Update the premerge testing system to use the LLVM-wide premerge
infrastructure. Also remove libcxx-restart-preempted-jobs.yaml, as this
should no longer be needed.
This patch removes setting the MAX_PARLLEL_COMPILE_JOBS and
MAX_PARALLEL_LINK_JOBS env variables in the windows runs. These were
originally used to control the parallelism on the old infrastructure and
we set them on the new infrastructure explicitly so that we could
maintain both at the same time. Now it does not make sense to keep them
explicitly set that we do not need to explicitly control the parallelism
given the amount of RAM we have on the machines. This also adds a
maintnenace cost as evidenced by the fact that these have been incorrect
(64 instead of 32) for quite a while.
As discussed in PR #142353, the current testsuite of the `clang` Python
bindings has several issues:
- It `libclang.so` cannot be loaded into `python` to run the testsuite,
the whole `ninja check-all` aborts.
- The result of running the testsuite isn't report like the `lit`-based
tests, rendering them almost invisible.
- The testsuite is disabled in a non-obvious way (`RUN_PYTHON_TESTS`) in
`tests/CMakeLists.txt`, which again doesn't show up in the test results.
All these issues can be avoided by integrating the Python bindings tests
with `lit`, which is what this patch does:
- The actual test lives in `clang/test/bindings/python/bindings.sh` and
is run by `lit`.
- The current `clang/bindings/python/tests` directory (minus the
now-subperfluous `CMakeLists.txt`) is moved into the same directory.
- The check if `libclang` is loadable (originally from PR #142353) is
now handled via a new `lit` feature, `libclang-loadable`.
- The various ways to disable the tests have been turned into `XFAIL`s
as appropriate. This isn't complete and not completely tested yet.
Tested on `sparc-sun-solaris2.11`, `sparcv9-sun-solaris2.11`,
`i386-pc-solaris2.11`, `amd64-pc-solaris2.11`, `i686-pc-linux-gnu`, and
`x86_64-pc-linux-gnu`.
Co-authored-by: Rainer Orth <ro@gcc.gnu.org>
Despite the error message for preempted jobs containing the words
"cancelled", these are considered workflow "failures" by github.
This is important, because if we fail to distinguish between "failed"
and "cancelled" jobs, the restarter will fight to restart jobs a user
intentionally cancelled (either by pressing the "cancel" button, or by
pushing an update to a PR).
This reverts commit 3ea7fc73397032e71fb20d27084f4552211bb1f6. This also
reverts earlier attempts to solve this problem by matching the messages
to detect manual cancellations.
This change also removes ldionne's test workflow, as its hard to
correctly keep in sync.
This change does not attempt to address the maintainability or
testability of this script, which continues to be an issue. If asked to
address these issues, my plan is to write the script in python (which
most people are more familar with), and turn this action into a "docker
action" using a container with the python action and dependencies built
into it. Let me know if that's a direction we're interested in heading.
As discussed in PR #142353, the current testsuite of the `clang` Python
bindings has several issues:
- It `libclang.so` cannot be loaded into `python` to run the testsuite,
the whole `ninja check-all` aborts.
- The result of running the testsuite isn't report like the `lit`-based
tests, rendering them almost invisible.
- The testsuite is disabled in a non-obvious way (`RUN_PYTHON_TESTS`) in
`tests/CMakeLists.txt`, which again doesn't show up in the test results.
All these issues can be avoided by integrating the Python bindings tests
with `lit`, which is what this patch does:
- The actual test lives in `clang/test/bindings/python/bindings.sh` and
is run by `lit`.
- The current `clang/bindings/python/tests` directory (minus the
now-subperfluous `CMakeLists.txt`) is moved into the same directory.
- The check if `libclang` is loadable (originally from PR #142353) is
now handled via a new `lit` feature, `libclang-loadable`.
- The various ways to disable the tests have been turned into `XFAIL`s
as appropriate. This isn't complete and not completely tested yet.
Tested on `sparc-sun-solaris2.11`, `sparcv9-sun-solaris2.11`,
`i386-pc-solaris2.11`, `amd64-pc-solaris2.11`, `i686-pc-linux-gnu`, and
`x86_64-pc-linux-gnu`.
Co-authored-by: Rainer Orth <ro@gcc.gnu.org>
The restarter will restart jobs that were canceled by a newer change,
causing the new and old change to fight it out.
