[flang-commits] [flang] [flang] Restrict O0 hlfir.assign scalar-to-array inlining to OpenMP target device (PR #201774)

[Sairudra More via flang-commits]

Saieiei wrote:

This failure looks unrelated to this PR.

The failing test is `libomptarget :: amdgcn-amd-amdhsa :: tools/omp-kernel-replay/record-replay-diff-threads.cpp`, built with `clang++`, and it fails during `llvm-omp-kernel-replay --verify` with a GPU memory access fault. This PR only changes Flang pipeline configuration/tests for `InlineHLFIRAssign` scheduling.

I don’t see a connection to the Flang HLFIR/FIR lowering change here, but I’m happy to rerun/monitor if needed.

https://github.com/llvm/llvm-project/pull/201774