Re: [RFC PATCH v3 07/20] lkdtm: Emit an indirect call for CFI tests

[Nick Desaulniers <ndesaulniers@google.com>]

On Fri, Jun 10, 2022 at 4:35 PM Sami Tolvanen <samitolvanen@google.com> wrote:
>
> Clang can convert the indirect calls in lkdtm_CFI_FORWARD_PROTO into
> direct calls. Move the call into a noinline function that accepts the
> target address as an argument to ensure the compiler actually emits an
> indirect call instead.

Thanks for the patch!
Reviewed-by: Nick Desaulniers <ndesaulniers@google.com>

>
> Signed-off-by: Sami Tolvanen <samitolvanen@google.com>
> ---
>  drivers/misc/lkdtm/cfi.c | 15 +++++++++------
>  1 file changed, 9 insertions(+), 6 deletions(-)
>
> diff --git a/drivers/misc/lkdtm/cfi.c b/drivers/misc/lkdtm/cfi.c
> index 666a7f4bc137..b6b375112a2a 100644
> --- a/drivers/misc/lkdtm/cfi.c
> +++ b/drivers/misc/lkdtm/cfi.c
> @@ -20,6 +20,13 @@ static noinline int lkdtm_increment_int(int *counter)
>
>         return *counter;
>  }
> +
> +/* Don't allow the compiler to inline the calls. */
> +static noinline void lkdtm_indirect_call(void (*func)(int *))
> +{
> +       func(&called_count);
> +}
> +
>  /*
>   * This tries to call an indirect function with a mismatched prototype.
>   */
> @@ -29,15 +36,11 @@ static void lkdtm_CFI_FORWARD_PROTO(void)
>          * Matches lkdtm_increment_void()'s prototype, but not
>          * lkdtm_increment_int()'s prototype.
>          */
> -       void (*func)(int *);
> -
>         pr_info("Calling matched prototype ...\n");
> -       func = lkdtm_increment_void;
> -       func(&called_count);
> +       lkdtm_indirect_call(lkdtm_increment_void);
>
>         pr_info("Calling mismatched prototype ...\n");
> -       func = (void *)lkdtm_increment_int;
> -       func(&called_count);
> +       lkdtm_indirect_call((void *)lkdtm_increment_int);
>
>         pr_err("FAIL: survived mismatched prototype function call!\n");
>         pr_expected_config(CONFIG_CFI_CLANG);
> --
> 2.36.1.476.g0c4daa206d-goog
>


-- 
Thanks,
~Nick Desaulniers