llvm dev - Apr 2011 - [LLVMdev] [x86 codegen] 3DNow! intrinsics not behaving as expected.

On Apr 14, 2011, at 12:47 PM, Eli Friedman wrote:
>> I looked at the program using a debugger, and the pfadd instruction is
>> executed correctly and the MMX register contains the correct values.
>> The code that prepares the stack for the printf call seems to be
>> messing it up.
> 
> I would call that "user error"; basically, using MMX instructions
> messes up the FP stack, and we assume the user is smart enough to make
> sure the two don't mix.
More specifically, if you use MMX/3dNow intrinsics, you have to call
"emms" at ABI boundaries.

-Chris

On Thu, Apr 14, 2011 at 5:37 PM, Chris Lattner <clattner at apple.com>
wrote:
>
> On Apr 14, 2011, at 12:47 PM, Eli Friedman wrote:
>
>>> I looked at the program using a debugger, and the pfadd instruction
is
>>> executed correctly and the MMX register contains the correct
values.
>>> The code that prepares the stack for the printf call seems to be
>>> messing it up.
>>
>> I would call that "user error"; basically, using MMX
instructions
>> messes up the FP stack, and we assume the user is smart enough to make
>> sure the two don't mix.
>
> More specifically, if you use MMX/3dNow intrinsics, you have to call
"emms" at ABI boundaries.
>
> -Chris
>
Ok, now it makes sense. Thanks. Now that I know this is right, are
these patches ok to commit? Or should I post them to the llvm-commits,
and cfe-commits list?

- Michael Spencer

On Apr 14, 2011, at 2:43 PM, Michael Spencer wrote:
> On Thu, Apr 14, 2011 at 5:37 PM, Chris Lattner <clattner at
apple.com> wrote:
>> 
>> On Apr 14, 2011, at 12:47 PM, Eli Friedman wrote:
>> 
>>>> I looked at the program using a debugger, and the pfadd
instruction is
>>>> executed correctly and the MMX register contains the correct
values.
>>>> The code that prepares the stack for the printf call seems to
be
>>>> messing it up.
>>> 
>>> I would call that "user error"; basically, using MMX
instructions
>>> messes up the FP stack, and we assume the user is smart enough to
make
>>> sure the two don't mix.
>> 
>> More specifically, if you use MMX/3dNow intrinsics, you have to call
"emms" at ABI boundaries.
>> 
>> -Chris
>> 
> 
> Ok, now it makes sense. Thanks. Now that I know this is right, are
> these patches ok to commit? Or should I post them to the llvm-commits,
> and cfe-commits list?
If they seem obvious to you, feel free to commit them directly.  Thanks Michael,

-Chris