Re: [PATCH v2] compiler.h: give up __compiletime_assert_fallback()

From: Kees Cook <keescook@chromium.org>
To: Greg KH <gregkh@linuxfoundation.org>
Cc: Nick Desaulniers <ndesaulniers@google.com>,
	Matthias Kaehlcke <mka@chromium.org>,
	Linus Torvalds <torvalds@linux-foundation.org>,
	Masahiro Yamada <yamada.masahiro@socionext.com>,
	Christopher Li <sparse@chrisli.org>,
	Sparse Mailing-list <linux-sparse@vger.kernel.org>,
	LKML <linux-kernel@vger.kernel.org>,
	Daniel Santos <daniel.santos@pobox.com>,
	Chris Wilson <chris@chris-wilson.co.uk>,
	Jani Nikula <jani.nikula@intel.com>
Subject: Re: [PATCH v2] compiler.h: give up __compiletime_assert_fallback()
Date: Wed, 26 Sep 2018 11:26:46 -0700	[thread overview]
Message-ID: <CAGXu5jLBz1YanL5X4HFZjhb8NZe5QK=Q1ySfBkBDNWqEg4hs5g@mail.gmail.com> (raw)
In-Reply-To: <CAKwvOd=xYG+r4KU+QurUTyaxkPB2AMane4YfWxuvKtONvVtLxQ@mail.gmail.com>

On Wed, Sep 26, 2018 at 11:03 AM, Nick Desaulniers
<ndesaulniers@google.com> wrote:
> On Wed, Sep 26, 2018 at 11:00 AM Matthias Kaehlcke <mka@chromium.org> wrote:
>>
>> On Fri, Aug 31, 2018 at 09:46:02AM -0700, Nick Desaulniers wrote:
>> > On Tue, Aug 28, 2018 at 4:00 PM Nick Desaulniers
>> > <ndesaulniers@google.com> wrote:
>> > >
>> > > On Mon, Aug 27, 2018 at 1:42 PM Daniel Santos <daniel.santos@pobox.com> wrote:
>> > > >
>> > > > Hello Nick,
>> > > >
>> > > > On 08/27/2018 03:09 PM, Nick Desaulniers wrote:
>> > > > >>> Let's give up __compiletime_assert_fallback().  This commit does not
>> > > > >>> change the current behavior since it just rips off the useless code.
>> > > > >> Clang is not the only target audience of
>> > > > >> __compiletime_assert_fallback().  Instead of ripping out something that
>> > > > >> may benefit builds with gcc 4.2 and earlier, why not override its
>> > > > > Note that with commit cafa0010cd51 ("Raise the minimum required gcc
>> > > > > version to 4.6") that gcc < 4.6 is irrelevant.
>> > > >
>> > > > Ah, I guess I'm not keeping up, that's wonderful news!  Considering that
>> > > > I guess I would be OK with its removal, but I still think it would be
>> > > > better if a similar mechanism to break the Clang build could be found.
>> > >
>> > > I'm consulting with our best language lawyers to see what combinations
>> > > of _Static_assert and __builtin_constant_p would do the trick.
>> >
>> > Linus,
>> > Can this patch be merged in the meantime?
>>
>> friendly ping :)
>>
>> With c5c2b11894f4 ("drm/i915: Warn against variable length arrays")
>> clang raises plenty of vla warnings about
>> __compiletime_error_fallback() in the i915 driver. Would be great to
>> get rid of those without having to revert that commit.
>
> I've been meaning to follow up on this, thanks Matthias.  I too would
> really like this patch.

Adding Greg to the thread. Between Masahiro's detailed commit log and
the Clang-familiar reviewers, I think this should land for 4.19 (as
part of the other Clang-sanity patches that are already in 4.19). This
has no impact on gcc now that we're requiring 4.6+.

https://lore.kernel.org/patchwork/patch/977668/

-Kees

-- 
Kees Cook
Pixel Security