[flang] [llvm] [OpenMP][Flang] Fix atomic operations on complex types (PR #165366)

[Tom Eccles via llvm-commits]

================
@@ -0,0 +1,34 @@
+! Test lowering of atomic read to LLVM IR for complex types.
+! This is a regression test for issue #165184.
+
+! RUN: %flang_fc1 -emit-llvm -fopenmp -o - %s | FileCheck %s
----------------
tblah wrote:

For future reference, we ordinarily do not write tests covering the whole translation from fortran to llvm IR - instead preferring to isolate a smaller part of the compilation flow (e.g. one particular pass, only lowering to HLFIR, only semantics, etc).

These more specific tests are easier to maintain because they are less likely to require modifications due to unrelated changes.

(We do have a few "system" tests which check more complete compilation flows, but these are for exceptional cases).

In this case, this test is written very minimally so I don't think it adds an undue maintenance burden so it can stay as it is. But as Kiran mentioned please do also add to the tests written specifically for OMPIRBuilder.

https://github.com/llvm/llvm-project/pull/165366