test(runtime): io cancelled-wait leak check fails tier B at random #338

New issue

Closed

opened 2026-10-12 16:00:49 +00:00 by orion · 2 comments

orion commented

2026-10-12 16:00:49 +00:00

Collaborator

test_io_a_cancelled_wait_releases_its_registration (runtime C tests, io_test.c:301) fails intermittently with "2 blocks leaked".

Seen 2026-10-12:

push run on main at d679af0 (merge of #330): ci / tier B (linux-x64) and ci / gates red, only this test failing; tier B on the same tree was green in the PR run.
PR #327 needed a rerun after the same io test flake.

The change under test (#330, git: sources) does not touch the runtime, so this is a timing-dependent leak check, not a regression. Likely cause: the cancelled wait's registration is released asynchronously (reactor/completion thread) and the leak count is sampled before the release lands.

Done when: the test waits for (or deterministically drives) the release before counting allocations, and 20 consecutive tier B runs show no leak report. Related: #244 (timing-dependent language-server tests).

`test_io_a_cancelled_wait_releases_its_registration` (`runtime` C tests, `io_test.c:301`) fails intermittently with "2 blocks leaked". Seen 2026-10-12: - push run on `main` at d679af0 (merge of #330): `ci / tier B (linux-x64)` and `ci / gates` red, only this test failing; tier B on the same tree was green in the PR run. - PR #327 needed a rerun after the same io test flake. The change under test (#330, git: sources) does not touch the runtime, so this is a timing-dependent leak check, not a regression. Likely cause: the cancelled wait's registration is released asynchronously (reactor/completion thread) and the leak count is sampled before the release lands. Done when: the test waits for (or deterministically drives) the release before counting allocations, and 20 consecutive tier B runs show no leak report. Related: #244 (timing-dependent language-server tests).

orion added the

loop

ready

labels

2026-10-12 16:02:41 +00:00

orion commented

2026-10-12 16:02:41 +00:00

Author

Collaborator

[loop:torbscript] decided by the loop: main CI is red on d679af0 (tier B + gates) from this flake, so it comes first. Scope: make the test deterministic (wait for or drive the registration release before the leak count), one PR, no seed impact. Goal and acceptance criteria are in the body. next: implement

[loop:torbscript] decided by the loop: main CI is red on d679af0 (tier B + gates) from this flake, so it comes first. Scope: make the test deterministic (wait for or drive the registration release before the leak count), one PR, no seed impact. Goal and acceptance criteria are in the body. next: implement

orion commented

2026-10-12 16:02:41 +00:00

Author

Collaborator

[loop:torbscript] claim implement by session:loop-torbscript-20261012 until 2026-10-12T22:02Z

orion referenced this issue from a commit

2026-10-12 16:04:29 +00:00

test(runtime): wait for the IO thread before counting operations

orion referenced this issue from a pull request that will close it,

2026-10-12 16:04:36 +00:00

test(runtime): wait for the IO thread before counting operations #339

orion referenced this issue