Re: [PATCH v3 08/17] slab: Remove folio references from kvfree()

[Vlastimil Babka <vbabka@suse.cz>]

Subject says kvfree() but it's kfree()

On 10/24/25 22:44, Matthew Wilcox (Oracle) wrote:
> As with memcg_slab_post_charge(), we save a call to compound_head()
> for large kmallocs.

I'd consider small kmallocs the more common path and so...

> This has a slight change of behaviour in that
> non-vmalloc, non-slab, non-kmalloc pointers will now cause a NULL
> pointer dereference rather than a warning.  We could add that back if
> really needed.

... considering also this...
> Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>

I'd rather restructure it to try page_slab() first, and free_large_kmalloc()
if it's unlikely(NULL), without checking the PageLargeKmalloc flag (let
free_large_kmalloc() do it for the WARN).

kvfree_rcu_cb() in previous patch too then, for consistency

> ---
>  mm/slub.c | 10 +++++-----
>  1 file changed, 5 insertions(+), 5 deletions(-)
> 
> diff --git a/mm/slub.c b/mm/slub.c
> index debe7132b343..100ed979312d 100644
> --- a/mm/slub.c
> +++ b/mm/slub.c
> @@ -6809,7 +6809,7 @@ void kvfree_rcu_cb(struct rcu_head *head)
>   */
>  void kfree(const void *object)
>  {
> -	struct folio *folio;
> +	struct page *page;
>  	struct slab *slab;
>  	struct kmem_cache *s;
>  	void *x = (void *)object;
> @@ -6819,13 +6819,13 @@ void kfree(const void *object)
>  	if (unlikely(ZERO_OR_NULL_PTR(object)))
>  		return;
>  
> -	folio = virt_to_folio(object);
> -	if (unlikely(!folio_test_slab(folio))) {
> -		free_large_kmalloc(&folio->page, (void *)object);
> +	page = virt_to_page(object);
> +	if (unlikely(PageLargeKmalloc(page))) {
> +		free_large_kmalloc(page, (void *)object);
>  		return;
>  	}
>  
> -	slab = folio_slab(folio);
> +	slab = page_slab(page);
>  	s = slab->slab_cache;
>  	slab_free(s, slab, x, _RET_IP_);
>  }