fix(ci): remove upgrade rollback signal race

packages/mosaic/framework/tools/quality/scripts/test-upgrade-rollback.sh

@@ -39,11 +39,12 @@ ORIG_PATH="$PATH"
 # loop — which would make the control a false negative. A root dotfile is
 # operator-owned (unknown→operator), so the sync loop skips it. Clean up on exit.
 STRIPPED="$FW/.install-rollback-control.tmp.sh"
+SIGNALED="$FW/.install-signal-control.tmp.sh"
 NOEXIT="$FW/.install-noexit-control.tmp.sh"
 D1CTRL="$FW/.install-d1guard-control.tmp.sh"
 D2CTRL="$FW/.install-d2guard-control.tmp.sh"
-rm -f "$STRIPPED" "$NOEXIT" "$D1CTRL" "$D2CTRL"
-trap 'rm -f "$STRIPPED" "$NOEXIT" "$D1CTRL" "$D2CTRL"' EXIT
+rm -f "$STRIPPED" "$SIGNALED" "$NOEXIT" "$D1CTRL" "$D2CTRL"
+trap 'rm -f "$STRIPPED" "$SIGNALED" "$NOEXIT" "$D1CTRL" "$D2CTRL"' EXIT
 
 pass=0; fail=0
 chk() { if eval "$2"; then echo "  ✓ $1"; pass=$((pass + 1)); else echo "  ✗ $1"; fail=$((fail + 1)); fi; }
@@ -180,41 +181,86 @@ chk "[control] without -E the mid-sync corruption survives (no rollback)" \
 # ── Part C: an INT/TERM interrupt must terminate, not resume (blocker-A) ──────
 # A bash signal trap that merely returns lets the script continue past the
 # interrupt — restoring the snapshot, then resuming the sync and reporting
-# success. We inject a SIGTERM mid-sync with a cp that SUCCEEDS (so set -e never
-# fires and ONLY the signal path governs), and assert the shipped installer
-# restores AND exits without reporting success. The control strips `exit 1` from
-# the trap and shows the buggy resume-to-success.
-make_term_shim() {
-  local dir="$1"
-  cat > "$dir/cp" <<SHIM
-#!/usr/bin/env bash
-dest="\${@: -1}"
-case "\$dest" in
-  */$POISON_REL)
-    kill -TERM "\$PPID" 2>/dev/null   # signal install.sh; the copy still succeeds
-    exec env PATH="$ORIG_PATH" cp "\$@" ;;
-esac
-exec env PATH="$ORIG_PATH" cp "\$@"
-SHIM
-  chmod +x "$dir/cp"
+# success. The earlier test used a child cp shim to signal its parent, making
+# child completion race Bash's interrupted wait. Concurrency is not part of the
+# guarded property: sync_framework_keep() runs in the installer's own Bash
+# process, and `kill` is a builtin. Generate two installer fixtures that signal
+# themselves at the same known mid-sync point. Their TERM handlers emit the same
+# observable before diverging, so missing signal delivery fails BOTH arms rather
+# than manufacturing a pass. The only semantic difference between fixtures is
+# the explicit `exit 1` whose load-bearing behavior this control proves.
+TERM_MARKER='[test-control] TERM handler entered'
+HANDLER_WITH_EXIT="trap 'echo \"$TERM_MARKER\" >&2; restore_snapshot; exit 1' TERM  # TEST-TERM-HANDLER"
+HANDLER_WITHOUT_EXIT="trap 'echo \"$TERM_MARKER\" >&2; restore_snapshot' TERM  # TEST-TERM-HANDLER"
+
+make_signal_installer() {
+  local output="$1" handler="$2"
+  local target_trap="trap 'restore_snapshot; exit 1' ERR INT TERM"
+  local target_cp='    cp "$abs" "$dst/$rel"'
+  local inject_open="    if [[ \"\$rel\" == \"$POISON_REL\" ]]; then"
+  local inject_kill='      kill -TERM "$$"  # TEST-TERM-INJECTION'
+  local inject_close='    fi'
+
+  if ! awk \
+    -v target_trap="$target_trap" -v target_cp="$target_cp" \
+    -v handler="$handler" -v inject_open="$inject_open" \
+    -v inject_kill="$inject_kill" -v inject_close="$inject_close" '
+      $0 == target_cp {
+        print inject_open
+        print inject_kill
+        print inject_close
+        injection_sites++
+      }
+      { print }
+      $0 == target_trap {
+        print handler
+        handler_sites++
+      }
+      END {
+        if (handler_sites != 1 || injection_sites != 1) exit 42
+      }
+    ' "$INSTALL" > "$output"; then
+    rm -f "$output"
+    fail "Could not construct the self-TERM control installer at the exact trap/copy sites"
+    exit 1
+  fi
+  chmod +x "$output"
 }
 
