[RFC v2 01/13] mm/kfence: Add a new kunit test test_use_after_free_read_nofault()

[Ritesh Harjani (IBM)]

Hello Kasan/kfence-devs, 

Wanted your inputs on this kfence kunit test [PATCH-1] and it's respective
powerpc fix [Patch-2]. The commit msgs has a good description of it. I
see that the same problem was noticed on s390 as well [1] a while ago.
So that makes me believe that maybe we should have a kunit test for the
same to make sure all architectures handles this properly. 

Thoughts?

[1]: https://lore.kernel.org/all/20230213183858.1473681-1-hca@linux.ibm.com/

-ritesh


"Ritesh Harjani (IBM)" <ritesh.list at gmail.com> writes:

> From: Nirjhar Roy <nirjhar at linux.ibm.com>
>
> Faults from copy_from_kernel_nofault() needs to be handled by fixup
> table and should not be handled by kfence. Otherwise while reading
> /proc/kcore which uses copy_from_kernel_nofault(), kfence can generate
> false negatives. This can happen when /proc/kcore ends up reading an
> unmapped address from kfence pool.
>
> Let's add a testcase to cover this case.
>
> Co-developed-by: Ritesh Harjani (IBM) <ritesh.list at gmail.com>
> Signed-off-by: Ritesh Harjani (IBM) <ritesh.list at gmail.com>
> Signed-off-by: Nirjhar Roy <nirjhar at linux.ibm.com>
> Cc: kasan-dev at googlegroups.com
> Cc: Alexander Potapenko <glider at google.com>
> Cc: linux-mm at kvack.org
> ---
>  mm/kfence/kfence_test.c | 17 +++++++++++++++++
>  1 file changed, 17 insertions(+)
>
> diff --git a/mm/kfence/kfence_test.c b/mm/kfence/kfence_test.c
> index 00fd17285285..f65fb182466d 100644
> --- a/mm/kfence/kfence_test.c
> +++ b/mm/kfence/kfence_test.c
> @@ -383,6 +383,22 @@ static void test_use_after_free_read(struct kunit *test)
>  	KUNIT_EXPECT_TRUE(test, report_matches(&expect));
>  }
>
> +static void test_use_after_free_read_nofault(struct kunit *test)
> +{
> +	const size_t size = 32;
> +	char *addr;
> +	char dst;
> +	int ret;
> +
> +	setup_test_cache(test, size, 0, NULL);
> +	addr = test_alloc(test, size, GFP_KERNEL, ALLOCATE_ANY);
> +	test_free(addr);
> +	/* Use after free with *_nofault() */
> +	ret = copy_from_kernel_nofault(&dst, addr, 1);
> +	KUNIT_EXPECT_EQ(test, ret, -EFAULT);
> +	KUNIT_EXPECT_FALSE(test, report_available());
> +}
> +
>  static void test_double_free(struct kunit *test)
>  {
>  	const size_t size = 32;
> @@ -780,6 +796,7 @@ static struct kunit_case kfence_test_cases[] = {
>  	KFENCE_KUNIT_CASE(test_out_of_bounds_read),
>  	KFENCE_KUNIT_CASE(test_out_of_bounds_write),
>  	KFENCE_KUNIT_CASE(test_use_after_free_read),
> +	KFENCE_KUNIT_CASE(test_use_after_free_read_nofault),
>  	KFENCE_KUNIT_CASE(test_double_free),
>  	KFENCE_KUNIT_CASE(test_invalid_addr_free),
>  	KFENCE_KUNIT_CASE(test_corruption),
> --
> 2.46.0