Re: [PATCH] x86: set fpmath to 387 to avoid gcc warnings

[H. Peter Anvin]

On 08/06/2012 04:36 AM, dragos.tatulea@xxxxxxxxx wrote:
> From: Dragos Tatulea <dragos.tatulea@xxxxxxxxx>
> 
> If gcc has a different default fpmath set (e.g. sse),
> many unuseful warnings will be spewed when compiling on an
> architecture that doesn't support it.
> 
> An example would be Google's Android 4.1 x86 toochain which
> is compiled with -mfpmath=sse.
> 
> This can affect out of tree modules that compile with
> -Werror (ex: systemtap stp scripts).
> 
> Signed-off-by: Dragos Tatulea <dragos.tatulea@xxxxxxxxx>
> ---
>  arch/x86/Makefile |    1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/arch/x86/Makefile b/arch/x86/Makefile
> index b1c611e..f4fefb8 100644
> --- a/arch/x86/Makefile
> +++ b/arch/x86/Makefile
> @@ -130,6 +130,7 @@ KBUILD_CFLAGS += -fno-asynchronous-unwind-tables
>  # prevent gcc from generating any FP code by mistake
>  KBUILD_CFLAGS += $(call cc-option,-mno-sse -mno-mmx -mno-sse2 -mno-3dnow,)
>  KBUILD_CFLAGS += $(call cc-option,-mno-avx,)
> +KBUILD_CFLAGS += $(call cc-option,-mfpmath=387,)
>  
>  KBUILD_CFLAGS += $(mflags-y)
>  KBUILD_AFLAGS += $(mflags-y)
> 

I guess we should do -msoft-float as well to try to get an error if FP
instructions are generated at all...

	-hpa
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/