[Mlir-commits] [mlir] [MLIR][OpenMP] Normalize lowering of omp.loop_nest (PR #127217)

[Sergio Afonso]

================
@@ -2348,57 +2364,31 @@ convertOmpSimd(Operation &opInst, llvm::IRBuilderBase &builder,
     llvm::Expected<llvm::CanonicalLoopInfo *> loopResult =
         ompBuilder->createCanonicalLoop(
             loc, bodyGen, lowerBound, upperBound, step,
-            /*IsSigned=*/true, /*InclusiveStop=*/true, computeIP);
+            /*IsSigned=*/true, loopOp.getLoopInclusive(), computeIP);
----------------
skatrak wrote:

Yes, we weren't looking at the `inclusive` attribute for `omp.simd`.

https://github.com/llvm/llvm-project/pull/127217