Re: [PATCH 3/4] mm: Add free()

[Matthew Wilcox]

On Fri, Mar 23, 2018 at 08:14:21AM -0700, Matthew Wilcox wrote:
> > One more thing, there is
> > some kasan checks on the main way of kfree(), and there is no guarantee they
> > reflected in kmem_cache_free() identical.
> 
> Which function are you talking about here?
> 
> slub calls slab_free() for both kfree() and kmem_cache_free().
> slab calls __cache_free() for both kfree() and kmem_cache_free().
> Each of them do their kasan handling in the called function.

... except for where slub can free large objects without calling slab_free():

        if (unlikely(!PageSlab(page))) {
                BUG_ON(!PageCompound(page));
                kfree_hook(object);
                __free_pages(page, compound_order(page));
                return;
        }
        slab_free(page->slab_cache, page, object, NULL, 1, _RET_IP_);

If you call kmalloc(16384, GFP_KERNEL), slub will hand back an order-2
page without setting PageSlab on it.  So if that gets passed to free(),
it'll call __put_page() which calls free_compound_page() which calls
__free_pages_ok().  Looks like we want another compound_dtor to be sure
we call the kfree hook.