From: Yu Kuai <yukuai3@huawei.com>

Some levels doesn't implement "pers->quiesce", for example
raid0_quiesce() is empty, and now that all levels will drop 'active_io'
until io is done, wait for 'active_io' to be 0 is enough to make sure all
normal io is done, and percpu_ref_kill() for 'active_io' will make sure
no new normal io can be dispatched. There is no need to call
"pers->quiesce" anymore from mddev_suspend().

Signed-off-by: Yu Kuai <yukuai3@huawei.com>
---
 drivers/md/md.c | 2 --
 1 file changed, 2 deletions(-)

diff --git a/drivers/md/md.c b/drivers/md/md.c
index abb616720393..962dacfd98cf 100644
--- a/drivers/md/md.c
+++ b/drivers/md/md.c
@@ -453,7 +453,6 @@ void mddev_suspend(struct mddev *mddev)
 		mddev->pers->prepare_suspend(mddev);
 
 	wait_event(mddev->sb_wait, percpu_ref_is_zero(&mddev->active_io));
-	mddev->pers->quiesce(mddev, 1);
 	clear_bit_unlock(MD_ALLOW_SB_UPDATE, &mddev->flags);
 	wait_event(mddev->sb_wait, !test_bit(MD_UPDATING_SB, &mddev->flags));
 
@@ -472,7 +471,6 @@ void mddev_resume(struct mddev *mddev)
 		return;
 	percpu_ref_resurrect(&mddev->active_io);
 	wake_up(&mddev->sb_wait);
-	mddev->pers->quiesce(mddev, 0);
 
 	set_bit(MD_RECOVERY_NEEDED, &mddev->recovery);
 	md_wakeup_thread(mddev->thread);
-- 
2.39.2

Hi,

在 2023/06/28 9:29, Yu Kuai 写道:
> From: Yu Kuai <yukuai3@huawei.com>
> 
> Some levels doesn't implement "pers->quiesce", for example
> raid0_quiesce() is empty, and now that all levels will drop 'active_io'
> until io is done, wait for 'active_io' to be 0 is enough to make sure all
> normal io is done, and percpu_ref_kill() for 'active_io' will make sure
> no new normal io can be dispatched. There is no need to call
> "pers->quiesce" anymore from mddev_suspend().

Just found that this patch actually prevent null_ptr_deref in some cases
that "pers->quiesce" is NULl for some personalities, and caller of
mddev_suspend() doesn't check this. For example, suspend_lo_store()
checks "pers->quiesce" is not NULL before calling mddev_suspend(), while
suspend_hi_store() doesn't.

Thanks,
Kuai
> 
> Signed-off-by: Yu Kuai <yukuai3@huawei.com>
> ---
>   drivers/md/md.c | 2 --
>   1 file changed, 2 deletions(-)
> 
> diff --git a/drivers/md/md.c b/drivers/md/md.c
> index abb616720393..962dacfd98cf 100644
> --- a/drivers/md/md.c
> +++ b/drivers/md/md.c
> @@ -453,7 +453,6 @@ void mddev_suspend(struct mddev *mddev)
>   		mddev->pers->prepare_suspend(mddev);
>   
>   	wait_event(mddev->sb_wait, percpu_ref_is_zero(&mddev->active_io));
> -	mddev->pers->quiesce(mddev, 1);
>   	clear_bit_unlock(MD_ALLOW_SB_UPDATE, &mddev->flags);
>   	wait_event(mddev->sb_wait, !test_bit(MD_UPDATING_SB, &mddev->flags));
>   
> @@ -472,7 +471,6 @@ void mddev_resume(struct mddev *mddev)
>   		return;
>   	percpu_ref_resurrect(&mddev->active_io);
>   	wake_up(&mddev->sb_wait);
> -	mddev->pers->quiesce(mddev, 0);
>   
>   	set_bit(MD_RECOVERY_NEEDED, &mddev->recovery);
>   	md_wakeup_thread(mddev->thread);
> 

On Wed, Jun 28, 2023 at 2:29 PM Yu Kuai <yukuai1@huaweicloud.com> wrote:
>
> Hi,
>
> 在 2023/06/28 9:29, Yu Kuai 写道:
> > From: Yu Kuai <yukuai3@huawei.com>
> >
> > Some levels doesn't implement "pers->quiesce", for example
> > raid0_quiesce() is empty, and now that all levels will drop 'active_io'
> > until io is done, wait for 'active_io' to be 0 is enough to make sure all
> > normal io is done, and percpu_ref_kill() for 'active_io' will make sure
> > no new normal io can be dispatched. There is no need to call
> > "pers->quiesce" anymore from mddev_suspend().
>
> Just found that this patch actually prevent null_ptr_deref in some cases
> that "pers->quiesce" is NULl for some personalities, and caller of
> mddev_suspend() doesn't check this. For example, suspend_lo_store()
> checks "pers->quiesce" is not NULL before calling mddev_suspend(), while
> suspend_hi_store() doesn't.

Let's refactor the code so that suspend_lo_store() and suspend_hi_store()
share code for this logic.

Thanks,
Song