[PATCH 09/11] kasan: fix memory corruption in kasan_bitops_tags test

[Alexander Potapenko]

On Tue, Jan 12, 2021 at 9:07 PM 'Andrey Konovalov' via kasan-dev
<kasan-dev at googlegroups.com> wrote:
>
> On Tue, Jan 12, 2021 at 9:30 AM Alexander Potapenko <glider at google.com> wrote:
> >
> > On Tue, Jan 5, 2021 at 7:28 PM Andrey Konovalov <andreyknvl at google.com> wrote:
> > >
> > > Since the hardware tag-based KASAN mode might not have a redzone that
> > > comes after an allocated object (when kasan.mode=prod is enabled), the
> > > kasan_bitops_tags() test ends up corrupting the next object in memory.
> > >
> > > Change the test so it always accesses the redzone that lies within the
> > > allocated object's boundaries.
> > >
> > > Signed-off-by: Andrey Konovalov <andreyknvl at google.com>
> > > Link: https://linux-review.googlesource.com/id/I67f51d1ee48f0a8d0fe2658c2a39e4879fe0832a
Reviewed-by: Alexander Potapenko <glider at google.com>

> > > ---
> > >  lib/test_kasan.c | 12 ++++++------
> > >  1 file changed, 6 insertions(+), 6 deletions(-)
> > >
> > > diff --git a/lib/test_kasan.c b/lib/test_kasan.c
> > > index b67da7f6e17f..3ea52da52714 100644
> > > --- a/lib/test_kasan.c
> > > +++ b/lib/test_kasan.c
> > > @@ -771,17 +771,17 @@ static void kasan_bitops_tags(struct kunit *test)
> > >
> > >         /* This test is specifically crafted for the tag-based mode. */
> > >         if (IS_ENABLED(CONFIG_KASAN_GENERIC)) {
> > > -               kunit_info(test, "skipping, CONFIG_KASAN_SW_TAGS required");
> > > +               kunit_info(test, "skipping, CONFIG_KASAN_SW/HW_TAGS required");
> > >                 return;
> > >         }
> > >
> > > -       /* Allocation size will be rounded to up granule size, which is 16. */
> > > -       bits = kzalloc(sizeof(*bits), GFP_KERNEL);
> > > +       /* kmalloc-64 cache will be used and the last 16 bytes will be the redzone. */
> > > +       bits = kzalloc(48, GFP_KERNEL);
> >
> > I think it might make sense to call ksize() here to ensure we have
> > these spare bytes.
>
> Calling ksize() will unpoison the whole object.

Ah, that's right.

> I think it's OK to make assumptions about KASAN internals in tests. I
> would actually say that we need more tests that check such internal
> properties.

Agreed.