betrusted-ed25519-verified/verification/selftest-axgate.sh
mrwulf 3bad795029 verification: pin the whole declaration surface (P1-b)
Phase 2b asks the kernel whether any AXIOM is declared under Proofs/. Phase 3
pins the cones of the named certificates. Between them sat every other
declaration in the corpus — around three thousand of them — and a helper lemma
quietly acquiring a hash oracle in its cone moved nothing either phase looked
at.

Phase 2c closes that. Ported from ltl-accumulator-verified, where a nine-attack
self-test proved a source-regex enumerator evadable by attributed, private,
indented and `instance` declarations and by a nested-namespace basename
collision. Reading the compiled environment sees what the kernel saw; no name
shape hides. Every constant contributes module, name, kind and full axiom cone,
and the observed set must equal inventory-allowlist.txt exactly in BOTH
directions, with a count trailer so a truncated run cannot pass as an empty
diff.

FOUR THINGS THIS BUILD GOT WRONG, each caught by a check rather than by review:

  - The number of inventory drivers is a per-repo FACT, not an assumption.
    dalek and anza cannot import their corpus as one environment (Proofs.Basic
    and Proofs.ConstSpecs both declare CurveFieldProofs.zero_spec); risc0 and
    betrusted have no Proofs.Basic at all. Determined by compiling a probe.
    check.sh now DISCOVERS its drivers from the filesystem instead of naming
    two, and the generator refuses to split out a module the repo lacks.

  - The split let one real declaration hide behind another's entry. Keyed on
    name alone, the two zero_specs produced byte-identical records, so 3022
    declarations were covered by 3021 allowlist entries. Caught by the count
    trailer. Every record now carries its originating module.

  - The gate's success line said "single sanctioned axiom", inherited from the
    accumulator's policy. This corpus permits NONE. A success message
    describing a different rule is how an assertion stops meaning anything.

  - selftest-axgate.sh lifted Phase 2b with a range ending at "Phase 3", so
    inserting Phase 2c between them made it swallow the new phase and die on
    variables only check.sh defines — surfacing as the BASELINE case failing,
    a self-test blaming a gate for its own extraction bug. Both self-tests now
    stop at the next phase marker whatever it is called, and refuse to run if
    they capture more than one phase. The guard is the fix; the range was the
    symptom.

WHAT THIS IS NOT, recorded in TRUSTED-BASE.md at the same length as the claim:

  - No independent cone walker. The accumulator cross-checks collectAxioms
    against a hand-written walker. Ported here it was wrong in BOTH directions
    on mathlib's inductive shapes: EdPoint gave [] against the kernel's three
    axioms, and once extended, ProjPoint gave three against the kernel's none.
    Two implementations disagreeing both ways are a second wrong answer, not a
    check. These cones rest on collectAxioms alone.

  - Thirteen Proofs/Scalar* modules are inventoried by nothing — the
    second-button seam, still open. Phase 2c names every uncovered module on
    every run so the omission is visible rather than inferred.

selftest-inventory.sh exercises the shipping gate with six cases, each
asserting a specific diagnostic, including the one that matters: a cone
widened by one oracle while name, module and kind stay put. Negative-tested by
disabling the gate's diff, which turns two cases red including one for the
wrong reason, correctly reported as such.

Verified green: 20 runs across the four repositories (four buttons, four
harness, four inventory, four axgate, four binding self-tests), zero red. The
four check-scalar.sh greens from the preceding sweep stand: that script neither
reads the pin file nor changed.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-07-30 01:20:20 +02:00

