[llvm-commits] CVS: llvm/lib/Transforms/Scalar/InstructionCombining.cpp

[Reid Spencer]

Andrew,

This change has broken several of the InstCombine regression tests.
Without your changes, the regression tests all pass so I have reverted
your patch. Could you please find a workaround that doesn't break
InstCombine? In the future, please do "make check" and ensure that all
changes of this nature pass all the Feature and Regression tests (at a
minimum) before you commit them. 

Here's the regression test output:

FAIL: /proj/llvm/build/../llvm/test/Regression/Transforms/InstCombine/2004-11-27-SetCCForCastLargerAndConstant.ll:
        %Y = cast sbyte %SB to uint             ; <uint> [#uses=1]
        %Y = cast sbyte %SB to int              ; <int> [#uses=1]
        %Y = cast sbyte %SB to int              ; <int> [#uses=1]
        %Y = cast ubyte %SB to uint             ; <uint> [#uses=1]
        %Y = cast ubyte %SB to int              ; <int> [#uses=1]
        %Y = cast sbyte %SB to uint             ; <uint> [#uses=1]
        %Y = cast sbyte %SB to int              ; <int> [#uses=1]
        %Y = cast sbyte %SB to int              ; <int> [#uses=1]
        %Y = cast ubyte %SB to uint             ; <uint> [#uses=1]
        %Y = cast ubyte %SB to int              ; <int> [#uses=1]

FAIL: /proj/llvm/build/../llvm/test/Regression/Transforms/InstCombine/JavaCompare.ll:
        %c1 = setgt int %A, %B          ; <bool> [#uses=1]
        %tmp = cast bool %c1 to int             ; <int> [#uses=1]
        %c2 = setlt int %A, %B          ; <bool> [#uses=1]
        %c31 = setlt int %tmp, 1                ; <bool> [#uses=1]
        %c3 = or bool %c2, %c31         ; <bool> [#uses=1]

FAIL: /proj/llvm/build/../llvm/test/Regression/Transforms/InstCombine/cast-set.ll:
        %B = cast bool %A to int                ; <int> [#uses=1]

FAIL: /proj/llvm/build/../llvm/test/Regression/Transforms/InstCombine/cast.ll:
        %c = cast int %X to long                ; <long> [#uses=1]

FAIL: /proj/llvm/build/../llvm/test/Regression/Transforms/InstCombine/set.ll:
        %C = seteq int %a, %b           ; <bool> [#uses=1]

Reid.


On Sat, 2005-06-04 at 07:44 -0500, Andrew Lenharth wrote:
> 
> Changes in directory llvm/lib/Transforms/Scalar:
> 
> InstructionCombining.cpp updated: 1.345 -> 1.346
> ---
> Log message:
> 
> hack to fix bzip2 (bug 571: http://llvm.cs.uiuc.edu/PR571 )
> 
> ---
> Diffs of the changes:  (+2 -0)
> 
>  InstructionCombining.cpp |    2 ++
>  1 files changed, 2 insertions(+)
> 
> 
> Index: llvm/lib/Transforms/Scalar/InstructionCombining.cpp
> diff -u llvm/lib/Transforms/Scalar/InstructionCombining.cpp:1.345 llvm/lib/Transforms/Scalar/InstructionCombining.cpp:1.346
> --- llvm/lib/Transforms/Scalar/InstructionCombining.cpp:1.345	Sat May 14 07:25:32 2005
> +++ llvm/lib/Transforms/Scalar/InstructionCombining.cpp	Sat Jun  4 07:43:56 2005
> @@ -3088,6 +3088,8 @@
>      return 0;
>    }
>  
> +  return 0; //The rest of this function is broken.  See bug 571
> +
>    // Okay, we have the two reduced sized operands.  If we are doing a <,>
>    // comparison, make sure we perform the compare with the same signedness as
>    // the DestTy.  We don't have to do this if the comparison is !=/== or if the
> 
> 
> 
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> http://mail.cs.uiuc.edu/mailman/listinfo/llvm-commits
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20050612/3b88f628/attachment.sig>