René's URL Explorer Experiment

{"@context":"https://schema.org","@type":"DiscussionForumPosting","headline":"test_runner: child process exits 0xC0000005 on Windows after its tests pass","articleBody":"### Version\n\nv24.19.0\n\n### Platform\n\n`windows-2025-vs2026` (GitHub-hosted `windows-latest` runner), x64\n\n### Subsystem\n\ntest_runner\n\n### What steps will reproduce the bug?\n\nRun a multi-file suite under `node --test` on Windows:\n\n```\nnode --test --test-concurrency=1 \\\n     --test-reporter=spec --test-reporter-destination=stdout \\\n     --test-reporter=junit --test-reporter-destination=test-results/junit.xml \\\n     \"dist/test/*.test.js\"\n```\n\nIntermittently one test file is reported as failed even though every test in it\npassed. I have not found a minimal reproducer — it is timing-dependent and does\nnot reproduce on demand.\n\n### How often does it reproduce? Is there a required condition?\n\nIntermittent, perhaps 1 run in 8 across a ~1330-test suite. Only on the Windows\nleg; the same commit passes on Linux, arm64 (QEMU) and macOS in the same\nmatrix, and passes on Windows when the job is re-run with no change.\n\n`--test-concurrency=1` does not prevent it.\n\n### What is the expected behavior? Why is that the expected behavior?\n\nA test file whose tests all pass should be reported as passing. The process\nexit status should reflect the test results.\n\n### What do you see instead?\n\nThe file is reported failed at `:1:1` — the whole-file marker rather than any\nassertion — while its own tests all show `✔`:\n\n```\n✔ resolveSignalkNetworks (7.2616ms)\n✖ dist\\test\\resolveSignalkNetworks.test.js (249.1325ms)\n...\nℹ tests 1327\nℹ pass 1323\nℹ fail 1\n  'test failed'\n```\n\nThe JUnit reporter contradicts this in the same run — zero failures overall:\n\n```\ntests=1340  failures=0  errors=0\n```\n\nand the suite for that file reports every test passing:\n\n```xml\n\u003ctestsuite name=\"resolveSignalkNetworks\" tests=\"5\" failures=\"0\" errors=\"0\">\n```\n\nwhile the *file-level* testcase in the same document carries:\n\n```xml\n\u003ctestcase name=\"dist\\test\\resolveSignalkNetworks.test.js\" time=\"0.249133\">\n  \u003cfailure type=\"testCodeFailure\" message=\"test failed\">\n[Error: test failed] {\n  code: 'ERR_TEST_FAILURE',\n  failureType: 'testCodeFailure',\n  cause: 'test failed',\n  exitCode: 3221225477,\n  signal: null\n}\n  \u003c/failure>\n\u003c/testcase>\n```\n\n`3221225477` is `0xC0000005` — `STATUS_ACCESS_VIOLATION`. The test file's child\nprocess appears to crash after completing its tests, rather than exiting\nnon-zero from user code.\n\n### Additional information\n\n- The affected file differs between occurrences and has been unrelated to the\n  change under test each time. The run above was on a PR touching a different\n  file entirely.\n- The console and JUnit totals disagree in an affected run (1327 vs 1340),\n  which suggests the crash also truncates reporter accounting.\n- I have the `exitCode` value from one retained CI artifact (7-day retention\n  expired the earlier ones). The behavioural pattern — all tests pass, file\n  reported failed, clean re-run — is from several occurrences.\n\nI am happy to gather more if it would help narrow this: a specific reporter\ncombination, `--test-concurrency` setting, or an approach to collecting a crash\ndump on a GitHub-hosted Windows runner.\n","author":{"url":"https://github.com/dirkwa","@type":"Person","name":"dirkwa"},"datePublished":"2026-09-03T06:42:04Z","url":"https://github.com/nodejs/node/issues/65756"}