[PATCH] kasan: test: Bypass __alloc_size checks

[Kees Cook]

Intentional overflows, as performed by the KASAN tests, are detected
at compile time[1] (instead of only at run-time) with the addition of
__alloc_size. Fix this by forcing the compiler into not being able to
trust the size used following the kmalloc()s.

[1] https://lore.kernel.org/lkml/20211005184717.65c6...@linux-foundation.org

Cc: Andrey Ryabinin <ryabin...@gmail.com>
Cc: Alexander Potapenko <gli...@google.com>
Cc: Andrey Konovalov <andre...@gmail.com>
Cc: Dmitry Vyukov <dvy...@google.com>
Cc: kasa...@googlegroups.com
Signed-off-by: Kees Cook <kees...@chromium.org>
---
lib/test_kasan.c | 10 +++++-----
lib/test_kasan_module.c | 2 +-
2 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/lib/test_kasan.c b/lib/test_kasan.c
index 8835e0784578..0e1f8d5281b4 100644
--- a/lib/test_kasan.c
+++ b/lib/test_kasan.c
@@ -435,7 +435,7 @@ static void kmalloc_uaf_16(struct kunit *test)
static void kmalloc_oob_memset_2(struct kunit *test)
{
char *ptr;
- size_t size = 128 - KASAN_GRANULE_SIZE;
+ volatile size_t size = 128 - KASAN_GRANULE_SIZE;

ptr = kmalloc(size, GFP_KERNEL);
KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr);
@@ -447,7 +447,7 @@ static void kmalloc_oob_memset_2(struct kunit *test)
static void kmalloc_oob_memset_4(struct kunit *test)
{
char *ptr;
- size_t size = 128 - KASAN_GRANULE_SIZE;
+ volatile size_t size = 128 - KASAN_GRANULE_SIZE;

ptr = kmalloc(size, GFP_KERNEL);
KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr);
@@ -459,7 +459,7 @@ static void kmalloc_oob_memset_4(struct kunit *test)
static void kmalloc_oob_memset_8(struct kunit *test)
{
char *ptr;
- size_t size = 128 - KASAN_GRANULE_SIZE;
+ volatile size_t size = 128 - KASAN_GRANULE_SIZE;

ptr = kmalloc(size, GFP_KERNEL);
KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr);
@@ -471,7 +471,7 @@ static void kmalloc_oob_memset_8(struct kunit *test)
static void kmalloc_oob_memset_16(struct kunit *test)
{
char *ptr;
- size_t size = 128 - KASAN_GRANULE_SIZE;
+ volatile size_t size = 128 - KASAN_GRANULE_SIZE;

ptr = kmalloc(size, GFP_KERNEL);
KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr);
@@ -483,7 +483,7 @@ static void kmalloc_oob_memset_16(struct kunit *test)
static void kmalloc_oob_in_memset(struct kunit *test)
{
char *ptr;
- size_t size = 128 - KASAN_GRANULE_SIZE;
+ volatile size_t size = 128 - KASAN_GRANULE_SIZE;

ptr = kmalloc(size, GFP_KERNEL);
KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr);
diff --git a/lib/test_kasan_module.c b/lib/test_kasan_module.c
index 7ebf433edef3..c8cc77b1dcf3 100644
--- a/lib/test_kasan_module.c
+++ b/lib/test_kasan_module.c
@@ -19,7 +19,7 @@ static noinline void __init copy_user_test(void)
{
char *kmem;
char __user *usermem;
- size_t size = 128 - KASAN_GRANULE_SIZE;
+ volatile size_t size = 128 - KASAN_GRANULE_SIZE;
int __maybe_unused unused;

kmem = kmalloc(size, GFP_KERNEL);
--
2.30.2

[Mark Rutland]

Hi Kees,

On Tue, Oct 05, 2021 at 08:55:22PM -0700, Kees Cook wrote:
> Intentional overflows, as performed by the KASAN tests, are detected
> at compile time[1] (instead of only at run-time) with the addition of
> __alloc_size. Fix this by forcing the compiler into not being able to
> trust the size used following the kmalloc()s.

It might be better to use OPTIMIZER_HIDE_VAR(), since that's intended to
make the value opaque to the compiler, and volatile might not always do
that depending on how the compiler tracks the variable.

Thanks,
Mark.

[Kees Cook]

On Wed, Oct 06, 2021 at 12:38:36PM +0100, Mark Rutland wrote:
> Hi Kees,
>
> On Tue, Oct 05, 2021 at 08:55:22PM -0700, Kees Cook wrote:
> > Intentional overflows, as performed by the KASAN tests, are detected
> > at compile time[1] (instead of only at run-time) with the addition of
> > __alloc_size. Fix this by forcing the compiler into not being able to
> > trust the size used following the kmalloc()s.
>
> It might be better to use OPTIMIZER_HIDE_VAR(), since that's intended to
> make the value opaque to the compiler, and volatile might not always do
> that depending on how the compiler tracks the variable.

