[llvm-commits] [llvm] r55886 - /llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp

[Bill Wendling]

It's not a bug in that this wasn't being done. I was doing a static  
analysis of the DAG combiner and noticed that this wasn't doing what  
the comment claimed it should be doing. If this transformation is  
performed in the selection DAG code, then it should probably be  
removed from here entirely (along with the analogous "ADD"  
transformation). What do you think?

-bw

On Sep 7, 2008, at 5:19 PM, Evan Cheng wrote:

> Is this fixing a bug? getNode should be performing constant folding.
> See SelectionDAG.cpp line 2441. If somehow that wasn't happening,
> please fix it there and provide a test case.
>
> Thanks,
>
> Evan
>
> On Sep 7, 2008, at 4:34 AM, Bill Wendling wrote:
>
>> Author: void
>> Date: Sun Sep  7 06:34:47 2008
>> New Revision: 55886
>>
>> URL: http://llvm.org/viewvc/llvm-project?rev=55886&view=rev
>> Log:
>> Convert
>>
>>   // fold (sub c1, c2) -> c1-c2
>>
>> from a no-op into an actual transformation.
>>
>> Modified:
>>   llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
>>
>> Modified: llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
>> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp?rev=55886&r1=55885&r2=55886&view=diff
>>
>> =
>> =
>> =
>> =
>> =
>> =
>> =
>> =
>> = 
>> =====================================================================
>> --- llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp (original)
>> +++ llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp Sun Sep  7
>> 06:34:47 2008
>> @@ -1111,7 +1111,7 @@
>>    return DAG.getConstant(0, N->getValueType(0));
>>  // fold (sub c1, c2) -> c1-c2
>>  if (N0C && N1C)
>> -    return DAG.getNode(ISD::SUB, VT, N0, N1);
>> +    return DAG.getConstant(N0C->getAPIntValue() - N1C-
>>> getAPIntValue(), VT);
>>  // fold (sub x, c) -> (add x, -c)
>>  if (N1C)
>>    return DAG.getNode(ISD::ADD, VT, N0,
>>
>>
>> _______________________________________________
>> llvm-commits mailing list
>> llvm-commits at cs.uiuc.edu
>> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits