defer_free() links pending objects using the slab's freelist offset
which is fine as they are not free yet. free_deferred_objects() then
clears this pointer to avoid confusing the debugging consistency checks
that may be enabled for the cache.

However, with CONFIG_SLAB_FREELIST_HARDENED, even the NULL pointer needs
to be encoded appropriately using set_freepointer(), otherwise it's
decoded as something else and triggers the consistency checks, as found
by the kernel test robot.

Use set_freepointer() to prevent the issue.

Fixes: af92793e52c3 ("slab: Introduce kmalloc_nolock() and kfree_nolock().")
Reported-by: kernel test robot <oliver.sang@intel.com>
Closes: https://lore.kernel.org/oe-lkp/202510101652.7921fdc6-lkp@intel.com
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
---
will be added to slab/for-next-fixes and sent later this week
---
 mm/slub.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/mm/slub.c b/mm/slub.c
index b1f15598fbfd..64c17afc375b 100644
--- a/mm/slub.c
+++ b/mm/slub.c
@@ -6443,15 +6443,16 @@ static void free_deferred_objects(struct irq_work *work)
 		slab = virt_to_slab(x);
 		s = slab->slab_cache;
 
+
+		/* Point 'x' back to the beginning of allocated object */
+		x -= s->offset;
 		/*
 		 * We used freepointer in 'x' to link 'x' into df->objects.
 		 * Clear it to NULL to avoid false positive detection
 		 * of "Freepointer corruption".
 		 */
-		*(void **)x = NULL;
+		set_freepointer(s, x, NULL);
 
-		/* Point 'x' back to the beginning of allocated object */
-		x -= s->offset;
 		__slab_free(s, slab, x, x, 1, _THIS_IP_);
 	}
 

---
base-commit: 3a8660878839faadb4f1a6dd72c3179c1df56787
change-id: 20251014-fix-freelist-29915edcc2f4

Best regards,
-- 
Vlastimil Babka <vbabka@suse.cz>

On 10/14/25 10:40, Vlastimil Babka wrote:
> defer_free() links pending objects using the slab's freelist offset
> which is fine as they are not free yet. free_deferred_objects() then
> clears this pointer to avoid confusing the debugging consistency checks
> that may be enabled for the cache.
> 
> However, with CONFIG_SLAB_FREELIST_HARDENED, even the NULL pointer needs
> to be encoded appropriately using set_freepointer(), otherwise it's
> decoded as something else and triggers the consistency checks, as found
> by the kernel test robot.
> 
> Use set_freepointer() to prevent the issue.
> 
> Fixes: af92793e52c3 ("slab: Introduce kmalloc_nolock() and kfree_nolock().")
> Reported-by: kernel test robot <oliver.sang@intel.com>
> Closes: https://lore.kernel.org/oe-lkp/202510101652.7921fdc6-lkp@intel.com
> Acked-by: Alexei Starovoitov <ast@kernel.org>
> Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
> ---
> will be added to slab/for-next-fixes and sent later this week
> ---
>  mm/slub.c | 7 ++++---
>  1 file changed, 4 insertions(+), 3 deletions(-)
> 
> diff --git a/mm/slub.c b/mm/slub.c
> index b1f15598fbfd..64c17afc375b 100644
> --- a/mm/slub.c
> +++ b/mm/slub.c
> @@ -6443,15 +6443,16 @@ static void free_deferred_objects(struct irq_work *work)
>  		slab = virt_to_slab(x);
>  		s = slab->slab_cache;
>  
> +

extra newline moved...

> +		/* Point 'x' back to the beginning of allocated object */
> +		x -= s->offset;

... here. noticed and fixed up locally after sending...

>  		/*
>  		 * We used freepointer in 'x' to link 'x' into df->objects.
>  		 * Clear it to NULL to avoid false positive detection
>  		 * of "Freepointer corruption".
>  		 */
> -		*(void **)x = NULL;
> +		set_freepointer(s, x, NULL);
>  
> -		/* Point 'x' back to the beginning of allocated object */
> -		x -= s->offset;
>  		__slab_free(s, slab, x, x, 1, _THIS_IP_);
>  	}
>  
> 
> ---
> base-commit: 3a8660878839faadb4f1a6dd72c3179c1df56787
> change-id: 20251014-fix-freelist-29915edcc2f4
> 
> Best regards,

On Tue, Oct 14, 2025 at 10:44:51AM +0200, Vlastimil Babka wrote:
> On 10/14/25 10:40, Vlastimil Babka wrote:
> > defer_free() links pending objects using the slab's freelist offset
> > which is fine as they are not free yet. free_deferred_objects() then
> > clears this pointer to avoid confusing the debugging consistency checks
> > that may be enabled for the cache.
> > 
> > However, with CONFIG_SLAB_FREELIST_HARDENED, even the NULL pointer needs
> > to be encoded appropriately using set_freepointer(), otherwise it's
> > decoded as something else and triggers the consistency checks, as found
> > by the kernel test robot.
> > 
> > Use set_freepointer() to prevent the issue.
> > 
> > Fixes: af92793e52c3 ("slab: Introduce kmalloc_nolock() and kfree_nolock().")
> > Reported-by: kernel test robot <oliver.sang@intel.com>
> > Closes: https://lore.kernel.org/oe-lkp/202510101652.7921fdc6-lkp@intel.com
> > Acked-by: Alexei Starovoitov <ast@kernel.org>
> > Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
> > ---
> > will be added to slab/for-next-fixes and sent later this week
> > ---
> >  mm/slub.c | 7 ++++---
> >  1 file changed, 4 insertions(+), 3 deletions(-)
> > 
> > diff --git a/mm/slub.c b/mm/slub.c
> > index b1f15598fbfd..64c17afc375b 100644
> > --- a/mm/slub.c
> > +++ b/mm/slub.c
> > @@ -6443,15 +6443,16 @@ static void free_deferred_objects(struct irq_work *work)
> >  		slab = virt_to_slab(x);
> >  		s = slab->slab_cache;
> >  
> > +
> 
> extra newline moved...
> 
> > +		/* Point 'x' back to the beginning of allocated object */
> > +		x -= s->offset;
> 
> ... here. noticed and fixed up locally after sending...

LGTM,
Reviewed-by: Harry Yoo <harry.yoo@oracle.com>

Thanks for fixing it!

> >  		/*
> >  		 * We used freepointer in 'x' to link 'x' into df->objects.
> >  		 * Clear it to NULL to avoid false positive detection
> >  		 * of "Freepointer corruption".
> >  		 */
> > -		*(void **)x = NULL;
> > +		set_freepointer(s, x, NULL);
> >  
> > -		/* Point 'x' back to the beginning of allocated object */
> > -		x -= s->offset;
> >  		__slab_free(s, slab, x, x, 1, _THIS_IP_);
> >  	}

-- 
Cheers,
Harry / Hyeonggon