[PATCH] pwm: sun4i: Initialize variables before use

[PATCH] pwm: sun4i: Initialize variables before use

[Thierry Reding]

GCC can't always determine that the duty, period and prescaler values
are initialized when returning from sun4i_pwm_calculate(), so help out a
little by initializing them to 0.

Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
---
 drivers/pwm/pwm-sun4i.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/pwm/pwm-sun4i.c b/drivers/pwm/pwm-sun4i.c
index 0decc7cde133..3e3efa6c768f 100644
--- a/drivers/pwm/pwm-sun4i.c
+++ b/drivers/pwm/pwm-sun4i.c
@@ -234,9 +234,9 @@ static int sun4i_pwm_apply(struct pwm_chip *chip, struct pwm_device *pwm,
 {
 	struct sun4i_pwm_chip *sun4i_pwm = to_sun4i_pwm_chip(chip);
 	struct pwm_state cstate;
-	u32 ctrl, duty, period, val;
+	u32 ctrl, duty = 0, period = 0, val;
 	int ret;
-	unsigned int delay_us, prescaler;
+	unsigned int delay_us, prescaler = 0;
 	unsigned long now;
 	bool bypass;
 
-- 
2.24.1

Re: [PATCH] pwm: sun4i: Initialize variables before use

[Uwe Kleine-König]

Hello Thierry,

On Mon, Jan 20, 2020 at 03:32:06PM +0100, Thierry Reding wrote:
> GCC can't always determine that the duty, period and prescaler values
> are initialized when returning from sun4i_pwm_calculate(), so help out a
> little by initializing them to 0.

Is it worth mentioning the gcc version you're using?

> Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
> ---
>  drivers/pwm/pwm-sun4i.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/pwm/pwm-sun4i.c b/drivers/pwm/pwm-sun4i.c
> index 0decc7cde133..3e3efa6c768f 100644

I don't find this object (0decc7cde133) in my tree or next. Which
version is this?

> --- a/drivers/pwm/pwm-sun4i.c
> +++ b/drivers/pwm/pwm-sun4i.c
> @@ -234,9 +234,9 @@ static int sun4i_pwm_apply(struct pwm_chip *chip, struct pwm_device *pwm,
>  {
>  	struct sun4i_pwm_chip *sun4i_pwm = to_sun4i_pwm_chip(chip);
>  	struct pwm_state cstate;
> -	u32 ctrl, duty, period, val;
> +	u32 ctrl, duty = 0, period = 0, val;

+     u32 ctrl, uninitialized_var(duty), uninitialized_var(period), val;

should fix the warnings, too, and additionally explicitly documents that
it's just the compiler that doesn't see there is no problem.

Best regards
Uwe

-- 
Pengutronix e.K.                           | Uwe Kleine-König            |
Industrial Linux Solutions                 | https://www.pengutronix.de/ |

Re: [PATCH] pwm: sun4i: Initialize variables before use

[Clément Péron]

Hi Uwe, Thierry

On Mon, 20 Jan 2020 at 21:09, Uwe Kleine-König
<u.kleine-koenig@pengutronix.de> wrote:
>
> Hello Thierry,
>
> On Mon, Jan 20, 2020 at 03:32:06PM +0100, Thierry Reding wrote:
> > GCC can't always determine that the duty, period and prescaler values
> > are initialized when returning from sun4i_pwm_calculate(), so help out a
> > little by initializing them to 0.
>
> Is it worth mentioning the gcc version you're using?

This issue has been trig by kbuild test robot.
I planned to submit a patch for it as it's due to my modification but
forget to submit it...

