Received-SPF: pass (google.com: domain of linux-kernel+bounces-182583-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:45d1:ec00::1 as permitted sender) client-ip=2604:1380:45d1:ec00::1;
Precedence: bulk
MIME-Version: 1.0
References: <20240517130118.759301-1-andrey.konovalov@linux.dev>
In-Reply-To: <20240517130118.759301-1-andrey.konovalov@linux.dev>
From: Nico Pache <npache@redhat.com>
Date: Fri, 17 May 2024 13:50:10 -0600
Message-ID: <CAA1CXcAdG=OFkBzjPqr7M_kC7VZUdj-+vH_2W4UidfbQwfQbeA@mail.gmail.com>
Subject: Re: [PATCH] kasan, fortify: properly rename memintrinsics
To: andrey.konovalov@linux.dev
Cc: Marco Elver <elver@google.com>, Andrey Konovalov <andreyknvl@gmail.com>, 
	Alexander Potapenko <glider@google.com>, Dmitry Vyukov <dvyukov@google.com>, 
	Andrey Ryabinin <ryabinin.a.a@gmail.com>, kasan-dev@googlegroups.com, 
	Andrew Morton <akpm@linux-foundation.org>, linux-mm@kvack.org, 
	Erhard Furtner <erhard_f@mailbox.org>, Daniel Axtens <dja@axtens.net>, linux-kernel@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

On Fri, May 17, 2024 at 7:02=E2=80=AFAM <andrey.konovalov@linux.dev> wrote:
>
> From: Andrey Konovalov <andreyknvl@gmail.com>
>
> After commit 69d4c0d32186 ("entry, kasan, x86: Disallow overriding mem*()
> functions") and the follow-up fixes, with CONFIG_FORTIFY_SOURCE enabled,
> even though the compiler instruments meminstrinsics by generating calls
> to __asan/__hwasan_ prefixed functions, FORTIFY_SOURCE still uses
> uninstrumented memset/memmove/memcpy as the underlying functions.
>
> As a result, KASAN cannot detect bad accesses in memset/memmove/memcpy.
> This also makes KASAN tests corrupt kernel memory and cause crashes.
>
> To fix this, use __asan_/__hwasan_memset/memmove/memcpy as the underlying
> functions whenever appropriate. Do this only for the instrumented code
> (as indicated by __SANITIZE_ADDRESS__).
>
> Reported-by: Erhard Furtner <erhard_f@mailbox.org>
> Reported-by: Nico Pache <npache@redhat.com>
> Closes: https://lore.kernel.org/all/20240501144156.17e65021@outsider.home=
/
> Fixes: 69d4c0d32186 ("entry, kasan, x86: Disallow overriding mem*() funct=
ions")
> Fixes: 51287dcb00cc ("kasan: emit different calls for instrumentable memi=
ntrinsics")
> Fixes: 36be5cba99f6 ("kasan: treat meminstrinsic as builtins in uninstrum=
ented files")
> Signed-off-by: Andrey Konovalov <andreyknvl@gmail.com>

Thank you for fixing this !! The test no longer panics :)

Now that the test progresses I also see rcu_uaf failing.
    # rcu_uaf: EXPECTATION FAILED at mm/kasan/kasan_test.c:870
    KASAN failure expected in "call_rcu(&global_rcu_ptr->rcu,
rcu_uaf_reclaim); rcu_barrier()", but none occurred
    not ok 31 rcu_uaf
 I can open a new thread for that if you'd like.

Tested-by: Nico Pache <npache@redhat.com>
Acked-by: Nico Pache <npache@redhat.com>

> ---
>  include/linux/fortify-string.h | 22 ++++++++++++++++++----
>  1 file changed, 18 insertions(+), 4 deletions(-)
>
> diff --git a/include/linux/fortify-string.h b/include/linux/fortify-strin=
g.h
> index 85fc0e6f0f7f..bac010cfc42f 100644
> --- a/include/linux/fortify-string.h
> +++ b/include/linux/fortify-string.h
> @@ -75,17 +75,30 @@ void __write_overflow_field(size_t avail, size_t want=
ed) __compiletime_warning("
>         __ret;                                                  \
>  })
>
> -#if defined(CONFIG_KASAN_GENERIC) || defined(CONFIG_KASAN_SW_TAGS)
> +#if defined(__SANITIZE_ADDRESS__)
> +
> +#if !defined(CONFIG_CC_HAS_KASAN_MEMINTRINSIC_PREFIX) && !defined(CONFIG=
_GENERIC_ENTRY)
> +extern void *__underlying_memset(void *p, int c, __kernel_size_t size) _=
_RENAME(memset);
> +extern void *__underlying_memmove(void *p, const void *q, __kernel_size_=
t size) __RENAME(memmove);
> +extern void *__underlying_memcpy(void *p, const void *q, __kernel_size_t=
 size) __RENAME(memcpy);
> +#elif defined(CONFIG_KASAN_GENERIC)
> +extern void *__underlying_memset(void *p, int c, __kernel_size_t size) _=
_RENAME(__asan_memset);
> +extern void *__underlying_memmove(void *p, const void *q, __kernel_size_=
t size) __RENAME(__asan_memmove);
> +extern void *__underlying_memcpy(void *p, const void *q, __kernel_size_t=
 size) __RENAME(__asan_memcpy);
> +#else /* CONFIG_KASAN_SW_TAGS */
> +extern void *__underlying_memset(void *p, int c, __kernel_size_t size) _=
_RENAME(__hwasan_memset);
> +extern void *__underlying_memmove(void *p, const void *q, __kernel_size_=
t size) __RENAME(__hwasan_memmove);
> +extern void *__underlying_memcpy(void *p, const void *q, __kernel_size_t=
 size) __RENAME(__hwasan_memcpy);
> +#endif
> +
>  extern void *__underlying_memchr(const void *p, int c, __kernel_size_t s=
ize) __RENAME(memchr);
>  extern int __underlying_memcmp(const void *p, const void *q, __kernel_si=
ze_t size) __RENAME(memcmp);
> -extern void *__underlying_memcpy(void *p, const void *q, __kernel_size_t=
 size) __RENAME(memcpy);
> -extern void *__underlying_memmove(void *p, const void *q, __kernel_size_=
t size) __RENAME(memmove);
> -extern void *__underlying_memset(void *p, int c, __kernel_size_t size) _=
_RENAME(memset);
>  extern char *__underlying_strcat(char *p, const char *q) __RENAME(strcat=
);
>  extern char *__underlying_strcpy(char *p, const char *q) __RENAME(strcpy=
);
>  extern __kernel_size_t __underlying_strlen(const char *p) __RENAME(strle=
n);
>  extern char *__underlying_strncat(char *p, const char *q, __kernel_size_=
t count) __RENAME(strncat);
>  extern char *__underlying_strncpy(char *p, const char *q, __kernel_size_=
t size) __RENAME(strncpy);
> +
>  #else
>
>  #if defined(__SANITIZE_MEMORY__)
> @@ -110,6 +123,7 @@ extern char *__underlying_strncpy(char *p, const char=
 *q, __kernel_size_t size)
>  #define __underlying_strlen    __builtin_strlen
>  #define __underlying_strncat   __builtin_strncat
>  #define __underlying_strncpy   __builtin_strncpy
> +
>  #endif
>
>  /**
> --
> 2.25.1
>