Received-SPF: pass (google.com: best guess record for domain of linux-crypto-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
MIME-Version: 1.0
References: <AM6PR09MB3523ADF4617CB97D59904616D21E0@AM6PR09MB3523.eurprd09.prod.outlook.com>
 <CAKv+Gu8bReGWAUm4GrCg7kefVR7U0Z8XBt_GVV4WEvgOpCtjpA@mail.gmail.com>
 <AM6PR09MB3523B77DE66DD5353F08A687D21E0@AM6PR09MB3523.eurprd09.prod.outlook.com>
 <20190529180731.Horde.NGHeOXuCgw23pVdGqjc0fw9@messagerie.si.c-s.fr> <AM6PR09MB35232561AF362BF5A9FE72FFD2180@AM6PR09MB3523.eurprd09.prod.outlook.com>
In-Reply-To: <AM6PR09MB35232561AF362BF5A9FE72FFD2180@AM6PR09MB3523.eurprd09.prod.outlook.com>
From:   Ard Biesheuvel <ard.biesheuvel@linaro.org>
Date:   Thu, 30 May 2019 12:25:06 +0200
Message-ID: <CAKv+Gu_DRDwQLuajk9TZ14vTce_1NhyU86dhtvY=-kQuvrP0AA@mail.gmail.com>
Subject: Re: Conding style question regarding configuration
To:     Pascal Van Leeuwen <pvanleeuwen@insidesecure.com>
Cc:     Christophe Leroy <christophe.leroy@c-s.fr>,
        "linux-crypto@vger.kernel.org" <linux-crypto@vger.kernel.org>
Content-Type: text/plain; charset="UTF-8"
Sender: linux-crypto-owner@vger.kernel.org
Precedence: bulk

On Thu, 30 May 2019 at 12:16, Pascal Van Leeuwen
<pvanleeuwen@insidesecure.com> wrote:
>
> > >> Yes. Code and data with static linkage will just be optimized away by
> > >> the compiler if the CONFIG_xx option is not enabled, so all you need
> > >> to guard are the actual statements, function calls etc.
> > >>
> > > Ok, makes sense. Then I'll just config out the relevant function bodies
> > > and assume the compiler will do the rest ...
> > >
> >
> > No need to config out function bodies when they are static.
> >
> Well, I got a complaint from someone that my driver updates for adding PCIE
> support wouldn't  compile properly on a platform without a PCI(E) subsystem.
> So I figure I do have to config out the references to PCI specific function
> calls to fix that.
>
> Or are you just referring to bodies of static subfunctions that are no
> longer being called? Would the compiler skip those entirely?
>

The idea is that, by doing something like

static int bar;

static void foo(void)
{
    bar = 1;
}

if (IS_ENABLED(CONFIG_FOO))
    foo();

the function foo() or the variable bar don't have to be decorated with
#ifdefs or anything. The compiler will not complain that they are
unused if CONFIG_FOO is not enabled, and the contents of foo() are
always visible to the compiler, and so any programming errors will be
caught regardless of whether CONFIG_FOO is set.