Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Mon, 25 Jul 2022 11:13:23 -0500
From:   Segher Boessenkool <segher@kernel.crashing.org>
To:     Michael Ellerman <mpe@ellerman.id.au>
Cc:     Michal =?iso-8859-1?Q?Such=E1nek?= <msuchanek@suse.de>,
        lkp@intel.com, linux-kernel@vger.kernel.org, npiggin@gmail.com,
        Julia.Lawall@inria.fr, paulus@samba.org,
        Jason Wang <wangborong@cdjrlc.com>,
        linuxppc-dev@lists.ozlabs.org, hbathini@linux.ibm.com,
        joel@jms.id.au
Subject: Re: [PATCH] powerpc: Remove the static variable initialisations to 0
Message-ID: <20220725161323.GT25951@gate.crashing.org>
References: <20220723092436.7149-1-wangborong@cdjrlc.com> <20220723133405.GF17705@kitsune.suse.cz> <20220723192211.GP25951@gate.crashing.org> <87fsipdg7b.fsf@mpe.ellerman.id.au>
Mime-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
In-Reply-To: <87fsipdg7b.fsf@mpe.ellerman.id.au>
User-Agent: Mutt/1.4.2.3i
Precedence: bulk

On Mon, Jul 25, 2022 at 01:27:52PM +1000, Michael Ellerman wrote:
> Segher Boessenkool <segher@kernel.crashing.org> writes:
> > On Sat, Jul 23, 2022 at 03:34:05PM +0200, Michal Such?nek wrote:
> >> Hello,
> >> 
> >> On Sat, Jul 23, 2022 at 05:24:36PM +0800, Jason Wang wrote:
> >> > Initialise global and static variable to 0 is always unnecessary.
> >> > Remove the unnecessary initialisations.
> >> 
> >> Isn't this change also unnecessary?
> >> 
> >> Initializing to 0 does not affect correctness, or even any kind of
> >> semantics in any way.
> >
> > It did make a difference when the kernel was still compiled with
> > -fcommon (which used to be the GCC default on most configurations, it is
> > traditional on Unix).  No explicit initialiser puts an object in .bss if
> > you use -fcommon.  This matters a bit for data layout.
> 
> The kernel has built with -fno-common since ~2002.

2001, yes (255649c18287).  And before that it was important to
initialise everything with static storage duration explicitly in the
source code.  It was part of the collective memory, I wondered if this
patch originated that way?

> I think the belief is that an explicit initialiser of 0 forces the
> variable into .data, but AFAICS that is not true with any compiler we
> support.

Exactly, you get identical code either way, if you use -fno-common.
People will still see this difference if they use a compiler before
GCC 10 for compiling most other things though.


Segher