[llvm] [llvm-exegesis] Timeout if subprocess executor hangs (PR #132861)

[Aiden Grossman via llvm-commits]

boomanaiden154 wrote:

Interesting. That test has nothing to do with the subprocess executor, but I can see why there might be some issues there.

It doesn't seem like the tests are even running on my system at all by default (Skylake-X, so definitely has LBR) so it looks like there might have been a lack of test coverage for this for a while. Although it passes if I remove the `-mcpu` flag so it can actually find a sched model with perf counter mappings and I bypass the `REQUIRES:`. Updating the timeout thresholds I think makes sense.  Something has definitely gone wrong if we're not getting results within a second. I'm not familiar enough with how LBR is handled in the perf subsystem to be able to see anything wrong by inspection.

@oontvoo I know this was a long time ago, but do you have any recollection of why you set up the timeout to be so long?

https://github.com/llvm/llvm-project/pull/132861