[CI] replace E1 pixel races with observed protocol wait controls

This commit is contained in:
2026-09-20 10:27:41 -04:00
parent 08c63f77e2
commit 10bfd29fc5
8 changed files with 324 additions and 49 deletions
+15 -30
View File
@@ -33,8 +33,10 @@
# CONTROL", "is the E3(a) NEGATIVE CONTROL"), and it is tempting to grep for that. It is not
# evidence: it is written at config load, by every process in the run, whatever happens next. A
# setup abort would carry it too. It proves the knob was READ, never that the knob caused the red.
# Only the failing case's own diagnostic does that. ID-53 gives each Split entry a private
# library log: E1 reads its Fatal there; E3(a) reads the scenario assertion in ctest output.
# Only the failing case's own diagnostic does that. E1 now observes deterministic transport
# waits in dedicated CPU tests; E3(a) retains its scenario assertion and private library log.
# P5e ID-122: ordinary run-ahead draws need not fail with VERB_BARRIER=0, and the old
# overlap Fatal cannot fire with the default BATCH_WAITS=1. Pixel failures are not E1 evidence.
#
# Usage: split_negative_controls.sh [--self-test]
# CTEST ctest binary (default: ctest)
@@ -139,20 +141,14 @@ run_control() {
cat "${out}"
# Inspect JUnit before exit status or private Fatal: a skipped pre-flight can carry both.
label='E3(a)'
[ "${evidence}" != private-barrier-fatal ] || label=E1
python3 "${log_helper}" results "${manifest}" "${filter}" "${result}" "${label}" || exit 1
python3 "${log_helper}" results "${manifest}" "${filter}" "${result}" 'E3(a)' || exit 1
if [ "${control_rc}" -eq 0 ]; then
echo "::error::${name} left ${matched} split entries GREEN, so the knob it turns is not load-bearing and the gate it controls proves nothing."
exit 1
fi
# E1's MGLOG_F sink is the private file, never ctest's status or transcript.
if [ "${evidence}" = "private-barrier-fatal" ]; then
python3 "${log_helper}" evidence "${manifest}" "${filter}" \
'Fatal\{BarrierViolation, "[A-Za-z_][A-Za-z_0-9]*"\}' || exit 1
elif ! python3 "${log_helper}" assertion "${manifest}" "${filter}" "${result}" "${evidence}"; then
if ! python3 "${log_helper}" assertion "${manifest}" "${filter}" "${result}" "${evidence}"; then
echo "::error::${name} FAILED: red lacks its persistent-map push diagnostic. Required: ${evidence}"
exit 1
fi
@@ -168,26 +164,15 @@ run_control() {
echo "${name} turned ${matched} selected entries red, and the red carries the scenario's own diagnostic, as it must"
}
# E1: c1 ClientSession::EmitAndWait emits MGLOG_F Fatal{BarrierViolation, "<slot>"}.
# This asserts an observed overlap with the applier, which is timing-dependent.
# Without that Fatal in a fresh selected file E1 fails; pixel/status fallbacks do not count.
# Keep the ID-53-approved SmallRing selection as well as the default lane.
# ...MINUS TheServerStampedAVerbBoundaryOnThisDrawingFrame, and that exclusion is load-bearing
# rather than tidying (P5e, ID-122). That case is package gl's ID-115 POSITIVE CONTROL: it reads
# a PipeStats window out of the lane's own private log, so it runs ONLY in the strict-arming
# lane and skips everywhere else BY DESIGN, whatever this knob is set to. E1's selection began
# matching it the moment gl added it, and ID-62 makes a SKIPPED SELECTED entry a hard failure -
# correctly, because a control that silently loses its subjects proves nothing. E1 then reported
# "the knob killed the pre-flight, not the entry" while the knob was in fact working: four other
# selected entries went red with the abort in the same run. So the bug was in WHO E1 selects,
# not in what it asserts, and the fix excludes a case that can never run here rather than
# teaching E1 to tolerate skips - which would have thrown ID-62 away.
run_control "negative control E1 (MOBILEGL_IPC_VERB_BARRIER=0)" \
'DirectGLES\.Split\.(SmallRing\.)?(Triangle|ClearThenReadPixels)' \
'private-barrier-fatal' \
'' \
'TheServerStampedAVerbBoundaryOnThisDrawingFrame' \
MOBILEGL_IPC_VERB_BARRIER=0
# E1: the default arm must actually park for a kWaitApplied record and for a
# kWaitNone record whose server has no run-ahead cap. Removing the verb barrier
# must fail those exact return-before-apply assertions. A third positive case
# proves that a cap-authorized kWaitNone record does not wait. The peer releases
# on an observed Park or on the client's return, not on a sleep or pixel race.
# The helper demands green baseline, two named reds, and green restoration;
# skipped/missing cases, timeouts and unrelated failures never count as evidence.
python3 "$(dirname "$0")/wait_boundary_negative_control.py" \
--ctest "${CTEST}" --out "${CONTROL_TMPDIR}/e1" || exit 1
# E3(a): PersistentMapTracker::PushBlocksFor stops at blockBytes == 0 - deliberately, because 0
# is the negative control and not "unlimited". Two independent halves are now required: