Re: [PATCH v2] mm: Make ksize() a reporting-only function

[Kees Cook]

On Sat, Nov 26, 2022 at 06:04:39PM +0100, Andrey Konovalov wrote:
> On Fri, Nov 18, 2022 at 4:57 AM Kees Cook <keescook@xxxxxxxxxxxx> wrote:
> >
> > With all "silently resizing" callers of ksize() refactored, remove the
> > logic in ksize() that would allow it to be used to effectively change
> > the size of an allocation (bypassing __alloc_size hints, etc). Users
> > wanting this feature need to either use kmalloc_size_roundup() before an
> > allocation, or use krealloc() directly.
> >
> > For kfree_sensitive(), move the unpoisoning logic inline. Replace the
> > some of the partially open-coded ksize() in __do_krealloc with ksize()
> > now that it doesn't perform unpoisoning.
> >
> > Adjust the KUnit tests to match the new ksize() behavior.
> >
> > Cc: Andrey Konovalov <andreyknvl@xxxxxxxxx>
> > Cc: Christoph Lameter <cl@xxxxxxxxx>
> > Cc: Pekka Enberg <penberg@xxxxxxxxxx>
> > Cc: David Rientjes <rientjes@xxxxxxxxxx>
> > Cc: Joonsoo Kim <iamjoonsoo.kim@xxxxxxx>
> > Cc: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
> > Cc: Roman Gushchin <roman.gushchin@xxxxxxxxx>
> > Cc: Hyeonggon Yoo <42.hyeyoo@xxxxxxxxx>
> > Cc: Andrey Ryabinin <ryabinin.a.a@xxxxxxxxx>
> > Cc: Alexander Potapenko <glider@xxxxxxxxxx>
> > Cc: Dmitry Vyukov <dvyukov@xxxxxxxxxx>
> > Cc: Vincenzo Frascino <vincenzo.frascino@xxxxxxx>
> > Cc: linux-mm@xxxxxxxxx
> > Cc: kasan-dev@xxxxxxxxxxxxxxxx
> > Acked-by: Vlastimil Babka <vbabka@xxxxxxx>
> > Signed-off-by: Kees Cook <keescook@xxxxxxxxxxxx>
> > ---
> > v2:
> > - improve kunit test precision (andreyknvl)
> > - add Ack (vbabka)
> > v1: https://lore.kernel.org/all/20221022180455.never.023-kees@xxxxxxxxxx
> > ---
> >  mm/kasan/kasan_test.c | 14 +++++++++-----
> >  mm/slab_common.c      | 26 ++++++++++----------------
> >  2 files changed, 19 insertions(+), 21 deletions(-)
> >
> > diff --git a/mm/kasan/kasan_test.c b/mm/kasan/kasan_test.c
> > index 7502f03c807c..fc4b22916587 100644
> > --- a/mm/kasan/kasan_test.c
> > +++ b/mm/kasan/kasan_test.c
> > @@ -821,7 +821,7 @@ static void kasan_global_oob_left(struct kunit *test)
> >         KUNIT_EXPECT_KASAN_FAIL(test, *(volatile char *)p);
> >  }
> >
> > -/* Check that ksize() makes the whole object accessible. */
> > +/* Check that ksize() does NOT unpoison whole object. */
> >  static void ksize_unpoisons_memory(struct kunit *test)
> >  {
> >         char *ptr;
> > @@ -829,15 +829,19 @@ static void ksize_unpoisons_memory(struct kunit *test)
> >
> >         ptr = kmalloc(size, GFP_KERNEL);
> >         KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr);
> > +
> >         real_size = ksize(ptr);
> > +       KUNIT_EXPECT_GT(test, real_size, size);
> >
> >         OPTIMIZER_HIDE_VAR(ptr);
> >
> > -       /* This access shouldn't trigger a KASAN report. */
> > -       ptr[size] = 'x';
> > +       /* These accesses shouldn't trigger a KASAN report. */
> > +       ptr[0] = 'x';
> > +       ptr[size - 1] = 'x';
> >
> > -       /* This one must. */
> > -       KUNIT_EXPECT_KASAN_FAIL(test, ((volatile char *)ptr)[real_size]);
> > +       /* These must trigger a KASAN report. */
> > +       KUNIT_EXPECT_KASAN_FAIL(test, ((volatile char *)ptr)[size]);
> > +       KUNIT_EXPECT_KASAN_FAIL(test, ((volatile char *)ptr)[real_size - 1]);
> 
> Hi Kees,
> 
> I just realized there's an issue here with the tag-based modes, as
> they align the unpoisoned area to 16 bytes.
> 
> One solution would be to change the allocation size to 128 -
> KASAN_GRANULE_SIZE - 5, the same way kmalloc_oob_right test does it,
> so that the last 16-byte granule won't get unpoisoned for the
> tag-based modes. And then check that the ptr[size] access fails only
> for the Generic mode.

Ah! Good point. Are you able to send a patch? I suspect you know exactly
what to change; it might take me a bit longer to double-check all of
those details.

-Kees

-- 
Kees Cook