Re: [PATCH 05/14] tracefs: replace call_rcu by kfree_rcu for simple kmem_cache_free callback

[Vlastimil Babka]

On 6/10/24 10:36 PM, Steven Rostedt wrote:
> On Mon, 10 Jun 2024 08:46:42 -0700
> "Paul E. McKenney" <paulmck@xxxxxxxxxx> wrote:
> 
>> > > index 7c29f4afc23d..338c52168e61 100644
>> > > --- a/fs/tracefs/inode.c
>> > > +++ b/fs/tracefs/inode.c
>> > > @@ -53,14 +53,6 @@ static struct inode *tracefs_alloc_inode(struct super_block *sb)
>> > >  	return &ti->vfs_inode;
>> > >  }
>> > >  
>> > > -static void tracefs_free_inode_rcu(struct rcu_head *rcu)
>> > > -{
>> > > -	struct tracefs_inode *ti;
>> > > -
>> > > -	ti = container_of(rcu, struct tracefs_inode, rcu);
>> > > -	kmem_cache_free(tracefs_inode_cachep, ti);  
>> > 
>> > Does this work?
>> > 
>> > tracefs needs to be freed via the tracefs_inode_cachep. Does
>> > kfree_rcu() handle specific frees for objects that were not allocated
>> > via kmalloc()?  
>> 
>> A recent change to kfree() allows it to correctly handle memory allocated
>> via kmem_cache_alloc().  News to me as of a few weeks ago.  ;-)
> 
> If that's the case then:
> 
> Acked-by: Steven Rostedt (Google) <rostedt@xxxxxxxxxxx>
> 
> Do we have a way to add a "Depends-on" tag so that anyone backporting this
> will know that it requires the change to whatever allowed that to happen?

Looks like people use that tag, although no grep hits in Documentation, so
Cc'ing workflows@ and Thorsten.

In this case it would be

Depends-on: c9929f0e344a ("mm/slob: remove CONFIG_SLOB")

> Or we need to update the change log to explicitly state that this should
> *not* be backported.
> 
> -- Steve