On Dec 17, 2007, at 4:27 PM, Chris Lattner wrote:

>
> On Dec 17, 2007, at 2:33 PM, Evan Cheng wrote:
>
>> Author: evancheng
>> Date: Mon Dec 17 16:33:23 2007
>> New Revision: 45131
>>
>> URL: http://llvm.org/viewvc/llvm-project?rev=45131&view=rev
>> Log:
>> Bring back int_x86_sse2_movl_dq intrinsic for backward
>> compatibility. Make sure
>> it's auto-upgraded to a shufflevector instruction.
>
> Nice, thanks Evan.
>
>> +++ llvm/trunk/include/llvm/IntrinsicsX86.td Mon Dec 17 16:33:23 2007
>> @@ -460,6 +460,8 @@
>>   def int_x86_sse2_packuswb_128 :
>> GCCBuiltin<"__builtin_ia32_packuswb128">,
>>               Intrinsic<[llvm_v8i16_ty, llvm_v8i16_ty,
>>                          llvm_v8i16_ty], [IntrNoMem]>;
>> +  def int_x86_sse2_movl_dq : GCCBuiltin<"__builtin_ia32_movqv4si">,
>> +              Intrinsic<[llvm_v4i32_ty, llvm_v4i32_ty],  
>> [IntrNoMem]>;
>
> I don't think this is need anymore?  The intrinsic can't get past the
> bc/ll readers now.

AutoUpgrade.cpp has to recognize it somehow?

     switch(F->getIntrinsicID()) {
     case Intrinsic::x86_sse2_movl_dq: {
        ...
     }

Evan

>
>
> -Chris
>
>
> _______________________________________________
> llvm-commits mailing list
> llvm-commits@cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

_______________________________________________
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

Reply via email to