linux-kernel - Re: [GIT PULL] hardening fixes for v6.6-rc3

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: <202309222034.F2B777F55@keescook>
Date:   Fri, 22 Sep 2023 20:49:21 -0700
From:   Kees Cook <keescook@...omium.org>
To:     Linus Torvalds <torvalds@...ux-foundation.org>
Cc:     linux-kernel@...r.kernel.org,
        Alexey Dobriyan <adobriyan@...il.com>,
        linux-hardening@...r.kernel.org
Subject: Re: [GIT PULL] hardening fixes for v6.6-rc3

On Fri, Sep 22, 2023 at 04:55:45PM -0700, Linus Torvalds wrote:
> On Fri, 22 Sept 2023 at 09:59, Kees Cook <keescook@...omium.org> wrote:
> >
> > - Fix UAPI stddef.h to avoid C++-ism (Alexey Dobriyan)
> 
> Ugh. Did we really have to make two different versions of that define?
> 
> Ok, so C++ did something stupid wrt an empty struct. Fine.
> 
> But I think we could have still shared the same definition by just
> using the same 'zero-sized array' trick, regardless of any 'empty
> struct has a size in C++'.
> 
> IOW, wouldn't this just work universally, without any "two completely
> different versions" hack?
> 
> #define __DECLARE_FLEX_ARRAY(TYPE, NAME)        \
>         struct { \
>                 char __empty_ ## NAME[0]; \
>                 TYPE NAME[]; \
>         }
> 
> I didn't test. I'm just hating on that '#ifdef __cplusplus'.

Yeah, I had same thought[1], but in the end I left it the way Alexey
suggested for one decent reason, and one weak reason:

1) As discovered[2] while porting this helper to ACPICA, using a flexible
   array in a struct like this does not fly with MSVC, so for MSVC
   ingesting UAPI, having the separate struct is likely more robust.

2) __cplusplus is relatively common in UAPI headers already:
   $ git grep __cplusplus -- include/uapi | wc -l
   58

-Kees

[1] https://lore.kernel.org/all/202309151208.C99747375@keescook/
[2] https://github.com/acpica/acpica/pull/837

-- 
Kees Cook