Re: Need advice with iput() deadlock during writeback

[Al Viro <viro@zeniv.linux.org.uk>]

On Wed, Sep 17, 2025 at 10:19:27PM +0200, Max Kellermann wrote:
> On Wed, Sep 17, 2025 at 10:14 PM Al Viro <viro@zeniv.linux.org.uk> wrote:
> > Looks rather dangerous - what do you do on fs shutdown?
> 
> Sorry, I'm new to this, I don't know how fs shutdown works - stupid
> question: is my code any more dangerous than what's already happening
> with ceph_queue_inode_work()?

umount /wherever/the/fuck/it/is/mounted

calls umount(2), which removes the mount from the tree, then calls
deactivate_super(), dropping the active reference to superblock.
If it hadn't been mounted elsewhere, that's the last reference and
we this:
		shrinker_free(s->s_shrink);
		fs->kill_sb(s);

		kill_super_notify(s);

		/*
		 * Since list_lru_destroy() may sleep, we cannot call it from
		 * put_super(), where we hold the sb_lock. Therefore we destroy
		 * the lru lists right now.
		 */
		list_lru_destroy(&s->s_dentry_lru);
		list_lru_destroy(&s->s_inode_lru);

		put_filesystem(fs);
		put_super(s);
At some point ->kill_sb() will call generic_shutdown_super() (in case of ceph
that's done via kill_anon_super()), where we get to evict_inodes().  Any
busy inode at that point is a bad problem...