Re: [PATCH 07/10] md: stop using for_each_mddev in md_notify_reboot

[Hannes Reinecke <hare@suse.de>]

On 7/18/22 08:34, Christoph Hellwig wrote:
> Just do a simple list_for_each_entry_safe on all_mddevs, and only grab a
> reference when we drop the lock.
> 
> Reviewed-by: Christoph Hellwig <hch@lst.de>
> ---
>   drivers/md/md.c | 12 +++++++++---
>   1 file changed, 9 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/md/md.c b/drivers/md/md.c
> index 687f320c7ec4a..44e4071b43148 100644
> --- a/drivers/md/md.c
> +++ b/drivers/md/md.c
> @@ -9587,11 +9587,13 @@ EXPORT_SYMBOL_GPL(rdev_clear_badblocks);
>   static int md_notify_reboot(struct notifier_block *this,
>   			    unsigned long code, void *x)
>   {
> -	struct list_head *tmp;
> -	struct mddev *mddev;
> +	struct mddev *mddev, *n;
>   	int need_delay = 0;
>   
> -	for_each_mddev(mddev, tmp) {
> +	spin_lock(&all_mddevs_lock);
> +	list_for_each_entry_safe(mddev, n, &all_mddevs, all_mddevs) {
> +		mddev_get(mddev);
> +		spin_unlock(&all_mddevs_lock);
>   		if (mddev_trylock(mddev)) {
>   			if (mddev->pers)
>   				__md_stop_writes(mddev);
> @@ -9600,7 +9602,11 @@ static int md_notify_reboot(struct notifier_block *this,
>   			mddev_unlock(mddev);
>   		}
>   		need_delay = 1;
> +		mddev_put(mddev);
> +		spin_lock(&all_mddevs_lock);
>   	}
> +	spin_unlock(&all_mddevs_lock);
> +
>   	/*
>   	 * certain more exotic SCSI devices are known to be
>   	 * volatile wrt too early system reboots. While the

I wish 'mddev_get()' could fail if 'mddev->active' is zero when we call 
it ... but maybe for another time.

Reviewed-by: Hannes Reinecke <hare@suse.de>

Cheers,

Hannes