Re: [PATCH v2] mm/slab_common: Deleting kobject in kmem_cache_destroy() without holding slab_mutex/cpu_hotplug_lock

[Roman Gushchin <roman.gushchin@linux.dev>]

On Wed, Aug 10, 2022 at 12:49:46PM -0400, Waiman Long wrote:
> A circular locking problem is reported by lockdep due to the following
> circular locking dependency.
> 
>   +--> cpu_hotplug_lock --> slab_mutex --> kn->active --+
>   |                                                     |
>   +-----------------------------------------------------+
> 
> The forward cpu_hotplug_lock ==> slab_mutex ==> kn->active dependency
> happens in
> 
>   kmem_cache_destroy():	cpus_read_lock(); mutex_lock(&slab_mutex);
>   ==> sysfs_slab_unlink()
>       ==> kobject_del()
>           ==> kernfs_remove()
> 	      ==> __kernfs_remove()
> 	          ==> kernfs_drain(): rwsem_acquire(&kn->dep_map, ...);
> 
> The backward kn->active ==> cpu_hotplug_lock dependency happens in
> 
>   kernfs_fop_write_iter(): kernfs_get_active();
>   ==> slab_attr_store()
>       ==> cpu_partial_store()
>           ==> flush_all(): cpus_read_lock()
> 
> One way to break this circular locking chain is to avoid holding
> cpu_hotplug_lock and slab_mutex while deleting the kobject in
> sysfs_slab_unlink() which should be equivalent to doing a write_lock
> and write_unlock pair of the kn->active virtual lock.
> 
> Since the kobject structures are not protected by slab_mutex or the
> cpu_hotplug_lock, we can certainly release those locks before doing
> the delete operation.
> 
> Move sysfs_slab_unlink() and sysfs_slab_release() to the newly
> created kmem_cache_release() and call it outside the slab_mutex &
> cpu_hotplug_lock critical sections.
> 
> Signed-off-by: Waiman Long <longman@redhat.com>
> ---
>  [v2] Break kmem_cache_release() helper into 2 separate ones.
> 
>  mm/slab_common.c | 54 +++++++++++++++++++++++++++++++++---------------
>  1 file changed, 37 insertions(+), 17 deletions(-)
> 
> diff --git a/mm/slab_common.c b/mm/slab_common.c
> index 17996649cfe3..7742d0446d8b 100644
> --- a/mm/slab_common.c
> +++ b/mm/slab_common.c
> @@ -392,6 +392,36 @@ kmem_cache_create(const char *name, unsigned int size, unsigned int align,
>  }
>  EXPORT_SYMBOL(kmem_cache_create);
>  
> +#ifdef SLAB_SUPPORTS_SYSFS
> +static void kmem_cache_workfn_release(struct kmem_cache *s)
> +{
> +	sysfs_slab_release(s);
> +}
> +#else
> +static void kmem_cache_workfn_release(struct kmem_cache *s)
> +{
> +	slab_kmem_cache_release(s);
> +}
> +#endif
> +
> +/*
> + * For a given kmem_cache, kmem_cache_destroy() should only be called
> + * once or there will be a use-after-free problem. The actual deletion
> + * and release of the kobject does not need slab_mutex or cpu_hotplug_lock
> + * protection. So they are now done without holding those locks.
> + */
> +static void kmem_cache_release(struct kmem_cache *s)
> +{
> +#ifdef SLAB_SUPPORTS_SYSFS
> +	sysfs_slab_unlink(s);
> +#endif
> +
> +	if (s->flags & SLAB_TYPESAFE_BY_RCU)
> +		schedule_work(&slab_caches_to_rcu_destroy_work);
> +	else
> +		kmem_cache_workfn_release(s);
> +}
> +
>  static void slab_caches_to_rcu_destroy_workfn(struct work_struct *work)
>  {
>  	LIST_HEAD(to_destroy);
> @@ -418,11 +448,7 @@ static void slab_caches_to_rcu_destroy_workfn(struct work_struct *work)
>  	list_for_each_entry_safe(s, s2, &to_destroy, list) {
>  		debugfs_slab_release(s);
>  		kfence_shutdown_cache(s);
> -#ifdef SLAB_SUPPORTS_SYSFS
> -		sysfs_slab_release(s);
> -#else
> -		slab_kmem_cache_release(s);
> -#endif
> +		kmem_cache_workfn_release(s);
>  	}
>  }
>  
> @@ -437,20 +463,10 @@ static int shutdown_cache(struct kmem_cache *s)
>  	list_del(&s->list);
>  
>  	if (s->flags & SLAB_TYPESAFE_BY_RCU) {
> -#ifdef SLAB_SUPPORTS_SYSFS
> -		sysfs_slab_unlink(s);
> -#endif
>  		list_add_tail(&s->list, &slab_caches_to_rcu_destroy);
> -		schedule_work(&slab_caches_to_rcu_destroy_work);

Hi Waiman!

This version is much more readable, thank you!

But can we, please, leave this schedule_work(&slab_caches_to_rcu_destroy_work)
call here? I don't see a good reason to move it, do I miss something?
It's nice to have list_add_tail() and schedule_work() calls nearby, so
it's obvious we can't miss the latter.

Thanks!