runtime: apparent deadlock in TestNoShrinkStackWhileParking on netbsd-arm64-bsiegert #49382

bcmills · 2021-11-05T14:05:25Z

greplogs --dashboard -md -l -e runtime_test\\.TestNoShrinkStackWhileParking\\.func4

@bsiegert, @coypoop: could you investigate? This is the only builder exhibiting this failure, and it occurs quite frequently.

The text was updated successfully, but these errors were encountered:

bcmills · 2021-11-05T14:07:09Z

(Note that this failure mode appears to predate #49209.)

gopherbot · 2021-11-05T14:13:25Z

Change https://golang.org/cl/361615 mentions this issue: runtime: skip TestNoShrinkStackWhileParking on netbsd/arm64

bcmills · 2021-11-05T14:16:01Z

This is a release-blocker (via #11811) until the test is skipped. (Then it can go in the backlog, because netbsd/arm64 is not a first-class port.)

This test appears to deadlock frequently on the only netbsd-arm64 builder we have (netbsd-arm64-bsiegert). Skip the test to provide more useful test coverage for other failures. For #49382 Change-Id: I3be32f58ce1e396f7c69163e70cf58f779f57ac6 Reviewed-on: https://go-review.googlesource.com/c/go/+/361615 Trust: Bryan C. Mills <[email protected]> Trust: Benny Siegert <[email protected]> Run-TryBot: Bryan C. Mills <[email protected]> Reviewed-by: Benny Siegert <[email protected]>

TestUDP/Messages and TestUDP/Messages-dialed occasionally failed because the expected messages were not received in a single RecvMsgs call, or because the messages were received out of order. Assuming that both messages are returned immediately from a single RecvMsgs call was a flawed expectation. Fixed by repeatedly invoking RecvMsgs until all expected messages have been received. While it certainly seems unusual that packets are reordered on a loopback device, it does appear to happen occasionally (on linux-mips). Fixed by sizing receive buffers such that messages in any order can be received correctly, and by allowing either order for the reassembled message. Combine "Messages" and "Messages-dialed" subtests with a simple table-driven test, to avoid the repetition. The same "Message" and "Message-dialed". Finally, make the test failure messages slightly more useful. Fixes golang/go#49382

bcmills added arch-arm64 OS-NetBSD labels Nov 5, 2021

bcmills added this to the Backlog milestone Nov 5, 2021

bcmills added the release-blocker label Nov 5, 2021

bcmills modified the milestones: Backlog, Go1.18 Nov 5, 2021

bcmills added NeedsInvestigation Someone must examine and confirm this is a valid issue and not a duplicate of an existing one. and removed release-blocker labels Nov 5, 2021

bcmills modified the milestones: Go1.18, Backlog Nov 5, 2021

prattmic mentioned this issue Mar 4, 2022

runtime: timeout in TestNoShrinkStackWhileParking on openbsd #51482

Open

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

runtime: apparent deadlock in TestNoShrinkStackWhileParking on netbsd-arm64-bsiegert #49382

runtime: apparent deadlock in TestNoShrinkStackWhileParking on netbsd-arm64-bsiegert #49382

bcmills commented Nov 5, 2021

bcmills commented Nov 5, 2021

gopherbot commented Nov 5, 2021

bcmills commented Nov 5, 2021

runtime: apparent deadlock in TestNoShrinkStackWhileParking on netbsd-arm64-bsiegert #49382

runtime: apparent deadlock in TestNoShrinkStackWhileParking on netbsd-arm64-bsiegert #49382

Comments

bcmills commented Nov 5, 2021

bcmills commented Nov 5, 2021

gopherbot commented Nov 5, 2021

bcmills commented Nov 5, 2021