← board

origin/master has advanced 1 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: test-core#src:test/csocket_loopback_b88.c red at 330f62af78d0 (auto-filed by twatch)

Repro

tools/testmgr.py --tier native --job 'test-core#src:test/csocket_loopback_b88.c' at 330f62af78d0971ee8c39d5aa9bf6dc294aeaab7

Range

bad 330f62af78d0, last good 7d8929633721, 58 commit(s) in range — the watcher narrows this by idle bisect; check tstate/TSTATE.md for the current range.

Log tail

pascal26:1: error: C include file not found: "socket.c" (searched: test/, lib/crtl/include/, lib/crtl/src/, /tmp/testmgr-scratch-989299/../lib/crtl/include/, /tmp/testmgr-scratch-989299/../../lib/crtl/include/, lib/crtl/include/, <host system dirs>)
(tail)
pascal26:1: error: C include file not found: "socket.c" (searched: test/, lib/crtl/include/, lib/crtl/src/, /tmp/testmgr-scratch-989299/../lib/crtl/include/, /tmp/testmgr-scratch-989299/../../lib/crtl/include/, lib/crtl/include/, <host system dirs>)

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

Triage (Track T, 2026-08-05) — attributed, and NOT the blamed sha

Cause: 8d7c47f8f — "fix(crtl): the socket veneer was never pulled, so calls fell back to glibc", which MOVED the file:

D  lib/crtl/src/socket.c
A  lib/crtl/src/sys/socket.c

test/csocket_loopback_b88.c line 1 still says #include "socket.c", so it no longer resolves:

pascal26:1: error: C include file not found: "socket.c"
  (searched: test/, lib/crtl/include/, <host system dirs>)

Reproduced at HEAD on this box. It is a compile failure, not a socket or networking problem — the loopback behaviour the test exists to check has not been exercised since the move.

The blamed sha is innocent. The stub names 330f62af7, which is tickets(B): record 0758d456e on the printf/strtod ticket — a docs-only commit that cannot break code. It was simply the sha the watcher happened to be testing; the real cause is earlier in the range. Same shape as [[bug-t-empty-range-regression-cannot-be-bisected]], which is why that ticket matters.

Lane: B (lib/crtl/** is Track B's ground, and the fix is in the test's include path — one line, either #include "sys/socket.c" or an include-path adjustment). T files, never fixes.

Resolved 2026-08-05

Fixed by 110949bfe before this ticket was noticed — the watcher auto-filed it against 330f62af78d0 while the fix was already in flight, so it sat at the head of Track B's ranked queue (prio 70) pointing at work that was done.

Root cause was as the summary said: 8d7c47f8f moved lib/crtl/src/socket.c to src/netinet/in.c (the path a header actually maps to, so the impl gets pulled), and the test reached in with #include "socket.c" through the Makefile's -Ilib/crtl/src. That include was itself a workaround for the very bug 8d7c47f8f fixed — the veneer was never pulled, so the test compiled the impl itself. With the pull working, the ordinary headers suffice.

Verified at HEAD:

tools/testmgr.py --tier native --job 'test-core#src:test/csocket_loopback_b88.c'
  PASS  test-core#459  test/csocket_loopback_b88.c
testmgr: GREEN

Process note: a watcher-filed regression outlives its fix unless someone closes it. This one had 58 commits in its bisect range and ranked above every real Track B item.