Triaging Polly Issues/Failures from LLVM Changes

10 views
Skip to first unread message

Leonard Chan

unread,
Jun 23, 2021, 2:34:16 PMJun 23
to Polly Development
Hi all,

As of https://reviews.llvm.org/D104732, it looks like `ScopInfo/fortran_array_param_nonmalloc_nonvectored.ll` is failing. See https://luci-milo.appspot.com/p/fuchsia/builders/toolchain.ci/clang-linux-x64/b8843641688849975440 for the full test failure. I pinged the author but was notified that the change does not warrant a revert because it is not considered a core-tier project. So a couple of questions:

- Is polly considered a peripheral-tier project?
- If so, what's the policy on getting polly fixes/updates out as a response to upstream LLVM changes?

Thanks,
Leonard

Michael Kruse

unread,
Jun 23, 2021, 3:17:19 PMJun 23
to Leonard Chan, Polly Development
The tier system was introduced when integrating Bazel into the
monorepo. Although the docs page does not explicitly mention what is
which tier, the discussion around it [1-3] makes it clear that Polly
is a core-tier project that makes it clear that "we should not break
it." It is part of the buildbot and pre-merge check system and
breaking it means people blame mails and pre-merge check failures.

I am working on a patch to repair the test-case.

[1] https://lists.llvm.org/pipermail/llvm-dev/2020-November/146337.html
[2] https://lists.llvm.org/pipermail/llvm-dev/2020-November/146349.html
[3] https://lists.llvm.org/pipermail/llvm-dev/2020-November/146357.html

Michael

--
Tardyzentrismus verboten!
Reply all
Reply to author
Forward
0 new messages