* Arnd Bergmann <a...@kernel.org> wrote:
> From: Arnd Bergmann <a...@arndb.de>
>
> Building with 'make W=1', gcc points out that casting between
> incompatible function types can be dangerous:
>
> arch/x86/math-emu/fpu_trig.c:1638:60: error: cast between incompatible
> function types from ‘int (*)(FPU_REG *, u_char)’ {aka ‘int (*)(struct
> fpu__reg *, unsigned char)’} to ‘void (*)(FPU_REG *, u_char)’ {aka ‘void
> (*)(struct fpu__reg *, unsigned char)’} [-Werror=cast-function-type]
> 1638 | fprem, fyl2xp1, fsqrt_, fsincos, frndint_, fscale, (FUNC_ST0)
> fsin, fcos
> | ^
>
> This one seems harmless, but it is easy enough to work around it by
> adding an intermediate function that adjusts the return type.
>
> Signed-off-by: Arnd Bergmann <a...@arndb.de>
> ---
> arch/x86/math-emu/fpu_trig.c | 11 ++++++++---
> 1 file changed, 8 insertions(+), 3 deletions(-)
>
> diff --git a/arch/x86/math-emu/fpu_trig.c b/arch/x86/math-emu/fpu_trig.c
> index 4a9887851ad8..aebfdb244cb4 100644
> --- a/arch/x86/math-emu/fpu_trig.c
> +++ b/arch/x86/math-emu/fpu_trig.c
> @@ -547,7 +547,7 @@ static void frndint_(FPU_REG *st0_ptr, u_char st0_tag)
> single_arg_error(st0_ptr, st0_tag);
> }
>
> -static int fsin(FPU_REG *st0_ptr, u_char tag)
> +static int do_fsin(FPU_REG *st0_ptr, u_char tag)
> {
> u_char arg_sign = getsign(st0_ptr);
>
> @@ -608,6 +608,11 @@ static int fsin(FPU_REG *st0_ptr, u_char tag)
> }
> }
>
> +static void fsin(FPU_REG *st0_ptr, u_char tag)
> +{
> + fsin(st0_ptr, tag);
> +}
> +
> static int f_cos(FPU_REG *st0_ptr, u_char tag)
> {
> u_char st0_sign;
The existing nomenclature in that code is:
static int f_cos(FPU_REG *st0_ptr, u_char tag)
static void fcos(FPU_REG *st0_ptr, u_char st0_tag)
So I'd suggest naming the variant with a return value f_sin().
Thanks,
Ingo
