Re: [PATCH] slab: align ZERO_SIZE_PTR to ARCH_KMALLOC_MINALIGN

[Karl Mehltretter]

On Mon, Aug 10, 2026 at 03:26:37PM +0100, Catalin Marinas wrote:
> On Sat, Aug 08, 2026 at 12:56:22PM +0200, Karl Mehltretter wrote:
> > diff --git a/include/linux/slab.h b/include/linux/slab.h
> > index 32c9f8ed7ae20..0798a714da87f 100644
> > --- a/include/linux/slab.h
> > +++ b/include/linux/slab.h
> > @@ -259,13 +259,16 @@ enum _slab_flag_bits {
> >  
> >  /*
> >   * ZERO_SIZE_PTR will be returned for zero sized kmalloc requests.
> > + * It satisfies the alignment promised by __assume_kmalloc_alignment
> > + * and keeps the historic value 16 where that is already aligned.
> >   *
> >   * Dereferencing ZERO_SIZE_PTR will lead to a distinct access fault.
> >   *
> >   * ZERO_SIZE_PTR can be passed to kfree though in the same way that NULL can.
> >   * Both make kfree a no-op.
> >   */
> > -#define ZERO_SIZE_PTR ((void *)16)
> > +#define ZERO_SIZE_PTR ((void *)(ARCH_KMALLOC_MINALIGN > 16 ? \
> > +				ARCH_KMALLOC_MINALIGN : 16))
> >  
> >  #define ZERO_OR_NULL_PTR(x) ((unsigned long)(x) <= \
> >  				(unsigned long)ZERO_SIZE_PTR)
> > @@ -622,6 +625,12 @@ static inline bool kmem_dump_obj(void *object) { return false; }
> >  #define KMALLOC_SHIFT_LOW ilog2(KMALLOC_MIN_SIZE)
> >  #endif
> >  
> > +/*
> > + * Keep ZERO_SIZE_PTR below PAGE_SIZE and the low pointer poison values.
> > + * 128 is the largest in-tree ARCH_KMALLOC_MINALIGN.
> > + */
> > +static_assert(ARCH_KMALLOC_MINALIGN <= 128);
> 
> I'm not sure we should bother with this, or at least make it strictly
> less than PAGE_SIZE since 128 doesn't have any meaning for the slab
> allocator.
> 

Thanks for the review!

The case I had in mind when choosing 128 was POISON_POINTER_DELTA == 0,
where LIST_POISON1 is 256 (0x100). With the current range check, an
architecture with non-coherent DMA and 256-byte cache lines would make
kfree(LIST_POISON1) a silent no-op.

Looking at this again, I think ZERO_OR_NULL_PTR() should match only NULL
and ZERO_SIZE_PTR, rather than the whole range below the sentinel. This
would also stop unrelated low pointer values from being silently
accepted. I'll include that change in v2.

Thanks,
Karl