Repository navigation
Conversation
HarryR
force-pushed
the
diag/ci-alert-wait-hang
branch
from
May 28, 2026 04:09
4695c00 to
84326c1
Compare
cpulimit-wraps QEMU at N% of one CPU. THROTTLE=10 reproduces the GitHub-Actions sysenter hang locally that doesn't surface on full- speed TCG or KVM. Diagnostic only; not for main.
HarryR
force-pushed
the
diag/ci-alert-wait-hang
branch
from
May 28, 2026 05:05
84326c1 to
15ad4a1
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
GitHub Actions runners have shown a repeatable hang at the sysenter 'alert wakes alertable wait' test that doesn't reproduce on local TCG/KVM or
gh act. Previously,th:wait()was unbounded — if the alert was lost, the test hung indefinitely and CI sat in selftest silence until the step wall ran out.Bound both waits (done:wait(2.0) keeps its prior bound; th:wait gains a 5.0s bound). On failure, enumerate every thread in every process via NtQuerySystemInformation and print thread_state / wait_reason / context_switches. This gives us — for the first time — a kernel-side post-mortem from the runner where the bug actually manifests.
Diagnostic branch; not for main until we know what the dump says.