146 lines
6.7 KiB
Bash
Executable file

#!/usr/bin/env bash
# ─────────────────────────────────────────────────────────────────────────────
# selftest-axgate.sh — adversarial self-test for check.sh Phase 2b.
#
# An untested guard is decoration. This script breaks the thing Phase 2b
# guards and asserts the gate goes red FOR THE STATED REASON — a rejection by
# some other gate, or with some other message, fails the test too.
#
# It extracts Phase 2b out of check.sh at run time, so it attacks THE SHIPPING
# GATE rather than a copy that can drift away from it.
#
# Requires: Proofs/*.olean already built (run check.sh first, or any prior
# green build). Takes ~10 s; compiles one tiny throwaway module.
# ─────────────────────────────────────────────────────────────────────────────
set -uo pipefail
source ~/aeneas-toolchain/env.sh
HERE="$(cd "$(dirname "$0")" && pwd)"
AENEAS_LEAN="$AENEAS_HOME/backends/lean"
TIMEOUT="${LEAN_TIMEOUT:-600}"
export LEAN_MEM_MB="${LEAN_MEM_MB:-8192}"
CORES="${LEAN_MAX_CORES:-0-3}"
ATTACK="$HERE/Proofs/ZZSelftestAttack.lean"
STASH="$(mktemp -d)"
FAILURES=0
# Recorded before anything is touched, so the restore check compares against
# reality rather than assuming a pristine checkout.
TREE_AT_START="$(cd "$(dirname "$0")/.." && git status --porcelain -- verification/Proofs)"
cleanup() {
rm -f "$ATTACK" "${ATTACK%.lean}.olean"
[ -f "$STASH/FeQ.olean" ] && mv "$STASH/FeQ.olean" "$HERE/Proofs/FeQ.olean"
rm -rf "$STASH"
rm -f "$HERE"/.axgate-*.lean "$HERE"/.axgate-*.olean
}
trap cleanup EXIT INT TERM
# Phase 2b, lifted verbatim from the shipping button.
DRIVER="$STASH/phase2b.sh"
{
echo 'set -euo pipefail'
echo 'source ~/aeneas-toolchain/env.sh'
echo "HERE=\"$HERE\""
echo 'AENEAS_LEAN="$AENEAS_HOME/backends/lean"'
echo "TIMEOUT=$TIMEOUT; CORES=\"$CORES\""
# Stop at the NEXT phase marker, whatever it is called. A hardcoded
# terminator ("...to Phase 3") silently widens the moment a phase is
# inserted between the two: adding Phase 2c made this driver swallow 2c as
# well and die on variables that phase expects check.sh to have defined,
# which surfaced as the BASELINE failing — a self-test blaming a gate for
# its own extraction bug.
awk '/^# ── Phase 2b/{f=1} f&&/^# ── Phase /&&!/Phase 2b/{exit} f{print}' "$HERE/check.sh"
} > "$DRIVER"
if [ "$(wc -l < "$DRIVER")" -lt 40 ]; then
echo "FATAL: could not lift Phase 2b out of check.sh — the phase markers moved."
echo "This self-test must attack the shipping gate; refusing to run against nothing."
exit 1
fi
if [ "$(grep -c '^# ── Phase ' "$DRIVER")" -ne 1 ]; then
echo "FATAL: the lifted block spans more than one phase; the extraction is wrong."
grep '^# ── Phase ' "$DRIVER" | sed 's/^/ /'
exit 1
fi
expect() { # expect <name> <expected-rc> <required-substring>
local name="$1" want_rc="$2" want_txt="$3"
local out rc
out=$(bash "$DRIVER" 2>&1); rc=$?
if [ "$rc" -ne "$want_rc" ]; then
echo " FAIL $name: exit $rc, expected $want_rc"; FAILURES=$((FAILURES+1)); return
fi
if ! grep -qF "$want_txt" <<<"$out"; then
echo " FAIL $name: exit code right but diagnostic wrong (rejected for the wrong reason)"
echo " wanted substring: $want_txt"
echo " got: $(tr '\n' '|' <<<"$out" | cut -c1-300)"
FAILURES=$((FAILURES+1)); return
fi
echo " ok $name"
}
echo "=== selftest-axgate: attacking check.sh Phase 2b ==="
# ── 1. Baseline: the untouched repo must pass, and say how much it covered.
expect "baseline green, coverage reported" 0 "none is an axiom"
# ── 2. The attack Phase 1's grep cannot see: an indented top-level axiom.
# Lean accepts it; the repo then proves False; the source-text gate is blind.
cat > "$ATTACK" <<'EOF'
namespace ZZSelftestAttack
axiom cheat : ∀ (P : Prop), P
theorem repo_proves_false : False := cheat _
end ZZSelftestAttack
EOF
if grep -rnE '^(private |protected |noncomputable )*axiom ' "$HERE"/Proofs/*.lean >/dev/null 2>&1; then
echo " FAIL premise: Phase 1's grep sees the attack — this test no longer tests what it claims"
FAILURES=$((FAILURES+1))
else
echo " ok premise: Phase 1's source-text grep is blind to this attack"
fi
(cd "$AENEAS_LEAN" && lake env bash -c "
set -euo pipefail
cd '$HERE/gen' && export LEAN_PATH=\"\$LEAN_PATH:\$PWD:$HERE\"
cd '$HERE'
LEAN_TIMEOUT=$TIMEOUT LEAN_MAX_CORES=$CORES '$HERE/lean-guard' 'Proofs/ZZSelftestAttack.lean'
") >/dev/null 2>&1 || { echo " FAIL setup: the attack module did not compile"; FAILURES=$((FAILURES+1)); }
expect "indented axiom caught kernel-side" 1 "AXIOM DECLARED under Proofs/"
rm -f "$ATTACK" "${ATTACK%.lean}.olean"
# ── 3. Vacuity: delete a compiled module. "Nothing found" must not pass for
# "nothing wrong" — the gate has to notice it stopped covering something.
mv "$HERE/Proofs/FeQ.olean" "$STASH/FeQ.olean"
expect "missing .olean is a failure, not a vacuous pass" 1 "COVERAGE MISMATCH"
mv "$STASH/FeQ.olean" "$HERE/Proofs/FeQ.olean"
# ── 4. Litter: neither path may leave the temp gate source or its artifact
# behind (this repo accumulated 101 orphan .olean files exactly that way).
if ls "$HERE"/.axgate-* >/dev/null 2>&1; then
echo " FAIL litter: temp gate files survived a run"; FAILURES=$((FAILURES+1))
else
echo " ok no litter left by either the green or the red path"
fi
# ── 5. Restored: the self-test must leave the working tree exactly as it found
# it. Compared against the state recorded at START, not against a pristine
# checkout — files can be legitimately uncommitted while work is in flight,
# and a test that assumes otherwise reports its own premise as a failure.
# Compare the two states AS STRINGS. Comparing `echo "$VAR"` against a raw
# command substitution is asymmetric: for a clean tree the variable is empty
# and `echo` still emits one blank line while the command emits none, so the
# check reports a spurious difference exactly when nothing is wrong.
TREE_NOW="$(cd "$HERE/.." && git status --porcelain -- verification/Proofs)"
if [ "$TREE_AT_START" != "$TREE_NOW" ]; then
echo " FAIL restore: Proofs/ differs from how this test found it:"
diff <(printf '%s\n' "$TREE_AT_START") <(printf '%s\n' "$TREE_NOW") | sed 's/^/ /'
FAILURES=$((FAILURES+1))
else
echo " ok working tree restored to its starting state"
fi
echo ""
if [ "$FAILURES" -eq 0 ]; then
echo "SELFTEST PASSED — Phase 2b rejects what it claims to reject, for the stated reason."
exit 0
fi
echo "SELFTEST FAILED: $FAILURES check(s) did not behave as claimed."
exit 1