On Thu, 22 Jun 2023, Nick Desaulniers wrote:
> On Thu, Jun 22, 2023 at 9:12 AM Nathan Chancellor <nathan@xxxxxxxxxx> wrote:
> >
> > Clang's kernel Control Flow Integrity (kCFI) is a compiler-based
> > security mitigation that ensures the target of an indirect function call
> > matches the expected type of the call and trapping if they do not match
> > exactly. The warning -Wincompatible-function-pointer-types-strict aims
> > to catch these issues at compile time, which reveals:
> >
> > drivers/leds/leds-mt6323.c:598:49: error: incompatible function pointer types assigning to 'int (*)(struct led_classdev *, enum led_brightness)' from 'int (struct led_classdev *, unsigned int)' [-Werror,-Wincompatible-function-pointer-types-strict]
> > 598 | leds->led[reg]->cdev.brightness_set_blocking =
> > | ^
> > 599 | mt6323_wled_set_brightness;
> > | ~~~~~~~~~~~~~~~~~~~~~~~~~~
> > drivers/leds/leds-mt6323.c:600:40: error: incompatible function pointer types assigning to 'enum led_brightness (*)(struct led_classdev *)' from 'unsigned int (struct led_classdev *)' [-Werror,-Wincompatible-function-pointer-types-strict]
> > 600 | leds->led[reg]->cdev.brightness_get =
> > | ^
> > 601 | mt6323_get_wled_brightness;
> > | ~~~~~~~~~~~~~~~~~~~~~~~~~~
> > 2 errors generated.
> >
> > While 'unsigned int' is ABI compatible with 'enum led_brightness' (hence
> > no warning from -Wincompatible-function-pointer-types) and the callers
> > of these callbacks use/pass the values as 'unsigned int', the mismatch
> > between the prototype and the called function will trip kCFI at runtime.
> >
> > Change the types in the implementations to match the prototypes, clearing
> > up the warning and avoiding kCFI failures.
> >
> > Fixes: 9bb0a9e0626c ("leds: leds-mt6323: Add support for WLEDs and MT6332")
> > Signed-off-by: Nathan Chancellor <nathan@xxxxxxxxxx>
>
> Thanks for the patch! Consider additionally having
> mt6323_get_wled_brightness return LED_OFF rather than 0 in its
> ternary.
https://elixir.bootlin.com/linux/latest/source/include/linux/leds.h#L32
Perhaps this is not relevant for this older driver though?
I'd really like some more information from Pavel on the history.
> Reviewed-by: Nick Desaulniers <ndesaulniers@xxxxxxxxxx>
>
>
> > ---
> > drivers/leds/leds-mt6323.c | 6 +++---
> > 1 file changed, 3 insertions(+), 3 deletions(-)
> >
> > diff --git a/drivers/leds/leds-mt6323.c b/drivers/leds/leds-mt6323.c
> > index e8fecfc2e90a..24f35bdb55fb 100644
> > --- a/drivers/leds/leds-mt6323.c
> > +++ b/drivers/leds/leds-mt6323.c
> > @@ -76,7 +76,7 @@ struct mt6323_led {
> > int id;
> > struct mt6323_leds *parent;
> > struct led_classdev cdev;
> > - unsigned int current_brightness;
> > + enum led_brightness current_brightness;
> > };
> >
> > /**
> > @@ -451,7 +451,7 @@ static int mtk_wled_hw_off(struct led_classdev *cdev)
> > return 0;
> > }
> >
> > -static unsigned int mt6323_get_wled_brightness(struct led_classdev *cdev)
> > +static enum led_brightness mt6323_get_wled_brightness(struct led_classdev *cdev)
> > {
> > struct mt6323_led *led = container_of(cdev, struct mt6323_led, cdev);
> > struct mt6323_leds *leds = led->parent;
> > @@ -471,7 +471,7 @@ static unsigned int mt6323_get_wled_brightness(struct led_classdev *cdev)
> > }
> >
> > static int mt6323_wled_set_brightness(struct led_classdev *cdev,
> > - unsigned int brightness)
> > + enum led_brightness brightness)
> > {
> > struct mt6323_led *led = container_of(cdev, struct mt6323_led, cdev);
> > struct mt6323_leds *leds = led->parent;
> >
> > ---
> > base-commit: 7bd932d9adbcc5c5370d968bdb0b00385606bf3a
> > change-id: 20230621-mt6323-wled-wincompatible-function-pointer-types-strict-334f06d92ffb
> >
> > Best regards,
> > --
> > Nathan Chancellor <nathan@xxxxxxxxxx>
> >
> >
>
>
> --
> Thanks,
> ~Nick Desaulniers
--
Lee Jones [李琼斯]
