[PATCH 0/2] kasan: a couple of test fixes

[PATCH 0/2] kasan: a couple of test fixes

[Andrey Konovalov]

The first one fixes the invalid-free test crashing the kernel, and the
second one fixes the memset tests working incorrectly due to compiler
optimizations.

Andrey Konovalov (2):
  kasan: fix invalid-free test crashing the kernel
  kasan: disallow compiler to optimize away memset in tests

 lib/Makefile     | 1 +
 lib/test_kasan.c | 8 ++++++++
 2 files changed, 9 insertions(+)

-- 
2.16.2.395.g2e18187dfd-goog

[PATCH 1/2] kasan: fix invalid-free test crashing the kernel

[Andrey Konovalov]

When an invalid-free is triggered by one of the KASAN tests, the object
doesn't actually get freed. This later leads to a BUG failure in
kmem_cache_destroy that checks that there are no allocated objects in the
cache that is being destroyed. Fix this by calling kmem_cache_free with
the proper object address after the call that triggers invalid-free.

Signed-off-by: Andrey Konovalov <andreyknvl@google.com>
---
 lib/test_kasan.c | 8 ++++++++
 1 file changed, 8 insertions(+)

diff --git a/lib/test_kasan.c b/lib/test_kasan.c
index 98854a64b014..ec657105edbf 100644
--- a/lib/test_kasan.c
+++ b/lib/test_kasan.c
@@ -567,7 +567,15 @@ static noinline void __init kmem_cache_invalid_free(void)
 		return;
 	}
 
+	/* Trigger invalid free, the object doesn't get freed */
 	kmem_cache_free(cache, p + 1);
+
+	/*
+	 * Properly free the object to prevent the "Objects remaining in
+	 * test_cache on __kmem_cache_shutdown" BUG failure.
+	 */
+	kmem_cache_free(cache, p);
+
 	kmem_cache_destroy(cache);
 }
 
-- 
2.16.2.395.g2e18187dfd-goog

Re: [PATCH 1/2] kasan: fix invalid-free test crashing the kernel

[Andrey Konovalov]

On Thu, Mar 1, 2018 at 6:15 PM, Andrey Konovalov <andreyknvl@google.com> wrote:
> When an invalid-free is triggered by one of the KASAN tests, the object
> doesn't actually get freed. This later leads to a BUG failure in
> kmem_cache_destroy that checks that there are no allocated objects in the
> cache that is being destroyed. Fix this by calling kmem_cache_free with
> the proper object address after the call that triggers invalid-free.

Note, that for this patch to fix the issue, it is supposed to go on
top of the "kasan, slub: fix handling of kasan_slab_free hook" patch I
sent recently.

https://patchwork.kernel.org/patch/10238179/

>
> Signed-off-by: Andrey Konovalov <andreyknvl@google.com>
> ---
>  lib/test_kasan.c | 8 ++++++++
>  1 file changed, 8 insertions(+)
>
> diff --git a/lib/test_kasan.c b/lib/test_kasan.c
> index 98854a64b014..ec657105edbf 100644
> --- a/lib/test_kasan.c
> +++ b/lib/test_kasan.c
> @@ -567,7 +567,15 @@ static noinline void __init kmem_cache_invalid_free(void)
>                 return;
>         }
>
> +       /* Trigger invalid free, the object doesn't get freed */
>         kmem_cache_free(cache, p + 1);
> +
> +       /*
> +        * Properly free the object to prevent the "Objects remaining in
> +        * test_cache on __kmem_cache_shutdown" BUG failure.
> +        */
> +       kmem_cache_free(cache, p);
> +
>         kmem_cache_destroy(cache);
>  }
>
> --
> 2.16.2.395.g2e18187dfd-goog
>

Re: [PATCH 1/2] kasan: fix invalid-free test crashing the kernel

[Andrey Ryabinin]

On 03/01/2018 08:15 PM, Andrey Konovalov wrote:
> When an invalid-free is triggered by one of the KASAN tests, the object
> doesn't actually get freed. This later leads to a BUG failure in
> kmem_cache_destroy that checks that there are no allocated objects in the
> cache that is being destroyed. Fix this by calling kmem_cache_free with
> the proper object address after the call that triggers invalid-free.
> 
> Signed-off-by: Andrey Konovalov <andreyknvl@google.com>

Acked-by: Andrey Ryabinin <aryabinin@virtuozzo.com>

[PATCH 2/2] kasan: disallow compiler to optimize away memset in tests

[Andrey Konovalov]

A compiler can optimize away memset calls by replacing them with mov
instructions. There are KASAN tests, that specifically test that KASAN
correctly handles memset calls, we don't want this optimization to
happen.

The solution is to add -fno-builtin flag to test_kasan.ko

Signed-off-by: Andrey Konovalov <andreyknvl@google.com>
---
 lib/Makefile | 1 +
 1 file changed, 1 insertion(+)

diff --git a/lib/Makefile b/lib/Makefile
index a90d4fcd748f..9c940c4c0593 100644
--- a/lib/Makefile
+++ b/lib/Makefile
@@ -52,6 +52,7 @@ obj-$(CONFIG_TEST_FIRMWARE) += test_firmware.o
 obj-$(CONFIG_TEST_SYSCTL) += test_sysctl.o
 obj-$(CONFIG_TEST_HASH) += test_hash.o test_siphash.o
 obj-$(CONFIG_TEST_KASAN) += test_kasan.o
+CFLAGS_test_kasan.o += -fno-builtin
 obj-$(CONFIG_TEST_KSTRTOX) += test-kstrtox.o
 obj-$(CONFIG_TEST_LIST_SORT) += test_list_sort.o
 obj-$(CONFIG_TEST_LKM) += test_module.o
-- 
2.16.2.395.g2e18187dfd-goog

Re: [PATCH 2/2] kasan: disallow compiler to optimize away memset in tests

[Andrey Ryabinin]

On 03/01/2018 08:15 PM, Andrey Konovalov wrote:
> A compiler can optimize away memset calls by replacing them with mov
> instructions. There are KASAN tests, that specifically test that KASAN
> correctly handles memset calls, we don't want this optimization to
> happen.
> 
> The solution is to add -fno-builtin flag to test_kasan.ko
> 
> Signed-off-by: Andrey Konovalov <andreyknvl@google.com>

Acked-by: Andrey Ryabinin <aryabinin@virtuozzo.com>