Re: [PATCH] x86: LLVMLinux: Wrap -mno-80387 with cc-option

[Behan Webster <behanw@converseincode.com>]

On 04/21/14 23:05, Ingo Molnar wrote:
> * behanw@converseincode.com <behanw@converseincode.com> wrote:
>
>> From: Behan Webster <behanw@converseincode.com>
>>
>> Wrap -mno-80387 gcc options with cc-option so they don't break clang.
>>
>> Signed-off-by: Behan Webster <behanw@converseincode.com>
>> ---
>>   arch/x86/Makefile | 4 +++-
>>   1 file changed, 3 insertions(+), 1 deletion(-)
>>
>> diff --git a/arch/x86/Makefile b/arch/x86/Makefile
>> index d1b7c37..ce6ad7e 100644
>> --- a/arch/x86/Makefile
>> +++ b/arch/x86/Makefile
>> @@ -83,7 +83,9 @@ else
>>           KBUILD_CFLAGS += -m64
>>   
>>           # Don't autogenerate traditional x87, MMX or SSE instructions
>> -        KBUILD_CFLAGS += -mno-mmx -mno-sse -mno-80387 -mno-fp-ret-in-387
>> +        KBUILD_CFLAGS += -mno-mmx -mno-sse
>> +        KBUILD_CFLAGS += $(call cc-option,-mno-80387)
>> +        KBUILD_CFLAGS += $(call cc-option,-mno-fp-ret-in-387)
> Is there a clang equivalent option that inhibits all things FPU opcode
> generation by the compiler?
Not that I've found yet. Still investigating.

> That's the general purpose of -no-80387.
Yes, I understand what this is trying to accomplish. At this point I'd 
just like this new code not to break the use of clang with v3.15.

I will submit another patch which adds similar functionality when 
compiled with clang once I have that answer.

Behan

-- 
Behan Webster
behanw@converseincode.com