Re: [PATCH] input: appletouch: shut up -Wmaybe-uninitialized warning

[Dmitry Torokhov <dmitry.torokhov@gmail.com>]

Hi Arnd,

On Tue, Jan 07, 2020 at 09:42:45PM +0100, Arnd Bergmann wrote:
> Building with -O3 introduces a false-positive warning in this file:
> 
> In file included from include/linux/compiler_types.h:68,
>                  from <command-line>:
> drivers/input/mouse/appletouch.c: In function 'atp_complete_geyser_3_4':
> include/linux/compiler-gcc.h:71:45: error: 'x_z' may be used uninitialized in this function [-Werror=maybe-uninitialized]
>  #define __UNIQUE_ID(prefix) __PASTE(__PASTE(__UNIQUE_ID_, prefix), __COUNTER__)
>                                              ^~~~~~~~~~~~
> drivers/input/mouse/appletouch.c:657:12: note: 'x_z' was declared here
>   int x, y, x_z, y_z, x_f, y_f;
>             ^~~
> In file included from include/linux/compiler_types.h:68,
>                  from <command-line>:
> drivers/input/mouse/appletouch.c: In function 'atp_complete_geyser_1_2':
> include/linux/compiler-gcc.h:71:45: error: 'x_z' may be used uninitialized in this function [-Werror=maybe-uninitialized]
>  #define __UNIQUE_ID(prefix) __PASTE(__PASTE(__UNIQUE_ID_, prefix), __COUNTER__)
>                                              ^~~~~~~~~~~~
> drivers/input/mouse/appletouch.c:517:12: note: 'x_z' was declared here
>   int x, y, x_z, y_z, x_f, y_f;
>             ^~~
> 
> The variables are not actually used here since the only usage
> happens when both 'x' and 'y' are non-zero, but this is something
> that gcc fails to track.
> 
> Fixes: mmtom ("init/Kconfig: enable -O3 for all arches")
> Signed-off-by: Arnd Bergmann <arnd@arndb.de>
> ---
>  drivers/input/mouse/appletouch.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/input/mouse/appletouch.c b/drivers/input/mouse/appletouch.c
> index 3f06e8a495d8..ec2139c43fcd 100644
> --- a/drivers/input/mouse/appletouch.c
> +++ b/drivers/input/mouse/appletouch.c
> @@ -330,7 +330,7 @@ static void atp_reinit(struct work_struct *work)
>  			retval);
>  }
>  
> -static int atp_calculate_abs(struct atp *dev, int offset, int nb_sensors,
> +static noinline int atp_calculate_abs(struct atp *dev, int offset, int nb_sensors,
>  			     int fact, int *z, int *fingers)

I wonder if instead of nolinine we can have "*fingers = *z = 0;" in this
function...

I feel noinline has to be reserved for exceptional cases...

Thanks.

-- 
Dmitry