This change attempts to address this by treating the "canceled" message
as an error.
As discussed in PR #142353, the current testsuite of the `clang` Python
bindings has several issues:
- If `libclang.so` cannot be loaded into `python` to run the testsuite,
the whole `ninja check-all` aborts.
- The result of running the testsuite isn't report like the `lit`-based
tests, rendering them almost invisible.
- The testsuite is disabled in a non-obvious way (`RUN_PYTHON_TESTS`) in
`tests/CMakeLists.txt`, which again doesn't show up in the test results.
All these issues can be avoided by integrating the Python bindings tests
with `lit`, which is what this patch does:
- The actual test lives in `clang/test/bindings/python/bindings.sh` and
is run by `lit`.
- The current `clang/bindings/python/tests` directory (minus the
now-superfluous `CMakeLists.txt`) is moved into the same directory.
- The check if `libclang` is loadable (originally from PR #142353) is
now handled via a new `lit` feature, `libclang-loadable`.
- The various ways to disable the tests have been turned into `XFAIL`s
as appropriate.
- AArch64 doesn't `FAIL` any longer, so no `XFAIL` is necessary.
- It keeps the `check-clang-python` target for use by the Clang Python
CI.
Tested on `sparc-sun-solaris2.11`, `sparcv9-sun-solaris2.11`,
`i386-pc-solaris2.11`, `amd64-pc-solaris2.11`, `i686-pc-linux-gnu`, and
`x86_64-pc-linux-gnu`.
This reverts commit 6f62979a5a5bcf70d65f23e0991a274e6df5955b.
The reapplies commit 80ea5f46df3e365a0a2112889bb91732167b6214.
That commit was reverted because it was causing compiler-rt test
failures due to tysan not having its dependencies set up properly within
CMake. That situation has since been rectified in
3cef099ceddccefca8e11268624397cde9e04af6.
Reviewers: lnihlen, rnk, gburgessiv, cmtice
Reviewed By: rnk, cmtice
Pull Request: https://github.com/llvm/llvm-project/pull/144033
A while ago, the test workflow was updated with a new preemption regex,
however it was only applied to the test job, and not the job
that's actually restarting the failed libc++ test runs.
This fix should correct the issue and get the restarter working
again.
As for now, It may be hard for people to get truth from long Discourse
discussion, so a link to official document may be enough to convince
changing email from private to public.
This patch migrates the premerge pipeline to use LLVM_ENABLE_RUNTIMES to
build libc and compiler-rt.
Reviewers: DavidSpickett, tstellar, cmtice, lnihlen
Reviewed By: DavidSpickett
Pull Request: https://github.com/llvm/llvm-project/pull/142696
Our CI is currently having major difficulties, which causes the Stage 3
CI to basically never succeed. As a result, our macOS CI jobs have not
been running recently. This patch gates the macOS CI jobs on Stage 2
instead of Stage 3 so that they actually run sometimes.
This patch removes the "test only please ignore" tagline from the
premerge job names. Now that we are looking to sunset the old
infrastructure pretty soon and the new infrastructure is reporting
errors, we want people to actually pay attention to the failures and
report anything erroneous.
This patch makes the new premerge system report failures when the build
errors out. We were previously not doing this to not notify people on
failures as we were testing out the infra.
This works towards making the new premerge system canonical and the
deprecation of the old system.
A launch announcement on Discourse will accompany this commit.
Reviewers: cmtice, tstellar, joker-eph, Keenuts, dschuff, lnihlen, gburgessiv
Reviewed By: tstellar, joker-eph, Keenuts
Pull Request: https://github.com/llvm/llvm-project/pull/139359
This patch stops running the premerge checks on the main branch. This is
based on some feedback in
https://discourse.llvm.org/t/rfc-running-premerge-postcommit-through-github-actions/86124.
We want to use buildbot for post commit testing. This should be covered
in the mean time by the current pemerge-monolithic-* bots even though
the configurations are not exactly the same. This also adds a bit of
capacity back to the cluster (although might not in the end when we
start running more substantial postcommit testing through buildbot).
This is primarily in preparation for turning the new premerge system on
as canonical. Without this, we run into warning fatigue issues as
described in the RFC above.
