Re: [syzbot] [mm?] WARNING: bad unlock balance in copy_process

[Sebastian Andrzej Siewior]

On 2025-09-18 10:35:24 [+0200], Vlastimil Babka wrote:
> > IMPORTANT: if you fix the issue, please add the following tag to the commit:
> > Reported-by: syzbot+80cb3cc5c14fad191a10@xxxxxxxxxxxxxxxxxxxxxxxxx
> > 
> > =====================================
> > WARNING: bad unlock balance detected!
> > syzkaller #0 Not tainted
> > -------------------------------------
> > syz.1.48/6865 is trying to release lock (&sighand->siglock) at:
> > [<ffff8000803b8634>] spin_unlock include/linux/spinlock.h:391 [inline]
> > [<ffff8000803b8634>] copy_process+0x22d4/0x31ec kernel/fork.c:2432
> 
> bad_fork_core_free:
>         sched_core_free(p);
>         spin_unlock(&current->sighand->siglock); <- here
> 
> Sebastian, I think it's your 7c4f75a21f63 ("futex: Allow automatic
> allocation of process wide futex hash") adding a "goto bad_fork_core_free;"
> from a place that doesn't yet have current->sighand->siglock locked?

Yes. Judging from -rc6, if futex_hash_allocate_default() fails we hold
neither siglock nor tasklist_lock. sched_core_free() looks also bad as
the cookie was allocated later in sched_core_fork(). sched_cgroup_fork()
does nothing special. So it should be

diff --git a/kernel/fork.c b/kernel/fork.c
index c4ada32598bd5..6ca8689a83b5b 100644
--- a/kernel/fork.c
+++ b/kernel/fork.c
@@ -2295,7 +2295,7 @@ __latent_entropy struct task_struct *copy_process(
 	if (need_futex_hash_allocate_default(clone_flags)) {
 		retval = futex_hash_allocate_default();
 		if (retval)
-			goto bad_fork_core_free;
+			goto bad_fork_cancel_cgroup;
 		/*
 		 * If we fail beyond this point we don't free the allocated
 		 * futex hash map. We assume that another thread will be created

Sebastian