[llvm] r332548 - [Thumb] preserve test intent by removing undef

via llvm-commits llvm-commits at lists.llvm.org
Wed May 16 20:15:44 PDT 2018


Hi Sanjay,

Your change is causing a failure on a bot, can you take a look?

http://lab.llvm.org:8011/builders/clang-cmake-armv8-selfhost-neon/builds/598/steps/ninja%20check%202/logs/stdio

FAIL: LLVM :: CodeGen/Thumb2/cross-rc-coalescing-2.ll (24923 of 38795)
******************** TEST 'LLVM :: CodeGen/Thumb2/cross-rc-coalescing-2.ll' FAILED ********************
Script:
--
/home/buildslave/buildslave/clang-cmake-armv8-selfhost-neon/stage2/bin/llc < /home/buildslave/buildslave/clang-cmake-armv8-selfhost-neon/llvm/test/CodeGen/Thumb2/cross-rc-coalescing-2.ll -mtriple=thumbv7-apple-darwin9 -mcpu=cortex-a8 -arm-atomic-cfg-tidy=0 | /home/buildslave/buildslave/clang-cmake-armv8-selfhost-neon/stage2/bin/FileCheck /home/buildslave/buildslave/clang-cmake-armv8-selfhost-neon/llvm/test/CodeGen/Thumb2/cross-rc-coalescing-2.ll
--
Exit Code: 2

Command Output (stderr):
--
/home/buildslave/buildslave/clang-cmake-armv8-selfhost-neon/stage2/bin/llc: <stdin>:71:34: error: '%34' defined with type 'i1'
  %35 = fadd float 0.000000e+00, %34
                                 ^
FileCheck error: '-' is empty.
FileCheck command line:  /home/buildslave/buildslave/clang-cmake-armv8-selfhost-neon/stage2/bin/FileCheck /home/buildslave/buildslave/clang-cmake-armv8-selfhost-neon/llvm/test/CodeGen/Thumb2/cross-rc-coalescing-2.ll

--

********************

Douglas Yung

> -----Original Message-----
> From: llvm-commits [mailto:llvm-commits-bounces at lists.llvm.org] On Behalf Of
> Sanjay Patel via llvm-commits
> Sent: Wednesday, May 16, 2018 15:48
> To: llvm-commits at lists.llvm.org
> Subject: [llvm] r332548 - [Thumb] preserve test intent by removing undef
> 
> Author: spatel
> Date: Wed May 16 15:47:51 2018
> New Revision: 332548
> 
> URL: http://llvm.org/viewvc/llvm-project?rev=332548&view=rev
> Log:
> [Thumb] preserve test intent by removing undef
> 
> We need to clean up the DAG floating-point undef logic.
> This process is similar to how we handled integer undef
> logic in D43141.
> 
> And as we did there, I'm trying to reduce the patch by
> changing tests that would probably become meaningless
> once we correct FP undef folding.
> 
> Modified:
>     llvm/trunk/test/CodeGen/Thumb2/cross-rc-coalescing-2.ll
> 
> Modified: llvm/trunk/test/CodeGen/Thumb2/cross-rc-coalescing-2.ll
> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/Thumb2/cross-
> rc-coalescing-2.ll?rev=332548&r1=332547&r2=332548&view=diff
> ==============================================================================
> --- llvm/trunk/test/CodeGen/Thumb2/cross-rc-coalescing-2.ll (original)
> +++ llvm/trunk/test/CodeGen/Thumb2/cross-rc-coalescing-2.ll Wed May 16
> 15:47:51 2018
> @@ -35,14 +35,14 @@ bb8:
>    %5 = fmul float %4, %0                          ; <float> [#uses=1]
>    %6 = fsub float %3, %5                          ; <float> [#uses=1]
>    %7 = fmul float %4, %1                          ; <float> [#uses=1]
> -  %8 = fadd float undef, %7                       ; <float> [#uses=2]
> +  %8 = fadd float %7, %7
>    %9 = load float, float* %fi.1, align 4                 ; <float> [#uses=2]
>    %10 = fsub float %9, %8                         ; <float> [#uses=1]
>    %11 = fadd float %9, %8                         ; <float> [#uses=1]
>    %12 = fsub float 0.000000e+00, %6               ; <float> [#uses=1]
> -  %13 = fsub float 0.000000e+00, undef            ; <float> [#uses=2]
> -  %14 = fmul float undef, %0                      ; <float> [#uses=1]
> -  %15 = fadd float %14, undef                     ; <float> [#uses=2]
> +  %13 = fsub float 0.000000e+00, %12
> +  %14 = fmul float %0, %0
> +  %15 = fadd float %14, %14
>    %16 = load float, float* %scevgep81, align 4           ; <float> [#uses=2]
>    %17 = fsub float %16, %15                       ; <float> [#uses=1]
>    %18 = fadd float %16, %15                       ; <float> [#uses=2]
> @@ -68,6 +68,6 @@ bb8:
>    br i1 %34, label %bb8, label %bb9
> 
>  bb9:                                              ; preds = %bb8
> -  %35 = fadd float 0.000000e+00, undef            ; <float> [#uses=1]
> +  %35 = fadd float 0.000000e+00, %34
>    br label %bb7
>  }
> 
> 
> _______________________________________________
> 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