[flang-commits] [flang] [flang] Modifications to ieee floating point environment procedures (PR #121949)

[via flang-commits]

vdonaldson wrote:

> > This PR allows different code to be generated for sparc / solaris environments vs. other environments since the relevant data sizes are substantially different. I expect the code should be correct for the various cases, but I have no way to verify that myself. Given that, for the tests I would like to find a directive that suppresses the tests for environments that generate alternate code.
> 
> Actually, you can verify things on SPARC yourself: there's both a Solaris/sparcv9 and a Linux/sparc64 system in the [Compile Farm](https://portal.cfarm.net) (and many other systems on top of those).
> 
> Besides, if you can tell me **exactly** what to check, I could do this for both Solaris and Linux. I think that would be better than blindly `XFAIL`ing the tests without understanding the issues.

The two tests are executable Fortran tests. Output for both tests lists expected output on the left side of each line, and actual output on the right. Is that output correct?

https://github.com/llvm/llvm-project/pull/121949