Original report
[linux-next:master 6586/9861] drivers/pwm/pwm-sun4i.c:57:34: warning:
'prescaler' may be used uninitialized in this function

tree:   https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git
master
head:   de970dffa7d19eae1d703c3534825308ef8d5dec
commit: 9f28e95b5286fce63a3d0d90dc7ca43eca8dda58 [6586/9861] pwm:
sun4i: Add support to output source clock directly
config: microblaze-randconfig-a001-20200118 (attached as .config)
compiler: microblaze-linux-gcc (GCC) 7.5.0
reproduce:
        wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross
-O ~/bin/make.cross
        chmod +x ~/bin/make.cross
        git checkout 9f28e95b5286fce63a3d0d90dc7ca43eca8dda58
        # save the attached .config to linux build tree
        GCC_VERSION=7.5.0 make.cross ARCH=microblaze

If you fix the issue, kindly add following tag
Reported-by: kbuild test robot <lkp@intel.com>

Note: it may well be a FALSE warning. FWIW you are at least aware of it now.
http://gcc.gnu.org/wiki/Better_Uninitialized_Warnings

>
> > Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
> > ---
> >  drivers/pwm/pwm-sun4i.c | 4 ++--
> >  1 file changed, 2 insertions(+), 2 deletions(-)
> >
> > diff --git a/drivers/pwm/pwm-sun4i.c b/drivers/pwm/pwm-sun4i.c
> > index 0decc7cde133..3e3efa6c768f 100644
>
> I don't find this object (0decc7cde133) in my tree or next. Which
> version is this?
>
> > --- a/drivers/pwm/pwm-sun4i.c
> > +++ b/drivers/pwm/pwm-sun4i.c
> > @@ -234,9 +234,9 @@ static int sun4i_pwm_apply(struct pwm_chip *chip, struct pwm_device *pwm,
> >  {
> >       struct sun4i_pwm_chip *sun4i_pwm = to_sun4i_pwm_chip(chip);
> >       struct pwm_state cstate;
> > -     u32 ctrl, duty, period, val;
> > +     u32 ctrl, duty = 0, period = 0, val;
>
> +     u32 ctrl, uninitialized_var(duty), uninitialized_var(period), val;
>
> should fix the warnings, too, and additionally explicitly documents that
> it's just the compiler that doesn't see there is no problem.
>
> Best regards
> Uwe
>
> --
> Pengutronix e.K.                           | Uwe Kleine-König            |
> Industrial Linux Solutions                 | https://www.pengutronix.de/ |

Re: [PATCH] pwm: sun4i: Initialize variables before use

[Thierry Reding]

[-- Attachment #1: Type: text/plain, Size: 1943 bytes --]

On Mon, Jan 20, 2020 at 09:09:17PM +0100, Uwe Kleine-König wrote:
> Hello Thierry,
> 
> On Mon, Jan 20, 2020 at 03:32:06PM +0100, Thierry Reding wrote:
> > GCC can't always determine that the duty, period and prescaler values
> > are initialized when returning from sun4i_pwm_calculate(), so help out a
> > little by initializing them to 0.
> 
> Is it worth mentioning the gcc version you're using?

I could, but what good is that going to be? I don't think this is
something that's limited to one specific version but I don't know
exactly which ones are impacted. Stating just one specific version
isn't all that useful in that case.

> > Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
> > ---
> >  drivers/pwm/pwm-sun4i.c | 4 ++--
> >  1 file changed, 2 insertions(+), 2 deletions(-)
> > 
> > diff --git a/drivers/pwm/pwm-sun4i.c b/drivers/pwm/pwm-sun4i.c
> > index 0decc7cde133..3e3efa6c768f 100644
> 
> I don't find this object (0decc7cde133) in my tree or next. Which
> version is this?

I made this on top of my local pwm/for-next when I was build-testing,
which I usually do before pushing, so it's not surprising that you
don't have this in your tree.

