Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

gcext test is often failing #50434

Closed
vtjnash opened this issue Jul 6, 2023 · 5 comments · Fixed by #50533
Closed

gcext test is often failing #50434

vtjnash opened this issue Jul 6, 2023 · 5 comments · Fixed by #50533
Assignees
Labels
domain:ci Continuous integration
Milestone

Comments

@vtjnash
Copy link
Sponsor Member

vtjnash commented Jul 6, 2023

The gcext test is failing. A quick fix would be to change the length(errlines) == 0 check to use isempty(errlines) instead, which has special printing in Test and would likely quickly reveal the issue.
example: https://buildkite.com/julialang/julia-master/builds/25595#01892862-243e-4599-b390-cf005b815ea4

@vtjnash vtjnash added the domain:ci Continuous integration label Jul 6, 2023
@vchuravy vchuravy added this to the 1.10 milestone Jul 6, 2023
@vchuravy
Copy link
Sponsor Member

Looked briefly into this, definitely a release blocker.
Diogo surmised that we likely broke conservative GC support with the page table changes

@d-netto
Copy link
Member

d-netto commented Jul 13, 2023

5939e2d (from #50137) is the first commit in which gcext is failing.

vtjnash pushed a commit that referenced this issue Jul 13, 2023
Ensure `internal_obj_base_ptr` checks whether objects past freelist
pointer are in freelist.

Fixes #50434
KristofferC pushed a commit that referenced this issue Jul 17, 2023
Ensure `internal_obj_base_ptr` checks whether objects past freelist
pointer are in freelist.

Fixes #50434

(cherry picked from commit dcca46b)
@fingolfin
Copy link
Contributor

I just stumbled over this issue. Yes we noticed in GAP.jl resp Oscar.jl new crashes related to our gcext code, see oscar-system/Oscar.jl#2543

I don't know yet if these are related or a separate issue (nor do
i exclude a possible bug i our code -- although nothing there has changed recently).

CC @benlorenz

@d-netto
Copy link
Member

d-netto commented Jul 17, 2023

Is this reproducible on the latest master?

@fingolfin
Copy link
Contributor

I am not sure, the problem with these issues is that we in general have a hard time reproducing them outside of CI (the CI runners have comparatively low memory compared to our dev machines, and exhibit the problems much more frequently).

The last one I am explicitly aware of was 5 days ago, the same day PR #50533 was merged, so perhaps that resolved it?

In general it would be great to get a ping when you observe gcext failures, because they almost always affect us, and then we can coordinate testing. If you wish I'll be happy to do the reverse (so we far we usually try to first figure out more about the problem in case it is on our side; but with the recent fast pace of GC changes, maybe it'd be better to notify you immediately? But we don't want to bother you guys unduly...)

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
domain:ci Continuous integration
Projects
None yet
4 participants