[PATCH] D68577: [LV] Apply sink-after & interleave-groups as VPlan transformations (NFC)

[Eric Christopher via llvm-commits]

I've gone ahead and reverted this here:

To github.com:llvm/llvm-project.git
   9f10cc2d124..009e032634b  master -> master

as we have a testcase.

Also it seems that NFC isn't quite right here :) Perhaps splitting out
the change into more NFC type commits?

-eric

On Wed, Nov 6, 2019 at 8:24 PM Richard Trieu via Phabricator via
llvm-commits <llvm-commits at lists.llvm.org> wrote:
>
> rtrieu added a comment.
>
> The recommited patch is hitting an assert.  See message and reduced test case below:
>
> assertion failed at llvm/lib/Transforms/Vectorize/VPRecipeBuilder.h:91 in llvm::VPRecipeBase *llvm::VPRecipeBuilder::getRecipe(llvm::Instruction *): Ingredient2Recipe[I] != nullptr && "Ingredient doesn't have a recipe"
>
>   "clang" \
>   "-cc1" \
>   "-triple" "x86_64-unknown-linux-gnu" \
>   "-emit-obj" \
>   "-O3" \
>   "-w" \
>   "-vectorize-loops" \
>   "-x" "c" "test.c"
>
>
>
>   int test(int I1, int *Arr1, int *Arr2) {
>     int i = 0, j = 0;
>     int I2 = 0;
>
>     int ret = 0;
>
>     for (i = 0; i < 5; i++) {
>       if ((Arr1[i] < 5)) {
>         I2 = I1;
>       }
>     }
>
>     for (i = 0; i < I2; i++, j++) {
>       Arr2[3] = (j < I1 ? Arr1[j] : 0);
>     }
>     if (j < I1) {
>       ret = 1;
>     }
>
>     return ret;
>   }
>
>
> Repository:
>   rG LLVM Github Monorepo
>
> CHANGES SINCE LAST ACTION
>   https://reviews.llvm.org/D68577/new/
>
> https://reviews.llvm.org/D68577
>
>
>
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at lists.llvm.org
> https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-commits