[Mlir-commits] [flang] [mlir] [flang][openmp] Add support for ordered regions in SIMD directives (PR #181012)

[llvmlistbot at llvm.org]

================
@@ -0,0 +1,102 @@
+// RUN: mlir-translate -mlir-to-llvmir %s | FileCheck %s
+
+// Test that linear variables in worksharing+SIMD loops with ordered regions
+// are correctly rewritten to use .linear_result in:
+// 1. The ordered region (omp.ordered.region)
+// 2. Code after the ordered region (omp_region.finalize)
+//
+// This tests "omp ordered simd" nested in  "omp do simd ordered" case
+// !$omp do simd ordered
+// do i = 1, n
+//     a(i) = b(i) * 10
+//         !$omp ordered simd
+//             print *, a(i)
+//         !$omp end ordered
+//     c(i) = a(i) * 2
+// end do
+// !$omp end do simd
+
+module {
+  omp.private {type = private} @i_private_i32 : i32
+
+  llvm.func @_FortranAioBeginExternalListOutput(i32, !llvm.ptr, i32) -> !llvm.ptr
+  llvm.func @_FortranAioOutputInteger32(!llvm.ptr, i32) -> i1
+  llvm.func @_FortranAioEndIoStatement(!llvm.ptr) -> i32
+
+  llvm.mlir.global internal constant @str("test.f90\00") {addr_space = 0 : i32}
----------------
NimishMishra wrote:

Same comment here as before

https://github.com/llvm/llvm-project/pull/181012