linux-kernel - Re: [PATCH] linux/bits.h: Fix compilation error with GENMASK

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: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CAHp75Vd+YL2uhn5aHiWbvNu3DeT9OSa7wc+iY8uDnRjdTPc_uA@mail.gmail.com>
Date:   Fri, 21 May 2021 13:15:16 +0300
From:   Andy Shevchenko <andy.shevchenko@...il.com>
To:     Rikard Falkeborn <rikard.falkeborn@...il.com>
Cc:     Andrew Morton <akpm@...ux-foundation.org>,
        Andy Shevchenko <andriy.shevchenko@...ux.intel.com>,
        Tetsuo Handa <penguin-kernel@...ove.sakura.ne.jp>,
        Yury Norov <yury.norov@...il.com>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        Ard Biesheuvel <ardb@...nel.org>,
        Peter Zijlstra <peterz@...radead.org>
Subject: Re: [PATCH] linux/bits.h: Fix compilation error with GENMASK

On Tue, May 11, 2021 at 11:41 PM Rikard Falkeborn
<rikard.falkeborn@...il.com> wrote:
>
> GENMASK() has an input check which uses __builtin_choose_expr() to enable
> a compile time sanity check of its inputs if they are known at compile
> time. However, it turns out that __builtin_constant_p() does not always
> return a compile time constant [0]. It was thought this problem was fixed
> with gcc 4.9 [1], but apparently this is not the case [2].
>
> Switch to use __is_constexpr() instead which always returns a compile
> time constant, regardless of its inputs.
>
> [0]: https://lore.kernel.org/lkml/42b4342b-aefc-a16a-0d43-9f9c0d63ba7a@rasmusvillemoes.dk
> [1]: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=19449
> [2]: https://lore.kernel.org/lkml/1ac7bbc2-45d9-26ed-0b33-bf382b8d858b@I-love.SAKURA.ne.jp

Thanks for fixing this issue!
Since there is a consensus about the place, I have no objection either.
Reviewed-by: Andy Shevchenko <andy.shevchenko@...il.com>

> Reported-by: Tetsuo Handa <penguin-kernel@...ove.SAKURA.ne.jp>
> Signed-off-by: Rikard Falkeborn <rikard.falkeborn@...il.com>
> ---
> Feedback on placing __is_constexpr() in const.h is welcome, at least the
> name is appropriate...
>
>  include/linux/bits.h        |  2 +-
>  include/linux/const.h       |  8 ++++++++
>  include/linux/minmax.h      | 10 ++--------
>  tools/include/linux/bits.h  |  2 +-
>  tools/include/linux/const.h |  8 ++++++++
>  5 files changed, 20 insertions(+), 10 deletions(-)
>
> diff --git a/include/linux/bits.h b/include/linux/bits.h
> index 7f475d59a097..87d112650dfb 100644
> --- a/include/linux/bits.h
> +++ b/include/linux/bits.h
> @@ -22,7 +22,7 @@
>  #include <linux/build_bug.h>
>  #define GENMASK_INPUT_CHECK(h, l) \
>         (BUILD_BUG_ON_ZERO(__builtin_choose_expr( \
> -               __builtin_constant_p((l) > (h)), (l) > (h), 0)))
> +               __is_constexpr((l) > (h)), (l) > (h), 0)))
>  #else
>  /*
>   * BUILD_BUG_ON_ZERO is not available in h files included from asm files,
> diff --git a/include/linux/const.h b/include/linux/const.h
> index 81b8aae5a855..435ddd72d2c4 100644
> --- a/include/linux/const.h
> +++ b/include/linux/const.h
> @@ -3,4 +3,12 @@
>
>  #include <vdso/const.h>
>
> +/*
> + * This returns a constant expression while determining if an argument is
> + * a constant expression, most importantly without evaluating the argument.
> + * Glory to Martin Uecker <Martin.Uecker@....uni-goettingen.de>
> + */
> +#define __is_constexpr(x) \
> +       (sizeof(int) == sizeof(*(8 ? ((void *)((long)(x) * 0l)) : (int *)8)))
> +
>  #endif /* _LINUX_CONST_H */
> diff --git a/include/linux/minmax.h b/include/linux/minmax.h
> index c0f57b0c64d9..5433c08fcc68 100644
> --- a/include/linux/minmax.h
> +++ b/include/linux/minmax.h
> @@ -2,6 +2,8 @@
>  #ifndef _LINUX_MINMAX_H
>  #define _LINUX_MINMAX_H
>
> +#include <linux/const.h>
> +
>  /*
>   * min()/max()/clamp() macros must accomplish three things:
>   *
> @@ -17,14 +19,6 @@
>  #define __typecheck(x, y) \
>         (!!(sizeof((typeof(x) *)1 == (typeof(y) *)1)))
>
> -/*
> - * This returns a constant expression while determining if an argument is
> - * a constant expression, most importantly without evaluating the argument.
> - * Glory to Martin Uecker <Martin.Uecker@....uni-goettingen.de>
> - */
> -#define __is_constexpr(x) \
> -       (sizeof(int) == sizeof(*(8 ? ((void *)((long)(x) * 0l)) : (int *)8)))
> -
>  #define __no_side_effects(x, y) \
>                 (__is_constexpr(x) && __is_constexpr(y))
>
> diff --git a/tools/include/linux/bits.h b/tools/include/linux/bits.h
> index 7f475d59a097..87d112650dfb 100644
> --- a/tools/include/linux/bits.h
> +++ b/tools/include/linux/bits.h
> @@ -22,7 +22,7 @@
>  #include <linux/build_bug.h>
>  #define GENMASK_INPUT_CHECK(h, l) \
>         (BUILD_BUG_ON_ZERO(__builtin_choose_expr( \
> -               __builtin_constant_p((l) > (h)), (l) > (h), 0)))
> +               __is_constexpr((l) > (h)), (l) > (h), 0)))
>  #else
>  /*
>   * BUILD_BUG_ON_ZERO is not available in h files included from asm files,
> diff --git a/tools/include/linux/const.h b/tools/include/linux/const.h
> index 81b8aae5a855..435ddd72d2c4 100644
> --- a/tools/include/linux/const.h
> +++ b/tools/include/linux/const.h
> @@ -3,4 +3,12 @@
>
>  #include <vdso/const.h>
>
> +/*
> + * This returns a constant expression while determining if an argument is
> + * a constant expression, most importantly without evaluating the argument.
> + * Glory to Martin Uecker <Martin.Uecker@....uni-goettingen.de>
> + */
> +#define __is_constexpr(x) \
> +       (sizeof(int) == sizeof(*(8 ? ((void *)((long)(x) * 0l)) : (int *)8)))
> +
>  #endif /* _LINUX_CONST_H */
> --
> 2.31.1
>


-- 
With Best Regards,
Andy Shevchenko