> 
> > --- a/drivers/pwm/pwm-sun4i.c
> > +++ b/drivers/pwm/pwm-sun4i.c
> > @@ -234,9 +234,9 @@ static int sun4i_pwm_apply(struct pwm_chip *chip, struct pwm_device *pwm,
> >  {
> >  	struct sun4i_pwm_chip *sun4i_pwm = to_sun4i_pwm_chip(chip);
> >  	struct pwm_state cstate;
> > -	u32 ctrl, duty, period, val;
> > +	u32 ctrl, duty = 0, period = 0, val;
> 
> +     u32 ctrl, uninitialized_var(duty), uninitialized_var(period), val;
> 
> should fix the warnings, too, and additionally explicitly documents that
> it's just the compiler that doesn't see there is no problem.

I haven't convinced myself fully yet that there really isn't a problem.
I'm fairly sure it's safe, but always initializing to 0 doesn't hurt.

Thierry

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [PATCH] pwm: sun4i: Initialize variables before use

[Uwe Kleine-König]

Hello Thierry,

On Tue, Jan 21, 2020 at 02:50:11PM +0100, Thierry Reding wrote:
> On Mon, Jan 20, 2020 at 09:09:17PM +0100, Uwe Kleine-König wrote:
> > On Mon, Jan 20, 2020 at 03:32:06PM +0100, Thierry Reding wrote:
> > > GCC can't always determine that the duty, period and prescaler values
> > > are initialized when returning from sun4i_pwm_calculate(), so help out a
> > > little by initializing them to 0.
> > 
> > Is it worth mentioning the gcc version you're using?
> 
> I could, but what good is that going to be? I don't think this is
> something that's limited to one specific version but I don't know
> exactly which ones are impacted. Stating just one specific version
> isn't all that useful in that case.

I think something like:

	gcc 4.6 reports [...], newer gccs are fine.

is useful. If you read that in a few years, it helps you either to
reproduce the problem or determine it is not important any more.
 
> > > Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
> > > ---
> > >  drivers/pwm/pwm-sun4i.c | 4 ++--
> > >  1 file changed, 2 insertions(+), 2 deletions(-)
> > > 
> > > diff --git a/drivers/pwm/pwm-sun4i.c b/drivers/pwm/pwm-sun4i.c
> > > index 0decc7cde133..3e3efa6c768f 100644
> > 
> > I don't find this object (0decc7cde133) in my tree or next. Which
> > version is this?
> 
> I made this on top of my local pwm/for-next when I was build-testing,
> which I usually do before pushing, so it's not surprising that you
> don't have this in your tree.

The reason I asked this (and also the gcc version) is to reproduce the
issue and work with it. With your reply I can only say that I expect
that uninitialized_var fixes the problem in a better way. If I knew the
exact circumstances, I could test them and claim that it indeed fixes it
(or see it doesn't and don't take your time with non-sense reviews).

> > > --- a/drivers/pwm/pwm-sun4i.c
> > > +++ b/drivers/pwm/pwm-sun4i.c
> > > @@ -234,9 +234,9 @@ static int sun4i_pwm_apply(struct pwm_chip *chip, struct pwm_device *pwm,
> > >  {
> > >  	struct sun4i_pwm_chip *sun4i_pwm = to_sun4i_pwm_chip(chip);
> > >  	struct pwm_state cstate;
> > > -	u32 ctrl, duty, period, val;
> > > +	u32 ctrl, duty = 0, period = 0, val;
> > 
> > +     u32 ctrl, uninitialized_var(duty), uninitialized_var(period), val;
> > 
> > should fix the warnings, too, and additionally explicitly documents that
> > it's just the compiler that doesn't see there is no problem.
> 
> I haven't convinced myself fully yet that there really isn't a problem.
> I'm fairly sure it's safe, but always initializing to 0 doesn't hurt.

Yes, I agree that initializing the variable fixes the warning. Still
using uninitialized_var is the better way (assuming it indeed fixes the
problem, see above).

Best regards
Uwe

-- 
Pengutronix e.K.                           | Uwe Kleine-König            |
Industrial Linux Solutions                 | https://www.pengutronix.de/ |