Re: [PATCH] kasan: adopt KUNIT tests to SW_TAGS mode

[David Gow <davidgow@google.com>]

On Sat, Oct 17, 2020 at 3:33 AM Andrey Konovalov <andreyknvl@google.com> wrote:
>
> Now that we have KASAN-KUNIT tests integration, it's easy to see that
> some KASAN tests are not adopted to the SW_TAGS mode and are failing.
>
> Adjust the allocation size for kasan_memchr() and kasan_memcmp() by
> roung it up to OOB_TAG_OFF so the bad access ends up in a separate
> memory granule.
>
> Add new kmalloc_uaf_16() and kasan_bitops_uaf() tests that rely on UAFs,
> as it's hard to adopt the existing kmalloc_oob_16() and kasan_bitops_oob()
> (rename from kasan_bitops()) without losing the precision.
>
> Disable kasan_global_oob() and kasan_alloca_oob_left/right() as SW_TAGS
> mode doesn't instrument globals nor dynamic allocas.
>
> Signed-off-by: Andrey Konovalov <andreyknvl@google.com>

This looks good to me. Though, as you mention, writing to freed memory
might not bode well for system stability after the test runs. I don't
think that needs to be a goal for these tests, though.

One thing which we're hoping to add to KUnit soon is support for
skipping tests: once that's in place, we can use it to mark tests as
explicitly skipped if they rely on the GENERIC mode. That'll take a
little while to get upstream though, so I wouldn't want to hold this
up for it.

Otherwise, from the KUnit side, this looks great.

I also tested it against the GENERIC mode on x86_64 (which is all I
have set up here at the moment), and nothing obviously had broken.
So:
Tested-by: David Gow <davidgow@google.com>

Cheers,
-- David