[polly] r331744 - [test] Replace undef with true/false to make test case less fragile

Michael Kruse via llvm-commits llvm-commits at lists.llvm.org
Wed May 9 16:00:23 PDT 2018


We have many bugpoint-reduced test cases using undef. What makes these
undefs more fragile than the others?

Michael

2018-05-08 2:24 GMT-05:00 Tobias Grosser via llvm-commits
<llvm-commits at lists.llvm.org>:
> Author: grosser
> Date: Tue May  8 00:24:05 2018
> New Revision: 331744
>
> URL: http://llvm.org/viewvc/llvm-project?rev=331744&view=rev
> Log:
> [test] Replace undef with true/false to make test case less fragile
>
> This test case does not require undef to be present in branch
> conditions. Replace these undef values with true/false values to clarify
> the control-flow required to reach the loop under testing.
>
> Modified:
>     polly/trunk/test/ScopInfo/scev-div-with-evaluatable-divisor.ll
>
> Modified: polly/trunk/test/ScopInfo/scev-div-with-evaluatable-divisor.ll
> URL: http://llvm.org/viewvc/llvm-project/polly/trunk/test/ScopInfo/scev-div-with-evaluatable-divisor.ll?rev=331744&r1=331743&r2=331744&view=diff
> ==============================================================================
> --- polly/trunk/test/ScopInfo/scev-div-with-evaluatable-divisor.ll (original)
> +++ polly/trunk/test/ScopInfo/scev-div-with-evaluatable-divisor.ll Tue May  8 00:24:05 2018
> @@ -11,14 +11,14 @@ target triple = "x86_64-unknown-linux-gn
>  ; Function Attrs: nounwind uwtable
>  define void @cfft2(i32 %n, double* %A) local_unnamed_addr #0 {
>  entry:
> -  br i1 undef, label %for.body.lr.ph, label %for.end
> +  br i1 true, label %for.body.lr.ph, label %for.end
>
>  for.body.lr.ph:                                   ; preds = %entry
>    br label %for.body
>
>  for.body:                                         ; preds = %for.inc, %for.body.lr.ph
>    %mj.017 = phi i32 [ 1, %for.body.lr.ph ], [ undef, %for.inc ]
> -  br i1 undef, label %if.else, label %if.then
> +  br i1 true, label %if.else, label %if.then
>
>  if.then:                                          ; preds = %for.body
>    br label %for.inc
> @@ -26,16 +26,16 @@ if.then:
>  if.else:                                          ; preds = %for.body
>    %mul.i44 = shl i32 %mj.017, 2
>    %div.i45 = sdiv i32 %n, %mul.i44
> -  br i1 undef, label %for.body.i58.lr.ph, label %for.inc
> +  br i1 true, label %for.body.i58.lr.ph, label %for.inc
>
>  for.body.i58.lr.ph:                               ; preds = %if.else
> -  br i1 undef, label %for.body.i58.us, label %for.body.i58.preheader
> +  br i1 false, label %for.body.i58.us, label %for.body.i58.preheader
>
>  for.body.i58.preheader:                           ; preds = %for.body.i58.lr.ph
>    br label %for.body.i58
>
>  for.body.i58.us:                                  ; preds = %for.body.i58.us, %for.body.i58.lr.ph
> -  br i1 undef, label %for.inc, label %for.body.i58.us
> +  br i1 false, label %for.inc, label %for.body.i58.us
>
>  for.body.i58:                                     ; preds = %for.body.i58, %for.body.i58.preheader
>    store double 0.0, double* %A
>
>
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at lists.llvm.org
> http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-commits


More information about the llvm-commits mailing list