Reviewers: Keenuts, joker-eph, dschuff, cmtice, gburgessiv, lnihlen, tstellar
Reviewed By: Keenuts, joker-eph
Pull Request: https://github.com/llvm/llvm-project/pull/139358
This patch partially prunes the windows container to reduce the image
size, primarily to improve image pull time which is currently a pretty
significant bottleneck in the new premerge due to autoscaling.
This patch removes the following:
- An extra copy of LLVM that is not needed anymore.
- An unneeded perl installation
- Some extra python packages that are specific to buildbot
This overall saves about 4GB on the uncompressed image, or about 20%.
I tested this locally against the premerge pipeline and everything
passes.
There are still several significant areas of opportunity, namely seeing
if we can move away from the 4.8 sdk image to just the
`windowsservercore` image (about 7GB of opportunity), and shrinking the
VS installation (in total about 5GB uncompressed currently opportunity
unknown).
This patch renames the new premerge job as suggested in
https://discourse.llvm.org/t/github-ci-notifications-and-main-branch/85868/10.
This uses more industry standard terms (like CI vs premerge checks which
might be somewhat of a LLVM CI idiom?) and makes it more generic if we
end up doing postcommit testing through Github.
This patch installs sccache using apt-get in the CI container build
process rather than manually downloading it from github releases. This
makes things quite a bit simpler and means we are not manually handling
things like hash/signature verification. This is only possible now that
we are using ubuntu 24.04 by default.
The comment was originally added in
b3af755254599f474601d0d7b7374dcc36af279a, but
2898c3e0bbf904c68efb194525bcdfeedb3014c2 greatly simplified things
making the previous comment invalid. We no longer need to look at
diffing main versus the PR and whether we are doing a two dot or three
dot diff as we are just diffing the merge commit.
When editing the premerge.yaml file it seems like I left some extra yaml
in there that ended up in the run section of one of the steps that ended
up throwing the error.
This patch renames premerge artifacts per OS. Without this, the premerge
artifacts will conflict with each other and the slower running job will
fail to upload the artifact.
The premerge pipeline currently creates an artifacts directory with some
statistics that gets uploaded on the buildkite side for later
inspection. This patch adds support for this on the Github side by using
the upload artifacts action.
Reviewers: Keenuts, lnihlen, mizvekov, tstellar, Endilll
Reviewed By: mizvekov
Pull Request: https://github.com/llvm/llvm-project/pull/135538
This patch bumps workflows depending upon the Linux CI container to
ubuntu 24.04. The 22.04 container is no longer being built as it was
recently bumped to 24.04, so this patch moves all of these workflows
over to the new container to keep them updated and ensure they are using
an actually maintained version of the container image.
In 0547e573c555, I introduced backdeployment testing on macOS using
Github-provided builders. This was done by basically building libc++ on
a slightly older macOS (like macOS 13) and then running against the
system library on that machine. However, that created a dependency that
libc++ must keep working on macOS 13, which doesn't support the
latest-released Xcode.
This patch solves that problem by moving the deployment testing to a
newer version of macOS which supports the latest-released version of
Xcode.
Sadly, that also reduces the backdeployment coverage we have since we're
not actually testing on older OSes, but is necessary to satisfy the
documented libc++ support policy. In the future, we could improve the
situation by providing a Lit configuration that allows compiling (but
not running) all the tests, building the tests on a supported macOS, and
then shipping those tests on an older backdeployment target in order to
run them against the system library. Since that requires significant
engineering, this isn't done at this time.
The Github runner version got bumped recently and it would be good to keep
this up to date. Also debugging an issue where Github ARC is failing to
create new pods and trying to see if it might be related to outdated
versions.
This patch migrates the CI over to the new compute_projects.py script
for calculating what projects need to be tested based on a change to
LLVM.
Reviewers: lnihlen, ldionne, tstellar, Endilll, joker-eph, Keenuts
Reviewed By: Keenuts, tstellar
Pull Request: https://github.com/llvm/llvm-project/pull/132642
This patch changes getting changed files in the pr code format job to
just checking out the previous two commits (the merge commit and its
porent, the current commit latest in main), which allows us to just diff
the merge commit. This means we do not have to checkout the ref through
the merge base, which should save approximately a minute per job (or
much more in some cases where the PR is particularly out of date).
