[PATCH] Teach BaseIndexOffset::match to identify base pointers in loops

[Juergen Ributzka]

Yup, it was already committed in r188936. I fixed the style in r189526 as you suggested.

Thanks

-Juergen

On Aug 28, 2013, at 3:20 PM, Eric Christopher <echristo at gmail.com> wrote:

> Haven't looked to see if it was ever committed (it also LGTM and
> awesome for the better comments), one style nit:
> 
> +    if (Ptr->getOperand(1)->getOpcode() == ISD::MUL) {
> +      return BaseIndexOffset(Ptr, SDValue(), 0, IsIndexSignExt);
> +    }
> 
> No need for braces here I don't think.
> 
> -eric
> 
> 
> On Wed, Aug 21, 2013 at 2:44 PM, Juergen Ributzka <juergen at apple.com> wrote:
>> Hi @ll,
>> 
>> the small utility function that pattern matches Base + Index + Offset patterns for loads and stores fails to recognize the base pointer for loads/stores from/into an array at offset 0 inside a loop. As a result DAGCombiner::MergeConsecutiveStores was not able to merge all stores.
>> 
>> Example:
>> for(int i=0; i <N; i+=4) {
>>        a[i+0] = 1;
>>        a[i+1] = 1;
>>        a[i+2] = 1;
>>        a[i+3] = 1;
>> }
>> 
>> In this example BaseIndexOffset::match would identify a different base pointer for a[i+0] then for all the others.
>> 
>> This small patch adds an additional pattern match to recognize this special case and a test case.
>> 
>> Cheers,
>> Juergen
>> 
>> 
>> _______________________________________________
>> llvm-commits mailing list
>> llvm-commits at cs.uiuc.edu
>> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20130828/ef75dc22/attachment.html>