Re: [PATCH 2/2] lib/test_kasan: Add stack overflow test

[Mark Rutland]

On Fri, Jul 19, 2019 at 03:28:18PM +0200, Marco Elver wrote:
> Adds a simple stack overflow test, to check the error being reported on
> an overflow. Without CONFIG_STACK_GUARD_PAGE, the result is typically
> some seemingly unrelated KASAN error message due to accessing random
> other memory.

Can't we use the LKDTM_EXHAUST_STACK case to check this?

I was also under the impression that the other KASAN self-tests weren't
fatal, and IIUC this will kill the kernel.

Given that, and given this is testing non-KASAN functionality, I'm not
sure it makes sense to bundle this with the KASAN tests.

Thanks,
Mark.

> 
> Signed-off-by: Marco Elver <elver@xxxxxxxxxx>
> Cc: Thomas Gleixner <tglx@xxxxxxxxxxxxx>
> Cc: Ingo Molnar <mingo@xxxxxxxxxx>
> Cc: Borislav Petkov <bp@xxxxxxxxx>
> Cc: "H. Peter Anvin" <hpa@xxxxxxxxx>
> Cc: Andrey Ryabinin <aryabinin@xxxxxxxxxxxxx>
> Cc: Alexander Potapenko <glider@xxxxxxxxxx>
> Cc: Dmitry Vyukov <dvyukov@xxxxxxxxxx>
> Cc: Andrey Konovalov <andreyknvl@xxxxxxxxxx>
> Cc: Mark Rutland <mark.rutland@xxxxxxx>
> Cc: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
> Cc: x86@xxxxxxxxxx
> Cc: linux-kernel@xxxxxxxxxxxxxxx
> Cc: kasan-dev@xxxxxxxxxxxxxxxx
> ---
>  lib/test_kasan.c | 36 ++++++++++++++++++++++++++++++++++++
>  1 file changed, 36 insertions(+)
> 
> diff --git a/lib/test_kasan.c b/lib/test_kasan.c
> index b63b367a94e8..3092ec01189d 100644
> --- a/lib/test_kasan.c
> +++ b/lib/test_kasan.c
> @@ -15,6 +15,7 @@
>  #include <linux/mman.h>
>  #include <linux/module.h>
>  #include <linux/printk.h>
> +#include <linux/sched/task_stack.h>
>  #include <linux/slab.h>
>  #include <linux/string.h>
>  #include <linux/uaccess.h>
> @@ -709,6 +710,32 @@ static noinline void __init kmalloc_double_kzfree(void)
>  	kzfree(ptr);
>  }
>  
> +#ifdef CONFIG_STACK_GUARD_PAGE
> +static noinline void __init stack_overflow_via_recursion(void)
> +{
> +	volatile int n = 512;
> +
> +	BUILD_BUG_ON(IS_ENABLED(CONFIG_STACK_GROWSUP));
> +
> +	/* About to overflow: overflow via alloca'd array and try to write. */
> +	if (!object_is_on_stack((void *)&n - n)) {
> +		volatile char overflow[n];
> +
> +		overflow[0] = overflow[0];
> +		return;
> +	}
> +
> +	stack_overflow_via_recursion();
> +}
> +
> +static noinline void __init kasan_stack_overflow(void)
> +{
> +	pr_info("stack overflow begin\n");
> +	stack_overflow_via_recursion();
> +	pr_info("stack overflow end\n");
> +}
> +#endif
> +
>  static int __init kmalloc_tests_init(void)
>  {
>  	/*
> @@ -753,6 +780,15 @@ static int __init kmalloc_tests_init(void)
>  	kasan_bitops();
>  	kmalloc_double_kzfree();
>  
> +#ifdef CONFIG_STACK_GUARD_PAGE
> +	/*
> +	 * Only test with CONFIG_STACK_GUARD_PAGE, as without we get other
> +	 * random KASAN violations, due to accessing other random memory (we
> +	 * want to avoid actually corrupting memory in these tests).
> +	 */
> +	kasan_stack_overflow();
> +#endif
> +
>  	kasan_restore_multi_shot(multishot);
>  
>  	return -EAGAIN;
> -- 
> 2.22.0.657.g960e92d24f-goog
>