[flang-commits] [flang] [flang][OpenMP] Fix use_device_addr handling for COMMON blocks in target data (PR #217105)

[via flang-commits]

https://github.com/agozillon approved this pull request.

The lowering changes look fine to me, please do wait on someone else to look at the semantic lowering before landing though. 

Could you please add a runtime regression test to offload/test/offloading/fortran please! 

And does it make sense to consider extending this change to use_device_ptr/has_device_addr (can be separate PRs, if they do need the increased coverage)?  

https://github.com/llvm/llvm-project/pull/217105