linux-kernel - Re: [PATCH] lkdtm: fix memory copy size for WRITE

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <CAKwvOd=b+ku7cd24KTYpNecYAUHxR5aBmrC1q+BL67+1YBHjzA@mail.gmail.com>
Date:   Mon, 25 Jan 2021 13:16:01 -0800
From:   Nick Desaulniers <ndesaulniers@...gle.com>
To:     Candle Sun <candlesea@...il.com>
Cc:     Kees Cook <keescook@...omium.org>, Arnd Bergmann <arnd@...db.de>,
        Greg KH <gregkh@...uxfoundation.org>,
        Nathan Chancellor <natechancellor@...il.com>,
        candle.sun@...soc.com, LKML <linux-kernel@...r.kernel.org>,
        clang-built-linux <clang-built-linux@...glegroups.com>
Subject: Re: [PATCH] lkdtm: fix memory copy size for WRITE_KERN

On Mon, Jan 25, 2021 at 12:56 AM Candle Sun <candlesea@...il.com> wrote:
>
> From: Candle Sun <candle.sun@...soc.com>
>
> Though do_overwritten() follows do_nothing() in source code, the final
> memory address order is determined by compiler. We can't always assume
> address of do_overwritten() is bigger than do_nothing(). At least the
> Clang we are using places do_overwritten() before do_nothing() in the
> object. This causes the copy size in lkdtm_WRITE_KERN() is *really*
> big and WRITE_KERN test on ARM32 arch will fail.
>
> Compare the address order before doing the subtraction.
>
> Signed-off-by: Candle Sun <candle.sun@...soc.com>
> ---
>  drivers/misc/lkdtm/perms.c | 19 +++++++++----------
>  1 file changed, 9 insertions(+), 10 deletions(-)
>
> diff --git a/drivers/misc/lkdtm/perms.c b/drivers/misc/lkdtm/perms.c
> index 2dede2ef658f..fbfbdf89d668 100644
> --- a/drivers/misc/lkdtm/perms.c
> +++ b/drivers/misc/lkdtm/perms.c
> @@ -31,13 +31,13 @@ static unsigned long ro_after_init __ro_after_init = 0x55AA5500;
>   * This just returns to the caller. It is designed to be copied into
>   * non-executable memory regions.
>   */
> -static void do_nothing(void)
> +static noinline void do_nothing(void)
>  {
>         return;
>  }
>
>  /* Must immediately follow do_nothing for size calculuations to work out. */
> -static void do_overwritten(void)
> +static noinline void do_overwritten(void)
>  {
>         pr_info("do_overwritten wasn't overwritten!\n");
>         return;
> @@ -110,15 +110,14 @@ void lkdtm_WRITE_RO_AFTER_INIT(void)
>
>  void lkdtm_WRITE_KERN(void)
>  {
> -       size_t size;
> -       volatile unsigned char *ptr;
> +       unsigned long value_dow = (unsigned long)do_overwritten;
> +       unsigned long value_do =  (unsigned long)do_nothing;
> +       size_t size = (size_t)(value_dow > value_do ?
> +                       value_dow - value_do : value_do - value_dow);
>
> -       size = (unsigned long)do_overwritten - (unsigned long)do_nothing;

Thanks for the patch! Might it be simpler to do:

size = abs((uintptr_t)do_overwritten - (uintptr_t)do_nothing));

Then I don't think much of this function needs to be changed.

> -       ptr = (unsigned char *)do_overwritten;
> -
> -       pr_info("attempting bad %zu byte write at %px\n", size, ptr);
> -       memcpy((void *)ptr, (unsigned char *)do_nothing, size);
> -       flush_icache_range((unsigned long)ptr, (unsigned long)(ptr + size));
> +       pr_info("attempting bad %zu byte write at %px\n", size, do_overwritten);
> +       memcpy((void *)value_dow, (void *)value_do, size);
> +       flush_icache_range(value_dow, value_dow + (unsigned long)size);
>         pr_err("FAIL: survived bad write\n");
>
>         do_overwritten();
> --
> 2.17.0
>


-- 
Thanks,
~Nick Desaulniers