[flang-commits] [flang] [flang] de-duplicate AbstractResult pass (PR #88867)

[Tom Eccles via flang-commits]

tblah wrote:

Thanks for taking a look

> One question I have looking at the code, it seems the pass can also run on `omp.declare_reduction` now. Is this intentional and if yes, should there be a test exercising this?
We don't currently need this pass to run on `omp.declare_reduction`. I included that operation because my intention is for as many passes as possible to share this same infrastructure. Some other passes do need to run on `omp.declare_reduction`, for example, CFGConversion. I don't want to set a precedent that every combination of pass and top level operation has to be tested because there are a very large number of passes so that would create a lot of work to add a new top level operation (say `omp.private`). Most of these top level operations will only ever have very specific code constructs inside of them (e.g. `fir.global`, `omp.*`) so I think it is easier to test only specifically those cases.


https://github.com/llvm/llvm-project/pull/88867