[flang-commits] [flang] [flang][OpenMP] Fix firstprivate not working with lastprivate in DO SIMD (PR #170163)

[Tom Eccles via flang-commits]

tblah wrote:

I think they take order(s) of magnitude more compute than running llvm's built in lit and unit tests. I don't think it would be impossible but there could be pushback. But in an ideal world I would like to see this too.

Last I checked, for the Fujitsu tests there's no upstream public list of what is expected to pass and fail (and the few that are known to be flaky). When I run them myself I have to compare results from before and after a change to see if anything regressed. Having something with a straightforward pass/fail for the fujitsu tests would be needed. If you'd like to upstream that I would be more than happy to help testing where I can as it would make my work a bit easier.

https://github.com/llvm/llvm-project/pull/170163