[flang-commits] [flang] [mlir] [RFC][Draft][flang][FIR] Model implict box allocations in side effects (PR #178412)

[Tom Eccles via flang-commits]

tblah wrote:

I am back now and had some time to look more closely at this. Flang is actually correct to not use implicit firstprivatization in this case, because the shared DSA in the enclosing parallel region sets the default for the task.

If modifying the effects in FIR is not an option I will have to investigate how feasible it would be to make outlined openmp operations isolated from above. 

https://github.com/llvm/llvm-project/pull/178412