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;
Date:   Mon, 5 Oct 2020 23:03:46 -0700
From:   Kees Cook <keescook@chromium.org>
To:     Nathan Chancellor <natechancellor@gmail.com>
Cc:     Andrew Morton <akpm@linux-foundation.org>,
        Linus Torvalds <torvalds@linux-foundation.org>,
        Ard Biesheuvel <ardb@kernel.org>,
        Arnd Bergmann <arnd@arndb.de>,
        Masahiro Yamada <masahiroy@kernel.org>,
        Michal Marek <michal.lkml@markovi.net>,
        Nick Desaulniers <ndesaulniers@google.com>,
        Marco Elver <elver@google.com>,
        Randy Dunlap <rdunlap@infradead.org>,
        Dmitry Vyukov <dvyukov@google.com>,
        George Popescu <georgepope@android.com>,
        Herbert Xu <herbert@gondor.apana.org.au>,
        Peter Oberparleiter <oberpar@linux.ibm.com>,
        Andrey Ryabinin <aryabinin@virtuozzo.com>,
        clang-built-linux@googlegroups.com, linux-kbuild@vger.kernel.org,
        linux-kernel@vger.kernel.org
Subject: Re: [PATCH 3/4] ubsan: Force -Wno-maybe-uninitialized only for GCC
Message-ID: <202010052301.CFBC03C72E@keescook>
References: <20201002221527.177500-1-keescook@chromium.org>
 <20201002221527.177500-4-keescook@chromium.org>
 <20201004071614.GC1650@Ryzen-9-3900X.localdomain>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20201004071614.GC1650@Ryzen-9-3900X.localdomain>
Precedence: bulk

On Sun, Oct 04, 2020 at 12:16:14AM -0700, Nathan Chancellor wrote:
> On Fri, Oct 02, 2020 at 03:15:26PM -0700, Kees Cook wrote:
> > Clang handles 'maybe-uninitialized' better in the face of using UBSAN,
> > so do not make this universally disabled for UBSAN builds.
> > 
> > Signed-off-by: Kees Cook <keescook@chromium.org>
> 
> Well this patch is not strictly necessary because Clang does not support
> -Wmaybe-uninitialized anyways :) its flags are -Wuninitialized and
> -Wsometimes-uninitialized so the warning stays enabled for UBSAN as it
> stands.

Ah, yes. Heh. Well... perhaps I can just drop this patch.

> However, something like this could still worthwhile because it would
> save us one call to cc-disable-warning (yay micro optimizations).
> 
> Maybe it just does not need to have a whole new symbol, just make it
> 
> ubsan-cflags-$(CONFIG_CC_IS_GCC)
> 
> instead of
> 
> ubsan-cflags-$(CONFIG_UBSAN)

If it gets kept, I'd still like it gated on CONFIG_UBSAN in some way
(e.g. the patch has an implicit depends due to the "if UBSAN" section).

But yes, this patch is rather a no-op.

-- 
Kees Cook