[PATCH] pwm: meson: handle unknown ID values

Arnd Bergmann arnd at arndb.de
Thu Sep 8 02:29:57 PDT 2016


On Thursday, September 8, 2016 11:23:54 AM CEST Thierry Reding wrote:
> On Tue, Sep 06, 2016 at 02:50:47PM +0200, Arnd Bergmann wrote:
> > When building with -Wmaybe-uninitialized, we get a couple of harmless
> > warnings about three functions in this new driver that don't look
> > safe to the compiler:
> > 
> > drivers/pwm/pwm-meson.c: In function 'meson_pwm_get_state':
> > drivers/pwm/pwm-meson.c:355:26: error: 'mask' may be used uninitialized in this function [-Werror=maybe-uninitialized]
> > drivers/pwm/pwm-meson.c: In function 'meson_pwm_disable':
> > drivers/pwm/pwm-meson.c:263:13: error: 'enable' may be used uninitialized in this function [-Werror=maybe-uninitialized]
> > drivers/pwm/pwm-meson.c: In function 'meson_pwm_apply':
> > drivers/pwm/pwm-meson.c:231:13: error: 'clk_shift' may be used uninitialized in this function [-Werror=maybe-uninitialized]
> > drivers/pwm/pwm-meson.c:231:36: error: 'enable' may be used uninitialized in this function [-Werror=maybe-uninitialized]
> > drivers/pwm/pwm-meson.c:231:24: error: 'clk_enable' may be used uninitialized in this function [-Werror=maybe-uninitialized]
> > 
> > Specifically, if we have a device with an id other than 0 or 1,
> > this would result in undefined behavior. This is currently not
> > possible, but the compiler cannot be expected to know this.
> > 
> > This patch adds a 'default' clause to let the compiler know
> > what to do instead, which shuts up the warning and makes the
> > code slightly more resiliant in case it gets extended to other
> > identifiers.
> > 
> > Signed-off-by: Arnd Bergmann <arnd at arndb.de>
> > ---
> >  drivers/pwm/pwm-meson.c | 10 ++++++++++
> >  1 file changed, 10 insertions(+)
> 
> Applied, though I left out the dummy assignment of state->enabled = 0
> because that's dead code anyway.

Sounds good. My first version had "break" instead of "return" there,
so it required the assignment,  but you are in the current version you
are right that we are better off without it.

	Arnd



More information about the linux-amlogic mailing list