← board

Track guessed as B from the FAILING STEP — line 1 of 2, stable_linux_amd64/default/pinned -Fulib/rtl/platform/posix test/lib_dns_aaaa.pas /tmp/lib_dns_aaaa, which names test/lib_dns_aaaa.pas. Not from the job's name or its src: those describe what the job is ABOUT, and this job's recipe spans 2 source file(s). The ranker reads frontmatter, so this line — not the body — decides who works it; correct it if the guess is wrong.

This commit CANNOT be the cause. The job builds only with $(PXX_STABLE), and this commit moved no stable_linux_amd64/** — so the bytes that compiled it are unchanged. Look at flakiness or box load, not at the named sha; the bisect is unsound here and has been skipped.

origin/master has advanced 12 commit(s) since this sha. Re-verify at current HEAD before acting — the callback is tagged to the sha that was tested, which may no longer be the state of the tree.

regression: lib-test#src:test/lib_dns_aaaa.pas at 021cd94f10a9 in step 1/2, stable_linux_amd64/default/pinned -Fulib/rtl/platform/posix test/lib_dns_aaaa.pas /tmp/lib_dns_aaaa (auto-filed by twatch)

Repro

tools/testmgr.py --tier full --job 'lib-test#src:test/lib_dns_aaaa.pas' at 021cd94f10a97c051c6c71d75d308028adc34c43

Range

The named sha 021cd94f10a9 CANNOT be the cause — it touches no buildable file (docs / tickets / tstate only). It is the sha that was TESTED, i.e. the upper bound of an untested range; the cause is somewhere below it.

bad 021cd94f10a9, last good 1e37a55f6748, 14 observable commit(s) in range (it builds with $(PXX_STABLE), so compiler/ commits cannot have caused it and are dropped; pin moves, lib/ and test/ are kept) — the watcher narrows this by idle bisect.

Log tail

pascal26:44: error: undefined variable (PalVfork)
(tail)
pascal26:44: error: undefined variable (PalVfork)
  near: boundP ) ; pid := PalVfork >>> ; if pid

Stub ticket: signal only. Track T agent (face 2) enriches or a dev track takes it from the repro line.

2026-09-01 (frankH) — resolved: one deliberate PAL rename, filed once per file

e2ba5a1e1 renamed the PAL entry PalVfork -> PalFork on purpose (the old name was false and had already cost a real conclusion in lib/crtl/src/unistd.c, which left fork() an ENOSYS stub by reasoning FROM the name, against a PAL whose body issued SYS_fork all along). The rename is right; ten call sites in test/ were missed.

Track T filed eight tickets, one per failing file. They are one bug. Fixed together in the commit subject fix(B): eight DNS tests still called PalVfork, a name e2ba5a1e1 deliberately retired — lib_dns_{resolve,facade,spoof,tcp, multins,chase}.pas plus cache_facade and aaaa with two sites each.

Pure rename, same signature; PalVforkAndExec deliberately untouched.

Verified: make lib-test EXIT=0 end to end against stable v399.

Worth recording that the two instruments agreed while failing differently: a grep for the shape in the working tree found exactly the same eight files the watcher found by running them on seven. That is corroboration; two greps would not have been.

For whoever tunes the auto-filer

Each of these tickets carries "look at flakiness or box load, not at the named sha". It is the only actionable line on them and it is wrong — nothing was flaky. Dropping compiler/ commits from a $(PXX_STABLE) job's bisect range is sound, but the fallback then assumes the residue is environmental. lib/ and test/ remain in range and are exactly where such a job's real regressions live; the advice should name them before it says "flakiness".