Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Wed, 6 Nov 2019 12:22:15 -0500
From:   Brian Foster <bfoster@redhat.com>
To:     Dave Chinner <david@fromorbit.com>
Cc:     linux-xfs@vger.kernel.org, linux-fsdevel@vger.kernel.org,
        linux-mm@kvack.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH 26/28] xfs: use xfs_ail_push_all in xfs_reclaim_inodes
Message-ID: <20191106172215.GD37080@bfoster>
References: <20191031234618.15403-1-david@fromorbit.com>
 <20191031234618.15403-27-david@fromorbit.com>
MIME-Version: 1.0
In-Reply-To: <20191031234618.15403-27-david@fromorbit.com>
User-Agent: Mutt/1.12.1 (2019-06-15)
Content-Type: text/plain; charset=WINDOWS-1252
Content-Transfer-Encoding: quoted-printable
Content-Disposition: inline
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Fri, Nov 01, 2019 at 10:46:16AM +1100, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
>=20
> If we are reclaiming all inodes, it is likely we need to flush the
> entire AIL to do that. We have mechanisms to do that without needing
> to push to a specific LSN.
>=20
> Convert xfs_relaim_all_inodes() to use xfs_ail_push_all variant so
> we can get rid of the hacky xfs_ail_push_sync() scaffolding we used
> to support the intermediate stages of the non-blocking reclaim
> changeset.
>=20
> Signed-off-by: Dave Chinner <dchinner@redhat.com>
> ---
>  fs/xfs/xfs_icache.c     | 17 +++++++++++------
>  fs/xfs/xfs_trans_ail.c  | 32 --------------------------------
>  fs/xfs/xfs_trans_priv.h |  2 --
>  3 files changed, 11 insertions(+), 40 deletions(-)
>=20
> diff --git a/fs/xfs/xfs_icache.c b/fs/xfs/xfs_icache.c
> index 71a729e29260..11bf4768d491 100644
> --- a/fs/xfs/xfs_icache.c
> +++ b/fs/xfs/xfs_icache.c
...
> @@ -1066,13 +1074,10 @@ xfs_reclaim_all_inodes(
>  =09=09=09=09      xfs_inode_reclaim_isolate, &ra, to_free);
>  =09=09xfs_dispose_inodes(&ra.freeable);
> =20
> -=09=09if (freed =3D=3D 0) {
> +=09=09if (freed =3D=3D 0)
>  =09=09=09xfs_log_force(mp, XFS_LOG_SYNC);
> -=09=09=09xfs_ail_push_all(mp->m_ail);
> -=09=09} else if (ra.lowest_lsn !=3D NULLCOMMITLSN) {
> -=09=09=09xfs_ail_push_sync(mp->m_ail, ra.lowest_lsn);
> -=09=09}
> -=09=09cond_resched();
> +=09=09else if (ra.dirty_skipped)
> +=09=09=09congestion_wait(BLK_RW_ASYNC, HZ/10);

Why not use xfs_ail_push_all_sync() in this function and skip the direct
stall? This is only used in the unmount and quiesce paths so the big
hammer approach seems reasonable. As it is, the former already calls
xfs_ail_push_all_sync() before xfs_reclaim_all_inodes() and the latter
calls xfs_log_force(mp, XFS_LOG_SYNC).

Brian

>  =09}
>  }
> =20
> diff --git a/fs/xfs/xfs_trans_ail.c b/fs/xfs/xfs_trans_ail.c
> index 3e1d0e1439e2..685a21cd24c0 100644
> --- a/fs/xfs/xfs_trans_ail.c
> +++ b/fs/xfs/xfs_trans_ail.c
> @@ -662,36 +662,6 @@ xfs_ail_push_all(
>  =09=09xfs_ail_push(ailp, threshold_lsn);
>  }
> =20
> -/*
> - * Push the AIL to a specific lsn and wait for it to complete.
> - */
> -void
> -xfs_ail_push_sync(
> -=09struct xfs_ail=09=09*ailp,
> -=09xfs_lsn_t=09=09threshold_lsn)
> -{
> -=09struct xfs_log_item=09*lip;
> -=09DEFINE_WAIT(wait);
> -
> -=09spin_lock(&ailp->ail_lock);
> -=09while ((lip =3D xfs_ail_min(ailp)) !=3D NULL) {
> -=09=09prepare_to_wait(&ailp->ail_push, &wait, TASK_UNINTERRUPTIBLE);
> -=09=09if (XFS_FORCED_SHUTDOWN(ailp->ail_mount) ||
> -=09=09    XFS_LSN_CMP(threshold_lsn, lip->li_lsn) < 0)
> -=09=09=09break;
> -=09=09if (XFS_LSN_CMP(threshold_lsn, ailp->ail_target) > 0)
> -=09=09=09ailp->ail_target =3D threshold_lsn;
> -=09=09wake_up_process(ailp->ail_task);
> -=09=09spin_unlock(&ailp->ail_lock);
> -=09=09schedule();
> -=09=09spin_lock(&ailp->ail_lock);
> -=09}
> -=09spin_unlock(&ailp->ail_lock);
> -
> -=09finish_wait(&ailp->ail_push, &wait);
> -}
> -
> -
>  /*
>   * Push out all items in the AIL immediately and wait until the AIL is e=
mpty.
>   */
> @@ -732,7 +702,6 @@ xfs_ail_update_finish(
>  =09if (!XFS_FORCED_SHUTDOWN(mp))
>  =09=09xlog_assign_tail_lsn_locked(mp);
> =20
> -=09wake_up_all(&ailp->ail_push);
>  =09if (list_empty(&ailp->ail_head))
>  =09=09wake_up_all(&ailp->ail_empty);
>  =09spin_unlock(&ailp->ail_lock);
> @@ -889,7 +858,6 @@ xfs_trans_ail_init(
>  =09spin_lock_init(&ailp->ail_lock);
>  =09INIT_LIST_HEAD(&ailp->ail_buf_list);
>  =09init_waitqueue_head(&ailp->ail_empty);
> -=09init_waitqueue_head(&ailp->ail_push);
> =20
>  =09ailp->ail_task =3D kthread_run(xfsaild, ailp, "xfsaild/%s",
>  =09=09=09ailp->ail_mount->m_fsname);
> diff --git a/fs/xfs/xfs_trans_priv.h b/fs/xfs/xfs_trans_priv.h
> index 1b6f4bbd47c0..35655eac01a6 100644
> --- a/fs/xfs/xfs_trans_priv.h
> +++ b/fs/xfs/xfs_trans_priv.h
> @@ -61,7 +61,6 @@ struct xfs_ail {
>  =09int=09=09=09ail_log_flush;
>  =09struct list_head=09ail_buf_list;
>  =09wait_queue_head_t=09ail_empty;
> -=09wait_queue_head_t=09ail_push;
>  };
> =20
>  /*
> @@ -114,7 +113,6 @@ xfs_trans_ail_remove(
>  }
> =20
>  void=09=09=09xfs_ail_push(struct xfs_ail *, xfs_lsn_t);
> -void=09=09=09xfs_ail_push_sync(struct xfs_ail *, xfs_lsn_t);
>  void=09=09=09xfs_ail_push_all(struct xfs_ail *);
>  void=09=09=09xfs_ail_push_all_sync(struct xfs_ail *);
>  struct xfs_log_item=09*xfs_ail_min(struct xfs_ail  *ailp);
> --=20
> 2.24.0.rc0
>=20