[Mlir-commits] [flang] [mlir] [Flang][OpenMP] Don't generate code for unreachable target regions. (PR #178937)

[Tom Eccles]

https://github.com/tblah commented:

Why not use MLIR's `mlir/Analysis/DataFlow/DeadCodeAnalysis.h`?

Using dominance instead of data flow analysis might be lighter weight in compilation time (I never checked this) but I'm not sure it will produce easy to interpret results in the presence of loops (maybe that would be okay in this case, I haven't thought hard about it).

https://github.com/llvm/llvm-project/pull/178937