Re: [PATCH 2/4] fs/namespace: add umounted mounts to umount_mnt_ns

[Al Viro]

On Thu, Oct 02, 2025 at 06:18:38PM +0530, Bhavik Sachdev wrote:

> @@ -1438,6 +1440,18 @@ static void mntput_no_expire(struct mount *mnt)
>  	mnt->mnt.mnt_flags |= MNT_DOOMED;
>  	rcu_read_unlock();
>  
> +	if (mnt_ns_attached(mnt)) {
> +		struct mnt_namespace *ns;
> +
> +		move_from_ns(mnt);
> +		ns = mnt->mnt_ns;
> +		if (ns) {
> +			ns->nr_mounts--;
> +			__touch_mnt_namespace(ns);
> +		}
> +		mnt->mnt_ns = NULL;
> +	}

Sorry, no.  You are introducing very special locking for one namespace's rbtree.
Not gonna fly.

NAKed-by: Al Viro <viro@xxxxxxxxxxxxxxxxxx>