[llvm-dev] [RFC] Compiled regression tests. (original) (raw)

Hal Finkel via llvm-dev llvm-dev at lists.llvm.org
Wed Jul 1 11:46:25 PDT 2020

On 7/1/20 1:43 PM, David Greene wrote:

Hal Finkel via llvm-dev <llvm-dev at lists.llvm.org> writes:

I definitely agree that we should not be trying to do this kind of checking using textual metadata-node matching in FileCheck. The alternative already available is to add an analysis pass with some kind of verifier output. This output, not the raw metadata itself, can be checked by FileCheck. We also need to check the verification code, but at least that's something we can keep just in one place. For parallel annotations, we already have such a thing (we can run opt -loops -analyze; e.g., in test/Analysis/LoopInfo/annotated-parallel-complex.ll). We also do this kind of thing for the cost model (by running with -cost-model -analyze). To what extent would making more-extensive use of this technique address the use cases you're trying to address? Analysis tests are definitely useful but I don't think they're sufficient. By their nature, analysis tests run in a "sanitized" environment. The test specifies exactly what is analyzed and no other passes are run.

Yes, but we can also combine them with a transformation pipeline.

From my quick glance at Michael's example test, it's checking that the inlining pass does something specific with metadata: ; Check that the access groups (llvm.access.group) are correctly merged. I suppose one could create an "analysis" that just dumps out metadata information. Is that what you have in mind? That seems like a lot of overhead just to write a test to check metadata.

Yes. From my perspective, it depends on how much resuse you get out of that work. Can the "dump out metadata information" part be reused for many tests? And is there a reason not to use a traditional unit test?

 -Hal

-David

-- Hal Finkel Lead, Compiler Technology and Programming Languages Leadership Computing Facility Argonne National Laboratory

More information about the llvm-dev mailing list