[lvc-project] [PATCH] [RFC] fs: prefer kfree_rcu() in fasync_remove_entry()

[Al Viro]

On Fri, Feb 09, 2024 at 03:52:19PM +0300, Dmitry Antipov wrote:
> In 'fasync_remove_entry()', prefer 'kfree_rcu()' over 'call_rcu()' with dummy
> 'fasync_free_rcu()' callback. This is mostly intended in attempt to fix weird
> https://syzkaller.appspot.com/bug?id=6a64ad907e361e49e92d1c4c114128a1bda2ed7f,
> where kmemleak may consider 'fa' as unreferenced during RCU grace period. See
> https://lore.kernel.org/stable/20230930174657.800551-1-joel@joelfernandes.org
> as well. Comments are highly appreciated.

That should go with mentioning that _these_ _days_ kfree() can be paired with
kmem_cache_alloc().  A reference to ae65a5211d90 "mm/slab: document kfree() as
allowed for kmem_cache_alloc() objects" might be a good idea; at the very least
it *must* come with "don't even think of backporting to any kernel that still
has SLOB support (i.e. anything prior to 6.4)".

Not sure if it's a good idea at this point - it doesn't look like it would get
mixed into anything that might need backporting, but...