1bio {
> >          sector_t                sector;
> >          int                     sectors;
> >          unsigned long           state;
> > -       unsigned long           start_time;
> >          struct mddev            *mddev;
> >          /*
> >           * original bio going to /dev/mdx
> > --
> > 2.39.2
>
> Hi Kuai
>
> After this patch, raid1 will have one more memory allocation in the
> I/O path. Not sure if it can affect performance. Beside this, the
> patch is good for me.

Yes, I'm aware of this additional memory allocation, however, raid1(and
similar to other levels) already need to allocate r1bio and some bios(1
for read, and copies for write), so this is not a none -> new case,
it's just a allocate 2 -> allocate 3 case.

I think performance under memory pressure are both bad with or without
this patch, and one one bio clone latency without memory reclaim should
be fine.

Thanks,
Kuai
> Reviewed-by: Xiao Ni <xni@xxxxxxxxxx>
>
> .