Given both you and Jann[1] have suggested this, I'll send a v2 with that.
:) Thanks!

-Kees

[1] https://lore.kernel.org/lkml/CAG48ez19raco+s+UF8eiXqTv...@mail.gmail.com/

--
Kees Cook

[Kees Cook]

Intentional overflows, as performed by the KASAN tests, are detected
at compile time[1] (instead of only at run-time) with the addition of
__alloc_size. Fix this by forcing the compiler into not being able to
trust the size used following the kmalloc()s.

[1] https://lore.kernel.org/lkml/20211005184717.65c6...@linux-foundation.org

Cc: Andrey Ryabinin <ryabin...@gmail.com>
Cc: Alexander Potapenko <gli...@google.com>
Cc: Andrey Konovalov <andre...@gmail.com>
Cc: Dmitry Vyukov <dvy...@google.com>
Cc: kasa...@googlegroups.com
Signed-off-by: Kees Cook <kees...@chromium.org>
---

v2: use OPTIMIZER_HIDE_VAR() (jann, mark)
v1: https://lore.kernel.org/lkml/20211006035522.5...@chromium.org/
---
lib/test_kasan.c | 8 +++++++-
lib/test_kasan_module.c | 2 ++
2 files changed, 9 insertions(+), 1 deletion(-)

diff --git a/lib/test_kasan.c b/lib/test_kasan.c
index 8835e0784578..8a8a8133f4cd 100644
--- a/lib/test_kasan.c
+++ b/lib/test_kasan.c
@@ -440,6 +440,7 @@ static void kmalloc_oob_memset_2(struct kunit *test)

ptr = kmalloc(size, GFP_KERNEL);
KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr);

+ OPTIMIZER_HIDE_VAR(size);
KUNIT_EXPECT_KASAN_FAIL(test, memset(ptr + size - 1, 0, 2));
kfree(ptr);
}
@@ -452,6 +453,7 @@ static void kmalloc_oob_memset_4(struct kunit *test)

ptr = kmalloc(size, GFP_KERNEL);
KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr);

+ OPTIMIZER_HIDE_VAR(size);
KUNIT_EXPECT_KASAN_FAIL(test, memset(ptr + size - 3, 0, 4));
kfree(ptr);
}
@@ -464,6 +466,7 @@ static void kmalloc_oob_memset_8(struct kunit *test)

ptr = kmalloc(size, GFP_KERNEL);
KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr);

+ OPTIMIZER_HIDE_VAR(size);
KUNIT_EXPECT_KASAN_FAIL(test, memset(ptr + size - 7, 0, 8));
kfree(ptr);
}
@@ -476,6 +479,7 @@ static void kmalloc_oob_memset_16(struct kunit *test)

ptr = kmalloc(size, GFP_KERNEL);
KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr);

+ OPTIMIZER_HIDE_VAR(size);
KUNIT_EXPECT_KASAN_FAIL(test, memset(ptr + size - 15, 0, 16));
kfree(ptr);
}
@@ -488,6 +492,7 @@ static void kmalloc_oob_in_memset(struct kunit *test)

ptr = kmalloc(size, GFP_KERNEL);
KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr);

+ OPTIMIZER_HIDE_VAR(size);
KUNIT_EXPECT_KASAN_FAIL(test,
memset(ptr, 0, size + KASAN_GRANULE_SIZE));
kfree(ptr);
@@ -497,7 +502,7 @@ static void kmalloc_memmove_invalid_size(struct kunit *test)
{
char *ptr;
size_t size = 64;
- volatile size_t invalid_size = -2;
+ size_t invalid_size = -2;

/*
* Hardware tag-based mode doesn't check memmove for negative size.
@@ -510,6 +515,7 @@ static void kmalloc_memmove_invalid_size(struct kunit *test)
KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr);

memset((char *)ptr, 0, 64);
+ OPTIMIZER_HIDE_VAR(invalid_size);
KUNIT_EXPECT_KASAN_FAIL(test,
memmove((char *)ptr, (char *)ptr + 4, invalid_size));
kfree(ptr);
diff --git a/lib/test_kasan_module.c b/lib/test_kasan_module.c
index 7ebf433edef3..b112cbc835e9 100644
--- a/lib/test_kasan_module.c
+++ b/lib/test_kasan_module.c
@@ -35,6 +35,8 @@ static noinline void __init copy_user_test(void)
return;
}

+ OPTIMIZER_HIDE_VAR(size);
+
pr_info("out-of-bounds in copy_from_user()\n");
unused = copy_from_user(kmem, usermem, size + 1);

--
2.30.2