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:   Wed, 7 Oct 2020 08:10:28 +0100
From:   Lee Jones <lee.jones@linaro.org>
To:     Michael Brunner <Michael.Brunner@kontron.com>
Cc:     "mibru@gmx.de" <mibru@gmx.de>,
        "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH] mfd: kempld-core: Mark kempld-acpi_table as
 __maybe_unused
Message-ID: <20201007071028.GC1763265@dell>
References: <37c55c13f9042dde06fd67c829b06765286d0580.camel@kontron.com>
 <20201002070134.GR6148@dell>
 <b7fb20174aa7fd8a15132b6de679498fe0b18fd1.camel@kontron.com>
 <20201006065313.GW6148@dell>
 <66d77a96a488078eff1d8f0a2180941d94ce0bdb.camel@kontron.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
In-Reply-To: <66d77a96a488078eff1d8f0a2180941d94ce0bdb.camel@kontron.com>
Precedence: bulk

On Tue, 06 Oct 2020, Michael Brunner wrote:

> On Tue, 2020-10-06 at 07:53 +0100, Lee Jones wrote:
> > On Mon, 05 Oct 2020, Michael Brunner wrote:
> > 
> > > On Fri, 2020-10-02 at 08:01 +0100, Lee Jones wrote:
> > > > On Thu, 01 Oct 2020, Michael Brunner wrote:
> > > > 
> > > > > The Intel 0-DAY CI Kernel Test Service reports an unused variable
> > > > > warning when compiling with clang for PowerPC:
> > > > > 
> > > > > > > drivers/mfd/kempld-core.c:556:36: warning: unused variable
> > > > > > > 'kempld_acpi_table' [-Wunused-const-variable]
> > > > >    static const struct acpi_device_id kempld_acpi_table[] = {
> > > > > 
> > > > > The issue can be fixed by marking kempld_acpi_table as
> > > > > __maybe_unused.
> > > > > 
> > > > > Fixes: e8299c7313af ("[PATCH] mfd: Add ACPI support to Kontron PLD
> > > > > driver")
> > > > > 
> > > > > Reported-by: kernel test robot <lkp@intel.com>
> > > > > Signed-off-by: Michael Brunner <michael.brunner@kontron.com>
> > > > > ---
> > > > >  drivers/mfd/kempld-core.c | 2 +-
> > > > >  1 file changed, 1 insertion(+), 1 deletion(-)
> > > > > 
> > > > > diff --git a/drivers/mfd/kempld-core.c b/drivers/mfd/kempld-core.c
> > > > > index 1dfe556df038..273481dfaad4 100644
> > > > > --- a/drivers/mfd/kempld-core.c
> > > > > +++ b/drivers/mfd/kempld-core.c
> > > > > @@ -553,7 +553,7 @@ static int kempld_remove(struct platform_device
> > > > > *pdev)
> > > > >  	return 0;
> > > > >  }
> > > > >  
> > > > > -static const struct acpi_device_id kempld_acpi_table[] = {
> > > > > +static const struct acpi_device_id __maybe_unused
> > > > > kempld_acpi_table[] = {
> > > > >  	{ "KEM0001", (kernel_ulong_t)&kempld_platform_data_generic },
> > > > >  	{}
> > > > >  };
> > > > 
> > > > This is not the right fix.  Better just to compile it out completely
> > > > in these circumstances.  I already have a fix for this in soak.
> > > 
> > > Ok - thank you for the other fix you submitted!
> > > 
> > > But just out of curiosity - in process/coding-style.rst is written that
> > > __maybe_unused should be preferred over wrapping in preprocessor
> > > conditionals, if a function or variable may potentially go unused in a
> > > particular configuration. So why is my patch not the right one here? At
> > > least in my tests it seemed to solve the issue.
> > 
> > It's a bone of contention for sure.  In these kinds of scenarios
> > (i.e. ACPI and OF tables) it is way more common to wrap them:
> > 
> > $ git grep -B3 'acpi_device_id\|of_device_id' | grep 'CONFIG_ACPI\|CONFIG_OF' | wc -l
> > 596
> > $ git grep -B3 'acpi_device_id\|of_device_id' | grep __maybe_unused | wc -l
> > 63
> > 
> > Parsing them out completely, also has the benefit of saving space,
> > reducing the size of the finalised binary.
> 
> Doesn't the compiler remove it anyway? At least in my test I didn't see
> a difference in the resulting object files.
> Doing a crosscheck, by adding __attribute__((used)) to the definition
> of kempld_acpi_table, the object file size increased and
> kempld_acpi_table showed up in the symbol table.
> 
> Nevertheless, I don't want to start a discussion. I am fine with using
> the preprocessor. Just wanted to make sure I understand the technical
> implications of both solutions.

This is what happened last time I submitted a patch using
__maybe_unused:

https://lkml.org/lkml/2020/8/17/1704

-- 
Lee Jones [李琼斯]
Senior Technical Lead - Developer Services
Linaro.org │ Open source software for Arm SoCs
Follow Linaro: Facebook | Twitter | Blog