On Tue, Feb 12, 2019 at 10:12 PM Andrew Morton
<[email protected]> wrote:
>
> On Mon, 11 Feb 2019 22:59:53 +0100 Andrey Konovalov <[email protected]>
> wrote:
>
> > With tag based KASAN page_address() looks at the page flags to see
> > whether the resulting pointer needs to have a tag set. Since we don't
> > want to set a tag when page_address() is called on SLAB pages, we call
> > page_kasan_tag_reset() in kasan_poison_slab(). However in allocate_slab()
> > page_address() is called before kasan_poison_slab(). Fix it by changing
> > the order.
> >
> > ...
> >
> > --- a/mm/slub.c
> > +++ b/mm/slub.c
> > @@ -1642,12 +1642,15 @@ static struct page *allocate_slab(struct kmem_cache
> > *s, gfp_t flags, int node)
> > if (page_is_pfmemalloc(page))
> > SetPageSlabPfmemalloc(page);
> >
> > + kasan_poison_slab(page);
> > +
> > start = page_address(page);
> >
> > - if (unlikely(s->flags & SLAB_POISON))
> > + if (unlikely(s->flags & SLAB_POISON)) {
> > + metadata_access_enable();
> > memset(start, POISON_INUSE, PAGE_SIZE << order);
> > -
> > - kasan_poison_slab(page);
> > + metadata_access_disable();
> > + }
> >
> > shuffle = shuffle_freelist(s, page);
>
> This doesn't compile when CONFIG_SLUB_DEBUG=n. Please review carefully:
Sorry, missed this. I think it makes more sense to move this memset
into another function CONFIG_SLUB_DEBUG ifdef, since all other
poisoning code is also there. I'll send a v2.
>
> --- a/mm/slub.c~kasan-slub-move-kasan_poison_slab-hook-before-page_address-fix
> +++ a/mm/slub.c
> @@ -1357,6 +1357,14 @@ slab_flags_t kmem_cache_flags(unsigned i
>
> #define disable_higher_order_debug 0
>
> +static inline void metadata_access_enable(void)
> +{
> +}
> +
> +static inline void metadata_access_disable(void)
> +{
> +}
> +
> static inline unsigned long slabs_node(struct kmem_cache *s, int node)
> { return 0; }
> static inline unsigned long node_nr_slabs(struct kmem_cache_node *n)
> _
>
