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
sync: TestPool failure on windows-amd64-2016 #20198
Comments
/cc @aclements for log history greppin'. |
It seems unlikely to be caused by CL 42090 indeed. That (SSE) instruction isn't generated by the compiler and not present in any of the |
Perhaps the goroutine was rescheduled to another P (due to preemption?). |
Also openbsd-386: |
@aclements, this started happening pretty frequently. Do you know what changed? |
CL https://golang.org/cl/42770 mentions this issue. |
The only recent sync.Pool change I can think of is CL 40913. |
That one seems safe. |
Oh, also CL 40918. |
I'm only seeing three failures ever. Did I grep this wrong?
|
@aclements, I've seen a number of trybot failures too, and those three URLs are all of the form used by non-trybot failures. |
Okay. If it's failing fairly regularly on the trybots, can we bisect this? |
Note that the test is formally broken/incorrect. So any innocent change can suddenly cause increased failure rate it in. Does cl/42770 help? We could also use procPin/procUnpin around the test, that should reflect the intention of the test -- within same P reuse order is LIFO. |
Just saw this on nacl-386: https://storage.googleapis.com/go-build-log/b9150e2c/nacl-386_6f3075c2.log |
CL https://golang.org/cl/44310 mentions this issue. |
Noticed on build.golang.org.
https://build.golang.org/log/a3f357b132ef8cb703e42085a50c8c4499e64301
First showed up after CL 42090, although that seems unlikely to be the cause.
cc @valyala @dvyukov
The text was updated successfully, but these errors were encountered: