[PHP] test: verify AF_UNIX listener queues across fork and exec#745
Merged
brandonpayton merged 1 commit intoJul 11, 2026
Conversation
This was referenced Jul 10, 2026
Exercise the existing shared-listener and in-place exec behavior with one fork-instrumented guest on Node and Chromium. Cover a connection queued before exec, a second connection after exec, the CLOEXEC/non-CLOEXEC alias split, FIFO acceptance, and exact queue consumption. This replaces PR 745’s obsolete socket serialization approach with regression coverage only; it does not change runtime behavior or the ABI.
32bd40b to
b49be60
Compare
11 tasks
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Purpose
Add durable cross-host regression coverage for the POSIX listener behavior now
owned by Batch 2: an AF_UNIX listener inherited across
fork()keeps oneaccept queue, and
exec()closes itsFD_CLOEXECalias without losing asurviving alias or an already-queued connection.
The original #745 production patch is not replayed. Its socket-wire and accept
queue intent was superseded by the reviewed in-place implementation already in
the aggregate (
e3c7ba3a9and9def16b2a).Reviewed result
SOCK_CLOEXEClistener and a non-CLOEXECdup();execve();EBADFafter exec while the surviving aliasremains a listening socket;
exactly once, and exchanges data over both.
requests, or HTTP error responses as well as incorrect guest results.
SocketInfocomments to describe the existing shared listenerqueue and clone-reset behavior.
exports, and host poll logging.
This changes no runtime behavior, ABI surface, package artifact, or VFS image.
Review view
Accept into Batch 2 as regression coverage via rebase merge. The production
behavior is already supplied by the aggregate; this PR makes it durable across
the real guest, Node, and browser paths. It does not independently make #871
ready for
main, and the aggregate ABI-17 and full ready-to-merge gates remain.Validation
On the exact reviewed commit:
cargo test -p kandelo --target aarch64-apple-darwin --lib— 1076/1076passed.
cd host && npx vitest run test/unix-listener-exec.test.ts— 1/1 passed.CI=true KANDELO_PLAYWRIGHT_PORT=5545 npx playwright test test/unix-listener-exec.spec.ts --project=chromium --workers=1— realChromium, 1/1 passed. Vite reported missing optional Node-demo artifacts
during its dependency scan; the isolated test-runner route and guest ran
successfully with no captured browser/runtime diagnostics.
bash scripts/check-abi-version.sh— passed; it reported only theaggregate's already-known backward-compatible additive exports.
git diff --check— clean.Earlier on the same repaired behavior tree,
scripts/build-musl.shandbash build.shalso completed. The full host, libc, POSIX, Sortix, PHPT,canonical browser, and manual
./run.sh browsergates were not run for thistest-only constituent; they remain aggregate-only under the locked Batch 2
policy.