[Mlir-commits] [flang] [llvm] [mlir] [flang][OpenMP] Support user-defined declare reduction with derived types (PR #190288)

[llvmlistbot at llvm.org]

MattPD wrote:

This replaces the merged and reverted commit e80604a from PR #184897.

Part 4, GPU codegen for non-descriptor by-ref reductions, addresses the ICE encountered in the original PR: These are the changes in llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp (I did run git-clang-format, FWIW) and mlir/lib/Target/LLVMIR/Dialect/OpenMP/OpenMPToLLVMIRTranslation.cpp

(I can see 1 failing check, "CI Checks / Build and Test Linux AArch64 (pull_request)". The failure is an LLDB test:
TestSyntheticCapping.py - Fatal Python error: Segmentation fault. It's a Python segfault in an LLDB data-formatter test not related to this patch.)


https://github.com/llvm/llvm-project/pull/190288