-# Run one keep-mode upgrade with the SIGTERM shim. Echoes "<exit>\t<out>\t<home>".
+make_signal_installer "$SIGNALED" "$HANDLER_WITH_EXIT"
+make_signal_installer "$NOEXIT" "$HANDLER_WITHOUT_EXIT"
+signal_fixture_ready() {
+  local fixture="$1" expected_handler="$2"
+  [[ "$(grep -cF '# TEST-TERM-INJECTION' "$fixture")" -eq 1 ]] \
+    && [[ "$(grep -cF '# TEST-TERM-HANDLER' "$fixture")" -eq 1 ]] \
+    && grep -Fqx "$expected_handler" "$fixture"
+}
+signaled_fixture_ready() { signal_fixture_ready "$SIGNALED" "$HANDLER_WITH_EXIT"; }
+noexit_fixture_ready() { signal_fixture_ready "$NOEXIT" "$HANDLER_WITHOUT_EXIT"; }
+chk "[signal] shipped fixture has exactly one self-TERM injection and marked handler" \
+  "signaled_fixture_ready"
+chk "[control] no-exit fixture has exactly one self-TERM injection and marked handler" \
+  "noexit_fixture_ready"
+chk "[control] removing the explicit TERM exit changes the fixture" \
+  "! cmp -s '$SIGNALED' '$NOEXIT'"
+
+# Run one keep-mode upgrade whose own shell delivers SIGTERM synchronously at
+# the selected copy. Echoes "<exit>\t<out>\t<home>".
 run_signal_upgrade() {
-  local installer="$1" H OUT SHIM rc
-  H=$(mktemp -d); OUT=$(mktemp); SHIM=$(mktemp -d)
+  local installer="$1" H OUT rc
+  H=$(mktemp -d); OUT=$(mktemp)
   seed_home "$H"
-  make_term_shim "$SHIM"
   set +e
-  PATH="$SHIM:$ORIG_PATH" \
+  PATH="$ORIG_PATH" \
     MOSAIC_HOME="$H" MOSAIC_INSTALL_MODE=keep MOSAIC_SYNC_ONLY=1 bash "$installer" >"$OUT" 2>&1
   rc=$?
   set -e 2>/dev/null || true
-  rm -rf "$SHIM"
   printf '%s\t%s\t%s\n' "$rc" "$OUT" "$H"
 }
 
-IFS=$'\t' read -r rcC OUTC HC < <(run_signal_upgrade "$INSTALL")
+IFS=$'\t' read -r rcC OUTC HC < <(run_signal_upgrade "$SIGNALED")
+chk "[signal] TERM handler observable fires exactly once" \
+  "[ \"\$(grep -cF '$TERM_MARKER' '$OUTC')\" -eq 1 ]"
 chk "[signal] SIGTERM mid-sync aborts non-zero (trap exits, does not resume)" \
   "[ '$rcC' -ne 0 ]"
 chk "[signal] restore_snapshot fires on the interrupt" \
@@ -222,13 +268,13 @@ chk "[signal] restore_snapshot fires on the interrupt" \
 chk "[signal] does NOT resume to report sync success after the interrupt" \
   "! grep -q 'file phase complete' '$OUTC'"
 
-# Control: strip `exit 1` from the signal trap → the handler returns, the script
-# resumes past the interrupt and wrongly reports success. In $FW so SOURCE_DIR resolves.
-sed "s/trap 'restore_snapshot; exit 1' ERR INT TERM/trap 'restore_snapshot' ERR INT TERM/" \
-  "$INSTALL" > "$NOEXIT"
-chk "[control] the exit-strip actually changed the installer" \
-  "! cmp -s '$INSTALL' '$NOEXIT'"
-IFS=$'\t' read -r _rcD OUTD HD < <(run_signal_upgrade "$NOEXIT")
+IFS=$'\t' read -r rcD OUTD HD < <(run_signal_upgrade "$NOEXIT")
+chk "[control] TERM handler observable fires exactly once" \
+  "[ \"\$(grep -cF '$TERM_MARKER' '$OUTD')\" -eq 1 ]"
+chk "[control] without 'exit 1' the handler restores before returning" \
+  "grep -q 'restoring previous state from snapshot' '$OUTD'"
+chk "[control] without 'exit 1' the installer exits zero after resuming" \
+  "[ '$rcD' -eq 0 ]"
 chk "[control] without 'exit 1' the trap resumes and reports sync success (the bug)" \
   "grep -q 'file phase complete' '$OUTD'"
 
@@ -309,10 +355,10 @@ chk "[control] without the D2 recovery line the operator gets no snapshot pointe
 # Reap any snapshot the reset-fail runs left in /tmp (reset failed → never cleaned).
 grep -o '/[^ ]*mosaic-snapshot[^ ]*' "$OUTH" 2>/dev/null | head -1 | while read -r s; do rm -rf "$s"; done
 
-# Cleanup ($STRIPPED / $NOEXIT / $D1CTRL / $D2CTRL are also removed by the EXIT trap).
+# Cleanup (generated installer controls are also removed by the EXIT trap).
 for d in "$HA" "$REFA" "$HB" "$REFB" "$HC" "$HD" "$HE" "$REFE" "$HF" "$REFF" "$HG" "$HH"; do rm -rf "$d"; done
 rm -f "$OUTA" "$OUTB" "$OUTC" "$OUTD" "$OUTE" "$OUTF" "$OUTG" "$OUTH" \
-  "$STRIPPED" "$NOEXIT" "$D1CTRL" "$D2CTRL"
+  "$STRIPPED" "$SIGNALED" "$NOEXIT" "$D1CTRL" "$D2CTRL"
 
 echo
 echo "RESULT: $pass passed, $fail failed"