Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
From:   Michael Ellerman <mpe@ellerman.id.au>
To:     Kees Cook <keescook@chromium.org>,
        Nick Desaulniers <ndesaulniers@google.com>
Cc:     Andrew Morton <akpm@linux-foundation.org>,
        Linus Torvalds <torvalds@linux-foundation.org>,
        Rasmus Villemoes <linux@rasmusvillemoes.dk>,
        Nathan Chancellor <nathan@kernel.org>,
        Masahiro Yamada <masahiroy@kernel.org>,
        Joe Perches <joe@perches.com>, Arnd Bergmann <arnd@kernel.org>,
        Stephen Rothwell <sfr@canb.auug.org.au>, llvm@lists.linux.dev,
        linux-kernel@vger.kernel.org
Subject: Re: [PATCH 00/10] raise minimum GCC version to 5.1
In-Reply-To: <202109101917.5BA95B87E0@keescook>
References: <20210910234047.1019925-1-ndesaulniers@google.com>
 <202109101917.5BA95B87E0@keescook>
Date:   Sat, 11 Sep 2021 20:42:49 +1000
Message-ID: <8735qbmlzq.fsf@mpe.ellerman.id.au>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

Kees Cook <keescook@chromium.org> writes:
> On Fri, Sep 10, 2021 at 04:40:37PM -0700, Nick Desaulniers wrote:
>> commit fad7cd3310db ("nbd: add the check to prevent overflow in
>> __nbd_ioctl()")
>>=20
>> raised an issue from the fallback helpers added in
>>=20
>> commit f0907827a8a9 ("compiler.h: enable builtin overflow checkers and a=
dd fallback code")
>>=20
>> Specifically, the helpers for checking whether the results of a
>> multiplication overflowed (__unsigned_mul_overflow,
>> __signed_add_overflow) use the division operator when
>> !COMPILER_HAS_GENERIC_BUILTIN_OVERFLOW. This is problematic for 64b
>> operands on 32b hosts.
>>=20
>> Also, because the macro is type agnostic, it is very difficult to write
>> a similarly type generic macro that dispatches to one of:
>> * div64_s64
>> * div64_u64
>> * div_s64
>> * div_u64
>
> Given that it's all compile-time type-aware goo, this isn't so bad. The
> gist[1] you linked off the bug report is pretty close. Needs some
> bikeshedding. ;)
>
>> Raising the minimum supported versions allows us to remove all of the
>> fallback helpers for !COMPILER_HAS_GENERIC_BUILTIN_OVERFLOW, instead
>> dispatching the compiler builtins.
>>=20
>> arm64 has already raised the minimum supported GCC version to 5.1, do
>> this for all targets now. See the link below for the previous
>> discussion.
>
> That said, I'd much prefer raising the minimum GCC -- no one appears
> to actually be building on 4.9 -- there are close to 200 errors (ne=C3=A9
> warnings) on x86_64 allmodconfig there currently.

I still do 4.9 builds on kisskb, but I agree there are a lot of
warnings, and no one ever has time to fix any.

cheers